Acupuncture for symptomatic gastroparesis

Abstract

Background

Gastroparesis, a state of delayed gastric emptying in the absence of mechanical obstruction of the stomach, has a substantial impact on people's daily function and quality of life when symptomatic. Current treatment options are based on limited evidence of benefits. Acupuncture is widely used to manage gastrointestinal disorders, although its role in people with symptomatic gastroparesis is unclear. We therefore undertook a systematic review of the evidence.

Objectives

To assess the benefits and harms of acupuncture, in comparison with no treatment, sham acupuncture, conventional medicine, standard care, or other non‐pharmacological active interventions for symptom management in people with gastroparesis.

Search methods

On 26 March 2018, we searched the Cochrane Neuromuscular Specialised Register, CENTRAL, MEDLINE, Embase, CINAHL Plus, PsycINFO, AMED, Korean medical databases (including Korean Studies Information, DBPIA, Korea Institute of Science and Technology Information, Research Information Centre for Health Database, KoreaMed, and the National Assembly Library), and Chinese databases (including the China Academic Journal). We also searched two clinical trials registries for ongoing trials. We imposed no language limitations.

Selection criteria

We selected all randomised controlled trials comparing the penetrating type of acupuncture with no treatment, sham acupuncture, conventional medicine, standard care, and other non‐pharmacological active interventions for people with symptomatic gastroparesis of any aetiology (i.e. surgical, diabetic, or idiopathic). Trials reporting outcomes at least four weeks from baseline (short‐term outcomes) were eligible. We defined long‐term outcomes as those measured after 12 weeks from baseline. The primary outcome was improvement of gastroparesis symptoms in the short term. Secondary outcomes were: improvement of symptoms measured after three months, change in the rate of gastric emptying, quality of life, use of medication, and adverse events in the short and long term.

Data collection and analysis

Two review authors independently selected eligible trials based on predefined selection criteria. Two review authors independently extracted data and evaluated the risk of bias. The review authors contacted investigators to obtain missing information wherever possible.

Main results

We included 32 studies that involved a total of 2601 participants. Acupuncture was either manually stimulated (24 studies) or electrically stimulated (8 studies). The aetiology of gastroparesis was diabetes (31 studies) or surgery (1 study). All studies provided data on the proportion of people with symptoms 'improved', although the definition or categorisation of improvement varied among the studies. Most measured only short‐term outcomes (28 studies), and only one study employed validated instruments to assess subjective changes in symptoms or reported data on quality of life or the use of medication. Reporting of harm was incomplete; minor adverse events were reported in only seven trials. Most studies had unclear risk of bias in terms of allocation concealment (29/32), outcome assessor blinding (31/32) and selective reporting (31/32), as well as high risk of bias in terms of participant/personnel blinding (31/32). Acupuncture was compared with sham acupuncture (needling on non‐acupuncture points), three different types of gastrokinetic drugs (domperidone, mosapride, cisapride), and a histamine H₂ receptor antagonist (cimetidine).

There was low‐certainty evidence that symptom scores of participants receiving acupuncture did not differ from those of participants receiving sham acupuncture at three months when measured by a validated scale.

There was very low‐certainty evidence that a greater proportion of participants receiving acupuncture had 'improved' symptoms in the short term compared to participants who received gastrokinetic medication (4 to 12 weeks) (12 studies; 963 participants; risk ratio (RR) 1.25; 95% confidence interval (CI) 1.17 to 1.33, I² = 8%). Short‐term improvement in overall symptom scores favouring acupuncture was also reported in five studies with considerable heterogeneity.

Acupuncture in combination with other treatments, including gastrokinetics, non‐gastrokinetics and routine care, was compared with the same treatment alone. There was very low‐certainty evidence in favour of acupuncture for the proportion of participants with 'improved' symptoms in the short term (4 to 12 weeks) (17 studies; 1404 participants; RR 1.22; 95% CI 1.16 to 1.28; I² = 0%). Short‐term improvement in overall symptom scores, favouring acupuncture, were also reported (two studies, 132 participants; MD ‐1.96, 95% CI ‐2.42 to ‐1.50; I² = 0%).

Seven studies described adverse events, including minor bleeding and hematoma, dizziness, xerostomia, loose stool, diarrhoea, abdominal pain, skin rash and fatigue. The rest of the trials did not report whether adverse events occurred.

Subgroup analyses revealed that short‐term benefits in terms of the proportion of people with 'improved' symptoms did not differ according to the type of acupuncture stimulation (i.e. manual or electrical). The sensitivity analysis revealed that use of a valid method of random sequence generation, and the use of objective measurements of gastric emptying, did not alter the overall effect estimate in terms of the proportion of people with 'improved' symptoms. The asymmetric funnel plot suggests small study effects and publication bias towards positive reporting.

Authors' conclusions

There is very low‐certainty evidence for a short‐term benefit with acupuncture alone or acupuncture combined with gastrokinetic drugs compared with the drug alone, in terms of the proportion of people who experienced improvement in diabetic gastroparesis. There is evidence of publication bias and a positive bias of small study effects. The reported benefits should be interpreted with great caution because of the unclear overall risk of bias, unvalidated measurements of change in subjective symptoms, publication bias and small study reporting bias, and lack of data on long‐term outcomes; the effects reported in this review may therefore differ significantly from the true effect. One sham‐controlled trial provided low‐certainty evidence of no difference between real and sham acupuncture in terms of short‐term symptom improvement in diabetic gastroparesis, when measured by a validated scale. No studies reported changes in quality of life or the use of medication.

Due to the absence of data, no conclusion can be made regarding effects of acupuncture on gastroparesis of other aetiologies. Reports of harm have remained largely incomplete, precluding assessments of the safety of acupuncture in this population. Future research should focus on reducing the sources of bias in the trial design as well as transparent reporting. Harms of interventions should be explicitly reported.

Plain language summary

Acupuncture for delayed emptying of the stomach

Review question

Is acupuncture beneficial for symptoms of delayed emptying of the stomach?

Background

Gastroparesis is a condition where the stomach empties more slowly than normal, in the absence of a blockage. Causes include damage to the nerves to the gut (autonomic failure, most commonly as a result of diabetes), viral infections, and complications of surgery. Symptoms include feeling full very soon after starting to eat, feeling uncomfortably full after eating, nausea, regurgitation, sickness, and bloating. Many people have no symptoms, but this Cochrane Review focuses on people with symptoms.

Study characteristics

We analysed 32 studies that involved a total of 2601 participants. Most trials involved people with diabetic gastroparesis, who received short‐term treatment (often four weeks). Non‐profit funding bodies (the Chinese government and a university) funded six out of the 32 studies and the others did not report the funding source. One study compared real acupuncture with sham acupuncture (needling on non‐acupuncture points). Twenty‐eight trials compared acupuncture to a drug, or acupuncture with a drug or to the drug alone. A small number of trials compared acupuncture plus a non‐drug treatment to the same treatment alone. The drugs in the trials were mostly gastrokinetic agents (such as domperidone, mosapride, and cisapride), which promote stomach emptying.

Key results

Despite there being a small reported effect, we are uncertain about any benefit of acupuncture for symptomatic gastroparesis in the short term, when used alone or added to other treatments for gastroparesis (gastrokinetics, other medicines, or 'routine care'), due to the evidence being of very low certainty. There is no information to help understand any long‐term benefits of acupuncture. The effects of acupuncture on symptoms of gastroparesis are probably little different from those of sham acupuncture in the short term. It is unclear whether acupuncture helps in gastroparesis after surgery or when the cause of gastroparesis is unknown, because there is not enough information. No trial studied the effects of acupuncture on quality of life or delayed emptying of the stomach. We do not know whether acupuncture is safe for people with diabetes who have delayed emptying of the stomach, because safety was incompletely reported in most trials.

Certainty of the evidence

Overall, the certainty of evidence is very low. Most studies had design issues. We suspected the existence of unpublished studies and it was not possible to be sure whether those we identified fully reported their findings. There was no consistent definition of improvement across studies. Any reported benefit may not be accurate and should be interpreted with caution. Future trials should focus on valid measures of treatment effects reported directly by patients, and assessment of stomach emptying. Trials should meet quality standards for design and transparent reporting.

The evidence is current to March 2018.

Background

Description of the condition

Gastroparesis is a state of delayed gastric emptying in the absence of mechanical obstruction of the stomach (Camilleri 2007). It is frequently, but not necessarily, accompanied by debilitating symptoms including fullness after eating, early satiety, nausea, vomiting and bloating. Several aetiologies are associated with gastroparesis, such as type I or II diabetes, postviral infection or postsurgical complications (Waseem 2009). The most common identifiable cause is diabetes mellitus (Camilleri 2008; Jung 2009). Diabetic gastroparesis often presents after at least a 10‐year history of diabetes (Camilleri 2007). The prevalence of upper gastrointestinal symptoms associated with gastroparesis in people with diabetes is related to the degree of acute hyperglycaemia and chronic glycaemic control (Khoo 2009).

Hospitalisations with gastroparesis have doubled between 1995 and 2004 in the United States, which suggests an increasing prevalence (Wang 2008a). However, the poor correlation between gastric emptying and upper gastrointestinal symptoms hinders exact estimation of prevalence (Jung 2009). One population‐based study reported that the age‐adjusted prevalence of gastroparesis, regardless of aetiology, was 24.2% when diagnosed using both cardinal symptoms of gastroparesis and scintigraphy‐confirmed delayed gastric emptying; whereas it was 50.5% when diagnosed by either method alone (Jung 2009). In other population‐based studies, 11% to 18% of people with diabetes reported symptoms consistent with upper gastrointestinal dysmotility, whereas the prevalence of gastroparesis assessed by gastric emptying studies in people with diabetes at an academic medical centre was 48% (Parkman 2004).

Although the prevalence of symptomatic gastroparesis is lower than that of asymptomatic disease (Parkman 2004), living with the symptoms can be challenging. In a multicentre trial of domperidone (n = 269), people with diabetes and symptoms suggestive of gastroparesis had a significantly reduced quality of life compared to both the general population and people with type II diabetes (Farub 1998). Of people with symptomatic idiopathic gastroparesis with delayed gastric emptying on scintigraphy (n = 243), 36% had severe anxiety and 18% had severe depression (Parkman 2011). Higher depression and anxiety levels were also associated with worse symptoms of gastroparesis (Hasler 2010). People with gastroparesis reported difficulties when eating out, in relationships with partners or family, in professional activities and social functions because of their symptoms (Bielefeldt 2009).

The impact of gastroparesis on mortality has been regarded as minimal (Wang 2008a). However, one population‐based study in the United States (n = 3604) showed that, regardless of the underlying condition, people with gastroparesis showed a significantly decreased five‐year survival rate compared to the regional control population (67% versus 81%) (Jung 2009). This study also revealed longer survival in people with idiopathic compared to non‐idiopathic gastroparesis and in nondiabetic compared to diabetic gastroparesis (Jung 2009). Whether the mortality risk of gastroparesis arises from gastroparesis itself or the associated diabetes remains uncertain.

The healthcare costs for the management of gastroparesis are significant. This is possibly due to the need for hospitalisation and diagnostic testing such as endoscopy and gastric emptying scintigraphy (Parkman 2004). In one study, gastroparesis had the longest hospitalisation among upper gastrointestinal conditions (Wang 2008a). The cause of the increasing number of gastroparesis‐related hospitalisations is not clear, but could be associated with the increasing prevalence of gastroparesis and diabetes; changes in the diagnostic criteria, severity and treatment of gastroparesis; or better recognition and diagnosis (Wang 2008a).

Possible pathophysiological mechanisms of delayed gastric emptying include autonomic neuropathy, enteric neuropathy, loss of Interstitial cells of Cajal (ICC, the pacemaker cells for initiating slow waves in the stomach) and acute and chronic hyperglycaemia (Waseem 2009). Cellular defects of ICC, abnormal immune infiltrates and impaired nerve fibres were found in full‐thickness gastric biopsy specimens of 83% of people with gastroparesis whether diabetic or idiopathic (n = 40) (Grover 2011). It is well established that acute hyperglycaemia can produce abnormal gastric emptying and alter the effect of gastrokinetic agents, whereas the effect of chronic hyperglycaemia is less clear (Kashyap 2010). Postsurgical gastroparesis can be due to damage to the vagus nerve following upper abdominal operations. This leads to reduced antral contractions and loss of pyloric relaxation (Tang 2011).

Current management of gastroparesis includes removal of exacerbating factors (for example drugs that can cause gastric dysmotility), dietary modification and nutritional support, normalisation of glycaemic control, pharmacological interventions (gastrokinetic and antiemetic agents), non‐pharmacological interventions (endoscopic injection of botulinum toxin) and surgery (implantation of devices for gastric electrical stimulation and for palliative purposes) (Camilleri 2007). However, none of these interventions is generally accepted as the gold standard (Ma 2009). Systematic reviews concluded that current evidence on oral erythromycin (Maganti 2003), domperidone (Sugumar 2008), gastric electrical stimulation (O'Grady 2009), and intrapyloric botulinum toxin injection (Bai 2010), is insufficient to support their use for people with gastroparesis. There is no Cochrane Review on this subject and only two published protocols which plan to investigate interventions for symptomatic gastroparesis (Camilleri 2008; Wijeratne 2016). Given the negative impact of gastroparesis on quality of life, further research is needed to improve current treatment strategies and build reliable evidence (Bielefeldt 2009). 

Description of the intervention

Acupuncture is a widely used non‐pharmacological intervention in contemporary East Asian medicine and complementary medicine that classically involves the insertion of fine needles into certain locations that are considered emerging points of vital energy (the 'qi' of traditional Asian medicine). The available points vary from classical acupuncture points and myofascial trigger points to non‐classical extra points and non‐predefined new points, reflecting the diverse clinical backgrounds of practitioners.

Acupuncture is widely used for a variety of illnesses, including chronic pain, stroke rehabilitation, and cancer care. Its efficacy, using sham acupuncture as a comparator, has been established under limited conditions including the prevention of postoperative nausea and vomiting, tension‐type headache, episodic migraine, and peripheral joint arthritis (Linde 2016; Linde 2016a; Manheimer 2010). Based on the results of randomised, sham‐controlled trials, there appear to be few specific effects of acupuncture needling and stimulation compared with sham acupuncture. However, the validity of sham acupuncture as a placebo for superficial needling or even non‐penetrating needling has been questioned because simple sensory stimulation from the proposed sham needling procedures can also provoke neurophysiological responses and may produce unexpected therapeutic effects (Lunderberg 2011). The effects of sham acupuncture may exceed those of inert placebo interventions used in most pharmaceutical trials (Kaptchuk 2006; Linde 2010). In Charlotte 2005, it was suggested that the use of placebo or sham‐controlled trial designs may not be appropriate when evaluating complex interventions such as acupuncture, as those designs often oversimplify important characteristic elements such as therapeutic interactions between the patient and practitioner, which may lead to false‐negative results. Given these issues, interpretation of the current evidence for the efficacy of acupuncture is not straightforward and remains controversial (Linde 2010).

How the intervention might work

Experimental studies have suggested that acupuncture may involve excitation of the vagus nerve (Noguchi 2010), serotonergic pathways (Chen 2008; Sugai 2004), opioidergic pathways (Yin 2010), and spinal or supraspinal reflexes (Noguchi 2010), all or any of which may modify gastric motility.

Acupuncture has been observed to affect gastric hormones in people with gastroparesis in one study (Chang 2001), but not in another (Wang 2008b). In Yang 2009, researchers reported increased activation of ICCs in the stomachs of rats after applying electroacupuncture at the ST36 acupuncture point. Clinical evidence of acupuncture having antiemetic effects in the postoperative population, and other symptom‐relieving effects in gastrointestinal disorders, is available in Cochrane Reviews (Lan 2014; Lee 2015; Manheimer 2012), however any neurophysiological features of those effects remain unclear (Cheng 2009).

Why it is important to do this review

The use of complementary medicine in people with gastrointestinal disorders is common (Hung 2015; Dossett 2014). Acupuncture is known as a treatment modality that has been most studied with respect to its effect on visceral hypersensitivity, gastric accommodation, and gastric emptying in complementary medicine (Lee 2015a). In contrast to the potential benefits of acupuncture suggested in experimental studies, previous systematic reviews of acupuncture for diabetic or postoperative gastroparesis reached no definitive conclusions because of a high or an unclear risk of bias (Cheong 2014; Tang 2010; Yang 2013a). The limitations of previous reviews are related to the primary focus on the specific aetiology (i.e. diabetic or postoperative) and the immediate outcomes that were observed after less than four weeks of treatment, which may have limited clinical value in people whose disease burden derives mainly from living with chronic conditions (Woodhouse 2017). The aim of this Cochrane Review was to comprehensively assess the evidence of acupuncture for symptomatic gastroparesis of any aetiology and focus on long‐term outcomes, thus providing more relevant information on the symptom management of this chronic condition.

Objectives

To assess the benefits and harms of acupuncture, in comparison with no treatment, sham acupuncture, conventional medicine, standard care, or other non‐pharmacological active interventions for symptom management in people with gastroparesis.

Methods

Criteria for considering studies for this review

Types of studies

We included all randomised controlled trials (RCTs) testing acupuncture for gastroparesis of any aetiology. There was no language restriction. We excluded quasi‐randomised, non‐randomised and uncontrolled studies. We would have used data from the first phase of randomised cross‐over trials to avoid carry‐over effects (Manheimer 2010). However, we did not identify any eligible cross‐over trials.

