Oral galactagogues (natural therapies or drugs) for increasing breast milk production in mothers of non‐hospitalised term infants

Abstract

Background

Many women express concern about their ability to produce enough milk, and insufficient milk is frequently cited as the reason for supplementation and early termination of breastfeeding. When addressing this concern, it is important first to consider the influence of maternal and neonatal health, infant suck, proper latch, and feeding frequency on milk production, and that steps be taken to correct or compensate for any contributing issues.

Oral galactagogues are substances that stimulate milk production. They may be pharmacological or non‐pharmacological (natural). Natural galactagogues are usually botanical or other food agents. The choice between pharmacological or natural galactagogues is often influenced by familiarity and local customs. Evidence for the possible benefits and harms of galactagogues is important for making an informed decision on their use.

Objectives

To assess the effect of oral galactagogues for increasing milk production in non‐hospitalised breastfeeding mother‐term infant pairs.

Search methods

We searched the Cochrane Pregnancy and Childbirth Group's Trials Register, ClinicalTrials.gov, the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP), Health Research and Development Network ‐ Phillippines (HERDIN), Natural Products Alert (Napralert), the personal reference collection of author LM, and reference lists of retrieved studies (4 November 2019).

Selection criteria

We included randomised controlled trials (RCTs) and quasi‐RCTs (including published abstracts) comparing oral galactagogues with placebo, no treatment, or another oral galactagogue in mothers breastfeeding healthy term infants. We also included cluster‐randomised trials but excluded cross‐over trials.

Data collection and analysis

We used standard Cochrane Pregnancy and Childbirth methods for data collection and analysis. Two to four review authors independently selected the studies, assessed the risk of bias, extracted data for analysis and checked accuracy. Where necessary, we contacted the study authors for clarification.

Main results

Forty‐one RCTs involving 3005 mothers and 3006 infants from at least 17 countries met the inclusion criteria. Studies were conducted either in hospitals immediately postpartum or in the community. There was considerable variation in mothers, particularly in parity and whether or not they had lactation insufficiency. Infants' ages at commencement of the studies ranged from newborn to 6 months. The overall certainty of evidence was low to very low because of high risk of biases (mainly due to lack of blinding), substantial clinical and statistical heterogeneity, and imprecision of measurements.

Pharmacological galactagogues

Nine studies compared a pharmacological galactagogue (domperidone, metoclopramide, sulpiride, thyrotropin‐releasing hormone) with placebo or no treatment.

The primary outcome of proportion of mothers who continued breastfeeding at 3, 4 and 6 months was not reported. Only one study (metoclopramide) reported on the outcome of infant weight, finding little or no difference (mean difference (MD) 23.0 grams, 95% confidence interval (CI) ‐47.71 to 93.71; 1 study, 20 participants; low‐certainty evidence).

Three studies (metoclopramide, domperidone, sulpiride) reported on milk volume, finding pharmacological galactagogues may increase milk volume (MD 63.82 mL, 95% CI 25.91 to 101.72; I² = 34%; 3 studies, 151 participants; low‐certainty evidence). Subgroup analysis indicates there may be increased milk volume with each drug, but with varying CIs.

There was limited reporting of adverse effects, none of which could be meta‐analysed. Where reported, they were limited to minor complaints, such as tiredness, nausea, headache and dry mouth (very low‐certainty evidence). No adverse effects were reported for infants.

Natural galactagogues

Twenty‐seven studies compared natural oral galactagogues (banana flower, fennel, fenugreek, ginger, ixbut, levant cotton, moringa, palm dates, pork knuckle, shatavari, silymarin, torbangun leaves or other natural mixtures) with placebo or no treatment.

One study (Mother's Milk Tea) reported breastfeeding rates at six months with a concluding statement of "no significant difference" (no data and no measure of significance provided, 60 participants, very low‐certainty evidence).

Three studies (fennel, fenugreek, moringa, mixed botanical tea) reported infant weight but could not be meta‐analysed due to substantial clinical and statistical heterogeneity (I² = 60%, 275 participants, very low‐certainty evidence). Subgroup analysis shows we are very uncertain whether fennel or fenugreek improves infant weight, whereas moringa and mixed botanical tea may increase infant weight compared to placebo. Thirteen studies (Bu Xue Sheng Ru, Chanbao, Cui Ru, banana flower, fenugreek, ginger, moringa, fenugreek, ginger and turmeric mix, ixbut, mixed botanical tea, Sheng Ru He Ji, silymarin, Xian Tong Ru, palm dates; 962 participants) reported on milk volume, but meta‐analysis was not possible due to substantial heterogeneity (I² = 99%). The subgroup analysis for each intervention suggested either benefit or little or no difference (very low‐certainty evidence). There was limited reporting of adverse effects, none of which could be meta‐analysed. Where reported, they were limited to minor complaints such as mothers with urine that smelled like maple syrup and urticaria in infants (very low‐certainty evidence).

Galactagogue versus galactagogue

Eight studies (Chanbao; Bue Xue Sheng Ru, domperidone, moringa, fenugreek, palm dates, torbangun, moloco, Mu Er Wu You, Kun Yuan Tong Ru) compared one oral galactagogue with another. We were unable to perform meta‐analysis because there was only one small study for each match‐up, so we do not know if one galactagogue is better than another for any outcome.

Authors' conclusions

Due to extremely limited, very low certainty evidence, we do not know whether galactagogues have any effect on proportion of mothers who continued breastfeeding at 3, 4 and 6 months. There is low‐certainty evidence that pharmacological galactagogues may increase milk volume. There is some evidence from subgroup analyses that natural galactagogues may benefit infant weight and milk volume in mothers with healthy, term infants, but due to substantial heterogeneity of the studies, imprecision of measurements and incomplete reporting, we are very uncertain about the magnitude of the effect. We are also uncertain if one galactagogue performs better than another. With limited data on adverse effects, we are uncertain if there are any concerning adverse effects with any particular galactagogue; those reported were minor complaints.

High‐quality RCTs on the efficacy and safety of galactagogues are urgently needed. A set of core outcomes to standardise infant weight and milk volume measurement is also needed, as well as a strong basis for the dose and dosage form used.

Plain language summary

Milk boosters (galactagogues) for mothers breastfeeding their healthy infants born at term

What is the issue?

We set out to determine the ability of milk boosters taken by mouth (medicine, herb or food) to increase milk production in breastfeeding mothers of healthy infants born at term. Poor milk supply is often given as the reason for early supplementation and weaning sooner than desired. A range of factors, including mother's and baby's health, baby's sucking skills, proper latch and frequency of feeds, can affect milk production. Every attempt should first be made to identify and correct the causes for low milk production before trying a milk booster.

Why is this important?

Inadequate milk production can be distressing for mothers and threatening to babies' health. The choice of milk booster is often influenced by familiarity or local customs. Some mothers may prefer medications, while others prefer natural remedies. Evidence for the possible benefits and harms of milk boosters is important to assist mothers in making informed decisions.

What evidence did we find?

We searched for evidence from randomised controlled studies up to 4 November 2019 and identified 41 eligible studies involving 3005 mothers and 3006 infants from at least 17 countries. The studies varied widely in babies ages, type of milk boosters investigated, how long they were taken, and how outcomes were reported. Medications included sulpiride, metoclopramide, domperidone and thyrotropin‐releasing hormone. Natural interventions included banana flower, fennel, fenugreek, ginger, ixbut, levant cotton, moringa, palm dates, pork knuckle, shatavari, silymarin, torbangun leaves, and a variety of natural mixtures as teas or soups.

Milk‐boosting medication

Nine studies compared a milk‐boosting medication with placebo or no treatment. None reported exclusive breastfeeding rates at 3. 4 or 6 months and only one (metoclopramide, 20 participants) reported on weight gain in infants receiving only their mothers' own milk, with better results in the milk booster group. Three studies that tracked milk volume (domperidone, metoclopramide, sulpiride; 151 participants) reported more milk in the booster groups, though the certainty of the evidence was low. Adverse effects were poorly reported. Where mentioned, they were limited to minor complaints, such as tiredness, nausea, decreased appetite, headache and dry mouth.

Natural milk boosters

Twenty‐seven studies compared natural milk boosters with placebo or no treatment. Only one (Mother’s Milk Tea; 60 participants) examined the impact on breastfeeding rates, reporting "no significant difference at 6 months" without providing any data (very low‐certainty evidence). Three studies (275 participants) reported infant weight, two of which (moringa, mixed botanical tea) reported higher gains in the milk booster group, while the other study (fennel and fenugreek) was inconclusive on whether infant weight gain improved with the milk boosters. In the 13 studies tracking changes in milk volume (Bu Xue Sheng Ru, Chan Bao, Cui Ru, banana flower, fenugreek, ginger, moringa, fenugreek, ginger and turmeric mix, ixbut, mixed botanical tea, Sheng Ru He Ji, silymarin, Xian Tong Ru, palm dates; 962 participants), some showed benefits and others little or no difference, so we are very uncertain about the results for milk volume. Adverse effects were poorly reported. Where mentioned, they were limited to minor complaints, such as mothers with urine that smells like maple syrup and rash in infants (very low‐certainty evidence).

One milk booster compared with another

Eight studies (Chanbao, Bue Xue Sheng Ru, domperidone, moringa, fenugreek, palm dates, torbangun, moloco, Mu Er Wu You, Kun Yuan Tong Ru) compared one milk booster with another. There was only one small study for each particular match‐up, hence we cannot be certain if any one milk booster truly worked better than another.

What does this mean?

There is limited evidence that milk‐boosting medications may increase milk volume and that natural milk boosters may improve milk volume and infants' weight, but we are very uncertain about the supporting evidence. Due to limited information, we are also uncertain if there are any risks to the mother or baby in taking any particular milk booster. More high‐quality studies are needed to increase our certainty about the effects of milk boosters.

Summary of findings

Summary of findings 1. Pharmacological oral galactagogues compared to placebo or no treatment for increasing breast milk production in mothers of non‐hospitalised term infants.

Pharmacological oral galactagogues compared to placebo or no treatment for increasing breast milk production in mothers of non‐hospitalisedterm infants
Patient or population: increasing breast milk production in mothers of non‐hospitalised term infants Setting: community settings in Belgium, Iran, Japan, Thailand Intervention: pharmacological oral galactagogues Comparison: placebo or no treatment
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with placebo or no treatment	Risk with pharmacological oral galactagogues
Proportion of mothers who continued breastfeeding (exclusive or any) at 3, 4 and 6 months	‐	‐	‐	‐	‐	No studies reported this outcome
Infant weight at the end of the study (grams) (in trials where they were only receiving own mother's milk)‐ metoclopramide (follow‐up: 15 days)	The mean infant weight gain in the control group was 328.5g	MD 23.00 grams higher (47.71 lower to 93.71 higher)	‐	20 (1 RCT, 1 intervention)	⊕⊕⊝⊝ Lowa
Volume of breast milk at the latest time measured (mL) ‐ domperidone, metoclopramide, sulpiride (follow‐up: 4 days to 8 days)	The mean milk volume ranged across control groups from 91.4 mL to 247.4 mL	MD 63.82 mL higher (25.91 higher to 101.72 higher)	‐	151 (3 RCTs, 3 interventions)	⊕⊕⊝⊝ Lowa	Subgroup analysis for each intervention suggested either benefit or little or no difference
Adverse effects for the infant or mother (follow‐up: 5 days to 28 days)	Mothers ‐ Metoclopromide: tiredness with or without headache or nausea (6/11 versus 3/14 controls) ‐ Domperidone: dry mouth (7/22) versus 0/23 controls); no extrapyramidal effects in either group ‐ Sulpiride: tiredness (2/14 versus 0/12 controls), headache (1/14 versus 0/12 controls) ‐ Thyrotropin‐releasing hormone: none reported Infants ‐ No adverse effects reported in infants		‐	133 (5 RCTs, 4 interventions)	⊕⊝⊝⊝ Very lowb,c	Adverse effects were generally poorly reported and those listed here are what is reported in the studies. An overall summary of adverse effects reported in the included studies in provided in Table 5.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; MD: mean difference; RCT: randomised controlled trial.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded two levels for serious imprecision due to a single study with sparse data (the 95% CI includes both important benefits and important harms).
^bDowngraded one level for inconsistency; studies and interventions were too diverse to be meaningfully combined.
^cDowngraded one level for imprecision due to sparse data and wide confidence intervals.

Summary of findings 2. Natural oral galactagogues compared to placebo or no treatment for increasing breast milk production in mothers of non‐hospitalised term infants.

Natural oral galactagogues compared to placebo or no treatment for increasing breast milk production in mothers of non‐hospitalisedterm infants
Patient or population: increasing breast milk production in mothers of non‐hospitalised term infants Setting: community settings in China, Egypt, Guatemala, Iran, Malaysia, Peru, Thailand, Turkey, the Philippines, USA Intervention: natural oral galactagogues Comparison: placebo or no treatment
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with placebo or no treatment	Risk with natural oral galactagogues
Proportion of mothers who continued breastfeeding (exclusive or any) at 3, 4 and 6 months	See comment		‐	60 (1 RCT, 1 intervention)	⊕⊕⊕⊝ Very lowa,b	No meta‐analysis. One study reported "no significant difference" in breastfeeding rates at six months for Mother's Milk Tea (no data and no measure of significance). Not reported at 3 or 4 months
Infant weight (where they were only receiving own mother's milk) at the end of the study (grams) (follow‐up at 1 month or at infant age 1 to 5 months)	Meta‐analysis was not conducted due to substantial heterogeneity (I2 = 60%)	‐	‐	275 (3 RCTs, 4 interventions)	⊕⊝⊝⊝ Very lowc,d	No meta‐analysis. Subgroup analysis for each intervention suggested either benefit or little or no difference.
Volume of breast milk at the latest time measured (mL) (follow‐up: ranged from infant age 2 days to 2 months)	Meta‐analysis was not conducted due to substantial heterogeneity (I2 = 99%). However, due to heterogeneity, imprecision and high risk of bias the overall certainty of evidence was very low.	‐	‐	962 (13 RCTs, 14 interventions)	⊕⊝⊝⊝ Very lowc,d,e	No meta‐analysis. Subgroup analysis for each intervention suggested either benefit or little or no difference.
Adverse effects for the infant or mother follow‐up: ranged from 36 weeks gestation up to the infant age of 12 months	Almost all reported "no adverse effects" Mothers Fenugreek, ginger and tumeric mix: maple syrup urine odour (2/25 versus 0/25 controls), excessive flatus (2/25 versus 2/25 controls) Infants Shirafza drops: nausea and urticaria (2 versus 0 controls, denominator not available)		‐	(10 RCTs, 7 interventions)	⊕⊝⊝⊝ Very lowb,f,g	Adverse effects were poorly reported. An overall summary of adverse effects reported in the included studies in provided in Table 5.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; MD: mean difference; RCT: randomised controlled trial.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded for limitations: very serious concerns as authors only reported as "no significant difference" without numerical values.
^bDowngraded two levels for serious imprecision due to a single study with sparse data. 
^cDowngraded one level for inconsistency: substantial statistical heterogeneity.
^dDowngraded one level for imprecision due to sparse data and wide confidence intervals. 
^eDowngraded one level for limitations: high or unclear risk of performance and detection bias.
^fDowngraded one level for limitations: lack of blinding in most studies.
^gDowngraded for inconsistency: studies and interventions were too diverse to be meaningfully combined.

Background

Description of the condition

The critical importance of breastfeeding for mothers and infants in both the short and long term has been well‐documented and recognized (AAP 2012; WHO 2003). The World Health Organization (WHO) recommends that infants be breastfed exclusively (no other liquids or solids) for the first 6 months, and that breastfeeding continue for a minimum of 2 years (WHO 2003). This recommendation is affirmed by a recent systematic review (Kramer 2012). If the infant routinely receives other liquids or solids in addition to any amount of breast milk, this is termed as ‘any breastfeeding' (Labbok 1990).

Breastfeeding is biologically symbiotic, with the mother's body depending upon feedback from the infant in order to produce and deliver sufficient milk (Kent 2007). Successful breastfeeding starts with adequate maternal mammary tissue, intact nerves and ducts, and a favourable hormonal milieu (Kent 2012; Livingstone 1997; Marasco 2020). Feeding practices such as effective positioning and attachment of the infant and nursing on demand both day and night are crucial to ensure that breastfeeding succeeds. Exclusively breastfed infants consume approximately 750 mL to 900 mL a day, but there is quite a range reported (478 mL to 1356 mL) among thriving infants (Kent 2006; Nielsen 2011). Milk production normally rises rapidly with the onset of secretory activation, with approximately 92% of mothers of full‐term infants reaching a minimum of 440 mL daily output by 2 weeks postpartum (Kent 1999; Kent 2016; Neville 1988). When breastfeeding is going well, most infants will regain their birthweight by 7 days of age (Kellams 2017; Macdonald 2003).

When the amount of milk produced is not measured, infant weight gain as frequently used in animal lactation studies may be an acceptable proxy for adequate milk production (Anderson 2013). However, breastfeeding success is not only measured by the volume of milk produced but by the duration of exclusive breastfeeding.

Many breastfeeding women express concern about their ability to produce sufficient milk for their infant (Amir 2006; Brown 2014; Wagner 2013). Insufficient milk production is a frequently cited reason for early termination of breastfeeding (Stuebe 2014a), and in many cases it is a misperception due to an incorrect interpretation of the situation (Stuebe 2014b). While it is difficult to quantify how many women actually have inadequate production, it is important to consider the influence of a range of factors, including maternal and neonatal health, good breast attachment, frequency of feeds, maternal rest, maternal confidence, family and peer support, pain (including painful nipples), supplementary bottle feeds, and so on (Amir 2006). Lactation is a symbiotic process, and as such the cause of problems may originate from the mother, the infant, or both (Amir 2006; Stuebe 2014b). Abnormal breast anatomy secondary to previous breast surgery, especially reduction or augmentation of tissue, or primary mammary gland hypoplasia (underdevelopment), may limit milk production capability (Arbour 2013; Cassar‐Uhl 2014; Huggins 2000; Neifert 1985). Conditions that interfere with or alter lactation hormones can also affect milk synthesis (Nommsen‐Rivers 2012). These include but are not limited to: diabetes (Nommsen‐Rivers 2012), polycystic ovarian syndrome (PCOS) (Harrison 2016; Marasco 2000), hyperandrogenism (Betzold 2004; Carlsen 2010), obesity (Bever Bavendure 2015; Nommsen‐Rivers 2016; Rasmussen 2007; Stuebe 2014a), thyroid disease (Alexander 2017; Marasco 2006; Speller 2012), gestational theca lutein cyst (Hoover 2002), and postpartum haemorrhage or hypopituitarism (Dökmetas 2006; Willis 1995). In a recent study, one‐third of mothers with gestational diabetes had delayed onset of lactation (Matias 2014), and another study hypothesized that lower insulin sensitivity may lead to over‐expression of a gene responsible for regulating an insulin signalling pathway, resulting in decreased milk production (Lemay 2013), a finding that is being validated by new research (Nommsen‐Rivers 2016; Nommsen‐Rivers 2017). Medications that intersect with the hormonal pathways responsible for lactation, such as hormonal contraceptives, can also have a suppressive effect on lactation (Berens 2015; Brownell 2011; Pieh 2015).

The infant can also cause low milk production in the mother by failing to remove enough milk in an effective manner, as milk production is dependent upon both the volume of milk removed and the quality of the sucking stimulation (Geddes 2008; Zhang 2016). Examples of infant factors include oromotor dysfunction, hypotonia, and abnormalities of the oral cavitysuch as clefts of the hard or soft palate, or a tight lingual frenulum (Amir 2006; Garbin 2013; Kent 2012; McClellan 2012). In practice, it is not always easy to isolate a single cause for low milk production, and at times the root cause may be multifactorial (Marasco 2015; Riddle 2017).

It has been postulated that for a well‐meaning mother who was prepared to exclusively breastfeed her infant, the inability to do so might result in a major psychological setback (Cannon 2005; Watkinson 2016; Williams 2002). Furthermore, a common reason reported by mothers for early termination of breastfeeding is the belief that their milk production is inadequate for the infant's needs (Amir 2006; Kent 2012; Stuebe 2014a; Stuebe 2014b; Winterfeld 2012). Optimising effective feeding practices should be the priority in managing low milk production for women with healthy full‐term infants (Brodribb 2018; Kamala 2015; Kent 2012). When these measures fail or are only partially effective, galactagogues could be the second‐tier option for improving milk production in order for the mother and infant to have a better breastfeeding experience (Kent 2012; Sim 2014).

Description of the intervention

Galactagogues are substances that stimulate breast milk production (Gabay 2002; Tabares 2014). These substances are differentiated from nutrients which are essential for lactation, and whose absence may contribute to a decrease in milk synthesis. Supplementation of an essential nutrient to a nutritionally deficient mother may improve lactation (a 'lactogenic effect'), but will not improve milk output if there are no deficiencies. For instance, zinc research elucidating its critical role in lactation suggests a likely lactogenic effect of supplementation for mothers with simple zinc deficiency (Lee 2016a; Lee 2016b; O'Brien 2007). The focus of the current review is on lactation‐stimulating substances that are absorbed through the digestive tract. We have excluded galactagogues that are absorbed exclusively in the oral mucosa such as sublingual or buccal routes.

Generally, oral galactagogues can be divided into pharmacological and natural. There are no pharmaceuticals manufactured for the purpose of stimulating lactation; all use is off label. Among the pharmacological galactagogues, domperidone is frequently regarded as the drug of choice and is perceived to be safe for both the mother and infant (Haase 2016; Winterfeld 2012), though not licensed for use in all countries (Anderson 2017; FDA 2013). Some concerns have been raised regarding possible cardiac effects in vulnerable individuals, but they seem to be rare (Buffery 2015; Grzeskowiak 2019; Hale 2018). Metoclopramide and sulpiride are also used but reportedly have more potential side effects. A worldwide survey of 1990 mothers comparing side effects of domperidone versus metoclopramide found that they occurred in a small percentage of women overall, with those taking domperidone less likely to report a side effect than those taking metoclopramide (Hale 2018).

For mothers who are not keen to use drugs, there is a wide range of natural galactagogues that have been used by generations of women around the world for increasing milk production. In many traditional cultures, foods and herbs form an integral part of the overall health strategy. New mothers typically are offered special foods or drinks to ensure a bountiful supply for the infant (Jacobson 2004; Rajith 2010). In the Phillippines and in parts of Africa and India, dishes featuring moringa are offered both prophylactically and as a treatment for low milk production (Rajith 2010). Europeans, on the other hand, are more likely to use galactagogue decoctions and brew tea infusions to treat low production, rather than for prophylaxis (Bruckner 1993). Nonculinary natural galactagogues such as goat's rue or blessed thistle are much less likely to be employed unless problems arise. In Western cultures utilising allopathic (mainstream) medicine, natural galactagogues are viewed with both suspicion and skepticism and may be employed only after all other measures have failed (Brodribb 2018; Forinash 2012; Kent 2012). When taken, they are often ingested in non‐traditional forms, such as capsules and tincture extracts that do not necessarily replicate historical usage (Humphrey 2003; Jacobson 2004). Some of the more popular natural galactagogues include, but are not limited to: fenugreek (Trigonella foenum‐graecum), fennel (Foeniculum vulgare), blessed thistle (Cnicus benedictus), torbangun leaves (Coleus amboinicus Lour), shatavari (Asparagus racemosus), anise or aniseed (Pimpinella anisum), milk thistle (Silybum marianum), barley (Hordeum vulgare), malunggay (Moringa oleifera), and goat's rue (Galega officinalis) (Abascal 2008; Bingel 1994; Bruckner 1993; Marasco 2020; Sim 2014).

How the intervention might work

The mechanism of galactagogue action likely varies (Bingel 1994; Grzeskowiak 2019; MacIntosh 2003; MacIntosh 2004; Whitten 2010), and can be understood through three different paradigms: those that work on milk synthesis directly, those that improve milk synthesis by correcting a lactation‐critical hormone or hormone receptor problem, and those that stimulate lactation by improving milk removal via a stronger milk ejection reflex. While pharmacological galactagogues are believed to work primarily by stimulating prolactin (Grzeskowiak 2019; Mortel 2013), less is known about the mechanism(s) by which natural galactagogues are able to enhance lactation. It has been hypothesized that milk output, and thus the shape of the ‘lactation curve,’ is a function of the number of milk‐secreting cells and their secretory rate (Capuco 2003). Any increase in the number of these cells or the rate of their activity, or both, would then increase milk production (Mortel 2013). Indeed, many reputed natural galactagogues have oestrogenic properties that may stimulate mammary alveolar growth, prolactin, or both (Tabares 2014). Others are considered oxytocic and are theorized to work indirectly by triggering a stronger milk ejection reflex, leading to increased milk removal and thus stimulation of a higher rate of milk production (Bingel 1994; Humphrey 2007; Javan 2017). As the underlying causes for low milk production are variable, galactagogues may not work equally in all circumstances (Humphrey 2003; Marasco 2020), especially in the context of hormonal problems or mammary hypoplasia (Arbour 2013).

In addition, particularly for natural galactagogues, the part of the plant used (leaf, root, seed, etc.), and the form in which these parts are administered (e.g. as a tea, tincture, decoction or powdered leaf) may influence therapeutic effects (Brinker 1999; Garg 2010; Wilinska 2015). Dosages for galactagogues are largely unquestioned and untested, and may or may not be sufficient to evoke the maximum therapeutic effect possible (Humphrey 2007; Marasco 2020). In the case of domperidone, 30 mg daily is commonly used in Australia, sometimes up to 60 mg (Grzeskowiak 2015), while mothers in North America may take 80 mg and sometimes up to double this amount (Papastergiou 2013; Sewell 2017). Yet, 30 mg is the dose most commonly used in efficacy research (Paul 2015). It is, therefore, important to take note of not just the substance, but the form and dosage as well (Betz 2014). It is beyond the scope of this review to determine if a particular study is using the appropriate dose for the intended effect.

Table 3 and Table 4 summarize the hypothesized mechanism of action and the potential adverse effects of some commonly used galactagogues.

1. Pharmacological galactagogues.

Oral pharmacological galactagogue	How it might work	Possible harms	References
Domperidone	Peripherally‐acting dopamine D2‐receptor antagonist, increases prolactin release from the pituitary gland	Headaches, somnolence, abdominal pain, diarrhoea; may also cause weight gain. Increased risk of cardiac problems if history of prolonged QT interval, especially at high doses	Anderson 2013; Barone 1999; Doggrell 2014; Forinash 2012; Hale 2007; Hale 2018; Zuppa 2010
Metoclopramide	Increases prolactin levels by anti‐dopaminergic effects	Crosses the blood brain barrier; may cause restlessness, drowsiness, fatigue, depression and involuntary body movements	Anderson 2013; Forinash 2012; Hale 2007; Hale 2018; Zuppa 2010
Sulpiride	Increases prolactin levels by anti‐dopaminergic effects	Anti‐psychotic medication; may cause headache, fatigue, weight gain, extrapyrimidal effects	Forinash 2012; Grzeskowiak 2019
Thyrotrophin‐ releasing hormone (TRH)	Increases prolactin, likely via stimulation of calcium release, which induces prolactin gene expression and release	Changes in blood pressure, headaches, nausea; could induce hyperthyroidism	Bingel 1994; Grzeskowiak 2019; Tabares 2014

2. Natural oral galactagogues.

Natural oral galactagogue	Botanical part	How it might work	Possible harms	References
Alfalfa (Medicago sativa)	Leaf	Phytoestrogens may stimulate prolactin, mammary tissue. Provides nutrients essential to milk production	Loose stools, may be allergenic for some people; seeds may increase risk of sunburn	Bingel 1994; Bnouham 2010; Goksugur 2014; Humphrey 2007; Javan 2017; LactMed 2006 ; Mills 2006; Nice 2015; Rajagopal 2016; Scott 2005
Anise or aniseed (Pimpinella anisum)	Fruit*	Contains trans‐anethole, considered weakly oestrogenic; the aromatic compound in anise may act as a dopamine receptor antagonist	Possible allergen for some people	Abascal 2008; Bingel 1994; Bruckner 1993; Goksugur 2014; Humphrey 2007; Low Dog 2009; Nice 2015; Romm 2010; Tabares 2014; Vargova 2018
Banana flower (Musa x paradisiaca)	Blossom	Increased prolactin levels in rats	None known. Commonly consumed food in Asia	Javan 2017; Luecha 2013; Mahmood 2012
Barley (Hordeum vulgare)	Grain	Polysaccharide stimulates prolactin	None known. Commonly consumed grain, also used to make beer	Bingel 1994; Humphrey 2007; Koletzko 2000; MacIntosh 2004; Nice 2015; Sawagado 1988; Scott 2005
Blackseed or Black cumin (Nigella sativa)	Seed	Stimulated mammary gland proliferation in rats	Possible allergic contact dermatitis with oil	Abu‐Rabia 2005; Dandotiya 2013; Humphrey 2007; Javan 2017; LactMed 2006; Luecha 2013; Rajagopal 2016; Yashmin 2017
Blessed thistle (Cnicus benedictus)	Aerial parts	Reputedly stimulates the flow of blood to the mammary glands	Possible allergen for some people	Abascal 2008; Bingel 1994; Goksugur 2014; Humphrey 2007; LactMed 2006; Nice 2015; Vargova 2018; Zapantis 2012
Caraway (Carum carvi)	Fruit*	Reputedly oestrogenic	Possible allergen for some people	Abu‐Rabia 2005; Bingel 1994; Goksugur 2014; Johri 2011; LactMed 2006; Nice 2015; Vargova 2018
Chasteberry (Vitex agnus‐castus)	Berry	In historically used dosages, appears to stimulate prolactin	Diarrhoea, heartburn, flatulence, itching, rash; large doses suppressed prolactin in men; impact on women unknown	Abu‐Rabia 2005; Bingel 1994; Ergol 2016; Goksugur 2014; Humphrey 2007; Javan 2017; Mills 2006; Nice 2015; Scott 2005; Zapantis 2012
Chickpea (Cicer arietinum)	Seed	Oestrogenic isoflavones may stimulate prolactin secretion	None known; common food	Javan 2017; Nice 2015; Scott 2005
Coriander (Coriandrum sativum)	Fruit*	Unknown	Allergic reactions, photosensitivity, contact dermatitis	Abu‐Rabia 2005; Ergol 2016; Goksugur 2014; LactMed 2006; Nice 2015
Cotton seed or Levant cotton (Gossypium herbaceum)	Seed	Stimulated prolactin in animal studies. May assist milk ejection reflex.	Hypokalaemia possible at high doses	Abascal 2008; Bingel 1994; LactMed 2006; Patil 2017; Rajagopal 2016
Cumin (Cuminum cyminum)	Fruit*	Reputedly oestrogenic; stimulated mammary growth in rats	None known	Bingel 1994; Ergol 2016; Johri 2011; LactMed 2006; Luecha 2013; Rajagopal 2016; Vargova 2018
Dandelion (Taraxacum officinale)	Leaf, root	Unknown; reputed to stimulate mammary tissue; provides essential nutrients	Allergenic for some people; diarrhoea, gastrointestinal upset (rare)	Brodribb 2018; Bingel 1994; Ergol 2016; Goksugur 2014; LactMed 2006; Nice 2015; Scott 2005
Date palm (Phoenix dactylifera or sylvestris)	Fruit	Increased prolactin in rats	None known	Bingel 1994; Ebrahimi 2017; Yashmin 2017
Dill (Anethum graveolens)	Fruit*	Oxytocic‐like activity may improve milk ejection and milk removal; lightly stimulated mammary gland growth in an unpublished rat study; contains linoleic acid and metabolites that are important to milk production	None known	Bingel 1994; Doaa 2016; Ergol 2016; Goksugur 2014; Javan 2017; LactMed 2006; Luecha 2013; Nice 2015; Vargova 2018
Fennel (Foeniculum vulgare)	Fruit*	May stimulate prolactin indirectly via trans‐anethole by decreasing the effect of dopamine on dopamine receptors. Alternatively, oestrogenic properties may stimulate prolactin. May also increase milk production indirectly by assisting the milk ejection reflex. Reputedly stimulates mammary growth	Photo sensitivity, atopic dermatitis, increased gastrointestinal motility. Essential oil may be toxic in large amounts.	Abascal 2008; Abu‐Rabia 2005; Bingel 1994; Bnouham 2010; Bruckner 1993; Ergol 2016; Goksugur 2014; Humphrey 2007; Javan 2017; Low Dog 2009; Mills 2006; Mortel 2013; Nice 2015; Patil 2017; Rajagopal 2016; Romm 2010; Vargova 2018; Zapantis 2012
Fenugreek (Trigonella foenum‐graecum)	Seed	Stimulated growth hormone in ruminants; may stimulate milk production through dopamine receptor antagonism. Phytoestrogens may also stimulate mammary growth. Oxytocic and anxiolytic properties may assist milk ejection reflex for better milk removal.	Digestive upset or loose stools (mother or infant), light headedness, lower blood sugar, maple smell in the urine and sweat; mild allergic reaction. Possible peanut allergen cross sensitivity	Abascal 2008; Bingel 1994; Bruckner 1993; Capasso 2009; Ergol 2016; Goksugur 2014; Humphrey 2007; Low Dog 2009; MacIntosh 2004; Mortel 2013; Nice 2015; Panda 1999; Rajagopal 2016; Romm 2010; Scott 2005; Shawahna 2018; Tabares 2014; Vargova 2018; Yashmin 2017; Zapantis 2012
Garden cress (Lepidium sativum)	Seed	May assist milk ejection reflex. Stimulated prolactin and mammary growth in rats. Provides iron and protein, essential to lactation	None known	Al‐Yawer 2006; Bingel 1994; Bnouham 2010; Patel 2018; Patil 2017; Rajagopal 2016; Shabbir 2018
Ginger (Zingiber officinale)	Rhizome	Unknown	None known	Bingel 1994; Ergol 2016; Luecha 2013; Mills 2006
Goat's rue (Galega officinalis)	Aerial parts	Contains galegin, a precursor to metformin. May exert effects via contents of steroidal saponins. Reputedly stimulates mammary growth	No data for humans. Minor abnormalities in blood and pathological specimens in rats	Abascal 2008; Bingel 1994; Bruckner 1993; Goksugur 2014; Humphrey 2007; MacIntosh 2004; Nice 2015; Rajagopal 2016; Rasekh 2008; Romm 2010; Scott 2005; Tabares 2014; Vargova 2018
Hops (Humulus lupulus)	Strobilus	Oestrogenic, mammary stimulating; relaxing properties may assist the milk ejection reflex to improve milk removal	Could worsen depression; contact allergy	Bingel 1994; Ergol 2016; Goksugur 2014; LactMed 2006; Nice 2015
Ixbut (Euphorbia lancifolia)	Leaf	Euphorbia increased serum prolactin in animal studies	Nausea and vomiting	Bingel 1994; LactMed 2006; Rosengarten 1982
Jivanti (Leptadenia reticulata)	Whole plant	May assist milk ejection reflex	None known	Bingel 1994; Patil 2017
Marshmallow (Malva sylvestris)	Root	Mucilage contains polysacharides that may stimulate prolactin secretion	None known	Brodribb 2018; Bingel 1994; Ergol 2016; Goksugur 2014; Humphrey 2007; Javan 2017; LactMed 2006; Nice 2015; Scott 2005
Milk thistle (Silybum marianum)	Aerial parts, seeds	Appears to stimulate prolactin; possibly oestrogenic	Nausea, flatulence, diarrhoea	Abascal 2008; Abu‐Rabia 2005; Bingel 1994; Capasso 2009; Goksugur 2014; Low Dog 2009; Mills 2006; Mortel 2013; Nice 2015; Zapantis 2012
Moringa (Moringa oleifera) also known as drumstick, kelor leave or malunggay	Leaf	Increases prolactin; provides essential nutrients	None known. Commonly consumed as a vegetable in the Philippines and elsewhere	Bingel 1994; King 2013; LactMed 2006; Nice 2015; Rajagopal 2016
Nettle or Stinging Nettle (Urtica dioica)	Leaf	Unknown. Provides essential nutrients	Itching and dermatitis (contact with fresh herb); gastrointestinal upset; allergenic for some people	Abascal 2008; Bingel 1994; Bnouham 2010; Goksugur 2014; Humphrey 2007; Mills 2006; Nice 2015; Scott 2005; Vargova 2018
Oats (Avena sativa)	Grain	Unknown. Reputed to stimulate mammary gland tissue	None known; commonly consumed food	Abu‐Rabia 2005; Bingel 1994; Ergol 2016; Goksugur 2014; Javan 2017; Monteban 2017; Nice 2015
Papaya (green) (Carica papaya)	Fruit	Stimulated higher prolactin levles and mammary weight in rats	Possible allergen for some people	Bingel 1994; Ergol 2016; Luecha 2013; Tossawanchuntra 2005
Quinoa (Chenopodium quinoa)	Grain	Unknown. Provides essential nutrients	None known; commonly consumed food	Javan 2017; Monteban 2017; Nice 2015
Red Clover (Trifolium pratense)	Flower	May stimulate prolactin via phytoestrogens	Headache, nausea, rash	Bingel 1994; Ergol 2016; Goksugur 2014; Mills 2006; Nice 2015
Red Raspberry (Rubus idaeus)	Leaf	Oxytocic activity may assist the milk ejection reflex and breast emptying. Women who used the extract during pregnancy had a shorter time to lactogenesis.	May have laxative effect	Bingel 1994; Ergol 2016; Goksugur 2014; Kong 2008; Nice 2015
Sesame (Sesamum indicum)	Seed	Stimulated mammary growth in rats.	None known; commonly consumed food	Al‐Bazii 2019; Bingel 1994; Bnouham 2010; Ergol 2016
Shatavari (Asparagus racemosus)	Root	Oestrogenic; may stimulate production by increasing prolactin. Increases weight of mammary gland in animal studies	Runny nose, itchy conjunctivitis, contact dermatitis and cough. May have laxative effect	Bingel 1994; Chaudhury 1983; Dandotiya 2013; Humphrey 2007; Mortel 2013; Rajagopal 2016; Tabares 2014; Vargova 2018; Zapantis 2012
Torbangun (Coleus amboinicus Lour)	Leaf	May stimulate proliferation of secretory mammary cells	Hypoglycaemia and stimulation of the thyroid gland	Bingel 1994; Humphrey 2007; Mortel 2013; Zapantis 2012
Vervain or Verbena (Verbena officinalis)	Aerial parts	Reputed oxytocic and anxiolytic properties may assist the milk ejection reflex and milk removal	Unknown	Bingel 1994; Ergol 2016; Goksugur 2014; Nice 2015

*Apiaceae fruits are commonly referred to as seeds.

