Pharmacological and non‐pharmacological strategies for obese women with subfertility

Abstract

Background

Clinicians primarily recommend weight loss for obese women seeking pregnancy. The effectiveness of interventions aimed at weight loss in obese women with subfertility is unclear.

Objectives

To assess the effectiveness and safety of pharmacological and non‐pharmacological strategies compared with each other, placebo, or no treatment for achieving weight loss in obese women with subfertility.

Search methods

We searched the CGF Specialised Register, CENTRAL, MEDLINE, Embase, PsycINFO, and AMED from inception to 18 August 2020. We also checked reference lists and contacted experts in the field for additional relevant papers.

Selection criteria

We included published and unpublished randomised controlled trials in which weight loss was the main goal of the intervention. Our primary effectiveness outcomes were live birth or ongoing pregnancy and primary safety outcomes were miscarriage and adverse events. Secondary outcomes included clinical pregnancy, weight change, quality of life, and mental health outcome.

Data collection and analysis

Review authors followed standard Cochrane methodology.

Main results

This review includes 10 trials. Evidence was of very low to low quality: the main limitations were due to lack of studies and poor reporting of study methods. The main reasons for downgrading evidence were lack of details by which to judge risk of bias (randomisation and allocation concealment), lack of blinding, and imprecision.

Non‐pharmacological intervention versus no intervention or placebo

Evidence is insufficient to determine whether a diet or lifestyle intervention compared to no intervention affects live birth (odds ratio (OR) 0.85, 95% confidence interval (CI) 0.65 to 1.11; 918 women, 3 studies; I² = 78%; low‐quality evidence). This suggests that if the chance of live birth following no intervention is assumed to be 43%, the chance following diet or lifestyle changes would be 33% to 46%. We are uncertain if lifestyle change compared with no intervention affects miscarriage rate (OR 1.54, 95% CI 0.99 to 2.39; 917 women, 3 studies; I² = 0%; very low‐quality evidence). Evidence is insufficient to determine whether lifestyle change compared with no intervention affects clinical pregnancy (OR 1.06, 95% CI 0.81 to 1.40; 917 women, 3 studies; I² = 73%; low‐quality evidence). Lifestyle intervention resulted in a decrease in body mass index (BMI), but data were not pooled due to heterogeneity in effect (mean difference (MD) ‐3.70, 95% CI ‐4.10 to ‐3.30; 305 women, 1 study; low‐quality evidence; and MD ‐1.80, 95% CI ‐2.67 to ‐0.93; 43 women, 1 study; very low‐quality evidence).

Non‐pharmacological versus non‐pharmacological intervention

We are uncertain whether intensive weight loss interventions compared to standard care nutrition counselling affects live birth (OR 11.00, 95% CI 0.43 to 284; 11 women, 1 study; very low‐quality evidence), clinical pregnancy (OR 11.00, 95% CI 0.43 to 284; 11 women, 1 study; very low‐quality evidence), BMI (MD ‐3.00, 95% CI ‐5.37 to ‐0.63; 11 women, 1 study; very low‐quality evidence), weight change (MD ‐9.00, 95% CI ‐15.50 to ‐2.50; 11 women, 1 study; very low‐quality evidence), quality of life (MD 0.06, 95% CI ‐0.03 to 0.15; 11 women, 1 study; very low‐quality evidence), or mental health (MD ‐7.00, 95% CI ‐13.92 to ‐0.08; 11 women, 1 study; very low‐quality evidence). No study reported on adverse events .

Pharmacological versus pharmacological intervention

For metformin plus liraglutide compared to metformin we are uncertain of an effect on the adverse events nausea (OR 7.22, 95% CI 0.72 to 72.7; 28 women, 1 study; very low‐quality evidence), diarrhoea (OR 0.31, 95% CI 0.01 to 8.3; 28 women, 1 study; very low‐quality evidence), and headache (OR 5.80, 95% CI 0.25 to 133; 28 women, 1 study; very low‐quality evidence). We are uncertain if a combination of metformin plus liraglutide vs metformin affects BMI (MD 2.1, 95% CI ‐0.42 to 2.62; 28 women, 1 study; very low‐quality evidence) and total body fat (MD ‐0.50, 95% CI ‐4.65 to 3.65; 28 women, 1 study; very low‐quality evidence).

For metformin, clomiphene, and L‐carnitine versus metformin, clomiphene, and placebo, we are uncertain of an effect on miscarriage (OR 3.58, 95% CI 0.73 to 17.55; 274 women, 1 study; very low‐quality evidence), clinical pregnancy (OR 5.56, 95% CI 2.57 to 12.02; 274 women, 1 study; very low‐quality evidence) or BMI (MD ‐0.3, 95% CI 1.17 to 0.57, 274 women, 1 study, very low‐quality evidence).

We are uncertain if dexfenfluramine versus placebo affects weight loss in kilograms (MD ‐0.10, 95% CI ‐2.77 to 2.57; 21 women, 1 study; very low‐quality evidence). No study reported on live birth, quality of life, or mental health outcomes.

Pharmacological intervention versus no intervention or placebo

We are uncertain if metformin compared with placebo affects live birth (OR 1.57, 95% CI 0.44 to 5.57; 65 women, 1 study; very low‐quality evidence). This suggests that if the chance of live birth following placebo is assumed to be 15%, the chance following metformin would be 7% to 50%. We are uncertain if metformin compared with placebo affects gastrointestinal adverse events (OR 0.91, 95% CI 0.32 to 2.57; 65 women, 1 study; very low‐quality evidence) or miscarriage (OR 0.50, 95% CI 0.04 to 5.80; 65 women, 1 study; very low‐quality evidence) or clinical pregnancy (OR 2.67, 95% CI 0.90 to 7.93; 96 women, 2 studies; I² = 48%; very low‐quality evidence). We are also uncertain if diet combined with metformin versus diet and placebo affects BMI (MD ‐0.30, 95% CI ‐2.16 to 1.56; 143 women, 1 study; very low‐quality evidence) or waist‐to‐hip ratio (WHR) (MD 2.00, 95% CI ‐2.21 to 6.21; 143 women, 1 study; very low‐quality evidence).

Pharmacological versus non‐pharmacological intervention

No study undertook this comparison.

Authors' conclusions

Evidence is insufficient to support the use of pharmacological and non‐pharmacological strategies for obese women with subfertility. No data are available for the comparison of pharmacological versus non‐pharmacological strategies. We are uncertain whether pharmacological or non‐pharmacological strategies effect live birth, ongoing pregnancy, adverse events, clinical pregnancy, quality of life, or mental health outcomes. However, for obese women with subfertility, a lifestyle intervention may reduce BMI. Future studies should compare a combination of pharmacological and lifestyle interventions for obese women with subfertility.

Plain language summary

Do pharmacological and non‐pharmacological strategies reduce weight in obese women with subfertility?

To assess the effectiveness and safety of pharmacological and non‐pharmacological strategies compared with each other, placebo, or no treatment for weight reduction in obese women with subfertility.

Background

To prevent the adverse effects of obesity, weight loss is recommended as the first line of treatment for obese women seeking pregnancy. The effectiveness of pharmacological and non‐pharmacological interventions for obese women with subfertility is unclear.

Study characteristics

We found 10 randomised trials comparing pharmacological and non‐pharmacological strategies in 1490 obese women with subfertility.

Key results

Lack of data is a major concern in interpretation of these data. Only 10 studies were included in the analysis. Three studies compared non‐pharmacological intervention versus no intervention or placebo. We are uncertain whether diet versus no intervention improves live birth, ongoing pregnancy, clinical pregnancy, or adverse events. A diet or lifestyle intervention may result in body mass index (BMI) weight change. Evidence was insufficient to show a difference in waist‐to‐hip ratio (WHR) with diet or lifestyle change compared to no intervention. No study reported on quality of life, or mental health outcomes for this comparison.

One study compared non‐pharmacological interventions ‐ intensive weight loss intervention versus standard of care nutrition counselling; however due to the very low quality of evidence, we are uncertain whether intensive weight loss interventions improve live birth, clinical pregnancy, quality of life, or mental health outcomes. No study reported on adverse events, or weight changes, for this comparison.

Three studies reported on pharmacological versus pharmacological interventions. Evidence was insufficient to show a difference in adverse events between metformin compared to metformin plus liraglutide. Evidence was insufficient to demonstrate a difference between the combination of metformin, clomiphene, and L‐carnitine versus metformin, clomiphene, and placebo for miscarriage. Evidence was insufficient to reveal a difference between the combination of metformin, clomiphene, and L‐carnitine versus metformin, clomiphene, and placebo or metformin plus liraglutide versus metformin in clinical pregnancy. Evidence was insufficient to demonstrate a difference between the combination of metformin, clomiphene, and L‐carnitine versus metformin, clomiphene, and placebo for weight change using BMI. Moreover, evidence was insufficient to reveal a difference between dexfenfluramine versus placebo or metformin plus liraglutide versus metformin for weight loss in kilograms, or per cent of total body fat. No study reported on live birth and change in quality of life or mental health outcomes for this comparison.

In the comparison of pharmacological intervention versus no intervention or placebo, three studies were included. Evidence was insufficient to show a difference between metformin and control groups related to live birth. Evidence was insufficient to reveal a difference between metformin compared to placebo in live birth, clinical pregnancy, or adverse events. Evidence was insufficient to demonstrate a difference between diet combined with metformin versus diet combined with placebo for weight change using BMI or WHR. No study reported on quality of life or mental health outcomes for this comparison.

We found no study comparing non‐pharmacological with pharmacological interventions.

Quality of the evidence

The evidence was of very low to low quality. The main limitations were due to lack of studies and poor reporting of study methods. The main reasons for downgrading of evidence were lack of details by which to judge risk of bias (randomisation and allocation concealment), lack of blinding, and imprecision.

Summary of findings

Summary of findings 1. Non‐pharmacological intervention compared to no intervention or placebo for obese women with subfertility.

Non‐pharmacological intervention compared to no intervention for obese women with subfertility
Patient or population: obese women with subfertility Setting: hospital Intervention: non‐pharmacological (diet and/or lifestyle changes) Comparison: no intervention
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№. of participants (studies)	Certainty of evidence (GRADE)	Comments
	Risk with non‐pharmacological intervention	Risk with no intervention
Live birth	395 per 1000 (334 to 462)	431 per 1000	OR 0.85 (0.65 to 1.12)	917 (3 RCTs)	⊕⊕⊝⊝ lowa,b
Ongoing pregnancy	536 per 1000 (453 to 617)	588 per 1000	OR 0.81 (0.58 to 1.13)	564 (1 RCT)	⊕⊕⊝⊝ lowc
Miscarriage	122 per 1000 (82 to 177)	83 per 1000	OR 1.54 (0.99 to 2.39)	917 (3 RCTs)	⊕⊝⊝⊝ very lowc,d
Clinical pregnancy	529 per 1000 (458 to 594)	514 per 1000	OR 1.06 (0.81 to 1.40)	917 (3 RCTs)	⊕⊕⊝⊝ lowa,b
Weight change BMI change (Einarsson 2017)	MD 3.70 kg/m² lower (4.10 lower to 3.36 lower)	Mean BMI change ranged from 0.04 to 0.7 kg/m²	‐	305 (1 RCT)	⊕⊕⊝⊝ lowc
BMI change (Sim 2014)	MD 1.80 kg/m² lower (2.67 lower to 0.93 lower)	Mean BMI change ranged from ‐1.3 to 0 kg/m²	‐	43 (1 RCT)	⊕⊝⊝⊝ very lowc,e
Quality of life/ Mental health outcome						No study reported this outcome
*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; OR: odds ratio; 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 one level for substantial heterogeneity: I² > 70%.

^bDowngraded one level for imprecision as reflected by the large confidence interval.

^cDowngraded two levels for serious imprecision.

^dDowngraded one level for indirectness due to differences in definition.

^eDowngraded one level for incomplete outcome data with an uneven dropout between groups.

Summary of findings 2. Non‐pharmacological intervention compared to non‐pharmacological intervention for obese women with subfertility.

Non‐pharmacological intervention compared to non‐pharmacological intervention for obese women with subfertility
Patient or population: obese women with subfertility Setting: hospital Intervention: intensive weight loss intervention Comparison: standard‐of‐care nutrition counseling
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№. of participants (studies)	Certainty of evidence (GRADE)	Comments
	Risk with weight loss	Risk with standard of care nutrition
Live birth	500 per 1000	0 per 1000 (0 to 0)	OR 11.00 (0.43 to 284)	11 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
Miscarriage	0 per 1000	0 per 1000				In the trial with 11 women, no pregnancy loss occurred
Clinical pregnancy	500 per 1000	0 per 1000 (0 to 0)	OR 11.00 (0.43 to 284)	11 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
Weight change Weight change (kg) Body mass index (BMI)	Mean weight change ranged from 5 to 6 kg	MD 9 kg lower (15.5 lower to 2.5 lower)	‐	11 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
	Mean body mass Index ranged from 2 to 3 kg/m²	MD 3 kg/m² lower (5.37 lower to 0.63 lower)	‐	11 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
Quality of life/Mental health outcome Mental health (at 12 weeks) Quality of life (at 12 weeks)	Mean mental health ranged from 2 to 3	MD 7 lower (13.92 lower to 0.08 lower)	‐	11 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
	Mean quality of life ranged from 0.01 to 0.02	MD 0.06 higher (0.03 lower to 0.15 higher)	‐	11 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
*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; OR: odds ratio; 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 one level for selection bias due to lack of details on random sequence, random allocation, and blinding.

^bDowngraded two levels for serious imprecision due to small sample size.

Summary of findings 3. Pharmacological intervention compared to pharmacological intervention for obese women with subfertility.

Pharmacological intervention compared to pharmacological intervention for obese women with subfertility
Patient or population: obese women with subfertility Setting: hospital Intervention: different pharmacological interventions Comparison: different pharmacological interventions
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№. of participants (studies)	Certainty of evidence (GRADE)	Comments
	Risk with pharmacological intervention	Risk with pharmacological intervention
Live birth						No study reported this outcome
Adverse events ‐ Metformin plus liraglutide vs metformin Nausea Diarrhoea Headache	71 per 1000	357 per 1000 (52 to 848)	OR 7.22 (0.72 to 73)	28 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
	71 per 1000	23 per 1000	OR 0.31 (0.01 to 8.3)	28 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
	0 per 1000	0 per 1000	OR 5.80 (0.25 to 133)	28 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
Clinical pregnancy Metformin plus liraglutide vs metformin Met + CC + L‐carnitine vs Met + CC + placebo	80 per 1000	311 per 1000 (184 to 476)	OR 5.20 (2.59 to 10.4)	28 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
	214 per 1000	500 per 1000 (160 to 839)	OR 3.67 (0.70 to 19.1)	274 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
Miscarriage Met + CC + L‐carnitine vs Met + CC + placebo	15 per 1000	51 per 1000 (11 to 208)	OR 3.58 ( 0.73 to 17.6)	274 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
Weight ‐ BMI change Metformin plus liraglutide vs metformin Met + CC + L‐carnitine vs Met + CC + placebo Weight change Dexfenfluramine vs placebo Weight ‐ body fat Metformin plus liraglutide vs metformin	Mean BMI change was set at 0	MD 2.1 higher (0.42 lower to 4.52 higher)	‐	28 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
	Mean BMI change was set at 0	MD 0.3 lower (1.17 lower to 0.57 higher)	‐	274 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
	Mean weight change was set at 0 kg	MD 0.1 kg lower (2.77 lower to 2.57 higher)	‐	21 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
	Mean change in % total body fat was set at 0	MD 0.5 lower (4.65 lower to 3.65 higher)	‐	28 (1 RCT)	⊕⊝⊝⊝ Very lowa,b
Quality of life/Mental health outcome						No study reported this outcome
*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; OR: odds ratio; 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 one level for selection bias due to lack of details on random sequence, random allocation, and blinding.

^bDowngraded two levels for serious imprecision as reflected by the large confidence interval.

Summary of findings 4. Pharmacological intervention compared to no intervention or placebo for obese women with subfertility.

Pharmacological intervention compared to no intervention/placebo for obese women with subfertility
Patient or population: obese women with subfertility Setting: hospital Intervention: metformin Comparison: no intervention/placebo
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№. of participants (studies)	Certainty of evidence (GRADE)	Comments
	Risk with pharmacological intervention	Risk with no intervention/placebo
Live birth	219 per 1000 (73 to 499)	152 per 1000	OR 1.57 (0.44 to 5.57)	65 (1 RCT)	⊕⊝⊝⊝ very lowa,b
Ongoing pregnancy						No study reported this outcome
Miscarriage	31 per 1000 (3 to 272)	61 per 1000	OR 0.50 (0.04 to 5.80)	65 (1 RCTs)	⊕⊝⊝⊝ very lowa,b
Adverse events (GI)	313 per 1000	333 per 1000	OR 0.91 (0.32 to 2.57)	65 (1 RCT)	⊕⊝⊝⊝ very lowa,b
Clinical pregnancy	237 per 1000 (95 to 480)	104 per 1000	OR 2.67 (0.90 to 7.93)	96 (2 RCTs)	⊕⊝⊝⊝ very lowa,b
Weight change BMI WHR	MD 0.3 kg/m² lower (2.16 lower to 1.56 higher)	Mean BMI change was set at 0	‐	143 (1 RCTs)	⊕⊝⊝⊝ very lowa,c
	MD 2 cm higher (2.21 lower to 6.21 higher)	Mean WHR change was set at 0	‐	143 (1 RCT)	⊕⊝⊝⊝ very lowa,c
Quality of life/Mental health outcome						No study reported this outcome
*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; OR: odds ratio; 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 twice for serious imprecision as reflected by the large confidence interval.