Types of participants

Trials involving participants with symptomatic gastroparesis, regardless of aetiology, were eligible. We defined symptomatic gastroparesis as having at least one cardinal symptom (i.e. early satiety, nausea, vomiting, bloating and upper abdominal discomfort), as well as delayed gastric emptying, measured by gastric emptying tests (i.e. gastric scintigraphy, radiopaque markers, ultrasonography, 13‐C labelled octanoate breath tests, magnetic resonance imaging, single‐photon emission computed tomography and other objective tests) (Parkman 2004; Waseem 2009). Although we planned to consider both subjective symptoms and objective evidence of delayed gastric emptying as criteria for symptomatic gastroparesis, we considered it likely that some studies would be conducted in countries where objective measurements of gastric emptying are restricted, due to limited resources. Studies without explicit diagnosis of gastroparesis (objective measurement of gastric emptying) were also included in our review.

We included studies involving people with diabetic gastroparesis if the diagnostic criteria for diabetes mellitus in the study corresponded to the World Health Organization (WHO) recommendations (i.e. fasting plasma glucose equal to or more than 7.0 millimoles per litre (mmol/L) or two‐hour postprandial plasma glucose equal to or more than 11.1 mmol/L) (WHO 2006).

Types of interventions

We evaluated studies of acupuncture that involve the penetration of skin or muscle by needles, regardless of the stimulation method. We excluded other techniques if used in the absence of needling (for example, acupressure, laser acupuncture, transcutaneous electrical stimulation (TENS) and moxibustion (heat stimulation on acupoints by heating the mugwort herb)), since inserting needles is the most common form of acupuncture practice. Studies employing needle acupuncture as a primary intervention, but also providing non‐needling acupuncture‐related techniques (e.g. acupressure or moxibustion) as minor ancillary treatments were included in the analysis, since the use of acupuncture‐related techniques to supplement the effects of needle acupuncture is common in clinical practice. In our review, we defined minor ancillary treatment as techniques provided on the needled point that enhance the stimulation of acupuncture points, or as those using a subset of acupuncture‐related techniques (e.g. moxibustion or warming needles) on less than 25% of the total acupuncture points used in the study. The minimum number of sessions should be at least six within 12 weeks, based on empirical evidence in the field of chronic pain (Ezzo 2000).

We compared the effects of acupuncture with those of no treatment, sham acupuncture, conventional medicine, standard care or other non‐pharmacological active interventions. We analysed these controls separately. We excluded studies comparing one type of acupuncture to another. We excluded studies in which cointerventions were not administered equally to all participants.

Types of outcome measures

We analysed outcomes that were evaluated at least four weeks after the start of treatment. For the meta‐analysis, we defined short‐term outcomes as those with a measurement point of at least four weeks but no more than three months, and longer‐term outcomes as those with a measurement point of greater than three months after the start of treatment. We dichotomised the categorical outcomes into 'improved' or 'not improved' where appropriate, to enable the estimation of relative risk of the defined outcomes. For instance, the participants' change after allocated intervention graded as 'recovery', 'markedly effective', 'effective' and 'ineffective' was dichotomised into 'improved' or 'not improved' by combining full or partial recovery of participants into 'improved', and the rest (i.e. 'ineffective') into 'not improved'.

Primary outcomes

Improvement in symptoms of gastroparesis, including nausea, vomiting, early satiety, abdominal pain or distension on validated scales for measuring upper gastrointestinal symptoms (for example, the Gastroparesis Cardinal Symptom Index (GCSI) (Revicki 2004) or Patient Assessment of Gastrointestinal Disorders Symptom Severity Index (PAGI‐SYM) (Rentz 2004)) at 4 to 12 weeks ('short term').

Secondary outcomes

• Improvement in symptoms of gastroparesis, including nausea, vomiting, early satiety, abdominal pain or distension on validated scales for measuring upper gastrointestinal symptoms (for example, the GCSI (Revicki 2004), or PAGI‐SYM (Rentz 2004)) at over 12 weeks ('long term').

• Proportion of people whose symptoms 'improved' in the short and long term, presented as risk ratios (RRs).

• Changes of medication for gastroparesis (for example, antiemetics, gastrokinetics or other medication for gastroparesis‐related symptoms) in the short and long term.

• Changes in quality of life (for example, general quality of life measured by the Short Form 36 Health Survey (SF‐36), and disease‐specific quality of life measured by participant assessment using the Patient Assessment of Upper Gastrointestinal Disorders‐Quality of Life (PAGI‐QOL) (de la Loge 2004)) in the short and long term.

• Changes in the rate of gastric emptying (measured by gastric scintigraphy, radio‐opaque markers, ultrasonography, 13‐C labelled octanoate breath test, electrogastrography and other objective tests) in the short and long term.

• Effects on glycaemic control (measured by, for example, level of haemoglobin A1C, fasting or postprandial blood glucose or oral glucose tolerance) in the short and long term.

• Adverse events.

We would have reported indirect outcomes such as depression and anxiety in the discussion, if reported, but there were no data on these in the included studies.

Search methods for identification of studies

Electronic searches

We searched the following databases.

• Cochrane Neuromuscular Specialised Register via the Cochrane Register of Studies (CRS‐Web) (26 March 2018) (Appendix 1)

• Cochrane Central Register of Controlled Trials (CENTRAL) via the Cochrane Register of Studies (CRS‐Web) (26 March 2018) (Appendix 2)

• MEDLINE (1946 to 26 March 2018) (Appendix 3)

• Embase (1980 to 26 March 2018) (Appendix 4)

• CINAHL Plus (1937 to 26 March 2018) (Appendix 5)

• PsycINFO (1806 to 26 March 2018) (Appendix 6)

• AMED (1985 to 26 March 2018) (Appendix 7)

• Korean medical databases (including Korean Studies Information, DBPIA; Korea Institute of Science and Technology Information; Research Information Centre for Health Database; KoreaMed and the National Assembly Library) (from the inception to 26 March 2018); and

• Chinese databases (including the China Academic Journal with China Doctoral and Master Dissertation) (1994 to 26 March 2018)

We searched the following trial registries for ongoing trials to 26 March 2018 (Appendix 8).

• World Health Organization International Clinical Trials Registry Platform (WHO ICTRP; apps.who.int/trialsearch)

• US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov)

A review author (TYC), who has used the Chinese language for academic purposes, searched the Chinese databases. We imposed no language limitations.

We used the following terms: acupuncture, acupoint, meridian, gastroparesis, gastropathy, gastric emptying, diabetes mellitus, diabetic.

Searching other resources

We reconsidered all RCTs included in previous systematic reviews of acupuncture for gastroparesis for inclusion in this review (Cheong 2014; He 2015; Shi 2012a; Tang 2010; Yang 2013; Wang 2014; Xu 2016; Zhang 2010; Zhang 2012; Zhang 2016). We checked all references in the identified trials and contacted authors, where contact details were provided, to identify any additional published or unpublished data.

Data collection and analysis

Selection of studies

Two authors (KHK and TYC) independently selected eligible trials based on the predefined selection criteria. They resolved disagreement through discussion with the other authors (THK and MSL), if necessary.

Data extraction and management

The first review author (KHK) extracted data and another review author (TYC) fluent in Chinese checked these for accuracy.

When data or information needed for the review were missing from published reports, a review author (TYC) contacted the corresponding author to obtain further information, where contact details were provided. When there were multiple reports of one study, we attempted to include all relevant outcomes without duplicating outcome data.

Assessment of risk of bias in included studies

Two review authors (KHK and TYC) independently assessed the risk of bias in the included studies. We evaluated the following items using the 'Risk of bias' tool in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We assessed the following domains.

• Random sequence generation (selection bias)

• Allocation concealment (selection bias)

• Blinding of participants and personnel (performance bias)

• Blinding of outcome assessment (detection bias)

• Incomplete outcome data (attrition bias)

• Selective outcome reporting (reporting bias)

• Other bias

We planned to rate the risk of bias for participant blinding as low only when results of the sham credibility test demonstrated successful participant blinding in the study; otherwise, we rated the risk of performance bias as unclear or high. We rated the risk of bias for selective outcome reporting as low only when the protocol of the study was available (for example, trial register information or by author contact), to find out whether authors reported all the planned and measured outcomes in their report. Otherwise, we rated the risk of bias as unclear or high.

Measures of treatment effect

We conducted statistical analysis using Review Manager (RevMan) 5, Cochrane's statistical software (RevMan 2014). For dichotomous outcomes (e.g. improved or not improved), we presented the results as RRs with 95% confidence intervals (CIs). When data were continuous, we used the mean difference (MD) and 95% CI, or the standardised mean difference (SMD) if different scales had been employed.

Unit of analysis issues

In the case of multiple intervention groups, we attempted to combine the groups into a single pairwise comparison wherever possible. If any studies had multiple observations for outcome variables, we classified the time frames of the included studies as short‐term (from four weeks to three months) or long‐term (over three months) follow‐ups, and we conducted a meta‐analysis for each group of studies. If there were more than two outcome measurement points in one time frame, we chose the time point that was measured last within the time frame. As there were fewer than 25% cross‐over trials, we only analysed data from the first phase. Where appropriate, we planned to combine the results of cross‐over trials with those of parallel‐group trials.

Dealing with missing data

We based analysis on the intention‐to‐treat principle whenever possible.

Assessment of heterogeneity

We used the I² statistic to assess heterogeneity (Higgins 2011). Following the rough guide to interpretation described in the Cochrane Handbook for Systematic Reviews of Interventions, we considered that I² values of 30% to 40% might not be important, 30% to 60% may represent moderate heterogeneity, 50% to 90% may represent substantial heterogeneity, and over 75% as demonstrating considerable heterogeneity (Higgins 2011).

Assessment of reporting biases

We considered at protocol stage that it was unlikely that we would identify sufficient studies to evaluate publication bias. However, we were able to examine a funnel plot for analyses of the effect estimates with regard to the proportion of participants whose symptoms 'improved' (a secondary outcome) to identify small study effects.

Data synthesis

We showed the pooled effect estimates from all included studies that provided quantitative data using MD, SMD or RR with 95% CIs, depending on the type and the availability of outcomes. We used the Mantel‐Haenzsel fixed‐effect method and the DerSimonian and Laird random‐effects method for dichotomous data. We used the inverse‐variance method for analysis of continuous data for both the fixed‐effect or random‐effects model (Deeks 2008). At the protocol stage, we assumed that the different types, point selections and lengths of acupuncture treatment, as well as different aetiologies of gastroparesis, would contribute to clinical heterogeneity in this review. We also expected methodological diversity among included trials in terms of the risk of bias and the study design. Therefore, we employed a random‐effects model for primary analysis to incorporate clinical and methodological heterogeneity. We also planned to conduct a fixed‐effect model analysis and to compare the results of the two models

'Summary of findings' tables

We created 'Summary of findings' tables for each comparison using GRADEproGDT software (GRADEpro GDT 2015) to present the following outcomes, all at either 4 to 12 weeks (i.e. short term) or more than 12 weeks (i.e. long term).

• Improvement in symptoms of gastroparesis on validated scales

• Proportion of people whose symptoms 'improved'

• Changes of medication for gastroparesis

• Changes in quality of life

• Adverse events

We used the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness and publication bias) to assess the certainty of a body of evidence (studies that contributed data for the prespecified outcomes). We used methods and recommendations described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We considered data from RCTs as high‐certainty evidence if the five aforementioned GRADE factors were not present to any serious degree, but we downgraded the certainty of evidence once if a GRADE consideration was serious and twice if it was very serious, with a maximum of three downgradings in total and judgements of moderate, low, or very low certainty for one, two, or three downgradings, respectively. We justified decisions to downgrade the certainty of studies using footnotes and made comments to aid readers' understanding of the review where necessary.

Subgroup analysis and investigation of heterogeneity

In the protocol stage, we planned to conduct the following subgroup analyses.

• People with different aetiologies of gastroparesis (people with diabetes, whether type I or II, versus people with gastroparesis of other aetiologies)

• Type of acupuncture stimulation (manual versus electrical needle stimulation);

• Type of control intervention (sham intervention, conventional medicine, routine/usual care, or other active non‐pharmacological interventions).

• Number of acupuncture sessions (six or more versus less than six)

Predefined subgroup analysis was possible only for the type of acupuncture stimulation (manual versus electrical), due to an insufficient number of trials for analysis by aetiology. We performed and reported the main analyses by type of control intervention, rather than doing this as a subgroup analysis as planned in the protocol. As we required eligible studies to employ six or more acupuncture sessions, subgroup analysis by number of sessions was not applicable.

We performed a post hoc subgroup analysis for type of gastrokinetic (domperidone, cisapride and mosapride).

Sensitivity analysis

If a sufficient number of RCTs had been available, we would have performed sensitivity analyses to explore the influence of the risk of bias on effect estimates by excluding studies with a high or an unclear risk of bias in terms of random sequence generation, concealment of allocation, and participant and assessor blinding. However, the risk of bias was largely consistent between trials and we could only perform a sensitivity analysis for the domain of random sequence generation. We conducted a sensitivity analysis to explore the influence of how gastroparesis was defined (using objective gastric emptying tests or not) on the results of our review. We also compared results with fixed‐effect versus random‐effects models.

Where continuous or ordinal scales were used that were either non‐validated or of uncertain validity, we planned in the protocol to exclude such trials in a sensitivity analysis. However, none of the trials included in meta‐analyses employed validated outcome measures and the sensitivity analysis was not possible.

Results

Description of studies

Results of the search

We screened all titles and abstracts from the initial search of the databases to identify potentially relevant studies and to exclude explicitly irrelevant reports. When it was not possible to determine eligibility based on information in the abstract, we obtained the full text. We identified 904 reports from the English and Chinese databases and an additional 20 reports from other sources (mainly references of previous systematic reviews). We identified no reports in the Korean databases. After screening and full‐text retrieval, we included 32 randomised controlled trials (RCTs) in the qualitative syntheses, and 29 RCTs in the meta‐analyses. Two studies could not be meta‐analysed due to erroneous reporting of results (Yuan 2004), and the time‐to‐event outcome (Mu 2010). Figure 1 shows a flow diagram of the study selection process. There were three ongoing studies from China, two comparing acupuncture to sham acupuncture for participants with diabetic gastroparesis, and one comparing acupuncture with routine management for those with postoperative gastroparesis (Characteristics of ongoing studies) (Figure 2). Three potentially eligible studies (also from China), comparing acupuncture to gastrokinetic medication for participants with diabetic gastroparesis, need further translation for evaluation and are awaiting classification (Li 2017; Wang 2017; Yu 2017). We will consider these in the updated version of this review.

1.

Study flow diagram.

2.

Ongoing study flow diagram.

Included studies

See Characteristics of included studies.

Design and setting

All 32 studies adopted a parallel‐group design. Twenty‐eight studies had two groups, three studies had three groups, and one study had four groups. In Zeng 2006 (four arms), and Lin 2013 (three arms), we performed pairwise comparisons by merging the different acupuncture groups into one for comparison with the control group. In Wang 2003 and Qi 2013, which had thee arms, we made separate pairwise comparisons according to the different type of control group. One trial compared acupuncture with sham acupuncture (Shi 2012). All included studies were conducted in China, and most were indexed in the Chinese database. Nine studies recruited participants from both inpatient and outpatient settings, 10 trials recruited solely from an outpatient setting, and four trials recruited from an inpatient setting. Nine studies did not report whether the recruited participants were from an inpatient or outpatient setting. Six studies received funding from public sources (i.e. governments or a university) (Ge 2015; Ge 2016; Shi 2012; Zeng 2006; Zeng 2008; Zhang 2013). None of the other studies reported the source of funding.

Participants

The number of participants in each study ranged from 30 to 194. In one study of postsurgical gastroparesis, people undergoing abdominal surgery due to various cancer or non‐cancer gastrointestinal conditions were enrolled (Mu 2010). Among the 31 trials in people with diabetic gastroparesis, 15 based the diagnosis of diabetes on criteria from WHO, four on American Diabetes Association guidelines, four on local guidelines, and one on current insulin and a glycosylated haemoglobin of less than 7.0%. Seven studies did not report any criteria for diagnosis of diabetes (Ge 2010; Ge 2015; Zhao 2004; Wang 2009; Xu 2012; Yang 2013; Zhang 2013). To diagnose delayed gastric emptying, 27 trials used an upper gastrointestinal barium radiographic study and one trial used ultrasonography (Wang 2009). Four studies did not report whether delayed gastric emptying was objectively measured (Liu 2011; Mu 2010; Xu 2012; Xue 2014).

Interventions

In 26 RCTs acupuncture was manually stimulated, while the remaining six trials used either a mixture of manual and electrical acupuncture stimulation (Li 2013; Liu 2011; Mu 2010; Xu 2012; Zhao 2011), or electrical stimulation alone (Zhang 2014). One study compared real acupuncture versus sham acupuncture (Shi 2012). Fifteen trials compared acupuncture versus drugs (Ge 2010; Ge 2015; Ge 2016; Kong 2009; Li 2013; Lin 2013; Liu 2009; Qi 2013; Wang 2003; Wang 2007; Wang 2009; Xue 2014; Yang 2013; Yuan 2004; Zhao 2004), while 18 RCTs compared acupuncture plus other treatments (i.e. routine care or drugs) with the same other treatments (with gastrokinetic agents: Cheng 2003; He 2006; Liu 2001; Liu 2011; Liu 2012; Mo 2005; Qi 2013; Wu 2015; Xu 2012; Zeng 2008; Zhang 2007; Zhang 2013; Zhao 2011; with omeprazole (Zhang 2007), or with usual care/glycaemic control: Mu 2010; Wang 2003; Xu 2016; Zeng 2006). (Two trials were three‐way comparisons that appear in two categories (Qi 2013; Wang 2003.)