Why it is important to do this review

The risks of artificial infant milk are well‐documented (Stuebe 2009; Spatz 2011), and yet frequently minimized to parents. When maternal milk production falls short, infant formula has become the default recommendation for supplementation, despite the fact that the WHO ranks infant formula fourth behind direct breastfeeding, expressed mother's own milk, and donor human milk (WHO 2003). As a result, the potential negative impact on the mother or infant's health, or both, and their well‐being are overlooked (Stuebe 2014b), as are alternatives for increasing maternal milk production. The distress a mother feels when she is unable to breastfeed exclusively is also frequently neglected. This is likely to be experienced as a loss, and hence associated with a grief response (Flaherman 2012; Jacobson 2016; Labbok 2008; Williams 2002). For vulnerable mothers and infants, even small increases in milk could be important.

A recent study surveyed mothers about their experiences of using herbal galactagogues. Some participants expressed frustration that they had not received information from their healthcare providers regarding the possibility of insufficient milk production, or the existence of various galactagogues as a possible treatment option. One strong theme that emerged was the need for better, and more available, evidence‐based information, along with healthcare providers who are sufficiently versed to provide supportive guidance (Sim 2014). Currently, there are several descriptive reviews on the use of galactagogues (Bazzano 2016; Forinash 2012; Mortel 2013; Zapantis 2012; Zuppa 2010). It is important to systematically review the evidence on the potential benefits and harms of these galactagogues in order to provide appropriate recommendations for mothers of term infants with low milk productiondespite optimal feeding practices (Abascal 2008; Sim 2014; Zheng 2019).

This review will not look at the effects of galactagogues to increase milk production for hospitalised preterm infants because of multiple differences between term and preterm mothers, both physiologically and logistically. For instance, in premature births, maternal mammary gland differentiation may be incomplete, possibly reducing early milk production compared to the mother's full‐term potential (Cregan 2007). Preterm infants may not be able to suckle or suckle well, and they are often separated from their mothers, which interferes with crucial time at the breast. The commencement of compensatory milk expression is frequently delayed, and the frequency of pumping or milk expression may suffer if mother must travel back and forth to visit the infant (Henderson 2008; Hopkinson 1988). Therefore, our population of interest is the ‘average woman' who has her healthy infant with her, yet has low milk production. For mothers with hospitalised preterm infants, there is an existing Cochrane Review entitled ‘Medications for increasing milk supply in mothers expressing breast milk for their preterm hospitalised infants’ (Donovan 2012).

Objectives

To assess the effect of oral galactagogues for increasing milk production in non‐hospitalised breastfeeding mother‐term infant pairs.

Methods

Criteria for considering studies for this review

Types of studies

We included only randomised controlled trials (RCTs) (including those using a cluster‐randomised design) and quasi‐RCTs. We also included trials published in abstract form only. We did not include cross‐over trials in this review because the natural increase in milk production during the first few weeks of breastfeeding would confound cross‐overs done during this period.

Types of participants

Non‐hospitalised mother‐term infant pairs either breastfeeding or expressing breast milk, or both, during the first 6 months of life. We excluded studies that specifically looked at mothers with hospitalised preterm infants. We included studies that had both term and preterm infants if most of the infants were term.

The term "non‐hospitalised" refers to healthy mothers and infants even though they may still physically be in the hospital at the time of initiation of intervention.

Types of interventions

We included any oral galactagogues and compared them as follows:

1. Pharmacological oral galactagogues compared with placebo or no treatment.

2. Natural (non‐pharmacological) oral galactagogues compared with placebo or no treatment.

3. Oral galactagogues compared with another oral galactagogue.

We did not report individual oral galactagogues as separate comparisons, but explored them within subgroup analysis.

Types of outcome measures

Primary outcomes

Effect on breast milk production as measured by the following:

1. Proportion of mothers who continued breastfeeding (exclusive or any) for at least 3, 4 and 6 months.

2. Infant weight in trials where the infants received only own mother's milk (g) at latest time measured. We will not include this outcome if infants in the study were given supplemental milk, as the additional milk will invalidate the outcome.

3. Volume of breast milk at the latest time measured (mL).

Secondary outcomes

1. Adverse effects for the infant, such as gastrointestinal disturbances or any other reported effects (see Additional Table 3 and Table 4 for details of potential adverse effects described in the literature).

2. Adverse effects for the mother, such as gastrointestinal disturbances or any other reported effects (see Additional Table 3 and Table 4 for details of potential adverse effects described in the literature).

3. Ability of mother to stop or reduce supplementation with formula milk.

4. Measures of maternal psychological status (e.g. maternal satisfaction, depression scale).

Search methods for identification of studies

The following search methods section of this review is based on a standard template used by Cochrane Pregnancy and Childbirth.

Electronic searches

We searched Cochrane Pregnancy and Childbirth's Trials Register by contacting their Information Specialist (4 November 2019).

The Register is a database containing over 25,000 reports of controlled trials in the field of pregnancy and childbirth. It represents over 30 years of searching. For full current search methods used to populate Pregnancy and Childbirth's Trials Register including the detailed search strategies for CENTRAL, MEDLINE, Embase and CINAHL, the list of handsearched journals and conference proceedings, and the list of journals reviewed via the current awareness service, please follow this link.

Briefly, Cochrane Pregnancy and Childbirth's Trials Register is maintained by their Information Specialist and contains trials identified from:

1. monthly searches of the Cochrane Central Register of Controlled Trials (CENTRAL);

2. weekly searches of MEDLINE (Ovid);

3. weekly searches of Embase (Ovid);

4. monthly searches of CINAHL (EBSCO);

5. handsearches of 30 journals and the proceedings of major conferences;

6. weekly current awareness alerts for a further 44 journals plus monthly BioMed Central email alerts.

Search results are screened by two people and the full text of all relevant trial reports identified through the searching activities described above is reviewed. Based on the intervention described, each trial report is assigned a number that corresponds to a specific Pregnancy and Childbirth review topic (or topics), and is then added to the Register. The Information Specialist searches the Register for each review using this topic number rather than keywords. This results in a more specific search set that has been fully accounted for in the relevant review sections (Included studies; Excluded studies; Studies awaiting classification; Ongoing studies).

In addition, we searched ClinicalTrials.gov and the WHO International Clinical Trials Registry Platform (ICTRP) for unpublished, planned and ongoing trial reports (4 November 2019) (see Appendix 1 for search terms used).

We also searched Health Research and Developmental Network ‐ Philippines (HERDIN) and Natural Products Alert (Napralert) (last searched on 4 November 2019) without any date or language restrictions. The search terms used here are shown in Appendix 1.

Searching other resources

We searched the reference lists of retrieved studies for other potential studies as well as a personal collection of studies belonging to an author (LM). We also contacted experts in the field and used journals and voluntary organizations related to breastfeeding to seek additional published and unpublished studies.

We did not apply any language, geographic or date restrictions.

Data collection and analysis

The following methods section of this review is based on a standard template used by Cochrane Pregnancy and Childbirth.

Selection of studies

FSC and TML independently assessed all potential studies identified from all our electronic searches for inclusion to the review. We resolved any disagreement through discussion and consultation with a third review author (LM). FSC and LM independently assessed all potential studies identified from our other resources. Any disagreement was resolved by consulting a third review author (TML).

Data extraction and management

Four review authors (FSC, FWC, LM and OJH) extracted the data from the included studies. We resolved discrepancies through discussion, and on several occasions consulted other review authors (TML, JJH). We entered data into Review Manager 5 software and checked for accuracy (Review Manager 2014).

When information regarding any of the studies was unclear, we contacted the study authors for further details.

Assessment of risk of bias in included studies

Three review authors (FSC, FWC and LM) independently assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We resolved any disagreement by discussion and by consulting other review authors (TML, JJH).

(1) Random sequence generation (checking for possible selection bias)

For each included study we described the method used to generate the allocation sequence in sufficient detail to allow an assessment of whether it should produce comparable groups.

We assessed the method as:

1. low risk of bias (any truly random process, e.g. random number table; computer random number generator);

2. high risk of bias (any non‐random process, e.g. odd or even date of birth; hospital or clinic record number); or

3. unclear risk of bias.   

(2) Allocation concealment (checking for possible selection bias)

For each included study we described the method used to conceal allocation to interventions prior to assignment and assessed whether intervention allocation could have been foreseen in advance of, or during recruitment, or changed after assignment.

We assessed the methods as:

1. low risk of bias (e.g. telephone or central randomisation; consecutively numbered sealed opaque envelopes);

2. high risk of bias (open random allocation; unsealed or non‐opaque envelopes, alternation; date of birth);

3. unclear risk of bias.   

(3.1) Blinding of participants and personnel (checking for possible performance bias)

For each included study we described the methods used, if any, to blind study participants and personnel from knowledge of which intervention a participant received. We considered that studies are at low risk of bias if they were blinded, or if we judged that the lack of blinding would be unlikely to affect results. We assessed blinding separately for different outcomes or classes of outcomes.

We assessed the methods as:

• low, high or unclear risk of bias for participants;

• low, high or unclear risk of bias for personnel.

(3.2) Blinding of outcome assessment (checking for possible detection bias)

For each included study we described the methods used, if any, to blind outcome assessors from knowledge of which intervention a participant received. We assessed blinding separately for different outcomes or classes of outcomes.

We assessed methods used to blind outcome assessment as:

1. low, high or unclear risk of bias for milk volume outcomes;

2. low, high or unclear risk of bias for self reported outcomes;

3. low, high or unclear risk of bias for infant weight outcomes.

(4) Incomplete outcome data (checking for possible attrition bias due to the amount, nature and handling of incomplete outcome data)

For each included study, and for each outcome or class of outcomes, we described the completeness of data including attrition and exclusions from the analysis. We stated whether attrition and exclusions were reported and the numbers included in the analysis at each stage (compared with the total randomised participants), reasons for attrition or exclusion where reported, and whether missing data were balanced across groups or were related to outcomes. Where sufficient information was reported, or was supplied by the trial authors, we reincluded missing data in the analyses that we undertook.

We assessed methods as being at:

1. low risk of bias (e.g. no missing outcome data; missing outcome data balanced across groups);

2. high risk of bias (e.g. numbers or reasons for missing data imbalanced across groups; 'as treated' analysis done with substantial departure of intervention received from that assigned at randomisation);

3. unclear risk of bias.

(5) Selective reporting (checking for reporting bias)

For each included study we described how we explored the possibility of selective outcome reporting bias and what we found.

We assessed the methods as being at:

1. low risk of bias (where it is clear that all of the study's prespecified outcomes and all expected outcomes of interest to the review have been reported);

2. high risk of bias (where not all the study's prespecified outcomes have been reported; one or more reported primary outcomes were not prespecified; outcomes of interest are reported incompletely and so cannot be used; study fails to include results of a key outcome that would have been expected to have been reported);

3. unclear risk of bias.

(6) Other bias (checking for bias due to problems not covered by (1) to (5) above)

For each included study we described any important concerns we have about other possible sources of bias.

We assess whether each study was free of other problems (such as baseline imbalance between the two groups) that could put it at risk of bias. We assess the studies as:

1. low risk of other bias;

2. high risk of other bias;

3. unclear whether there is risk of other bias.

(7) Overall risk of bias

We made explicit judgements about whether studies are at high risk of bias, according to the criteria given in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). With reference to (1) to (6) above, we assessed the likely magnitude and direction of the bias and whether we consider it as likely to impact on the findings. We explored the impact of the level of bias through undertaking Sensitivity analysis.

Measures of treatment effect

Dichotomous data

For dichotomous data, we presented the results as risk ratios (RRs) with 95% confidence intervals (CIs). 

Continuous data

For continuous data, we used mean difference (MD) with 95% CIs. We used change scores together with postintervention scores where needed. If we had encountered outcomes with different scales, we would have considered using standardised mean differences (SMDs).

Unit of analysis issues

Cluster‐randomised trials

There were no cluster‐randomised trials included in this review.

If these trials were included, we would have analysed them along with individually randomised trials. We would have adjusted for cluster size using the methods described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), using an estimate of the intracluster correlation coefficient (ICC) derived from the trial (if possible), from a similar trial or from a study of a similar population. If we had used ICCs from other sources, we would have reported this and conducted sensitivity analyses to investigate the effect of variation in the ICC. If we had identified both cluster‐randomised trials and individually‐randomised trials, we would have planned to synthesise the relevant information. We would have considered it reasonable to combine the results from both if there was little heterogeneity between the study designs, and the interaction between the effect of intervention and the choice of randomization unit was considered to be unlikely. We would have also acknowledged heterogeneity in the randomization unit and performed a subgroup analysis to investigate the effects of the randomization unit.

Other unit of analysis issues

We did not encounter any studies that specifically randomised twins or multiples. In the one study which included twins (Xu 2000), we considered the mother as a single unit of analysis, as all outcomes were not related to the infants.

For studies using one or more treatment groups (multi‐arm studies), where appropriate, we combined groups to create a single pair‐wise comparison using the methods described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). When a multi‐arm study contributed more than one comparison to a particular meta‐analysis, we combined treatment groups or divided the control group, so that the inclusion of data from the same woman more than once in the same analysis would be avoided. In studies where the intervention for the third arm was also given to the first and second arms (example: water), we excluded the third arm and treated the studies as a two‐armed study.

Dealing with missing data

For each included study, we noted the levels of attrition. We explored the impact of including studies with high levels of missing data in the overall assessment of treatment effect by using sensitivity analysis.

For all outcomes, we carried out analyses, as far as possible, on an intention‐to‐treat (ITT) basis, that is, we included all participants randomised to each group in the analyses, and all participants were analysed in the group to which they were allocated, regardless of whether or not they received the allocated intervention. The denominator for each outcome in each trial was the number randomised minus any participants whose outcomes were known to be missing.

Missing summary statistics, such as standard deviations (SD), were estimated by conversion of available statistics, such as standard errors (SEs).

Assessment of heterogeneity

We assessed statistical heterogeneity in each meta‐analysis using the T², I² and Chi² statistics. We regarded heterogeneity as substantial if I² was greater than 30% and either T² was greater than zero, or there was a low P value (less than 0.10) in the Chi² test for heterogeneity. 

Assessment of reporting biases

In future updates, if there are 10 or more studies in the meta‐analysis, we will investigate reporting biases (such as publication bias) using funnel plots. We will assess funnel plot asymmetry visually. If asymmetry is suggested by a visual assessment, we will perform exploratory analyses to investigate it.

Data synthesis

We carried out statistical analysis using the Review Manager 5 software (Review Manager 2014). Most of our included studies were not clinically similar enough and were not considered appropriate to meta‐analyse given the differences in participants, interventions and outcome measurements. Where it was reasonable to assume that studies were estimating the same underlying treatment effect, we performed meta‐analyses. We stratified galactagogues into separate comparisons (e.g. pharmacological and natural) and explored specific individual galactagogues by subgroup analysis. Due to the variety of galactagogues used and the differences in the participants, we chose to employ a random‐effects meta‐analysis.

Subgroup analysis and investigation of heterogeneity

We identified substantial clinical heterogeneity in our included studies. We had planned to investigate subgroup differences by interaction tests available within Review Manager (Review Manager 2014), and report the results by quoting the Chi² statistic and P value, and the interaction test I² value.

1. Mothers who only express milk versus mothers who directly breastfeed the infant.

2. Diabetic mothers (of any form) versus non‐diabetic mothers.

3. Mothers breastfeeding twins or multiples.

In the process of conducting this review, (not as a result of examining the data), we added the following three additional subgroups.

1. Age of the infant when the outcome was measured: less than 2 weeks old and more than 2 weeks old.

2. Mothers with lactation insufficiency, as defined by study authors.

3. Specific individual galactagogues within each comparison.

Sensitivity analysis

We conducted sensitivity analyses by study quality, where necessary, to test the impact of the following biases: selection, performance and attrition biases as defined in the 'Risk of bias' tool to explore the effect of including studies with a high or unclear risk of bias. This was done only with our primary outcomes. Only one study had a pair of twins included (Xu 2000), thus it was not necessary to conduct a sensitivity analysis for twin or multiples.

Assessment of the certainty of the evidence using the GRADE approach and 'Summary of findings' table

We assessed the certainty of our evidence using the GRADE approach, as outlined in the GRADE Handbook (Schünemann 2013). For each outcome, we assessed certainty based on five factors ‐ study limitations, consistency of effect, imprecision, indirectness and publication bias. We used GRADEPro GDT software to create a 'Summary of findings' table (GRADEpro GDT 2015). The 'Summary of findings' table had the following components: summarized key findings (participants, comparison and baseline information, outcome); summarized statistical results; summarized certainty of evidence; summarized magnitude of the effect, including the source of any external information used in the 'Assumed risk' column or any departures from the standard methods; and reasons behind the decisions.

We included the following outcomes in our 'Summary of findings' table.

1. Proportion of mothers who continued breastfeeding (exclusive or any) at 3, 4 and 6 months.

2. Infant weight at the end of the study (g) (in trials where the infants were only receiving their own mother's milk).

3. Volume of breast milk at the latest time measured (mL).

4. Adverse effects.

Results

Description of studies

Results of the search

Our search retrieved a total of 239 records (97 from PCG, ICTRP and ClinicalTrials.gov, 70 from Herdin, zero from Napralert and 72 from other sources), see Figure 1.

1.

Study flow diagram.

We excluded 75 by title and assessed 162 full texts and abstracts (of 146 studies) for eligibility.

We needed clarification or additional data for 36 studies but were only able to obtain current contact information for 27 of them. Of these, 15 authors replied to our queries (NCT00264719; NCT00477776; Damanik 2006; Di Pierro 2008; Ghasemi 2018; Manjula 2014; NCT02190448; Nordin 2019; Paritakul 2016; Sakha 2008; Sharma 1996; Sy 2012; Tirak 2008; Wagner 2019), but we did not get a response from the remaining 12 (Balahibo 2002; Briton‐Medrano 2002; Bumrungpert 2018; Gupta 2011; Khairani 2017; Inam 2013; Mathew 2018; CTRI/2016/01/006547; Petraglia 1985; Turkyilmaz 2011; Ylikorkala 1982; Yulinda 2017). We could not reach 10 authors (Aono 1982; Barguno 1988; NCT02233439; De Gezelle 1983; De Leo 1986; Espinosa‐Kuo 2005; Kauppila 1985; Mukherjee 1987; Ushiroyama 2007; Yabes‐Almirante 1996a), despite writing to their co‐authors, universities or workplaces.

We included 41 studies (51 records) in the review and excluded 93 studies (98 records).

We also identified 12 ongoing studies (13 records). See Characteristics of ongoing studies. An overview of the types of galactagogues used in these studies is also found in Table 5.

3. Overview of ongoing studies by galactagogue.

Pharmacological interventions
Intervention	Galactagogue form	Study ID	No. of estimated participants	Infant's age at point of enrolment	Mother with lactation deficiency	Dose	Duration of intervention
Domperidone	Tablet	TCTR20170811003	120	At birth	Not mentioned	1 tablet three times a day	Not mentioned
Metoclopramide	Tablet	NCT00264719	160	At birth	No	10 mg three times a day then 10 mg once a day	7 days then 2 days
Metoclopramide	Tablet	NCT00477776	160	At birth	No	10 mg three times a day then 10 mg once a day	7 days then 2 days
Natural interventions
Intervention	Galactagogue form	Study ID	No. of estimated participants	Infant's age at point of enrolment	Mother with lactation deficiency	Dose	Duration of intervention
Avuyedic medicine containing Jivanti, Kamboji, Vidarikand, Shatavari, Amalaki, Methi (fenugreek), Godanti bhasma and Suwa (Promolact)	Granules	CTRI/2016/01/006547	80	At birth	Yes	2 spoonfuls twice a day	10 days
Ayurved Siriraj Prasa‐Nam‐Nom Recipe	Capsule	TCTR20190218004	54	At birth	Yes	500 mg (frequency not mentioned)	Not mentioned
Chicken extract	Liquid extract	JPRN‐UMIN000027159	80	At birth	Not mentioned	70 mL twice a day	At least 2 weeks
Combination of Silybum marianum and Galega officinalis	Tea	NCT02233439	210	At birth	Yes	Once a day	6 weeks
Lysiphyllum Strychifolium leaves	Tea	TCTR20190716001	84	At birth	Not mentioned	At least 3 cups (1 cup is approximately 200 mL) daily	7 days
Mixed herbal galactagogue tea (Mansel Still Tee)	Tea	NCT02740751	90	At birth	No	200 mL three times a day	4 weeks
Mixed herbal galactagogue tea (Mother's Milk Tea)	Tea	NCT02190448	60	2 weeks old	No	3 to 5 cups daily	4 weeks
Prasaplai (Thai herbal drug containing 50% of this mix: kaffir lime, sweet flag, garlic, Eleutherine americana, black pepper, long pepper, Zingiber officinale Roscoe, white turmeric, black seed, sodium chloride, camphor, and 50% rhizome powder of Zingiber cassumunar Roxb)	Capsule	TCTR20180808007	60	At birth	No	Not mentioned	Not mentioned
Wang Nam Yen Herbs	Tea	TCTR20170811003	120	At birth	Not mentioned	200 cc three times a day	Not mentioned
Yeast powder	Capsule	ACTRN12619000704190	40	At birth	No	9 capsules daily (1 capsule = 5 g)	28 days

cc: cubic centimetre

mg: milligram

mL: milliliter

Included studies

We included 41 studies in the review (see Characteristics of included studies). An overview of the main clinical characteristics of the included studies by type of galactagogue is found in Table 6.

4. Overview of included studies by galactagogue.

Pharmacological galactagogue versus placebo or no intervention
Intervention	Galactagogue form	Study ID	No. of participants	Infant's age at point of enrolment	Mother with lactation deficiency	Dose	Duration of intervention
Domperidone	Tablet	Inam 2013	100	0 to 7 days	Yes	10 mg three times a day	7 days
Domperidone	Tablet	Jantarasaengaram 2012	44	Newborn	Not reported	10 mg four times a day	4 days
Metoclopramide	Tablet	De Gezelle 1983	13	0 to 8 days	Not reported	10 mg three times a day	8 days
Metoclopramide	Tablet	Kauppila 1985	33	4 to 20 weeks	Yes	10 mg three times a day	3 weeks
Metoclopramide	Tablet	Sakha 2008	20	A few months old	Yes	10 mg three times a day	15 days
Sulpiride	Tablet	Aono 1982	96	3 to 9 days	Yes	50 mg twice a day	4 days
Sulpiride	Tablet	Barguno 1988	66	1 to 90 days	Not reported	100 mg three times a day then 50 mg three times a day	4 days then 86 days
Sulpiride	Tablet	Ylikorkala 1982	28	0 to 4 months	Yes	50 mg three times a day	4 weeks
Thyrotropin‐releasing hormone	Capsule	Zarate 1976	16 (first part of study) 9 ('conjoint study')	2 days (first part of study) 2 weeks ('conjoint study')	No	20 mg three times a day	4 weeks (first part of study) 1 week ('conjoint study')
Natural galactagogue versus placebo or no intervention
Intervention	Galactagogue form	Study ID	No. of participants	Infant's age at point of enrolment	Mother with lactation deficiency	Dose	Duration of intervention
Mixed galactagogue with Shatavari (Asparagus racemosus) as main ingredient	Powder	Sharma 1996	64	14 to 90 days	Yes	2 teaspoonsful twice a day	4 weeks
Cui Ru (催乳汤)	Soup	Su 2008	108	7 to 13 days	Yes	No specified amount but was given twice a day	7 days
Banana flower flour	Biscuits	Nordin 2019	58	2 to 6 months	Not reported	3.24g (2 pieces of biscuits) daily	3 weeks
Fennel (Foeniculum vulgare)	Powder in tea	Ghasemi 2018	39	0 to 4 months	Not reported	7.5 g three times a day	4 weeks
Fenugreek (Trigonella foenum‐graecum L)	Seed in tea	Ghasemi 2018	39	0 to 4 months	Not reported	7.5 g three times a day	4 weeks
Galactagogue foods	Food	Thaweekul 2014	233	Newborn	Not reported	"2500 kcal diet with 70 g protein per day"	Not specified
Galactagogue herbal medicine with fenugreek, ginger and turmeric	Capsule	Bumrungpert 2018	50	1 month	Not reported	200 mg fenugreek seed, 120 mg ginger, and 100 mg turmeric per capsule three times a day	4 weeks
Ginger (Zingiber officinale)	Capsules	Paritakul 2016	68	Newborn	Not reported	500 mg twice a day	7 days
Humana Still Tee	Tea	Tirak 2008	78	Newborn	Not reported	9 g three times a day	1 month
Humana Still Tee	Tea	Turkyilmaz 2011	66	Newborn	No	600 mL daily	Not specified
Ixbut (Euphorbia lancifolia)	Infusion	Chan 2005	34	30 to 90 days	Yes	20 leaves daily	3 days
Levant cotton (Gossypium herbaceum Linn) kernels	Capsules	Manjula 2014	48	10 to 180 days	Yes	10 g daily	1 months
Moringa leaves	Capsules	Balahibo 2002	60	Newborn	Not reported	250 mg daily or bd, 500 mg daily or twice a day	8 weeks
Moringa leaves	Capsules	Briton‐Medrano 2002	53	Before birth	Not reported	700 mg three times a day	From 35 weeks' gestation till delivery of infant
Moringa leaves	Capsules	Espinosa‐Kuo 2005	82	3 days	Not reported	700 mg daily	6 days
Moringa leaves	Capsules	Yabes‐Almirante 1996a	116	Newborn	Not reported	250 mg twice a day	4 months
Mother's Milk Tea (MMT)	Tea	Wagner 2019	60	2 to 12 weeks	Not reported	240 mL 3 to 5 times a day	Unclear (possible for 4 weeks)
Palm dates	Flesh	Sakka 2014	75	Newborn	Not reported	100 g three times a day	Not stated
Palm dates	Extracts	Yulinda 2017	Not mentioned	Not mentioned	Not reported	Dose and frequency not mentioned	3 days
Pork knuckle soup	Soup	Xu 2000	82	8 to 10 days	Not reported	300 mL 2 to 3 hourly	Not stated
Shatavari (Asparagus racemosus)	Capsules	Gupta 2011	60	Average 2.8 months	Yes	60 mg/kg/day	30 days
Sheng Ru He Ji (生乳合剂)	Oral liquid	Yin 2005	200	Newborn	Not reported	100 mL twice a day	3 days
Shirafza; combination alcohol extraction of fennel (Foeniculum vulgare), anise (Pimpinella anisum), green cumin (Cuminum cyminum), dill (Anethum gravolens), parsley (Petroselinum crispum), black seed (Nigella sativa)	Drops	Shariati 2004	158	0 to 6 months	Yes	30 drops three times a day	4 weeks
Silymarin (Silybum marianum)	Sachet	Di Pierro 2008	50	0 to 63 days	Yes	420 mg daily	63 days
Xian Tong Ru (先通乳)	Soup	Huang 2000	85	At birth	Not reported	50 mL twice a day	3 days
Pharmacological galactagogues versus natural galactagogues
Intervention	Galactagogue form	Study ID	No. of participants	Infant's age at point of enrolment	Mother with lactation deficiency	Dose	Duration of intervention
Domperidone versus moringa leaves	Tablet versus capsule	Sy 2012	26	2 weeks to 6 months	No	Domperidone 10 mg three times a day, moringa leaves 250 mg twice a day	7 days
Natural galactagogues versus natural galactagogues
Intervention	Galactagogue form	Study ID	No. of participants	Infant's age at point of enrolment	Mother with lactation deficiency	Dose	Duration of intervention
Chan Bao (产宝) versus Bu Xue Sheng Ru (补血生乳)	Oral liquid versus capsule	Jiang 2006	60	2 days	Not reported	ChanBao 10 mg twice a day, Bu Xue Sheng Ru 4 mg twice a day	2 days
Fennel (Foeniculum vulgare) versus Fenugreek (Trigonella foenum‐graecum L)	Seeds versus powder in tea	Ghasemi 2018	78	0 to 4 months	Not reported	7.5 g three times a day	4 weeks
Fennel (Foeniculum vulgare) versus Fenugreek (Trigonella foenum‐graecum L)	Seeds in tea	Mathew 2018	30	10 days to 3 months	Not reported	14 grams in 2 litres of water. 300 mL to be taken everyday	7 days
Mu Er Wu You (母儿无忧汤) versus Kun Yuan Tong Ru soup (坤元通乳口服液)	Soup versus soup	Li 2010	90	Newborn	Yes	Unclear	4 days
Ru Quan Chong Ji (乳泉冲剂) versus Shengruzhi (生乳汁)	Soup versus soup	Fang 2003	120	Not reported	Yes	15 g twice a day	3 days
Torbangun (Coleus amboinicus L) versus Fenugreek (Trigonella foenum‐graecum L) versus Molocco (placental extract)	Soup versus capsule versus tablet	Damanik 2006	75	2 days	Not reported	150 g daily	30 days

g: gram

kg: kilogram

mg: milligram

mL: milliliter

Design

Of the 41 included studies, two were four‐arm studies (Balahibo 2002; Khairani 2017), six were three‐arm studies (Damanik 2006; Ghasemi 2018; Jiang 2006; Sakka 2014; Tirak 2008; Turkyilmaz 2011), and the remaining were two‐arm parallel studies. There were no cluster‐randomised studies.

Sample sizes

Overall there was a total of 3005 included mothers and 3006 infants: one study included one set of twins. One study did not report the number of included mothers (Yulinda 2017). The sample size of each study ranged from nine to 233 participants.

Setting

Included studies were conducted in very diverse geographical and cultural settings, which played a large role in the heterogeneity of the interventions, especially those that were non‐pharmacological. Geographically, they were spread throughout Asia, Europe and Latin America.

• Eight were from East Asia (Aono 1982; Fang 2003; Huang 2000; Jiang 2006; Li 2010; Su 2008; Xu 2000; Yin 2005).

• Thirteen were from South‐East Asia (Balahibo 2002; Briton‐Medrano 2002; Bumrungpert 2018; Damanik 2006; Espinosa‐Kuo 2005; Jantarasaengaram 2012; Khairani 2017; Sy 2012; Nordin 2019; Paritakul 2016; Thaweekul 2014; Yabes‐Almirante 1996a; Yulinda 2017).

• Six were from South Asia (Gupta 2011; Inam 2013; Manjula 2014; Mathew 2018; Mukherjee 1987; Sharma 1996).

• Six were from Central Asia (Ghasemi 2018; Sakha 2008; Sakka 2014; Shariati 2004; Tirak 2008; Turkyilmaz 2011).

• Three were from Europe (De Gezelle 1983; Kauppila 1985; Ylikorkala 1982).

• Three were from Latin America (Chan 2005; Di Pierro 2008; Zarate 1976).

• One was from North America (Wagner 2019).

• One did not specify the study location (Barguno 1988).

The studies were conducted in a range of economic situations (World Bank 2016), as follows.

• Sixteen were conducted in low‐middle‐income countries (Balahibo 2002; Briton‐Medrano 2002; Chan 2005; Damanik 2006; Espinosa‐Kuo 2005; Gupta 2011; Inam 2013; Khairani 2017; Manjula 2014; Mathew 2018; Mukherjee 1987; Sakka 2014; Sharma 1996; Sy 2012; Yabes‐Almirante 1996a; Yulinda 2017).

• Nineteen were conducted in upper‐middle‐income countries (Bumrungpert 2018; Di Pierro 2008; Fang 2003; Ghasemi 2018; Huang 2000; Jantarasaengaram 2012; Jiang 2006; Li 2010; Nordin 2019; Paritakul 2016; Sakha 2008; Shariati 2004; Su 2008; Thaweekul 2014; Tirak 2008; Turkyilmaz 2011; Xu 2000; Yin 2005; Zarate 1976).

• Five were conducted in high‐income countries (Aono 1982; De Gezelle 1983; Kauppila 1985; Wagner 2019; Ylikorkala 1982).

The studies were conducted either in the hospital during the immediate postpartum period or in the community.

Trial authors' declarations of interest

Sixteen authors made declarations of interest statements in their reports (Bumrungpert 2018; Ghasemi 2018; Jantarasaengaram 2012; Jiang 2006; Kauppila 1985; Manjula 2014; Mathew 2018; Paritakul 2016; Sakha 2008; Sharma 1996; Thaweekul 2014; Tirak 2008; Wagner 2019; Yabes‐Almirante 1996a; Ylikorkala 1982; Zarate 1976), but this information was absent from the remaining 25 studies' reports.

Sources of funding

Ten studies received university or government funding (Aono 1982; Damanik 2006; Ghasemi 2018; Jantarasaengaram 2012; Jiang 2006; Manjula 2014; Nordin 2019; Paritakul 2016; Sakha 2008; Thaweekul 2014).

Nine studies received some form of commercial funding, such as the provision of the intervention or outcome assessment tools (Bumrungpert 2018; Di Pierro 2008; Kauppila 1985; Sharma 1996; Tirak 2008; Wagner 2019; Yabes‐Almirante 1996a; Ylikorkala 1982; Zarate 1976).

One reported self‐funding (Mathew 2018), while the remaining 21 studies did not report their source of funding.

Trial dates

Of the 41 included studies, only 23 reported their trial dates (which ranged from 1996 to 2016) (Balahibo 2002; Briton‐Medrano 2002; Chan 2005; Espinosa‐Kuo 2005; Fang 2003; Huang 2000; Inam 2013; Jantarasaengaram 2012; Jiang 2006; Li 2010; Manjula 2014; Nordin 2019; Paritakul 2016; Sakka 2014; Su 2008; Thaweekul 2014; Tirak 2008; Turkyilmaz 2011; Wagner 2019; Xu 2000; Yabes‐Almirante 1996a; Yin 2005; Yulinda 2017).

The remaining studies did not report their trial dates.

Participants

There was considerable variation in the demographic and clinical characteristics of mother‐infant pairs in the included studies.