^bDowngraded once in view of inconsistency in clinical pregnancy, which could not be seen in studies in view of the presence of only a single trial with 65 women for the outcomes live birth and miscarriage.

^cDowngraded once for selection bias due to lack of details on random sequence, random allocation, and blinding.

Background

Description of the condition

Obesity rates have been rising worldwide, creating a global health problem. The World Health Organization has defined obesity as body mass index (BMI) > 30 kg/m². According to estimates, in 2015, about 2.3 billion men and women were overweight (BMI 25 to 30 kg/m²), and 700 million were obese (BMI > 30 kg/m²). The prevalence of obesity is estimated to range from 5% in some developing countries to more than 30% in developed countries (Hossain 2007; Ogden 2006).

The obesity epidemic has contributed to fertility problems. Obese women have a lesser chance of conceiving naturally than non‐obese women (Chong 1986; Crosignani 1994; Hamilton‐Fairley 1992), and they are at greater risk of miscarriage (Boots 2011). Obesity can result in anovulation and a reduced chance of conceiving among ovulatory subfertile women (Metwally 2007; van der Steeg 2008). Furthermore, pregnancy and live birth rates following IVF appear to be lower in obese women (Koloszar 2002; Wang 2002; Fedorcsák 2004; Sermonade 2019). Moreover, literature suggests that obesity is related to maternal and neonatal complications such as congenital anomalies, hypertensive disorder, gestational diabetes, prolonged labour, macrosomia, and shoulder dystocia (Edwards 1996; Garbaciak 1985; Waller 1994; Weiss 2004).

To prevent the adverse effects of obesity, weight loss is recommended as the first line of treatment for obese women seeking pregnancy (Thessaloniki 2008).

Description of the intervention

Treatment for obesity can involve both non‐pharmacological and pharmacological strategies.

Non‐pharmacological strategies

Diet

Generally, weight loss occurs when energy intake is lower than energy expenditure. In two small studies, replacing protein for carbohydrate within the context of an energy‐restricted diet using 12‐week and 1‐month dietary intervention among the overweight or obese target population (polycystic ovarian syndrome (PCOS) patients) provided the same improved reproductive outcomes that were achieved in the control group (Moran 2005; Stamets 2004), although postprandial glucose response was 3.5 times lower in the group with a higher‐protein diet. Lifestyle modification through diet and exercise programmes in obese women with PCOS improves reproductive outcomes (Clark 1998; Huber‐Buchholz 1999). An important point is that a minimal amount of weight loss (5% to 10%) over as little as four weeks is sufficient to improve the presentation of PCOS despite patients remaining clinically overweight or obese (Clark 1998; Hamilton‐Fairley 1993; Wahrenberg 1999).

Exercise

Exercise is an important component of any lifestyle modification and weight management programme. The results of two studies that examined effects of exercise on insulin resistance in overweight or obese women with PCOS (who were followed for 16 to 24 weeks and for 6 months) were different. Neither of these studies reported changes in hormone or reproductive parameters (Brown 2009; Randeva 2002). In these studies, the impact of exercise was not evaluated separately from diet, and the result may suffer from reporting bias (Thomson 2010).

Behavioural

Behavioural might contribute to greater weight loss when combined with medical therapy and diet. In one study (Wadden 2005), participants were followed for 52 weeks, during which time counselling was given including regular supportive and motivational personal or group sessions. Behaviour therapy improved weight loss as well as weight maintenance and control.

Complementary and traditional healthcare approaches

The terms 'complementary' and 'alternative' describe practices and products that people choose as adjuncts or alternatives to Western medical approaches (Kaptchuk 2001; Straus 2004). The National Institutes of Health has grouped such interventions into five somewhat overlapping domains as follows (nccam.nih.gov/health/whatiscam).

• Biologically based practices. These include use of a vast array of vitamins and mineral supplements, natural products such as chondroitin sulphate, which is derived from bovine or shark cartilage, and herbals such as ginkgo biloba and echinacea.

• Manipulative and body‐based approaches. These types of therapies, which include massage, have been used throughout history. In the 19th century, additional formal manipulative disciplines emerged in the United States: chiropractic medicine and osteopathic medicine, which had a great influence in complementary medicine.

• Mind‐body medicine. Many ancient cultures assumed that the mind exerts powerful influences on bodily functions and vice versa. Attempts to reassert proper harmony between these bodily systems led to the development of mind‐body medicine, an array of approaches that incorporate spiritual, meditative, and relaxation techniques.

• Alternative medical systems. Whereas the ancient Greeks postulated that health requires a balance of vital humors, Asian cultures considered that health depends on the balance and flow of vital energies through the body. This latter theory underlies the practice of acupuncture, for example, which asserts that vital energy flow can be restored by placing needles at critical body points.

• Energy medicine. This approach uses therapies that involve the use of energy ‐ biofield‐based or bioelectromagnetic‐based interventions. An example of the former is Reiki therapy, which aims to realign and strengthen healthful energies through the intervention of energies radiating from the hands of a master healer.

Pharmacological strategies

Numerous anti‐obesity medications are prescribed for weight loss. These drugs may be classified as follows.

• Drugs acting on the gastrointestinal tract (GIT): lipase inhibitors (orlistat).

• Centrally acting anti‐obesity agents: catecholaminergic agents (phentermine).

• Serotonin and noradrenaline reuptake inhibitors such as sibutramine, selective serotonin reuptake inhibitors (SSRIs) (sertraline).

• Dopamine reuptake antagonists (bupropion), anti‐depressants (fluoxetine).

• Exercise mimetics ephedrine, caffeine, synephrine, beta 3 adrenergic agonists, uncoupling proteins 2 and 3 (thermogenin).

• Leptin‐related agents: therapeutic leptin, leptin analogues, leptin receptor agonists.

All of these drugs have side effects, and side effect profiles vary per drug. Sibutramine is associated with modest increases in heart rate and blood pressure; gastrointestinal symptoms predominate with the use of orlistat; phentermine can induce cardiovascular and gastrointestinal side effects; fluoxetine is associated with agitation and nervousness, in addition to gastrointestinal side effects; bupropion with paraesthesia, insomnia, and central nervous system effects; and topiramate with paraesthesia and changes in taste (Li 2005).

How the intervention might work

The aetiology of obesity is believed to be multi‐factorial, with both genetic and environmental contributions. A key determinant of obesity is the balance between ingested calories and the body's basal energy expenditure. Obesity therefore results when small positive energy balances accumulate over a long time (Flegal 2010; Swinburn 2009). Weight loss can be achieved through lifestyle intervention programmes incorporating the combination of a healthy diet, increased physical activity, behavioural modification, and use of complementary and traditional healthcare approaches and medications.

An adverse effect of obesity on female fertility could be mediated by several mechanisms. First, obesity potentially contributes to excess oestrogen as a result of extraglandular aromatisation of androgen precursors. Moreover, sex hormone–binding globulin levels are diminished, resulting in more bioavailable oestrogen and androgen for aromatisation. Second, obesity increases leptin levels. The actions of leptin on the hypothalamus‐pituitary‐ovary (HPO) axis are believed to have differential effects on central and peripheral components of the reproductive system. In the central nervous system, leptin has been shown to modulate gonadotropin‐releasing hormone (GnRH) pulse frequency in vitro (Scott 2009). On the gonadal level, leptin has been found in ovarian follicular fluid, and leptin receptor has been localised to human granulosa and theca cells. In humans, leptin may interrupt normal oocyte maturation (Smith 2002). Weight loss improves the metabolic, endocrine, and reproductive profile of obese women (Falsetti 1992; Hollmann 1996; Kumar 1993). Evidence indicates that a 5% weight loss improves both natural and induced conception, as well as the chance of a healthy live birth (Khaskheli 2013).

Weight loss can be achieved by pharmacological treatments and non‐pharmacological intervention programmes. Pharmacological treatment for obesity is considered an option for infertile women who are overweight or obese because the safety of these treatments has not been fully studied (Johansson 2015; Kominiarek 2017). A systematic review suggested that sibutramine, orlistat, phentermine, probably diethylpropion, probably fluoxetine, bupropion, and topiramate might promote modest weight loss for at least six months when given along with recommendations for diet (and possibly other behavioural and exercise interventions) (Li 2005). Medications act on the mechanisms regulating appetite and satiety and help combat the pathophysiological adaptations that drive weight regain (Garvey 2013). Due to the potential risks associated with surgery or weight loss medications, health organisations have recommended that infertile women who are overweight or obese should follow lifestyle changes (ASRM 2015). Recent international guidelines strongly support the importance of pre‐pregnancy lifestyle interventions in an interdisciplinary situation to encourage healthy lifestyles and maintain weight loss for obese women (Brauer 2015; Kominiarek 2017). In addition, knowledge about the effects of supplements and complementary therapy (herbal medicine and acupuncture) is emerging, but evidence for the overall effects of these interventions is incomplete.

Lifestyle modification, which generally consists of a combination of nutrition, physical activity, and behavioural modification, is an oft‐used strategy to help patients achieve weight loss and maintenance (Berkel 2005; Lang 2006). It has been suggested that complementary and alternative medicine including acupuncture might improve weight loss by (1) regulating obesity‐related neuropeptides (Cabioglu 2006; Gucel 2012); (2) regulating hypothalamus‐pituitary‐adrenal cortex and sympathetic adrenal cortex (Yin 2005); and (3) conferring lipid‐lowering effects (Abdallah 2011). A systematic review and meta‐analysis suggested that acupuncture for obesity might be beneficial compared to placebo or lifestyle control, but results were limited by the clinical heterogeneity and poor methodological quality of the included trials (Cho 2009).

Why it is important to do this review

With the growing incidence of infertility among obese women, it is becoming increasingly common for women to utilise assisted reproductive treatment to become pregnant (Kupta 2014). Overweight and obese women have poor maternity outcomes (Koning 2012; Pandey 2010; Rittenberg 2011), while weight reduction improves reproductive outcomes for these patients (Crosignani 2003; Pandey 2010). The effectiveness of pharmacological and non‐pharmacological interventions for obese women with subfertility is unclear. Moreover, despite the fact that non‐pharmacological interventions are commonly recommended for management of obese subfertile women, their effectiveness in comparison with pharmacological strategies has not been previously examined in a systematic review (Kim 2020).

Objectives

To assess the effectiveness and safety of pharmacological and non‐pharmacological strategies compared with each other, placebo, or no treatment, for achieving weight loss in obese women with subfertility.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs) and cross‐over randomised trials. For cross‐over trials, only data from the first phase will be included in meta‐analyses, as the cross‐over is not a valid design in this context.

Types of participants

Obese women (BMI ≥ 30 kg/m², or as appropriate for the ethnicity of women in the primary study) of childbearing age (post‐menarche and pre‐menopause) of any ethnic origin who have been unable to conceive for at least 12 months (including primary and secondary subfertility), with or without reasons (anovulatory, unexplained, tubal disease, endometriosis, uterine abnormalities, male factor), and with all types of fertility treatment including intrauterine insemination (IUI), in vitro fertilisation (IVF), and expectant management.

Types of interventions

We included all studies in which weight loss was the main treatment intervention or weight loss interventions were part of a subfertility management programme.

Eligible comparisons are pharmacological, non‐pharmacological, and no intervention or placebo.

We considered the following comparisons.

• Non‐pharmacological versus pharmacological intervention (e.g. acupuncture versus SSRI).

• Non‐pharmacological versus non‐pharmacological intervention (e.g. acupuncture versus exercise, one type of exercise versus another).

• Pharmacological versus pharmacological intervention (e.g. one SSRI versus one type of pharmacological intervention, one SSRI versus another).

• Pharmacological intervention versus no intervention or placebo.

• Non‐pharmacological intervention versus no intervention or placebo.

The following interventions will be considered.

Non–pharmacological interventions

• Behaviour: behaviour modification, behaviour change, brief intervention, brief advice, nurse counselling, physician counselling, psychological counselling, waiting list for treatment with the promise of treatment upon achieving a weight target, behavioural advice, behaviour therapy, Internet‐based support, self‐directed support, social support, group therapy, family therapy, psychotherapy, support group, relaxation, health education, health promotion, motivation, meditation, religious intervention.

• Diet: diet modification, dietician‐led dietary advice, self‐directed dietary instruction, low‐carbohydrate diet, low‐fat diet, hypocaloric diet.

• Exercise: walking, jogging, running, swimming, aerobics, structured exercise referral/interventions, weight‐lifting/training, gymnastics, resistance training, fitness training, endurance training, cycling, boxing, kick‐boxing, pedometry, exercise therapy, sports therapy.

• Complementary and traditional healthcare approaches: acupuncture; electro‐therapy; physical therapy; aromatherapy; auricular stimulation; body therapy; acupuncture ‐ Moxibustion; Tai‐chi; phytooestrogens; soy products; phytovitamins; dietary supplements and herbal products including conjugated linoleic, pyruvate, ephedra sinica (ma huang), chromium, hydroxy citric acid (Garcinia cambogia), and Chitosan.

Pharmacological interventions

• Drugs acting on the GIT: lipase inhibitors (e.g. orlistat (Xenical), tetrahydrolipstatin), bulking agents (e.g. methylcellulose (Celevac), ispaghula husk, sterculia, bran, guar gum), insulin sensitisers, gastrointestinal peptides, glucagon‐like peptide‐1, enterostatin.

• Centrally acting anti‐obesity agents: catecholaminergic agents (e.g. phentermine, mazindol, diethylpropion, phenylpropanolamine), serotonergic agents (e.g. fenfluramine, dexfenfluramine, fluoxetine), combined catecholaminergic plus serotoninergic agents (e.g. phentermine plus fenfluramine).

• Serotonin and noradrenaline reuptake inhibitors (e.g. sibutramine), selective serotonin reuptake inhibitors (SSRIs; e.g. sertraline).

• Dopamine reuptake antagonists (e.g. bupropion), anti‐depressants (e.g. fluoxetine).

• Exercise mimetics (e.g. ephedrine, caffeine, synephrine), beta 3 adrenergic agonists, uncoupling proteins 2 and 3 (Thermogenin).

• Leptin‐related agents (e.g. therapeutic leptin, leptin analogues, leptin receptor agonists).

We excluded surgical interventions.

Types of outcome measures

Primary outcomes

• Live birth or ongoing pregnancy (when live birth is not available)

  • Live birth is defined as delivery of a live fetus after 20 completed weeks of gestation

  • Ongoing pregnancy is defined as evidence of a gestational sac with fetal heart motion at 12 weeks, confirmed by ultrasound

• Adverse events: miscarriage (loss of pregnancy during the first 20 weeks of gestation) or gastrointestinal symptoms (e.g. nausea, vomiting, diarrhoea)

Secondary outcomes

• Clinical pregnancy: defined as evidence of a gestational sac, confirmed by ultrasound

• Weight change (e.g. body mass index (BMI), waist to hip ratio (WHR), percentage of body fat or total body fat)

• Change in endocrine parameters: total and free testosterone (ng/dL or nmol/L), sex hormone‐binding globulin (SHBG; nmol/L), testosterone‐to‐SHBG ratio, diabetic tests such as glucose tolerance test (GTT; mmol/L), glycated haemoglobin (HbA1c; mmol/mol)

• Quality of life or mental health outcome. If studies reported more than one scale, preference will be given to the SF‐36 (36‐Item Short Form Health Survey), then to other validated generic scales, and finally, to condition‐specific scales

Search methods for identification of studies

In consultation with the Cochrane Gynaecology and Fertility Group (CGF) Information Specialist, we formulated a comprehensive search strategy to identify all RCTs of pharmacological and non‐pharmacological strategies for obese women with subfertility regardless of language or publication status (published, unpublished, in press, or in progress).

Electronic searches

We searched the following electronic databases, trial registers, and websites:

• Cochrane Gynaecology and Fertility Group (CGF) Specialised Register of Controlled Trials, ProCite platform, searched on 18 August 2020 (Appendix 1).

• Cochrane Central Register of Controlled Trials (CENTRAL), via the Cochrane Central Register of Studies Online (CRSO), Web platform, searched on 18 August 2020 (Appendix 2).

• MEDLINE, Ovid platform, searched from 1946 to 18 August 2020 (Appendix 3).

• Embase, Ovid platform, searched from 1980 to 18 August 2020 (Appendix 4).

• PsycINFO, Ovid platform, searched from 1806 to 18 August 2020 (Appendix 5).

• Allied and Complementary Medicine Database (AMED), Ovid platform, searched from 1985 to 18 August 2020 (Appendix 6).