All trials used the traditional Chinese medicine (TCM) style of acupuncture. Twenty‐four trials used a fixed acupuncture regimen (i.e. a predefined set of acupuncture points in all participants), and the remaining eight trials used semi‐individualised points (i.e. use of a predefined set of acupuncture points combined with optional points that could be selected based on the diagnosis and symptom differentiation). The median number of acupuncture points applied per session was 11 (range: 3 to 25). The median number of total acupuncture sessions was 28 (range: 10 to 120). Most trials provided intensive acupuncture treatments (i.e. once daily or five times a week in 30 trials and twice daily in three trials). One study had two different acupuncture frequencies (i.e. once or twice daily) and appeared in both categories (Qi 2013). The duration of the acupuncture treatments was two to four weeks (in 28 studies), five weeks (in one study), six weeks (in two studies), or eight weeks (in one study).

Outcomes

All trials measured outcomes within 12 weeks. Two studies also measured outcomes at 20 and 24 weeks (Cheng 2003; Liu 2012), and one study measured outcomes at 1 year after baseline (Qi 2013). One study measured the number of participants with delayed gastric emptying, the number of improved participants, the level of serum gastric/motilin and Gastroparesis Cardinal Symptom Index (GCSI) score at 2 and 12 weeks after baseline (Shi 2012); however, we used only the GCSI score measured at 12 weeks for our analysis. The most frequently reported outcome was the proportion of improved participants after treatment (32 studies). Studies used either the improvement of clinical symptoms or of gastric emptying tests, or both, to define 'improved cases'. These definitions were not consistent among the studies. Other outcomes included symptom improvement measured by continuous variables (12 studies), changes in gastric emptying (eight studies), changes in diabetic control (six studies), and the occurrence of adverse events (six studies). No study assessed quality of life, or changes in medication to manage gastroparesis. Four studies measured the level of serum motilin or gastrin, or both. Four studies failed to report the objective measurement of gastric emptying as either a diagnostic criterion or as an outcome assessment of gastroparesis (Liu 2011; Mu 2010; Xue 2014; Xu 2016).

Excluded studies

We describe some key excluded studies in the Characteristics of excluded studies section. Reasons for exclusion of these studies include the following: they provided outcomes measured at less than four weeks (n = 15), had unbalanced cointerventions between groups (n = 6), were quasi‐randomised (n = 2), duplicated (n = 1), not for people with gastroparesis, and compared different acupuncture treatments (n = 1). For the full number and reasons of excluded studies, see Figure 1.

Risk of bias in included studies

We created a graphical summary of the assessment of the risk of bias for the included studies based on the seven 'Risk of bias' domains (Figure 3)

3.

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

Allocation

Sixteen of the 32 included RCTs described adequate methods for random sequence generation (i.e. random number table for 15 trials and drawing lots for one trial) and we rated them as having a low risk of bias for the domain of random sequence generation. We rated the remaining RCTs as having an unclear risk of bias due to a lack of reporting of randomisation methods.

Three trials reported adequate concealment of allocation (i.e. opaque sealed envelope method) and we thus rated them as having a low risk of bias (Liu 2012; Shi 2012; Zhang 2013). The other 29 trials did not describe allocation concealment and we assessed the risk of bias as unclear.

Blinding

Thirty‐one trials compared acupuncture with an active comparison or no treatment. One study used a sham control group, and reported that the participants were blinded (Shi 2012); however, there was information on the success of the blinding (e.g. results of blinding credibility test) and we accordingly rated it as having unclear risk of bias. In the rest of the trials, the participants could not be blinded; we rated them as having a high risk of performance bias.

Thirty‐one trials did not report the methods for blinding outcome assessors and so we rated them as having an unclear risk of detection bias. We rated one study that reported outcome assessor blinding as having a low risk of detection bias (Shi 2012).

Incomplete outcome data

Thirty‐one of 32 trials analysed all participants who were randomised and we rated these studies as having a low risk of attrition bias. In Yuan 2004, the number of participants who completed the outcome measure (n = 128) was not the same as the number of randomised participants (n = 75). We rated this study as having an unclear risk of bias.

Selective reporting

One study, Shi 2012, predefined the primary and secondary outcomes and reported all predefined measurements. We rated this study as having a low risk of reporting bias. We rated the rest of the trials as having an unclear risk of reporting bias, since they did not describe whether the study was registered or whether the protocol of the study could be publicly assessed. Thus, we could not assess the consistency between the study protocol and the study results.

Other potential sources of bias

Four trials did not report objective measurement of delayed gastric emptying (Liu 2011; Mu 2010; Xu 2012; Xue 2014). In Liu 2009, only some of the participants (64%) showed delayed gastric emptying at baseline. Nine of 31 trials for participants with diabetic gastroparesis did not specify the diagnostic criteria of diabetes. Potential bas due to the misclassification of disease might be introduced if patients with inadequate diagnosis of gastroparesis were not randomly distributed between intervention groups and included in analyses. However, we are not sure about the direction and the magnitude of such bias. Therefore, we rated trials with incomplete reporting of diagnostic criteria of gastroparesis or diabetes as having unclear risk of other bias.

Effects of interventions

See: Table 1; Table 2; Table 3; Table 4; Table 5; Table 6

Summary of findings for the main comparison. Acupuncture compared to sham acupuncture for symptomatic gastroparesis.

Acupuncture versus sham acupuncture for symptomatic gastroparesis
Patient or population: symptomatic gastroparesis Setting: inpatient setting, China Intervention: acupuncture Comparison: sham acupuncture
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with sham acupuncture	Risk with acupuncture
Improvement in symptoms of gastroparesis (short term: 4 to 12 weeks) Assessed with: Gastroparesis Cardinal Symptom Index (range of scores, 0 to 45; higher scores reflect worse outcome) Follow up: 12 weeks	Mean overall symptom score was 11.26	MD 0.17 higher (0.62 lower to 0.96 higher)	‐	60 (1 RCT)	⊕⊝⊝⊝ LOW1	‐
Improvement in symptoms of gastroparesis (long term: more than 12 weeks) ‐ not measured	‐	‐	‐	‐	‐	‐
Proportion of people whose symptoms 'improved' (short term: 4 to 12 weeks) ‐ not reported	‐	‐	‐	‐	‐	‐
Proportion of people whose symptoms 'improved' (long term: more than 12 weeks) ‐ not measured	‐	‐	‐	‐	‐	‐
Changes of medication for gastroparesis (either short or long term) ‐ not measured	‐	‐	‐	‐	‐	‐
Quality of life (either short or long term) ‐ not measured	‐	‐	‐	‐	‐	‐
Adverse events	In the sham acupuncture group, one patient experienced a vasovagal response (i.e. fainting) and another reported local bleeding and haematoma			60 (1 RCT)	⊕⊝⊝⊝ LOW1,2	‐
*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; RR: risk ratio; MD: mean difference; RCT: randomised controlled trial
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

¹ We downgraded the certainty of evidence twice to low: once for study limitations (risk of bias from lack of practitioner blinding) and once for imprecision (n = 60).
 ² Occurrence of adverse events were systematically assessed in a trial setting.

Summary of findings 2. Acupuncture compared to gastrokinetic agents for symptomatic gastroparesis.

Acupuncture compared to gastrokinetic agents for symptomatic gastroparesis
Patient or population: people with symptomatic gastroparesis Setting: inpatient or outpatient setting, China Intervention: acupuncture Comparison: gastrokinetic agent
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with gastrokinetic agent	Risk with Acupuncture
Improvement in symptoms of gastroparesis (short term: 4 to 12 weeks) Assessed with overall symptom scales (range of scores, 0 to 15 or 0 to 20; higher scores reflect worse outcome)	All 5 studies reported improvements in the acupuncture group compared with the gastrokinetic agent (mosapride or domperidone) group.		‐	360 (5 RCTs)	⊕⊝⊝⊝ VERY LOW1,2	We are uncertain whether acupuncture improves symptoms of gastroparesis.
Improvement in symptoms of gastroparesis (long term: more than 12 weeks) ‐ not measured	‐	‐	‐	‐	‐	No study assessed this outcome in the long term
Proportion of people whose symptoms 'improved' (short term: 4 to 12 weeks)	Study population		RR 1.25 (1.17 to 1.33)	963 (12 RCTs)	⊕⊝⊝⊝ VERY LOW1,3	We are uncertain whether acupuncture improves symptoms of gastroparesis.
	706 per 1,000	882 per 1,000 (826 to 939)
Proportion of people whose symptoms 'improved' (long term: more than 12 weeks) Follow up: mean 12 months	Study population		RR 0.79 (0.36 to 1.73)	85 (1 RCT)	⊕⊝⊝⊝ VERY LOW1,3	The number of improved participants in the acupuncture group was lower than those in the gastrokinetic agent group. We are uncertain whether acupuncture improves symptoms of gastroparesis.
	26 per 100	21 per 100 (9 to 45)
Changes of medication for gastroparesis (either short or long term) ‐ not measured	‐	‐	‐	‐	‐	No study reported this outcome
Quality of life (either short or long term) ‐ not measured	‐	‐	‐	‐	‐	No study reported this outcome
Adverse events	1 trial reported diarrhoea (n = 1), fatigue (n = 1), dermatitis (n = 1), and dry mouth (n = 2) and a 2nd trial reported mild upper abdominal pain (n = 1) in the gastrokinetic group. Mild dizziness occurred in 1 person both in the acupuncture group and the gastrokinetic group in the 2nd trial.		‐	180 (2 RCTs)	⊕⊝⊝⊝ VERY LOW4	12 of 14 trials provided no information on adverse events.
*The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RR: risk 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

¹ We downgraded the evidence twice for study limitations: unclear risk of selection bias in the domains of allocation concealment, detection bias in the domains of outcome assessor blinding and for high risk of performance bias in the domains of blinding of participants/personnel.
 ² We downgraded the certainty of the evidence once for inconsistency; The pooled effects estimates showed considerable heterogeneity; therefore, we did not meta‐analyse the results.
 ³ We downgraded the certainty of the evidence once for indirectness related to the use of unvalidated outcome measures and the acupuncture protocol. The frequency of sessions was relatively intensive and may not be applicable in other contexts. Whether the response rate is validated and can be used in other contexts is unclear.
 ⁴ We downgraded the certainty of the evidence once for imprecision and twice for study limitations (risk of detection bias due to assessor unblinding and risk of selective reporting). The sample size was small and events sparse. Most trials (12/14) provided no data on adverse events.

Summary of findings 3. Acupuncture compared to non‐gastrokinetic agents for symptomatic gastroparesis.

Acupuncture versus non‐gastrokinetic agents for symptomatic gastroparesis
Patient or population: symptomatic gastroparesis Setting: unclear (not reported; assumed to be inpatient setting given the intensive acupuncture sessions), China Intervention: acupuncture Comparison: non‐gastrokinetic agents
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with non‐gastrokinetic agent (cimetidine, a H₂ receptor antagonist)	Risk with acupuncture
Improvement in symptoms of gastroparesis (short term: 4 to 12 weeks) Follow up: 4 weeks	The study reported individual symptoms (see text). Fewer participants in the acupuncture‐treated group experienced abdominal distension than in the cimetidine‐treated group. Slightly fewer people in the acupuncture group than in the cimetidine‐treated group had nausea and vomiting, epigastric pain, or postprandial discomfort, but the CI did not rule out the possibility of little or no difference between groups. Results for diarrhoea and diarrhoea‐constipation were very imprecise, with few events and high attrition.		‐	68 (1 RCT)	⊕⊝⊝⊝ VERY LOW1	We are uncertain whether acupuncture improves symptoms of gastroparesis.
Improvement in symptoms of gastroparesis (long term: more than 12 weeks) ‐ not measured	‐	‐	‐	‐	‐	‐
Proportion of people whose symptoms 'improved' (short term: 4 to 12 weeks)	833 per 1,000	852 per 1,000 (630 to 1,000)	RR 1.38 (1.02 to 1.86)	68 (1 RCT)	⊕⊝⊝⊝ VERY LOW1	We are uncertain whether acupuncture improves symptoms of gastroparesis.
Proportion of people whose symptoms 'improved' (long term: more than 12 weeks) ‐ not measured	‐	‐	‐	‐	‐	‐
Changes of medication for gastroparesis (either short or long term) ‐ not measured	‐	‐	‐	‐	‐	‐
Quality of life (either short or long term) ‐ not measured	‐	‐	‐	‐	‐	‐
Adverse events ‐ not reported	‐	‐	‐	‐	‐	‐
*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; RR: risk ratio; MD: mean difference; RCT: randomised controlled trial
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

¹ We downgraded the certainty of evidence three times to very low: twice for study limitations and once for imprecision. The trial was open, it is unclear whether outcome assessment was blinded and the risk of bias for several other criteria was unclear. The single study that provided data had 68 participants and the CI were wide.

Summary of findings 4. Acupuncture plus gastrokinetic agents compared to gastrokinetic agents alone for symptomatic gastroparesis.

Acupuncture plus gastrokinetic agents compared to gastrokinetic agents alone for symptomatic gastroparesis
Patient or population: symptomatic gastroparesis Setting: inpatient or outpatient setting, China Intervention: acupuncture plus gastrokinetic agents Comparison: gastrokinetic agents alone
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with gastrokinetic agents alone	Risk with acupuncture plus gastrokinetic agents
Improvement in symptoms of gastroparesis (short term: 4 to 12 weeks) Assessed with: overall symptom scales (range of scores, 0 to 15 or 18; higher scores reflect worse outcome) Follow up: range 4 weeks to 5 weeks	The mean improvement in symptoms of gastroparesis (short term: 4 to 12 weeks) was 5.61	MD 1.96 lower (2.42 lower to 1.5 lower)	‐	132 (2 RCTs)	⊕⊝⊝⊝ VERY LOW1,2	A lower score on a symptom scale equates to a better outcome We are uncertain whether acupuncture improves symptoms of gastroparesis. 11/13 trials did not report improvement on overall symptom scales
Improvement in symptoms of gastroparesis (long term: more than 12 weeks) ‐ not reported	‐	‐	‐	‐	‐	‐
Proportion of people whose symptoms 'improved' (short term: 4 to 12 weeks)	742 per 1000	898 per 1000 (853 to 942)	RR 1.21 (1.15 to 1.27)	1107 (13 RCTs)	⊕⊝⊝⊝ VERY LOW1,2,3	We are uncertain whether acupuncture when added to gastrokinetic agents improves symptoms of gastroparesis in the short term.
Proportion of people whose symptoms 'improved' (long term: more than 12 weeks)	571 per 1000	806 per 1000 (680 to 949)	RR 1.41 (1.19 to 1.66)	237 (3 RCTs)	⊕⊝⊝⊝ VERY LOW1,2,3	We are uncertain whether acupuncture when added to gastrokinetic agents improves symptoms of gastroparesis in the short term.
Changes of medication for gastroparesis (either short or long term) ‐ not measured	‐	‐	‐	‐	‐	‐
Quality of life (either short or long term) ‐ not measured	‐	‐	‐	‐	‐	‐
Adverse events	Reported events were dizziness, dry skin, skin rash, loose stool or diarrhoea, abdominal pain and the local bruise at the needled sites. All were mild and spontaneously resolved without requiring further treatments.		‐	408 (4 RCTs)	⊕⊝⊝⊝ VERY LOW4	9 of 13 studies did not provide information on adverse events.
*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). The mean value for the continuous outcome is the mean of the control group values in the analysed studies. CI: confidence interval; RR: risk ratio; MD: mean difference; RCT: randomised controlled trial
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

¹ We downgraded the certainty of evidence twice, for unclear risk of selection bias in the domain of allocation concealment, detection bias in the domain of outcome assessor blinding and for high risk of performance bias in the domain of blinding of participants/personnel.
 ² We downgraded the certainty of evidence once further for indirectness related to the acupuncture protocol. The frequency of sessions was relatively intensive and may not be applicable in other contexts.
 ³ The use of unvalidated outcome measures also contributed to the very low certainty of the evidence. Whether the response rate is validated and can be used in other contexts is unclear.
 ⁴ We downgraded the certainty of the evidence twice for study limitations and once for imprecision. In addition to the high risk of selection bias and a lack of blinding, most trials (9/13) provided no data on adverse events. The sample size was insufficient for the detection of rare adverse events.

Summary of findings 5. Acupuncture plus non‐gastrokinetic agents compared to non‐gastrokinetic agents alone for symptomatic gastroparesis.