Mothers' characteristics

Eighteen studies specifically enrolled mothers who had lactation deficiency (Aono 1982; Chan 2005; Di Pierro 2008; Fang 2003; Gupta 2011; Inam 2013; Jiang 2006; Khairani 2017; Kauppila 1985; Li 2010; Manjula 2014; Mukherjee 1987; Sakha 2008; Shariati 2004; Sharma 1996; Su 2008; Ylikorkala 1982; Zarate 1976). Five studies included only primiparous mothers (Barguno 1988; De Gezelle 1983; Khairani 2017; Sakha 2008; Yin 2005). Two studies only included postcaesarean mothers (Jantarasaengaram 2012; Xu 2000). One study only included working mothers who were away from their infants eight hours a day (Nordin 2019). One study enrolled participants prior to delivery of their infants (Briton‐Medrano 2002). None of the studies included mothers with gestational diabetes.

Infants' characteristics

Twenty‐one studies enrolled mothers whose infants were less than 2 weeks old (Aono 1982; Balahibo 2002; Barguno 1988; Briton‐Medrano 2002; Damanik 2006; De Gezelle 1983; Espinosa‐Kuo 2005; Huang 2000; Inam 2013; Jantarasaengaram 2012; Jiang 2006; Li 2010; Paritakul 2016; Sakka 2014; Su 2008; Thaweekul 2014; Tirak 2008; Turkyilmaz 2011; Xu 2000; Yabes‐Almirante 1996a; Yin 2005); sixteen enrolled mothers whose infants' ages ranged from newborn to 6 months of age (Bumrungpert 2018; Chan 2005; Di Pierro 2008; Ghasemi 2018; Gupta 2011; Kauppila 1985; Manjula 2014; Mathew 2018; Nordin 2019; Sakha 2008; Shariati 2004; Sharma 1996; Sy 2012; Wagner 2019; Ylikorkala 1982; Zarate 1976). Four studies did not report the age of the infants at enrolment (Fang 2003; Khairani 2017; Mukherjee 1987; Yulinda 2017). One study only included female infants due to hypothetical concerns that the herb might affect male fertility (Ghasemi 2018). One study included a set of twins (Xu 2000).

Interventions and comparison

The included studies tested 33 different interventions. The types of interventions used were as follows:

Pharmacological galactagogues

• Domperidone (Inam 2013; Jantarasaengaram 2012), taken as tablets.

• Metoclopramide (De Gezelle 1983; Kauppila 1985; Sakha 2008), taken as tablets.

• Sulpiride (Aono 1982; Barguno 1988; Ylikorkala 1982), taken as tablets.

• Thyrotropin‐releasing hormone (Zarate 1976), taken as capsules.

Natural galactagogues

• Herbal remedies or culinary preparations: banana flower (Nordin 2019); Bu Xue Sheng Ru (补血生乳) (Jiang 2006); Chanbao Oral Liquid (产宝) (Jiang 2006); Cui Ru (催乳汤) (Su 2008); fennel (Foeniculum vulgare) (Ghasemi 2018); fenugreek (Trigonella foenum‐graecum L) (Ghasemi 2018; Sakka 2014); galactagogue foods (Thaweekul 2014); ginger (Zingiber officinale) (Paritakul 2016); mixed botanical teas (Humana Still Tee) (Tirak 2008; Turkyilmaz 2011)/(Mother's Milk Tea) (Wagner 2019); ixbut (Euphorbia lancifolia) (Chan 2005); Lactare (Mukherjee 1987); levant cotton (Gossypium herbaceum Linn) kernels (Manjula 2014); moringa leaves (Balahibo 2002; Briton‐Medrano 2002; Espinosa‐Kuo 2005; Khairani 2017; Yabes‐Almirante 1996a); mixed fenugreek, ginger and tumeric capsules (Bumrungpert 2018); mixed galactagogue with Shatavari (Asparagus racemosus) as main ingredient (Sharma 1996); palm dates (Sakka 2014; Yulinda 2017); pork knuckle soup (Xu 2000); shatavari (Asparagus racemosus) (Gupta 2011); Sheng Ru He Ji (生乳合剂) (Yin 2005); Shirafza: combination alcohol extraction of fennel (Foeniculum vulgare), anise (Pimpinella anisum), green cumin (Cuminum cyminum), dill (Anethum gravolens), parsley (Petroselinum crispum), black seed (Nigella sativa) (Shariati 2004); silymarin (Silybum marianum) (Di Pierro 2008); torbagun leaves (Damanik 2006); and Xian Tong Ru (先通乳) (Huang 2000). See Appendix 2 for the ingredients of teas, soups and galactagenic foods.

All comparisons were with placebo, another galactagogue or no intervention. Eight studies compared pharmacological galactagogues against placebo or no treatment (Aono 1982; Barguno 1988; De Gezelle 1983; Jantarasaengaram 2012; Kauppila 1985; Sakha 2008; Ylikorkala 1982; Zarate 1976). Twenty‐seven studies compared natural galactagogues against placebo or no treatment (Balahibo 2002; Briton‐Medrano 2002; Bumrungpert 2018; Chan 2005; Di Pierro 2008; Espinosa‐Kuo 2005; Ghasemi 2018; Gupta 2011; Huang 2000; Jiang 2006; Khairani 2017; Manjula 2014; Mukherjee 1987; Nordin 2019; Paritakul 2016; Sakka 2014; Shariati 2004; Sharma 1996; Su 2008; Thaweekul 2014; Tirak 2008; Turkyilmaz 2011; Wagner 2019; Xu 2000; Yabes‐Almirante 1996a; Yin 2005; Yulinda 2017), and eight compared one galactagogue against another galactagogue (Damanik 2006; Fang 2003; Ghasemi 2018; Jiang 2006; Li 2010; Mathew 2018; Sakka 2014; Sy 2012).

With regards to studies that used a placebo, six did not describe the placebo. Of those that did (26 studies), many tried to make the placebo into a form that resembled the intervention used, e.g. capsules, tablets, coloured water, soups, teas or food (e.g. biscuits).

Duration of intervention: the duration of the interventions ranged from 3 days to 4 months and started after delivery of the infant with the exception of Briton‐Medrano 2002, where the intervention was started after 35 weeks of pregnancy and terminated once the infant was born.

Breastfeeding routines

Breastfeeding routines were not consistent across the studies.

Three studies were conducted prior to the Baby Friendly Hospital Initiative, and all three prescribed a regimental breastfeeding schedule of "every 3 hours" for a total of six feedings per day (Aono 1982; Barguno 1988; De Gezelle 1983).

Fourteen studies reported on‐demand feeding (Damanik 2006; Ghasemi 2018; Huang 2000; Nordin 2019; Paritakul 2016; Sakha 2008; Sakka 2014; Shariati 2004; Sharma 1996; Tirak 2008; Turkyilmaz 2011; Wagner 2019; Yabes‐Almirante 1996a; Ylikorkala 1982).

In one study (Briton‐Medrano 2002), the infants were not breastfed directly but fed with expressed milk from their mothers.

The remaining studies did not describe how the mothers breastfed their infants.

Outcomes

Proportion of mothers who continued breastfeeding (exclusive or any) at 3, 4 and 6 months

One study reported the proportion of mothers still breastfeeding at 3 and 6 months (Wagner 2019). Two other studies reported the number of infants breastfeeding exclusively at one month (Aono 1982; Manjula 2014).

Infant weight in trials where the infants received only own mother's milk (g)

Nine studies in which the infants only received their own mother's milk reported infant weight outcomes using a variety of ways. Three reported mean percentage weight gain: Balahibo 2002 reported it every 2 weeks for 8 weeks, Gupta 2011 reported it after 1 month, and Sakka 2014 reported it after 3, 7 and 14 days. Three studies reported infant weight gain over varying periods of time: Shariati 2004 reported weight gain per week for 4 weeks, Sakha 2008 reported weight gain per 15 days, and Barguno 1988 reported weight gain per every 11 to 30 days. Three studies compared pre‐ and postintervention infant weights (Ghasemi 2018; Tirak 2008; Yabes‐Almirante 1996a). We did not include the results from four studies because the infants received additional infant formula (Li 2010; Manjula 2014; Mathew 2018; Sharma 1996).

Volume of breast milk at the latest time measured (mL)

Twenty‐eight studies reported outcomes related to milk volume. Twelve reported milk volume per day (Aono 1982; Bumrungpert 2018; Damanik 2006; Di Pierro 2008; Espinosa‐Kuo 2005; Huang 2000; Kauppila 1985; Jiang 2006; Li 2010; Nordin 2019; Paritakul 2016; Ylikorkala 1982), six reported milk volume per expression (Briton‐Medrano 2002; De Gezelle 1983; Sakka 2014; Su 2008; Sy 2012; Turkyilmaz 2011), two reported volume per two expressions (Jantarasaengaram 2012; Yin 2005), and one reported total volume per expression over three consecutive days (Chan 2005). Four studies reported milk volume as categorical data: Inam 2013 reported the number of mothers having milk volume equal to or greater than 50 mL (defined as 'efficacious' in the study); Khairani 2017 and Mukherjee 1987 reported milk volume as 'good,' 'moderate or sufficient' and 'less or poor;' Xu 2000 reported the number of mothers with different milk volume categories (0 to 100 mL, 101 to 250 mL, 251 to 400 mL, more than 400 mL). One study reported milk volume as part of an 'overall symptom score' (Fang 2003). Another study did not report how milk volume was measured and only stated 'no change' in the results (Zarate 1976), while Yulinda 2017 measured volume without specifying the specific time of measurement.

Of note, the methods for measuring or estimating milk volume varied considerably across the studies and can broadly be categorized to the following.

1. Weighing the infant before and after feeds to determine total milk transfer, then adding to this the residual milk expressed after the feeding (Aono 1982; Di Pierro 2008).

2. Weighing the infant before and after feeds without including the residual milk volume (Damanik 2006; De Gezelle 1983; Huang 2000; Jiang 2006; Kauppila 1985; Mathew 2018; Ylikorkala 1982).

3. Expression of milk using hand or breast pump (Bumrungpert 2018; Chan 2005; Espinosa‐Kuo 2005; Jantarasaengaram 2012; Nordin 2019; Sakka 2014; Su 2008; Sy 2012; Turkyilmaz 2011; Yin 2005; Yulinda 2017).

4. Measuring the dimensions of the breast before and after feeding to calculate milk removed, then adding to this the residual milk volume expressed after the feeding (Xu 2000).

5. Paritakul 2016: the authors attempted to apply an alternative method of calculating daily milk production as described by Lai 2010, by having the mother empty her breasts fully, then pumping again after an hour (i.e. second milk expression point) and multiplying the yield by 24. However, the Lai method actually describes an initial emptying of the breasts, repeated again one and two hours later for three total expressions. The third is the "second hour expression" that should be multiplied by 24 for total daily milk production. Use of the second milk expression point rather than the third could overestimate milk production.

The unit used for milk volume was either grams or millilitres (mL). We considered grams and mL to be equivalent in this review. Only one study specifically reported that they used the factor 0.93 to convert grams to mL (Damanik 2006).

Adverse effects

Of the 41 included studies, 20 reported on adverse effects. Only three studies prespecified in their methods or protocol specific adverse effects of interest. Briton‐Medrano 2002 looked for constipation and hypersensitivity reaction in mothers; De Gezelle 1983 looked for breast engorgement or tenderness and milk leakage; and Jantarasaengaram 2012 looked for headache, dry mouth, diarrhoea, muscle cramps, itching or allergic reactions in mothers. Five studies reported in their methods that they would look for adverse effects (Bumrungpert 2018; Damanik 2006; Espinosa‐Kuo 2005; Sy 2012; Wagner 2019), but did not specify any in particular. The remaining did not prespecify that they would measure adverse effects but did report the presence or absence of adverse effects in their results.

Seven studies reported on adverse effects separately for both mother and infant (Bumrungpert 2018; De Gezelle 1983; Shariati 2004; Turkyilmaz 2011; Wagner 2019; Ylikorkala 1982; Zarate 1976); seven studies reported on adverse effects in mothers only (Briton‐Medrano 2002; Espinosa‐Kuo 2005; Jantarasaengaram 2012; Kauppila 1985; Khairani 2017; Sharma 1996; Sy 2012), while in the remaining six (Damanik 2006; Li 2010; Manjula 2014; Su 2008; Yabes‐Almirante 1996a; Yin 2005), it was not clear if they were reporting adverse effects in the mother, the infant, or both.

Ability of mother to stop or reduce supplementation

Five studies reported this outcome as volume of supplemental feeds given before and after the intervention (Jiang 2006; Li 2010; Manjula 2014; Sharma 1996; Thaweekul 2014), and one reported the number of mother‐infant dyads who were able to reduce or terminate supplemental feeds after intervention (Ylikorkala 1982).

Measures of maternal psychological status

Quality of life and maternal satisfaction were recorded in four studies. Wagner 2019 used the World Health Organization Quality of Life assessment (WHOQOL), satisfaction with Life Scale, State/Trait Anxiety Inventory, Edinburgh Postnatal Depression Scale, Breastfeeding Self‐Efficacy Scale and mothers' perception of infant satisfaction. Gupta 2011 and Manjula 2014 used non‐validated ordinal scales, while Chan 2005 interviewed the mothers on their perception of milk production.

Excluded studies

After examining the abstract or full text, we excluded 93 studies from this review. Of the excluded studies, 39 investigated pharmacological interventions, 49 investigated natural interventions, two studies examined both pharmacological and natural interventions together, two were unclear as to what the intervention was and one was a review of galactagogues. See Appendix 3 for overview of types of galactagogues in excluded studies.

Reasons for exclusion were: trials reported in a way that made it difficult to confirm whether they were RCTs (23 studies); not RCTs (23 studies); cross‐over trial (1 study); systematic review of galactagogues (1 study); trials terminated before completion without any results (2 studies); trials included infants that were sick, preterm or required hospitalization (11 studies); trials studied a galactagogue but were not designed to look at the galactagenic effect of the intervention (7 studies); trials used ergometrine, which is a breast milk suppressant as their placebo (2 studies); trials included interventions that were not taken orally (16 studies); trial included animals and combined the results of humans and animals (1 study); trials included interventions that we do not consider to be galactagogues (3 studies); trials of which we only have titles and all attempts to contact the authors for the abstract or full text have failed (2 studies).

Details of the excluded studies are found in Characteristics of excluded studies.

Risk of bias in included studies

Details of the 'Risk of bias' assessment for each of the included studies are presented in Characteristics of included studies. Summary descriptions of the assessments are presented in Figure 2 and Figure 3.

2.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

3.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

We judged 18 studies to have low risk of bias for sequence generation (Briton‐Medrano 2002; Bumrungpert 2018; Damanik 2006; Di Pierro 2008; Ghasemi 2018; Gupta 2011; Inam 2013; Jantarasaengaram 2012; Jiang 2006; Manjula 2014; Mathew 2018; Paritakul 2016; Sakha 2008; Sakka 2014; Sharma 1996; Tirak 2008; Wagner 2019; Ylikorkala 1982). We judged four to have high risk of bias because they were quasi‐randomised studies (Aono 1982; Mukherjee 1987; Nordin 2019; Thaweekul 2014). Nineteen did not adequately describe how sequence generation was accomplished.

We judged 13 studies to have low risk of bias for allocation concealment (Briton‐Medrano 2002; Bumrungpert 2018; Di Pierro 2008; Gupta 2011; Jantarasaengaram 2012; Paritakul 2016; Sakha 2008; Sakka 2014; Sharma 1996; Sy 2012; Tirak 2008; Wagner 2019; Ylikorkala 1982). We judged four as high risk: Aono 1982 allocated participants according to the delivery week, Thaweekul 2014 allocated participants according to the admission month, Damanik 2006 did not conceal the allocation, and Nordin 2019 allocated the first batch of participants to the intervention group and the subsequent batch to the placebo group. Twenty‐four studies did not adequately describe how allocation concealment was performed and so we judged them at unclear risk of bias.

Blinding

Blinding of participants and personnel (performance bias)

We judged 10 studies to have low risk of bias for blinding of participants and personnel (Barguno 1988; Briton‐Medrano 2002; De Gezelle 1983; Di Pierro 2008; Jantarasaengaram 2012; Manjula 2014; Paritakul 2016; Sakha 2008; Ylikorkala 1982; Zarate 1976). We judged eight studies at unclear risk of bias: four because they did not adequately describe the intervention and placebo (Aono 1982; Balahibo 2002; Espinosa‐Kuo 2005; Kauppila 1985); three because there could be potential differences (taste, smell, colour or texture) that could be distinguished by participants (Bumrungpert 2018; Khairani 2017; Wagner 2019), and one that did not mention the type of control used (Yulinda 2017). We judged 23 studies at high risk of bias for the following reasons:

• Intervention and placebo were distinguishable by appearance: Damanik 2006 (soup versus capsule versus tablet); Fang 2003 (liquid versus powder); Nordin 2019 (banana flour biscuits versus wheat flour biscuits); Sy 2012 (capsule versus tablet); Tirak 2008 (granule versus tea bags).

• Intervention and placebo were distinguishable by smell or taste: Chan 2005; Jiang 2006; Ghasemi 2018; Li 2010; Mathew 2018; Sakka 2014; Shariati 2004; Su 2008; Thaweekul 2014; Turkyilmaz 2011; Xu 2000.

• Intervention and placebo were distinguishable by labels: Gupta 2011; Mukherjee 1987; Sharma 1996; Yabes‐Almirante 1996a.

• No placebo for the control group: Huang 2000; Inam 2013; Yin 2005.

Blinding of outcome assessment (detection bias)

We assessed detection bias separately for infant weight, milk volume and self‐reported outcomes. We chose to judge detection bias as low (no) risk in cases where these outcomes were not a part of the included studies or were not used in our analyses.

Infant weight: we judged all studies to have low risk of detection bias.

Milk volume: we judged 19 studies to have low risk of detection bias (Balahibo 2002; Barguno 1988; Briton‐Medrano 2002; De Gezelle 1983; Di Pierro 2008; Ghasemi 2018; Gupta 2011; Jantarasaengaram 2012; Manjula 2014; Mathew 2018; Paritakul 2016; Sakha 2008; Shariati 2004; Sharma 1996; Thaweekul 2014; Tirak 2008; Yabes‐Almirante 1996a; Ylikorkala 1982; Zarate 1976). We judged six at high risk of detection bias because the outcome assessors were not blinded or blinding was not successful (Chan 2005; Damanik 2006; Fang 2003; Mukherjee 1987; Nordin 2019; Sakka 2014). In these studies, milk volume was determined by milk expression, and knowledge of the intervention could have influenced the effort each mother took while expressing her milk. We judged 16 at unclear risk because blinding for the outcome assessor was either not reported or could not be determined (Aono 1982; Bumrungpert 2018; Espinosa‐Kuo 2005; Huang 2000; Inam 2013; Jiang 2006; Kauppila 1985; Khairani 2017; Li 2010; Su 2008; Sy 2012; Turkyilmaz 2011; Xu 2000; Yin 2005; Wagner 2019; Yulinda 2017).

Self‐reported outcomes: these are outcomes that were reported by the mothers and included satisfaction, duration of breastfeeding, amount of supplemental feeding, and adverse effects. We judged 24 studies to have low risk of detection bias. We judged 12 at high risk of detection bias because these mothers were unlikely to be blinded and this could affect how they perceived satisfaction and adverse effects (Chan 2005; Fang 2003; Gupta 2011; Li 2010; Mukherjee 1987; Shariati 2004; Sharma 1996; Su 2008; Sy 2012; Thaweekul 2014; Turkyilmaz 2011; Yabes‐Almirante 1996a). We judged five as unclear risk because they did not adequately describe the intervention and placebo and thus we were unable to determine if this would affect the mother's perception of these subjective outcomes (Aono 1982; Espinosa‐Kuo 2005; Kauppila 1985; Khairani 2017; Wagner 2019).

Incomplete outcome data

We judged 25 studies to have low risk of attrition bias (Bumrungpert 2018; Damanik 2006; Di Pierro 2008; Fang 2003; Ghasemi 2018; Huang 2000; Inam 2013; Jantarasaengaram 2012; Jiang 2006; Kauppila 1985; Khairani 2017; Li 2010; Manjula 2014; Mathew 2018; Nordin 2019; Paritakul 2016; Sakha 2008; Su 2008; Thaweekul 2014; Turkyilmaz 2011; Wagner 2019; Xu 2000; Yabes‐Almirante 1996a; Yin 2005; Ylikorkala 1982). We judged six at high risk of attrition bias: Barguno 1988, Briton‐Medrano 2002, Shariati 2004, Sy 2012, and Tirak 2008 had high dropout rates, while Mukherjee 1987 did not analyse all recruited participants. We judged 10 judged as having an unclear risk of attrition bias: Aono 1982, Balahibo 2002, Chan 2005, De Gezelle 1983, Espinosa‐Kuo 2005, Gupta 2011, Sakka 2014, Sharma 1996, Yulinda 2017, and Zarate 1976 did not describe the flow of participants in the study, and hence we do not know what happened to all the participants at the end of the study.

Selective reporting

We judged 33 studies to have low risk of reporting bias (Aono 1982; Balahibo 2002; Barguno 1988; Briton‐Medrano 2002; Bumrungpert 2018; Chan 2005; De Gezelle 1983; Di Pierro 2008; Espinosa‐Kuo 2005; Fang 2003; Ghasemi 2018; Gupta 2011; Huang 2000; Inam 2013; Jantarasaengaram 2012; Jiang 2006; Kauppila 1985; Li 2010; Manjula 2014; Paritakul 2016; Sakha 2008; Sakka 2014; Shariati 2004; Sharma 1996; Su 2008; Thaweekul 2014; Tirak 2008; Turkyilmaz 2011; Wagner 2019; Xu 2000; Yin 2005; Yabes‐Almirante 1996a; Ylikorkala 1982). We judged seven to have high risk of reporting bias because of the following reasons: Damanik 2006, Mukherjee 1987, and Sy 2012 did not report the adverse effects although it was stated as an outcome in the study methods; Mathew 2018 and Yulinda 2017 did not report standard deviation; and Nordin 2019 and Zarate 1976 had differences between planned methods and results reported. We judged one study as unclear because the methods section did not specify the outcomes of the study (Khairani 2017).

Other potential sources of bias

We judged 28 studies to have low risk of other biases (Aono 1982; Briton‐Medrano 2002; Bumrungpert 2018; Chan 2005; Di Pierro 2008; Espinosa‐Kuo 2005; Fang 2003; Ghasemi 2018; Huang 2000; Jantarasaengaram 2012; Jiang 2006; Li 2010; Manjula 2014; Mathew 2018; Mukherjee 1987; Nordin 2019; Paritakul 2016; Sakka 2014; Shariati 2004; Sharma 1996; Sy 2012; Thaweekul 2014; Tirak 2008; Turkyilmaz 2011; Wagner 2019; Yabes‐Almirante 1996a; Ylikorkala 1982; Zarate 1976). Of the remaining studies, we judged one at high risk of other bias because of baseline imbalances in the characteristics of mothers in the intervention and placebo groups (Kauppila 1985), while we judged 12 at unclear risk because there was no baseline information available for the intervention and placebo groups (Balahibo 2002; Barguno 1988; Damanik 2006; De Gezelle 1983; Gupta 2011; Inam 2013; Khairani 2017; Sakha 2008; Su 2008; Xu 2000; Yin 2005; Yulinda 2017).

Effects of interventions

See: Table 1; Table 2

Forty‐one studies involving 3005 mothers and 3006 infants (1 set of twins) were included in this review.

Comparison 1: pharmacological oral galactagogues compared with placebo or no treatment

Nine parallel studies compared pharmacological galactagogues with a placebo or no treatment (Aono 1982; Barguno 1988; De Gezelle 1983; Inam 2013; Jantarasaengaram 2012; Kauppila 1985; Sakha 2008; Ylikorkala 1982; Zarate 1976).

Types of pharmacological galactagogues contributing to this comparison were: domperidone (Inam 2013; Jantarasaengaram 2012), metoclopramide (De Gezelle 1983; Kauppila 1985; Sakha 2008), sulpiride (Aono 1982; Barguno 1988; Ylikorkala 1982), and thyrotropin‐releasing hormone (Zarate 1976).

Primary outcomes

Proportion of mothers who continued breastfeeding (exclusive or any) at 3, 4 and 6 months

None of the studies in this comparison reported this outcome.

Infant weight in trials where the infants received only their own mother's milk (g)

One study, using metoclopramide, reported this outcome (Sakha 2008), which found no difference in infant weight gain with metoclopramide compared to placebo at 15 days (mean difference (MD) 23.00, 95% confidence interval (CI) ‐47.71 to 93.71; 1 study, 20 participants; low‐certainty evidence; Analysis 1.1).

1.1. Analysis.

Comparison 1: Pharmacological oral galactagogues versus placebo or no treatment, Outcome 1: Infant weight (where they were only receiving own mother's milk) at the end of the study (grams)

Another study tested two different doses of sulpiride (Barguno 1988), but the data were only presented in a graph. We were unable to extract any data from the graph and could not contact the authors. It was reported in the publication that there was a larger weight gain with a higher dose of sulpiride during the first two weeks of the study, but there was no difference between the two groups as the study progressed with a lower dose of sulpiride.

Volume of breast milk at the latest time measured (mL)

Seven studies reported milk volume at the end of the study. Three studies, Jantarasaengaram 2012 (domperidone), De Gezelle 1983 (metoclopramide), and Aono 1982 (sulpiride), suggested that pharmacological galactagogues may increase milk volume (MD 63.82 mL, 95% CI 25.91 to 101.72; I² = 34%; 3 studies, 151 participants; low certainty‐evidence; Analysis 1.2).

1.2. Analysis.

Comparison 1: Pharmacological oral galactagogues versus placebo or no treatment, Outcome 2: Milk volume subgroup by type of galactagogue(please refer to footnotes for details on units used)

The four remaining studies did not provide data that could be analysed. A narrative summary is as follows: Kauppila 1985 (metoclopramide) presented individual participants' milk volumes over 20 days in a graph, concluding that "the milk yield increased significantly" in the intervention group. Ylikorkala 1982 (sulpiride) also reported their data in a graph showing the changes in the daily milk yield between groups. In the intervention group, there was an increase in milk yield while the milk yield was reduced in the control group. Inam 2013 (domperidone) reported more mothers in the galactagogue group having milk volume equal to or greater than 50 mL (defined as 'efficacious' in the study). Zarate 1976 (thyrotopin‐releasing hormone) reported "no change" in the amount of milk produced for both groups without reporting any numerical values.

Secondary outcomes

Adverse effects for the infant or mother

We were unable to perform a meta‐analysis because the results could not be meaningfully combined. Table 7 provides an overall summary of the prespecified adverse effects, reported adverse effects and adverse effects that were not reported in the included studies.

5. Overview of adverse effects reported in the included studies.

Galactagogue	Study ID	Adverse effects prespecified in study method	Adverse effects reported*
Asparagus racemosus	Gupta 2011	Nothing prespecified	Nothing reported
Banana flower	Nordin 2019	Nothing prespecified	Nothing reported
Chan Bao oral liquid or Bu Xue Sheng Ru	Jiang 2006	Nothing prespecified	Nothing reported
Cui Ru soup	Su 2008	Nothing prespecified	None occurred (not specified for mother or infant)
Domperidone	Inam 2013	Nothing prespecified	Nothing reported
Domperidone	Jantarasaengaram 2012	Headache, dry mouth, diarrhoea, muscle cramps, itching or allergic reactions (prespecified for mothers)	Dry mouth (intervention 7/22 mothers, control 0/23 mothers). Extrapyramidal effects (intervention 0/22 mothers, control 0/23 mothers)
Fennel or fenugreek	Ghasemi 2018	Nothing prespecified	Nothing reported
Fennel or fenugreek	Mathew 2018	Nothing prespecified	Nothing reported
Fenugreek or palm dates	Sakka 2014	Nothing prespecified	Nothing reported
Ginger	Paritakul 2016	Nothing prespecified	Nothing reported
Gossypium herbaceum Linn	Manjula 2014	Nothing prespecified	None occurred (not specified for mother or infant)
Humana Still Tea	Tirak 2008	Nothing prespecified	Nothing reported
Humana Still Tea with fenugreek listed as main ingredient	Turkyilmaz 2011	Nothing prespecified	None occurred in mothers and infants
Ixbut	Chan 2005	Nothing prespecified	Nothing reported
Lactare	Mukherjee 1987	Nothing prespecified	Nothing reported
Moringa	Balahibo 2002	Nothing prespecified	Nothing reported
Moringa	Briton‐Medrano 2002	Constipation, hypersensitivity reactions (prespecified for mothers)	None occurred in mothers.
Moringa	Espinosa‐Kuo 2005	Nothing prespecified	None occurred in mothers
Moringa	Khairani 2017	Nothing prespecified	None occurred in mothers. Nothing reported on infants
Moringa	Yabes‐Almirante 1996a	Nothing prespecified	None occurred (not specified for mother or infant)
Moringa and domperidone	Sy 2012	Nothing prespecified	Decrease appetite (domperidone 1/9 mothers, moringa 0/8 mothers)
Metoclopramide	De Gezelle 1983	Breast engorgement or tenderness, milk leakage (prespecified for mothers). None for infants	None occurred in mothers and infants
Metoclopramide	Kauppila 1985	Nothing prespecified	Tiredness (intervention 4/11 mothers, control 3/14 mothers). Tiredness and headache (intervention 1/11 mothers, control 0/14 mothers). Tiredness and nausea (intervention 1/11 mothers, control 0/14 mothers). Dizziness and sweating (intervention 0/11 mothers, control 1/14 mothers)
Metoclopramide	Sakha 2008	Nothing prespecified	Nothing reported
Mixed fenugreek, ginger, tumeric	Bumrungpert 2018	Nothing prespecified	Maple syrup urine odour (intervention 2/25 mothers, control 0/25 mothers). Excessive rectal gas (intervention 2/25 mothers, control 2/25 mothers). None detected in infants
Mixed galactagogue with asparagus as main ingredient	Sharma 1996	Nothing prespecified	No biochemical liver cell dysfunctions or any adverse effects were reported in mothers of either group
Mother's Milk Tea (MMT)	Wagner 2019	Nothing prespecified	None occurred in mothers and infants
Mu Er Wu You or Kun Yuan Tong Ru	Li 2010	Nothing prespecified	No adverse effects occurred (not specified for mother or infant)
Palm dates	Yulinda 2017	Nothing prespecified	Nothing reported
Pork leg soup with spring onion	Xu 2000	Nothing prespecified	Nothing reported
Ru Quan Chong Ji or Sheng Ru Hi Zhi	Fang 2003	Nothing prespecified	Nothing reported
Sheng Ru He Ji	Yin 2005	Nothing prespecified	None occurred (not specified for mother or infant)
Shirafza drops	Shariati 2004	Nothing prespecified	"No difference" between 2 groups for flatulence and headache in both groups (mother). Nausea (intervention 2 infants, control 0 infants). Urticaria (intervention 2 infants, control 0 infants). Total participants not mentioned
Silymarin	Di Pierro 2008	Nothing prespecified	Nothing reported
Sulpiride	Aono 1982	Nothing prespecified	Nothing reported
Sulpiride	Barguno 1988	Nothing prespecified	Nothing reported
Sulpiride	Ylikorkala 1982	Nothing prespecified	Headache (intervention 1/14 mothers, control 0/12 mothers) Tiredness (intervention 2/14 mothers, control 0/12 mothers). None occurred for infants
Thai food	Thaweekul 2014	Nothing prespecified	Nothing reported
Torbangun or fenugreek or moloco	Damanik 2006	Stated that they would be looking for adverse effects but did not specify what adverse effect	Nothing reported for mothers and infants
Thyrotropin‐releasing hormone	Zarate 1976	Nothing prespecified	"No clinical hyperthyroidism in infant and mother"
Xian Tong Ru oral liquor	Huang 2000	Nothing prespecified	Nothing reported
* "Nothing reported" signifies that there was no mention in the results section about adverse effects. "None occurred" means that adverse effects were specifically looked for, but none were identified.

For mothers

De Gezelle 1983 (metoclopramide) prespecified breast engorgement or tenderness and milk leakage; the study reported that none of these occurred. Jantarasaengaram 2012 (domperidone) prespecified headache, dry mouth, diarrhoea, muscle cramps, itching or allergic reactions; and reported dry mouth in seven out of 25 mothers in the domperidone group and none in the placebo group; no other adverse effects were reported. This study also reported no 'extrapyramidal effects' which was a non‐prespecified outcome. Zarate 1976 (thyrotropin‐releasing hormone) reported “no clinical hyperthyroidism.” Ylikorkala 1982 (sulpiride) reported headache (1 mother) and tiredness (2 mothers) out of the 14 mothers in the sulpiride group, and none in the placebo group. Kauppila 1985 (metoclopramide) reported that "Six women (out of 11) taking MC complained of side‐effects; four of tiredness alone, one of tiredness and headache, and one of tiredness and nausea. Three women (out of 14) receiving the placebo suffered from tiredness and one from dizziness and sweating."

For infants

De Gezelle 1983 (metoclopramide) and Ylikorkala 1982 (sulpiride) reported "no adverse effects.” Zarate 1976 (thyrotropin‐releasing hormone) reported "no clinical hyperthyroidism."

Ability of mother to stop or reduce supplementation with formula milk

Only one study reported the number of infants stopping supplemental feeding (Ylikorkala 1982), which was greater in the galactagogue group (4 out of 14 participants) compared to the placebo group (0 out of 14 participants).

Measures of maternal psychological status

No studies reported this outcome.

Comparison 2: natural (non‐pharmacological) oral galactagogues compared with placebo or no treatment

Twenty‐seven studies compared natural galactagogues with placebo: two were four‐arm studies (Balahibo 2002; Khairani 2017); five were three‐arm studies (Ghasemi 2018; Jiang 2006; Sakka 2014; Tirak 2008; Turkyilmaz 2011), and 20 were parallel studies (Briton‐Medrano 2002; Bumrungpert 2018; Chan 2005; Di Pierro 2008; Espinosa‐Kuo 2005; Gupta 2011; Huang 2000; Manjula 2014; Mukherjee 1987; Nordin 2019; Paritakul 2016; Sharma 1996; Shariati 2004; Su 2008; Thaweekul 2014; Wagner 2019; Xu 2000; Yabes‐Almirante 1996a; Yin 2005; Yulinda 2017).

Types of natural galactagogues contributing to this comparison were: banana flower (Nordin 2019); Bu Xue Sheng Ru (补血生乳) (Jiang 2006); Chanbao Oral Liquid (产宝) (Jiang 2006); Cui Ru (催乳汤) (Su 2008); fennel (Foeniculum vulgare) (Ghasemi 2018); fenugreek (Trigonella foenum‐graecum L) (Ghasemi 2018; Sakka 2014); galactagogue foods (Thaweekul 2014); ginger (Zingiber officinale) (Paritakul 2016); mixed botanical teas (Humana Still Tee) (Tirak 2008; Turkyilmaz 2011)/(Mother's Milk Tea) ( Wagner 2019); ixbut (Euphorbia lancifolia) (Chan 2005); Lactare (Mukherjee 1987); levant cotton (Gossypium herbaceum Linn) kernels (Manjula 2014); moringa leaves (Balahibo 2002; Briton‐Medrano 2002; Espinosa‐Kuo 2005; Khairani 2017; Yabes‐Almirante 1996a); mixed fenugreek, ginger and tumeric capsules (Bumrungpert 2018); mixed galactagogue with Shatavari (Asparagus racemosus) as main ingredient (Sharma 1996); palm dates (Sakka 2014; Yulinda 2017); pork knuckle soup (Xu 2000); shatavari (Asparagus racemosus) (Gupta 2011); Sheng Ru He Ji (生乳合剂) (Yin 2005); Shirafza: combination alcohol extraction of fennel (Foeniculum vulgare), anise (Pimpinella anisum), green cumin (Cuminum cyminum), dill (Anethum gravolens), parsley (Petroselinum crispum), black seed (Nigella sativa) (Shariati 2004), silymarin (Silybum marianum) (Di Pierro 2008), and Xian Tong Ru (先通乳) (Huang 2000).

Primary outcomes

Proportion of mothers who continued breastfeeding (exclusive or any) at 3, 4 and 6 months

Wagner 2019 (Mother's Milk Tea) reported "no significant difference" in breastfeeding rates at 6 months between the groups with no data or level of significance.

Infant weight in trials where the infants received only their own mother's milk (g)

Seven studies reported this outcome, but only three had data that could be analysed. We used the infant weight at the end of the study for our analysis, as the outcome was measured at different time points in all the studies. Out of these three, there was only one study for each type of natural galactagogue: fennel and fenugreek (Ghasemi 2018), Humana Still Tee (Tirak 2008), and moringa leaves (Yabes‐Almirante 1996a).