• Cumulative Index to Nursing and Allied Health Literature (CINAHL), Ebsco platform, searched from 1961 to 26 September 2019 (Appendix 7). (CINAHL references are now included in CENTRAL; therefore the CENTRAL search on 18 August 2020 included CINAHL references).

The MEDLINE search from inception to 18 August 2020 was combined with the Cochrane highly sensitive search strategy for identifying randomised trials, which appears in the Cochrane Handbook for Systematic Reviews of Interventions (Lefebvre 2020, Version 6.1, Chapters 4, 4.4.7; 4S1). Searches of Embase and CINAHL from inception to 18 August 2020 were combined with trial filters developed by the Scottish Intercollegiate Guidelines Network (SIGN) (www.sign.ac.uk/what-we-do/methodology/search-filters/).

Searching other resources

Other sources of trials below were searched on 18 August 2020:

• Trial registers for ongoing and registered trials: www.clinicaltrials.gov (a service of the US National Institutes of Health), the World Health Organization International Trials Registry Platform search portal, at www.who.int/trialsearch/Default.aspx.;

• Database of Abstracts of Reviews of Effects (DARE), part of the Cochrane Library, at onlinelibrary.wiley.com/o/cochrane/cochrane_cldare_articles_fs.html (for reference lists from relevant non‐Cochrane reviews);

• Relevant non‐Cochrane reviews;

• Web of Science wokinfo.com/ (another source of trials and conference abstracts);

• OpenGrey at www.opengrey.eu/ (for unpublished literature from Europe);

• PubMed (for recent trials not yet indexed in MEDLINE);

• ProQuest.

We checked the reference lists of relevant trials, reviews, and textbooks. We contacted experts in the field for relevant trials and to obtain additional data. Output of all searches was managed with EndNote, which lists all studies and removes duplicates.

Data collection and analysis

Selection of studies

After an initial screen of titles and abstracts retrieved by the search, conducted by two review authors (FB and ST), we retrieved the full text of all potentially eligible studies. These review authors independently examined these full‐text articles for compliance with the inclusion criteria and selected eligible studies. We corresponded with study investigators, as required, to clarify study eligibility. Any disagreement about whether to include or exclude a study was discussed with a third review author (SJ) until consensus was achieved. We have listed excluded studies and reasons for their exclusion in the Characteristics of excluded studies tables. See Figure 1 (PRISMA flow chart) for details of the screening and selection process.

1.

Study flow diagram.

Data extraction and management

Two review authors (FB and ST) independently extracted data from eligible studies using a data extraction form that had been designed and pilot‐tested by review authors. Any disagreements were resolved by discussion. Data extracted included study characteristics and outcome data. When studies had multiple publications, review authors collated multiple reports of the same under a single study ID with multiple references. We corresponded with study investigators to request further data on methods and/or results, as required. Data extracted included population characteristics (e.g. female age, BMI, waist‐hip ratio, ethnicity), study characteristics, and outcome data.

Assessment of risk of bias in included studies

Two review authors (FB and ST) assessed risk of bias using the Cochrane 'Risk of bias' assessment tool to assess selection bias (random sequence generation, allocation concealment); performance bias (blinding of women and personnel); detection bias (blinding of outcome assessors); attrition bias (incomplete outcome data); reporting bias (selective reporting); and other biases (Higgins 2011). Disagreements were resolved by consensus or by discussion with a third review author (SJ).

Random sequence generation was scored at low risk of bias when an appropriate method of sequence generation was described according to Cochrane methods (Higgins 2011).

Allocation concealment was considered at low risk of bias if opaque and numbered envelopes or a centralised Internet‐based randomisation procedure was used. Lack of blinding is unlikely to affect live birth (scored at low risk of bias) but might affect adverse events (scored at high risk of bias). Attrition bias was scored low when all or most (> 95%) of the women randomised were analysed. Reporting bias was scored low when all relevant outcomes were reported as planned in the protocol, as described in published protocols in journals or in trial registers. To score other forms of bias, we looked at differences in baseline values and treatment details. If these issues were unclear, we scored the risk of bias as unclear.

We corresponded with study authors to identify any within‐trial selective reporting. We sought published protocols and compared outcomes between the protocol and the final published study. The 'Risk of bias' table is presented with the table Characteristics of included studies. All judgements are fully described. Conclusions are presented in the ’Risk of bias’ table and are incorporated into the interpretation of review findings through sensitivity analyses.

With respect to within‐trial selective reporting, when identified studies failed to report the primary outcome of live birth but did report interim outcomes such as pregnancy, we assessed whether the interim values were similar to those reported in studies that also reported live birth.

Measures of treatment effect

We performed a statistical analysis in accordance with statistical guidelines provided in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We used a fixed‐effect model for all analyses.

For dichotomous data (e.g. live birth rates), we used numbers of events in control and intervention groups of each study to calculate odds ratios (ORs). For reporting purposes, we translated primary outcomes to absolute risks.

For continuous data (e.g. weight loss), if all studies reported exactly the same outcomes, we planned to calculate mean differences (MDs) between treatment groups. If similar outcomes were reported on different scales (e.g. change in weight, quality of life), we planned to calculate standardised mean differences (SMDs). We planned to reverse the direction of effect of individual studies, if required, to ensure consistency across trials. We planned to treat ordinal data (e.g. quality of life scores) as continuous data. We planned to present 95% confidence intervals for all outcomes.

When data to calculate RRs or MDs were not available, we planned to utilise the most detailed numerical data available that may facilitate similar analyses of included studies (e.g. test statistics, P values). We planned to assess whether estimates calculated in the review for individual studies were compatible in each case with estimates reported in study publications.

Because cluster‐RCTs are included in the review, we planned to first make an assessment as to whether the trial had been analysed in such a way as to account for clustering, and if not, we planned to make an adjustment to the trial results using one of several available approaches including a generic inverse‐variance method via effect estimates and their standard errors extracted from cluster‐RCTs.

Unit of analysis issues

The primary analysis was planned to be per woman randomised; per pregnancy data may also be included for some outcomes (e.g. miscarriage). Data that do not allow valid analysis (e.g. 'per cycle' data) were planned to be briefly summarised in an additional table and not to be meta‐analysed. Multiple births were planned to count as one live birth event. Only first‐phase data from cross‐over trials were planned to be included.

Dealing with missing data

We planned to analyse data on an intention‐to‐treat basis as far as possible (i.e. including all randomised women in the analysis, in the groups to which they were randomised). We planned to attempt to obtain missing data from the original study authors. When these were unobtainable, we planned to undertake imputation of individual values for live birth only: live birth was planned to be assumed not to have occurred in women without a reported outcome. For other outcomes, we planned to analyse only available data. Any imputation undertaken was planned to be subjected to Sensitivity analysis.

If studies reported sufficient detail to calculate mean differences but no information on associated SD, we planned to assume the outcome to have an SD equal to the highest SD from other studies within the same analysis.

Assessment of heterogeneity

We planned to consider whether the clinical and methodological characteristics of included studies were sufficiently similar for meta‐analysis to provide a clinically meaningful summary. We planned to assess statistical heterogeneity by using the I² value. An I² measurement greater than 50% was planned to be taken to indicate substantial heterogeneity (Higgins 2011).

Assessment of reporting biases

In view of the difficulty of detecting publication bias and other biases, review authors aimed to minimise their potential impact by ensuring a comprehensive search for eligible studies, and by being alert for duplication of data. If 10 or more studies were included in an analysis, we planned to use a funnel plot to assess the potential for publication bias.

Data synthesis

If studies were sufficiently similar, we planned to combine data using a fixed‐effect model for the following comparisons.

• Non‐pharmacological intervention versus no intervention/placebo.

• Non‐pharmacological versus non‐pharmacological intervention.

• Pharmacological versus pharmacological intervention.

• Pharmacological intervention versus no intervention/placebo.

• Pharmacological versus non‐pharmacological intervention.

See Types of interventions for the exact study interventions we planned to investigate.

An increase in the risk of all outcomes was planned to be displayed graphically in meta‐analyses to the right of the centre‐line, and a decrease in the risk of an outcome to the left of the centre‐line.

Subgroup analysis and investigation of heterogeneity

When sufficient data were available (at least five RCTs), we planned to perform the following subgroup analyses for primary outcomes only.

• Duration of intervention (short: 2 to 4 weeks, medium: 4 weeks to 6 months, long: longer than 6 months) (Lim 2019).

• Cause of infertility: anovulatory versus unexplained versus other causes.

• Maternal age: ≤ 35 or ≥ 36 years.

• Severity of obesity (BMI): 30.0 < BMI < 34.9 (class I obesity) versus 35.0 < BMI < 39.9 (class II obesity) versus BMI ≥ 40.0 (class III obesity).

Sensitivity analysis

We planned to conduct the following sensitivity analyses for primary outcomes, to examine stability regarding pooled outcomes.

• Restriction to studies without high risk of bias.

• Use of a random‐effects model.

• Use of risk ratio rather than odds ratio.

Summary of findings and assessment of the certainty of the evidence

We prepared a 'Summary of findings' table using GRADEpro (GRADEpro GDT 2014) and Cochrane methods (Higgins 2011). This table evaluates the overall quality of the body of evidence for the primary review outcomes (live birth or ongoing pregnancy, adverse events, clinical pregnancy, miscarriage) for the main review comparison (pharmacological versus non‐pharmacological strategies). Additional 'Summary of findings' tables were also prepared for the main review outcomes for other important comparisons.

We assessed the quality of the evidence using GRADE criteria: risk of bias, consistency of effect, imprecision, indirectness and publication bias. Judgements about evidence quality (high, moderate, low or very low) were made by two review authors working independently, with disagreements resolved by discussion. Judgements were justified, documented, and incorporated into reporting of results for each outcome.

We extracted study data, formatted our comparisons in data tables and prepared a 'Summary of findings' table using the GRADEpro Guideline Development Tool (GDT) (GRADEpro GDT 2014) before writing the results and conclusions of our review.

Results

Description of studies

We have reported the characteristics of included and excluded studies in the Characteristics of included studies and Characteristics of excluded studies tables. We did not identify any studies from the reference lists.

Results of the search

Our search retrieved 5577 articles (910 duplicates were removed). Of these articles, 4667 studies were screened by title and abstract and 68 studies were assessed at full text for eligibility. Finally, we included 10 studies that met the inclusion criteria for the review (Figure 1). The 10 included trials varied in size from 11 to 564 women (Einarsson 2017; El 2019; Galletly 1996; Johnson 2010; Khorram 2006; Mutsaerts 2016; Rothberg 2016; Salamun 2018; Sim 2014; Tang 2006).

Included studies

Study design

One study was a cross‐over randomised clinical trial (Galletly 1996) for which we tried to extract pre‐cross‐over data, and the others were RCTs (Einarsson 2017; El 2019; Johnson 2010; Khorram 2006; Mutsaerts 2016; Rothberg 2016; Salamun 2018; Sim 2014; Tang 2006). Einarsson 2017, El 2019, and Johnson 2010 used a computerised randomisation programme for the randomisation process, and Tang 2006 used a table of random numbers. Khorram 2006 explained that randomisation was done by picking a card out of a box. A web‐based randomisation program was adopted for Mutsaerts 2016. It is not clear how randomisation was done in Galletly 1996, Sim 2014, Salamun 2018, and Rothberg 2016.

Sample size

The number of women included in the studies ranged from 11 to 564.

Setting

All studies except one were conducted in high‐income countries (Egypt; El 2019). One study was undertaken in the UK (Tang 2006), one in The Netherlands (Mutsaerts 2016), two in the USA (Khorram 2006; Rothberg 2016), one in Sweden (Einarsson 2017), one in New Zealand (Johnson 2010), one in Slovenia (Salamun 2018), and two in Australia (Galletly 1996; Sim 2014). All trials recruited women in hospital settings.

Participants

All women in the included studies met our inclusion criteria. We found 10 randomised trials including 1490 obese women with subfertility. Einarsson 2017 included 305 participants, Sim 2014 48 participants, Mutsaerts 2016 564 participants, Rothberg 2016 11 participants, Salamun 2018 28 participants, El 2019 274 participants, Galletly 1996 21 participants, Tang 2006 143 participants, Khorram 2006 31 participants, and Johnson 2010 65 participants.

The main inclusion criterion was:

• age 18 to 37 (Sim 2014), 38 (Einarsson 2017), 39 (Mutsaerts 2016Tang 2006 ), 40 (Rothberg 2016), or ≤ 38 years (Salamun 2018).

The main exclusion criteria were:

• insulin‐dependent diabetes mellitus (Einarsson 2017);

• endocrinopathy (El 2019; Johnson 2010; Mutsaerts 2016Rothberg 2016Salamun 2018Sim 2014Tang 2006Khorram 2006);

• binge eating disorder (Einarsson 2017);

• current psychiatric condition (Sim 2014);

• recent (within three months) participation in treatment known to affect diet or body weight (Sim 2014);

• taking anti‐obesity drugs or appetite suppressants within the past two months (Rothberg 2016);

• previous bariatric surgery or gastrointestinal disease (Rothberg 2016);

• use of medications that affect reproductive or metabolic functions within the past six weeks (Tang 2006), or in the past two or three months (Rothberg 2016; Salamun 2018, respectively);

• smoker (El 2019); and

• drug user (El 2019Khorram 2006).

There were no significant differences between baseline characteristics in all studies (El 2019Galletly 1996Johnson 2010; Khorram 2006; Mutsaerts 2016; Rothberg 2016; Sim 2014; Tang 2006), except termination of pregnancy in Einarsson 2017 and 120‐minute overload of insulin levels in Salamun 2018.

Interventions

Non‐pharmacological intervention versus no intervention or placebo

One study compared diet versus no intervention (IVF) (Einarsson 2017). In the intervention group, weight reduction was done before IVF, starting with 12 weeks of a low‐calorie liquid formula diet (LCD) of 880 kcal/d and thereafter weight stabilisation for two to five weeks. In the control group, only IVF was done.

One study compared lifestyle versus no intervention (Mutsaerts 2016). In Mutsaerts 2016, the lifestyle intervention consisted of a six‐month structured programme aiming at a weight loss of 5% to 10% of original body weight. It included six structured outpatient visits and four telephone consultations with a pre‐trained intervention coach. Daily dietary energy intake was reduced by 600 kcal and was maintained at a minimum of 1200 kcal/d. Physical activity was stimulated to a level of 10,000 steps a day and at least 30 minutes of exercise two to three times a week. Behavioural changes were facilitated by motivational counselling. After the six‐month programme was completed, or when weight loss of 5% to 10% had been achieved, women started with appropriate infertility treatment if they were not yet pregnant. The control group received appropriate infertility treatment immediately after randomisation.

One study compared diet versus no intervention in a 12‐week intervention consisting of a very low‐energy diet for the first six weeks followed by a hypocaloric diet, combined with a weekly group multi‐disciplinary programme (Sim 2014). The control group received recommendations for weight loss and the same printed material as the intervention group.

Non‐pharmacological versus non‐pharmacological intervention

One study compared two different diet methods (Rothberg 2016): an intensive weight loss intervention (IWL) and standard of care nutrition counselling (SCN). IWL consisted of 12 weeks of a very low‐energy diet (800 kcal/d) and four weeks of a low‐calorie conventional food‐based diet (CFD) to promote 15% weight loss. SCN consisted of 16 weeks of CFD to promote 5% weight loss.

Pharmacological versus pharmacological intervention

One study compared metformin versus metformin combined with liraglutide (Salamun 2018). Metformin (MET) was initiated at a dose of 500 mg once per day and was increased by 500 mg every three days up to 1000 mg twice daily. In the group given metformin 1000 mg twice daily combined with 1.2 mg liraglutide once daily subcutaneously (COMBI), there was a run‐in period of 12 days to titrate metformin up to 1000 mg twice daily before liraglutide was added. Liraglutide was initiated at a dose of 0.6 mg injected subcutaneously once per day and was increased to 1.2 mg after three days. Medical treatment in both groups lasted 12 weeks.

One study compared dexfenfluramine/placebo versus placebo/dexfenfluramine (Galletly 1996). Dexfenfluramine and placebo were given in a cross‐over design. Dexfenfluramine dosage was 15 mg twice daily, and the duration of each treatment condition was 12 weeks.

One study compared L‐carnitine versus placebo (El 2019). Group 1 (clomiphene citrate (CC) plus metformin and L‐carnitine) received 150 mg/d CC from day 3 to day 7 of the menstrual cycle plus oral L‐carnitine (3 g) and metformin 850 mg (1 tablet daily); the dose was doubled after one week to 1700 mg/d (2 tablets daily). Metformin was ingested before a meal once daily during the first week, and thereafter twice daily. L‐carnitine and metformin were stopped only when pregnancy was documented. Group 2 (CC plus metformin and placebo) received 150 mg/d CC plus metformin (as above) and placebo capsules that were designed to look exactly like L‐carnitine capsules.

Pharmacological intervention versus no intervention or placebo

One study compared diet combined with metformin versus diet combined with placebo (Tang 2006). The intervention group took metformin (850 mg) twice daily over six months. The control group received placebo over six months.