Acupuncture plus non‐gastrokinetic agents compared to non‐gastrokinetic agents alone for symptomatic gastroparesis
Patient or population: symptomatic gastroparesis Setting: inpatient or outpatient setting, China Intervention: acupuncture plus omeprazole Comparison: omeprazole alone
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with omeprazole alone	Risk with acupuncture plus omeprazole
Improvement in symptoms of gastroparesis (short term: 4 to 12 weeks) Assessed with: overall symptom scales (range of scores, 0 to unclear upper limit; higher scores reflect worse outcome)	The study reported individual symptoms (see text). The mean score of each symptom was lower in the acupuncture combined with omeprazole group than those in the omeprazole alone group; bloating, stomach pain, belching, abnormal defecation and decreased oral intake.		‐	80 (1 RCT)	⊕⊝⊝⊝ VERY LOW1	We are uncertain whether acupuncture improves symptoms of gastroparesis.
Improvement in symptoms of gastroparesis (long term: more than 12 weeks) ‐ not reported	‐	‐	‐	‐	‐	‐
Proportion of people whose symptoms 'improved' (short term: 4 to 12 weeks)	675 per 1000	878 per 1000 (682 to 1.66)	RR 1.30 (1.01 to 1.66)	80 (1 RCT)	⊕⊝⊝⊝ VERY LOW1	We are uncertain whether acupuncture improves symptoms of gastroparesis.
Proportion of people whose symptoms 'improved' (long term: more than 12 weeks)	‐	‐	‐	‐	‐	‐
Changes of medication for gastroparesis (either short or long term) ‐ not measured	‐	‐	‐	‐	‐	‐
Quality of life (either short or long term) ‐ not measured	‐	‐	‐	‐	‐	‐
Adverse events ‐ not reported	‐	‐	‐	‐	‐	‐
*The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RR: risk ratio; MD: mean difference; RCT: randomised controlled trial
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

¹We downgraded the certainly of evidence once for study limitations, once for imprecision and once for indirectness. The trial provided no information on allocation concealment and blinding of outcome assessors. CI include the possibility of no effect. 'Improvement' was not defined and it is unclear whether this outcome is validated and can be used in other contexts.

Summary of findings 6. Acupuncture plus routine care/diabetes control compared to the same treatment alone (routine care/diabetes control) for symptomatic gastroparesis.

Acupuncture with routine care/diabetes control compared to routine care/diabetes control alone) for symptomatic gastroparesis
Patient or population: symptomatic gastroparesis Setting: inpatient or outpatient setting, China Intervention: acupuncture with routine care/diabetes control Comparison: the same routine care/diabetes control
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with routine care/diabetes control alone	Risk with acupuncture plus routine care/diabetes control
Improvement in symptoms of gastroparesis (short term: 4 to 12 weeks) Assessed with: overall symptom scales ‐ not reported	‐	‐	‐	‐	‐	‐
Improvement in symptoms of gastroparesis (long term: more than 12 weeks) ‐ not measured	‐	‐	‐	‐	‐	‐
Proportion of people whose symptoms 'improved' (short term: 4 to 12 weeks)	617 per 1000	907 per 1000 (654 to 1000)	RR 1.47 (1.06 to 2.04)	217 (3 RCTs)	⊕⊝⊝⊝ VERY LOW1,2,3	We are uncertain whether acupuncture improves symptoms of gastroparesis.
Proportion of people whose symptoms 'improved' (long term: more than 12 weeks) ‐ not measured	‐	‐	‐	‐	‐	‐
Changes of medication for gastroparesis (either short or long term) ‐ not measured	‐	‐	‐	‐	‐	‐
Quality of life (either short or long term) ‐ not measured	‐	‐	‐	‐	‐	‐
Adverse events ‐ not reported	‐	‐	‐	‐	‐	‐
*The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RR: risk ratio; MD: mean difference; RCT: randomised controlled trial
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

¹ We downgraded the certainty of evidence twice, for unclear risk of selection bias in the domain of allocation concealment, detection bias in the domain of outcome assessor blinding and for high risk of performance bias in the domain of blinding of participants/personnel.
 ² We downgraded the certainty of evidence once further for indirectness related to the acupuncture protocol in the outcome of proportion of people whose symptoms 'improved' (the only reported outcome).The frequency of sessions was relatively intensive and may not be applicable in other contexts.
 ³ The use of unvalidated outcome measures also contributed to the very low uncertainty of the evidence. Whether the response rate is validated and can be used in other contexts is unclear.

We suspect that the data from Ge 2015 were duplicated in Ge 2016 and therefore we excluded it from Analysis 2.1, Analysis 2.3 and Analysis 2.6. We attempted to obtain additional information from the authors, although only five studies provided contact details. We received a response from one author (Liu 2012), and are awaiting feedback from others.

2.1. Analysis.

Comparison 2 Acupuncture versus gastrokinetic agents, Outcome 1 Improvement in symptoms of gastroparesis ‐ overall symptom scores (short‐term: 4 to 12 weeks).

2.3. Analysis.

Comparison 2 Acupuncture versus gastrokinetic agents, Outcome 3 Proportion of people whose symptoms 'improved'(short‐term: 4 to 12 weeks).

2.6. Analysis.

Comparison 2 Acupuncture versus gastrokinetic agents, Outcome 6 Changes in the rate of gastric emptying‐ gastric emptying‐barium test.

Acupuncture versus sham acupuncture

One study (Shi 2012, 60 participants) compared acupuncture with sham acupuncture (i.e. needling on non‐acupuncture points). See Table 1.

Primary outcome: improvement in symptoms of gastroparesis at 4 to 12 weeks ('short term')

The Gastroparesis Cardinal Symptom Index (GCSI; range of scores, 0 to 45; higher scores reflect worse outcome) was measured at 2 and 12 weeks from baseline, and the latter contributed to the short‐term effect analysis (Analysis 1.1), which showed no clear difference between groups (mean difference (MD) 0.17, 95% confidence interval (CI) ‐0.62 to 0.96; n = 60). We considered the certainty of the evidence to be low due to study limitations (risk of performance bias from potential unmasking of participants and lack of blinding among practitioners and imprecision).

1.1. Analysis.

Comparison 1 Acupuncture versus sham acupuncture, Outcome 1 Improvement in symptoms of gastroparesis ‐ overall symptom scores (short‐term: 4 to 12 weeks).

Secondary outcomes

Improvement in symptoms of gastroparesis at over 12 weeks ('long term')

No study measured the outcome.

Proportion of people whose symptoms 'improved' in the short and long term

No data were available for the analysis.

Changes of medication for gastroparesis in the short and long term

No data were available for the analysis.

Changes in quality of life in the short and long term

No data were available for the analysis.

Changes in the rate of gastric emptying in the short and long term

No data were available for the analysis.

Effects on glycaemic control in the short and long term

No data were available for the analysis.

Adverse events

In the sham acupuncture group, one patient experienced a vasovagal response (i.e. fainting) and another reported local bleeding and haematoma (n = 60; very low‐certainty evidence due to study limitations and serious imprecision).

Acupuncture versus gastrokinetic agents

Fourteen studies (949 participants) compared the effects of acupuncture to those of gastrokinetic agents (Ge 2010; Ge 2015; Ge 2016; Kong 2009; Li 2013; Lin 2013; Liu 2009; Qi 2013; Wang 2003; Wang 2007; Wang 2009; Xue 2014; Yang 2013; Yuan 2004). See Table 2.

Primary outcome: improvement in symptoms of gastroparesis at 4 to 12 weeks ('short term')

Overall, scaled symptom scores were measured in six trials (Ge 2015; Ge 2016; Li 2013; Wang 2007; Xue 2014; Yang 2013) (range from 0 to 15 or 20 with higher scores indicating worse outcomes). Since we suspect that the data from Ge 2015 were duplicated with those from Ge 2016, only five studies contributed to the analysis (360 participants). We did not pool the results due to considerable heterogeneity (I² = 89%). Compared with mosapride, acupuncture showed significant improvement in symptoms of gastroparesis in three studies: Wang 2007 (MD ‐2.50, 95% CI ‐4.10 to ‐0.90; n = 60), Yang 2013 (MD ‐1.77, 95% CI ‐2.74 to ‐0.80; n = 70), and Xue 2014 (MD ‐3.47, 95% CI ‐5.44 to ‐1.50; n = 60). The other two trials also favoured acupuncture when compared with domperidone, though with smaller effects (Ge 2016: MD ‐3.30, 95% CI ‐3.48 to ‐3.12; n = 100; and Li 2013: MD ‐1.94, 95% CI ‐2.41 to ‐1.47; n = 70). We judged the certainty of evidence for this outcome as very low. See Analysis 2.1.

One study, Wang 2007, reported a modest benefit for acupuncture versus mosapride in terms of individual symptoms of bloating, stomach pain, belching, and abnormal defecation, but not loss of appetite (Analysis 2.2).

2.2. Analysis.

Comparison 2 Acupuncture versus gastrokinetic agents, Outcome 2 Improvement in symptoms of gastroparesis ‐ individual symptom scores (acupuncture versus mosapride).

Secondary outcomes

Improvement in symptoms of gastroparesis at over 12 weeks ('long term')

No study measured the outcome.

Proportion of people whose symptoms 'improved' in the short and long term

Twelve trials reported the proportion of participants whose symptoms 'improved' (Ge 2010; Ge 2016; Kong 2009; Li 2013; Lin 2013; Liu 2009; Qi 2013; Wang 2003; Wang 2007; Wang 2009; Xue 2014; Yang 2013). Data from Ge 2015 were discarded, as above. Yuan 2004 reportedly randomised 75 participants but reported the outcome in 127 participants; therefore, we did not include the trial in the analysis.

The proportion of participants whose symptoms improved favoured acupuncture versus gastrokinetic drugs in the short term (risk ratio (RR) 1.25, 95% CI 1.17 to 1.33; I² = 8%; 12 RCTs; n = 963; very low‐certainty evidence) (Analysis 2.3) (Figure 4).

4.

Forest plot of comparison: 2 Acupuncture versus gastrokinetic agents, outcome: 2.3 Proportion of people whose symptoms 'improved'(short‐term: 4 to 12 weeks).

Qi 2013 reported that fewer participants improved with acupuncture than with conventional agents in the long term (RR 0.79, 95% CI 0.36 to 1.73; 1 RCT; n = 85; very low‐certainty evidence) (Analysis 2.4). No other study reported long‐term outcomes.

2.4. Analysis.

Comparison 2 Acupuncture versus gastrokinetic agents, Outcome 4 Proportion of people whose symptoms 'improved' (long‐term: more than 12 weeks) (acupuncture versus mosapride).

Changes of medication for gastroparesis in the short and long term

No study measured the outcome.

Changes in quality of life in the short and long term

No study measured the outcome.

Changes in the rate of gastric emptying in the short and long term

One study, Wang 2009, observed gastric emptying by stomach ultrasonography and used the change in peristaltic flow and emptying time as parameters of improvement. Participants with complete or partial improvement of the peristaltic flow waves and less than six hours of gastric emptying time, accompanied by symptomatic improvement, were regarded as improved. More participants in the acupuncture group than in the mosapride group had improved peristaltic flow (RR 1.80, 95% CI 1.28 to 2.54; n = 76) and an improved gastric emptying time (RR 1.53, 95% CI 1.11 to 2.11; n = 76) (Analysis 2.5).

2.5. Analysis.

Comparison 2 Acupuncture versus gastrokinetic agents, Outcome 5 Changes in the rate of gastric emptying ‐ response rate (acupuncture versus mosapride).

We did not pool the results on the amount of residual barium meal fraction four to six hours following intake of 20 barium strips due to there being considerable heterogeneity (I² = 93%). The number of residual barium strips were significantly less in the acupuncture group than the domperidone group in Ge 2016 (MD ‐2.82, 95% CI ‐3.29 to ‐2.35; n = 100) but not in Wang 2007 (MD ‐1.64, 95% CI ‐3.62 to 0.64; n = 60). One study reported the percentage reduction of barium strips which favoured acupuncture compared to the control (Li 2013: MD ‐18.20%, 95% CI ‐22.99% to ‐13.41%). (Analysis 2.6).

Effects on glycaemic control in the short and long term

There was a reduction in serum HbA1c levels that may not be clinically important (according to Svensson 2017) in the acupuncture group versus the mosapride group in the short term (MD ‐1.30%, 95% CI ‐1.78 to ‐0.82; one RCT; n = 70) (Analysis 2.7). There was little or no difference in the fasting blood sugar (FBS) level in the short term in Wang 2009 (MD 0.36 mg/dL, 95% CI ‐1.61 to 2.33; one RCT; n = 76), whereas Xue 2014 showed a decrease favouring acupuncture (MD ‐6.70 mg/dL, 95% CI ‐9.84 to ‐3.56). Heterogeneity was too high (I² = 93%) for these data to be pooled (Analysis 2.8).

2.7. Analysis.

Comparison 2 Acupuncture versus gastrokinetic agents, Outcome 7 Effects on glycaemic control ‐ control of diabetes‐HbA1c (%) (acupuncture versus mosapride).

2.8. Analysis.

Comparison 2 Acupuncture versus gastrokinetic agents, Outcome 8 Effects on glycaemic control ‐ control of diabetes‐FBS (mg/dL) (acupuncture versus mosapride).

Adverse events

One study, Wang 2007, reported adverse events of diarrhoea (n = 1), fatigue (n = 1), dermatitis (n = 1), and dry mouth (n = 2), all of which occurred in the control group (n = 30). Another study (Lin 2013) measured liver and renal function, blood biochemistry, electrocardiogram and symptomatic adverse events and reported mild dizziness both in the acupuncture group (n = 1) and gastrokinetic drug group (n = 1), and mild upper abdominal pain in the gastrokinetic drug group (n = 1). All symptoms were temporary and resolved spontaneously. There were no adverse events in terms of liver and renal function, blood biochemistry and electrocardiogram. The remaining 12 trials did not provide information on the occurrence of adverse events.

Acupuncture versus non‐gastrokinetic agents

One trial (68 participants) compared acupuncture to a non‐gastrokinetic agent (cimetidine, a histamine H₂ receptor antagonist) and measured effects after four weeks (Zhao 2004). See Table 3.

Primary outcome: improvement in symptoms of gastroparesis at 4 to 12 weeks ('short term')

The study reported individual symptom scores. Fewer participants in the acupuncture‐treated group experienced abdominal distension than in the cimetidine‐treated group. Slightly fewer people in the acupuncture group had nausea and vomiting, epigastric pain, or postprandial discomfort, but the CI did not rule out the possibility of little or no difference between groups. Results for diarrhoea and diarrhoea‐constipation were very imprecise, with few events and high attrition (Analysis 3.1). We graded the evidence as of very low certainty because of serious study limitations and imprecision.

3.1. Analysis.

Comparison 3 Acupuncture versus non‐gastrokinetic agents, Outcome 1 Proportion of people whose symptoms 'improved' ‐ individual symptoms (acupuncture versus cimetidine).

Secondary outcomes

Improvement in symptoms of gastroparesis at over 12 weeks ('long term')

No study measured the outcome.

Proportion of people whose symptoms 'improved' in the short and long term

The proportion of participants whose symptoms improved was greater with acupuncture versus cimetidine in the short term (RR 1.38, 95% CI 1.02 to 1.86; one RCT; n = 68; very low‐certainty evidence) (Analysis 3.2). No study measured long‐term outcomes.

3.2. Analysis.

Comparison 3 Acupuncture versus non‐gastrokinetic agents, Outcome 2 Proportion of people whose symptoms 'improved'.

Changes of medication for gastroparesis in the short and long term

No study measured the outcome.

Changes in quality of life in the short and long term

No study measured the outcome.

Changes in the rate of gastric emptying in the short and long term

No study measured the outcome.

Effects on glycaemic control in the short and long term

No study measured the outcome.

Adverse events

No study provided information on the occurrence of adverse events.

Acupuncture plus gastrokinetic agents versus gastrokinetic agents alone

Thirteen studies (1107 participants) contributed to this comparison (Cheng 2003; He 2006; Liu 2001; Liu 2011; Liu 2012; Mo 2005; Qi 2013; Wu 2015; Xu 2012; Zeng 2008; Zhao 2011; Zhang 2007; Zhang 2013). (Cheng 2003 was a comparison of acupuncture plus domperidone and vitamin B6 versus domperidone and vitamin B6.) See Table 4.

Primary outcome: improvement in symptoms of gastroparesis at 4 to 12 weeks ('short term')

Scores on overall symptom scales (possible range of scores 0 to 15 or 18, a lower score indicates a better outcome) were measured in two trials (Zeng 2008; Zhang 2007). A reduction of overall symptoms related to gastroparesis was found in the acupuncture group compared with the control group in the short term (MD ‐1.96, 95% CI ‐2.42 to ‐1.50; I² = 0%; two RCTs; n = 132; very low‐certainty evidence) (Analysis 4.1).

4.1. Analysis.

Comparison 4 Acupuncture plus gastrokinetic agents versus gastrokinetic agents alone, Outcome 1 Improvement in symptoms of gastroparesis ‐ overall symptom scores (short‐term: 4 to 12 weeks) (gastrokinetic: domperidone).

Secondary outcomes

Improvement in symptoms of gastroparesis at over 12 weeks ('long term')

No study measured the outcome.

Proportion of people whose symptoms 'improved' in the short and long term

A greater proportion of participants 'improved' with acupuncture plus gastrokinetic drugs than with gastrokinetic drugs alone in the short term (RR 1.21, 95% CI 1.15 to 1.27; I² = 0; 13 RCTs; n = 1107) (Analysis 4.2; Analysis 8.1). Long‐term response rates, in terms of the proportion of participants 'improved', at 5, 6 or 12 months from baseline favoured the acupuncture group over the control group (RR 1.41, 95% CI 1.19 to 1.66; I² = 1%; three RCTs; n = 237; very low‐certainty evidence) (Analysis 4.3).