The point estimates for all of the four galactagogues compared to control fell on the side that favoured galactagogues, with very wide 95% CIs (3 studies, 275 participants; very low‐certainty evidence; Analysis 2.1). We did not perform any meta‐analysis as there was substantial heterogeneity between the different galactagogues (I² = 63.9%).

2.1. Analysis.

Comparison 2: Natural oral galactagogues versus placebo or no treatment, Outcome 1: Infant weight subgroup by type of galactagogue (where they were only receiving own mother's milk) at the end of the study (grams)

We could not analyse data from the other four studies. Balahibo 2002, Gupta 2011 and Sakka 2014 reported mean percentage weight gain (all reported higher weight gain in the galactagogue group). Shariati 2004 did not report the number of participants in each group, but reported "no difference in infant weight gain per week at the end of intervention."

Volume of breast milk at the latest time measured (mL)

Seventeen studies reported milk volume at the end of the study using various methods. Only 13 had data that could be analysed: two with moringa leaves (Briton‐Medrano 2002; Espinosa‐Kuo 2005), and one each with the following galactagogues: mixed fenugreek, ginger and tumeric capsules (Bumrungpert 2018); ixbut (Euphorbia lancifolia) (Chan 2005); silymarin (Silybum marianum) (Di Pierro 2008); Xian Tong Ru (先通乳) (Huang 2000); banana flower (Nordin 2019 ); Bu Xue Sheng Ru (补血生乳) and Chanbao Oral Liquid (产宝) (Jiang 2006); ginger (Zingiber officinale) (Paritakul 2016); Cui Ru (催乳汤) (Su 2008); Humana Still Tee (Turkyilmaz 2011); and Sheng Ru He Ji (生乳合剂) (Yin 2005). There was also one three‐arm study comparing fenugreek, palm dates and placebo (Sakka 2014).

The point estimates for all of the galactagogue types fell on the side that favoured galactagogues, with very wide 95% CIs (13 studies, 962 participants; very low‐certainty evidence; Analysis 2.2). We did not perform any meta‐analysis as there was high heterogeneity between the subgroups (I² = 98.9%).

2.2. Analysis.

Comparison 2: Natural oral galactagogues versus placebo or no treatment, Outcome 2: Milk volume subgroup by type of galactagogue (refer to footnotes for details on units used)

We could not analyse data from the other four studies. Khairani 2017 (moringa), Mukherjee 1987 (Lactare), and Xu 2000 (pork knuckle soup) presented the results as categorical data and concluded that more participants in the intervention group did better. Yulinda 2017 (palm dates) reported that mean milk volume was higher in the intervention group but did not present the number of participants or a measure of dispersion.

Secondary outcomes

Adverse effects for the infant or mother

We were unable to perform a meta‐analysis because the results could not be meaningfully combined. A narrative summary is provided below. Table 7 provides an overall summary of the prespecified adverse effects, reported adverse effects and adverse effects that were not reported in the included studies.

For mothers

Briton‐Medrano 2002 (moringa) prespecified constipation and hypersensitivity reactions; it was reported that none of these occurred. Bumrungpert 2018 (fenugreek, ginger and turmeric mix) reported two mothers in the galactagogue group having urine that smelled like maple syrup and excessive flatus in two mothers each in both groups. Espinosa‐Kuo 2005, Khairani 2017 (moringa), Turkyilmaz 2011 (Humana Still Tee), and Wagner 2019 (Mother's Milk Tea) reported "no adverse effects". Shariati 2004 (Shirafza) reported "no difference" between groups on the occurrence of flatulence and headache. Sharma 1996 (mixed galactagogue with Shatavari (Asparagus racemosus) as main ingredient) reported "no biochemical liver cell dysfunctions noted in any of the subjects in either group and nor were any significant side effects".

For infants

Bumrungpert 2018 (fenugreek, ginger and turmeric mix) reported "adverse effects were not found in infants." Turkyilmaz 2011 (Humana Still Tee) and Wagner 2019 (Mother's Milk Tea) reported “no adverse effects.” Shariati 2004 (Shirafza) reported the occurrence of nausea (6 infants) and urticaria (2 infants) in the Shirafza group and none in the placebo group.

Unclear if the report was for mothers or infants

Manjula 2014 (levant cotton (Gossypium herbaceum Linn) kernels) reported "No side effects". Shariati 2004 (Shirafza) reported "itchiness and redness" (1 participant) in the placebo group and none in the Shirafza group. Su 2008 (Cui Ru (催乳汤) reported no adverse effects.Yabes‐Almirante 1996a (moringa) reported "no reported adverse effects from the study" in the text, but an infant death in the galactagogue group was recorded in one of the tables. There was insufficient information to judge whether this was related to the intervention and we were unable to contact the authors for clarification. Yin 2005 (Sheng Ru He Ji) and Yabes‐Almirante 1996a (moringa) reported "no adverse effects."

Ability of mother to stop or reduce supplementation with formula milk

Three studies reported this outcome as volume of supplemental feeds given before and after the intervention (Manjula 2014; Sharma 1996; Thaweekul 2014).

Two studies reporting volume of supplemental feeding could be presented in a forest plot. There was only one study per galactagogue type: Gossypium herbaceum Linn (Manjula 2014) and Shatavari (Sharma 1996). The point estimates for both the galactagogue types fell on the side that favoured galactagogues (2 studies, 109 participants; very low‐certainty evidence; Analysis 2.3). We did not perform any meta‐analysis as there was high heterogeneity between the subgroups (I² = 74.7%).

2.3. Analysis.

Comparison 2: Natural oral galactagogues versus placebo or no treatment, Outcome 3: Volume of supplement beyond mother's own milk (mL)

One study had no data that could be analysed. Thaweekul 2014 (galactagenic food) reported "no significant difference" in the median and interquartile range of the volume of supplemental feeding between the intervention and control group.

Measures of maternal psychological status

Four studies reported this outcome, but only one used validated scales (Wagner 2019). There was no difference between the two groups for the WHO Quality of Life scale (WHOQOL‐BREF) (MD ‐0.01, 95% CI ‐3.84 to 3.82; 1 study, 60 participants; low‐certainty evidence; Analysis 2.4), the Breastfeeding Self‐Efficacy Scale (MD ‐1.74, 95% CI ‐4.47 to 0.99; 1 study, 60 participants; low‐certainty evidence; Analysis 2.5), or the Edinburgh Postnatal Depression Scale (MD 1.38, 95% CI ‐0.21 to 2.97; 1 study, 60 participants; low‐certainty evidence; Analysis 2.6).

2.4. Analysis.

Comparison 2: Natural oral galactagogues versus placebo or no treatment, Outcome 4: Quality of life using WHO QOL Scale

2.5. Analysis.

Comparison 2: Natural oral galactagogues versus placebo or no treatment, Outcome 5: Breastfeeding self‐efficacy

2.6. Analysis.

Comparison 2: Natural oral galactagogues versus placebo or no treatment, Outcome 6: Postpartum Depression Scale

Three other studies reported maternal satisfaction (Chan 2005; Gupta 2011; Manjula 2014), but used different scales and criteria, none of which were validated. All reported better maternal satisfaction in the galactagogue group.

Comparison 3: oral galactagogues compared with another oral galactagogue

There were four three‐arm studies (Damanik 2006; Ghasemi 2018; Jiang 2006; Sakka 2014), and four parallel studies (Fang 2003; Li 2010; Mathew 2018; Sy 2012), leading to nine pair‐wise comparisons.

The comparisons were: torbagun leaves versus fenugreek; fenugreek versus Moloco tablet; torbagun leaves versus Moloco tablets (Damanik 2006); Ruquan‐Chongji (乳泉冲剂) versus Shengruzhi (生乳汁) (Fang 2003); Chanbao oral liquid (产宝) versus Bu Xue Sheng Ru (补血生乳) (Jiang 2006); Mu Er Wu You soup (母儿无忧汤) versus Kun Yuan Tong Ru soup (坤元通乳口服液) (Li 2010); fennel versus fenugreek (Ghasemi 2018); moringa versus domperidone (Sy 2012); and fenugreek versus palm dates (Sakka 2014).

Primary outcomes

Proportion of mothers who continued breastfeeding (exclusive or any) at 3, 4 and 6 months

None of the included studies contributed to this outcome.

Infant weight in trials where the infants received only own mother's milk (g)

Fenugreek tea compared to fennel tea (Ghasemi 2018): there was no difference in infant weight with fenugreek tea compared to fennel tea at one month (MD ‐10.25, 95% CI ‐462.91 to 442.41; 1 study, 78 participants; very low‐certainty evidence; Analysis 3.1).

3.1. Analysis.

Comparison 3: Oral galactagogue versus another galactagogue: infant weight (grams), Outcome 1: Fenugreek tea versus Fennel tea

Fenugreek tea compared to palm dates (Sakka 2014): data was not analysed as the study reported mean percentage increase, with no difference between the two groups.

Volume of breast milk at the latest time measured (mL)

Chanbao oral liquid compared to Bu Xue Sheng Ru capsules (Jiang 2006): there was no difference in milk volume between the two groups (MD ‐9.50, 95% CI ‐25.65 to 6.65; 1 study, 40 participants; low‐certainty evidence; Analysis 4.1).

4.1. Analysis.

Comparison 4: Oral galactagogue versus another galactagogue: milk volume (mL), Outcome 1: Chanbao oral liquid versus Bu Xue Sheng Ru capsules

Domperidone compared to moringa capsules (Sy 2012): there was no difference in milk volume between the two groups (MD ‐0.38, 95% CI ‐10.64 to 9.88; 1 study, 26 participants; very low‐certainty evidence; Analysis 4.2).

4.2. Analysis.

Comparison 4: Oral galactagogue versus another galactagogue: milk volume (mL), Outcome 2: Domperidone versus Moringa leave capsules

Fenugreek tea compared to palm dates (Sakka 2014): there was a lower milk volume with fenugreek tea (MD ‐16.80, 95% CI ‐27.22 to ‐6.38; 1 study, 50 participants; very low‐certainty evidence; Analysis 4.3).

4.3. Analysis.

Comparison 4: Oral galactagogue versus another galactagogue: milk volume (mL), Outcome 3: Fenugreek versus Palm dates

Fenugreek capsule compared to torbagun leaves (Damanik 2006): there was no difference in milk volume between the two groups (MD‐78.40, 95% CI ‐188.83 to 32.03; 1 study, 45 participants; very low‐certainty evidence; Analysis 4.4).

4.4. Analysis.

Comparison 4: Oral galactagogue versus another galactagogue: milk volume (mL), Outcome 4: Fenugreek capsules versus Torbangun soup

Fenugreek capsules compared to Molocco tablets (Damanik 2006): there was no difference in milk volume between the two groups (MD 15.20, 95% CI ‐108.08 to 138.48; 1 study, 44 participants; very low‐certainty evidence;Analysis 4.5).

4.5. Analysis.

Comparison 4: Oral galactagogue versus another galactagogue: milk volume (mL), Outcome 5: Fenugreek capsules versus Moloco tablets

Mu Er Wu You soup (母儿无忧汤) compared to Kun Yuan Tong Ru soup (坤元通乳口服液) (Li 2010): there was a higher milk volume with Mu Er Wu You soup (母儿无忧汤) (MD 19.95, 95% CI 6.88 to 33.02; 1 study, 90 participants; very low‐certainty evidence; Analysis 4.6).

4.6. Analysis.

Comparison 4: Oral galactagogue versus another galactagogue: milk volume (mL), Outcome 6: Mu Er Wu You soup versus Kun Yuan Tong Ru soup

Torbangun leaves compared to Molocco tablets (Damanik 2006): there was no difference in milk volume between the two groups (MD 93.60, 95% CI ‐12.39 to 199.59; 1 study, 45 participants; very low‐certainty evidence; Analysis 4.7).

4.7. Analysis.

Comparison 4: Oral galactagogue versus another galactagogue: milk volume (mL), Outcome 7: Torbangun soup versus Moloco tablets

Ruquan‐Chongji (乳泉冲剂 )compared to Shengruzhi (生乳汁) (Fang 2003): the Ruquan‐Chongji (乳泉冲剂) group had a larger mean change (improvement) in the overall symptom score (which included volume as a symptom) compared to the Shengruzhi (生乳汁) group.

Fenugreek tea compared to fennel tea (Mathew 2018): the authors reported mean pre‐ and post‐ "lactational levels" for each group but did not present a measure of dispersion.

Secondary outcomes

Adverse effects for the infant or mother

We were unable to perform a meta‐analysis because the results could not be meaningfully combined. A narrative summary is provided below. Table 7 provides an overall summary of the prespecified adverse effects, reported adverse effects and adverse effects that were not reported in the included studies.

For mothers

Sy 2012 (domperidone versus moringa) reported decreased appetite in one mother out of nine taking domperidone (Sy 2018 [pers comm]) versus none in moringa group.

For infants

No studies reported adverse effects in infants.

Unclear if the report was for mothers or infants

Li 2010 (Mu Er Wu You versus Kun Yuan Tong Ru) reported 'no adverse effects;' Damanik 2006 (torbangun, fenugreek and Molocco) did not report anything about adverse effects, although they stated that they would be looking for adverse effects in the methods.

Ability of mother to stop or reduce supplementation with formula milk

Two studies reported volume of supplemental feeds (mL) (Jiang 2006; Li 2010). There was no difference in the volume of supplemental formula milk in the Chan Bao group compared to the Bu Xue Sheng Ru group (MD ‐2.50, 95% CI ‐8.45 to 3.45; 40 participants; low‐certainty evidence; Analysis 5.1). There was reduction in supplemental formula milk volume in the Mu Er Wu You group compared to Kun Yuan Tong Ru soup (MD ‐12.25, 95% CI ‐15.63 to ‐8.87; 90 participants; very low‐certainty evidence; Analysis 5.2).

5.1. Analysis.

Comparison 5: Oral galactagogue versus another galactagogue: volume of supplementary feeds (mL), Outcome 1: Chanbao oral liquid versus Bu Xue Sheng Ru capsules

5.2. Analysis.

Comparison 5: Oral galactagogue versus another galactagogue: volume of supplementary feeds (mL), Outcome 2: Mu Er Wu You soup versus Kun Yuan Tong Ru soup

Measures of maternal psychological status

None of the included studies reported this outcome.

Discussion

Summary of main results

Our review included 41 studies with 3005 mothers, 3006 infants (1 set of twins) and 33 types of galactagogue interventions. The studies were very diverse with substantial differences in the participants, interventions, and the outcomes reported.

Pharmacological oral galactagogues compared to placebo or no intervention

We have no data on the effects of pharmacological galactagogues on proportion of mothers who continued breastfeeding (exclusive or any) at 3, 4 and 6 months as none of the included studies reported this outcome. There was no difference in infant weight measured at the end of the studies (low‐certainty evidence). However, metoclopramide, domperidone and sulpiride, at the tested dosages, are probably effective in increasing milk volume (low‐certainty evidence). Adverse effects were poorly reported and thus we are unable to comment on the risk of adverse effects with pharmacological galactagogues. For the other outcomes under this comparison, there was very little data available and we do not know the effects of pharmacological galactagogues on the ability to stop supplemental formula milk and on maternal psychological status.

Natural (non‐pharmacological) oral galactagogues compared with placebo or no treatment

We have very limited data on the effects of natural galactagogues on proportion of mothers who continued breastfeeding (exclusive or any) at 3, 4 and 6 months. The one study that measured this outcome did not provide data but reported "no significant difference" in the rates of breastfeeding at 6 months (very low‐certainty evidence). There is very low‐certainty evidence that natural galactagogues might increase infant weight, milk volume and reduce the use of supplemental formula. However, due to substantial heterogeneity, it is not possible to estimate the size of this effect. Adverse effects were also poorly reported here, and thus we are unable to comment on the risk of adverse effects with natural galactagogues. There are very limited data available from the one study which reported "no difference in mother's quality of life, breastfeeding self‐efficacy, and postnatal depression scale with natural galactagogues compared to control" (low‐certainty evidence).

Oral galactagogues compared with another oral galactagogue

We do not know the effects of one galactagogue compared to another for all outcomes because there was only one study per pair‐wise comparison.

Overall completeness and applicability of evidence

We cast a wide net for studies, using the standard Cochrane Pregnancy and Childbirth search strategy, as well as regional and content‐specific databases. The reference list of included studies led to further articles of additional galactagogue studies not found elsewhere. Through these sources and personal contacts with experts in the field, we retrieved 18 of our 41 included studies, many of which were published in non‐indexed journals or did not contain any of our search terms. However, we excluded 23 studies because it was unclear whether or not they were randomised controlled trials (RCTs).

We took measures to ensure completeness of data by attempting to contact authors of included studies for clarifications. Some studies were too old and lacked contact details. Of the 36 authors whom we managed to locate, we did not get a response from 10 of them. Those who did reply were willing to discuss their work and provide clarification or further data, but some gaps remained. Additionally, 11 of our included studies were not published in English. Translation of non‐English studies was done by more than one person to avoid misinterpreting the data or missing any pertinent information.

We encountered substantial unexplained heterogeneity for our primary outcomes. Therefore, for all but one of our outcomes we were unable to estimate the overall effect size. We attempted to explore possible explanations. One clear explanation would be the type of galactagogue. However, we were not able to establish this because on subgroup analysis for each galactagogue there was only one study for almost all galactagogue types. Another possible explanation for heterogeneity might be that some studies recruited normal women, some included women with lactation insufficiency, and some did not specify this. Where reported in the studies, we recorded this in the Characteristics of included studies. We tested this in a subgroup analysis, but it did not explain the heterogeneity (analysis not shown). This might be due to the lack of standardised criteria for defining lactation insufficiency, resulting in misclassification bias.

The heterogeneity in the results could also be due to other reasons, such as variation in the babies ages, preparation and dose of galactagogues, and the duration of treatment. Authors rarely explained their rationale for the dose that was used, potentially impacting efficacy, and only a few provided a rationale for how the intervention might work. With particular reference to natural galactagogues, it is important to identify the material used (leaf, root, seed, etc.) and the reasons for that choice, the form chosen (powdered, tincture, tea, standardised extract, etc.) and the dosage tested (Betz 2014), as this might also contribute to heterogeneity. Results may vary when different parts or preparations of the same natural galactagogue are compared (Betz 2014; Brinker 1999; Garg 2010).

The measurement of milk volume as an outcome was particularly affected by the diverse methodology of the studies. First, the age at recruitment in included studies ranged from birth to 6 months, and during the first 6 to 8 weeks especially, milk volume naturally increases according to infant demand (Kent 2016; Neville 1988). Differences in the age at recruitment and the time point of measurement, such as time elapsed from baseline to first and subsequent measurements, further complicates this issue.

In addition, milk volume was measured in a variety of ways, including volume expressed, infant test weighing, and test weights plus expressed residual milk, with some measured once a day and others measured several times a day; all of these contributed further to the heterogeneity observed. When milk was expressed, the variation in methods (hand, manual pump or electric pump (Becker 2016)), and the timing of milk expression in respect to the previous feeding, could also affect milk yields and alter results.

Infant weight as an outcome was difficult to interpret. In studies of mothers with normal milk production, it is believed that authors assumed that if a galactagogue increases milk production, the infant will drink more and this can be measured through weight gain. However, an infant who has already taken plenty of milk may or may not drink more just because it is there, while an infant who needs more is likely to take more unless a suck problem limits her/his ability to do so. Each scenario presents a confounding variable that can influence weight gain and thus the validity of the conclusions. In studies that included mothers with low milk production, necessary use of supplementation required us to exclude them from the infant weight outcome analysis. For all of these reasons, we consider infant weight to be a poor surrogate marker of milk production.

Although this review looked at the effect of galactagogues on milk production, the ultimate goal is to increase the exclusivity and duration of breastfeeding. Therefore, we considered the proportion of women breastfeeding at 3, 4 and 6 months to be the most important of our three primary outcomes, something only one study measured but did not fully report (Wagner 2019).

The reporting of adverse effects in our included studies was poor, as many failed to specify whether the effect occurred in the mother or her infant. Many studies did not specify if they were actively looking for adverse effects.

One of the outcomes reported in several of the included studies was maternal serum prolactin levels: (Aono 1982; Barguno 1988; De Gezelle 1983; Gupta 2011; Kauppila 1985; Li 2010; Paritakul 2016; Sharma 1996; Turkyilmaz 2011; Yabes‐Almirante 1996a; Ylikorkala 1982). We did not include this outcome for this review because a wide range of factors influence prolactin levels (Hill 2009; Stuebe 2015; Zhang 2016), and there is no good evidence of a particular cutoff value of prolactin that might be associated with adequate milk production (Brodribb 2018; Cox 1996; Stuebe 2015). It is understood that there likely is a minimum threshold for prolactin needed for proper secretory activation (Nedkova 1995), and that the degree of the prolactin surge response to a feeding may be more important than serum levels (Eglash 2015).

Certainty of the evidence

Using the GRADE assessment tool, the overall certainty of evidence for our primary outcomes ranged from low to very low. The main reasons for downgrading the evidence were for imprecision, inconsistency, risk of bias, and indirectness.

We downgraded all outcomes for imprecision because the number of participants was small and did not meet the optimal information size. In addition, we downgraded several outcomes twice because of wide CIs. As discussed in the section on Overall completeness and applicability of evidence, the diversity of galactagogues and methodology used resulted in downgrading the outcomes for inconsistency. We also had serious concerns with risk of bias, mainly due to lack or poor descriptions of blinding.

Potential biases in the review process

We aimed to minimize bias at each stage of the review process. At least two review authors independently assessed eligibility for inclusion, carried out data extraction and assessed risk of bias. However, we had to make many subjective judgements in this review because many of the studies had major issues with reporting, particularly with the methods of random sequence generation and allocation concealment, as well as the description of the flow of participants in the study. Language issues may have played a role in how the studies were described by authors whose native language was not English. We have tried to be as transparent as possible regarding our judgements so that readers can choose to agree or disagree with our judgements. These judgements included making several posthoc changes to the review process, including adding three additional subgroup analyses and a restructuring of our comparisons. We also made minor changes to the words specified in the protocol for the primary outcomes to reflect the evidence across studies and note that these differences are unlikely to introduce bias given the similarity in definitions and that downgrades have already been made to the certainty of this evidence.These are detailed in the section Differences between protocol and review.

We excluded a large number of studies (n = 23) because we could not determine from the papers if the studies were RCTs. These were mainly older studies and studies from non‐standard database sources. Poorly reported studies may have been misjudged as to not be eligible, usually due to lack of clarity about whether the study was a RCT. We could essentially be wrong about excluding these studies, and if so, have excluded a considerable body of evidence from the review. However, we could not contact the authors of these studies to clarify and our intention was to keep this review strictly to RCTs and quasi‐RCTs.

During our search for evidence, we were made aware of a number of non‐English databases, such as Japanese and Chinese language databases, which could potentially contain galactagogue studies. However, from the studies we identified in our search, we found that most of the Chinese and Japanese studies we identified in these databases used traditional medicine concepts such as 'body harmony' and these could not be translated into the type of outcomes of interest to this review. Such studies are not typically reported as RCTs, therefore the yield from these databases would not justify any further searching.

Agreements and disagreements with other studies or reviews

There are a number of other systematic reviews published on galactagogues. One was a Cochrane Review on mothers with preterm hospitalised infants (Donovan 2012), which concluded from limited data that, with the use of domperidone, there is improvement in expressed milk volume in mothers of preterm infants who have insufficient milk.

The are several other systematic reviews that looked at a variety of pharmacological and natural galactagogues (Bazzano 2016; Budzynska 2012; Forinash 2012; Mortel 2013; Zapantis 2012; Zuppa 2010). Their included studies looked at fenugreek, torbangun, milk thistle, shatavari, domperidone and metoclopramide, which were also galactagogues used in the studies included in our review. There were also other reviews focusing on individual galactagogues, for example, Osadchy 2012 reviewed the effects of domperidone, King 2013 reviewed moringa and Khan 2018 reviewed fenugreek.

The overall conclusion from these reviews was that existing evidence for the efficacy of oral galactagogues is still weak and insufficient for creating guidelines, although some of them had relatively stronger recommendations about certain galactagogues (Bazzano 2016; Forinash 2012; Osadchy 2012; Zuppa 2010).

All of these other reviews found studies that we also identified, but the decision to include or exclude a particular study differed between reviews. For example, Petraglia 1985 was included in Osadchy 2012 but was excluded from our review because we could not find evidence that this was a RCT. On the other hand, we included Briton‐Medrano 2002 in our review while King 2013 excluded it because the intervention commenced prior to delivery.

Authors' conclusions

Implications for practice.

Due to extremely limited, very low certainty evidence, we do not know whether galactagogues have any effect on proportion of mothers who continued breastfeeding at 3, 4 and 6 months. However, there is very low‐certainty evidence that the oral galactagogues reported in this review might improve infant weight and milk volume in mothers breastfeeding their healthy term babies when compared to placebo or no intervention. We are very uncertain about the magnitude of this effect because of substantial heterogeneity of the studies, imprecision of measurement methods and incomplete reporting. We are uncertain if one galactagogue is more superior to another. With limited data available, we are uncertain if using an oral galactagogue would result in any harm.

Implications for research.

Taking into consideration the current widespread use of these substances, there is an urgent need for high‐quality randomised controlled trials (RCTs) on the efficacy and safety of galactagogues for mothers breastfeeding their healthy term infants. In considering future research, a set of core outcomes to standardize measurements, as well as a strong basis for the dosages and form used, could improve the certainty of the evidence. We highlight the following three areas to consider for future research.

Improvement in the quality of galactagogue studies

Galactagogue studies are complex and challenging to conduct due to many factors. Researchers might consider the following suggestions to improve the usability of future trials.

1. Utilization of the same rigor in the methodology and reporting of studies as required for RCTs involving a pharmacological intervention.

2. Provision of lactation support, preferably by a qualified lactation consultant, for every mother‐baby dyad.

3. Consider recruiting mothers with infants of similar ages to avoid difficulties in interpreting outcome measurements.

4. If a botanical galactagogue is used, it would be helpful to have a description of the part and purity of the plant(s) used, the preparation of the plant and the rationale for the test dosage.

5. When measuring milk production as an outcome, consider adding expressed residual milk from the breasts to the pre‐ and post‐breastfeeding infant weights for a more accurate picture of milk supply.

6. Report the total number of breastfeeds and/or milk expressions over 24 hours, because frequency of milk removal is a critical variable for assessing the efficacy of a galactagogue.

7. Consider reporting the duration of both 'any' and 'exclusive' breastfeeding to 6 months since an increase in one or both is the ultimate goal of treatment.

8. Attempt to identify, track and report potential side or adverse effects of the galactagogue to the mother and the infant. These include but are not limited to general symptoms such as breast engorgement or bowel changes and rashes in mother or infant, as well as effects specific to a particular substance, such as changes in blood sugar, body odour, appetite or weight. It could be useful to look for any long‐term effects of the galactagogues on the infants as well.

Priority for future galactagogue studies

1. Participants: women with insufficient milk production. For these women, consider identifying the aetiology of the low milk production because some galactagogues may work better under one condition than another.

2. Intervention: consider testing the more commonly used galactagogues first.

3. Intervention: consider testing multiple dosages to provide valuable clinical insights in determining the most effective therapeutic dosage.

Other related areas of research

1. Determining a standard for defining lactation insufficiency: currently, most studies rely on maternal perception of low milk production but this is very subjective.

2. Determining a standard method to measure the outcome 'milk volume,' including both measurement tools and duration of measurement. Measuring milk over a 24‐hour period is necessary for accuracy but may become burdensome and result in participant dropouts. More research on measuring milk volume over a shorter period of time to extrapolate 24‐hour milk volume is needed (Lai 2010).

3. Determining mechanisms by which a galactagogue may increase milk production. Data from animal studies may provide important clues. Our understanding of the mechanism of action could lead to more strategic choices of a particular galactagogue under different situations.

History

Protocol first published: Issue 4, 2015
Review first published: Issue 5, 2020

Appendices

Appendix 1. Additional search terms

HERDIN (The Health Research and Developmental Network ‐ Philippines) and Napralert (Natural Products Alert):

[Date of search 4 November 2019: 70 records from HERDIN, no records from Napralert]

galactagogue

galactagogue

lactogogue

lactagogue

lactogenic

galactagenic

galaktagog  (Turkish)

galactagoga (German)

laktagogon

laktagogum

galactagogic

galactogoguic

galactopoietic

galactokinetic

galactogenous

“promote lactation”

“stimulating lactation”

“stimulate lactation”

“enhancement of lactation”

ClinicalTrials.gov and the WHO International Clinical Trials Registry Platform (ICTRP) (4 November 2019)

galactogogue OR galactagogue OR lactogogue OR lactagogue

insufficient milk

lactational insufficiency

 

Appendix 2. Ingredients of galactagenic teas, soups, food, capsules and tablets

Cui Ru soup (催乳汤): one or two pork knuckles (猪蹄), soya bean (黄豆) 50 g and peanuts (花生) 50 g together with 5 herbs: bei qi (北芪) 30 g, dang sheng (党参) 15 g, dang gui (当归) 10 g, wang bu liu xing (王不留行) 20 g, and tong cao (通草) 12 g.

Galactagoue food from a hospital in Thailand: hot basil, lemon basil, sweet basil, banana blossom, garlic, garlic chives, ginger, pepper.

Humana Still Tee: hibiscus (Hibiscus tiliaceus): amber flower extract 2.6 g; fennel extract (Foeniculum vulgare): fennel 0.2 g; fennel oil: 0.02 g; Rooibos (Aspalatus linearis): red bush; Verbena Herb 0.2 g (Verbena officinalis): Mine flower; 0.2 g, Raspberry Leaves (Rubus idaeus): Raspberry, 0.2 g: Fenugreek (Trigonella foenum‐graecum): Fenugreek: 0.1 g and Goat's Rue Herb (Galega officinalis): Keçisedefi grass; 0.1 g.

Lactare: Shatavari (Asparagus racemosus), ashwagandha (Withania sominfera), licorice (Glycyrrhiza glabra), fenugreek (Trigonella foenum‐graecum), and garlic (Allium sativum).

Molocco: placental extract and vitamin B12.

Mu Er Wu You soup (母儿无忧汤): ren shen (人参) 10 g, huang qi (黄芪) 20 g, dang gui (当归) 15 g, mai men dong (麦门冬) 10 g, tian hua fen (天花粉) 10 g, chai mu (祡胡) 10 g, yi mu cao (益母草) 15 g, wang bu liu xing (王不留行) 10 g, and jie geng (桔梗) 10 g.

Sheng Ru He Ji (生乳合剂): zhu ti jia (猪蹄甲) 80 g, and wang bu liu xing (王不留行) 20 g.

Shirafza: fennel (Foeniculum volgare), anise (Pimpinella anisum), green cumin (Cuminumcyminum), dill (Nigella sativa), parsley (Anetom gravolen), and nigella (black seed) (Petroselinum crispum) via alcohol extraction.

Mother's Milk Tea: bitter fennel fruit 560 mg, anise fruit 350 mg, coriander fruit 210 mg, fenugreek seed 35 mg, blessed thistle herb 35 mg, propriety blend containing spearmint leaf, West Indian lemon grass, lemon verbena leaf and marshmallow root 560 mg.