One study compared metformin versus no intervention (Khorram 2006). All patients received CC 100 mg per day on cycle days 5 through 9 only. In group 1 (CC + MET), participants received MET 500 mg three times a day, given on cycle days 1 through 14, with cycle day 1 defined as the first day of menstrual flow. Group 2 (CC) received CC 100 mg per day on cycle days 5 through 9 only.

One study compared metformin versus placebo (Johnson 2010). Women with BMI > 32 received no treatment other than advice and encouragement on a lifestyle intervention (which included advice on calorie restriction and on increasing aerobic exercise to 30 minutes at least five times per week combined with an opportunity to see a dietician and an exercise therapist if required (i.e. standard care)). Women were then allocated to placebo or intervention groups. The intervention group (in addition to standard care) received metformin 500 mg three times daily at a gradually increasing dose over two weeks for six months.

Pharmacological versus non‐pharmacological intervention

No study was found for this comparison.

Excluded studies

We excluded 58 studies from the review for the following reasons.

• 49 of 58 included women not of interest to this review.

• 4 of 58 reported outcomes not of interest to this review and unlikely to ever be measured, as the objectives were different from the objective of this review.

• 1 of 58 reported interventions not of interest to this review.

• 4 of 58 studies are awaiting classification.

Risk of bias in included studies

We have summarised the risk of bias of included studies in Figure 2 and Figure 3.

2.

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

3.

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

Allocation

Random sequence generation

The method of random sequence generation was associated with a low risk of bias in seven trials (Einarsson 2017, El 2019, Johnson 2010, Khorram 2006, Mutsaerts 2016, Tang 2006Sim 2014 ). The three remaining trials were rated as unclear risk of bias as they did not report the methods of randomisation (Rothberg 2016, Galletly 1996, Salamun 2018).

Allocation concealment

Five studies were at low risk of selection bias, for allocation to intervention was Internet‐based or sequentially opaque envelopes were used (El 2019; Einarsson 2017; Johnson 2010; Mutsaerts 2016; Tang 2006). One study was at high risk of selection bias related to allocation concealment through open random allocation (Khorram 2006). Four studies were at unclear risk of selection bias, as they not report adequate details to establish whether an appropriate method of allocation and/or concealment had been used (Galletly 1996; Rothberg 2016; Salamun 2018; Sim 2014).

Blinding

Three studies described blinding of women and were at low risk of performance bias (El 2019; Johnson 2010; Tang 2006). In five studies (Einarsson 2017; Mutsaerts 2016; Rothberg 2016; Salamun 2018; Sim 2014), blinding of treatment assignments was not possible, and the studies were at high risk of performance bias. Two studies did not describe blinding; we judged them to be at unclear risk of bias (Galletly 1996; Khorram 2006).

Related to detection bias (outcome), three studies described blinding of both women and outcome assessors and were at low risk of this bias (El 2019; Johnson 2010; Tang 2006). Three study used no blinding, and we judged this trial to be at high risk (Mutsaerts 2016; Rothberg 2016; Salamun 2018). Four studies were judged to have unclear detection bias (Galletly 1996; Khorram 2006; Sim 2014; Einarsson 2017).

Incomplete outcome data

Eight studies mentioned dropouts or withdrawals, but the numbers were balanced across groups with similar reasons for missing data; therefore, we judged them as having low risk of attrition bias (Einarsson 2017; El 2019; Johnson 2010; Mutsaerts 2016; Rothberg 2016; Salamun 2018; Tang 2006). One study was considered to have high risk of bias for incomplete outcome data in view of the high (20%) drop‐out rate (Sim 2014). We judged two studies to be at unclear risk of attrition bias, as they did not mention dropouts or withdrawals (Galletly 1996; Khorram 2006).

Selective reporting

Seven studies were at low risk of selective bias, as the protocols for these articles were available and their aims were pre‐specified (Einarsson 2017; El 2019; Johnson 2010; Mutsaerts 2016; Rothberg 2016; Salamun 2018; Sim 2014). Three studies were at high risk of selective bias, as the protocols were unavailable (Galletly 1996; Khorram 2006; Tang 2006).

Other potential sources of bias

We did not identify any other potential sources of bias in the included studies, and we judged each of the included studies to be at low risk of other potential sources of bias.

Publication bias

We did not assess potential publication bias using a funnel plot or other corrective analytical methods in view of the small number of studies that could be included in the meta‐analysis (maximally, three RCTs) (Egger 1997).

Effects of interventions

See: Table 1; Table 2; Table 3; Table 4

Non‐pharmacological intervention versus no intervention or placebo

Three studies compared diet or a lifestyle intervention with no intervention: One study compared diet versus no IVF intervention (Einarsson 2017); another study compared diet versus no intervention (Sim 2014); and another study compared lifestyle versus no intervention (Mutsaerts 2016).

Primary outcomes

Live birth rate or ongoing pregnancy

All three studies that compared diet or a lifestyle intervention alone versus no intervention provided data on live birth rates. We found no conclusive evidence of a difference in live birth (odds ratio (OR) 0.85, 95% confidence interval (CI) 0.65 to 1.11; 3 studies, 917 women; I² = 78%; low‐quality evidence; Analysis 1.1). The corresponding relative risk was 0.91 (95% CI 0.78 to 1.06), and use of a random‐effects model resulted in an OR of 1.17 (95% CI 0.55 to 2.48). This suggests that if the chance of live birth following no intervention is assumed to be 43%, the chance following diet or lifestyle change would be 33% to 46%. This result needs to be interpreted with caution in view of high statistical heterogeneity. Heterogeneity in results was caused mainly by the smallest study of 48 women (Sim 2014); excluding this study resulted in an OR for live birth of 0.87 (95% CI 0.75 to 1.02) and I² of 49%.

1.1. Analysis.

Comparison 1: Non‐pharmacological intervention versus no intervention or placebo, Outcome 1: Live birth

In a two‐year follow‐up study of the Einarsson 2017 trial, the cumulative live birth rate was 57% (87/152) and 54% (82/153) for the weight loss/IVF group versus the IVF only group (OR 1.07, 95% CI 0.87 to 1.31).

Adverse events

We are uncertain whether miscarriage may occur more often following diet or lifestyle change compared to no intervention (OR 1.54, 95% CI 0.99 to 2.39; 3 studies, 917 women; I² = 0%; low‐quality evidence; Analysis 1.3).

1.3. Analysis.

Comparison 1: Non‐pharmacological intervention versus no intervention or placebo, Outcome 3: Miscarriage

Secondary outcomes

Clinical pregnancy

For diet alone versus no intervention, we found insufficient evidence of a difference for clinical pregnancy (OR 1.06, 95% CI 0.81 to 1.40, 1.13; 3 studies, 917 women; I² = 73%; low‐quality evidence; Analysis 1.4). This result needs to be interpreted with caution in view of high statistical heterogeneity. Heterogeneity in results was caused mainly by the smallest study of 48 women (Sim 2014); excluding this study resulted in an OR for live birth of 0.98 (95% CI 0.74 to 1.29) and I² of 11%.

1.4. Analysis.

Comparison 1: Non‐pharmacological intervention versus no intervention or placebo, Outcome 4: Clinical pregnancy

Weight change

All three studies found lower BMI in the lifestyle intervention group. One study reported BMI at three months following lifestyle change compared with no intervention (mean difference (MD) ‐1.00, 95% CI ‐1.02 to ‐0.98; 1 study, 574 women) and at six months (MD ‐1.30, 95% CI ‐1.32 to ‐1.28; 1 study, 574 women), but as a large proportion of pregnant women had to be excluded from the analysis, these results are difficult to interpret (Mutsaerts 2016). Diet resulted in a decrease in BMI in two studies (Einarsson 2017; Sim 2014), but data were not pooled due to heterogeneity in effect (MD ‐3.70, 95% CI ‐4.10 to ‐3.36; 305 women, 1 study; low‐quality evidence; and MD ‐1.80, 95% CI ‐2.67 to ‐0.93; 43 women, 1 study; very low‐quality evidence).

In a two‐year follow‐up study of the Einarsson 2017 trial, women in the weight loss group had regained their pre‐trial weight (mean (SD) BMI after 2 years was 32.5 (3.5) and 33.1 (3.0) for the weight loss/IVF versus IVF only arm).

Change in endocrine parameters

This outcome was not reported.

Quality of life or mental health outcome

This outcome was not reported.

Non‐pharmacological versus non‐pharmacological intervention

One study with 11 women compared two different diet methods including intensive weight loss intervention (IWL) versus standard of care nutrition counselling (SCN) (Rothberg 2016).

Primary outcomes

Live birth rate or ongoing pregnancy

There was uncertainty about the effects of IWL versus SCN on live birth (OR 11, 95% CI 0.43 to 284; 1 study, 11 women; very low‐quality evidence; Analysis 2.1). An absolute translation could not be calculated as there were no live births in the SCN group.

2.1. Analysis.

Comparison 2: Non‐pharmacological intervention versus non‐pharmacological intervention, Outcome 1: Live birth

Adverse events

In the trial with 11 women, no pregnancy loss occurred.

Secondary outcomes

Clinical pregnancy

Evidence of a difference in clinical pregnancy for IWL versus SCN was insufficient (OR 11, 95% CI 0.43 to 284; 1 study, 11 women; very low‐quality evidence; Analysis 2.2).

2.2. Analysis.

Comparison 2: Non‐pharmacological intervention versus non‐pharmacological intervention, Outcome 2: Clinical pregnancy

Weight change

In the single trial with 11 women, we found greater weight loss and a larger decrease in BMI for IWL versus SCN (MD ‐9.00 kg, 95% CI ‐15.5 to ‐2.5; and MD ‐3.00, 95% CI ‐5.37 to ‐0.63, respectively), but In view of the very low quality of the evidence, we are uncertain whether IWL reduces weight and BMI to a greater extent than SCN.

Change in endocrine parameters

No data on endocrine parameters were available.

Quality of life or mental health outcome

Evidence of a difference between IWL and SCN in quality of life (MD 0.06, 95% CI ‐0.03 to 0.15; 1 study, 11 women; very low‐quality evidence) and in mental health outcome (MD ‐7.00, 95% CI ‐13.92 to ‐0.08; 1 study, 11 women; very low‐quality evidence) was insufficient.

Pharmacological versus pharmacological intervention

One study compared metformin plus liraglutide versus metformin (Salamun 2018); one study compared a combination of metformin, clomiphene, and L‐carnitine versus metformin, clomiphene, and placebo (El 2019); and one study compared dexfenfluramine versus placebo (Galletly 1996); for this last study, only weight loss as an outcome could be retrieved from the pre‐cross‐over data.

Primary outcomes

Live birth rate or ongoing pregnancy

No comparison of live birth rate or ongoing pregnancy was available.

Adverse events

For metformin plus CC plus L‐carnitine versus metformin plus CC plus placebo (El 2019), we found insufficient evidence of a difference in miscarriage (OR 3.58, 95% CI 0.73 to 17.55; 243 women, 1 study; very low‐quality evidence; Analysis 3.1).

3.1. Analysis.

Comparison 3: Pharmacological intervention versus pharmacological intervention, Outcome 1: Miscarriage

For metformin plus liraglutide versus metformin (Salamun 2018), we found insufficient evidence of a difference for nausea (OR 7.22, 95% CI 0.72 to 72.7; 28 women, 1 study; very low‐quality evidence; Analysis 3.2); diarrhoea (OR 0.31, 95% CI 0.01 to 8.29; 28 women, 1 study; very low‐quality evidence; Analysis 3.3) and headache (OR 5.80, 95% CI 0.25 to 133; 28 women, 1 study; very low‐quality evidence; Analysis 3.4).

3.2. Analysis.

Comparison 3: Pharmacological intervention versus pharmacological intervention, Outcome 2: Adverse event ( nausea)

3.3. Analysis.

Comparison 3: Pharmacological intervention versus pharmacological intervention, Outcome 3: Adverse event (diarrhoea)

3.4. Analysis.

Comparison 3: Pharmacological intervention versus pharmacological intervention, Outcome 4: Adverse event (headache)

Secondary outcomes

Clinical pregnancy

For metformin plus liraglutide versus metformin (Salamun 2018), we found insufficient evidence of a difference in clinical pregnancy (OR 3.67, 95% CI 0.70 to 19.12; 28 women, 1 study; very low‐quality evidence; Analysis 3.5).

3.5. Analysis.

Comparison 3: Pharmacological intervention versus pharmacological intervention, Outcome 5: Clinical pregnancy

For metformin plus clomiphene citrate (CC) plus L‐carnitine versus metformin plus CC plus placebo (El 2019), we found evidence of a difference in favour of L‐carnitine for clinical pregnancy but considered this evidence to be of very low quality (OR 5.56, 95% CI 2.57 to 12.02; 274 women, 1 study; very low‐quality evidence; Analysis 3.5).

Weight change

For metformin plus liraglutide versus metformin (Salamun 2018), we found insufficient evidence of a difference in BMI (MD 2.10, 95% CI ‐0.42 to 4.62; 28 women, 1 study; very low‐quality evidence).

For dexfenfluramine versus placebo (Galletly 1996), we are unsure whether the data for weight loss presents pre‐cross‐over data. We found insufficient evidence of a difference in mean weight loss (MD ‐0.10, 95% CI ‐2.77 to 2.57; 21 women, 1 study; very low‐quality evidence). The mean weight loss ranged from 3 to 4 kg in the intervention group .

One study provided no evidence of a difference for metformin plus liraglutide versus metformin in percentage of body fat (MD ‐0.50, 95% CI ‐4.65 to 3.65; 28 women, 1 study; very low‐quality evidence) (Salamun 2018).

Change in endocrine parameters

For metformin plus liraglutide versus metformin (Salamun 2018), we found insufficient evidence of a difference in the oral glucose tolerance test (OGTT) (MD ‐0.30, 95% CI ‐1.92 to 1.3; 28 women, 1 study; very low‐quality evidence), free testosterone (MD 0.80, 95% CI ‐3.0 to 4.60; 28 women, 1 study; very low‐quality evidence), total testosterone (MD 0.20, 95% CI ‐0.21 to 0.61; 28 women, 1 study; very low‐quality evidence), and sex hormone‐binding globulin (SHBG) (MD 0.30, 95% CI ‐12.22 to 12.82; 28 women, 1 study; very low‐quality evidence).

Quality of life or mental health outcome

No comparison of quality of life or mental health outcome was available.

Pharmacological intervention versus no intervention or placebo

We found three studies: one study compared diet combined with metformin versus diet combined with placebo (Tang 2006); one study compared metformin versus no intervention (Khorram 2006); another study compared metformin versus placebo (Johnson 2010).

Primary outcomes

Live birth rate or ongoing pregnancy

Only one study that compared metformin with placebo reported on live birth, resulting in insufficient evidence of a difference (OR 1.57, 95% CI 0.44 to 5.57; 1 study, 65 women; very low‐quality evidence; Analysis 4.1) (Johnson 2010). This suggests that if the chance of live birth following placebo is assumed to be 15%, the chance following metformin would be 7.3% to 50%.

4.1. Analysis.

Comparison 4: Pharmacological intervention versus no intervention/placebo, Outcome 1: Live birth

Adverse events

One study compared pharmacological intervention versus no treatment or placebo, with no conclusive evidence of a difference for miscarriage (OR 0.50, 95% CI 0.04 to 5.80; 65 women, 1 study; very low‐quality evidence; Analysis 4.2) (Johnson 2010).

4.2. Analysis.

Comparison 4: Pharmacological intervention versus no intervention/placebo, Outcome 2: Miscarriage

We found insufficient evidence of a difference between metformin and placebo in gastrointestinal adverse events (OR 0.91, 95% CI 0.32 to 2.57; 1 study, 65 women; very low‐quality evidence; Analysis 4.3) (Johnson 2010).

4.3. Analysis.

Comparison 4: Pharmacological intervention versus no intervention/placebo, Outcome 3: Adverse event (GI)

Secondary outcomes

Clinical pregnancy

Two studies compared metformin versus no treatment or placebo, with no conclusive evidence of a difference for pregnancy (OR 2.67, 95% CI 0.90 to 7.93; 96 women, 2 studies; I² = 48%; very low‐quality evidence; Analysis 4.4) (Johnson 2010; Khorram 2006)

4.4. Analysis.

Comparison 4: Pharmacological intervention versus no intervention/placebo, Outcome 4: Clinical pregnancy

Weight change

One study reported on weight change, with no conclusive evidence of a change in BMI (MD ‐0.30, 95% CI ‐2.16 to 1.56; 143 women, 1 study; very low‐quality evidence) or WHR (MD 2.00, 95% CI ‐2.21 to 6.21; 1 study, 143 women; very low‐quality evidence) for metformin versus placebo (Tang 2006).

Change in endocrine parameters

One study reported on endocrine parameters (Khorram 2006). Changes in endocrine parameters were inconclusive for free testosterone (MD 0.7 nmol/L, 95% CI ‐1.01 to 2.42; 31 women, 1 study; very low‐quality evidence). Total testosterone and SHBG were higher following metformin versus placebo, but evidence was of very low quality (total testosterone: MD 6.20 nmol/L, 95% CI 0.46 to 11.94; 1 study, 31 women; very low‐quality evidence; SHBG: 5.50 nmol/L, 95% CI 3.79 to 7.21; 1 study, 31 women; very low‐quality evidence).