4.2. Analysis.

Comparison 4 Acupuncture plus gastrokinetic agents versus gastrokinetic agents alone, Outcome 2 Proportion of people whose symptoms 'improved' (short‐term: 4 to 12 weeks).

8.1. Analysis.

Comparison 8 Acupuncture versus gastrokinetic agents: sensitivity analyses, fixed‐effect versus random‐effects, Outcome 1 Proportion of people whose symptoms 'improved' (4 to 12 weeks): random effects model.

4.3. Analysis.

Comparison 4 Acupuncture plus gastrokinetic agents versus gastrokinetic agents alone, Outcome 3 Proportion of people whose symptoms 'improved' (long‐term: more than 12 weeks).

Changes of medication for gastroparesis in the short and long term

No study measured the outcome.

Changes in quality of life in the short and long term

No study measured the outcome.

Changes in the rate of gastric emptying in the short and long term

No study measured the outcome.

Effects on glycaemic control in the short and long term

There was a reduction in serum HbA1c levels in the acupuncture plus gastrokinetic drug group versus the gastrokinetic drugs alone group in the long term (MD ‐0.80%, 95% CI ‐1.32 to ‐0.28; one RCT; n = 70) (Liu 2012; Analysis 4.4); however, there was no difference in the fasting blood sugar (FBS) level in the long term (MD ‐1.8 mg/dL, 95% CI ‐26.80 to 23.20; one RCT; n = 72) (Cheng 2003; Analysis 4.5). The average number of hypoglycaemic events per day was lower in the acupuncture group in the short term (MD ‐0.53, 95% CI ‐0.62 to ‐0.44; one RCT; n = 70) and in the long term (MD ‐0.38, 95% CI ‐0.63 to ‐0.13; one RCT; n = 70) (Liu 2012; Analysis 4.6).

4.4. Analysis.

Comparison 4 Acupuncture plus gastrokinetic agents versus gastrokinetic agents alone, Outcome 4 Effects on glycaemic control ‐ control of diabetes‐HbA1c (%): long term) (gastrokinetic: mosapride).

4.5. Analysis.

Comparison 4 Acupuncture plus gastrokinetic agents versus gastrokinetic agents alone, Outcome 5 Effects on glycaemic control ‐ control of diabetes‐FBS) (mg/dL) (long‐term) (gastrokinetic: domperidone (plus vitamin B6)).

4.6. Analysis.

Comparison 4 Acupuncture plus gastrokinetic agents versus gastrokinetic agents alone, Outcome 6 Effects on glycaemic control ‐ control of diabetes‐hypoglycaemic events ‐ mean frequency/day (gastrokinetic: mosapride).

Adverse events

Four studies provided information on the occurrence of adverse events (Cheng 2003; Liu 2012; Zhang 2007; Zhang 2013). One study reported no adverse events in the acupuncture plus domperidone and vitamin B6 group (n = 36), but reported dizziness (n = 3), xerostomia (n = 3), and loose stool (n = 6) in the domperidone plus vitamin B6 group (n = 36) at six‐month follow‐up (Cheng 2003). Another study of domperidone with or without acupuncture reported no occurrence of adverse events (Zhang 2007). In Liu 2012, six participants in the acupuncture plus mosapride group (n = 35) and three participants in the mosapride alone group (n = 35) experienced adverse events such as abdominal pain, diarrhoea, and skin rash. In Zhang 2013, mild and temporary local hematoma on needled sites occurred in three of 98 participants in the acupuncture plus domperidone group. Abdominal pain (n = 12), diarrhoea (n = 12) and dizziness (n = 2) occurred in the domperidone alone group (n = 96); these events were mild and spontaneously resolved without requiring further treatments. We graded the evidence on adverse events as of very low certainty.

Acupuncture plus non‐gastrokinetic agents versus non‐gastrokinetic agents alone

One study (80 participants) compared acupuncture combined with omeprazole, a non‐gastrokinetic agent (proton‐pump inhibitor), versus omeprazole alone (Zhang 2014).

Primary outcome: improvement in symptoms of gastroparesis at 4 to 12 weeks ('short term')

One study (80 participants) reported a modest short‐term benefit in terms of individual symptoms of bloating, stomach pain, belching, abnormal defecation, and decreased oral intake for the combination of acupuncture and omeprazole versus omeprazole alone (Analysis 5.1) (Zhang 2014).

5.1. Analysis.

Comparison 5 Acupuncture plus omeprazole versus omeprazole alone, Outcome 1 Individual symptom scores (short term: 4 to 12 weeks).

Secondary outcomes

Improvement in symptoms of gastroparesis at over 12 weeks ('long term')

No study measured the outcome.

Proportion of people whose symptoms 'improved' in the short and long term

Acupuncture combined with omeprazole showed marginally better results than the same medication alone (RR 1.30, 95% CI 1.01 to 1.66; one RCT; n = 80; very low‐certainty evidence) (Analysis 5.2).

5.2. Analysis.

Comparison 5 Acupuncture plus omeprazole versus omeprazole alone, Outcome 2 Proportion of people whose symptoms 'improved' (short‐term: 4 to 12 weeks).

Changes of medication for gastroparesis in the short and long term

No study measured the outcome.

Changes in quality of life in the short and long term

No study measured the outcome.

Changes in the rate of gastric emptying in the short and long term

Data from Zhang 2014 indicated that the amount of residual barium meal fraction four hours after intake was not significantly reduced in the acupuncture plus omeprazole group compared with that resulting from omeprazole alone (MD ‐0.08; 95% CI ‐0.67 to 0.51; n = 80) (Analysis 5.3).

5.3. Analysis.

Comparison 5 Acupuncture plus omeprazole versus omeprazole alone, Outcome 3 Changes in the rate of gastric emptying ‐ barium test (short term: 4 to 12 weeks).

Effects on glycaemic control in the short and long term

There was no difference in the fasting blood sugar level in the short term with acupuncture plus omeprazole versus omeprazole alone (MD ‐7.4 mg/dL, 95% CI ‐14.95 to 0.15; one RCT; n = 80) (Analysis 5.4).

5.4. Analysis.

Comparison 5 Acupuncture plus omeprazole versus omeprazole alone, Outcome 4 Effects on glycaemic control ‐ control of diabetes‐FBS) (mg/dL) (short‐term: 4 to 12 weeks).

Adverse events

The study did not provide information on the occurrence of adverse events.

Acupuncture plus routine care/glycaemic control versus routine care/glycaemic control alone

Four studies (178 participants) compared the effects of acupuncture versus routine care (Mu 2010; Wang 2003; Xu 2016; Zeng 2006).

Primary outcome: improvement in symptoms of gastroparesis at 4 to 12 weeks ('short term')

No study measured the outcome.

Secondary outcomes

Improvement in symptoms of gastroparesis at over 12 weeks ('long term')

No study measured the outcome.

Proportion of people whose symptoms 'improved' in the short and long term

There were no clear differences between acupuncture plus routine care and routine care alone (routine glycaemic control including medication and exercise without acupuncture) in the short term, in terms of the proportion of participants whose symptoms 'improved' (RR 1.47, 95% CI 1.06 to 2.04; I² = 62%; three RCTs; n = 217) (Analysis 6.1). There were no short‐term benefits of acupuncture combined with antidiabetic medications versus antidiabetic medications without acupuncture in the short term (RR 1.36, 95% CI 0.97 to 1.90; one RCT; n = 80) (Zeng 2006). In a trial of people with postsurgical gastroparesis (Mu 2010), 79% of the participants (19 of 24) recovered within two weeks in the acupuncture group, compared with 47% (8 of 17) participants in the routine postsurgical management group. All participants recovered within four weeks.

6.1. Analysis.

Comparison 6 Acupuncture plus routine care/gycaemic control versus routine care/glycaemic control alone, Outcome 1 Proportion of people whose symptoms 'improved' (short‐term: 4 to 12 weeks).

Changes of medication for gastroparesis in the short and long term

No study measured the outcome.

Changes in quality of life in the short and long term

No study measured the outcome.

Changes in the rate of gastric emptying in the short and long term

One trial reported the dominant frequency and amplitude of gastric waves measured by electrogastrography (Zeng 2006). Three different acupuncture groups (acupuncture with strong, moderate, and mild manual stimulation) and the antidiabetic medication group were compared. When different acupuncture groups were combined to assess the overall effects of acupuncture plus diabetes medication compared with medication for diabetes control alone, a reduced dominant frequency in the gastric antrum (MD ‐0.83 cycles per minute; 95% CI ‐1.02 to ‐0.64; n = 80) and gastric body (MD ‐0.72 cycles per minute; 95% CI ‐0.93 to ‐0.51; n = 80) and an increased amplitude in the gastric antrum (MD 20.50 microvolts (µV), 95% CI 16.78 to 24.21; n = 80) and gastric body (MD 13.50 µV, 95% CI 8.67 to 18.32; n = 80) were observed (Analysis 6.2).

6.2. Analysis.

Comparison 6 Acupuncture plus routine care/gycaemic control versus routine care/glycaemic control alone, Outcome 2 Changes in the rate of gastric emptying ‐ electrogastrography (short term: 4 to 12 weeks).

Effects on glycaemic control in the short and long term

No study measured the outcome.

Adverse events

No study provided information on the occurrence of adverse events.

Subgroup and sensitivity analyses

We planned subgroup analyses for three variables (differences in aetiologies of gastroparesis, types of acupuncture treatments, and control interventions). There were insufficient data to perform subgroup analysis by aetiology of gastroparesis. A statistical test of subgroup differences for types of acupuncture treatment found no statistically significant heterogeneity of effect estimates between the two substrata (manual versus electrical stimulation) (Analysis 7.1). We created separate comparisons for different categories of concomitant and control intervention (gastrokinetic agents, individual non‐gastrokinetic agents and usual care (including hypoglycaemic agents)) but performed post hoc subgroup analyses for the type of gastrokinetic agent. These showed no subgroup effects (Analysis 2.1; Analysis 2.3; Analysis 4.2; Analysis 4.3).

7.1. Analysis.

Comparison 7 Acupuncture versus gastrokinetic agents: subgroup analysis, manual acupuncture versus electroacupuncture, Outcome 1 Proportion of people whose symptoms 'improved' (4 to 12 weeks): acupuncture versus gastrokinetics.

We planned sensitivity analyses for five variables (low or high/unclear risk of bias in terms of randomisation, concealment of allocation, participant blinding, and outcome blinding, as well as objective measurements of gastric emptying at baseline), but data only allowed sensitivity analyses for the variables of randomisation and the use of objective gastric emptying measures. We conducted sensitivity analyses for the proportion of 'improved' patients (i.e. response rate) in the short‐term for acupuncture compared with gastrokinetic drug (Analysis 8.1), and for acupuncture plus the control interventions compared with the control interventions alone (Analysis 9.1). There were no significant differences in the overall effect estimates based on the relevant studies when the analyses were restricted to trials of valid methods of random sequence and those that measured gastric emptying (Analysis 8.1; Analysis 9.1). Comparison of fixed‐effect versus random‐effects model analyses revealed similar effect estimates and 95% CIs (Analysis 8.1; Analysis 8.2; Analysis 9.1; Analysis 9.2). A post hoc sensitivity analysis to explore whether the exclusion of quasi‐RCTs (i.e. Hong 2010; Xue 2006; and Zhang 2007a) made a difference in the effect estimate showed no difference (results not shown).

9.1. Analysis.

Comparison 9 Acupuncture plus treatment (gastrokinetics, other drugs or routine care) versus the same treatment alone: sensitivity analyses, valid randomisation and objective emptying measures, Outcome 1 Proportion of people whose symptoms 'improved' (4 to 12 weeks): random effects model.

8.2. Analysis.

Comparison 8 Acupuncture versus gastrokinetic agents: sensitivity analyses, fixed‐effect versus random‐effects, Outcome 2 Proportion of people whose symptoms 'improved' (4 to 12 weeks): fixed effect model.

9.2. Analysis.

Comparison 9 Acupuncture plus treatment (gastrokinetics, other drugs or routine care) versus the same treatment alone: sensitivity analyses, valid randomisation and objective emptying measures, Outcome 2 Proportion of people whose symptoms 'improved' (4 to 12 weeks): fixed effects model.

Publication bias

We identified clear asymmetry in the funnel plot for the short‐term response rate of acupuncture compared with gastrokinetic drugs (Figure 5), which may be associated with publication bias. For the same outcome of adjunctive acupuncture compared with various control interventions alone (figure not shown), the figure showed almost symmetric clustered results, other than an outlying value for the comparison of acupuncture with routine diabetic care. An analysis of small study effects for other outcomes was not possible as there were too few included studies.

5.

Funnel plot of comparison: 2 Acupuncture versus gastrokinetic agents, outcome: 2.3 Proportion of people whose symptoms 'improved'(short‐term: 4 to 12 weeks).

Discussion

Summary of main results

Among the 924 studies screened, we identified 32 randomised controlled trials (RCTs), involving 2601 participants. All included studies utilised a fixed acupuncture regimen (formula acupuncture) or a semi‐individualised regimen (formula acupuncture combined with additional individualised acupuncture), with manual or electrical stimulation. Most included trials involved people with diabetic gastroparesis. There were few data on people with gastroparesis of other aetiologies.

When compared with a gastrokinetic drug, acupuncture either as standalone or as complementary intervention showed short‐term benefits in terms of the proportion of participants experiencing improvement, as defined in various ways across the trials (Table 2; Table 4). However, this evidence is of very low certainty. A few studies provided information on the effects of acupuncture compared with other types of treatment, such as non‐gastrokinetic drugs or routine care. In one RCT, people treated with acupuncture showed more rapid recovery of postsurgical gastroparesis compared with those who received routine postsurgical management alone (Mu 2010). These effects are also uncertain as the quality of evidence was very low. Low‐certainty evidence from one RCT showed no difference between real and sham acupuncture at three months in terms of symptoms of gastroparesis, which contradicts the trials reporting short‐term benefits. Long‐term benefits of acupuncture remain unclear because of the absence of reliable data.

Reporting of adverse events was incomplete in most included trials; all adverse events reported were minor, although we could not exclude the possibility of under‐reporting. None of the included trials reported quality‐of‐life outcomes or changes in medication use, which may be important from a patient perspective (Bielefeldt 2009).

Subgroup analyses showed that manual acupuncture stimulation and electrical stimulation yielded similar outcomes in terms of the proportion of participants whose symptoms improved. In our sensitivity analyses, trials with valid randomisation methods or with objective gastric emptying measurements for the diagnosis of gastroparesis showed no differences, possibly due to the more important risk of bias from unclear methods of allocation concealment, outcome assessor blinding, and high risk of performance bias common to all trials. These findings suggest that the type of acupuncture stimulation and objective measurements for the diagnosis of gastroparesis may not influence treatment outcome in any important way, although these findings should be regarded as a hypothesis, at best.

Overall completeness and applicability of evidence

We searched comprehensively for RCTs of acupuncture for symptomatic gastroparesis of any cause, regardless of language of publication. However, most included trials involved people with diabetic gastroparesis and were conducted and published in China. Consequently, these review findings are not different to previous reviews that also relied heavily on the Chinese literature and may not be applicable to people with gastroparesis due to other aetiologies, in different contexts, or in other countries.

We created a funnel plot, which is clearly skewed and demonstrates evidence of either publication bias or small study bias favouring positive outcomes.

We included only trials that provided at least six sessions of acupuncture, assuming that the number of sessions would be an important factor for ensuring clinical validity of acupuncture treatments. However, this assumption may limit the applicability of our findings to only people who can afford to pay for multiple sessions (which may mean higher socioeconomic status) or who are healthy enough to tolerate multiple sessions of acupuncture. Studies providing outcomes measured at less than four weeks were excluded, which may be relevant for people seeking immediate reduction of symptoms.

Acupuncture is considered a safe intervention when performed by qualified practitioners in the general population (MacPherson 2001; Witt 2009). However, the safety of acupuncture in people with gastroparesis, who commonly have chronic underlying conditions, such as diabetes, cannot be assumed based on the review findings. Only RCTs were eligible for our review, and we did not consider data on adverse events from other types of clinical study.

One sham‐controlled trial showed no benefit of acupuncture in terms of short‐term symptom improvement. Penetrating acupuncture on the non‐acupuncture points served as the sham intervention, which may not be inert and exert favourable physiological responses (Lunderberg 2011). In an individual patient data meta‐analysis of acupuncture for chronic pain, use of penetrating‐type sham acupuncture was associated with smaller effect sizes than use of non‐penetrating sham or sham control without needles (MacPherson 2014). We are uncertain whether similar findings would present in people with gastroparesis, unless sufficient reliable data are available.

Quality of the evidence

Most trials comparing acupuncture with active comparators were of low methodological quality, and the meta‐analysis results should be interpreted with caution. We downgraded the certainty of evidence because of the unclear risk of selection bias due to the incomplete reporting of concealment of allocation, the high risk of detection and performance bias due to the unblinding of participants/personnel and outcome assessors, and indirectness from the lack of use of validated outcome measurements. The certainty of the evidence was low in the sham‐controlled trial because of an unclear risk of performance bias.

Potential biases in the review process

A substantial number of the included RCTs did not report the method of randomisation or details of allocation concealment. We could not contact the authors in most cases, due to an absence of contact information, and relied on the reporting within the study. We judged the risk of selection bias in these studies to be unclear. However, limited information on study design may have contributed to the potential misclassification of studies and biased our effect estimates, although the direction and the magnitude of such potential bias is unclear.