Appendix 3. Overview of galactagogues in excluded studies

Pharmacological galactagogues
Intervention	Study ID
Arginine aspartate	Tagliareni 1977
Domperidone	Campbell‐Yeo 2007; De Leo 1986; Hofmeyr 1985; Ivanyi 2006; Knoppert 2013; Petraglia 1985
Growth hormone	Breier 1993; Milsom 1992; Milsom 1998
Lugol's iodine or iodine solution	Robinson 1947; Nicholson 1948; Dean 1950
Luteotropin, Sol. lugoli, hydrocortisonacetate, insulin and superlutin	Sapak 1969
Metoclopramide	Dastgerdi 2012; Ertl 1991; Guzman 1979; Kauppila 1981; Lewis 1980; Seema 1997; Rath 1983
Metoclopramide, domperidone and ferolactan	Rolfini 1989
Obron multivitamin	von Jaisle 1958
Orgametril	Gyõry 1968
Oxytocin	Douglas 1962; Erb 1968; Espenhain 1970; Friedman 1961; Huntingford 1961; Luhman 1963; Pontuch 1970; Ruis 1981; Thummel 1969
Oxytocin and sulpiride	Ylikorkala 1984
Oestrogen and progestagen	Toaff 1969
Pitocin	Volet 1965
Pseudoephedrine	Aljazaf 2003
Sulpiride	Aono 1979
Thyrotrophin‐releasing hormone	Peters 1991
Natural galactagogues
Intervention	Study ID
Chasteberry (Vitex agnus‐castus)	Amann 1966; Bautze 1953; Mohr 1954
Collagen soup	Zhang 1996
Fenugreek (Trigonella foenum‐graecum L)	Ahmed 2015; Hale 2009; Reeder 2011
Fenugreek (Trigonella foenum‐graecum L), garlic (Allium sativum), galactagogue mix	Srinivas 2014
Goat's rue (Galegran, derived from Galega officinalis)	Heiss 1968; Typl 1961
Goat's rue (Galega officinalis) and Silymarin (Silybum marianum)	Zecca 2016
Garlic (Allium sativum)	Mennella 1991; Mennella 1993
Glutamic acid	Vogulkina 1966
Hedge nettle (Stachys sylvatica)	Aronova 1977; Filippova 1975; Stegaĭlo 1980
Humana Still Tee	Kavurt 2013
Kyuki‐choketu granules ((Xiong‐gui‐tiao‐xue‐yin)	Narimatsu 2001; Ushiroyama 2007
Lactare: (shatavari (Asparagus racemosus), ashwagandha (Withania sominfera), licorice (Glycyrrhiza glabra), fenugreek (Trigonella foenum‐graecum); garlic (Allium sativum)	Geetha 1987; Ghosh 1986; Bakshi 1986; Rajarathnam 1986; Sholapurkar 1986; Subramaniam 1986
Leptaden: Jivanti (Leptadenia reticulata) and Kamboji (Breynia patens)	Akhtar 1972; Bhandari 1979; Deshpande 1962; Gupta 1966; Gokhale 1965; Lal 1980; Patel 1982; Tablb 1977
Moringa	Yabes‐Almirante 1996b
Maternal nutrition supplementation	Huynh 2016
Milk and eggs	Achalapong 2016
Motherlove Herbal's More MIlk Plus Alcohol Free	Demirci 2016
Mu‐ying‐le	Qi 1996
Oligoplex (Vitex agnus‐castus)	Janke 1941; Noack 1943
Pectin‐rich plant extract	Sepehri 1998
Placental extract (Moloco)	Keldenich 1976
Sesame	Zhu 2005
Shatavari (Asparagus racemosus)	Joglekar 1967
Torbangun (Coleus amboinicus L)	Damanik 2001; Damanik 2009
Various Japanese Kampo medicine	Kawakami 2003
Yangxueshengru oral liquor	Chen 1995
Mixed pharmacologicals and naturals
Intervention	Study ID
Metformin and fenugreek	Nommsen‐Rivers 2019
Metoclopramide, domperidone and moringa leaves	Co 2002
Others
Various interventions	Zhang 1987
Unclear what was the intervention used	Trivedi 1956
A galactagogue review	Tustanofsky 1996

Data and analyses

Comparison 1. Pharmacological oral galactagogues versus placebo or no treatment.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Infant weight (where they were only receiving own mother's milk) at the end of the study (grams)	1	20	Mean Difference (IV, Random, 95% CI)	23.00 [‐47.71, 93.71]
1.1.1 Metoclopramide	1	20	Mean Difference (IV, Random, 95% CI)	23.00 [‐47.71, 93.71]
1.2 Milk volume subgroup by type of galactagogue(please refer to footnotes for details on units used)	3	151	Mean Difference (IV, Random, 95% CI)	63.82 [25.91, 101.72]
1.2.1 Domperidone	1	45	Mean Difference (IV, Random, 95% CI)	99.90 [37.92, 161.88]
1.2.2 Metoclopramide	1	13	Mean Difference (IV, Random, 95% CI)	42.60 [13.02, 72.18]
1.2.3 Sulpiride	1	93	Mean Difference (IV, Random, 95% CI)	80.57 [‐4.55, 165.69]

Comparison 2. Natural oral galactagogues versus placebo or no treatment.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Infant weight subgroup by type of galactagogue (where they were only receiving own mother's milk) at the end of the study (grams)	3		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.1.1 Fennel	1	59	Mean Difference (IV, Random, 95% CI)	374.58 [‐274.45, 1023.61]
2.1.2 Fenugreek	1	58	Mean Difference (IV, Random, 95% CI)	369.48 [‐280.77, 1019.73]
2.1.3 Moringa	1	116	Mean Difference (IV, Random, 95% CI)	1342.00 [786.71, 1897.29]
2.1.4 Mixed botanical tea (Humana Still Tea)	1	42	Mean Difference (IV, Random, 95% CI)	594.00 [326.60, 861.40]
2.2 Milk volume subgroup by type of galactagogue (refer to footnotes for details on units used)	13		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.2.1 Bu Xue Sheng Ru (补血生乳)	1	30	Mean Difference (IV, Random, 95% CI)	154.00 [140.02, 167.98]
2.2.2 Chan Bao (产宝)	1	30	Mean Difference (IV, Random, 95% CI)	144.50 [134.55, 154.45]
2.2.3 Cui Ru (催乳汤) soup	1	108	Mean Difference (IV, Random, 95% CI)	26.60 [21.76, 31.44]
2.2.4 Banana flower	1	58	Mean Difference (IV, Random, 95% CI)	93.70 [14.20, 173.20]
2.2.5 Fenugreek	1	37	Mean Difference (IV, Random, 95% CI)	15.30 [6.93, 23.67]
2.2.6 Ginger	1	63	Mean Difference (IV, Random, 95% CI)	56.00 [22.99, 89.01]
2.2.7 Moringa	2	135	Mean Difference (IV, Random, 95% CI)	132.59 [‐88.10, 353.28]
2.2.8 Mixed fenugreek, ginger, turmeric	1	50	Mean Difference (IV, Random, 95% CI)	503.00 [360.81, 645.19]
2.2.9 Ixbut	1	34	Mean Difference (IV, Random, 95% CI)	6.99 [1.27, 12.71]
2.2.10 Mixed botanical tea (Humana Still‐tea)	1	44	Mean Difference (IV, Random, 95% CI)	34.40 [11.03, 57.77]
2.2.11 Sheng Ru He Ji (生乳合剂) oral liquid	1	200	Mean Difference (IV, Random, 95% CI)	17.97 [15.75, 20.19]
2.2.12 Silymarin (milk thistle)	1	50	Mean Difference (IV, Random, 95% CI)	418.68 [359.77, 477.59]
2.2.13 Xian Tong Ru (先通乳) soup	1	85	Mean Difference (IV, Random, 95% CI)	37.70 [26.27, 49.13]
2.2.14 Palm dates	1	38	Mean Difference (IV, Random, 95% CI)	32.10 [23.81, 40.39]
2.3 Volume of supplement beyond mother's own milk (mL)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
2.3.1 Gossypium herbaceum L	1	45	Mean Difference (IV, Random, 95% CI)	‐186.70 [‐267.24, ‐106.16]
2.3.2 Shatavari (Asparagus racemosus)	1	64	Mean Difference (IV, Random, 95% CI)	‐46.80 [‐158.81, 65.21]
2.4 Quality of life using WHO QOL Scale	1	60	Mean Difference (IV, Fixed, 95% CI)	‐0.01 [‐3.84, 3.82]
2.4.1 Mother's Milk Tea	1	60	Mean Difference (IV, Fixed, 95% CI)	‐0.01 [‐3.84, 3.82]
2.5 Breastfeeding self‐efficacy	1	60	Mean Difference (IV, Fixed, 95% CI)	‐1.74 [‐4.47, 0.99]
2.5.1 Mother's Milk Tea	1	60	Mean Difference (IV, Fixed, 95% CI)	‐1.74 [‐4.47, 0.99]
2.6 Postpartum Depression Scale	1	60	Mean Difference (IV, Fixed, 95% CI)	1.38 [‐0.21, 2.97]
2.6.1 Mother's Milk Tea	1	60	Mean Difference (IV, Fixed, 95% CI)	1.38 [‐0.21, 2.97]

Comparison 3. Oral galactagogue versus another galactagogue: infant weight (grams).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Fenugreek tea versus Fennel tea	1	78	Mean Difference (IV, Fixed, 95% CI)	‐10.25 [‐462.91, 442.41]

Comparison 4. Oral galactagogue versus another galactagogue: milk volume (mL).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
4.1 Chanbao oral liquid versus Bu Xue Sheng Ru capsules	1	40	Mean Difference (IV, Fixed, 95% CI)	‐9.50 [‐25.65, 6.65]
4.2 Domperidone versus Moringa leave capsules	1	26	Mean Difference (IV, Fixed, 95% CI)	‐0.38 [‐10.64, 9.88]
4.3 Fenugreek versus Palm dates	1	50	Mean Difference (IV, Fixed, 95% CI)	‐16.80 [‐27.22, ‐6.38]
4.4 Fenugreek capsules versus Torbangun soup	1	45	Mean Difference (IV, Random, 95% CI)	‐78.40 [‐188.83, 32.03]
4.5 Fenugreek capsules versus Moloco tablets	1	44	Mean Difference (IV, Fixed, 95% CI)	15.20 [‐108.08, 138.48]
4.6 Mu Er Wu You soup versus Kun Yuan Tong Ru soup	1	90	Mean Difference (IV, Fixed, 95% CI)	19.95 [6.88, 33.02]
4.7 Torbangun soup versus Moloco tablets	1	45	Mean Difference (IV, Fixed, 95% CI)	93.60 [‐12.39, 199.59]

Comparison 5. Oral galactagogue versus another galactagogue: volume of supplementary feeds (mL).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
5.1 Chanbao oral liquid versus Bu Xue Sheng Ru capsules	1	40	Mean Difference (IV, Fixed, 95% CI)	‐2.50 [‐8.45, 3.45]
5.2 Mu Er Wu You soup versus Kun Yuan Tong Ru soup	1	90	Mean Difference (IV, Fixed, 95% CI)	‐12.25 [‐15.63, ‐8.87]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Aono 1982.

Study characteristics
Methods	Quasi‐randomised controlled trial in Japan Trial dates: not mentioned
Participants	96 healthy mothers with low milk production and their infants. 44% of mothers were primiparous. Age of infants at start of study: third day of life. The average birthweight was 3222 grams. Inclusion criteria: "Healthy mothers, aged 21 to 35 years, who had completed an uncomplicated term gestation with normal delivery, and who had poor secretion of milk (total yield of milk not exceeding 50 mL for the first three postpartum days). The infants were term, at least three days old." Exclusion criteria: none mentioned Breastfeeding method: timed feeding ("six times a day for approximately 20 minutes on a 3‐hour schedule from 0630 hour"). The breasts were emptied after feeding by manual or pump expression. Supplemental feeding was allowed throughout the study for both groups.
Interventions	Arm 1: oral sulpiride 50 mg (Dogmatyl, Delagrange‐Fujisawa) twice a day for 4 days, starting from third postpartum day (n = 48) Arm 2: placebo "given in similar manner" (n = 48)
Outcomes	Mean milk volume per day on first to third day of intervention, measured by weighing infants before and after nursing and adding this to the volume of milk expressed until flow almost ceased. Milk expression was done either manually or with a breast pump. Mean serum prolactin level on day of delivery and up to the sixth day post‐delivery Proportion of mothers with complete breastfeeding, breastfeeding plus bottle feeding and complete bottle feeding at 1 month after delivery
Funding and Declaration of interest	The prolactin radioimmunoassay kits were given by National Institute of Arthritis, Metabolism and Digestive Diseases, Bethesda, Maryland, USA (NIAMDD). No other funding or declaration of interest statement found
Notes	No contact details of authors available for clarifications The authors had reported results for primipara mothers separately from multipara mothers. For milk volume, we combined the 2 groups together using the method suggested in the Cochrane Handbook section 7.7.3.8 (Higgins 2011), for entry into Review Manager. The standard deviation was derived from the standard error reported by the authors using the RevMan calculator (Review Manager 2014). The authors had reported the proportion of mothers breastfeeding at one month as a graph and the numbers entered into RevMan were estimated from the graph.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	"Mothers were alternately selected for the sulpiride or placebo groups according to the delivery week."
Allocation concealment (selection bias)	High risk	No concealment done. Allocation was according to the delivery week.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Blinding was unlikely for personnel because the allocation was done according to the delivery week. A placebo was "given with the same schedule" but we were unable to judge whether the placebo was identical with the intervention treatment.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	Blinding of the outcome assessor (mother) was not reported.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Unclear risk	There was a lack of description of the placebo.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	There was no description of the flow of participants, so we do not know what happened to all participants at the end of the study.
Selective reporting (reporting bias)	Low risk	All major outcomes stated in the methods were reported in the results.
Other bias	Low risk	None detected

Balahibo 2002.

Study characteristics
Methods	4‐arm randomised control trial in the Philippines Trial dates: June 2000 to January 2001
Participants	60 healthy mothers and their infants. 50% of mothers were primiparous and 68% had normal vaginal delivery. Age of infants at start of study: first day of life Inclusion criteria: women with "healthy infants of 38 to 42 weeks gestation weighing 2500 to 5000 g who agreed not to feed their infants any milk formula for two months starting from date of birth, will not give their infants solids or semi solid food for the duration of the study and were healthy throughout their pregnancy and the study period." Exclusion criteria: smokers, alcoholics, taking medication which could affect foetal growth and development during pregnancy Breastfeeding method: the authors did not mention if the infants were allowed to breastfeed on demand. However, no supplemental feeding was allowed during the study
Interventions	Arm 1: moringa leaves capsule 250 g (Natalac) once a day for 8 weeks (n = unknown) Arm 2: moringa leaves capsule 250 g (Natalac) twice a day for 8 weeks (n = unknown) Arm 3: placebo once a day for 8 weeks (n = unknown) Arm 4: placebo twice a day for 8 weeks (n = unknown)
Outcomes	Mean percentage infant weight gain at second, fourth, sixth and eighth weeks of intervention. Mean percentage infant length increase at second, fourth, sixth and eighth weeks of intervention.
Funding and Declaration of interest	No funding or declaration of interest statement found.
Notes	This study was reported in 3 different publications. Contact was made with author Zea Baldovino, but there was no response to our specific questions.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	How this was done was not mentioned.
Allocation concealment (selection bias)	Unclear risk	How this was done was not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	A placebo was given with the same schedule, but we were unable to judge whether the placebo was identical with the intervention treatment.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	Milk volume was not a part of this study.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	Self‐reported outcomes were not a part of this study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	"The doctor and students who took measurements were blinded as to the contents of the capsules for each trial group, and as to which trial group the respondents belonged."
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	There was no description of the flow of participants so we do not know what happened to all participants at the end of the study.
Selective reporting (reporting bias)	Low risk	All major outcomes stated in the methods were reported in the results.
Other bias	Unclear risk	Only demographics of all participants were reported. No baseline information was reported for the 4 groups.

Barguno 1988.

Study characteristics
Methods	Randomised controlled trial, country not specified Trial dates: not mentioned
Participants	66 mothers and their infants Age of infants at start of study: first day of life. Birthweight ranged from 2700 g to 4000 g. Inclusion criteria: primiparous women aged 18 to 35 years from the same socioeconomic milieu with normal pregnancies, and had given birth to normal infants, weighing 2700 g to 4000 g. All expressed their wish to breastfeed their infants. Exclusion criteria: no previous history of abortion Breastfeeding method: timed feeding ("Daily scheduled of 6 nursing episodes of 30‐minute maximal duration without bottle supplements")
Interventions	Arm 1: oral sulpiride 100 mg (Tepavil) 3 times a day for 4 days then 50 mg 3 times a day for next 86 days (n = 34) Arm 2: placebo with "similar appearance given to control group" (n = 32)
Outcomes	Mean infant weight gain on fourth to 15th, 15th to 30th, 30th to 60th and 60th to 90th post delivery days Mean basal and 30‐minute post‐feeding plasma prolactin level on first, fourth, 15th, 30th, 60th and 90th post‐delivery days Concentration of sulpiride in breast milk
Funding and Declaration of interest	No funding or declaration of interest statement found
Notes	Attempts to contact authors for clarifications failed. Mean infant weight gain was presented as a bar chart with standard error.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	How this was done was not described, although the authors stated that "66 women were randomly assigned to a control group of 32 women receiving placebo or another group of 34 women treated with sulpiride."
Allocation concealment (selection bias)	Unclear risk	How this was done was not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Placebo tablets of similar appearance was supplied to the control group."
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	Milk volume was not a part of this study.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	Self‐reported outcomes were not a part of this study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Quote: "Placebo tablets of similar appearance were supplied to the control group."
Incomplete outcome data (attrition bias) All outcomes	High risk	Of the 66 women enrolled, 11 (34%) were excluded from the control group and 14 (41%) from the sulpiride group. The authors mentioned that "The exclusions were due to failure to follow protocol specifications, intercurrent illnesses interfering with normal lactation and shortening the period of lactation due to work reasons." Comment: although the attrition was fairly evenly distributed between the 2 groups, we judged that the attrition rate was high enough to affect infant weight gain. In addition, the reasons for attrition in each group were not given and we could not judge if they were evenly distributed across the groups.
Selective reporting (reporting bias)	Low risk	All major outcomes stated in the methods were reported in the results.
Other bias	Unclear risk	No baseline information was reported for the 2 groups.

Briton‐Medrano 2002.

Study characteristics
Methods	Randomised controlled trial in the Philippines Trial dates: January 2004 to November 2004
Participants	Healthy mothers and their infants (the mothers were given the intervention during pregnancy until the delivery of their infants). The number of mothers recruited and randomised is unclear. 53 mothers were included in the analysis. The average parity of the mothers was 1.8. Age of infants at start of study: not born yet, but at least 35 weeks' gestation (average gestation is 36 weeks). Almost all of the infants were delivered at term. Inclusion criteria: "All pregnant women at least 35 weeks of gestation with regular prenatal checkups..." Exclusion criteria: "Pregnant women gravida four or more with no history of abortion or still births, with maternal conditions that would have an effect on breast milk production and contraindications to breastfeeding (e.g. advanced pulmonary tuberculosis, extrapulmonary spread of tuberculosis, rubella, renal problems, fever, retracted nipples, anaemia and pneumonia)." Breastfeeding method: direct breastfeeding was not allowed but how the infants were fed was not clearly described. Mothers expressed their milk at the sixth hour after birth and subsequently every 4 hours for 2 days. There was no mention if supplementation with formula milk was allowed.
Interventions	Arm 1: moringa leaf capsules (Prolacta) 700 mg 3 times a day from time of recruitment until delivery (n = 27) Arm 2: placebo (similar appearance) given in a similar manner (n = 26)
Outcomes	Onset of breast milk production; measured as the time to the 'first milk drip,' time to 'significant amount' (10 mL or more) and time to 'adequate amount' (30 mL or more) Mean volume of milk expressed (per 4 hours) on first to second day after delivery; using battery operated or electric breast pump Adverse effects to the mothers
Funding and Declaration of interest	No funding or declaration of interest statement found
Notes	Intervention was stopped once the infant was delivered. All outcomes were measured after intervention was stopped. Attempts to contact authors for clarifications failed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization using the table of random numbers was prepared prior to the start of the study."
Allocation concealment (selection bias)	Low risk	"An assigned third party gave each subject a sealed envelope which had been arranged according to the prepared randomisation sequence."
Blinding of participants and personnel (performance bias) All outcomes	Low risk	"Placebos provided had the same appearance as the moringa capsules."
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	"Placebos provided had the same appearance as the moringa capsules."
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	"Placebos provided had the same appearance as the moringa capsules."
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	High risk	The authors reported that 28 out of the 82 enrolled mothers dropped out of the study. The reasons for dropping out were given as "11 did not deliver at OSMAK, 14 had inadequate intake, one delivered a baby with poor APGAR score secondary to perinatal asphyxia, one refused breast pumping and one had a congenital anomaly (hydrocephalus, meningocoele and spina bifida)." However, we do not know which group the dropouts were from. Furthermore, the number of participants reported in the baseline characteristics and outcomes do not tally ‐ we could not account for a further two missing participants.
Selective reporting (reporting bias)	Low risk	All expected outcomes were reported.
Other bias	Low risk	None detected

Bumrungpert 2018.

Study characteristics
Methods	Randomised control trial in Thailand Trial dates: not mentioned
Participants	50 exclusively breastfeeding mothers and their infants Age of infants at start of study: 1 month of life Inclusion criteria: aged 20 to 40 years, 1 month postpartum with exclusive breastfeeding, willing to participate Exclusion criteria: having chronic disease, using a galactagogue herb or medicine, smoking, drinking, twins, separated from their infants Breastfeeding method: mothers were exclusively breastfeeding but most likely only expressed milk feeding during the intervention period
Interventions	Arm 1: Galactagogue herbal medicine (200 mg fenugreek seed, 120 mg ginger, and 100 mg turmeric per capsule) given three times per day before meals for 4 weeks (n = 25) Arm 2: Placebo corn starch capsules (n = 25)
Outcomes	General characteristics and blood chemistry, including age, weight, body mass index, blood pressure, heart rate, haemoglobin, haematocrit, blood glucose, total cholesterol, triglyceride, low density lipoprotein‐cholesterol, high density lipoprotein‐cholesterol, aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, blood urea nitrogen, creatinine and albumin Dietary intake Energy and nutrient content of milk samples Milk volume Adverse effects in mothers and infants
Funding and Declaration of interest	The authors declared “No competing financial interests exist." They thanked Herbal Acharn's Home (Thailand) Co. Ltd. that provided the Fenucaps and placebo "and the partial support publication by the China Medical Board (CMB), Faculty of Public Health, Mahidol University, Bangkok, Thailand.”
Notes	Further clarification about the study was needed, but attempts to contact the authors failed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"The random allocation sequence was provided by an independent consultant and was computer generated using a randomization plan from www.randomization.com with randomization in blocks of 10. A list of consecutive study numbers was generated.”
Allocation concealment (selection bias)	Low risk	“Herb supplement groups were allocated by research assistant, but the allocation was concealed by assigning each participant with a unique number.”
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	The taste of fenugreek, ginger and tumeric may be detected, however it is less likely considering they were in a capsule.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	Details of milk expression not described. Therefore, it is unclear how much lack of blinding would affect this outcome.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	Self‐reported outcomes were not a part of this study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were included in the analysis.
Selective reporting (reporting bias)	Low risk	All expected outcomes were reported in the results.
Other bias	Low risk	None detected

Chan 2005.

Study characteristics
Methods	Randomised controlled trial in Guatemala Trial dates: March 2003 to September 2003
Participants	34 healthy mothers with lactation insufficiency and their infants. The average parity of mothers was 5 (range 1 to 11). Age of infants at start of study: 30th to 90th day of life; "in the second and third postpartum month" Inclusion criteria: healthy breastfeeding mothers "during the second and third postpartum month" with term infants, normal delivery, exclusive or mixed breastfeeding with hypogalactia, had not consumed Ixbut before and permanent residents of communities in the study site. Breastfeeding method: this was not described, but the mothers had to go to the Hospital of Coatepeque for 6 consecutive days before initiation of intervention and 6 consecutive days after 3 days of intervention, between 0800 and 1000 hours, with instructions to avoid breastfeeding the baby for at least 1 hour prior to the collection of breast milk samples. Supplemental feeding was allowed.
Interventions	Arm 1: Ixbut (Euphorbia lancifolia) infusion (20 fresh leaves infused in a litre of water) taken once a day for 3 days (n = 17) Arm 2: placebo (green and yellow vegetable coloured water), given in a similar manner (n = 17)
Outcomes	Mean change in volume of the milk expressed before and after intervention. Milk was expressed at 10 am daily for 3 consecutive days prior to intervention and repeated for another 3 consecutive days after completing the intervention. Composition of breast milk before and after intervention Mother's perception of milk production (faster milk flow, breast fullness, thickness and whiteness of milk, infants' satisfaction)
Funding and Declaration of interest	No funding or declaration of interest statement found
Notes	The original paper is an unpublished thesis written in Spanish. The above information was obtained from the English translation of the article. Attempts to locate the author were unsuccessful.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"Participants were randomly divided into two groups." Comment: no further description was available.
Allocation concealment (selection bias)	Unclear risk	How this was done was not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Although the placebo (green and yellow vegetable coloured water) might have looked like the ixbut, it would not have the distinct taste and smell.
Blinding of outcome assessment (detection bias) Milk volume outcomes	High risk	The outcome assessors (mothers) were not blinded.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	High risk	Lack of blinding could have affected the mothers' perception of milk production.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	There was no description of the flow of participants, so we do not know what happened to all participants at the end of the study.
Selective reporting (reporting bias)	Low risk	All expected outcomes were reported in the results.
Other bias	Low risk	The baseline weight of the mothers in the control group was more than the intervention group. However, the body mass indexes were not different between both groups. Therefore we judge this as low risk.

Damanik 2006.

Study characteristics
Methods	3‐arm, randomised controlled trial in Indonesia Trial dates: not mentioned
Participants	75 healthy mothers and their infants Age of infants at start of study: second day of life Inclusion criteria: women "aged between 20 and 40 years in their last trimester of pregnancy, apparently healthy, have no symptoms of malnutrition or chronic diseases, not take any medication on regular basis or have no medical conditions or complications during previous pregnancies or deliveries and intend to breast feed their infants exclusively for at least 4 months." The infants must be "healthy term (gestation of 37 to 43 weeks) infants and have a birth weight at least 2.5 kg." Exclusion criteria: regular smoking or alcohol intake Breastfeeding method: not described if feeding was timed or untimed. However, as per the inclusion criteria, no supplemental feeding was allowed.
Interventions	Arm 1: Torbagun leaves (Colleus amboinicus leaves) 150 g daily for 6 days a week as a soup for 30 days (n = 23) Arm 2: Fenugreek capsule 600 mg three times a day for 30 days (n = 22) Arm 3: "Placebo." Sugar‐coated vitamin B12 and placental extract tablets (Moloco‐B12) 1 tablet 3 times a day for 30 days (n = 22) Duration of intervention was for 30 days for all participants, but the study continued up until 60 days.
Outcomes	Mean milk volume per day on 14th, 28th, 42nd and 56th day post‐delivery, measured by calculating the difference in the baby's weight before and after feed and multiplying with 0.983 to convert into millilitres Mean percentage of change in milk volume on 14th, 28th, 42nd and 56th day post‐delivery Composition of breast milk (micro and macro nutrient) Adverse effects (not specified if for mother or infant)
Funding and Declaration of interest	No funding or declaration of interest statement found. Correspondance with the author revealed that the Australian International Development and Assistance Bureau (AusAID) and the Indonesian government, especially, local government in Simalungun District, North Sumatera Province Indonesia provided research grant for this study.
Notes	"Placebo" contained Molocco, a placental extract sometimes considered lactogenic (Hammett 1918; Keldenich 1976; Soykova‐Pachnerova 1954)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Participants were "randomly assigned to one of the three groups ..." Correspondence with the authors revealed that they used the lottery method to assign the participants.
Allocation concealment (selection bias)	High risk	Not mentioned in the paper but correspondence with the author revealed that no allocation concealment was done.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Torbagun tea was in the form of a soup, fenugreek was given as capsules and the vitamin B12 was given as a tablet.
Blinding of outcome assessment (detection bias) Milk volume outcomes	High risk	There was no blinding of the outcome assessor (trained research assistant) (correspondence with author).
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	Self‐reported outcomes were not a part of this study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	A total of 8 participants dropped out and were excluded from the analysis. Correspondence with the author revealed that 2 were from the Torbangun group (1 participant refused to have blood samples taken and 1 participant moved to another village outside the study area), 3 were from the fenugreek group (2 participants refused to have blood samples taken and 1 participant moved to another village outside the study area), 3 were from the Moloco (B12) group (2 participants delivered low‐weight infants and 1 participant refused to have blood samples taken).
Selective reporting (reporting bias)	High risk	Adverse effects were not reported, although the authors stated that "Structured conversation between the subjects and researchers during visits provided information about the general health status of the subjects ....Any complaints or concerns were also recorded."
Other bias	Unclear risk	No baseline information was reported for the 3 groups.

De Gezelle 1983.

Study characteristics
Methods	Randomised controlled trial in Belgium Trial dates: not mentioned
Participants	13 healthy mothers and their infants Age of infants at start of study: first day of life Inclusion criteria: "Healthy nursing primiparas with normal infants" Exclusion criteria: none mentioned Breastfeeding method: timed feeding ("Three hour schedule basis starting at 0630 hours"). No mention if supplemental feeding was allowed.
Interventions	Arm 1: oral metoclopramide 10 mg 3 times a day for 8 days (n = 7) Arm 2: placebo. No description available (n = 6)
Outcomes	Mean milk volume of second daily feeding, measured by weighing infants before and after feed on third to eighth day of life; breasts were not emptied after feeds Mean serum prolactin level on third until 28th day of life Composition of breast milk (fat and amino acid concentrations) "Two minutes after the onset of the second daily feed, 10 mL of milk sample was obtained by breast pump on fourth, sixth, eighth, 14th, 21st and 28th day." Adverse effects for the mothers and infants.
Funding and Declaration of interest	No funding or declaration of interest statement found
Notes	Further clarification about the study was needed, but attempts to contact the authors failed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The authors reported that this was a placebo controlled double blind study. "... were randomly selected." No further elaboration on how the mothers were assigned to groups
Allocation concealment (selection bias)	Unclear risk	How this was done was not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	The nature of the placebo was not described, but the authors stated that (quote) "Neither the mother or the nursing staff were aware of the nature of the tablets."
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	The nature of the placebo was not described, but the authors stated that (quote) "Neither the mother or the nursing staff were aware of the nature of the tablets."
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	The nature of the placebo was not described, but the authors stated that (quote) "Neither the mother or the nursing staff were aware of the nature of the tablets."
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	The number of participants analysed for the outcomes is unclear.
Selective reporting (reporting bias)	Low risk	All major outcomes stated in the methods were reported.
Other bias	Unclear risk	No baseline information was reported for the 2 groups.

Di Pierro 2008.

Study characteristics
Methods	Randomised controlled trial in Peru Trial dates: not mentioned
Participants	50 mothers with less than expected milk production and their infants Age of infants at start of study: this was not reported in the paper but correspondence with the author revealed that the average age of the infants was 4 months. Inclusion criteria: healthy breastfeeding mothers who had borderline milk production (< 700 mL/day) Exclusion criteria: mothers with anomalies or diseases that can affect breastfeeding Breastfeeding method: not stated if feeding was timed or untimed. The breasts were emptied after feeding by pump expression. Correspondence with the author revealed that there was use of supplemental feeding in the first week of study for 5 infants in the control group and supplemental feeding for an average of 3 days for 3 infants in the treatment group.
Interventions	Arm 1: oral micronized silymarin (Silybum marianum) (BIO‐C) 420 mg daily for 63 days (n = 25) Arm 2: placebo, "undistinguishable from the active." No further description available (n = 25)
Outcomes	Mean milk volume per day, measured by weighing the baby before and after sucking and the milk expressed with breast pump after each sucking to void the gland at baseline, 30th and 63rd day of intervention Percentage of increase in milk volume from baseline to 63th day of intervention Composition of breast milk (water, fats, carbohydrates and proteins)
Funding and Declaration of interest	"Silymarin and placebo were obtained with the support from Indena S.p. A, S.I.I.T. s.r.l, Trezzanno S/N and Milte S.p.A (Milano, Italy)." Through correspondence, the first author Francesco di Pierro stated that she is the scientific and research director of Velleja Research, which is a biopharmaceutical research company which does not market any products. She is also the editor‐in‐chief of Nutrafoods (Springer). She informed us that "Milte Italia supported the study with breast pumps." No money was given to the physicians and/or to the midwifes involved.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Participants were divided into two groups .....considering their age, number of sons and age of their last born." Communication with author revealed that the division was done by toss of coin. "We first used the coin to decide if woman number 1 was in group A or B, then we select a similar woman to put for the opposite group. So we did that for 25 times. At the end we tossed the last coin to decide if group A or group B was treatment or control."
Allocation concealment (selection bias)	Low risk	Based on the above communication, given that the last coin was used to determine whether Group A was treatment or control, we think allocation concealment is adequate.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Placebo was "indistinguishable from the active."
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	Placebo was "indistinguishable from the active."
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	Placebo was "indistinguishable from the active."
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	We judged this as low risk because the authors stated that "During the study, no single drop out was recorded in both groups," and correspondence with the author confirmed that all participants were analysed.
Selective reporting (reporting bias)	Low risk	All expected outcomes were reported in the results.
Other bias	Low risk	None detected

Espinosa‐Kuo 2005.

Study characteristics
Methods	Randomised controlled trial in the Phillipines Trial dates: January 2003 to October 2003
Participants	82 healthy mothers and infants. The median parity of mothers was 2 to 2.1 Age of infants at start of study: third day of life. The average gestation at birth was 39 weeks with an average birthweight of 3048 grams. Inclusion criteria: women aged 18 to 38 years who delivered term infants via vaginal delivery and who will breastfeed their infants. These mothers lived within a 5 kilometre radius from the recruitment centre. Exclusion criteria: presence of hypertension, diabetes mellitus, chorioamnionitis, breast abnormalities, chronic illness, acute illness such as upper respiratory tract infection or urinary tract infection, mothers taking medication on regular basis except multivitamins and iron supplements; and infants with illness or congenital anomalies Breastfeeding method: it was not clearly reported if the mothers were allowed to directly breastfeed their infants. Mothers were given breast pumps and instructed to express milk every 4 hours for at least 5 minutes. It was not clear if any breastfeeding occurred before or after the expression. There was no mention if supplemental feeding was allowed.
Interventions	Arm 1: moringa leaves capsule (Prolacta) 700 mg a day for 8 days starting from third day post delivery (n = 41) Arm 2: placebo (flour containing capsule) given in a similar manner (n = 41)
Outcomes	Mean milk volume expressed per day from third to 10th day of intervention, as measured by using a pump on each breast every 4 hours for at least 5 minutes Adverse effects for mothers
Funding and Declaration of interest	No funding or declaration of interest statement found.
Notes	Attempts to contact the authors failed. Quoted percentages for dropouts were incorrect because the wrong denominator was used (total recruitment numbers instead of allocation group numbers).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No random sequence generation was described. "The researcher randomly assigned subject to the treatment and placebo groups."
Allocation concealment (selection bias)	Unclear risk	How this was done was not reported.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	"...those who belonged to the placebo group were given flour‐containing capsules in identical containers prepared by a pharmaceutical company." Note: we were unable to judge whether the placebo capsules were identical with the intervention capsules.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	Blinding or the outcome assessor (mother) was not reported.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Unclear risk	We judge this to be unclear risk of bias because it was unclear whether the mothers were blinded, and they were the ones who reported adverse effects.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	The authors reported that 6 mothers dropped out from the intervention group due to "failure to take moringa capsules even on just 1 occasion during treatment course or failure to comply with procedures, such as completing data entry in the notebook provided or the failure and improper use of breast pump" and 3 mothers dropped out from the placebo group because they "did not take moringa capsules on schedule." However, we were unclear how many participants were analysed in the results.
Selective reporting (reporting bias)	Low risk	All expected outcomes were reported in the results.
Other bias	Low risk	None detected

Fang 2003.

Study characteristics
Methods	Single centre, quasi‐randomised trial in China Trial dates: October 2001 to October 2002
Participants	110 mothers with lactation insufficiency and their infants Age of infants at start of study: this was not reported Inclusion criteria: ".....mothers aged 21 to 34 years.....with any of these 6 problems after delivery: produced little to no breast milk for feeding; could not produce milk easily and steadily; experienced pain in their chest and ribs; swollen and tender breasts; poor appetite; mild depression." Exclusion criteria: ".... those with endocrine disorders, vascular diseases, malignant tumours, liver, kidney as well as haematological diseases." "Those that did not follow the protocol" were later excluded from the analysis. Breastfeeding method: this was not described and there was no mention if supplemental feeding was allowed.
Interventions	Arm 1: Ruquan‐Chongji (乳泉冲剂) 15 g twice a day for 3 days (n = 80) Ruquan‐Chongji (Registration number: WS3‐B‐1362‐93) is a herbal mixture containing Cowherb Seed (Semen Baccariae) 210 g, Squama mantitis 25 g, Radix tricho santhis 90 g, Liquorice 90 g, Radix angelicae sinensis, Angelica sinensis (Oliv)Diels) 150 g, Radix rhapontici 90 g Arm 2: Shengruzhi (生乳汁) 100 mL bottle twice a day for 3 days (n = 30) Shengruzhi (Registration number: WS3‐B‐0528‐9) is a herbal mixture containing Radix Angelica sinensis (Oliv)Diels) 35 g, Radix reh manniae 25 g, Milk beteh (astragalus root) 5 g, Dangshen (radix codonopsitis 5 g, Ningpo figwort 25 g, Ophiopogon root 5 g, Squama manitis 15 g, Anemarrhena asphodeloids Bge 10 g
Outcomes	Change in the 'Overall symptom score' after intervention Note: the "Overall Symptom Score" is based on the six symptoms listed in the inclusion criteria i.e. produced little to no breast milk for feeding could not produce milk easily and steadily (experienced pain in their chest and ribs swollen and tender breasts poor appetite mild depression A score of '1,' '2' or '3' was given to each symptom corresponding with 'mild,' 'moderate' and 'severe', respectively. All six scores were then added to form the 'Overall symptom score' ranging 6‐18, which was then categorized in severity: ≥ 18: severe problem 10 to 17: moderately severe problem < 10: mild problem The change in the 'Overall symptom score' before and after intervention was tabulated and categorized into one of four categories: 'recovery' (90% reduction in symptoms), 'significant effect' (70% to 89% reduction in symptoms), 'effective' (30% to 69% reduction in symptoms), 'no effect' (less than 30% change in symptoms).
Funding and Declaration of interest	No funding or declaration of interest statement found.
Notes	This study was published in Chinese. The above information was obtained from the English translation of the original paper. No contact details of authors were available for clarifications. We also could not contact the translator for clarification.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	How this was done was not mentioned.
Allocation concealment (selection bias)	Unclear risk	How this was done was not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding was not possible due to the difference in the nature of the treatment (prepared and package in sachets) and placebo (prepared in the form of liquid in bottles).
Blinding of outcome assessment (detection bias) Milk volume outcomes	High risk	Mother's perception of milk volume was recorded, hence lack of blinding could have influenced the results.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	High risk	All outcomes in this study were reported by the mothers, hence lack of blinding could have influenced the results.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were included in the analysis.
Selective reporting (reporting bias)	Low risk	All outcomes mentioned in the methods were reported in the results.
Other bias	Low risk	None detected.

Ghasemi 2018.

Study characteristics
Methods	3‐arm randomised controlled trial in Iran Trial dates: not mentioned
Participants	117 healthy mothers with lactation insufficiency and their baby girls Age of infants at start of study: at birth until 4 months of life. Mean age 69 days Inclusion criteria: "Girl infants aged 0‐4 months old, term, birth weight between 2500‐4000 g, normal ability of sucking." Exclusion criteria: infants given infant formula or complementary feeding, mothers on herbal and chemical galactagogues, mother and infant with HIV infection, addiction to narcotic substances and alcohol, untreated active tuberculosis, using medication such as Phenobarbital and Ergotamine, women receiving breast cancer treatment, women with breast problems, such as inverted nipples, breast abscess, mastitis, and underlying diseases such as asthma, cardiac diseases, bleeding disorders and diabetes Breastfeeding method: breastfeeding on demand and no supplementary feeding allowed
Interventions	Arm 1: fennel seed (Foeniculum vulgare) powder 7.5 g in black tea 3 times a day for 4 weeks (n = 39) (this arm was reported in Ghasemi (Fennel) 2014) Arm 2: fenugreek seed (Trigonella foenum‐graecum) powder 7.5 g in black tea 3 times a day for 4 weeks (n = 39) (this arm was reported in Ghasemi (Fenugreek) 2015) Arm 3: placebo was black tea (3 g) 3 times a day for 4 weeks (n = 39)
Outcomes	Mean infant weight Mean infant length Mean infant head circumference Number of wet diapers per day Frequency of defecation per day Frequency of infant feeding per day All outcomes were measured weekly from the first until fourth week.
Funding and Declaration of interest	This study was funded by Tehran University of Medical Sciences. The herbal tea was produced by the Iranian Institute of Medicinal Plants (AECR). No declaration of interest statement found
Notes	This was a three‐arm study (fenugreek, fennel and placebo) published in four different papers (as fenugreek versus placebo in one paper; as fennel versus placebo in another paper; as a three‐arm study in the third paper (published in English) and as a three‐arm study in the fourth paper (published in Persian). Correspondence with the author confirmed that all referred to the same 3‐arm study.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The participants were divided into 2 groups by lottery method. Cards numbered "1" and "2" were drawn to determine which group the mothers would be assigned to.
Allocation concealment (selection bias)	Unclear risk	It was not stated whether the cards were concealed or if they could be changed.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Fennel and fenugreek are aromatic herbs. Their smell and taste are distinct. Therefore it its likely that participants would be able to guess what they were drinking.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	Milk volume was not a part of this study.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	The self‐reported outcomes in this study were not outcomes of this review.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	It was not reported in the paper but correspondence with the author revealed that the study personnel who measured the infant's weight was blinded to what the participant was receiving.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All 117 participants recruited were included in the analysis.
Selective reporting (reporting bias)	Low risk	All the outcomes stated in the methods were reported.
Other bias	Low risk	There was no baseline imbalance between the 2 groups except the frequency of feeding. However, we judged that the absolute difference is not clinically important.