Quality of life or mental health outcome

No comparison of quality of life or mental health outcome was available.

Non‐pharmacological versus pharmacological intervention

No study reported this comparison.

For this review, we did not gather enough data to perform subgroup analyses. Moreover, for this version of the review, we identified insufficient studies to perform meaningful sensitivity analyses.

Discussion

Summary of main results

Based on available data, we are uncertain about the effectiveness and safety of pharmaceutical and non‐pharmaceutical interventions for weight reduction in obese women with subfertility due to evidence of very low to low quality. We included 10 trials that varied in size from 11 to 564 women. The best evidence was found for diet/lifestyle interventions (three randomised controlled trials (RCTs)). Compared to no intervention or placebo, the lifestyle intervention did not appear to improve live birth or clinical pregnancy; however, lifestyle intervention may lower body mass index (BMI) among obese women. It is unclear whether intensive weight loss interventions compared to nutrition counselling had a positive or negative effect on any outcomes.

For pharmaceutical interventions, all outcomes were scored to have evidence of very low quality. Whether pharmaceutical interventions such as liraglutide, L‐carnitine as addition to metformin, and clomiphene result in improved fertility or weight outcomes is unclear. Similarly, we found no conclusive evidence for metformin versus placebo or no intervention in terms of live birth, clinical pregnancy, or weight change outcomes.

We found no trials that compared pharmacological versus non‐pharmacological interventions.

Overall completeness and applicability of evidence

There is a serious lack of evidence in the field of weight reduction in obese women with subfertility. The included studies only partially addressed the objectives of this review. Outcome data could not be retrieved for several of the comparisons and outcomes that we sought to investigate. The high heterogeneity of pharmacological and non‐pharmacological strategies in included studies may limit the generalisability of trial results regarding the effectiveness of pharmacological and non‐pharmacological strategies for obese women with subfertility. The included studies were clinically heterogeneous and differed in factors such as duration of treatment, type of intervention, dosage, and length of follow‐up. Given the very low quality of evidence for the pharmaceutical interventions, the applicability of those findings is limited and does not allow us to draw conclusions.

It was our intention to include studies with obese women with a BMI of at least 30. In two trials the minimal BMI was 29 (Khorram 2006, Mutsaerts 2016) with medians or means above 36. The study groups consider the effect of including a few women with a BMI of 29 instead of 30 negligible.

Quality of the evidence

Although evidence generated in this review was based on 10 RCTs, this evidence was of very low to low quality as determined by GRADE methods (see 'Summary of findings' for the main comparison and 'Summary of findings 2'). The main limitations were due to lack of studies and poor reporting of study methods. The quality of individual studies was generally low, with over 40% failing to describe adequate methods of blinding of participants and personnel, outcome assessment, and selective reporting (see Figure 2 and Figure 3).

Potential biases in the review process

To minimise bias and issues related to subjectivity of judgement, any disagreements that occurred during the review process were discussed among all review authors until consensus was reached. Two review authors independently carried out data extraction. The accuracy of data was further checked by a third review author. Potential risk of bias in each study and the overall quality of evidence for each outcome were assessed by two independent review authors. We adopted a highly sensitive search strategy. However, the literature identified was predominantly written in English, and most studies were conducted in high‐income countries.

Agreements and disagreements with other studies or reviews

One systematic review evaluated the effectiveness of non‐pharmacological interventions for overweight or obese infertile women (Kim 2020). On the basis of 21 RCTs, it was suggested that non‐pharmacological interventions could have a positive effect on pregnancy and natural conception rates, whereas it remains unclear whether they improve the live birth rate. Not all women in the included studies were overweight, and most were not obese, which may explain the medium to high heterogeneity of effect sizes for pregnancy rates and live birth rates in all non‐pharmacological interventions.

An earlier systematic review evaluated first whether weight loss interventions for infertile patients achieve their goal in reducing weight, and second whether they result in improved fertility outcomes (Best 2017). A total of 40 studies were included, of which 14 were RCTs. Results suggest that weight loss interventions, particularly diet and exercise, may improve pregnancy rates and ovulatory status.

In a Cochrane Review aiming to assess the effectiveness of lifestyle treatment (diet, exercise, behavioural, or combined treatments) for women with PCOS, lifestyle intervention improved body composition, hyperandrogenism, and insulin resistance, but evidence for an effect of diet on reproductive outcomes was lacking (Lim 2019).

Our review differs from previous reviews in that it is focused on obese women only. We conducted an extensive search until 2020 and could include data from 10 studies. In contrast to previous reviews, we are uncertain whether pharmacological or non‐pharmacological strategies improve live birth, ongoing pregnancy, adverse events, clinical pregnancy, quality of life, or mental health outcomes. However, for obese women with subfertility, a lifestyle intervention may reduce BMI.

Authors' conclusions

Implications for practice.

Evidence is yet insufficient to support the use of pharmacological and non‐pharmacological strategies for weight reduction for obese women with subfertility. No data are available for the comparison of pharmacological versus non‐pharmacological strategies. We are uncertain whether pharmacological or non‐pharmacological strategies improve live birth, ongoing pregnancy, adverse events, clinical pregnancy, quality of life, or mental health outcomes. Our findings were based on very low‐quality evidence from studies that may contribute to these outcomes. Thus, we are not able to draw firm conclusions with regard to the impact of pharmacological and non‐pharmacological strategies for weight reduction. However, limited information suggests that lifestyle intervention for obese women with subfertility may reduce BMI. Although obese women should be counselled about the risk of obesity for fertility and pregnancy outcomes, healthcare providers should caution women that the existing research is of poor quality and of limited quantity.

Implications for research.

In view of the high prevalence of obesity‐related subfertility and lack of good evidence, we need well‐designed and well‐conducted RCTs with double‐blinding and adequately powered trials reporting methods such as randomisation and allocation concealment in detail and aiming to share data after the trial. Future studies are advised to focus on lifestyle interventions with or without pharmacological interventions for obese women with subfertility. Follow‐up time needs to be long enough for these clinically relevant outcome data to be obtained. The duration of follow‐up for assessing outcomes should be at least three months, but longer follow‐up is advised to enable reporting of natural pregnancies. Finally, the economic impact of different weight reduction methods should be reported.

What's new

Date	Event	Description
1 April 2021	Amended	Typo corrected in abstract and PLS

History

Protocol first published: Issue 4, 2017
Review first published: Issue 3, 2021

Date	Event	Description
8 February 2020	Amended	Correction to affiliation of 2 authors. Changes to the methods as noted in the section Differences between protocol and review.

Appendices

Appendix 1. Cochrane Gynaecology and Fertility Specialised Register search strategy

ProCite platform

Searched on 18 August 2020

Keywords CONTAINS "IVF" or "ICSI" or "in‐vitro fertilisation " or "in‐vitro fertilisation procedure" or "in vitro fertilization" or "intracytoplasmic sperm injection" or "intracytoplasmic morphologically selected sperm injection" or "superovulation" or "superovulation induction" or "IUI" or "insemination, intrauterine " or "Intrauterine Insemination" or "ART" or "artificial insemination" or "assisted reproduction techniques" or "subfertility‐Female" or "pregnancy" or "live birth" or"unexplained and endometriosis related infertility" or"unexplained infertility" or "unexplained subfertility" or "anovulation" or "infertile"or "infertility" or "ovulation" or "subfertility" or "ovarian hyperstimulation"or "ovarian stimulation"or"controlled ovarian "or "timed intercourse" or "in vivo maturation" or"in vitro maturation" or "IMSI" or "implantation" or "oocyte"or"oocytes"or"embryo" or "polycystic ovary syndrome" or "PCOS" or "assisted reproduction" or "assisted reproductive technology" or Title CONTAINS "infertile"

AND

Keywords CONTAINS "*Obesity" or "obese women" or "obese" or "overweight" or "overweight‐to‐obese" or "fat body mass" or "fat distribution" or "BMI" or "body mass index" or "body composition" or "body fat distribution" or "body fat mass" or "Body Mass" or "Body Mass Index" or"Body weight"or "Weight Loss"or "Diet" or"diet therapy"or "dietary intervention"or"Weight "or"Weight Gain" or Title CONTAINS "*Obesity" or "obese women" or "obese" or"overweight"

or "overweight‐to‐obese" or "fat body mass" or "fat distribution" or "BMI" or "body mass index" or "body composition" or "body fat distribution" or "body fat mass" or "Body Mass" or "Body Mass Index" or "Body weight" or "Weight Loss" or "Diet" or "diet therapy" or "dietary intervention" or "Weight" or"Weight Gain"

1167 records

Appendix 2. CENTRAL via the Cochrane Central Register of Studies Online (CRSO) search strategy

Web platform

Searched on 18 August 2020

#1 MESH DESCRIPTOR Obesity EXPLODE ALL TREES 13487

#2 MESH DESCRIPTOR Overweight EXPLODE ALL TREES 16016

#3 (Obesity or obese or overweight):TI,AB,KY 43053

#4 MESH DESCRIPTOR Obesity, Morbid EXPLODE ALL TREES 1165

#5 MESH DESCRIPTOR Body Mass Index EXPLODE ALL TREES 9957

#6 (High BMI or BMI above or BMI greater):TI,AB,KY 562

#7 (High body mass index or body mass index above or body mass index greater):TI,AB,KY 348

#8 #1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 48495

#9 MESH DESCRIPTOR Infertility, Female EXPLODE ALL TREES 1390

#10 MESH DESCRIPTOR Polycystic Ovary Syndrome EXPLODE ALL TREES 1482

#11 MESH DESCRIPTOR Fertilization in Vitro EXPLODE ALL TREES 2050

#12 (infertil* or subfertil*):TI,AB,KY 8830

#13 (Polycystic Ovar*):TI,AB,KY 3520

#14 PCOS:TI,AB,KY 2823

#15 (ivf or icsi):TI,AB,KY 6480

#16 (intrauterine insemination*):TI,AB,KY 981

#17 iui:TI,AB,KY 881

#18 MESH DESCRIPTOR Ovulation Induction EXPLODE ALL TREES 1340

#19 (Ovulation Induction):TI,AB,KY 2571

#20 (ovar* hyperstimulation):TI,AB,KY 1625

#21 (ovar* adj2 stimulation):TI,AB,KY 2263

#22 MESH DESCRIPTOR Reproductive Techniques, Assisted EXPLODE ALL TREES 3136

#23 (assisted reproduct*):TI,AB,KY 1401

#24 #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20 OR #21 OR #22 OR #23 16115

#25 #8 AND #24 1098

Appendix 3. MEDLINE search strategy

Ovid platform

Searched from 1946 to 18 August 2020

1 (Obesity or obese or overweight).tw. (315125)
2 exp Obesity/ or exp Overweight/ or exp Body Weight/ (466482)
3 exp Body Composition/ or exp Body Fat Distribution/ (54864)
4 exp Body Mass Index/ (126990)
5 exp Obesity, Morbid/ or exp Waist‐Hip Ratio/ (23856)
6 (High BMI or BMI above).tw. (4009)
7 (BMI adj3 over).tw. (1563)
8 Body Mass Index.tw. (183850)
9 or/1‐8 (743472)
10 exp Infertility, Female/ (28492)
11 exp Polycystic Ovary Syndrome/ (14478)
12 exp Fertilization in Vitro/ (35811)
13 (infertil$ adj5 female$).tw. (3518)
14 (infertil$ adj5 wom?n).tw. (9574)
15 (subfertil$ adj5 wom?n).tw. (635)
16 (subfertil$ adj5 female$).tw. (253)
17 Polycystic Ovar$.tw. (16819)
18 PCOS.tw. (11371)
19 exp Amenorrhea/ (9855)
20 amenorrh?ea.tw. (13347)
21 oligomenorrh$.tw. (1437)
22 exp Oligomenorrhea/ (720)
23 exp Hyperandrogenism/ (2019)
24 hyperandrogenism.tw. (3816)
25 (ivf or icsi).tw. (27791)
26 intrauterine insemination$.tw. (2524)
27 iui.tw. (1785)
28 exp Ovarian Hyperstimulation Syndrome/ (2262)
29 exp Ovulation Induction/ (13151)
30 ovar$ hyperstimulation.tw. (5118)
31 ovar$ stimulation.tw. (6023)
32 exp Reproductive Techniques, Assisted/ (69677)
33 assisted reproduct$.tw. (15166)
34 or/10‐33 (139274)
35 randomized controlled trial.pt. (511179)
36 controlled clinical trial.pt. (93800)
37 randomized.ab. (489167)
38 randomised.ab. (97651)
39 placebo.tw. (215911)
40 clinical trials as topic.sh. (192520)
41 randomly.ab. (338981)
42 trial.ti. (223326)
43 (crossover or cross‐over or cross over).tw. (85722)
44 or/35‐43 (1373935)
45 exp animals/ not humans.sh. (4725508)
46 44 not 45 (1265255)
47 9 and 34 and 46 (1266)

Appendix 4. Embase search strategy

Ovid platform

Searched from 1980 to 18 August 2020

1 exp obesity/ (512611)
2 (obesity or obese or overweight).tw. (457990)
3 exp body weight/ or exp body mass/ (832100)
4 waist hip ratio/ (14604)
5 body mass index.tw. (268127)
6 (High BMI or BMI above).tw. (7500)
7 (BMI adj3 over).tw. (3104)
8 (weight adj3 above).tw. (1994)
9 (weight adj3 over).tw. (10556)
10 or/1‐9 (1244347)
11 exp female infertility/ or exp ovary polycystic disease/ or exp fertilization in vitro/ (122255)
12 (infertil$ adj5 female$).tw. (5121)
13 (infertil$ adj5 wom?n).tw. (14133)
14 (subfertil$ adj5 wom?n).tw. (1029)
15 (subfertil$ adj5 female$).tw. (369)
16 Polycystic Ovar$.tw. (23277)
17 PCOS.tw. (17262)
18 exp amenorrhea/ (17330)
19 amenorrh?ea.tw. (15407)
20 oligomenorrh$.tw. (1973)
21 exp oligomenorrhea/ or exp "amenorrhea and oligomenorrhea"/ (27942)
22 exp hyperandrogenism/ (7408)
23 hyperandrogenism.tw. (5559)
24 (ivf or icsi).tw. (47745)
25 intrauterine insemination$.tw. (3780)
26 iui.tw. (3278)
27 exp ovary hyperstimulation/ or exp ovulation induction/ (22001)
28 ovar$ hyperstimulation.tw. (7484)
29 ovar$ stimulation.tw. (9753)
30 or/11‐29 (177709)
31 exp Diet Therapy/ (340480)
32 diet$.tw. (681613)
33 (weight adj3 reduc$).tw. (48577)
34 (body mass index adj2 loss).tw. (677)
35 (body mass index adj2 reduc$).tw. (1197)
36 (body mass index adj2 decreas$).tw. (1359)
37 (BMI adj2 loss).tw. (1727)
38 (BMI adj2 redu$).tw. (4064)
39 (BMI adj2 decreas$).tw. (4800)
40 exercise$.tw. (376424)
41 exp sport/ (160081)
42 (run$ or jog$).tw. (253096)
43 (sport$ or walk$).tw. (255446)
44 (swim$ or cycl$).tw. (1430222)
45 (train or training).tw. (553768)
46 exp cognitive therapy/ or exp psychotherapy/ (235459)
47 (cognitive adj2 therap$).tw. (29865)
48 Psychotherapy.tw. (40737)
49 exp behavior therapy/ (42495)
50 exp lifestyle/ (129810)
51 (lifestyle adj2 change$).tw. (15252)
52 (lifestyle adj2 intervention$).tw. (11337)
53 exp social support/ (91720)
54 (social adj2 support).tw. (49798)
55 weight loss.tw. (138913)
56 (weight adj2 control).tw. (12071)
57 dynamic exercise/ or isotonic exercise/ or exercise/ or aquatic exercise/ or leg exercise/ or anaerobic exercise/ or stretching exercise/ or aerobic exercise/ or isometric exercise/ (289306)
58 behavio?r modif$.tw. (3258)
59 exp weight control/ or exp weight reduction/ (39254)
60 behavio?r therap$.tw. (9492)
61 low calorie$.tw. (4447)
62 fitness.tw. (81610)
63 exp health behavior/ (409721)
64 (decrease adj2 weight).tw. (3343)
65 hypnosis.tw. (7140)
66 group therap$.tw. (5887)
67 or/31‐66 (4363927)
68 exp antiobesity agent/ (4884)
69 antiobesity.tw. (2254)
70 lipase inhibitor$.tw. (1255)
71 (xenical or tetrahydrolipstatin or orlistat).tw. (3754)
72 exp tetrahydrolipstatin/ (6440)
73 exp metformin/ or exp pioglitazone/ or exp insulin sensitizing agent/ (75303)
74 insulin sensitizer$.tw. (2157)
75 exp antidiabetic agent/ (461406)
76 exp 2,4 thiazolidinedione derivative/ (12894)
77 exp thiazole derivative/ (168176)
78 metformin.tw. (35181)
79 (Appetite adj3 (suppress$ or depress$)).tw. (2798)
80 exp serotonin uptake inhibitor/ (256753)
81 exp serotonin antagonist/ (221959)
82 exp antidepressant agent/ (419880)
83 exp noradrenalin uptake inhibitor/ (216863)
84 (Reductil or sibutramine or fenfluramine).tw. (4271)
85 fenfluramine/ (5653)
86 exp dexfenfluramine/ (2264)
87 exp phentermine/ (2547)
88 exp phenylpropanolamine/ (3042)
89 exp fluoxetine/ (46337)
90 antidepressant$.tw. (89208)
91 anti depressant$.tw. (3456)
92 exp mazindol/ (1611)
93 exp amfepramone/ (1320)
94 antiandrogen/ (11465)
95 androgen$ antagonist$.tw. (323)
96 bulking agent$.tw. (1853)
97 fluoxetine.tw. (16185)
98 (methylcellulose or celevac).tw. (7244)
99 guar gum.tw. (2212)
100 anti obesity.tw. (4963)
101 exp Ephedra/ (1143)
102 ephedra.tw. (1024)
103 bupropion.tw. (6245)
104 exp amfebutamone/ (18316)
105 (Wellbutrin or Zyban or Amfebutamone).tw. (2481)
106 zonisamide.tw. (2062)
107 (Excegran or Zonegran).tw. (379)
108 sertraline.tw. (6775)
109 exp sertraline/ (25989)
110 (Serad or Serlain or Tresleen or Zoloft).tw. (2616)
111 leptin/ae, dt [Adverse Drug Reaction, Drug Therapy] (853)
112 topiramate.tw. (7652)
113 or/68‐112 (1157634)
114 (lap band$ or lapband$).tw. (629)
115 roux‐en‐y.tw. (19567)
116 bariatric surger$.tw. (30089)
117 exp gastroplasty/ or exp bariatric surgery/ (46918)
118 exp gastrectomy/ (52328)
119 (GASTROPLASTY or Gastrectomy or gastric surgery or Gastric Bypass or gastric band$).tw. (60946)
120 (Biliopancreatic Diversion$ or biliopancreatic bypass$ or gastro$gastrostomy or restrictive surgery).tw. (2030)
121 (obesity adj3 surg$).tw. (6397)
122 (obese adj3 surg$).tw. (3076)
123 (jejunoileal bypass$ or jejuno ileal bypass$).tw. (789)
124 exp gastric banding/ (7411)
125 or/114‐124 (106478)
126 67 or 113 or 125 (5321490)
127 Clinical Trial/ (970755)
128 Randomized Controlled Trial/ (611928)
129 exp randomization/ (87710)
130 Single Blind Procedure/ (39825)
131 Double Blind Procedure/ (172019)
132 Crossover Procedure/ (63930)
133 Placebo/ (339738)
134 Randomi?ed controlled trial$.tw. (234558)
135 Rct.tw. (38031)
136 random allocation.tw. (2043)
137 randomly allocated.tw. (35686)
138 allocated randomly.tw. (2562)
139 (allocated adj2 random).tw. (822)
140 Single blind$.tw. (25054)
141 Double blind$.tw. (204347)
142 ((treble or triple) adj blind$).tw. (1172)
143 placebo$.tw. (305406)
144 prospective study/ (618991)
145 or/127‐144 (2217034)
146 case study/ (71157)
147 case report.tw. (408717)
148 abstract report/ or letter/ (1110703)
149 or/146‐148 (1579736)
150 145 not 149 (2163027)
151 10 and 30 and 126 and 150 (3462)