Most studies reported the proportion of improved participants as the outcome, making interpretation difficult because the definition of improvement was heterogeneous among the studies. Lack of information on the actual cut‐off values for outcome categorisation and scales used can cause bias due to selective reporting from transformations of an outcome (Page 2013). We rated the risk of attrition bias as low when the total number of participants randomised and those reported in the table of responders were equivalent. The studies did not explicitly report whether any data were missing due to attrition (e.g. from loss to follow‐up or withdrawal from trials) or whether any such missing data were appropriately addressed. Therefore, our assumption that all randomised participants contributed to the analyses as randomised may not be valid and may be considered as a potential source of bias.

The asymmetric funnel plot (Figure 5) suggests the presence of small‐study effects, which may be associated with publication bias. This may have contributed to an overestimation of the effects of acupuncture. Although in our protocol we planned to include both true RCTs and quasi‐RCTs, we excluded quasi‐RCTs from the review due to concerns about the high risk of selection bias. Inclusion of quasi‐RCTs did not alter the magnitude, direction or heterogeneity of the effect estimates (The analysis was not shown.).

Agreements and disagreements with other studies or reviews

We identified seven systematic reviews of acupuncture for diabetic gastroparesis (He 2015; Tang 2010; Wang 2014; Xu 2016; Yang 2013; Zhang 2010; Zhang 2016), one for postoperative gastroparesis (Cheong 2014), and two reviews with systematic inclusion of RCTs to assess the reporting quality of those RCTs using the CONSORT statements (Shi 2012a; Zhang 2012). All but two reviews were published in Chinese databases (Cheong 2014; Yang 2013). The differences from our review include the search date, the population (diabetic or postoperative gastroparesis alone), the searched databases (Chinese databases alone or Chinese and core English databases), the eligibility criteria used for the studies (e.g. the use of concomitant interventions including Tuina or Chinese herbal medicine in addition to acupuncture treatment), and the studies analysed. Some of the studies included in these non‐Cochrane reviews were not eligible for this review due to the use of different inclusion criteria. For example, some trials were excluded from our review because they used acupuncture treatments for less than four weeks, or included concomitant treatments that were not provided equally to the intervention groups. Despite such differences, these non‐Cochrane reviews reflect our findings that the current evidence is very susceptible to risk of bias and that few, if any, trials provide reliable information concerning the effects and safety of acupuncture for symptomatic gastroparesis.

Authors' conclusions

Implications for practice.

Due to the unclear risk of bias and the limited generalisability of the study findings, we can suggest only very limited implications for practice. Our findings are based on studies that mostly included people with diabetic gastroparesis. Therefore, whether the findings of our review are applicable to people with gastroparesis of other aetiologies is unknown. There was very low‐certainty evidence that formula acupuncture alone or in combination with gastrokinetic medication may provide short‐term benefits for the management of symptoms of diabetic gastroparesis. Whether any benefits of acupuncture are sustained beyond four to five weeks is unknown. We have very limited information on the safety of acupuncture for symptomatic gastroparesis; therefore, careful observation of adverse events is required, especially in people with diabetes and other comorbid conditions. In most included studies acupuncture was provided as an intensive course of treatment. We do not know whether such a treatment schedule is acceptable or tolerable to people with gastroparesis in clinical, cultural, or geographical settings outside China. Whether a less intensive course of acupuncture treatment can be effective is unknown. Overall, it is uncertain whether acupuncture improves symptoms of gastroparesis because the quality of evidence is very low.

Implications for research.

Future researchers should completely and transparently report the methods and results of their study to enable readers to understand the study design, conduct, analysis, and interpretation (Turner 2012). The Consolidated Standards of Reporting Trials (CONSORT) statement can be a good reference for transparent reporting (Moher 2010). In our review, most studies reported benefits of acupuncture that are possibly due to the poor methodological quality of the included studies and the possibility of reporting bias. Future studies should incorporate rigorous trial methodologies, especially in the domains of randomisation, concealment of allocation, blinding of (at least) outcome assessors in the design, and the conduct and analysis phases of the trials. Prospective registration of the study protocol in an accessible manner is needed to avoid a risk of bias from selective outcome reporting. Future trials conducted in countries other than China are warranted to improve the limited external generalisability of the current evidence. More studies are needed for postsurgical gastroparesis because the current evidence is mainly based on trials examining patients with diabetic gastroparesis. Researchers should use validated outcome measurements when assessing symptom changes to reduce potential bias from the use of unvalidated outcomes, as seen in the studies included in this review. Patient‐relevant outcomes, including quality of life or daily function, need to be addressed in future trials. Long‐term benefits and adverse events of acupuncture should be assessed. In particular, adverse events of acupuncture should be more thoroughly investigated and transparently reported to improve the current lack of evidence for the safety of acupuncture in this condition. The optimal mode, frequency, and duration of acupuncture to achieve the best patient‐relevant outcomes should be investigated.

Appendices

Appendix 1. Cochrane Neuromuscular Specialised Register (CRS‐W) search strategy

#1 acupuncture or acupoint* or meridian* AND INSEGMENT
 #2 gastroparesis or gastric AND INSEGMENT
 #3 #1 and #2 AND INSEGMENT
 #4 (#1 and #2) AND (INREGISTER)

Appendix 2. Cochrane Central Register of Controlled Trials (CENTRAL) (CRS‐W) search strategy

#1 acupuncture or acupoint* or meridian* AND CENTRAL:TARGET
 #2 gastroparesis or gastric AND CENTRAL:TARGET
 #3 #1 and #2 AND CENTRAL:TARGET

Appendix 3. MEDLINE (OvidSP) search strategy

Database: Ovid MEDLINE(R) Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations, Ovid MEDLINE(R) Daily and Ovid MEDLINE(R) <1946 to Present>
 Search Strategy:
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
 1 randomized controlled trial.pt. (456087)
 2 controlled clinical trial.pt. (92247)
 3 randomized.ab. (405922)
 4 placebo.ab. (187323)
 5 clinical trials as topic.sh. (182997)
 6 randomly.ab. (287027)
 7 trial.ti. (179425)
 8 or/1‐7 (1139504)
 9 exp animals/ not humans.sh. (4435919)
 10 8 not 9 (1049435)
 11 Acupuncture Therapy/ (14629)
 12 acupuncture/ (1491)
 13 (acupuncture or acupoint$ or meridian$).tw. (24455)
 14 or/11‐13 (27734)
 15 Gastroparesis/ (1559)
 16 Gastric Emptying/ (9546)
 17 (gastrop$ or gastric).tw. (232500)
 18 or/15‐17 (234061)
 19 10 and 14 and 18 (95)
 20 remove duplicates from 19 (95)

Appendix 4. Embase (OvidSP) search strategy

Database: Embase <1980 to 2018 Week 13>
 Search Strategy:
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
 1 crossover‐procedure.sh. (54687)
 2 double‐blind procedure.sh. (144930)
 3 single‐blind procedure.sh. (30760)
 4 randomized controlled trial.sh. (490967)
 5 (random* or crossover* or cross over* or placebo* or (doubl* adj blind*) or allocat*).tw,ot. (1472826)
 6 trial.ti. (239478)
 7 controlled clinical trial/ (457082)
 8 or/1‐7 (1774231)
 9 exp animal/ or exp invertebrate/ or animal.hw. or non human/ or nonhuman/ (25277249)
 10 human/ or human cell/ or human tissue/ or normal human/ (19331811)
 11 9 not 10 (5974515)
 12 8 not 11 (1578763)
 13 limit 12 to (conference abstracts or embase) (1333343)
 14 exp acupuncture/ (40275)
 15 (acupuncture or acupoint$ or meridian$).mp. (44184)
 16 14 or 15 (46239)
 17 stomach paresis/ (5553)
 18 stomach emptying/ (17844)
 19 (gastropa$ or gastric or stomach).mp. (488747)
 20 or/17‐19 (488747)
 21 13 and 16 and 20 (199)
 22 remove duplicates from 21 (199)

Appendix 5. CINAHL Plus (EBSCOhost) search strategy

Monday, March 26, 2018 12:24:02 PM
 
 S29 S27 AND S28 3
 S28 EM 20170101‐ Limiters ‐ Exclude MEDLINE records
 Search modes ‐ Boolean/Phrase 324,812
 S27 S26 Limiters ‐ Exclude MEDLINE records
 Search modes ‐ Boolean/Phrase 6
 S26 S18 and S21 and S22 and S25 11
 S25 S23 or S24 175,027
 S24 diabet* 174,341
 S23 (MH "Diabetes Mellitus+") 125,674
 S22 gastropa* or gastric or stomach 28,512
 S21 S19 or S20 14,914
 S20 acupuncture or acupoint* or meridian* 14,580
 S19 (MH "Acupuncture+") 12,838
 S18 S1 or S2 or S3 or S4 or S5 or S6 or S7 or S8 or S9 or S10 or S11 or S12 or S13 or S14 or S15 or S16 or S17 1,071,562
 S17 ABAB design* 106
 S16 TI random* or AB random* 246,110
 S15 ( TI (cross?over or placebo* or control* or factorial or sham? or dummy) ) or ( AB (cross?over or placebo* or control* or factorial or sham? or dummy) ) 491,164
 S14 ( TI (clin* or intervention* or compar* or experiment* or preventive or therapeutic) or AB (clin* or intervention* or compar* or experiment* or preventive or therapeutic) ) and ( TI (trial*) or AB (trial*) ) 187,845
 S13 ( TI (meta?analys* or systematic review*) ) or ( AB (meta?analys* or systematic review*) ) 65,362
 S12 ( TI (single* or doubl* or tripl* or trebl*) or AB (single* or doubl* or tripl* or trebl*) ) and ( TI (blind* or mask*) or AB (blind* or mask*) ) 37,692
 S11 PT ("clinical trial" or "systematic review") 153,244
 S10 (MH "Factorial Design") 1,069
 S9 (MH "Concurrent Prospective Studies") or (MH "Prospective Studies") 341,576
 S8 (MH "Meta Analysis") 31,506
 S7 (MH "Solomon Four‐Group Design") or (MH "Static Group Comparison") 94
 S6 (MH "Quasi‐Experimental Studies") 9,485
 S5 (MH "Placebos") 10,733
 S4 (MH "Double‐Blind Studies") or (MH "Triple‐Blind Studies") 38,141
 S3 (MH "Clinical Trials+") 237,238
 S2 (MH "Crossover Design") 16,012
 S1 (MH "Random Assignment") or (MH "Random Sample") or (MH "Simple Random Sample") or (MH "Stratified Random Sample") or (MH "Systematic Random Sample") Display

Appendix 6. PsycINFO (OvidSP) search strategy

Database: PsycINFO <1806 to March Week 3 2018>
 Search Strategy:
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
 1 (acupuncture or acupoint$ or meridian$).mp. (2744)
 2 gastrointestinal disorders/ (1503)
 3 (gastrop$ or gastric or stomach).mp. (5727)
 4 2 or 3 (6991)
 5 (random$ or rct or cct or factorial$ or crossover$ or cross over$ or cross‐over$ or placebo$ or (doubl$ adj blind$) or (singl$ adj blind$) or assign$ or allocat$ or volunteer$).tw. (318440)
 6 1 and 4 and 5 (3)

Appendix 7. AMED (OvidSP) search strategy

Database: AMED (Allied and Complementary Medicine) <1985 to March 2018>
 Search Strategy:
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
 1 Randomized controlled trials/ (1908)
 2 Random allocation/ (314)
 3 Double blind method/ (643)
 4 Single‐Blind Method/ (90)
 5 exp Clinical Trials/ (3711)
 6 (clin$ adj25 trial$).tw. (6811)
 7 ((singl$ or doubl$ or treb$ or trip$) adj25 (blind$ or mask$ or dummy)).tw. (2856)
 8 placebos/ (583)
 9 placebo$.tw. (3070)
 10 random$.tw. (17240)
 11 research design/ (1935)
 12 Prospective Studies/ (1047)
 13 meta analysis/ (239)
 14 (meta?analys$ or systematic review$).tw. (3364)
 15 control$.tw. (34658)
 16 (multicenter or multicentre).tw. (995)
 17 ((study or studies or design$) adj25 (factorial or prospective or intervention or crossover or cross‐over or quasi‐experiment$)).tw. (12547)
 18 or/1‐17 (53855)
 19 (acupuncture or acupoint$ or meridian$).mp. (10308)
 20 (gastropa$ or gastric or stomach).mp. (1429)
 21 gastrointestinal motility/ (99)
 22 20 or 21 (1505)
 23 18 and 19 and 22 (58)
 24 remove duplicates from 23 (58)

Appendix 8. Trial registries search strategy

Database: the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP) and ClinicalTrials.gov <1985 to January 2018>

Search Strategy:
 ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐

1 acupuncture

2 gastroparesis

3 1 and 2

Appendix 9. STRICTA components of acupuncture treatments in included RCTs

Cheng 2003

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: semi‐individualised based on pattern identification of Chinese medicine

Number of acupuncture points: 19 (assumed bilateral needling)

Names of points used:

• spleen‐stomach deficiency, spleen‐kidney deficiency with cold pattern: ST36, CV12, BL20, LB13, BL23. (moxibustion on GV4 and CV4 when a patient presented with obvious deficiency cold patterns)

• rebellious liver‐qi invading the spleen, stomach‐yang deficiency pattern: ST36, CV12, BL21, BL18, SP6, GB34, LR14

Depth of insertion: not reported

Needle sensation: not reported

Methods of point stimulation: manual stimulation using plain tonification and reduction techniques

Needle retention time: 30 minutes

Needle type: not reported

Number of acupuncture sessions: 28

Frequency/duration of acupuncture treatments: once daily for four weeks (two weeks of acupuncture and a three‐day treatment‐free interval were repeated two times)

Practitioner information: not reported

Ge 2010

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 5

Names of points used: CV12, ST36, PC6 (all bilateral)

Depth of insertion: 20 to 30 mm

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using rapid rotation techniques

Needle retention time: 30 minutes

Needle type: 0.30 × 40 mm sterilised disposable acupuncture needles

Number of acupuncture sessions: 20

Frequency/duration of acupuncture treatments: once daily for four weeks (five days of acupuncture and a two‐day treatment‐free interval were repeated four times)

Practitioner information: not reported

Ge 2015

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 5 (assumed bilateral needling)

Names of points used: CV12 and bilateral ST36 and PC6

Depth of insertion: 30 mm for CV12 and ST36, 25 mm for PC6

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using rapid rotation techniques

Needle retention time: 30 minutes

Needle type: 0.30 × 40 mm disposable filiform needles

Number of acupuncture sessions: 20

Frequency/duration of acupuncture treatments: once daily for four weeks (a two‐day treatment‐free interval after five days of treatment)

Practitioner information: not reported

Ge 2016

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 5 (assumed bilateral needling)

Names of points used: CV12 and bilateral ST36 and PC6

Depth of insertion: 30 mm for CV12 and ST36, 25 mm for PC6

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using rapid rotation techniques

Needle retention time: 30 minutes

Needle type: 0.30 × 40 mm disposable filiform needles

Number of acupuncture sessions: 20

Frequency/duration of acupuncture treatments: once daily for four weeks (a two‐day treatment‐free interval after five‐day of treatment)

Practitioner information: not reported

He 2006

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 17

Names of points used: CV13, CV12, CV10, BL18, BL20, BL21, ST36, PC6, SP4, SP6 (all bilateral)

Depth of insertion: not reported

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using rapid insertion, rotation, and lifting/thrusting techniques

Needle retention time: 20 to 30 minutes

Needle type: 0.32 × 40 mm stainless acupuncture needles

Number of acupuncture sessions: 28

Frequency of acupuncture treatments: once daily for four weeks

Practitioner information: not reported

Kong 2009

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 9

Names of points used: CV12, PC6, ST25, ST36, SP4 (all bilateral)

Depth of insertion: 0.8 to 1.2 cun (20 to 30 mm)

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation

Needle retention time: 30 minutes

Needle type: not reported

Number of acupuncture sessions: 28

Frequency/duration of acupuncture treatments: once daily for four weeks

Practitioner information: not reported

Li 2013

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 5 (assumed bilateral needling)

Names of points used:

• (1) CV13, CV12, ST37, ST36, PC6, ST36

• (2) BL20, BL21, Chest 11,12 jiaji (Ex‐B2), SP6

• They were crossed (1) and (2) points on once a day.