Gupta 2011.

Study characteristics
Methods	Randomised controlled trial in India Trial dates: not mentioned
Participants	60 mothers with lactation insufficiency and their infants Age of infants at start of study: birth up to 6 months of life. The mean infant age at the start of study was 2.8 months. Inclusion criteria: "mothers aged 20 to 40 years.... infants up to 6 months, ... having 1 or more of the following symptoms: deficient lactation, infant's crying just after feeding, painful sensation in breast during the time of feeding, loss of appetite in mother or the manifestation of any anxiety disorder which could effect the lactation." Exclusion criteria: none mentioned Breastfeeding method: mothers were advised to "use normal feeding techniques and schedule for their infants." There was no mention if supplemental feeding was allowed.
Interventions	Arm 1: Asparagus racemosus Willd. capsules (root powder of Asparagus racemosus Willd.) 3 times a day for 30 days. (Total daily dose 60 mg/kg) (n = 30) Arm 2: Placebo was identical looking capsules containing rice powder given in the same way as the intervention group. (n = 30)
Outcomes	Serum prolactin level before and after treatment Changes in maternal weight Changes in infant weight Maternal satisfaction measured by a graded scale ranging to 1 to 5 (1 = unsatisfactory, 5 = highly satisfactory) Well‐being and happiness of infants measured by a graded scale ranging to 1to 5 (1 = unsatisfactory, 5 = highly satisfactory)
Funding and Declaration of interest	No funding or declaration of interest statement was found.
Notes	Attempts to contact the authors for clarifications failed
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"The treatment allocation schedule was based on computer‐generated random numbers. The treatment codes resided with the principal investigator and the local investigators were not aware of treatment assignments. No treatment code was broken before the last follow‐up visit completion."
Allocation concealment (selection bias)	Low risk	"The study medication was provided in white paper boxes, numbered consecutively with a medication number."
Blinding of participants and personnel (performance bias) All outcomes	High risk	"The root powder was put into capsules depending on the bodyweight of each subject and labelled as "R." Similarly placebo capsules were prepared with fine rice powder and labelled "C." Comment: it is difficult to understand how the treatment could be blinded if they were labelled C and R and if 1 person was able to guess which treatment they were on, the code would be broken.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	Milk volume was not a part of this study.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	High risk	Maternal satisfaction and maternal perception of infant well‐being would likely be affected by the lack of blinding.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Serum prolactin levels, maternal and infant weigh assessment would not likely be affected by lack of blinding.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	"A total of 10 patients, who did not participate in the entire trial or did not turn up for regular follow‐up visits, were excluded from the study." Comment: we were unable to contact the authors to see if these dropouts were equally distributed between the 2 groups.
Selective reporting (reporting bias)	Low risk	All major outcomes stated in the methods were reported in the results.
Other bias	Unclear risk	There was no baseline comparison between the 2 groups.

Huang 2000.

Study characteristics
Methods	A randomised controlled trial in China Trial dates: January 1996 to December 1998
Participants	85 mothers and their infants Age of infants at start of study: at birth Inclusion criteria: mothers with infants born at 37 to 42 weeks' gestations, aged between 23 and 28 years old who had normal menstruation and normal mammary gland development, without any complications or endocrine diseases Exclusion criteria: not mentioned Breastfeeding method: 3‐hourly feeding. There was no mention if supplemental feeding was allowed.
Interventions	Arm 1: Xian Tong Ru (先通乳) oral liquid 50 mL twice a day for 3 days (n = 45) Arm 2: no intervention (n = 40) Note: Xian Tong Ru (先通乳) oral liquid is a herbal mixture containing danggui (当归), huangqi (黄芪), shudi (熟地), tongcao (通草) and wangbuliuxing (王不留行). 1 mL of the intervention contained 1 g of the herbal mixture
Outcomes	Onset of milk production Milk volume per half a day, measured by weighing the infant before and after breastfeeding on the first and third day of life Serum prolactin levels
Funding and Declaration of interest	No funding or declaration of interest statement was found.
Notes	The study was published in Chinese. The above information was obtained from the English translation of the article. No contact details of the authors were available.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	How this was done was not mentioned.
Allocation concealment (selection bias)	Unclear risk	How this was done was not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding was not possible because the control group received no intervention.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	Blinding of the outcome assessor was not reported.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	Self‐reported outcomes were not a part of this study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants completed the trial and were analysed.
Selective reporting (reporting bias)	Low risk	All outcomes mentioned in the methods were reported in the results.
Other bias	Low risk	None detected

Inam 2013.

Study characteristics
Methods	Randomised controlled trial in Pakistan Trial dates: March 2012 to September 2012
Participants	100 mothers with inadequate breast milk production and their infants Age of infants at start of study: day 6 of life Inclusion criteria: mothers who delivered at term with "inadequate breast milk production (≤ 10 mL breast milk per single expression (both breasts) at 6th postnatal day)." Exclusion criteria: "...women with medical diseases like chronic renal diseases or tuberculosis possibly decreasing milk output. Malnutrition with BMI < 18kg/m2, women with some breast diseases like abscess, mastitis, or malignancy on clinical examination and medical records and women with known allergy or prior reaction to Domperidone..." Breastfeeding method: mothers were taught proper breastfeeding techniques and practices including good diet and proper positioning. No mention if supplemental feeding was allowed.
Interventions	Arm 1: domperidone 10 mg 3 times a day for 7 days (n = 50) Arm 2: the control group were just given training and explanation on proper breastfeeding techniques and practices including good diet and proper positioning (n = 50)
Outcomes	Milk volume per expression (of both breasts) after 7 days of intervention
Funding and Declaration of interest	No funding or declaration of interest statements were found.
Notes	Attempts to contact authors for clarifications failed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"All women were randomly allocated in two groups by lottery method..."
Allocation concealment (selection bias)	Unclear risk	We judge this to be unclear because we were unable to tell if the allocation on the lottery ticket could be seen and be at risk of being interfered with.
Blinding of participants and personnel (performance bias) All outcomes	High risk	The control group was not given a placebo.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	Blinding of the outcome assessor was not reported.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	Self‐reported outcomes were not a part of this study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants completed the trial and were analysed.
Selective reporting (reporting bias)	Low risk	All outcomes mentioned in the methods were reported in the results.
Other bias	Unclear risk	Incomplete baseline information was reported for the 2 groups.

Jantarasaengaram 2012.

Study characteristics
Methods	Randomised controlled trial in Thailand Trial dates: July 2008 to August 2008
Participants	50 postcaesarean section mothers and their infants. 38% of mothers were primiparous. 46% had prior breastfeeding experience. Age of infants at start of study: first day of life. The average gestation of the infants was 38.5 weeks. Inclusion criteria: ".... healthy infants via cesarean delivery under regional anaesthesia after normal singleton term pregnancy." Exclusion criteria: "...body mass index (BMI, calculated as weight in kilograms divided by the square height in meters) of more than 24; postpartum bleeding of more than 1 L; underlying chronic medical disease; history of allergies to domperidone; history of smoking or substance abuse; gross breast or nipple abnormalities; and other conditions that would contraindicate breastfeeding." Breastfeeding method: "All participants were encouraged to breastfeed their infants within 24 hours of delivery." No further details were available and there was no mention if supplemental feeding was allowed.
Interventions	Arm 1: oral domperidone 10 mg 4 times a day for 4 days within 24 hours of delivery (n = 25) Arm 2: placebo was vitamin B6 tablets 25 mg given in a similar manner (n = 25)
Outcomes	Mean milk volume per 2 expressions in a day (first until fifth postpartum day), measured by milk expression using electric breast pump 2 hours after last breastfeed Adverse effects for mothers
Funding and Declaration of interest	The study was funded by Rajavithin Hospital (Department of Medical Services, Ministry of Public Health of Thailand) and the authors had declared no conflicts of interest.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Simple randomisation was achieved using a table of random numbers."
Allocation concealment (selection bias)	Low risk	"Both study medications were prepackaged, in sealed opaque packages, for four consecutive days of postpartum use."
Blinding of participants and personnel (performance bias) All outcomes	Low risk	"Neither the study participants nor the clinical staff members were aware of the medication assignments. Both tablets were similar in size, shape and colour."
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	"Neither the study participants nor the clinical staff members were aware of the medication assignments. Both tablets were similar in size, shape and colour."
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	"Neither the study participants nor the clinical staff members were aware of the medication assignments. Both tablets were similar in size, shape and colour."
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Comment: dropout rate was less than 15% in both groups, and the authors described reasons for dropout clearly. Quote: "In the domperidone group, 3 women withdrew because of neonatal hypoglycaemia (2 infants) and severe neonatal jaundice (one infant). In the placebo group, 2 women withdrew because of neonatal hypoglycaemia (one infant) and severe neonatal jaundice (one infant). The final analysis, therefore, included 22 women in the domperidone group and 23 women in the placebo group (Fig. 1)."
Selective reporting (reporting bias)	Low risk	All outcomes were reported in the results.
Other bias	Low risk	None detected

Jiang 2006.

Study characteristics
Methods	Randomised controlled trial in China Trial dates: January 2004 to June 2004
Participants	60 mothers with lactation insufficiency and their infants Age of infants at start of study: day 2 of life. The average birthweight was 3667 grams. Inclusion criteria: mothers with lactation insufficiency, and syndrome of qi and blood based on the Chinese medicine theory at 48 hours after delivery. Exclusion criteria: not mentioned Breastfeeding method: although the authors mentioned that supplemental feeding was not allowed, one of the study's outcome was the volume of supplemental feeds.
Interventions	Arm 1: Chanbao Oral Liquid (产宝) 10 mL twice a day for 5 days (n = 20) Arm 2: Bu Xue Sheng Ru (补血生乳) capsules 4 g twice a day for 5 days (n = 20) Arm 3: no intervention (n = 20)
Outcomes	Milk volume per day, measured by weighing the infant before and after breastfeeding at baseline, first, second, third, fourth and fifth day postintervention Serum prolactin Supplemental feed
Funding and Declaration of interest	Government‐funded study
Notes	The study was published in Chinese. The above information was obtained from the English translation of the article. No contact details of the authors were available.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random table was used.
Allocation concealment (selection bias)	Unclear risk	How this was done was not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding was not possible due to the difference in the nature of the interventions (Chanbao Oral Liquid was a form of liquid while Bu Xue Sheng Ru were capsules) and the third group was not given any intervention.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	Blinding of the outcome assessor was not reported.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	Self‐reported outcomes were not a part of this study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for in the results.
Selective reporting (reporting bias)	Low risk	All expected outcomes were reported.
Other bias	Low risk	None detected

Kauppila 1985.

Study characteristics
Methods	Randomised controlled trial in Finland Trial dates: not mentioned
Participants	33 healthy mothers (13 with lactation insufficiency) and their infants Age of infants at start of study: 4 to 20 weeks of life Inclusion criteria: none reported Exclusion criteria: none reported Breastfeeding method: this was not mentioned and there was no mention if supplemental feeding was allowed
Interventions	Arm 1: oral metoclopramide 10 mg 3 times a day for 3 weeks (n = 15) Arm 2: placebo tablet 1 tablet 3 times a day for 3 weeks (n = 18)
Outcomes	Maternal serum prolactin, thyroid stimulating hormone and free thyroxin levels at baseline, seventh and 21st day postintervention Infant serum prolactin, thyroid stimulating hormone and free thyroxin levels at baseline, seventh and 21st day postintervention Milk volume per day, measured by weighing the infant before and after breastfeeding at baseline, sixth and 21st postintervention (only mothers with obvious lactation deficiency had their milk volume measured (metoclopramide: n = 8; placebo: n = 5) Adverse effects (we assume this referred to all mothers included in the study)
Funding and Declaration of interest	"The drugs in this study were kindly supplied by from Neofarma Oy, Helsinki, Finland." No other declaration of interest found.
Notes	Attempts to contact authors for clarifications failed. The authors had measured milk volume only for the subgroup of mothers with lactation insufficiency. This was reported as individual participant data in a graph. We estimated and calculated the values for entry into Revman.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Comment: How this was done was not mentioned. Quote: "After one control day, when the daily milk yield was accurately measured, the women were randomised to receive metoclopramide (10 mg 3 times daily orally) or a placebo (one tablet 3 times daily) for a period of three weeks."
Allocation concealment (selection bias)	Unclear risk	How this was done was not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	A placebo was given in a similar manner but we were unable to judge whether the placebo looked identical with the intervention treatment.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	We judged this to be unclear because we are unsure of how successful blinding was with the outcome assessors (mothers).
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Unclear risk	We are unclear as to how well the mothers were blinded, therefore unclear how this would have affected their perception of adverse effects.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	The authors mentioned that "four women taking metoclopramide and four taking the placebo discontinued the trial for unknown reasons." Although not clearly reported, the mothers who dropped out were likely to be from the subgroup with normal lactation. Thus, it did not have any effect on the number of mothers with lactation insufficiency (those with outcomes that mattered).
Selective reporting (reporting bias)	Low risk	All expected outcomes were reported.
Other bias	High risk	There was baseline imbalance between the 2 groups.

Khairani 2017.

Study characteristics
Methods	4‐arm randomised control trial in Indonesia Trial dates not mentioned
Participants	24 primiparous mothers and their infants Age of infants at start of study: not mentioned Inclusion criteria: postpartum primipara mothers Exclusion criteria: none mentioned Breastfeeding method: not mentioned
Interventions	Arm 1: kelor leaf (moringa) powder 250mg 3 times per day for 10 days (n = 6) Arm 2: kelor leaf (moringa) powder 350mg 3 times per day for 10 days (n = 6) Arm 3: kelor leaf (moringa) 450mg 3 times per day for 10 days (n = 6) Arm 4: Placebo (not described) for 10 days (n = 6)
Outcomes	Breast milk production Adverse effects for mother
Funding and Declaration of interest	Funding and declaration of interest was not mentioned.
Notes	Further clarification about the study was needed but attempts to contact the authors failed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The authors used "Simple Random Sampling technique" but how this was done was not described.
Allocation concealment (selection bias)	Unclear risk	How this was done was not described.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	We were unable to judge whether the different capsules with the different dosages and the placebo looked identical.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	Milk volume was subjectively assessed by mothers as "Good, sufficient or less" and we are unable to judge if they were blinded.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Unclear risk	We are unable to judge if the mothers were blinded.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for in the results.
Selective reporting (reporting bias)	Unclear risk	Outcomes not specified in methods. No protocol available
Other bias	Unclear risk	Baseline not reported

Li 2010.

Study characteristics
Methods	Randomised controlled trial in China Trial dates: probably February 2006 to June 2009
Participants	90 mothers with lactation insufficiency and their infants. The average parity of mothers was 1.3 Age of infants at start of study: first day of life Inclusion criteria: women aged 21 to 36 years with insufficient breast milk but no breast engorgement who delivered healthy term infants Major criteria: lactation insufficiency, lack of breast engorgement Minor criteria: pallor, lethargy, mood instability; red tongue with white moss coating, thready weak pulse. 2 major criteria and 1 minor criterion need to be fulfilled to be included in the study. Exclusion criteria: women with endocrine, cardiac, liver, kidney, breast, blood, neurological or psychological problems Breastfeeding method: breastfeeding to start 30 minutes after delivery. No further details regarding feeding times were reported. Supplemental feeds were allowed.
Interventions	Arm 1: Mu Er Wu You soup (母儿无忧汤) twice a day at third day post‐delivery; 4 days duration as 1 complete course. Mu Er Wu You soup contains: ren shen (人参) 10 g, huang qi (黄芪) 20 g, dang gui (当归) 15 g, mai men dong (麦门冬) 10 g, tian hua fen (天花粉 10 g, chai mu (祡胡) 10 g, yi mu cao (益母草) 15 g, wang bu liu xing (王不留行) 10 g and jie geng (桔梗) 10 g (n = 45) Note: English translation of Mu Er Wu You: "Carefree mother and child." Arm 2: Kun Yuan Tong Ru soup (坤元通乳口服液). 60 mL taken twice daily. The herbs in the Kun Yuan Tong Ru soup was also not listed but from its name 'Tong Ru' which means milk flow, it was likely to contain galactagogues (n = 45)
Outcomes	Infant weight gain Amount of formula milk supplemented before and after intervention Breast fullness Milk volume measured by infant weight before and after feeding Mothers' general appearance: tiredness, tongue changes, appetite, pulse condition Serum prolactin and oestrogen levels Adverse effects (not specified if for mothers or infants)
Funding and Declaration of interest	No funding or declaration of interest statement reported.
Notes	The study was published in Chinese. The above information was obtained from the English translation of the article. No contact details of the authors were available.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The English abstract reported that "......... were divided into treatment and control groups randomly." Comment: how this was done was not mentioned.
Allocation concealment (selection bias)	Unclear risk	How this was done was not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	High risk	We judged this as high risk of bias because both the soups would taste different.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	Blinding of the outcome assessor was not reported.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	High risk	We judged assessment of milk supplementation at high risk of bias because lack of blinding could have affected the way these were measured.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	We judged infant weight measurements to be unlikely to be affected even if there was no blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for in the results.
Selective reporting (reporting bias)	Low risk	Although the methods section did not state the intended outcomes, all expected outcomes for this study were reported.
Other bias	Low risk	None detected

Manjula 2014.

Study characteristics
Methods	Randomised controlled trial in India Trial dates: December 2010 to April 2012
Participants	48 mothers with lactation insufficiency and their infants. The average parity of mothers was 1.8 Age of infants at start of study: between 10 and 180 days of life. The average infant age at start of study was 95 days. The average birthweight was 2838.5 grams. Inclusion criteria: mixed parity "lactating mothers who delivered at term without complications whose infants weighed not less than 2000 g at birth. Their infants had to be between 10 and 180 days of age and had either failed to regain birth weight at 15 days of life or required supplementing feed ≥ 250 mL/day after 4 weeks of birth." Exclusion criteria: "... mothers with breast abscess, cracked nipples, epilepsy, psychosis, alcohol addiction, mastitis, previous breast surgery, chronic diseases such as tuberculosis, malignancy, and acquired immunodeficiency syndrome; and infants who were premature, had inborn errors or if they weighed less than 2000 g." Breastfeeding method: this was not described but supplemental feeding was allowed
Interventions	Arm 1:Gossypium herbaceum Linn seed kernels 10 g a day for 1 month (n = 32) Arm 2: placebo (roast wheat flour) given in the same manner (n = 16)
Outcomes	Mean infant weight before and after intervention Number of mothers fully breastfeeding, partially breastfeeding or no response after intervention Mean volume of supplemental feeds before and after intervention Maternal satisfaction, measured on a scale of 1 to 5 (1 = unsatisfactory and 5 = highly satisfactory) Maternal perception on breast fullness, contralateral ejection of milk and breast milk increase, measured on a scale of 1 to 5 (1 = unsatisfactory and 5 = highly satisfactory) Adverse effects (not specified if for mothers or infants)
Funding and Declaration of interest	Correspondence with the lead author revealed that the Ministry of Ayurveda, Yoga and Naturopathy, Unani, Siddha and Homoeopathy (AYUSH) India funded the study as an unrestricted educational grant for a postgraduate thesis.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization was carried out by lottery method. Under this method, small and identical paper slips were numbered, which were folded and mixed together in a drum thoroughly. A blindfold selection was then made of the number slips that were required for this study. After drawing out 1 slip and noting the number, the slip was again put back in the drum. The drum was reshuffled and second slip was drawn. This process was repeated till the sample size was completed. The slip that was drawn for second time was rejected."
Allocation concealment (selection bias)	Unclear risk	How this was done was not described.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	"The lactating mothers were blinded for the study by dispensing the test drug and placebo in same colour capsules."
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	Correspondence with the lead author confirmed that outcome assessors were blinded.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	"The lactating mothers were blinded for the study by dispensing the test drug and placebo in same colour capsules."
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Correspondence with the lead author confirmed that outcome assessors were blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	2 participants from the Gossypium group and 1 participant from the control group dropped out and were excluded from the final analysis. Correspondence with the lead author revealed that participants dropped out because they were staying too far away from the hospital.
Selective reporting (reporting bias)	Low risk	All the outcomes mentioned in the methods were reported in the results.
Other bias	Low risk	None detected

Mathew 2018.

Study characteristics
Methods	"Randomised controlled pre‐test post‐test design" in India Trial dates: not mentioned
Participants	30 mothers and their infants. 60% of mothers were primiparous and 93% had normal vaginal delivery. Age of infants at start of study: 10 days to three months of life Inclusion criteria: lactating mothers 20 to 35 years old, 10 days up to 3 months of postpartum Exclusion criteria: none mentioned Breastfeeding method: direct breastfeeding during the intervention period. No mention if supplemental feeding was allowed.
Interventions	Arm 1: fennel tea (14 grams of in two liters of water) and 300 mL of this was given per day for seven consecutive days (n = 15) Arm 2: fenugreek tea (14 grams in two liters of water) and 300 mL of this was given each day for seven days (n = 15)
Outcomes	To compare the effects of fenugreek and fennel on lactation among lactating women. To compare the average ideal weight gain of babies according to their age and the obtained weight gain. To find an association between selected variables and lactation among lactating women.
Funding and Declaration of interest	Self‐funded. The authors declared no conflict of interest.
Notes	Further clarification about the study was needed but attempts to contact the authors failed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Lottery method
Allocation concealment (selection bias)	Unclear risk	How this was done was not described.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Fennel and fenugreek are aromatic herbs. Their smell and taste are distinct. Therefore it its likely that participants would be able to guess what they were drinking.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	Milk volume was assessed via test weighing, thus unlikely to be affected.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	This was not an outcome in the study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	This was not an outcome in the study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants recruited were included in the analysis.
Selective reporting (reporting bias)	High risk	The outcomes mentioned in the methods did not match results of study. Measures of dispersion was not included in the results.
Other bias	Low risk	None detected

Mukherjee 1987.

Study characteristics
Methods	Quasi‐randomised trial in India Trial dates: not mentioned
Participants	100 mothers with insufficient or failure of lactation, and their infants. 40% of mothers were primiparous. Age of infants at start of study: unable to tell Inclusion criteria: mothers with insufficient or failure of lactation validated by clinical examination of the breasts, infants requiring supplemental feeding, infants who were not gaining weight Exclusion criteria: mothers with any systemic disorders or local disease of the breast and with infants either grossly premature of seriously ill Breastfeeding method: this was not described but supplemental feeding was allowed
Interventions	Arm 1: Lactare 2 capsules 3 times a day for 30 days (n = 50) Arm 2: placebo (of same size, colour and shape) given in a similar manner (n = 50) Lactare: was a mix of shatavari (Asparagus racemosus), ashwagandha (Withania sominfera), licorice (Glycyrrhiza glabra), fenugreek (Trigonella foenum‐graecum) and garlic (Allium sativum)
Outcomes	Efficacy of Lactare as a galactagogue categorized as "Good" (mother's milk increased so infant could solely breastfeed and gained weight), "Moderate" (infant's supplemental feeds reduced and infant gained weight), "Poor" (no change in lactation or infant's weight)
Funding and Declaration of interest	Funding and declaration of interest was not mentioned.
Notes	144 mothers were recruited, but only the first 50 from each arm who completed the trial were analysed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	"Alternate patients was allotted to Group A and Group B"
Allocation concealment (selection bias)	Unclear risk	How this was done was not described.
Blinding of participants and personnel (performance bias) All outcomes	High risk	"Lactare capsules marked Code A and Code B containing the drug and placebo of same size, colour and shape were supplied and were given to Group A and Group B patients respectively."
Blinding of outcome assessment (detection bias) Milk volume outcomes	High risk	This was a self‐reported outcome, hence the results could be affected if the mother knew what she was taking.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	High risk	This was a self‐reported outcome, hence the results could be affected if the mother knew what she was taking.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	We judged infant weight measurements to be unlikely to be affected even if there was no blinding.
Incomplete outcome data (attrition bias) All outcomes	High risk	A total of 144 participants were recruited but the trial stopped when 100 participants completed the 30 days of study.
Selective reporting (reporting bias)	High risk	Did not report adverse effects although it was specified in the methods section as one of the outcomes.
Other bias	Low risk	None detected

Nordin 2019.

Study characteristics
Methods	Randomised controlled trial in Malaysia Trial dates: October 2016 to 2017
Participants	58 working mothers and their infants. 19% of mothers were primiparous. Age of infants at start of study: two to six month's of life. The average infant age at start of study was 4.42 months. Inclusion criteria: mothers were 18 to 40 years old, "working hours of within 9 hours", exclusively breastfeeding. Infants at full‐term, healthy, 2 to 6 months old, not started weaning diet Exclusion criteria: mothers with a history of smoking, alcohol, drugs or herbs to improve breast milk production. Infant with low birth weight, low APGAR score, intrauterine growth retardation, any illness or congenital abnormalities Breastfeeding method: exclusive breastfeeding on demand
Interventions	Arm 1: banana flower flour biscuits 3.24 g (2 pieces of biscuits) daily for 3 weeks (n = 29) Arm 2: placebo (wheat flour biscuits) given in a similar way (n = 29)
Outcomes	Frequency and amount of breast milk expression for five working days before and after intervention Anthropometric indices of mothers and infants before and after intervention
Funding and Declaration of interest	Sponsored by the Malaysian Ministry of Higher Education. Authors declared no conflict of interest.
Notes	We managed to contact the corresponding author who provided details not available in the published paper.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	"Convenient sampling was done on 58 mother‐infant pairs which were recruited via social media and volunteers from member of Breastfeeding Mother;s Support Group of Pahang (KUSSIP).” Correspondence with the author confirmed that no random sequence was generated. The first 29 mothers were allotted to the intervention group, and the subsequent 20 mothers were allotted to the placebo group.
Allocation concealment (selection bias)	High risk	As above
Blinding of participants and personnel (performance bias) All outcomes	High risk	Correspondence with the author revealed that the banana flower biscuits looked darker compared to the placebo wheat flour biscuits.
Blinding of outcome assessment (detection bias) Milk volume outcomes	High risk	Blinding of the outcome assessors (mothers) could potentially be compromised because the biscuits looked slightly different.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	This was not a part of the study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study. The authors only measured the infant's BMI.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for in the results.
Selective reporting (reporting bias)	High risk	Infants' BMI was not reported postintervention, although this was specified as a secondary outcome in the methods.
Other bias	Low risk	None detected

Paritakul 2016.

Study characteristics
Methods	Randomised controlled trial in Thailand Trial dates: August 2015 to April 2016
Participants	68 healthy mothers and their infants. 38% of mothers were primiparous and 58% had normal vaginal delivery. Age of infants at start of study: first day of life. The average gestation of the infants was 38.6 weeks. The average birthweight was 3066 grams. Inclusion criteria: healthy women 18 years and above who aimed to exclusively breastfeed for at least 6 months Exclusion criteria: mothers with "serious medical conditions presumed to result in mother‐infant separation and decreased breast‐feeding frequency (e.g. postpartum haemorrhage, postpartum sepsis), allergic to ginger, or have a contraindication to breastfeeding such as HIV infection." Breastfeeding method: breastfeeding on demand. Supplemental feeding was probably not allowed as the mothers had to be exclusively breastfeeding to be in the inclusion criteria.
Interventions	Arm 1: ginger (Zingiber officinale) capsule 500 mg twice a day for 7 days (n = 34) Arm 2: placebo was corn starch capsules 500 mg given in a similar manner (n = 34) Both intervention and placebo were identical looking and were prepared by Abhaibhubejhr Herbal company.
Outcomes	Mean milk volumes on third and seventh day of intervention, measured in very different ways* Serum prolactin level *Milk volume measurement on third day: test weighing of the infant before and after each feeding for a period of 24 hours Milk volume measurement on seventh day: extrapolate 24‐hour milk production from 1 expression (see notes below)
Funding and Declaration of interest	Research funds from Faculty of Medicine, Srinakarinwirot University. Both intervention and placebo were prepared by Abhaibhubejhr Herbal company.
Notes	The study investigators misemployed Lai 2010 to estimate the milk volume on the seventh day. The investigators had asked mothers to "first empty their breast using an electronic breast pump (first expression). After an hour, both breasts were then pumped again for 15 minutes to measure the 1‐hour breast milk volume (second expression)." The 1‐hour breast milk volume was then used to extrapolate the 24‐hour milk production. This is in contrast to the method originally described by Lai 2010, which used the average of the amounts from the third and fourth expression to estimate the 24‐hour milk production. The main author clarified that the number of dropouts from the control group was 12, not 11 as reported in the paper.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	This was done using a "Computer‐generated list with block of four method."
Allocation concealment (selection bias)	Low risk	This was done using "…sequentially numbered sealed envelopes."
Blinding of participants and personnel (performance bias) All outcomes	Low risk	The mothers were given an "identical looking placebo." "Neither the midwife nor the patient was aware of the treatment group."
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	The mothers were given an "identical looking placebo." "Neither the midwife nor the patient was aware of the treatment group."
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	Self‐reported outcomes were not a part of this study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	For milk volume assessed on third day, 3 from the intervention group withdrew due to personal reasons and 1 due to puerperal sepsis. 1 from the placebo group withdrew due to personal reasons. We judge this as low risk because although it appears that the intervention group had slightly higher attrition rates, the absolute number is actually small.
Selective reporting (reporting bias)	Low risk	All prespecified outcomes were reported.
Other bias	Low risk	None detected

Sakha 2008.

Study characteristics
Methods	Randomised controlled trial in Iran Trial dates: not mentioned
Participants	20 primiparous mothers with perceived lactation insufficiency and their infants Age of infants at start of study: not stated clearly except that they were "a few months old." Inclusion criteria: primigravida nursing mothers with perceived lactation inadequacy and their infant's weight gain was less than 500 g/month. All these mothers were given a short training on proper breastfeeding techniques before starting the study. Exclusion criteria: "...preterm or low‐birth‐weight infants; working mothers; mothers with infants who had cardiac, pulmonary, musculoskeletal, metabolic, genetic, and neurological disorders or anomalies; mothers who had tried bottle feeding before counselling; mothers with multi‐foetal delivery; mothers with anatomical abnormalities of the breast; mothers who had been admitted to hospital more than three days after delivery and mothers whose new‐born infant had been admitted to the hospital more than three days after birth." Breastfeeding method: breastfeeding on demand and no supplemental feeding was allowed
Interventions	Arm 1: metoclopramide tablets 10 mg 3 times per day for 15 days (n = 10) Arm 2: placebo (starch tablet), 3 times a day for 15 days (n = 10) Both groups received a short training session on breastfeeding focusing on positioning, latch and importance of exclusive breastfeeding before the study.
Outcomes	Infant weight gain after intervention
Funding and Declaration of interest	"Our study was sponsored by Research Vice‐Chancellor of Tabriz University of Medical Sciences" (correspondence with the author)
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"We used computer generated randomisation" (correspondence with author).
Allocation concealment (selection bias)	Low risk	"Yes, we did concealment by using to types of opaque envelopes named as: A (placebo) and B (metoclopramide) delivered to participants according to their random number that was determined by computer" (correspondence with author).
Blinding of participants and personnel (performance bias) All outcomes	Low risk	"Drug delivery was done in Tabriz Children's Hospital by pharmacy personnel who were not aware of this study. Placebo was starch tablet with white colour and the same size of metoclopramide (10 mg) tablet" (correspondence with author).
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	Milk volume was not a part of this study.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	No self‐reported outcomes
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	"Weighing was done by a nurse in Infants' Clinic of Tabriz Children's Hospital who was not aware of this study" (correspondence with author).
Incomplete outcome data (attrition bias) All outcomes	Low risk	"Indeed, we had only 10 participants in each group and no one left our study protocol" (correspondence with author).
Selective reporting (reporting bias)	Low risk	All prespecified outcomes were reported.
Other bias	Unclear risk	No baseline information was reported for the 2 groups.

Sakka 2014.

Study characteristics
Methods	Randomised controlled trial in Egypt Trial dates: April 2012 to November 2012
Participants	75 mothers who delivered vaginally and their infants. 52% of mothers were primiparous. Age of infants at start of study: first day of life. Average gestation of the infants was 38.4 weeks. The average birthweight was 3335 grams. Inclusion criteria: mothers who were willing to exclusively breastfeed and pump their breasts Exclusion criteria: Mothers: high‐risk pregnancy, such as diabetes or hypertension; inverted nipples; history of asthma; allergy to peanuts Infants: premature or low birthweight infants; infants with cleft lip or palate or gross congenital malformations or genetic syndromes; preterm or low birthweight. Mothers with poor compliance to the intervention, used supplementation of any type or the infant had illness requiring medications were removed from the study. Breastfeeding method: exclusive breastfeeding on demand and no supplemental feeding was allowed
Interventions	Arm 1: fenugreek herbal tea, 1 cup 3 times daily. Each cup contained approximately 2 g of Grade A fenugreek (n = 25) Arm 2: palm dates (Grade A) approximately 100 g 3 times daily (n = 25) Arm 3: no treatment: "...mothers who consumed no galactagogues" (n = 25) The exact duration for each intervention was not mentioned
Outcomes	Infant weight at birth and on the third, seventh and fourteenth day of life Milk volume expressed using a manual breast pump before the first feed in the morning of the third day postpartum
Funding and Declaration of interest	No funding or declaration of interest statement was found.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"..., we used a random number table ..."
Allocation concealment (selection bias)	Low risk	Sealed opaque envelopes were used.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding was not possible because of the nature of the intervention used.
Blinding of outcome assessment (detection bias) Milk volume outcomes	High risk	The outcome assessors (mothers) were not blinded.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	Self‐reported outcomes was not a part of this study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight measurements would unlikely be affected by lack of blinding.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	The number of participants analysed for the outcomes is unclear.
Selective reporting (reporting bias)	Low risk	All prespecified outcomes were reported.
Other bias	Low risk	None detected

Shariati 2004.

Study characteristics
Methods	Randomised controlled trial in Iran Trial dates: not mentioned
Participants	158 mothers with lactation insufficiency and their infants Age of infants at start of study: from birth until 6 months of life. The median infant age at start of study was 50.5 days Inclusion criteria: healthy mothers who did not smoke, singleton pregnancy, had not more than 4 previous children, had normal breasts and no nipple retraction and not on drugs. Their infants must be term, weighing between 2.5 kg to 4 kg, healthy but had growth charts that showed plateauing growth or growth below what was expected. Exclusion criteria: mothers with infants who received supplemental feeding or had malnutrition that required hospital admission. Mothers who had disease or mental problems or were taking other drugs, mothers who could not come for follow‐up measurements and mothers who could not breastfeed properly. Breastfeeding method: breastfeeding on demand and no supplemental feeding was allowed.
Interventions	Arm 1: Shirafza* drops 30 drops 3 times a day for 4 weeks (n = unknown) Arm 2: placebo 3 times a day (chlorophyl in alcohol 1:1000, no further details given) to be taken 3 times a day (n = unknown) * Shirafza was made up of 6 herbs: fennel (Foeniculum volgare), anise (Pimpinella anisum), green cumin (Cuminumcyminum), dill (Anethum graveolens), parsley (Petroselinum crispum) and black seed (Nigella sativa) via alcohol extraction
Outcomes	Weekly infant's weight, length and head circumference from first until fourth week Adverse effects in the form of a self‐reported questionnaire for mothers and infants
Funding and Declaration of interest	No funding or declaration of interest reported in the paper.
Notes	This study was published in Persian. The data above were obtained from the English translation of the paper. There was no contact address in the paper for us to contact the authors for further information.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No random sequence generation was described. "Participants were randomly allocated to 2 groups."
Allocation concealment (selection bias)	Unclear risk	How this was done was not reported.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Shirafza drops were made up of aromatic herbs which would smell and taste different from chlorophyl. Therefore it its likely that participants would be able to guess what they were taking.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	Milk volume was not a part of this study.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	High risk	We judge this to be high risk of bias because the mothers were unlikely to be blinded, hence the results of the self‐reported outcomes would be affected.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight, length and head circumference measurements would unlikely be affected by lack of blinding.
Incomplete outcome data (attrition bias) All outcomes	High risk	The authors reported "high dropout rates." It was also not clear what the actual number was, and from which group they were from.
Selective reporting (reporting bias)	Low risk	All expected outcomes were reported in the results.
Other bias	Low risk	None detected

Sharma 1996.