Appendix 5. PsycINFO search strategy

Ovid platform

Searched from 1806 to 18 August 2020

1 exp Obesity/ (24601)
2 (obesity or obese or overweight).tw. (44923)
3 exp Body Weight/ or exp Body Mass Index/ (56375)
4 Body Mass Index.tw. (20444)
5 Body Weight.tw. (16156)
6 BMI.tw. (17893)
7 or/1‐6 (88421)
8 exp Infertility/ (2185)
9 (infertil$ adj5 female$).tw. (357)
10 (infertil$ adj5 wom?n).tw. (742)
11 (subfertil$ adj5 wom?n).tw. (16)
12 (subfertil$ adj5 female$).tw. (6)
13 Polycystic Ovar$.tw. (426)
14 PCOS.tw. (290)
15 exp Amenorrhea/ (264)
16 amenorrh?ea.tw. (830)
17 oligomenorrh$.tw. (52)
18 exp Menstrual Disorders/ (1270)
19 hyperandrogenism.tw. (97)
20 (ivf or icsi).tw. (604)
21 intrauterine insemination$.tw. (32)
22 iui.tw. (41)
23 exp Reproductive Technology/ (1847)
24 ovar$ hyperstimulation.tw. (13)
25 ovar$ stimulation.tw. (26)
26 or/8‐25 (6184)
27 7 and 26 (572)
28 random.tw. (59084)
29 control.tw. (448804)
30 double‐blind.tw. (23062)
31 clinical trials/ (11727)
32 placebo/ (5682)
33 exp Treatment/ (1052281)
34 or/28‐33 (1452988)
35 27 and 34 (225)

Appendix 6. AMED search strategy

Ovid platform

Searched from 1985 to 18 August 2020

1 exp Obesity/ (2115)
2 (obesity or obese or overweight).tw. (3015)
3 exp Body weight/ (848)
4 BMI.tw. (1134)
5 body mass index.tw. (2127)
6 weight.tw. (9971)
7 or/1‐6 (13101)
8 exp Fertility/ (68)
9 exp Infertility female/ (215)
10 exp Ovarian disease/ (275)
11 (infertil$ adj5 female$).tw. (227)
12 (infertil$ adj5 wom?n).tw. (57)
13 (subfertil$ adj5 wom?n).tw. (2)
14 (subfertil$ adj5 female$).tw. (0)
15 Polycystic Ovar$.tw. (98)
16 PCOS.tw. (54)
17 exp Amenorrhea/ (39)
18 amenorrh?ea.tw. (95)
19 oligomenorrh$.tw. (11)
20 exp Menstruation disorders/ (589)
21 hyperandrogenism.tw. (6)
22 (ivf or icsi).tw. (52)
23 intrauterine insemination$.tw. (8)
24 iui.tw. (6)
25 ovar$ hyperstimulation.tw. (5)
26 or/8‐25 (1202)
27 7 and 26 (64)

Appendix 7. CINAHL search strategy

Ebsco platform

Searched from 1961 to 26 September 2019. The CENTRAL search on 18 August 2020 includes CINAHL references.

#	Query	Results
S35	S20 AND S34	771
S34	S21 OR S22 or S23 or S24 OR S25 OR S26 OR S27 OR S28 OR S29 OR S30 OR S31 OR S32 OR S33	1,352,629
S33	TX allocat* random*	11,021
S32	(MH "Quantitative Studies")	23,323
S31	(MH "Placebos")	11,475
S30	TX placebo*	59,197
S29	TX random* allocat*	11,021
S28	(MH "Random Assignment")	56,644
S27	TX randomi* control* trial*	175,687
S26	TX ( (singl* n1 blind*) or (singl* n1 mask*) ) or TX ( (doubl* n1 blind*) or (doubl* n1 mask*) ) or TX ( (tripl* n1 blind*) or (tripl* n1 mask*) ) or TX ( (trebl* n1 blind*) or (trebl* n1 mask*) )	1,033,852
S25	TX ( (trebl* n1 blind*) or (trebl* n1 mask*) )	241
S24	TX ( (trebl* n1 blind*) or (trebl* n1 mask*) )	241
S23	TX clinic* n1 trial*	251,068
S22	PT Clinical trial	86,871
S21	(MH "Clinical Trials+")	267,348
S20	S9 AND S19	3,829
S19	S10 OR S11 OR S12 OR S13 OR S14 OR S15 OR S16 OR S17 OR S18	27,803
S18	TX intrauterine insemination	462
S17	(MH "Fertilization in Vitro") OR (MH "Fertility")	9,858
S16	TX (ivf or icsi)	4,846
S15	TX hyperandrogenism	697
S14	TX subfertil*	885
S13	TX infertil*	16,371
S12	TX Polycystic Ovar*	4,117
S11	(MM "Polycystic Ovary Syndrome")	2,587
S10	(MM "Infertility")	7,396
S9	S1 OR S2 OR S3 OR S4 OR S5 OR S6 OR S7 OR S8	343,194
S8	TX weight	211,089
S7	TX BMI	43,745
S6	TX Body Mass Index	102,302
S5	(MM "Body Mass Index")	11,041
S4	TX overweight	27,892
S3	TX obese	35,589
S2	TX Obesity	119,085
S1	(MH "Obesity+")	88,246

Data and analyses

Comparison 1. Non‐pharmacological intervention versus no intervention or placebo.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Live birth	3	918	Odds Ratio (M‐H, Fixed, 95% CI)	0.85 [0.65, 1.11]
1.2 Ongoing pregnancy	1	564	Odds Ratio (M‐H, Fixed, 95% CI)	0.81 [0.58, 1.13]
1.3 Miscarriage	3	917	Odds Ratio (M‐H, Fixed, 95% CI)	1.54 [0.99, 2.39]
1.4 Clinical pregnancy	3	917	Odds Ratio (M‐H, Fixed, 95% CI)	1.06 [0.81, 1.40]
1.5 BMI change	2		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

1.2. Analysis.

Comparison 1: Non‐pharmacological intervention versus no intervention or placebo, Outcome 2: Ongoing pregnancy

1.5. Analysis.

Comparison 1: Non‐pharmacological intervention versus no intervention or placebo, Outcome 5: BMI change

Comparison 2. Non‐pharmacological intervention versus non‐pharmacological intervention.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Live birth	1	11	Odds Ratio (M‐H, Fixed, 95% CI)	11.00 [0.43, 284.30]
2.2 Clinical pregnancy	1	11	Odds Ratio (M‐H, Fixed, 95% CI)	11.00 [0.43, 284.30]
2.3 BMI change	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
2.4 Weight change	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
2.5 Mental health	1	11	Mean Difference (IV, Fixed, 95% CI)	‐7.00 [‐13.92, ‐0.08]
2.6 Quality of life	1	11	Mean Difference (IV, Fixed, 95% CI)	0.06 [‐0.03, 0.15]

2.3. Analysis.

Comparison 2: Non‐pharmacological intervention versus non‐pharmacological intervention, Outcome 3: BMI change

2.4. Analysis.

Comparison 2: Non‐pharmacological intervention versus non‐pharmacological intervention, Outcome 4: Weight change

2.5. Analysis.

Comparison 2: Non‐pharmacological intervention versus non‐pharmacological intervention, Outcome 5: Mental health

2.6. Analysis.

Comparison 2: Non‐pharmacological intervention versus non‐pharmacological intervention, Outcome 6: Quality of life

Comparison 3. Pharmacological intervention versus pharmacological intervention.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Miscarriage	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
3.2 Adverse event ( nausea)	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
3.3 Adverse event (diarrhoea)	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
3.4 Adverse event (headache)	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
3.5 Clinical pregnancy	2		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
3.5.1 Metformin plus liraglutide vs metformin	1	28	Odds Ratio (M‐H, Fixed, 95% CI)	3.67 [0.70, 19.12]
3.5.2 Metformin + CC + L‐carnitine vs metformin + CC + placebo	1	274	Odds Ratio (M‐H, Fixed, 95% CI)	5.56 [2.57, 12.02]
3.6 BMI change	2		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
3.6.1 Metformin + CC + L‐carnitine vs metformin + CC + placebo	1	274	Mean Difference (IV, Fixed, 95% CI)	‐0.30 [‐1.17, 0.57]
3.6.2 Metformin plus liraglutide vs metformin	1	28	Mean Difference (IV, Fixed, 95% CI)	2.10 [‐0.42, 4.62]
3.7 Weight loss	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
3.7.1 Dexfenfluramine vs placebo	1	21	Mean Difference (IV, Fixed, 95% CI)	‐0.10 [‐2.77, 2.57]
3.8 Percent of total body fat	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
3.8.1 Metformin plus liraglutide vs metformin	1	28	Mean Difference (IV, Fixed, 95% CI)	‐0.50 [‐4.65, 3.65]
3.9 Glucose test (OGTT)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
3.9.1 Metformin plus liraglutide vs metformin	1	28	Mean Difference (IV, Fixed, 95% CI)	‐0.30 [‐1.90, 1.30]
3.10 Free testosterone	2		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
3.10.1 Metformin plus liraglutide vs metformin	1	28	Mean Difference (IV, Fixed, 95% CI)	0.80 [‐3.00, 4.60]
3.10.2 Metformin + CC + L‐carnitine vs metformin + CC + placebo	1	274	Mean Difference (IV, Fixed, 95% CI)	‐1.76 [‐2.06, ‐1.46]
3.11 Total testosterone	1	28	Mean Difference (IV, Fixed, 95% CI)	0.20 [‐0.21, 0.61]
3.11.1 Metformin plus liraglutide vs metformin	1	28	Mean Difference (IV, Fixed, 95% CI)	0.20 [‐0.21, 0.61]
3.12 SHBG	1	28	Mean Difference (IV, Fixed, 95% CI)	0.30 [‐12.22, 12.82]
3.12.1 Metformin plus liraglutide vs metformin	1	28	Mean Difference (IV, Fixed, 95% CI)	0.30 [‐12.22, 12.82]

3.6. Analysis.

Comparison 3: Pharmacological intervention versus pharmacological intervention, Outcome 6: BMI change

3.7. Analysis.

Comparison 3: Pharmacological intervention versus pharmacological intervention, Outcome 7: Weight loss

3.8. Analysis.

Comparison 3: Pharmacological intervention versus pharmacological intervention, Outcome 8: Percent of total body fat

3.9. Analysis.

Comparison 3: Pharmacological intervention versus pharmacological intervention, Outcome 9: Glucose test (OGTT)

3.10. Analysis.

Comparison 3: Pharmacological intervention versus pharmacological intervention, Outcome 10: Free testosterone

3.11. Analysis.

Comparison 3: Pharmacological intervention versus pharmacological intervention, Outcome 11: Total testosterone

3.12. Analysis.

Comparison 3: Pharmacological intervention versus pharmacological intervention, Outcome 12: SHBG

Comparison 4. Pharmacological intervention versus no intervention/placebo.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
4.1 Live birth	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
4.2 Miscarriage	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
4.3 Adverse event (GI)	1	65	Odds Ratio (M‐H, Fixed, 95% CI)	0.91 [0.32, 2.57]
4.4 Clinical pregnancy	2	96	Odds Ratio (M‐H, Fixed, 95% CI)	2.67 [0.90, 7.93]
4.5 BMI change	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
4.6 WHR	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
4.7 Total testosterone	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
4.8 Free testosterone	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
4.9 SHBG	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

4.5. Analysis.

Comparison 4: Pharmacological intervention versus no intervention/placebo, Outcome 5: BMI change

4.6. Analysis.

Comparison 4: Pharmacological intervention versus no intervention/placebo, Outcome 6: WHR

4.7. Analysis.

Comparison 4: Pharmacological intervention versus no intervention/placebo, Outcome 7: Total testosterone

4.8. Analysis.

Comparison 4: Pharmacological intervention versus no intervention/placebo, Outcome 8: Free testosterone

4.9. Analysis.

Comparison 4: Pharmacological intervention versus no intervention/placebo, Outcome 9: SHBG

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Einarsson 2017.

Study characteristics
Methods	Multi‐centre, multi‐disciplinary, prospective, randomised controlled trial
Participants	Setting: 9 infertility clinics in Sweden, Denmark, and Iceland Inclusion: infertile women between 18 and 38 years of age with indications for IVF and planning to start their first, second, or third IVF treatment and with BMI ≥ 30 and < 35 kg/m² Exclusion: women were excluded from the trial if they had insulin‐dependent diabetes mellitus and other exclusion factors such as planned oocyte donation, planned pre‐implantation genetic diagnosis, husband with azoospermia known at randomisation, less than adequate knowledge of the local language, binge eating disorder, or previous study participation
Interventions	Comparison Intervention group: weight reduction before IVF, starting with 12 weeks of a low‐calorie liquid formula diet (LCD) of 880 kcal/d and thereafter weight stabilisation for 2 to 5 weeks Control group: IVF only
Outcomes	Live birth rate Weight reduction (change in BMI) Clinical pregnancy Ongoing pregnancy Miscarriage
Notes	Conflicts of interest: some issues are presented in the article Funding: Sahlgrenska University Hospital (ALFGBG‐70 940), Merck AB Solna Sweden (an affiliate of Merck KGaA, Darmstadt, Germany), Impolin AB, Hjalmar Svensson Foundation, and Jane and Dan Olsson Foundation for Science. Funders had no role in the design of the study, statistical analysis, or interpretation of study results, nor in writing the article or deciding to submit it for publication Date study was conducted: 10 May 2010 Clinical trial registration number: NCT01566929 Trial authors contacted: Ann Thurin‐Kjellberg
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computerised randomisation programme
Allocation concealment (selection bias)	Low risk	Online concealed allocation of patients in the proportion of 1:1. Optimal allocation was applied according to Pocock’s minimisation technique for sequential randomisation, taking account of the number of previously performed fresh IVF cycles and age of the woman, as well as parity, polycystic ovarian syndrome (PCOS), fertilisation method planned, tubal factor, smoking, BMI, and waist circumference
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding was not possible for patients or physicians
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Embryologists and statisticians were unaware as to which group participants were allocated but as the women and investigators were aware we can't rule out detection bias
Incomplete outcome data (attrition bias) All outcomes	Low risk	Missing outcome data were balanced in numbers across intervention groups, with similar reasons for missing data across groups In the weight reduction and IVF group, 1 patient did not receive the allocated intervention and 7 discontinued the intervention. In the IVF only group, 2 patients did not receive the allocated intervention and 2 discontinued the intervention. No patients were lost to follow‐up
Selective reporting (reporting bias)	Low risk	The study protocol is available, and all of the study's pre‐specified outcomes have been reported in the results https://clinicaltrials.gov/ct2/show/NCT01566929
Other bias	Low risk	We did not identify any other potential sources of bias in the study, and we judged low risk for other potential sources of bias

El 2019.