Depth of insertion: not reported

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation following electrical stimulation

Needle retention time: 20 min

Needle type: not reported

Number of acupuncture sessions: 20

Frequency/duration of acupuncture treatments: once daily for five days, total four weeks

Practitioner information: not reported

Lin 2013

Style of acupuncture: traditional Chinese medicine abdominal acupuncture

Extent to which treatment was varied: fixed

Number of acupuncture points: 5 (assumed bilateral needling)

Names of points used: ST36, PC6, CV4

Depth of insertion: 0.5˜1.2寸

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using rapid rotation techniques

Needle retention time: 20 minutes

Needle type: 0.30 × 40 mm sterilised disposable acupuncture needles

Number of acupuncture sessions: 24

Frequency/duration of acupuncture treatments: once daily for six days, total four weeks

Practitioner information: not reported

Liu 2001

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 9 (assumed bilateral needling)

Names of points used: CV12, LI10, SP6, ST36 (for acupuncture), GV4, CV4 (for moxibustion)

Depth of insertion: not reported

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using tonification techniques

Needle retention time: 30 minutes

Needle type: not reported

Number of acupuncture sessions: not reported (28 sessions when assuming the session was provided once daily)

Frequency/duration of acupuncture treatments: frequency not reported (assuming once daily) for four weeks

Practitioner information: not reported

Liu 2009

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 4 to 5 points among 8 points

Names of points used: ST36, PC6, CV12, BL21, BL20, CV6, CV4 and SP4

Depth of insertion: not reported

Needle sensation: not reported

Methods of point stimulation: manual stimulation using plain tonification and reduction techniques

Needle retention time: 20 to 30 minutes

Needle type: 0.35 × 25 to 40 mm filiform needles

Number of acupuncture sessions: 40

Frequency/duration of acupuncture treatments: once daily for 10 days as a treatment session and a total of four treatment sessions

Practitioner information: not reported

Liu 2011

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 11 (assumed bilateral needling)

Names of points used: CV12, ST36, PC6, ST25, ST37, ST39

Depth of insertion: not reported

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using plain tonification/reduction techniques following electrical stimulation

Needle retention time: 20 minutes

Needle type: not reported

Number of acupuncture sessions: 28

Frequency/duration of acupuncture treatments: once daily for four weeks

Practitioner information: not reported

Liu 2012

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 9 (assumed bilateral needling)

Names of points used: CV12, BL20, LI11, SP6, ST36

Depth of insertion: not reported

Needle sensation: not reported

Methods of point stimulation: manual stimulation using plain tonification/reduction techniques

Needle retention time: 20 minutes

Needle type: not reported

Number of acupuncture sessions: 28

Frequency/duration of acupuncture treatments: once daily for four weeks

Practitioner information: not reported

Mo 2005

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 8

Names of points used: CV12, PC6, ST36, SP4 (all bilateral)

Depth of insertion: 1 to 1.5 cun

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using rotation and/or lifting/thrusting tonification/reduction techniques

Needle retention time: 30 minutes

Needle type: not reported

Number of acupuncture sessions: 24 (reported six working days in a week)

Frequency/duration of acupuncture treatments: once daily for four weeks (six days of acupuncture and a one‐day treatment‐free interval repeated four times)

Practitioner information: not reported

Mu 2010

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 22 (assumed bilateral)

Names of points used: ST21, ST25, SP15, ST30, ST36, ST37, ST39, ST44, PC6, GB34, SP6

Depth of insertion: not reported

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation followed by electrical stimulation

Needle retention time: not reported

Needle type: 1.5 to 2‐cun piriform acupuncture needles

Number of acupuncture sessions: unclear (at least 10 with an assumption of five working days in a week)

Frequency/duration of acupuncture treatments: once daily until the recovery of postsurgical gastroparesis

Practitioner information: not reported

Qi 2013

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 10 (assumed bilateral needling)

Names of points used: CV10, CV11, CV12, CV13, ST25, ST36, PC6

Depth of insertion: 0.5˜1.2寸

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using rapid rotation techniques

Needle retention time: 30 minutes

Needle type: 0.30 × 40 mm sterilised disposable acupuncture needles

Number of acupuncture sessions: 24

Frequency/duration of acupuncture treatments: once daily for seven days, total eight weeks (seven days of acupuncture and a two‐day treatment‐free interval repeated for eight weeks)

Practitioner information: not reported

Shi 2012

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 3 (assumed bilateral needling)

Names of points used: CV12 and bilateral BL21

Depth of insertion: 0.5 ‐ 1 Cun for CV12 and 0.5 to 0.8 Cun for BL21

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using plain tonification and reduction techniques

Needle retention time: 30 minutes

Needle type: 0.30 × 25 to 50 mm (1 Cun to 1.5 Cun) disposable filiform needles

Number of acupuncture sessions: 10

Frequency/duration of acupuncture treatments: once daily for two weeks (a two‐day treatment‐free interval after five days of treatment)

Practitioner information: acupuncture and moxibustion graduate professional with qualification certificate of TCM practitioner

Wang 2003

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: semi‐fixed

Number of acupuncture points: 21 (assumed bilateral)

Names of points used:

• CV12, ST36, PC6, SP4, BL20, BL21 as common points

• liver‐stomach disharmony pattern: additional use of LI11, GB34, LR3

• spleen‐stomach deficiency: additional use of CV6, CV4, SP6 (moxibustion on CV6, CV4, CV12, ST36)

Depth of insertion: not reported

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using plain tonification/reduction techniques

Needle retention time: 30 minutes

Needle type: not reported

Number of acupuncture sessions: 20

Frequency/duration of acupuncture treatments: once daily for 20 days (10 days of acupuncture and five days of treatment‐free intervals were repeated twice)

Practitioner information: not reported

Wang 2007

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 19 (assumed bilateral needling)

Names of points used: BL21, CV12, BL20, LR13, BL23, BL18, LR14, GB25, ST25, ST36

Depth of insertion: 0.5 – 0.8 CUN for BL21, BL23, BL18, LR13, LR14 and GB25, 1.0 to 1.5 CUN for CV12, ST25

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation

Needle retention time: 30 minutes

Needle type: 0.30 × 40 mm sterile acupuncture needles

Number of acupuncture sessions: 30

Frequency/duration of acupuncture treatments: once daily for 10 days as a treatment session and a total of three treatment sessions

Practitioner information: not reported

Wang 2009

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 11 (assumed bilateral)

Names of points used: CV12, ST36, SP6, ST34, LR3, Ashi points (1 cun lateral from CV 12)

Depth of insertion: not reported

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using plain tonification/reduction techniques

Needle retention time: 30 to 40 minutes

Needle type: not reported

Number of acupuncture sessions: 26

Frequency/duration of acupuncture treatments: once daily for six days over a 30‐day period (six days of acupuncture and a one‐day treatment‐free interval repeated for 30 days)

Practitioner information: not reported

Wu 2015

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 7 (assumed bilateral needling)

Names of points used:

• Zongqi therapy Ⅰ: PC6, CV14, CV12, CV10 and ST21

• Zongqi therapy Ⅱ: PC6, CV12, ST23 and ST19

Depth of insertion: check the ‘Methods of point stimulation’ section

Needle sensation: check the ‘Methods of point stimulation’ section

Methods of point stimulation: Using 0. 30 × 50 mm needles, acupuncture needle tip obliquely stabbed at 7 0 ° ˜ 8 0 ° angle, with the lifting/thrusting techniques to the local warm feeling; or using 0. 30 mm × 75 mm acupuncture needles, penetrating the abdominal wall straight piercing to the anterior wall of the stomach, only a deep stab without retention and manual stimulation of needles; or using 0. 30 mm × 40 mm needles for PC6 acupuncture, a sense of ache to the local soreness, should not be transmitted to the distal part

Needle retention time: 30 minutes

Needle type: 0. 30 × 50 mm, 0. 30 mm × 75 mm and 0. 30 mm × 40 mm filiform needles

Number of acupuncture sessions: 24

Frequency/duration of acupuncture treatments: once daily for 12 days (12 days of acupuncture and a one‐day treatment‐free interval were repeated two times)

Practitioner information: not reported

Xu 2012

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 11

Names of points used:

• type 1: CV12, ST25, ST36, LR3 (all bilateral)

• type 2: BL20, BL21, Weiwanxiashu (extraordinary points; located at 1.5 cun bilateral points on the spinous process of the 8th thoracic vertebrae) (all bilateral)

• (two types of acupuncture regimens were alternately used each day)

Depth of insertion: not reported

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation following electrical stimulation

Needle retention time: 30 minutes

Needle type: not reported

Number of acupuncture sessions: 15

Frequency/duration of acupuncture treatments: once daily for five days over a three‐week period (five days of acupuncture and a two‐day treatment‐free interval repeated for three weeks)

Practitioner information: not reported

Xue 2014

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: semi‐individualised based on pattern identification of Chinese medicine

Number of acupuncture points: 13 – 17 (assumed bilateral needling)

Names of points used:

• essential points: CV12, bilateral ST36, ST37, ST39, PC6, SP4 and ST25

• additional points: LR3 (for Liver qi invading the stomach), ST34 (for Phlegm‐dampness center obstruction), SP6 (for Stomach yin deficiency) or BL20/BL21 (for Spleen‐stomach deficiency)

Depth of insertion: 1.0 to 1.5 Cun for ST36, ST37, ST39, ST25, SP6, BL20 and BL21; 0.5 to 1.0 Cun for PC6, SP4 and LR3

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using plain tonification and reduction techniques

Needle retention time: 30 min

Needle type: 0.30 × 40 mm and 0.30 × 100 mm disposable filiform needles

Number of acupuncture sessions: 30

Frequency/duration of acupuncture treatments: once daily for 10 days as a treatment session and a total of 3 treatment sessions

Practitioner information: not reported

Xu 2016

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: semi‐individualised based on pattern identification of Chinese medicine

Number of acupuncture points: over 15 points (assumed bilateral needling)

Names of points used:

• common points: PC6, ST36, ST25 (above points for bilateral needling), CV12, SP6, KI3, BL20 and BL21

• severe phlegm‐dampness: add LR2, LR14, BL20 and BL21

• severe qi stagnation: add LI4 and LI6

• severe liver gi depression: add LR3

Depth of insertion: not reported

Needle sensation: de‐qi

Methods of point stimulation: manual stimulation using plain tonification and reduction techniques

Needle retention time: 30 minutes

Needle type: 0.30 × 40mm filiform needles

Number of acupuncture sessions: 45

Frequency/duration of acupuncture treatments: once daily for 15 days (15 days of acupuncture were repeated three times)

Practitioner information: not reported

Yang 2013

Style of acupuncture: traditional Chinese medicine abdominal acupuncture

Extent to which treatment was varied: fixed

Number of acupuncture points: 19 (assumed bilateral needling)

Names of points used: CV12, CV10, CV6, CV4, ST25,ST24, SP15, KI16, KI13

Depth of insertion: not reported

Needle sensation: not reported

Methods of point stimulation: not reported

Needle retention time: 30 minutes

Needle type: 0.22 × 40 mm sterilised disposable acupuncture needles

Number of acupuncture sessions: 30

Frequency/duration of acupuncture treatments: once daily for 30 days

Practitioner information: not reported

Yuan 2004

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 15 (assumed bilateral needling)

Names of points used: CV13, CV12, CV10, BL13, BL20, BL17, ST36, SP6, SP9

Depth of insertion: not reported

Needle sensation: not reported

Methods of point stimulation: manual stimulation using plain tonification and reduction techniques

Needle retention time: 15 minutes

Needle type: not reported

Number of acupuncture sessions: 30

Frequency/duration of acupuncture treatments: once daily for 30 days

Practitioner information: not reported

Zeng 2006

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 5 (assumed bilateral needling)

Names of points used: ST36, CV12, SP6

Depth of insertion: not reported

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using rapid rotation techniques

Needle retention time: not reported

Needle type: not reported

Number of acupuncture sessions: 28

Frequency/duration of acupuncture treatments: once daily for two weeks, two weeks constituting one course, total two courses

Practitioner information: not reported

Zeng 2008

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 7 (assumed bilateral needling)

Names of points used: CV12, ST36, PC6, SP6

Depth of insertion: not reported

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using rapid insertion, rotation, and lifting/thrusting techniques

Needle retention time: 30 minutes

Needle type: 0.25 × 40 mm sterilised disposable acupuncture needles

Number of acupuncture sessions: 14

Frequency/duration of acupuncture treatments: once daily for two weeks

Practitioner information: not reported

Zhang 2007

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 25 (assumed bilateral needling)

Names of points used: LI11, LI4, CV12, ST36, ST40, SP9, SP6, SP10, SP8, LR3, SP4, TE6, ST25

Depth of insertion: not reported

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using plain tonification/reduction techniques

Needle retention time: 30 minutes

Needle type: 0.30 × 50 mm and 0.30 × 60 mm sterilised disposable acupuncture needles

Number of acupuncture sessions: 60

Frequency/duration of acupuncture treatments: twice each day, 10 days constituting one course, with an interval of two days, total three courses

Practitioner information: not reported

Zhang 2013

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 5 (assumed bilateral needling)

Names of points used: ST36, PC6, CV12

Depth of insertion: 0.5˜1.2 cun (寸)

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using plain tonification/reduction techniques

Needle retention time: 20 minutes

Needle type: 0.25 × 40 mm sterilised disposable acupuncture needles

Number of acupuncture sessions: 15

Frequency/duration of acupuncture treatments: once daily for five days over a three‐week period

Practitioner information: not reported

Zhang 2014

Style of acupuncture: traditional Chinese medicine abdominal acupuncture

Extent to which treatment was varied: fixed

Number of acupuncture points: 9 (assumed bilateral needling)

Names of points used: CV4, ST36, ST25, ST21, ST37

Depth of insertion: not reported

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation following electrical stimulation

Needle retention time: 20 minutes

Needle type: 0.30 × 40 mm sterilised disposable acupuncture needles

Number of acupuncture sessions: 10

Frequency/duration of acupuncture treatments: once daily for five days over a two‐week period (five days of acupuncture and a two‐day treatment‐free interval repeated for three weeks)

Practitioner information: not reported

Zhao 2004

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: fixed

Number of acupuncture points: 7 (assumed bilateral needling)

Names of points used: PC6, CV12, ST37, ST36, BL18, BL20, BL21

Depth of insertion: 1˜3 cun (寸)

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using rapid insertion, rotation, and lifting/thrusting techniques

Needle retention time: 30 minutes

Needle type: not reported

Number of acupuncture sessions: 60

Frequency/duration of acupuncture treatments: twice daily for 30 days

Practitioner information: not reported

Zhao 2011

Style of acupuncture: traditional Chinese medicine

Extent to which treatment was varied: semi‐individualised based on pattern identification of Chinese medicine

Number of acupuncture points: 15 (assumed bilateral needling)

Names of points used:

• fixed: CV10, CV11, CV12, CV13, ST25

• individualised based on with points: LI11, LI4, ST40, ST36, SP9, SP6, SP10, SP8

Depth of insertion: not reported

Needle sensation: de‐qi sensation

Methods of point stimulation: manual stimulation using rapid rotation techniques and following electrical stimulation

Needle retention time: 30 minutes

Needle type: not reported

Number of acupuncture sessions: 30

Frequency/duration of acupuncture treatments: once daily for 10 days, 10 days constituting one course, total three courses

Practitioner information: not reported

Data and analyses

Comparison 1. Acupuncture versus sham acupuncture.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Improvement in symptoms of gastroparesis ‐ overall symptom scores (short‐term: 4 to 12 weeks)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

Comparison 2. Acupuncture versus gastrokinetic agents.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Improvement in symptoms of gastroparesis ‐ overall symptom scores (short‐term: 4 to 12 weeks)	5		Mean Difference (IV, Random, 95% CI)	Totals not selected
1.1 Acupuncture versus domperidone	2		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
1.2 Acupuncture versus mosapride	3		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
2 Improvement in symptoms of gastroparesis ‐ individual symptom scores (acupuncture versus mosapride)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
2.1 Bloating	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
2.2 Stomach pain	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
2.3 Loss of appetite	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
2.4 Belching	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
2.5 Abnormal defecation	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
3 Proportion of people whose symptoms 'improved'(short‐term: 4 to 12 weeks)	12	963	Risk Ratio (M‐H, Random, 95% CI)	1.25 [1.17, 1.33]
3.1 Acupuncture versus cisapride	1	90	Risk Ratio (M‐H, Random, 95% CI)	1.16 [1.00, 1.35]
3.2 Acupuncture versus domperidone	7	567	Risk Ratio (M‐H, Random, 95% CI)	1.31 [1.20, 1.43]
3.3 Acupuncture versus mosapride	4	306	Risk Ratio (M‐H, Random, 95% CI)	1.18 [1.02, 1.35]
4 Proportion of people whose symptoms 'improved' (long‐term: more than 12 weeks) (acupuncture versus mosapride)	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
5 Changes in the rate of gastric emptying ‐ response rate (acupuncture versus mosapride)	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
6 Changes in the rate of gastric emptying‐ gastric emptying‐barium test	4		Mean Difference (IV, Random, 95% CI)	Totals not selected
6.1 Acupuncture versus domperidone	4		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
7 Effects on glycaemic control ‐ control of diabetes‐HbA1c (%) (acupuncture versus mosapride)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
7.1 HbA1c (short‐term)	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
8 Effects on glycaemic control ‐ control of diabetes‐FBS (mg/dL) (acupuncture versus mosapride)	2		Mean Difference (IV, Random, 95% CI)	Totals not selected
8.1 Fasting blood sugar (short term)	2		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]

Comparison 3. Acupuncture versus non‐gastrokinetic agents.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Proportion of people whose symptoms 'improved' ‐ individual symptoms (acupuncture versus cimetidine)	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
1.1 Early satiety	1		Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
1.2 Nausea and vomiting	1		Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
1.3 epigastric pain	1		Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
1.4 postprandial discomfort	1		Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
1.5 abdominal distension	1		Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
1.6 diarrhoea	1		Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
1.7 diarrhoea‐constipation	1		Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]
2 Proportion of people whose symptoms 'improved'	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
2.1 Acupuncture versus cimetidine	1		Risk Ratio (M‐H, Random, 95% CI)	0.0 [0.0, 0.0]