Study characteristics
Methods	Randomised controlled trial in India Trial dates: not mentioned
Participants	64 mothers with lactation insufficiency and their infants Age of infants at start of study: 14th to 90th day of life Inclusion criteria: "Mothers who had delivered at term without complications who, between 14th to 90th postpartum day reported lactation inadequacy. ....which was defined as: i) failure to regain birth weight at 15 days of life or ii) infant weight gain < 15 g/day or iii) mothers supplementing > 250 mL/day of milk after four weeks of birth." All mothers were motivated to exclusively breastfeed the infants after 1 week of exclusive breastfeeding, had a weight gain of < 15 g/day. Exclusion criteria: "Infants with malformations that could affect feeding or growth, infants with illness, mothers with severe illness or severe malnutrition..." Breastfeeding method: mothers were motivated to exclusively breastfeed, advised on position and frequency of feeds, adequate rest and nutrition. Supplemental feeding was allowed.
Interventions	Arm 1: mixed galactagogue (dose not specified), 2 times a day for 4 weeks (each 100 g of the mixed galactagogue contained Shatavari (Asparagus racemosus) 15.0 g, Sowa (Anethum sowa) 1.0 g, Bidarikand (Epomea digitata linn) 1.0 g, Palak (Spinacia oleracea linn) 2.5 g), Safed jeera (Cuminum cyminum) 0.5 g, Panchatrinamol 1.0 g (n = 32) Arm 2: placebo given in the same manner (n = 32)
Outcomes	Serum prolactin levels before and after intervention Infant weight gain before and after intervention Volume and frequency of supplementary feeds before and after intervention Adverse effects for mothers Mother's liver function test before and after intervention
Funding and Declaration of interest	No funding or declaration of interest reported in the paper. Correspondence with the author revealed that the study was funded by Dabur Research Foundation.
Notes	We managed to contact the author for clarification but she only had the answer to a few of our queries as this study was conducted more than 20 years ago.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	“The mothers were randomised to receive either placebo or galactagogue ….”. Correspondence with author revealed that this was done by computer‐generated random numbers.
Allocation concealment (selection bias)	Low risk	How this was done was not mentioned. Correspondence with author revealed that this was done using sequentially sealed opaque envelopes.
Blinding of participants and personnel (performance bias) All outcomes	High risk	"Both placebo and the galactagogue had similar colour, consistency, taste and packing." However the packaging was labelled as Drug A and B making it possible for the mother to guess which group she was randomised to.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	Milk volume was not a part of this study.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	High risk	Lack of blinding would likely affect mother's perception of adverse effects and the need for supplementation.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Lack of blinding would be unlikely to affect outcome assessment for serum levels and weight gain.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	11 participants did not complete the trial but we were not able to determine what the reasons were and which group they were from.
Selective reporting (reporting bias)	Low risk	All the outcomes mentioned in the methods section were reported.
Other bias	Low risk	None detected

Su 2008.

Study characteristics
Methods	Randomised controlled trial in China Trial dates: January 2003 to January 2006
Participants	108 healthy mothers with lactation insufficiency and their infants Age of infants at start of study: 7 to 13 days old Inclusion criteria: healthy women with milk insufficiency and have insufficient qi and 'blood' Major criteria: inadequate lactation, lack of breast engorgement Minor criteria: pallor, lethargy, loss of appetite, red tongue with white moss coating, thready weak pulse Western medicine criteria: infants do not produce continuous suckling sounds, restlessness and crying after feeding, not able to sleep, and less than 6 times of urination in 24 hours Exclusion criteria: mothers with liver problems, gynaecological problems, kidney problems, psychological problems, mothers who do not know how to breastfeed; infants with 'short tongue' Breastfeeding method: this was not described. Supplemental feeding was allowed.
Interventions	Arm 1: Cui Ru soup (催乳汤) (made with 1 or 2 pork knuckles (猪蹄), soya bean (黄豆) 50 g and peanuts (花生) 50 g together with 5 herbs: bei qi (北芪) 30 g, dang sheng (党参)15 g, dang gui (当归) 10 g, wang bu liu xing (王不留行) 20 g, tong cao (通草) 12 g 2 times a day for 7 days. Amount taken not described (n = 60) Note: *Cuiru means "Promote lactation" Arm 2: similar soup without the herbs given in a similar manner (n = 48)
Outcomes	A collective score to measure the disappearance of 'bad signs.' This includes physician's observation (breast engorgement, insufficient qi and 'blood,' tiredness, poor appetite), volume of breast milk expressed and infants' satisfaction. Milk volume before and after treatment Adverse effects (not specified if for mothers or infants)
Funding and Declaration of interest	No funding or declaration of interest statements were found.
Notes	This study was published in Chinese. The data above were obtained from the English translation of the paper. No contact details of authors were available for clarifications.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"Patients were randomly divided into two groups." How this was done was not reported.
Allocation concealment (selection bias)	Unclear risk	How this was done was not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Soup with the herbs would smell and taste differently from soup without the herbs. Blinding of personnel not described.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	Blinding of the outcome assessor (research assistants or investigators) was not reported.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	High risk	Most of the items in the collective score were self‐reported outcomes. Lack of blinding would likely influence this score.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants completed the trial and were analysed.
Selective reporting (reporting bias)	Low risk	All major outcomes stated in the methods were reported in the results.
Other bias	Unclear risk	No baseline information was reported for the 2 groups.

Sy 2012.

Study characteristics
Methods	Randomised controlled trial in Philliphines Trial dates: not mentioned
Participants	26 healthy mothers and their infants. 23.5% of mothers were primiparous Age of infants at start of study: 2 weeks to 6 months of age Inclusion criteria: exclusive or almost exclusive breastfeeding, "delivered at term two weeks to six months postpartum." Exclusion criteria: mothers with medical conditions contraindicated to breastfeeding, on medications contraindicated to moringa and domperidone, or currently taking other galactagogues Breastfeeding method: mothers were given breastfeeding education at the start of study. Supplemental feeding to a maximum of 2 formula bottles per day was allowed.
Interventions	Arm 1: moringa capsules 250 mg 2 times a day for 7 days (n = 12) Arm 2: domperidone tablets 10 mg 3 times a day for 7 days (n =14)
Outcomes	Mean milk volume per day measured by expressing milk (15 minutes of pumping) for 3 consecutive hours and the volume of the last expression (the fourth pump) was used to extrapolate the 24‐hour milk volume Percentage of mothers with significant increase in milk volume postintervention Adverse effects (we assume this referred to all mothers included in the study)
Funding and Declaration of interest	No funding or declaration of interest statements were found.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	How this was done was not mentioned.
Allocation concealment (selection bias)	Low risk	"....randomly given sealed brown envelopes with a number label."
Blinding of participants and personnel (performance bias) All outcomes	High risk	Domperidone were tablets given 3 times a day and moringa were capsules given twice a day.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	Blinding of the outcome assessor was not reported.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	High risk	Lack of blinding could have affected mother's perception of adverse effects.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	High risk	The dropout rate was around 35% across both groups.
Selective reporting (reporting bias)	High risk	In the methods, it was reported that milk was estimated by serial pumping every hour for 3 consecutive hours, and the amount at the fourth session was recorded, then the sum of the amount from both breasts obtained on the fourth pumping was multiplied by 24 to obtain the daily breast milk production. However in the results, the volume of milk reported was likely to be a single expression volume.
Other bias	Low risk	None detected

Thaweekul 2014.

Study characteristics
Methods	Quasi‐randomised trial in Thailand Trial dates: January 2012 to August 2012
Participants	233 healthy mothers and their infants. The mean parity of the mothers was 1.8 Age of infants at start of study: first day of life Inclusion criteria: ".... normal delivery of a healthy singleton infant with birth weight between 2500 ‐ 4000 g..." Exclusion criteria: mothers with serious medical conditions affecting lactation, receiving breastfeeding contraindicated medications. Infants admitted to the neonatal intensive care unit. Breastfeeding method: not specified but assumed to be breastfeeding on demand as the infants had early initiation of breastfeeding and were roomed with their mothers. Supplemental feeding was allowed.
Interventions	Arm 1: hospital‐based food programme "Galactagogue foods composed of hot basil, lemon basil, sweet basil, banana blossom, garlic, garlic chives, ginger, pepper .... 2500 kcal diet with 70 g protein per day" given from the first day postpartum through day of discharge (n = 106) Arm 2: healthy diet with the same amount of calorie and protein for a similar duration (n = 127)
Outcomes	Number of infants with excessive weight loss at 24 and 48 hours Maternal perception of breast fullness or heaviness, let‐down reflex and leakage of milk or colostrum in the first 48 hours Daily LATCH score till discharge Frequency of breastfeeding per day till discharge Timing of first breastfeed Volume of supplemental feed ("non breast milk fluid") given at 24 and 48 hours Amount of maternal food intake
Funding and Declaration of interest	Funding was from the Thammasat Research Grant, Thammasat University Thailand. The authors declared no conflict of interest.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Not done. This was a quasi‐randomised study where "Subjects were divided into two groups by a monthly admission."
Allocation concealment (selection bias)	High risk	No allocation concealment as the "Subjects were divided into two groups by a monthly admission."
Blinding of participants and personnel (performance bias) All outcomes	High risk	All the 'galactagogue' food have a unique taste and smell that would be impossible to blind.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	Milk volume was not a part of this study.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	High risk	Lack of blinding could influence how each mother perceived breast fullness and to some extent, let‐down reflex as well.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	The outcome assessors were not blinded, but measurement of the infant weight is unlikely influenced by the lack of blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for and analysed.
Selective reporting (reporting bias)	Low risk	All outcomes were reported adequately.
Other bias	Low risk	None detected

Tirak 2008.

Study characteristics
Methods	3‐arm randomised controlled trial in Turkey (we excluded the routine advice arm from the analysis) Trial dates: January 2007 to April 2007
Participants	63 mothers and their infants. 58.7% of mothers were primiparous, 44% had normal vaginal delivery and 76% had prior breastfeeding experience. Correspondence with the author revealed that 93 mothers were recruited but 15 dropped out. Age of infants at start of study: first day of life. The mean birthweight was 3240 grams. Inclusion criteria: mothers who delivered a term infant (more than 37 weeks' gestation) by the same doctor were enrolled. Exclusion criteria: mothers with chronic diseases and mothers on medication. Infants less than 35 weeks' gestation, infants with neonatal jaundice, congenital anomalies, pneumonia, sepsis and dehydration Breastfeeding method: breastfeeding on demand with no supplemental feeding
Interventions	Arm 1: Humana Still Tee granules, 9 g in 200 mL water 3 times a day for a month (n = 21). (Humana Still Tee consists of Hibiscus (Hibiscus tiliaceus): Amber flower extract 2.6 g, Fennel extract (Foeniculum vulgare): Fennel 0.2 g, Fennel oil: 0.02 g, Rooibos (Aspalatus linearis): red bush; 0.2 g, Verbena Herb (Verbena officinalis): Mine flower; 0.2 g, Raspberry Leaves (Rubus idaeus): Raspberry 0.2 g: Fenugreek (Trigonella foenum‐graecum): Fenugreek: 0.1 g and Goat's Rue Herb (Galega officinalis): Keçisedefi grass 0.1 g) Arm 2: Linden tea (1.5 g of lime blossom, Tilia silvestris) in 200 mL water 3 times a day for a month (n = 21) Arm 3: water 200 mL 3 times a day for a month (n = 21) Mothers in all 3 arms were encouraged to drink an addition of 2 litres of fluids a day.
Outcomes	Infant weight gain at the end of intervention
Funding and Declaration of interest	The paper reported that the study was funded by Sam Mamsel drug and TA S. and correspondence with the author added that this study was supported by Humana–Mamsel Pharmaceutical Company, Istanbul, Turkey.
Notes	This study was published in Turkish. All information was obtained from the English translation and correspondence with the author. The correct spelling for the first author is Tiras.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was done by using a table of random numbers.
Allocation concealment (selection bias)	Low risk	Allocation concealment was done by sequentially sealed opaque envelopes.
Blinding of participants and personnel (performance bias) All outcomes	High risk	There was no blinding as the appearance and taste of each intervention were different (granules versus tea bags versus water).
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	Milk volume was not a part of this study.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	Self‐reported outcomes were not a part of this study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	The assessor was blinded to the intervention given.
Incomplete outcome data (attrition bias) All outcomes	High risk	15 mothers were excluded from the analysis: 9 mothers from the Humana Still Tee group did not come for follow‐up, 3 mothers from the Linden tea group "did not use tea as suggested" and 3 infants from the water group were hospitalised due to jaundice.
Selective reporting (reporting bias)	Low risk	The outcome stated in the methods was reported in the results.
Other bias	Low risk	None detected

Turkyilmaz 2011.

Study characteristics
Methods	3‐arm randomised controlled trial in Turkey (we excluded the routine advice arm from the analysis) Trial dates: November 2006 to April 2007
Participants	66 mothers and their infants Age of infants at start of study: first day of life Inclusion criteria: "Mothers with healthy term infants, who were (1) willing to exclusively breast feed their infants, (2) having consented to follow‐up visits until infants catch‐up their birth weight following a period of postnatal weight loss, and (3) having agreed to pump the breast by electrical pump on the third day following delivery." Exclusion criteria: "Mothers who had chronic illness such as diabetes, hypertension, bronchial asthma, any allergies, and any breast problems such as inverted nipples, mastitis; a history of smoking, alcohol,or any drug use...... or infants that had low birth weight, low Apgar scores, intrauterine growth retardation, and any illnesses or congenital abnormalities." Breastfeeding method: "Mothers were supported by the same lactation consultant nurse throughout the study period. They were given similar breastfeeding education." Since exclusive breastfeeding was an inclusion criteria, we assume supplemental feeding was not allowed.
Interventions	Arm 1: Humana Still Tee, at least 3 cups (600 mL) a day (n = 22) (Humana Still Tee consists of Hibiscus (Hibiscus tiliaceus): Amber flower extract 2.6 g, Fennel extract (Foeniculum vulgare): Fennel 0.2 g, Fennel oil: 0.02 g, Rooibos (Aspalatus linearis): red bush 0.2 g, Verbena Herb (Verbena officinalis): Mine flower 0.2 g, Raspberry Leaves (Rubus idaeus): Raspberry 0.2 g: Fenugreek (Trigonella foenum‐graecum): Fenugreek: 0.1 g and Goat's Rue Herb (Galega officinalis): Keçisedefi grass 0.1 g) Arm 2: placebo (granule apple tea) at least 3 cups (600 mL) a day (n = 22) Arm 3: control (routine advice) (n = 22) The exact duration of intervention was not reported. Participants in all 3 arms received routine advice.
Outcomes	Mean milk volume on the third day after delivery, measured by pumping both breast consecutively for 15 minutes Infants' weight loss Time to regain birthweight (day) Adverse effects for mothers and infants
Funding and Declaration of interest	The authors reported that "No financial conflicts exist."
Notes	Attempts to contact the authors for further clarifications failed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"The mother‐infant pairs were randomly assigned to three groups." How this was done was not mentioned.
Allocation concealment (selection bias)	Unclear risk	How this was done was not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "...apple tea as placebo, which is the same colour and form as the galactagogue tea." Comment: however, both teas have distinct taste and smell and could not possibly be blinded.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	Blinding of the outcome assessor was not reported.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	High risk	Lack of blinding could influence the mothers' perception of adverse effects.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	"All mother–infant pairs were followed by the same nurse and paediatrician blinded to the study." Lack of blinding would also unlikely affect measurements of infant weight.
Incomplete outcome data (attrition bias) All outcomes	Low risk	"No mother was required to be excluded from the study due to noncompliance." Outcome data from 2 mothers in the control group (third arm of the study) were not reported. However, as this group was not included in our meta‐analysis, we judge this as low risk.
Selective reporting (reporting bias)	Low risk	All expected outcomes were reported in the results.
Other bias	Low risk	None detected

Wagner 2019.

Study characteristics
Methods	Randomised controlled trial in the USA Trial dates: March 2012 to March 2016
Participants	72 exclusively breastfeeding healthy mothers no milk production difficulties and their infants. 25% of mothers had normal vaginal delivery. Age of infants at start of study: 2 weeks to 3 months of life. The mean infant age at start of study was 5.1 weeks Inclusion criteria: exclusively breastfeeding, healthy with no milk production difficulties, 18 to 45 years old, BMI < 50 or without morbid obesity. Infants who are singleton, between 2 weeks and 3 months of age and ≥37 weeks gestation Exclusion criteria: mother with known allergic reactions or sensitivity to the component herbs in Mother Milk Tea or cross‐reactive plant species; had specific chronic illnesses (diabetes, hypertension, bronchial asthma, gastroesophageal reflux disease, atopic dermatitis, coeliac disease or gluten sensitivity, Crohn’s disease, ulcerative colitis, eating disorders, breast cancer, blood disorder, mental health disorders); pre‐pregnancy BMI > 50 consistent with morbid obesity; history of alcohol, drug abuse, or cigarette smoking; and reported intake of diuretics, pseudoephedrine, anticholinergics, warfarin (or any anticoagulant agent), oestrogen‐containing birth control pill or oestrogen‐containing device, selective serotonin reuptake inhibitors, or drugs/herbals used to induce milk production Breast feeding method: exclusive breastfeeding, most likely demand feeding with both direct and expressed milk feeding
Interventions	Arm 1: Mother's Milk Tea 3 to 5 cups per day for 4 weeks (n = 38) Arm 2: Placebo (Lemon Verbena leaf tea) 3 to 5 cups per day for 4 weeks (n = 34) Each tea bag of Mother's Milk Tea contained: 560 mg bitter fennel fruit PhEur (Foeniculum vulgare Miller ssp. vulgare var. vulgare, Apiaceae); 350 mg anise fruit PhEur (Pimpinella anisum L, Apiaceae); 210 mg coriander fruit PhEur (Coriandrum sativum L Apiaceae); 35 mg fenugreek seed PhEur (Trigonella foenum‐graecum L, Fabaceae); 35 mg blessed thistle herb DAC (Cnicus benedictus L, Asteraceae); 560 mg proprietary blend of flavoring botanicals in order of predominance: Spearmint leaf PhFr (Mentha spicata L, Lamiaceae), West Indian lemongrass leaf MFR (Cymbopogon citratus [DC. Ex Nees] Stapf, Poaceae), Lemon verbena leaf PhEur (Aloysia citrodora Palau, Verbenaceae), and marshmallow root PhEur (Althaea officinalis L, Malvaceae)
Outcomes	Safety and tolerance of Mother's Milk Tea among women and infants up to one year Quality of life Self‐perception of lactation sufficiency Perception of their infant’s satisfaction with the amount of milk Self‐reported milk volume using a scale from 0 to 10, with 0 being no milk to 10 being engorged Adverse effects for mother and infants
Funding and Declaration of interest	"Funding was provided in full, as an investigator‐initiated research study agreement 017066 ‐001 to MUSC, from Traditional Medicinals®, Sebastopol, CA, who also provided the coded teas used in the study. Neither the company, Traditional Medicinals®, Sebastopol, CA,nor employees of Traditional Medicinals® in any way at any point in time directed the study, altered the results, or influenced the study participants or research team, who were blinded to the study treatments until the study was completed.” One of the authors declared that he was a scientist at Traditional Medicinals and the six other authors declared no conflict of interest.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	" Women were randomised to MMT [Mother's Milk Tea] or placebo using a computer‐generated block design”
Allocation concealment (selection bias)	Low risk	"the tea had a randomised numeric code unknown to investigators, study team and participants."
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	We were unable to judge if the treatment and placebo teas would taste or smell different.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	We were unable to judge if the treatment and placebo teas would taste or smell different.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Unclear risk	We were unable to judge if the treatment and placebo teas would taste or smell different.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	This was not an outcome of the study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	9 out of 38 from the placebo group versus 3 out of 34 from intervention group dropped out. There was more attrition due to adverse effects in the placebo group. (Gardner 2020 [pers comm])
Selective reporting (reporting bias)	Low risk	All the prespecified main outcomes were reported.
Other bias	Low risk	None detected

Xu 2000.

Study characteristics
Methods	Randomised controlled trial in China Trial dates: May 1999 to October 1999
Participants	82 postcaesarean section mothers and their infants. 1 mother had a pair of twins. Age of infants at start of study: 8 to 10 days of life Inclusion criteria: postcaesarean section mothers without any complications with infants weighing 2500 g to 4150 g Exclusion criteria: not mentioned Breastfeeding method: this was not clearly reported. Authors reported that breastfeeding was encouraged for 5 to 20 minutes (10 minutes on average) before milk expression. There was no mention if supplemental feeding was allowed.
Interventions	Arm 1: pork leg soup (猪蹄汤) 300 mL every 2 to 3 hours beginning 6 hours postcaesarean section. Duration was not reported. The women were also encouraged to drink the soup as replacement for drinking water (n = 41). (pork leg soup was made with pork leg 500 g, salt 3 g, "a little spring onion" cooked in 3000 mL water) Arm 2: rice water and carrot soup (n = 41)
Outcomes	Milk volume at the 24th, 48th and 72nd hours post‐delivery. This was measured by the sum of 1) milk removal by the infant calculated mathematically from measurements of the difference in breast size before and after breastfeeding, and 2) the volume of residual milk expressed by hand after breastfeeding. Infant weight loss at first day of life Time to regain birthweight Number of infants at least regaining birthweight by eighth day of life
Funding and Declaration of interest	No funding or declaration of interest statement was found.
Notes	This study was published in Chinese. The above information was obtained from the English translation of the paper. No contact details of authors were available for clarification.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	".. women undergoing Caesarian section were randomly divided into two groups." How this was done was not reported.
Allocation concealment (selection bias)	Unclear risk	How this was done was not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Both soups have a distinct taste and smell and participants could not possibly be blinded. Blinding of personnel not described
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	Blinding of the outcome assessors (investigators or research assistants) were not reported.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	Self‐reported outcomes were not a part of this study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight measurements would unlikely be affected even if the outcome assessors were not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Low risk	All outcomes were reported in the results.
Other bias	Unclear risk	No baseline information was reported for the 2 groups.

Yabes‐Almirante 1996a.

Study characteristics
Methods	Randomised controlled trial in the Philippines Trial dates: November 1994 to September 1995
Participants	120 healthy mothers and their infants. 13.5% of mothers were primiparous. Age of infants at start of study: first day of life. The mean birthweight was 3033 grams. Inclusion criteria: healthy mothers who gave birth to healthy, term infants weighing 2500 g to 5000 g and did not plan to feed milk formula or introduce solid foods to their infants before 4 months of age Exclusion criteria: mothers with chronic illness and were taking any medication on a regular basis Breastfeeding method: "Infant suckling was started within 6 to 12 hours after delivery for a duration of 10 to 15 minutes to each breast, every 2 and a half to 3 hours for a total of eight to ten times a day." We assume supplemental feeding was not allowed as that was part of the inclusion criteria.
Interventions	Arm 1: Natalac capsules (moringa leaves) 250 mg 2 times a day for 4 months (n = 60) Arm 2: placebo capsules given in a similar manner (n = 60) *The capsules had been coded at source: 58 Natalac (NATC‐T) and 58 Placebo (NATC‐F)." Further description about the placebo was not available.
Outcomes	Infant's weight at birth, 2 weeks, 1 month, 4 months of age Serum prolactin levels before treatment Serum prolactin levels 48 hours after delivery Serum prolactin levels 4 months after delivery Time to breast engorgement Time to milk let down Adverse effects (not specified if for mothers or infants)
Funding and Declaration of interest	"Natalac capsules were provided by Tynor Health Supplement Division (a division of Tynor Drug House, Incorporated)." No declaration of interest statement was found.
Notes	Attempts to contact the authors for further clarification failed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	"Immediately after delivery, patients were given capsules, the content of which the researchers do not know. The capsules have been coded at source: 58 Natalac (NATC‐T) and 58 Placebo (NATC‐F)." How this was done was not mentioned.
Allocation concealment (selection bias)	Unclear risk	How this was done was not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	High risk	"Immediately after delivery, patients were given capsules, the content of which the researchers do not know. The capsules have been coded at source: 58 Natalac (NATC‐T) and 58 Placebo (NATC‐F)." Further description about the placebo was not available.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	Milk volume was not a part of this study.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	High risk	Time to breast engorgement and milk let down were likely to be influenced by lack of blinding.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Lack of blinding would unlikely affect measurements of infant weight and serum prolactin levels.
Incomplete outcome data (attrition bias) All outcomes	Low risk	4 out of the 120 participants were excluded from the analysis: 1 infant had died, 3 moved away.
Selective reporting (reporting bias)	Low risk	All outcomes mentioned in the methods were reported.
Other bias	Low risk	None detected

Yin 2005.

Study characteristics
Methods	Randomised controlled trial in China Trial dates: January 2004 to May 2005
Participants	200 mothers and their infants Age of infants at start of study: at birth Inclusion criteria: primipara mothers who had a normal singleton pregnancy Exclusion criteria: not mentioned Breastfeeding method: breastfeeding method was not mentioned but supplementary feeding was not allowed
Interventions	Arm 1: Sheng Ru He Ji (生乳合剂) 100 mL twice a day for 4 days. (n = 100) Arm 2: no intervention (n = 100) Sheng Ru He Ji (生乳合剂) contains 80 g zhu ti jia (猪蹄甲) and 20 g wang bu liu xing (王不留行)
Outcomes	Milk volume measured by hand expression in the morning and adding to another expression by hand and pump 4 hours later on baseline and fourth day postpartum. The mothers were not allowed to breastfeed their infants in between this interval. Serum prolactin levels Breast milk prolactin level Adverse effects (not specified if for mothers or infants)
Funding and Declaration of interest	No funding or declaration of interest statement was found.
Notes	The study was published in Chinese. The above information was obtained from the English translation of the article. No contact details of the authors were available.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	How this was done was not mentioned.
Allocation concealment (selection bias)	Unclear risk	How this was done was not mentioned.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding was not possible because the control group received no intervention.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	Blinding of the outcome assessor (a research assistant) was not reported.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	Self‐reported outcomes were not a part of this study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of this study.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were accounted for in the results.
Selective reporting (reporting bias)	Low risk	All expected outcomes were reported.
Other bias	Unclear risk	No baseline information was reported for the 2 groups.

Ylikorkala 1982.

Study characteristics
Methods	Randomised controlled trial in Finland Trial dates: not mentioned
Participants	28 mothers with lactation insufficiency and their infants. The mean parity of the mothers was 1.4 Age of infants at start of study: within the first 4 months of life. The average infant age at start of study was 58.8 days Inclusion criteria: women with lactation insufficiency (breast milk yield was less than 165 mL/kg/day) in the first 4 months after delivery Exclusion criteria: "Mothers with breast or other diseases possibly responsible for poor lactation..." Breastfeeding method: breastfeeding on demand. Supplemental feeding was allowed
Interventions	Arm 1: Sulpiride 50 mg 3 times a day for 4 weeks (n = 14) Arm 2: placebo given in a similar manner (n = 14)
Outcomes	Milk volume at baseline, third, fifth, seventh, 14th, 21st and 28th day of treatment measured by weighing the infant before and after breastfeeding. No mentioned if residual milk extraction was done Volume of supplementary feeds needed at baseline, third, fifth, seventh, 14th, 21st and 28th day of treatment Duration supplemental feeding needed (days) Infant weight gain Serum prolactin levels Adverse effects for mothers and infants
Funding and Declaration of interest	Sulpiride was donated by Leiras Ltd, Turku, Finland; a pharmaceutical company. No other declaration of interest reported.
Notes	Attempts to contact authors for further clarifications failed. The authors reported mean change in milk volume in the 2 groups in the form of a graph with a standard error (SE). This SE was converted to standard deviation (SD) using the Revman calculator (Review Manager 2014).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"The mothers were given consecutively numbered packages containing, in random order, either 50 mg sulpiride tablets or an identical‐looking placebo."
Allocation concealment (selection bias)	Low risk	"The mothers were given consecutively numbered packages containing, in random order, either 50 mg sulpiride tablets or an identical‐looking placebo."
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Mothers in the control group were given "..identical‐looking placebo."
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	"identical‐looking placebo"
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	"identical‐looking placebo"
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	"identical‐looking placebo"
Incomplete outcome data (attrition bias) All outcomes	Low risk	"Two women in the placebo group discontinued the trial because of lack of effect of the treatment and were excluded from the final analysis." We judged this to be of low risk of bias because the number was small and they were from the placebo group and was thus unlikely to be related to the intervention.
Selective reporting (reporting bias)	Low risk	All outcomes mentioned in the methods were reported.
Other bias	Low risk	None detected

Yulinda 2017.

Study characteristics
Methods	We judged this as a randomised controlled trial done in Indonesia
Participants	Number of mothers was not mentioned Age of infants at start of study: not mentioned Inclusion criteria: not mentioned Exclusion criteria: not mentioned Breastfeeding method:not mentioned. No mention if supplemental feeding was allowed.
Interventions	Arm 1: Palm date extract given for three days. Dose and frequency not mentioned Arm 2: No description on what the control group received
Outcomes	Prolactin levels Milk volume
Funding and Declaration of interest	No funding or declaration of interest statement reported
Notes	The study was published in Indonesian language and we translated the paper ourselves. Further clarification about the study was needed but attempts to contact the authors failed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	How this was done was not described.
Allocation concealment (selection bias)	Unclear risk	How this was done was not described.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	We are unable to judge because we do not know what the control group received.
Blinding of outcome assessment (detection bias) Milk volume outcomes	Unclear risk	We are unable to judge because we do not know what the control group received.
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	This was not an outcome of the study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	This was not an outcome of the study.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	There was no description of the flow of participants so we do not know what happened to all participants at the end of the study.
Selective reporting (reporting bias)	High risk	No standard deviation reported
Other bias	Unclear risk	No baseline values reported

Zarate 1976.

Study characteristics
Methods	Randomised controlled trial in Mexico. There were 2 parts to this study, the first part with 16 participants and a "conjoint study" with 9 participants. Trial dates: not mentioned
Participants	25 mothers and their infants (16 with no lactation insufficiency and 9 with lactation insufficiency) Age of infants at start of study: first part: 2 days of life; "conjoint study": 2 weeks of life Inclusion criteria: first part of the study: women with normal menstrual history, at least 1 previous pregnancy and no more than 2 deliveries, and had given proof of adequate lactation for more than 3 weeks in a previous prepuerium. and uncomplicated normal delivery. "Conjoint study": "...women who had been nursing for two weeks but showing a decreased in milk production." Exclusion criteria: not described Breastfeeding method: breastfeeding method was not described. Supplemental feeding was not given to participants in the first part of the study but was allowed for the participants in the "conjoint study."
Interventions	Arm 1: synthetic thyrotropin‐releasing hormone capsules 20 mg 3 times a day for 1 week (n = unknown for first part, 5 for "conjoint study") Arm 2: placebo (identical looking capsules) given in a similar manner (n = unknown for first part, 4 for "conjoint study")
Outcomes	Serum prolactin, follicle‐stimulating hormone and luteinizing hormone concentration before and after 30 minutes of suckling at baseline and end of intervention (first part) Milk composition (fat, protein and lactose) taken before and after 30 minutes of the first breastfeed at baseline and end of intervention (first part) Milk volume (method of measurement was not described) at baseline and end of intervention ("conjoint study") Adverse effects for mothers and infants
Funding and Declaration of interest	The study was funded in part by Academia Nacional de Medicina, and synthetic TRH was supplied by Farbwerke Hoechst of Frankfurt.
Notes	Number of participants was reported as 8 in the methods section for the "conjoint study" but in the results, data for 9 women were presented. We were unable to contact the authors for clarifications.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	How this was done was not described.
Allocation concealment (selection bias)	Unclear risk	How this was done was not described.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	"Identical capsules containing the placebo were utilized for the study"
Blinding of outcome assessment (detection bias) Milk volume outcomes	Low risk	"Identical capsules containing the placebo were utilized for the study"
Blinding of outcome assessment (detection bias) Self reported outcomes (adverse effects and measures of maternal psychological status)	Low risk	There were no self‐reported outcomes in the study.
Blinding of outcome assessment (detection bias) Infant weight outcomes	Low risk	Infant weight was not a part of the study.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	There was no description of the flow of participants so we do not know what happened to all participants at the end.
Selective reporting (reporting bias)	Unclear risk	The number of participants reported in the methods for the "conjoint study" did not tally with that which was reported in the results table. No other description about the methods used in this study apart from "Double‐blind trial."
Other bias	Low risk	None detected