Study characteristics
Methods	Double‐blinded randomised controlled clinical trial
Participants	Setting: Department of Obstetrics and Gynecology, Faculty of Medicine, Zagazig University, Zagazig, Egypt Inclusion:obesity, infertility Exclusion: smokers, drug users, those with other causes of infertility such as male factor or tubal factor, those with endocrine disorders such as thyroid dysfunction or hyperprolactinaemia
Interventions	Group 1 (CC plus metformin and L‐carnitine): received 150 mg/d CC from day 3 till day 7 of menstrual cycle plus oral L‐carnitine 3 g and metformin 850 mg (1 tablet daily), then the dose was doubled after 1 week to 1700 mg/d (2 tablets daily). Metformin was ingested before a meal once daily during the first week and thereafter twice daily. L‐carnitine and metformin were stopped only when pregnancy was documented. Group 2 (CC plus metformin and placebo): received 150 mg/d CC plus metformin (as above) and placebo capsules that were designed to look exactly like L‐carnitine capsules.
Outcomes	Pregnancy rate Miscarriage rate BMI Free testosterone
Notes	Conflicts of interest: no Funding: Zagazig University Date study was conducted: January 2017 Clinical trial registration number: NCT03108963 Trial authors contacted:El Sharkwy I.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computerised randomisation programme
Allocation concealment (selection bias)	Low risk	Allocation was concealed in opaque, sealed, serially numbered envelopes
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Women, treating physicians, and investigators were blinded to treatment allocation
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Women, treating physicians, and investigators were blinded to treatment allocation
Incomplete outcome data (attrition bias) All outcomes	Low risk	Missing outcome data were balanced in numbers across intervention groups, with similar reasons for missing data across groups A total of 375 women were assessed for eligibility. Of them, 95 were excluded (60 did not meet the inclusion criteria and 35 refused to participate). The consenting 280 women were randomly allocated to group 1 (n = 140) or group 2 (n = 140). Six women were excluded from analysis due to loss to follow‐up: 2 women in group 1 and 4 women in group 2
Selective reporting (reporting bias)	Low risk	The study protocol is available, and all of the study's pre‐specified outcomes have been reported in the results
Other bias	Low risk	We did not identify any other potential sources of bias in the study, and we judged low risk for other potential sources of bias

Galletly 1996.

Study characteristics
Methods	Double‐blind, cross‐over design
Participants	Setting: reproductive medicine clinic, Australia Inclusion: obese infertile women Exclusion: not mentioned
Interventions	Comparison: dexfenfluramine and placebo were given in a double‐blind cross‐over design Dexfenfluramine dosage was 15 mg twice daily, and the duration of each treatment condition was 12 weeks for placebo/dexfenfluramine (n = 11) and dexfenfluramine/placebo (n = 10). We tried to extract only the pre‐cross‐over data
Outcomes	Weight loss, this was the only useable outcome but not sure whether truly pre‐cross‐over data Self‐esteem score Depression score Anxiety score
Notes	Data regarding self‐esteem, depression, and BMI are not presented separately for each group. Conflicts of interest: not mentioned Funding: not mentioned Date study was conducted: not mentioned Clinical trial registration number: not mentioned Trial authors contacted: Cherrie Galletly
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information about sequence generation
Allocation concealment (selection bias)	Unclear risk	Method of concealment not described
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Insufficient information to permit judgement
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Insufficient information to permit judgement
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	The study did not address this outcome
Selective reporting (reporting bias)	High risk	The study protocol is unavailable, and all of the study's pre‐specified outcomes have not been reported in the results At the beginning, midpoint, and end of the study, patients completed the Rosenberg Self‐Esteem Scale and the Hospital Anxiety and Depression Scale. Weight patients were weighed weekly throughout the study BMI is not an outcome measure but it was evaluated
Other bias	Low risk	We did not identify any other potential sources of bias in the study, and we judged low risk for other potential sources of bias

Johnson 2010.

Study characteristics
Methods	Double‐blinded multi‐centre randomised trial
Participants	Setting: multi‐centre, New Zealand Inclusion: anovulatory or oligo‐ovulatory women with PCOS Exclusion: couples who had undergone previous fertility treatment involving more than 5 months of treatment with CC or metformin; with any other important infertility factor known to be present, including known tubal factor in which at least 1 fallopian tube was blocked (although a tubal potency test was not a prerequisite for trial entry); with important medical disorders in women
Interventions	Comparison Women with BMI > 32 received no treatment other than advice and encouragement on lifestyle intervention (which included advice on calorie restriction and on increasing aerobic exercise to 30 minutes at least 5 times per week with an opportunity to see a dietician and an exercise therapist if required (i.e. standard care)): Placebo Intervention: metformin (in addition to standard care) for 6 months; metformin 500 mg 3 times daily at a gradually increasing dose over 2 weeks was given; for CC, 50 mg was the initial dose and 150 mg the highest dose used
Outcomes	Clinical pregnancy Live birth Spontaneous abortion Adverse events In this study, patients were divided according to BMI < 32 and BMI ≥ 32. We included only data from BMI ≥ 32 in this review
Notes	Conflicts of interest: NPJ reports receiving travel support from Serono, Organon, Bayer‐Schering, and Device Technologies New Zealand, and funding for a research meeting from Serono. VPS reports receiving travel support from Serono Funding: Auckland Medical Research Foundation, Mercia Barnes Trust, and University of Auckland Research Committee. Funders played no role in the design, the conduct of the research, or the decision to publish Date study was conducted: August 2003 Clinical trial registration number: NCT00795808 Trial authors contacted: NP Johnson
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	In this study, it was mentioned that complete description of patients is provided Randomisation was done through a computer‐generated randomisation process
Allocation concealment (selection bias)	Low risk	Concealment was strictly maintained by a telephone call from the recruiting research nurse to the pharmacy
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Blinding (masking) of all parties (women and personnel) was maintained in all cases by placebo control until the end of the course of treatment or, in the event of pregnancy, until after the pregnancy
Blinding of outcome assessment (detection bias) All outcomes	Low risk	The research pharmacist executed the assignment by dispensing pre‐prepared drugs by true third party randomisation. Blinding (masking) of all parties was maintained in all cases by placebo control until the end of the course of treatment or, in the event of pregnancy, until after the pregnancy
Incomplete outcome data (attrition bias) All outcomes	Low risk	Of 33 women with BMI ≥ 32 kg/m² receiving placebo, 30 completed treatment and follow‐up (2 of whom were not fully adherent to treatment ‐ 1 experienced side effects so took a reduced dose; 1 misunderstood the gradual increase in dose and did this every month) 3 breached the protocol by stopping trial medications; 25 women who had not had confirmed ovulation 3 months into the trial received CC thereafter. Among 32 women with BMI ≥ 32 kg/m² receiving metformin, 29 completed treatment and follow‐up (all of whom were fully adherent to treatment) ‐ 2 women were lost to follow‐up, 1 of whom was pregnant at the time of emigration to Australia, and 1 who stopped trial medication; 22 women who had not had confirmed ovulation 3 months into the trial received CC thereafter
Selective reporting (reporting bias)	Low risk	The study protocol is available, but all of the study's pre‐specified outcomes have been reported in the results https://clinicaltrials.gov/ct2/show/NCT00795808
Other bias	Low risk	We did not identify any other potential sources of bias in the study, and we judged low risk for other potential sources of bias

Khorram 2006.

Study characteristics
Methods	A randomised prospective trial
Participants	Setting: university‐based medical centre, California Inclusion: anovulatory or oligo‐ovulatory cycles (35 days or 8 cycles/year), polycystic ovaries on a baseline ultrasound, hyperandrogenism (hirsutism, acne, alopecia, or elevated testosterone), BMI > 29 kg/m², desire for fertility Exclusion: pregnancy, hepatic disease, renal disease, heart disease, alcoholism, pulmonary disease, thyroid disease, prolactinoma, congenital adrenal hyperplasia, androgen‐secreting tumour
Interventions	Comparison Group 1 (CC + met): participants received MET 500 mg 3 times a day, given on cycle days 1 through 14, with cycle day 1 defined as the first day of menstrual flow after a 10‐day course of medroxyprogesterone acetate (10 mg daily) challenge in combination with CC 100 mg per day taken on days 5 through 9 of the cycle. In the group receiving MET, the entire dose of the medication (1500 mg/d) was taken from the start of treatment (n = 16) Group 2 (CC): participants received CC 100 mg per day on cycle days 5 through 9 only. The dose of CC chosen (100 mg/d) was based on significant obesity in the population and the known ineffectiveness of lower doses of CC in these patients (n = 15)
Outcomes	Free glucose Total and free Testosterone SHBG Pregnancy rate
Notes	Conflicts of interest: not mentioned Funding: not mentioned Date study was conducted: not found Clinical trial registration number: not found, at https://clinicaltrials.gov/ Trial authors contacted:Khorram O
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was done by picking a card out of a box, which had 1 or the other treatment written on it; the participant was assigned to that group
Allocation concealment (selection bias)	High risk	Randomisation was done by picking a card out of a box, which had 1 or the other treatment written on it; the participant was assigned to that group
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Probably not blinded
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Probably not blinded
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	The study lacks details of dropout rates
Selective reporting (reporting bias)	High risk	The study protocol is unavailable
Other bias	Low risk	We did not identify any other potential sources of bias in the study, and we judged low risk for other potential sources of bias

Mutsaerts 2016.

Study characteristics
Methods	Secondary analysis of a multi‐centre RCT (randomised controlled trial)
Participants	Setting: 6 university medical centres and 17 general hospitals in The Netherlands Inclusion: infertile women between 18 and 39 years of age with BMI ≥ 29 Exclusion: women with severe endometriosis, premature ovarian failure, or endocrinopathy (e.g. women with type 1 diabetes, women with Cushing’s syndrome). Those who were eligible for donor insemination because of azoospermia were excluded, as were women with untreated preexisting hypertension and those with hypertension‐related complications in a previous pregnancy
Interventions	Comparison Intervention: lifestyle intervention consisted of a 6‐month structured programme aiming at weight loss of 5% to 10% of original body weight. It included 6 structured outpatient visits and 4 telephone consultations with a pre‐trained intervention coach. Daily dietary energy intake was reduced by 600 kcal and was maintained at a minimum of 1200 kcal/d Physical activity was stimulated to a level of 10,000 steps a day and at least 30 minutes of exercise 2 to 3 times a week. Behavioural changes were facilitated by motivational counselling. After completion of the 6‐month programme, or when weight loss of 5% to 10% had been achieved, women started to receive appropriate infertility treatment if they were not yet pregnant Control: appropriate infertility treatment immediately after randomisation.
Outcomes	Live birth Ongoing pregnancy Clinical pregnancy Miscarriage
Notes	This lifestyle project and some data are presented in the papers below Mutsaerts MA, Groen H, ter Bogt NC, Bolster JH, Land JA, Bemelmans WJ, Kuchenbecker WK, Hompes PG, Macklon NS, Stolk RP, et al. The LIFESTYLE study: costs and effects of a structured lifestyle program in overweight and obese subfertile women to reduce the need for fertility treatment and improve reproductive outcome. A randomised controlled trial. BMC Womens Health 2010;10:22‐6874‐10‐22 Mutsaerts MA, van Oers AM, Groen H, Burggraaff JM, Kuchenbecker WK, Perquin DA, Koks CA, van Golde R, Kaaijk EM, Schierbeek JM, et al. Randomized trial of a lifestyle program in obese infertile women. N Engl J Med 2016;374:1942–1953 Karsten MDA, van Oers AM, Groen H, Mutsaerts MAQ, van Poppel MNM, Geelen A, van de Beek C, Painter RC, Mol BWJ, Roseboom TJ , Hoek A; LIFEstyle study group.Determinants of successful lifestyle change during a 6‐month preconception lifestyle intervention in women with obesity and infertility. Eur J Nutr van Oers AM, Mutsaerts MAQ, Burggraaff JM, Kuchenbecker WKH, Perquin DAM, Koks CAM, van Golde R, Kaaijk EM, Schierbeek JM, Klijn NF, van Kasteren YM, Land JA, Mol BWJ, Hoek A, Groen H; LIFEstyle Study Group. Cost-effectiveness analysis of lifestyle intervention in obese infertile women. Hum Reprod 2017;32(7):1418‐1426. doi: 10.1093/humrep/dex092 van Oers AM, Mutsaerts MAQ. Association between periconceptional weight loss and maternal and neonatal outcomes in obese infertile women. PLoS One 2018;13(3):e0192670. doi: 0.1371/journal.pone.0192670 Conflicts of interest: the Department of Obstetrics and Gynaecology of the UMCG received an unrestricted educational grant from Ferring Pharmaceuticals BV, The Netherlands. BWJM is a consultant for ObsEva, Geneva Funding: Netherlands Organization for Health Research and Development Date study was conducted: 1 April 2009 Clinical trial registration number: NTR1530 Trial authors contacted: Zon MW
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was performed through a web‐based randomisation programme and was stratified according to trial centre and ovulatory status
Allocation concealment (selection bias)	Low risk	Web‐based randomisation programme
Blinding of participants and personnel (performance bias) All outcomes	High risk	Blinding to treatment assignments was not possible
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding to treatment assignments was not possible
Incomplete outcome data (attrition bias) All outcomes	Low risk	Missing outcome data were balanced in numbers across intervention groups, with similar reasons for missing data across groups We randomised 577 women, of whom 290 were randomised to the lifestyle intervention preceding infertility treatment arm (9 were lost to follow‐up and 1 withdrew informed consent) and 287 to the prompt infertility treatment arm (1 was lost to follow‐up and 2 withdrew informed consent). So, in total, data for 280 women in the intervention group and for 284 women in the control group were available for analysis
Selective reporting (reporting bias)	Low risk	The study protocol is available, and all of the study's pre‐specified outcomes have been reported in the results
Other bias	Low risk	We did not identify any other potential sources of bias in the study, and we judged low risk for other potential sources of bias

Rothberg 2016.

Study characteristics
Methods	Open‐label, single‐site, pilot study
Participants	Setting: single‐site, academic institution, Michigan Inclusion: 18 to 35 years old. Obesity (BMI ≥ 35 and ≤ 45 kg/m²). Infertility. Evidence of normal uterine anatomy and at least 1 patent tube documented by hysterosalpingogram or saline infusion sonogram. Polycystic ovary syndrome. Ovulatory dysfunction (amenorrhoea, irregular cycles, or progesterone level < 10 ng/mL in the luteal phase). Partner with semen analysis demonstrating ≥ 20 million sperm/mL, 50% motility, and normal morphology by Kruger criteria of 8% Exclusion: women who were using donor sperm, had FSH > 10 mIU/mL, or had endometriosis American Fertility Society class III or IV; were taking anti‐obesity drugs or appetite suppressants within the past 2 months; had previous bariatric surgery or gastrointestinal disease; used hormone medications within the past 2 months; had elevated prolactin, type 1 diabetes, uncorrected thyroid disease, or evidence of adrenal disease; had evidence of conditions that would complicate pregnancy (liver disease, kidney disease, autoimmune disorders such as systemic lupus erythematosus, significant anaemia, history of clotting disorder, uncontrolled hypertension, heart disease, or cancer)
Interventions	Comparison Intensive weight loss intervention (IWL): IWL consisted of 12 weeks of very low‐energy diet (800 kcal/d) and 4 weeks of a low‐calorie conventional food‐based diet (CFD) to promote 15% weight loss (n = 6) Standard of care nutrition counselling (SCN): SCN consisted of 16 weeks of CFD to promote 5% weight loss (n = 5)
Outcomes	Live birth Weight loss (BMI) Fasting glucose Confirmed pregnancy Depression (IDS‐SR) Quality of life (EQ‐5D health score) Adverse events
Notes	Thirty‐nine women were screened; 25 (64%) were eligible to participate, and 14 of those eligible (56%) agreed to be randomised, 7 to each group. One withdrew from the IWL group and 2 from the SCN group Conflicts of interest: not mentioned Funding: supported by a grant from the Michigan Institute for Clinical Research (grant U040012 PI to A.R.); core services of the Michigan Nutrition Obesity Research Center (grant DK089503); and the Michigan Center for Diabetes Research (grant P30DK020572) Date study was conducted: October 2013 Clinical trial registration number: NCT01894074 Trial authors contacted:Rothberg A
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Information insufficient to permit judgement
Allocation concealment (selection bias)	High risk	Open‐label trial, not concealed
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open‐label trial, not blinded
Blinding of outcome assessment (detection bias) All outcomes	High risk	Open‐label trial, not blinded
Incomplete outcome data (attrition bias) All outcomes	Low risk	We screened 39 women, of whom 25 were found to be eligible and 14 agreed to participate. Our inability to achieve a target sample size of 32 was due to the extremely restrictive eligibility criteria and the reluctance of eligible women to delay ovulation induction. After the baseline oral glucose tolerance test, participants were randomly allocated to treatment. Seven participants were randomised to each group. One participant withdrew from the IWL group (did not tolerate diet) after 1 week of starting the dietary intervention and 2 participants withdrew (decided not to pursue pregnancy and were dissatisfied with the randomisation arm) from the SCN group before the dietary intervention
Selective reporting (reporting bias)	Low risk	The study protocol is available, but all of the study's pre‐specified outcomes have been reported in the results https://clinicaltrials.gov/ct2/show/NCT01894074
Other bias	Low risk	We did not identify any other potential sources of bias in the study, and we judged low risk for other potential sources of bias

Salamun 2018.

Study characteristics
Methods	A prospective, randomised, open‐label study
Participants	Setting: Slovenia Inclusion: PCOS diagnosed according to revised Rotterdam criteria, body mass index (BMI) ≥ 30 kg/m², age ≤ 38, first or second IVF attempt; no severe male infertility Exclusion: type 1 or type 2 diabetes mellitus; history of carcinoma; personal or family history of MEN2; significant cardiovascular, kidney, or liver disease; use of medications other than metformin known or suspected to affect reproductive or metabolic functions or statins within 90 days before study entry; no coexisting ovarian pathology
Interventions	Comparison Metformin (MET) 1000 mg twice daily: metformin was initiated at a dose of 500 mg once per day and was increased by 500 mg every 3 days up to 1000 mg twice daily Metformin 1000 mg twice daily combined with 1.2 mg liraglutide every day subcutaneously (COMBI): in the COMBI arm, there was a run‐in period of 12 days to titrate metformin up to 1000 mg twice daily before liraglutide was added. Liraglutide was initiated at a dose of 0.6 mg injected subcutaneously once per day and increased to 1.2 mg after 3 days. Medical treatment in both groups lasted 12 weeks
Outcomes	Clinical pregnancy Weight loss (BMI) Total body fat (%) Glu OGTT Free T Total T SHBG Adverse events
Notes	Conflicts of interest: no Funding: grant number 20140031 of the University Medical Center, Ljubljana, Slovenia Date study was conducted: 1 September 2014 Clinical trial registration number: NCT03353948 Trial authors contacted: Vesna Salamun
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	This is not described in sufficient detail to allow a definitive judgement
Allocation concealment (selection bias)	High risk	Open‐label trial, not concealed
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open‐label trial, not blinded
Blinding of outcome assessment (detection bias) All outcomes	High risk	Open‐label trial, not blinded
Incomplete outcome data (attrition bias) All outcomes	Low risk	Missing outcome data were balanced in numbers across intervention groups, with similar reasons for missing data across groups 28 patients started the study, and 27 (14 on MET, 13 on COMBI) completed treatment according to the protocol and were included in the analysis (Fig. 1). 1 patient in the COMBI group discontinued the study because of protocol violation, 2 patients in the MET group refused IVF. 1 patient in the COMBI group and 1 in the MET group conceived spontaneously immediately after treatment completion. The remaining 23 patients (11 in the MET group and 12 in the COMBI group) attended an IVF after treatment completion
Selective reporting (reporting bias)	Low risk	The study protocol is available, and all of the study's pre‐specified outcomes have been reported in the results https://clinicaltrials.gov/ct2/show/NCT03353948
Other bias	Low risk	We did not identify any other potential sources of bias in the study, and we judged low risk for other potential sources of bias

Sim 2014.

Study characteristics
Methods	Evaluator‐blinded, randomised controlled trial
Participants	Setting: Australian Inclusion: obese (BMI ≥ 30 kg/m²) female patients aged 18 to 37 years, intending to commence their IVF, ICSI, or cryo‐stored embryo transfer treatment at RPAH Fertility Unit Exclusion: current psychiatric condition (i.e. bulimia nervosa, overt psychosis, severe depression, drug or alcohol abuse); significant physical condition (i.e. acute cerebrovascular or cardiovascular disease, malignancy, significant hepatic or renal dysfunction, musculoskeletal disease); endocrine condition other than polycystic ovarian syndrome (PCOS) (i.e. type 1 diabetes, uncontrolled thyroid disease, Cushing’s syndrome, hyperprolactinaemia (> 450 IU L –1)); pancreatitis; porphyria; recent (within 3 months) participation in treatment known to affect diet or body weight; unable to follow both verbal and written English instructions; unwilling to suspend fertility treatment for up to 3 months
Interventions	Comparison Intervention (n = 27): 12‐week intervention consisting of a very low‐energy diet for the first 6 weeks followed by a hypocaloric diet, combined with a weekly group multi‐disciplinary programme Control (n = 22): received recommendations for weight loss and the same printed material as the intervention group
Outcomes	Clinical pregnancy rate Live birth Changes in anthropometric measures (weight, BMI, and waist circumference (WC)) Miscarriage rate
Notes	Conflicts of interest: no Funding: not mentioned. Date study was conducted: February 2007 Clinical trial registration number: 12606000448549 Trial authors contacted: KA Sim
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The code was prepared before study start, as the allocation was performed using sealed envelopes
Allocation concealment (selection bias)	Low risk	Randomisation was done by the sequentially numbered, opaque‐sealed envelope method
Blinding of participants and personnel (performance bias) All outcomes	High risk	The dietician, midwives, counsellor, fertility fellow, and participants were aware of randomisation, but fertility specialists were not
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	The fertility fellow who was aware of randomisation was not involved with cycle management and did not perform any assisted conception procedures for these patients. The dietician, midwives, counsellor, fertility fellow, and participants were aware of randomisation, but fertility specialists and patients were not. This may have affected detection bias
Incomplete outcome data (attrition bias) All outcomes	High risk	Very small study (26 versus 22). Missing outcome data problematic, especially for weight loss data as a 20% loss can have a lot of impact on the outcomes.  An estimated 86 patients were identified in consultation at the fertility unit and were approached by the research leader for discussion of participation in the study. Forty‐nine were screened for eligibility, and 37 chose not to participate. The physicians identifying patients referred only 86 for an initial discussion; it was estimated that more were eligible. All of the 49 participants who entered the trial completed a baseline assessment. 6 separate weight loss intervention programmes were run, with groups of participants varying in number between 3 and 8. At 12 weeks, 10 (20%) participants had withdrawn or dropped out of the trial. No participants reported any serious adverse events during the dietary intervention. There were no statistically significant differences between completers and dropouts in both groups in terms of pregnancy rates, fertility treatment outcomes, or maternal and fetal complications (results not shown)
Selective reporting (reporting bias)	Low risk	The study protocol is available, and all of the study's pre‐specified outcomes have been reported in the results https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=81639
Other bias	Low risk	We did not identify any other potential sources of bias in the study, and we judged low risk for other potential sources of bias

Tang 2006.

Study characteristics
Methods	A randomised, placebo‐controlled, double‐blind study
Participants	Setting: England Inclusion: anovulatory PCOS and BMI ≥ 30 kg/m², between 18 and 39 years of age inclusive, a desire to conceive, presence of ≥ 1 patent fallopian tube and normal semen analysis from male partner. All participants had normal serum prolactin concentrations; thyroid, renal, and liver function; and haematological indices, including serum B12 concentrations Exclusion: concurrent hormone therapy within the previous 6 weeks; any chronic disease that could interfere with absorption, distribution, metabolism, or excretion of metformin; renal or liver disease. Patients with significant systemic disease or diabetes (type 1 or 2) were excluded. Patients with irregular menstrual bleeding were thoroughly assessed to exclude pathology of the genital tract other than PCOS, and a negative pregnancy test was a prerequisite for commencing treatment
Interventions	Comparison Intervention: metformin (850 mg) twice daily over 6 months (n = 69) Control: placebo over 6 months (n = 74) All received the same advice from a dietician
Outcomes	Change in anthropometric measurements (BMI, WHR) Testosterone SHBG Pregnancy rate
Notes	Randomised: 143 patients: 69 metformin (56 completed) and 74 placebo (66 completed) Conflicts of interest: not mentioned Funding: not mentioned Date study was conducted: 1999 Clinical trial registration number: not found Trial authors contacted: Adam H. Balen
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A block‐of‐four randomisation technique was performed using random tables
Allocation concealment (selection bias)	Low risk	Double‐blind, placebo tablets for metformin were identical in appearance (size and colour) to metformin.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	The randomisation process was carried out by the clinical trials office in the pharmacy department and blinded to patients and investigators.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	The randomisation process was carried out by the clinical trials office in the pharmacy department and blinded to patients and investigators.
Incomplete outcome data (attrition bias) All outcomes	Low risk	A total of 8 centres took part in the recruitment process. A total of 183 women were screened for inclusion in the study. Of these, 40 women were excluded due to previously undiagnosed tubal disease or coexisting male factor infertility. As a result, a total of 143 subjects were randomised to receive metformin (n = 69) or to receive placebo (n = 74) (Fig. 1). In the metformin arm, 13 participants withdrew within the first 4 months of the trial (11 due to side effects and 2 due to spontaneous pregnancies). 8 women withdrew from the placebo arm (6 due to ‘side effects’ and 2 due to spontaneous pregnancies) within the first 2 months of the study. The difference in dropout rates, excluding those due to pregnancy (metformin 15.9% versus placebo 8.0%) was not significant (P = 0.229; 95% CI –2.69 to 18.5). At the end of the study, the numbers of participants who completed the trial in the metformin and placebo arms were 56 and 66, respectively. Compliance was high and the dropout rate relatively low, as these patients were motivated by a desire to conceive and the knowledge that they needed to attain BMI < 30 kg/m² to qualify for ovulation induction
Selective reporting (reporting bias)	High risk	The study protocol is unavailable. It is not described in sufficient detail to allow a definitive judgement
Other bias	Low risk	We did not identify any other potential sources of bias in the study, and we judged low risk for other potential sources of bias

BMI: body mass index.

CC: clomiphene citrate.

CFD: conventional food‐based diet.

EQ‐5D: EuroQoL Group Quality of Life Questionnaire based on 5 dimensions.

FSH: follicle‐stimulating hormone.

ICSI: intracytoplasmic sperm injection.

IDS‐SR: Inventory of Depressive Symptomatology‐Self‐Report.

IVF: in vitro fertilisation.

IWL: intensive weight loss intervention.

MEN2: multiple endocrine neoplasia type 2.

MET: metformin.

OGTT: oral glucose tolerance test.

PCOS: polycystic ovarian syndrome.

RCT: randomised controlled trial.

SCN: standard of care nutrition counselling.

SD: standard deviation.

SHBG: sex hormone‐blinding globulin.

WC: waist circumference.

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Asemi 2014	Wrong patient population
Asemi 2014a	Wrong patient population
Asemi 2015	Wrong patient population
Ashoush 2016	Wrong patient population
Attarzadeh 2012	Wrong patient population
Baillargeon 2004	Wrong patient population
Becker 2015	Wrong patient population
Cheraghi 2014	Wrong patient population
Gambineri 2004	Wrong patient population
Hanjalic‐Beck 2010	Wrong patient population
Janez 2017	Wrong patient population
Janez 2018	Wrong patient population
Jensterle 2017	Wrong patient population
Karsten 2018	Wrong outcome
Kocak 2002	Wrong patient population
Legro 2007	Wrong patient population
Liao 2011	Wrong patient population
Ma 2007	Wrong patient population
Moran 2006	Wrong patient population
Morin‐Papunen 1998	Wrong patient population
Morin‐Papunen 2000	Wrong patient population
Morin‐Papunen 2012	Wrong patient population
Pasquali 1986	Wrong patient population
Pasquali 2000	Wrong patient population
Pastore 2011	Wrong patient population
Penna 2005	Wrong patient population
Petranyi 2011	Wrong patient population
Pourmatroud 2015	Wrong patient population
Premawardhana 1994	Wrong patient population
Qin 2016	Wrong patient population
Qublan 2007	Wrong patient population
Seibel 2008	Wrong patient population
Shahebrahimi 2016	Wrong patient population
Siebert 2009	Wrong patient population
Sonmez 2005	Wrong patient population
Sordia‐Hernandez 2016	Wrong patient population
Sorensen 2012	Wrong patient population
Stamets 2004	Wrong patient population
Swora‐Cwynar 2016	Wrong patient population
Tang 2006a	Wrong patient population
Toscani 2011	Wrong patient population
Tsagareli 2006	Wrong patient population
Turner‐McGrievy 2014	Wrong patient population
Vandermolen 2001	Wrong patient population
Vanky 2004	Wrong patient population
Van Oers 2016a	Wrong outcome
Van Oers 2017	Wrong outcome
van Oers 2018	Wrong outcome
van Santbrink 2005	Wrong patient population
Vigerust 2012	Wrong patient population
Vosnakis 2013	Wrong patient population
Wiweko 2017	Wrong patient population
Yang 2005	Wrong patient population
Yang 2017	Wrong patient population
Yin 2018	Wrong patient population
Zhang 2015	Wrong patient population
Zhang 2017	Wrong study design
Zheng 2013	Wrong patient population

Characteristics of studies awaiting classification [ordered by study ID]

Duval 2015.

Methods	RCT
Participants	Obese infertile women
Interventions	Comparison Intervention: lifestyle intervention without fertility treatment for the first 6 months Control: standard fertility treatment Participants were followed for 18 months or until the end of pregnancy
Outcomes	Pregnancy rate Spontaneous pregnancy rate Live birth Weight loss
Notes	This study did not report the number of participants

Egbase 2001.

Methods	RCT
Participants	68 obese infertile PCOS patients
Interventions	Group 1: patients received metformin (500 mg tds) from day 2 of same index treatment menstrual cycle and continued until the day of administration of trigger dose of HCG Group 2: control (no drug)
Outcomes	Serum levels of T DHAS SHBG FBS/FBI ratio Fertilization and cleavage rates Clinical pregnancy rate Implantation rate Miscarriage rate
Notes	Metformin reduced insulin resistance in obese PCOS patients with a statistically significant positive effect on folliculogenesis (follicular synchrony) and embryo quality (good quality embryos and implantation rates)

Li 2015.

Methods	RCT
Participants	Obese women with infertility with polycystic ovary syndrome
Interventions	Control group: metformin Observation group: auxiliary acupuncture 75 cases in each group
Outcomes	Cycle ovulation rate Pregnancy rate
Notes	Cycle ovulation rate and pregnancy rate of the observation (auxiliary acupuncture) group were significantly better than those of the control (metformin) group

Pfüller 2004.

Methods	RCT
Participants	46 infertile women with PCOS, between 22 and 39 years of age; mean BMI 38.1, range 28.1 to 49.0 kg/m²
Interventions	Group 1: M and lifestyle modification Group 2: Lifestyle modification only; patients received placebo
Outcomes	In the metformin group, mean BMI decreased significantly (‐5.6 kg vs.‐2.2 kg). Neither metformin nor placebo modified levels of testosterone (T), free T, FAI (free androgen index), LH/FSH ratio. After treatment, SHBG concentrations were significantly increased in the metformin‐taking group (+7.5 vs. +3.2). Cholesterol, LDL, and triglycerides have not revealed any changes in either group. Serum HDL levels were significantly higher in the drug‐treated group (+3.7 vs –1.7)
Notes

BMI: body mass index.

DHAS: dehydroepiandrosterone sulphate.

FBS/FBI: fasting blood sugar/fasting blood insulin.

FSH: follicle‐stimulating hormone.

HCG: human chorionic gonadotropin.

HDL: high‐density lipoprotein.

LDL: low‐density lipoprotein.

LH: luteinising hormone.

PCOS: polycystic ovarian syndrome.

RCT: randomised controlled trial.

SHBG: sex hormone‐binding globulin.

Differences between protocol and review

The review objective was amended to improve clarity that the interventions are targeted to weight reduction in obese women with subfertility. The inclusion of cluster‐randomised trials was excluded from the review as they are not an appropriate study design. Lastly, the measure of treatment effect for dichotomous data was calculated using odds ratios (OR) rather than risk ratios (RR). We chose the OR while this is considered a mathematically more stable effect measure .

Contributions of authors

FB and SAT were involved in the design and conduct of the review, data analysis, drafting of the manuscript, and critical discussion. SHJ and MvW were involved in the design and conduct of the review, supervised data analysis, checked data extraction, and were involved in manuscript revision and critical discussion. All review authors read and approved the final manuscript.

Sources of support

Internal sources

• None, Other

External sources

• None, Other

Declarations of interest

None of the review authors have or have had affiliations or involvement in organisations with an interest in the review findings. MvW is a Cochrane Editor and Deputy Editor of Human Reproduction.

Edited (no change to conclusions)