BMI: body mass index

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Achalapong 2016	This was a 4‐arm randomized controlled trial. This clinical trial had 120 mothers given either extra eggs (n = 30) or extra milk (n = 30), or extra eggs and milk (n = 30) or regular diet (n = 30) in their first 3 postpartum days to see if increasing protein and caloric intake improves lactation. We do not consider nutrients that are critical to lactation to be general galactagogues.
Ahmed 2015	There was insufficient information in the methodology to determine if this was a randomised controlled trial. This clinical trial had 20 healthy mothers given either fenugreek capsule (n = 10) or placebo (n = 10). Outcomes measured were prolactin level and maternal weight.
Akhtar 1972	This was an observational study with no control group. 30 mothers with history of or currently experiencing lactation deficiency were given Leptaden (Jeevanti (Leptadenia reticulata) and Kamboji (Breynia patens)) and tracked to observe the ability to obtain full lactation.
Aljazaf 2003	Cross‐over trial looking at anti‐galactagogue effects of pseudoepinephrine
Amann 1966	This was a case report reviewing use of Agnolyt tincture. Agnolyt tincture contains Chasteberry (Vitex agnus‐castus).
Aono 1979	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This clinical trial had 130 healthy mothers given either oral sulpiride (n = 66) or placebo (n = 64). Outcomes included milk volume, prolactin level, infant weight gain, proportions of mothers exclusively breastfeeding, concentration of sulpiride in breast milk and composition of breast milk.
Aronova 1977	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This clinical trial had 130 healthy mothers given either Stachys sylvatica extract (Hedge nettle) (n = 75) or no treatment. (n = 55). Outcomes included milk volume, number of women with persistent low milk supply, time to cessation of breastfeeding and composition of breast milk.
Bakshi 1986	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This 3‐arm trial had 60 mothers with low milk supply allocated equally to Lactare (Shatavari (Asparagus racemosus), Ashwagandha (Withania somnifera), Licorice/Jeshtamadh (Glycyrrhiza glabra), Fenugreek/Methi (Trigonella foenum‐graecum) and Garlic/Lasun (Allium sativum)), metoclopramide or placebo. Outcomes included infant weight change, time to cessation of supplemental feeding, mother's milk ejection perception, serum prolactin and adverse effects.
Bautze 1953	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This study had 300 mothers who were given either Alyt 1, Alyt 2 or no treatment. Alyt is a Chasteberry (Vitex agnus‐castus) tincture. Outcomes included infant weight gain, proportions of mothers exclusively breastfeeding, milk volume and time to onset of lactogenesis II.
Bhandari 1979	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This study had 242 mothers who were given either Leptaden (Jeevanti (Leptadenia reticulata) and Kamboji (Breynia patens)) or nothing. Outcomes included infant weight gain and maternal improvement in breastfeeding experience.
Breier 1993	Growth hormone was administered subcutaneously, thus not an oral galactagogue.
Campbell‐Yeo 2007	Participants in this study were mothers with preterm infants. Mothers were given either domperidone or placebo. Outcomes included breast milk composition and volume, serum prolactin levels, infant weight and breastfeeding rates.
Chen 1995	This study involved human and animals and the reported results included animals. Mothers with lactation insufficiency were given Yangxueshengru Oral Liquor. Outcomes included improvement with breast milk secretion and breast milk composition.
Co 2002	Participants in this study were mothers with preterm infants. Mothers were given metoclopramide, domperidone and malunggay leaves (Moringa oleifera). Outcomes included milk volume, serum prolactin levels and adverse effects.
Damanik 2001	This was a survey reviewing use of torbangun (Coleus amboinicus Lour).
Damanik 2009	This was a focus group survey reviewing use of torbangun (Coleus amboinicus Lour).
Dastgerdi 2012	This study on metoclopramide looked at mothers with preterm infants.
De Leo 1986	The mothers in this trial were not randomized. This clinical trial had 32 mothers with either a current or past history of low milk supply. They were given either domperidone (multiparous women n = 8; primiparous women n = 9) or placebo (multiparous women n = 7; primiparous women n = 8). Outcomes included milk volume and plasma prolactin levels. (This study was published in Italian, the above information was obtained from the English translation of the article, and an independent translator confirmed that the study was not randomized.)
Dean 1950	This was an observational study. This study had 200 healthy mothers given either iodine solution mixed with milk or no treatment. Outcomes included daily milk yield, number of mothers who could exclusively breastfeed.
Demirci 2016	This observational study enrolled 11 women delivering late preterm or early term babies between 34 and 37 weeks gestation who intended to breast feed, did not have any known conditions that could potentially affect milk production, and had perceived low or insufficient milk. Mothers were randomised to 1 of 2 complementary alternative medicine treatments: meditation/relaxation exercises or Motherlove Herbal's More MIlk Plus Alcohol free tincture (fenugreek seed, blessed thistle, nettle, fennel seed, de‐ionized water, vegetable glycerin) for 9 days; there was no control group. Outcomes included average milk transfer by pre‐post feeding weights; average expressed milk volume, proportion of breast milk feeds, infant weight gain.
Deshpande 1962	This was a case series of 50 mothers treated with Leptaden (Jeevanti (Leptadenia reticulata) and Kamboji (Breynia patens)) tablets and observed for time to effect of intervention as well as "flow of milk".
Douglas 1962	This study looked at oxytocin which was administered via the buccal mucosa (not an orally ingested galactagogue).
Erb 1968	This study looked at oxytocin which was administered via the buccal mucosa (not an orally ingested galactagogue).
Ertl 1991	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This study tested metoclopramide versus no treatment in 20 healthy mothers with healthy infants. Outcomes included breast milk volume, milk prolactin concentration, milk sodium concentration, and plasma prolactin of the newborn.
Espenhain 1970	This study looked at oxytocin which was administered via the buccal mucosa (not an orally ingested galactagogue).
Filippova 1975	There was insufficient information in the methodology to determine if this was a randomised controlled trial. This study had 188 healthy mothers given either Stachys sylvatical liquid extract (SSLE‐V.N) (Hedge nettle extract) or no treatment. Outcomes included breast milk volume and breast milk composition.
Friedman 1961	This study looked at sublingual and intranasal oxytocin (not orally ingested galactagogues).
Geetha 1987	This was an observational study of 30 mothers taking Lactare (Shatavari (Asparagus racemosus), Ashwagandha (Withania somnifera), Licorice/Jeshtamadh (Glycyrrhiza glabra), Fenugreek/Methi (Trigonella foenum‐graecum) and Garlic/Lasun (Allium sativum)) with no control group.
Ghosh 1986	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This study had 90 healthy mothers given either Lactare (Shatavari (Asparagus racemosus), Ashwagandha (Withania somnifera), Licorice/Jeshtamadh (Glycyrrhiza glabra), Fenugreek/Methi (Trigonella foenum‐graecum) and Garlic/Lasun (Allium sativum)) or no treatment. Outcomes included prolactin levels, liver function tests, adverse effects and maternal perception of milk yield.
Gokhale 1965	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This study had 25 mothers with lactation failure or insufficiency who were given Leptaden.
Gupta 1966	This was an observational study of Leptaden (Jeevanti (Leptadenia reticulata) and Kamboji (Breynia patens)) with no control group. 150 patients were selected by previous poor lactation history and treatment (1 tablet 3 times daily) commenced between 32 and 38 weeks of pregnancy, continuing until birth, when the dosage was increased variably according to individual situations. Outcome was response to the drug, rated as good, partial or nil.
Guzman 1979	There was insufficient information in the methodology to determine if the first part of this study was a randomized controlled trial (the second part of the study was a cross‐over trial). This study had 21 healthy mothers with lactation insufficiency given either metoclopramide or placebo. Outcomes included prolactin levels and milk yield.
Gyõry 1968	This study of orgametril was about suppressing lactation rather than augmenting lactation
Hale 2009	This clinical trial of fenugreek was terminated before completion. Correspondence with the author revealed that only 2 participants were recruited and there were no data to share.
Heiss 1968	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This study had 200 mothers given either Galegran or no treatment. (Galegran was a commercial preparation extracted from Galega officinalis, phosphorus, calcium and ferric salts.) Outcomes included breast milk volume and breast milk composition.
Hofmeyr 1985	This study was not designed to determine the galactagenic effect of the domperidone. Outcome measured was mean serum levels of prolactin.
Huntingford 1961	This study looked at intranasal oxytocin (not an orally ingested galactagogue).
Huynh 2016	Milk supplement was used to increase the nutritional status of pregnant women with lower nutritional status in Vietnam. One of the outcomes of the study was exclusive breastfeeding at 12 weeks postpartum. We do not consider supplements that are critical to general health to be a galactagogue.
Ivanyi 2006	This clinical trial of domperidone was terminated before completion. No response from author when contacted to see if there were any usable data.
Janke 1941	This was an observational study of 30 mothers with perceived low milk production who were administered Oligoplex (Vitex agnus‐castus) and observed for changes in milk production. In addition, 6 wet‐nurses were administered Oligoplex off and on to observe effects of start and discontinuation of the drug.
Joglekar 1967	There was insufficient information in the title to determine if this was a randomized controlled trial looking at Shatavari (Asparagus racemosus). We do not have the abstract or full‐text of this study. All efforts to contact the authors have failed.
Kauppila 1981	This was a cross‐over study, not a randomized controlled trial. This study had 45 mothers with lactation insufficiency given 1 of 3 different doses of metoclopramide (5, 10 or 15 mg) and placebo. Outcomes included prolactin levels, breast milk production, need for supplementary feeds, and adverse events.
Kavurt 2013	This study was not designed to determine the galactagenic effect of Humana still‐tee. Outcomes included total antioxidant capacity, total oxidant status and the oxidative stress index of breast milk samples.
Kawakami 2003	This was an observational study of 4 different combinations of Kampo herbs and breast massage on 72 mothers with perceived low milk production.
Keldenich 1976	There was insufficient information in the methodology to determine if this 3‐arm study was a randomised controlled trial. 133 mothers with hypogalactia were graded by severity of milk production deficit (based on quantity of milk produced the third and seventh post intervention days). Those trying to breast feed were assigned to either placental extract Moloco or placebo, while those who stopped breastfeeding served as controls to compare growth against formula‐feeding. Main outcomes were changes in milk output determined by pre‐ and ‐post feeding test weights at 2 time points, average daily milk intake during last 3 days of treatment, and infant weight gain.
Knoppert 2013	Participants in this non‐randomised study were mothers of preterm infants of less than 33 weeks' gestation. This study had mothers given two different doses of domperidone, and had decreasing frequency of domperidone administration. Outcome measured was milk volume.
Lal 1980	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This study had 100 mothers with lactation insufficiency given either Leptaden (Jeevanti (Leptadenia reticulata) and Kamboji (Breynia patens)) or placebo. Outcomes included infant weight gain, breast milk flow and breast milk composition.
Lewis 1980	This was a randomized controlled trial of 20 mothers who underwent either elective or emergency Caesarean section and then were randomly assigned to either metoclopramide (n = 10) or placebo (n = 10). However, a high percentage of the infants of these mothers were either premature or ill, and had to be nursed in the intensive care ward.
Luhman 1963	This study looked at intranasal oxytocin (not an orally ingested galactagogue).
Mennella 1991	This study was not designed to determine the galactagenic effect of garlic. The outcome measured was the odour of the mother’s breast milk and the suckling behaviour of her infant (including amount of milk taken) 4 hours after taking the intervention.
Mennella 1993	This study was not designed to determine the galactagenic effect of the garlic. This was a 3‐arm study using garlic capsules as the intervention. Group 1 (n = 10) received placebo and Groups 2 and 3 (both groups n = 10) received the garlic capsules for 4 days but on different days. At the end of the study all 3 groups received the garlic capsules before the final outcome measurement (time at the nipple, total milk intake and number of feeding in 4 hours) was taken. This study was excluded because all 3 groups received the intervention.
Milsom 1992	This study looked at growth hormone which was administered subcutaneously (not an oral galactagogue).
Milsom 1998	This study looked at growth hormone which was administered subcutaneously (not an oral galactagogue).
Mohr 1954	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This clinical trial had 715 healthy mothers given Agnolyt tincture in the treatment group (n = 353). We do not know what the control group received (n = 362). Outcomes included milk volume, milk composition, number of mothers who could breast feed and averse effects.
Narimatsu 2001	This randomized controlled trial in Japan had 80 healthy mothers given either Kyuki‐choketu granules (Xiong‐gui‐tiao‐xue‐yin in Chinese) or 'hysterotonics'. Outcomes included uterine contraction pain, milk volume, infant weight loss, number of mothers exclusively breastfeeding at 1 month, infant weight gain, episodes of mastitis, post partum depression, adverse events in mother and infant. 'Hysterotonics' are likely substances that promote uterine contractions such as ergometrine. 1 of the authors in this study conducted another study (Ushiroyama 2007) which also used Kyuki‐choketu as intervention and ergometrine as placebo. This study was excluded because ergometrine, which was used as the placebo, is a breast milk suppressant.
Nicholson 1948	This study's reported methodology did not meet current definition of a randomized controlled trial. 43 mothers with lactation failure at fifth postpartum day were given Lugol's solution (5% iodine in 10% aqueous potassium iodide) "to alternate cases". The outcome measured was mean daily milk yield.
Noack 1943	Based on the English summary, this was an observational study. 125 early postpartum mothers identified with insufficient milk production after "feeding more often" 6 times per day and pumping were given Oligoplex (Vitex agnus‐castus) tincture and their ability to reach full breastfeeding versus partial (some supplementation) was recorded.
Nommsen‐Rivers 2019	This was a randomized controlled trial comparing metformin to placebo in 15 mothers with signs of insulin resistance. Excluded because 60% of participants in the intervention group had a co‐intervention (fenugreek) compared to only 20% in the control group.
Patel 1982	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This clinical trial had 60 healthy mothers given either Leptaden (n = 30) or placebo (n = 30). (17% of the infants were preterm infants with a birthweight of less than 2500 g).Outcomes include number of women with 21 to 60 g, 61 to 120 g and 121 to 180 g increase in breast milk before and after intervention, measured by weighing infants before and after nursing; and adverse effects.
Peters 1991	This study looked at thyrotrophin‐releasing hormone which was administered by nasal spray (not an oral galactagogue).
Petraglia 1985	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This clinical trial had 32 mothers who were divided into 2 groups: Group A (n = 15) were multiparous mothers with a prior history of failure of lactogenesis II. Group B (n = 17) primiparous mothers with inadequate lactation at 2 weeks postpartum. They were given either domperidone or placebo. Outcomes included plasma prolactin levels and change in daily milk volume.
Pontuch 1970	This study looked at oxytocin which was administered via the buccal mucosa (not an orally ingested galactagogue).
Qi 1996	There was insufficient information in the methodology to determine if this was a randomized controlled trial. 30 mothers with lactation insufficiency given 125 g Mu‐ying‐le per day compared with 30 mothers who could choose any traditional Chinese galactagogue. Outcome was self‐reported efficacy of the intervention.
Rajarathnam 1986	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This study had 75 mothers with lactation insufficiency given either Lactare (Shatavari (Asparagus racemosus), Ashwagandha (Withania somnifera), Licorice/Jeshtamadh (Glycyrrhiza glabra), Fenugreek/Methi (Trigonella foenum‐graecum) and Garlic/Lasun (Allium sativum)) or placebo. Outcomes included breast milk yield enhancement, prolactin levels and adverse effects.
Rath 1983	This was an observational study of 20 mothers with lactation insufficiency given metoclopramide. Outcomes included decrease in supplementation.
Reeder 2011	This trial on fenugreek was on mothers with premature infants.
Robinson 1947	This was an observational study of 78 mothers with lactation failure. 27 mothers were given Lugol's solution (5% iodine in 10% aqueous potassium iodide taken with milk twice daily), 11 were given breast massage, 19 were given intramuscular saline and 21 had no treatment. Outcome measured was mean daily milk yield.
Rolfini 1989	This 3‐armed trial study was published in Italian and all information was derived from an English translation of the paper. 30 mothers with delayed lactogenesis II or prior history of lactation insufficiency were given either ferolactan (human prolactin, Pfizer, 1 vial intramuscular for 10 days), metoclopramide (10 mg 3 times per day for 3 weeks) or domperidone (10 mg 3 times per day, third till sixth postpartum days). No control was indicated and there was insufficient information in the methodology to determine if this was a randomised controlled trial. In addition, ferolactin was administered intramuscularly, and the duration of treatment for each type varied. Outcomes included milk volume.
Ruis 1981	This was a trial looked intranasal oxytocin for enhancement of the onset of lactation among mothers with premature infants.
Sapak 1969	Based on English summary, this was an observational study that looked at various substances including luteotropin, Sol. lugoli, hydrocortisonacetate, insulin and superlutin.
Seema 1997	This study was done on mothers who were attempting relactation for their sick hospitalised infants. This study had 50 mothers with partial or complete lactation failure given either metoclopramide or standard care. Outcomes included time to relactation, infant weight and amount of supplemental feeding.
Sepehri 1998	This study was not designed to determine the galactagenic effect of the pectin‐rich plant extract. The outcome measured was the C3 and C4 complement concentration and antibacterial effect of in women's colostrum.
Sholapurkar 1986	This was an observational study of 10 women with "scanty breast milk" who were given Lactare (Shatavari (Asparagus racemosus), Ashwagandha (Withania somnifera), Licorice/Jeshtamadh (Glycyrrhiza glabra), Fenugreek/Methi (Trigonella foenum‐graecum) and Garlic/Lasun (Allium sativum)). Maternal perceptions of improvement in milk output were recorded.
Srinivas 2014	This clinical trial in India had 30 mothers given either fenugreek, garlic or galactagogue mix. Correspondence with the author revealed that this was not a randomised trial. "Those who were willing with the their consent to consume galactagogues were assigned into experimental groups and the ones who expressed their willingness to participate in the study but did not want consume any supplements were considered for control group." Outcomes included maternal prolactin levels, infant catch up weight, mother's perception of efficacy of milk production.
Stegaĭlo 1980	This was an observational study looking at the effects of an extract of betonica hedge nettle (Stachys betonicaeflora Rupr) on milk production and composition in hypogalactic women.
Subramaniam 1986	There was insufficient information in the methodology to determine if this was a randomized controlled trial. This study had 165 mothers with lactation insufficiency given either Lactare [Shatavari (Asparagus racemosus), Ashwagandha (Withania somnifera), Licorice/Jeshtamadh (Glycyrrhiza glabra), Fenugreek/Methi (Trigonella foenum‐graecum) and Garlic/Lasun (Allium sativum)] or placebo. Outcomes included breast milk yield enhancement, prolactin levels and adverse effects.
Tablb 1977	There was insufficient information in the title to determine if this was a randomised controlled trial looking at Leptaden. We do not have the abstract or full text of this study. All efforts to contact the authors have failed.
Tagliareni 1977	This was an observational study of 70 agalactic/hypogalactic mothers given arginine aspartate twice a day. Changes in milk output were recorded and outcomes grouped as 'good' (60%), 'great' (20%) or 'no change' (20%).
Thummel 1969	This study looked at oxytocin which was administered via the buccal mucosa (not an orally ingested galactagogue).
Toaff 1969	This study was not designed to determine the galactagenic effect of the intervention. The outcome measured was the effects of oestrogen and progestagen on the composition of human milk.
Trivedi 1956	There was insufficient information in the title to determine if this was a randomized controlled trial. We do not have the abstract or full text of this study. All efforts to contact the authors have failed.
Tustanofsky 1996	This was a galactagogue review, not a clinical trial.
Typl 1961	This was an observational study of 336 cases of primary or secondary hypogalactia treated with Galegran, (main component Galega officinalis). Outcomes recorded was the increase in milk production.
Ushiroyama 2007	This randomized controlled trial in Japan had 82 healthy mothers given either Xiong‐gui‐tiao‐xue‐yin (Kyuki‐choketu granules in Japanese) (Kanebo Pharmaceutical Co. Ltd. Tokyo, Japan) or ergometrine (methylergometrine maleate). Outcomes included milk volume, plasma prolactin and oxytocin levels. This study was excluded because ergometrine, which was used as the placebo, is a breast milk suppressant.
Vogulkina 1966	This was an observational study to see the stimulating effects of glutamic acid on milk production in 65 women with low milk supply presenting variously in the first week, first month, second month or third month.
Volet 1965	This study looked at oxytocin which was administered via the buccal mucosa (not an orally ingested galactagogue).
von Jaisle 1958	This study looked at Obron given to mothers in the first 10 postpartum days. Obron is a multivitamin, mineral and iron containing capsule. We do not consider nutrients that are critical to lactation to be general galactagogues.
Yabes‐Almirante 1996b	The majority of participants in this study on malunggay had infants born at less than 37 weeks' gestation.
Ylikorkala 1984	There was insufficient information in the methodology to determine if this was a randomised controlled trial. This clinical trial had 36 healthy mothers and the researchers looked at the additive effects of oxytocin to sulpiride versus placebo.
Zecca 2016	This prospective, double‐blind randomized trial recruited mothers of preterm infants of 27 to 33 weeks' gestation. 100 mothers were randomised to receive either an Intervention product consisting of silymarin phosphatidylserine and galega (goat's rue) or a placebo. Outcomes included daily and total milk volumes, with a target of 200 mL per day.
Zhang 1987	This was an observational study of various methods to promote breast milk production. Interventions used included acupoint injections and breast massages which were not oral galactagogues.
Zhang 1996	This study observed 2 groups of mothers: 1 group was given a combination of 4 to 5 important elements: early breast massage, frequent breast massage and nipple stimulation, early skin to skin, collagen soup and strict daily activities/routine. The other group was just given 'breast care'. The intervention group received a combination of interventions of which were mostly not oral galactagogues but the control group had none of the co‐interventions.
Zhu 2005	This was an observational study on sesame containing food.

Characteristics of ongoing studies [ordered by study ID]

ACTRN12619000704190.

Study name	Randomised controlled study of the effects of yeast based supplement on milk production in breastfeeding women
Methods	Randomised controlled trial in New Zealand
Participants	Inclusion criteria Healthy infants 1 to 4 months, breastfeeding, either directly from the breast, with expressed milk or mixed feeding (1 to 2 formula feeds a day with the total amount not over 100 mL) Healthy participants, minimum 16 years age Exclusion criteria Started complementary feeding, allergy to yeast, taking medicines such as Phenelzine (Nardil), Tranylcypromine (Parnate), Selegiline (Ensam, Eldepryl), Isocarboxazid (Marplan) and Meperidine (Demerol) or any other medications containing monoamine oxidase inhibitors, having health conditions or taking medications that can influence milk secretion related hormones, or milk supply, Crohn’s disease, diabetes, compromised immunity, treatment for fungal infections
Interventions	Arm 1: yeast powder 5g/day (9 capsules) for 28 days Arm 2: placebo (starch)
Outcomes	Milk volume Breastfeeding pattern Postnatal distress Perceived insufficient milk production The duration of exclusively breastfeeding Side effect Milk fat, IgA, IL‐6, IL‐10, IGF‐1, leptin, ghrelin, protein Infant weight, length, head circumference
Starting date	26 February 2019
Contact information	Public queries: Janet Weber, School of food and advanced technology, College of Science, Massey University Private Bag 11222, Palmerston North 4442, +6463569099 Ext 84562, J.L.Weber@massey.ac.nz Scientific queries: Ms Li Li Jia, School of Food and Advanced Technology, College of Science, Massey Univeristy, Private Bag 11222, Palmerston North 4442, New Zealand, l.jia@massey.ac.nz, +6469516367
Notes

CTRI/2016/01/006547.

Study name	A clinical study to evaluate the stanyjanana (galactagogue) effect of promolact capsules and granules
Methods	Randomised controlled trial in India
Participants	Mothers with insufficient breast milk Target: 80 mothers. Inclusion criteria: 18 to 40 year old mothers who delivered at term with the presence of stana mlanata 6 hours after feeding, ...,"absence of dripping down of milk following a long term non‐feeding," pumping yield less than 40 mL per session Exclusion criteria: mothers with history of serious illness, infectious disease, mothers on "medications like AED," infant with congenital anomalies, preterm infants
Interventions	Arm 1: Promolact capsules 2 capsules twice daily with milk for 10 days Arm 2: Promolact granules 1 table spoon twice daily for 10 days Promolact contains Jivanti, Kamboji, Vidarikand, Shatavari, Amalaki, Methi (fenugreek), Godanti bhasma and Suwa
Outcomes	Adequate breast milk after 4 days Maternal and infant satisfaction Improved immunity for the infant Maternal haemoglobin levels
Starting date	19 June 2015
Contact information	Dr KS Patel (drkspatel2007@yahoo.co.in)
Notes	No response from author regarding status of the study

JPRN‐UMIN000027159.

Study name	Effect of chicken extract on breast milk production of primiparous mothers in Japan: a randomised experimental study
Methods	Randomised controlled trial in Japan
Participants	Primiparous mothers with insufficient breast milk Target sample size: 80 mothers Inclusion criteria: primiparous women who wanted to breastfeed, were healthy, and whose pregnancies proceeded normally Exclusion criteria A) Exclusion criteria before commencement of test: chicken allergy, morphological abnormalities of nipple, caesarean section, hypertension, diabetes, mental illness, anaemia (person with Hb of less than 9.0 mg/dL), medicating medicine, judged by physicians to be inappropriate for participation B) Cancellation criteria after commencement of test: before 38 weeks gestation, cases where they wish to stop the examination, special mother's or child's abnormality, caesarean section, participants diagnosed by doctors and deemed unsuitable for continued testing
Interventions	Arm 1: chicken extract 70 mL twice a day for at least 2 weeks when the mother is at the 36th week of pregnancy Arm 2: water taken in a similar manner
Outcomes	1. Breast milk production on postpartum days 2 and 4
Starting date	1 May 2011
Contact information	Name: Masayo Awano Phone: 076‐265‐2500 Email: masayo@po2.nsknet.or.jp
Notes	Sponsored by NPO Science Research Center Alternative Medicine

NCT00264719.

Study name	Metoclopramide to aid establishment of breastfeeding: a randomised controlled trial
Methods	Randomised control trial in Singapore
Participants	Target: 160 Inclusion criteria: all pregnant women who intend to breastfeed, from 28 weeks to term, who have not met the exclusion criteria Exclusion criteria Patients who have epilepsy or on anti‐seizure medications Patients who have a history of significant depression or are on antidepressant drugs Patients who have pheochromocytoma or uncontrolled hypertension Patients who have intestinal bleeding or obstruction Patients who have a known allergy or prior reaction to metoclopramide, or any other contraindications to the use of metoclopramide Patients who have diabetes and hyperprolactinaemia Patients with HIV infection Current pregnancy complicated by foetal congenital anomalies and multiple foetuses
Interventions	Arm 1: oral metoclopramide 10 mg 3 times a day for 7 days followed by 2 times a day for the eighth to 10th day, and once a day for the 11th to 12th day (total duration of intervention 12 days) Arm 2: placebo given in a similar manner
Outcomes	Successful initiation of lactation, as determined by lactogenesis II markers at 7 days postpartum Weight change in infant 7 days after birth at 14 days postpartum Breastfeeding status at 14 days, 6 weeks, 3 months and 6 months after delivery
Starting date	January 2006
Contact information	YS Chong, Department of Obstetrics and Gynaecology, National University Singapore
Notes	Personal communication with 1 of the co‐investigators confirmed that the study has been completed and we will be informed when the study is published (Mattar 2017 [pers comm]).

NCT00477776.

Study name	Metoclopramide to improve lactogenesis II in diabetic women: a randomised controlled trial
Methods	Randomised control trial in Singapore
Participants	Target: 160. Inclusion criteria: all pregnant women with pregestational or gestational diabetes under diet or insulin control Exclusion criteria: Patients who have epilepsy or on anti‐seizure medications Patients who have a history of significant depression or are on antidepressant drugs Patients who have pheochromocytoma or uncontrolled hypertension Patients who have intestinal bleeding or obstruction Patients with known allergy or prior reaction to metoclopramide Patients with HIV infection Current pregnancy complicated by foetal congenital anomalies and multiple foetuses
Interventions	Arm 1: oral metoclopramide 10 mg 3 times a day for 7 days followed by 2 times a day for the 8th to 10th day, and once a day for the 11th to 12th day (total duration of intervention 12 days) Arm 2: placebo given in a similar manner
Outcomes	Primary outcomes Successful initiation of lactation, as determined by lactogenesis II markers Maternal perception and timing of successful establishment of lactogenesis II Secondary outcomes Amount of breast milk, determined by test weighing and expressed milk volumes, weight change on day 7 and breastfeeding status up to 6 months
Starting date	April 2006
Contact information	YS Chong, Department of Obstetrics and Gynaecology, National University Singapore
Notes	Personal communication with one of the co‐investigators confirmed that the study has been completed and we will be informed when the study is published (Mattar 2017 [pers comm]).

NCT02190448.

Study name	Randomised, placebo‐controlled study of an herbal tea to support lactation
Methods	Randomised controlled trial
Participants	Target: 60 mothers Women between the ages of 18 and 45 years who are healthy with term, singleton births 2 to 12 weeks postpartum who is successfully breastfeeding exclusively at the time they enter the study and intend to fully breastfeed their infants for the following 4 weeks at the time of enrolment
Interventions	Arm 1: herbal galactagogue tea (Mother's Milk Tea) 3 to 5 cups (8 ounces each) per day for 4 weeks Arm 2: placebo herbal tea 3 to 5 cups (8 ounces each) per day for 4 weeks
Outcomes	Primary outcomes To determine quality of life with the following measurement tools The Satisfaction with Life Scale ‐ an overall global life satisfaction questionnaire Secondary outcomes Oxytocin in maternal blood Safety Prolactin in maternal blood Composition of breast milk, measuring the quantities and quality of breast milk
Starting date	March 2013
Contact information	Bernadette Marriott, Medical University of South Carolina, email: marriobp@musc.edu
Notes	The authors have informed us that they completed the study in August 2015, with the exception of the follow‐up phone calls to the mothers concerning continuation of breastfeeding their infants. This follow‐up phase of the study will be completed in March 2016. They will be working on the manuscript for publication and inform us when the study is published. Personal communication with one of the co‐investigators confirmed that the study has been completed and we will be informed when the study is published (Marriott 2016 [pers comm]).

NCT02233439.

Study name	Double‐blind, placebo controlled randomised trial on the efficacy of herbal galactagogues
Methods	3‐arm randomised controlled trial in Italy
Participants	Target: 210 mothers recruited through the University Hospital Ospedale di Circolo e Fondazione Macchi, Garrison f. Del Ponte Varese, Italy Inclusion criteria: singleton, term delivery, greater than 2.5 kg newborn weight, lactation deficiency Exclusion criteria: neonatal intensive care unit admission, use of galactagogue drugs, allergy
Interventions	Arm 1: Piùlatte Plus (MILTE ITALIA SpA), a combination product of Silybum marianum 400 mg and Galega officinalis 150 mg, once a day for 6 weeks Arm 2: placebo identical in appearance to galactagogue product Arm 3: usual care, no treatment
Outcomes	Primary outcome Percentage of mothers who are exclusively breastfeeding their child at 6 weeks postpartum Secondary outcomes Rate of breastfeeding (exclusive or supplementing) at 6 weeks and 3 months Volume and frequency of formula use at 6 months postpartum Infant weight gain at 6 weeks postpartum Serum prolactin level at baseline and after 6 weeks of treatment Rate of maternal allergic reactions and gastrointestinal side effects during treatment
Starting date	September 2014
Contact information	Antonella Cromi, Università degli Studi dell'Insubria, email: antonella.cromi@uninsubria.it
Notes	Estimated completion date was reported as March 2015. No response from author when contacted regarding status of the study

NCT02740751.

Study name	Efficacy of herbal galactagogues on weight gain of the newborns within the first month of life in breastfeeding mothers
Methods	3‐arm randomised controlled trial in Turkey
Participants	Target: 90 mothers Inclusion criteria Infants with gestational age of 35 to 42 weeks breastfed infants Lactating mothers Infants admitted to outpatient clinics of neonatology within the first week of life Exclusion criteria Infants with gestational age of less than 35 weeks Non‐breastfeeding mothers Infants admitted to outpatient clinics of neonatology within the first week of life
Interventions	Arm 1: Still Tee (Mamsel) 3 cups (200 mL each) daily for 4 weeks Arm 2: placebo tea Arm 3: no intervention. The mothers will receive water.
Outcomes	Primary outcome: infant weight gain Secondary outcome: breast milk volume
Starting date	April 2016
Contact information	Dilek Dilli, email: mailto:dilekdilli2%40yahoo.com?subject=nct02740751, samiulusch‐trials‐stilltee, efficacy of herbal galactogogues in breastfeeding mothers
Notes	Study in process, completion estimated April 2017

TCTR20170811003.

Study name	Efficacy of Wong Nam Yen Herbal Tea on breast milk production: a factorial randomised controlled trial (Tea4Milk)
Methods	Randomised controlled trial in Thailand
Participants	Inclusion criteria: 15 to 40 years old, GA 28‐40 weeks, caesarean delivery Exclusion criteria: HIV infection, severe intrapartum complication, severe PPH, eclampsia, history of domperidone and herbal tea allergy
Interventions	Arm 1: Wan Nam Yen herbal yea and placebo domperidone Arm 2: Domperidone and placebo Wan Nam Yen herbal tea Arm 3: Placebo domperidone and Placebo Wan Name Yen herbal tea
Outcomes	Breast milk volume Nausea and vomiting Palpitations
Starting date	14 February 2017
Contact information	Koollachart Saejueng 0813924761 knot.md.28@gmail.com Sunpasithiprasong hospital, 122 Sunpasit roast Amphur, Ubonrachathani, 34000, Thailand
Notes

TCTR20180808007.

Study name	Effectiveness of Prasaplai as a galactagogue
Methods	Randomised controlled trial in Thailand
Participants	Inclusion criteria: 18 to 40 years old, singleton, healthy, birth weight 2500 g to 4000 g Exclusion criteria: contraindications for breastfeeding, medical disease affecting breastfeeding, twins, congenital anomalies
Interventions	Arm 1: Prasaplai capsule Arm 2: placebo (corn starch capsule)
Outcomes	Milk volume at 48 hours postpartum Side effects at 48 hours postpartum
Starting date	8 August 2018
Contact information	Public query: Panurak Ketpong, 0858316835, exitbird1991@gmail.com, 40/25.M.10Banlen Bang Pa‐In Scientific query: Worrawan Sirichai, 0639715942, ob_gyn@ymail.com
Notes

TCTR20190218004.

Study name	Effectiveness of Ayurved Siriraj Prasa‐Nam‐Nom recipe on breast milk volume in early postpartum women: a randomised, double‐blind, placebo‐controlled trial (Prasa‐Nam‐Nom)
Methods	Randomised controlled trial in Thailand
Participants	Target Sample Size: 54 mothers Inclusion criteria 18 to 80 years First vaginal delivery at term Inadequate milk volume 0 to 49 mL Exclusion criteria Medical disorders, such as hypertension type 2 diabetes, thyrotoxicosis and hypothyroidism, allergy to the intervention or its ingredients Inability to eat vegetables or spicy food Severe breastfeeding problems, such as abnormal breast anatomy, short nipples, mastitis, breast abscess, and severe foetal tongue tie Taking galactagogues, such as domperidone, metoclopramide, antidepressive drugs caffeine and alcohol
Interventions	Arm 1: Ayurved Siriraj Prasa‐Nam‐Nom capsule Arm 2: placebo capsule
Outcomes	Primary outcomes: volume Secondary outcomes: milk quality, prolactin II assay, safety
Starting date	26 August 2012
Contact information	Thapthep ‐ Thippayacharoentam +66+02+4198906 reab.tip@mahidol.edu
Notes

TCTR20190716001.

Study name	The clinical study of Lysiphyllum Strychifolium on breast milk production
Methods	Randomised controlled trial in Thailand
Participants	Target sample size: 84 Inclusion criteria Females, 20 to 40 years old, nursing women Normal delivery, exclusive breastfeeding No postpartum haemorrhage, haematocrit less than 10 of the original haematocrit before birth and blood loss less than 1000 cc No history of drug food or herb allergy Consent to join the study and consent to get the serum prolactin levels checked and have normal dietary intake Healthy full‐term neonates Exclusion criteria Chronic illness, such as diabetes, hypertension, bronchial asthma, and any allergies Breast problems, such as inverted nipples, mastitis, engorgement cracks, etc. History of smoking, alcohol or any drug use for improving breast milk production Twins Hepatitis B, AIDS, syphilis Low birth weight Low APGAR scores and or intrauterine growth retardation Congenital abnormalities
Interventions	Arm 1: Lysiphyllum strychnifolium tea Arm 2: warm water
Outcomes	Primary outcomes: quality of breast milk at day 4 and day 10, breast milk volume on day 10 Secondary outcomes: plasma prolactin level, oxytocin level, infants weight, quality of breast milk on day 10
Starting date	13 May 2019
Contact information	Scientific query: Somboon Kietinun, 085‐0645224, sbk9749@hotmail.com Public query: Suwanna Maenpuen, 089‐0695437, maenpuen.s@windowslive.com
Notes

AED: anti‐epileptic drugs

Hb: haemoglobin

PPH: postpartum haemorrhage

Differences between protocol and review

We made several changes from our protocol (Foong 2015) during the course of this review.

1. We edited the title to better reflect what the review is about.

2. We updated the background information.

3. We rearranged comparisons by type of galactagogue to have a more meaningful comparison.

4. We made minor changes to the words specified in the protocol for the primary outcomes to reflect the evidence across studies: “Proportion of mothers who continued breastfeeding” instead of “Proportion of infants breastfeeding;” “Infant weight in trials where the infants received only own mother's milk (g) at latest time measured” instead of “Infant weight gain (g/week; in trials where formula milk supplementation was not used);” “Volume of breast milk at the latest time measured (mL)” instead of “Volume of breast‐milk expressed in a specified time.”

5. We included the methods for assessing certainty of evidence (GRADE approach) and 'Summary of findings' tables.

6. We added three subgroup analyses: age of the infant when the outcome was measured; mothers with lactation insufficiency; and specific individual galactagogues within each comparison.

7. We added a statement about reporting the results of subgroup analyses by quoting the Chi² statistic and P value, and the interaction test I² value.

Contributions of authors

Siew Cheng Foong is the guarantor for the review. Siew Cheng Foong, May Loong Tan, Wai Cheng Foong, Lisa A Marasco, and Jacqueline J Ho all contributed to the overall writing of this review. Joo Howe Ong helped substantially with data extraction, data handling and translation of Chinese language papers.

Sources of support

Internal sources

• RCSI & UCD Malaysia Campus (formerly Penang Medical College), Malaysia

External sources

• No sources of support supplied

Declarations of interest

Siew Cheng Foong: none known
May Loong Tan: none known
Wai Cheng Foong: none known
Lisa A Marasco is co‐author of a breastfeeding book 'Making More Milk: The Breastfeeding Guide to Increasing Your Milk Production' (published by McGraw‐Hill) which discusses galactogogues. She also speaks on various lactation‐related topics, including galactogogues, providing training and continuing education units to lactation consultants and other health care providers.
Jacqueline J Ho is co‐coordinator of the local governance board of the World Alliance for Breastfeeding Action (WABA). A member of her family works for a pharmaceutical company.
Joo Howe Ong: none known

New