Comparison 4. Acupuncture plus gastrokinetic agents versus gastrokinetic agents alone.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Improvement in symptoms of gastroparesis ‐ overall symptom scores (short‐term: 4 to 12 weeks) (gastrokinetic: domperidone)	2	132	Mean Difference (IV, Random, 95% CI)	‐1.96 [‐2.42, ‐1.50]
2 Proportion of people whose symptoms 'improved' (short‐term: 4 to 12 weeks)	13	1107	Risk Ratio (M‐H, Random, 95% CI)	1.21 [1.15, 1.27]
2.1 Acupuncture plus cisapride versus cisapride alone	2	110	Risk Ratio (M‐H, Random, 95% CI)	1.14 [0.97, 1.33]
2.2 Acupuncture plus domperidone versus domperidone alone	11	997	Risk Ratio (M‐H, Random, 95% CI)	1.22 [1.15, 1.29]
3 Proportion of people whose symptoms 'improved' (long‐term: more than 12 weeks)	3	237	Risk Ratio (M‐H, Random, 95% CI)	1.41 [1.19, 1.66]
3.1 Acupuncture plus domperidone versus domperidone alone	1	72	Risk Ratio (M‐H, Random, 95% CI)	1.42 [1.11, 1.81]
3.2 Acupuncture versus mosapride versus mosapride alone	2	165	Risk Ratio (M‐H, Random, 95% CI)	1.48 [1.00, 2.18]
4 Effects on glycaemic control ‐ control of diabetes‐HbA1c (%): long term) (gastrokinetic: mosapride)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
5 Effects on glycaemic control ‐ control of diabetes‐FBS) (mg/dL) (long‐term) (gastrokinetic: domperidone (plus vitamin B6))	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
6 Effects on glycaemic control ‐ control of diabetes‐hypoglycaemic events ‐ mean frequency/day (gastrokinetic: mosapride)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
6.1 Short term	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
6.2 Long term	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]

Comparison 5. Acupuncture plus omeprazole versus omeprazole alone.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Individual symptom scores (short term: 4 to 12 weeks)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
1.1 Bloating	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
1.2 Stomach pain	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
1.3 Belching	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
1.4 Abnormal defecation	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
1.5 Decreased oral intake	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
2 Proportion of people whose symptoms 'improved' (short‐term: 4 to 12 weeks)	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected
3 Changes in the rate of gastric emptying ‐ barium test (short term: 4 to 12 weeks)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
4 Effects on glycaemic control ‐ control of diabetes‐FBS) (mg/dL) (short‐term: 4 to 12 weeks)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
4.1 Fasting blood sugar (short term: 4 to 12 weeks)	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]

Comparison 6. Acupuncture plus routine care/gycaemic control versus routine care/glycaemic control alone.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Proportion of people whose symptoms 'improved' (short‐term: 4 to 12 weeks)	3	217	Risk Ratio (M‐H, Random, 95% CI)	1.47 [1.06, 2.04]
2 Changes in the rate of gastric emptying ‐ electrogastrography (short term: 4 to 12 weeks)	1		Mean Difference (IV, Random, 95% CI)	Totals not selected
2.1 Frequency measured at gastric antrum (cycles per minute)	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
2.2 Frequency measured at gastric body (cycles per minute)	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
2.3 Amplitude measured at gastric antrum (μV)	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
2.4 Amplitude measured at gastric body (μV)	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]

Comparison 7. Acupuncture versus gastrokinetic agents: subgroup analysis, manual acupuncture versus electroacupuncture.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Proportion of people whose symptoms 'improved' (4 to 12 weeks): acupuncture versus gastrokinetics	12	963	Risk Ratio (M‐H, Random, 95% CI)	1.25 [1.17, 1.33]
1.1 Manual stimulation	11	873	Risk Ratio (M‐H, Random, 95% CI)	1.23 [1.15, 1.32]
1.2 Electrical stimulation	1	90	Risk Ratio (M‐H, Random, 95% CI)	1.41 [1.12, 1.79]
2 Proportion of people whose symptoms 'improved' (4 to 12 weeks): acupuncture combined with gastrokinetics versus gastrokinetics alone	13		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
2.1 Manual stimulation	8	769	Risk Ratio (M‐H, Random, 95% CI)	1.23 [1.15, 1.31]
2.2 Electrical stimulation	5	338	Risk Ratio (M‐H, Random, 95% CI)	1.18 [1.08, 1.28]

7.2. Analysis.

Comparison 7 Acupuncture versus gastrokinetic agents: subgroup analysis, manual acupuncture versus electroacupuncture, Outcome 2 Proportion of people whose symptoms 'improved' (4 to 12 weeks): acupuncture combined with gastrokinetics versus gastrokinetics alone.

Comparison 8. Acupuncture versus gastrokinetic agents: sensitivity analyses, fixed‐effect versus random‐effects.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Proportion of people whose symptoms 'improved' (4 to 12 weeks): random effects model	12		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.1 All studies	12	963	Risk Ratio (M‐H, Random, 95% CI)	1.25 [1.17, 1.33]
1.2 Valid random only	5	391	Risk Ratio (M‐H, Random, 95% CI)	1.27 [1.11, 1.45]
1.3 Objective emptying measures	7	607	Risk Ratio (M‐H, Random, 95% CI)	1.22 [1.10, 1.36]
2 Proportion of people whose symptoms 'improved' (4 to 12 weeks): fixed effect model	12		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
2.1 All studies	12	963	Risk Ratio (M‐H, Fixed, 95% CI)	1.27 [1.19, 1.36]
2.2 Valid random only	5	391	Risk Ratio (M‐H, Fixed, 95% CI)	1.33 [1.19, 1.49]
2.3 Objective emptying measures	7	607	Risk Ratio (M‐H, Fixed, 95% CI)	1.25 [1.14, 1.36]

Comparison 9. Acupuncture plus treatment (gastrokinetics, other drugs or routine care) versus the same treatment alone: sensitivity analyses, valid randomisation and objective emptying measures.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Proportion of people whose symptoms 'improved' (4 to 12 weeks): random effects model	17		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.1 All studies	17	1404	Risk Ratio (M‐H, Random, 95% CI)	1.22 [1.16, 1.28]
1.2 Valid random only	9	738	Risk Ratio (M‐H, Random, 95% CI)	1.20 [1.12, 1.28]
1.3 Objective emptying measures	14	1064	Risk Ratio (M‐H, Random, 95% CI)	1.24 [1.17, 1.31]
2 Proportion of people whose symptoms 'improved' (4 to 12 weeks): fixed effects model	17		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
2.1 All studies	17	1404	Risk Ratio (M‐H, Fixed, 95% CI)	1.26 [1.20, 1.33]
2.2 Valid random only	9	738	Risk Ratio (M‐H, Fixed, 95% CI)	1.22 [1.13, 1.31]
2.3 Objective emptying measures	14	1064	Risk Ratio (M‐H, Fixed, 95% CI)	1.29 [1.21, 1.37]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Cheng 2003.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 6 months
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 53 (12) in the treatment group; 51 (15) in the control group Gender (M/F): 33/39 Major eligibility criteria: 1) type 2 diabetes defined by WHO 1985 criteria, 2) a diabetes diagnosis of at least 5 years, 3) cardinal symptoms, 4) delayed gastric emptying confirmed by a barium meal study (more than 6 hours of gastric emptying), and 5) no evidence of underlying gastric diseases confirmed by upper GI endoscopy or radiography
Interventions	Manual acupuncture once daily plus domperidone 30 mg and vitamin B6 60 mg daily (n = 36) versus the same medication alone daily (n = 36) for 4 weeks
Outcomes	Overall symptom scores Proportion of people whose symptoms 'improved' Fasting blood glucose Measurements: at 4 weeks and 6 months
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (3‐point scales) (traditional Chinese medicine (TCM) clinical research guidelines) Markedly effective: complete improvement of clinical symptoms, normalised gastric motility or less than 4 hours of gastric emptying time measured by an upper gastrointestinal barium radiographic study Effective: significant improvement of clinical symptoms, improved gastric motility, or 4 to 6 hours of gastric emptying time measured by upper gastrointestinal barium radiographic study Ineffective: no significant improvement of clinical symptoms or no change in gastric emptying time measured by an upper gastrointestinal barium radiographic study Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A random number card was used.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline

Ge 2010.

Methods	Study design: parallel‐group RCT Duration: January 2008 to June 2009 Follow‐up: 4 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (range)): 58 (40 to 61) in the acupuncture group; 58.5 (42 to 60) in the control group Gender (M/F): 33/27 Major eligibility criteria: 1) type 2 diabetes, 2) a diabetes duration of at least 5 years, 3) cardinal symptoms, 4) delayed gastric emptying measured by a barium meal study, and 5) no evidence of underlying gastric diseases confirmed by endoscopy
Interventions	Manual acupuncture once daily (n = 30) versus domperidone 10 mg three times daily (n = 30) for 4 weeks
Outcomes	Proportion of people whose symptoms 'improved' Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (4‐point scales) Cardinal and related symptoms (bloating, nausea/vomiting, anorexia, early satiety, and fatigue) were scored as 3 (severe), 2 (moderate), 1 (mild), or 0 (none). The sum of each symptom score consisted of the overall symptom scores. The percentage change in overall symptom scores before and after treatment was used to determine Ratings of symptoms 'improved'. Recovery: percentage change in overall symptom scores equal to or greater than 95% Markedly effective: percentage change in overall symptom scores equal to or greater than 70% and less than 95% Effective: percentage change in overall symptom scores equal to or greater than 30% and less than 70% Ineffective: percentage change in overall symptom scores less than 30% Ratings of symptoms 'improved' were dichotomised into 'improved' (combining recovery, markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A random number table was used.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed
Selective reporting (reporting bias)	Unclear risk	Study protocol was not available
Other bias	Unclear risk	Delayed gastric emptying was objectively measured at baseline. The study did not report any criteria for diabetes (bias related to misclassification of diabetic gastroparesis).

Ge 2015.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 4 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 63.9 (5.7) in the treatment group; 64.2 (5.3) in the control group Gender (M/F): 45/35 Major eligibility criteria: 1) a type 2 diabetes duration of at least 5 years, 2) cardinal symptoms, 3) delayed gastric emptying confirmed by radiopaque markers in the stomach 4 hours after eating, 4 no evidence of underlying gastric diseases confirmed by upper gastrointestinal (GI) endoscopy or radiography, and 5) a spleen deficiency pattern in the TCM diagnosis
Interventions	Manual acupuncture five times a week (n = 40) versus domperidone 10 mg three times a day (n = 40) for 4 weeks
Outcomes	Response rate Overall symptom scores Gastric emptying test (barium X‐ray test) Measurements: at 4 weeks
Funding	Funding source: public funding
Conflicts of interest	No information
Notes	Criteria for symptoms 'improved' (4‐point scales) Cardinal and related symptoms (bloating, nausea/vomiting, anorexia, early satiety, and fatigue) were scored as 3 (severe), 2 (moderate), 1 (mild), or 0 (none). The sum of each symptom score consisted of the overall symptom scores. The percentage change in overall symptom scores before and after treatment was used to determine ratings of symptoms 'improved'. Recovery: 75% to 100% reduction of overall symptom scores Markedly effective: 50% to 75% reduction of overall symptom scores Effective: 25% to 50% reduction of overall symptom scores equal to or greater than 30% and less than 70% Ineffective: less than 25% reduction of overall symptom scores Ratings of symptoms 'improved' were dichotomised into 'improved' (combining recovery, markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Unlikely to be blinded (open trial)
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Unclear risk	Delayed gastric emptying was objectively measured at baseline and post‐treatment assessment. The study did not report any criteria for diabetes (bias related to misclassification of diabetic gastroparesis)

Ge 2016.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 4 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 58.0 (7.5) in the treatment group; 58.5 (5.7) in the control group Gender (M/F): 53/47 Major eligibility criteria: 1) aged 30 years or older, 2) type 2 diabetes based on the diagnostic criteria in the 2010 Guidelines for Prevention and Treatment of Type‐2 Diabetes by Chinese Medical Association, 3) a diabetes diagnosis of at least 5 years, 4) cardinal symptoms, 5) delayed gastric emptying confirmed by radiopaque markers in the stomach 4 hours after eating, 6) no evidence of underlying gastric diseases confirmed by upper GI endoscopy or radiography, and 7) a spleen deficiency pattern in the TCM diagnosis
Interventions	Manual acupuncture five times a week (n = 50) versus domperidone 10 mg three times a day (n = 50) for 4 weeks
Outcomes	Overall symptom scores Gastric emptying test (barium X‐ray test) Plasma motilin and serum gastrin concentration Proportion of people whose symptoms 'improved' Measurements: at 4 weeks
Funding	Funding source: public funding (2011 Science and Technology Planning Project of Wenzhou City, Zhejiang Province)
Conflicts of interest	Authors declared no conflicts of interest
Notes	Criteria for symptoms 'improved' (4‐point scales) (The Guiding Principles for Clinical Study of New Chinese Medicines) Cardinal and related symptoms (bloating, nausea/vomiting, anorexia, early satiety, and fatigue) were scored as 3 (severe), 2 (moderate), 1 (mild), or 0 (none). The sum of each symptom score consisted of the overall symptom scores. The percentage change in overall symptom scores before and after treatment was used to determine the ratings of symptoms 'improved' . Recovery: 75% to 100% reduction of overall symptom scores Markedly effective: 50% to 75% reduction of overall symptom scores Effective: 25% to 50% reduction of overall symptom scores Ineffective: less than 25% reduction of overall symptom scores Ratings of symptoms 'improved' were dichotomised into 'improved' (combining recovery, markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The method of random sequence generation was not reported.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Unlikely to be blinded (open trial)
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline and as a treatment outcome.

He 2006.

Methods	Study design: parallel‐group RCT Duration: June 2001 to June 2004 Follow‐up: 4 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (range)): 59 (45 to 73) in the acupuncture group; 58.5 (43 to 74) in the control group Gender (M/F): 28/32 Major eligibility criteria: 1) diabetes defined by 1997 American Diabetes Association guidelines, 2) a diabetes diagnosis of at least 5 years, 3) cardinal symptoms, 4) delayed gastric emptying confirmed by a barium meal study (more than 6 hours of gastric emptying), 5) reduced gastric motility measured by electrogastrography, 6) no evidence of underlying gastric diseases confirmed by upper GI endoscopy, and 7) spleen‐stomach deficiency syndrome defined by traditional Chinese medicine theory
Interventions	Manual acupuncture once daily plus cisapride 5 mg three times daily (n = 30) versus the same medication alone (n = 30) for 4 weeks
Outcomes	Proportion of people whose symptoms 'improved' Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (3‐point scales) (TCM clinical research guidelines) Markedly effective: complete improvement of clinical symptoms, normalised gastric motility, or less than 4 hours of gastric emptying time measured by an upper gastrointestinal barium radiographic study. Effective: significant improvement of clinical symptoms, improved gastric motility, or 4 to 6 hours of gastric emptying time measured by an upper gastrointestinal barium radiographic study. Ineffective: no significant improvement of clinical symptoms, no change of gastric emptying time measured by an upper gastrointestinal barium radiographic study. Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A random number table was used.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline.

Kong 2009.

Methods	Study design: parallel‐group RCT Duration: January 2004 to November 2008 Follow‐up: 4 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 66.93 (7.77) in the acupuncture group; 65.07 (8.05) in the control group Gender (M/F): 48/42 Major eligibility criteria: 1) type 2 diabetes defined by 1997 American Diabetes Association guidelines, 2) a diabetes duration of at least 5 years, 3) cardinal symptoms, 4) delayed gastric emptying and reduced gastric motility measured by the barium meal study, and 5) no evidence of underlying gastric diseases confirmed by endoscopy or ultrasonography
Interventions	Manual acupuncture once daily (n = 45) versus cisapride 10 mg three times daily (n = 45) for 4 weeks
Outcomes	Proportion of people whose symptoms 'improved' Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (3‐point scales) Markedly effective: complete improvement of clinical symptoms, normalised gastric emptying time (no reporting on the measurement method of gastric emptying) Effective: improvement of clinical symptoms and improved but still delayed gastric emptying time Ineffective: no improvement of clinical symptoms and no change in gastric emptying time Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A random number table was used.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	Study protocol was not available
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline.

Li 2013.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 4 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 51 (11) in the acupuncture group and 50 (12) in the control group Gender (M/F): 31/37 Major eligibility criteria: 1) type 2 diabetes and diabetic gastroparesis defined by the 1999 WHO Diagnostic Criteria, 2) a diabetes diagnosis of at least 5 years, 2) cardinal symptoms, 3) delayed gastric emptying confirmed by a barium meal study (more than 6 hours of gastric emptying), 4) decreased gastric motility, and 5) no evidence of underlying gastric diseases confirmed by upper GI endoscopy and ultrasound
Interventions	Manual and electroacupuncture five times a week (n = 35) versus domperidone 10 mg three times a day (n = 33) for 4 weeks
Outcomes	Symptom score Proportion of people whose symptoms 'improved' Gastric emptying rate Measurements: at 4 weeks
Funding	No information
Conflicts of interest	No information
Notes	Criteria for symptoms 'improved' (4‐point scales) Cardinal and related symptoms (bloating, nausea/vomiting, belching, anorexia, epigastric pain) were scored as 3 (severe), 2 (moderate), 1 (mild), or 0 (none). The sum of each symptom score consisted of the overall symptom scores. The percentage change in overall symptom scores before and after treatment was used to determine ratings of symptoms 'improved' . Markedly effective: 100% to 75% reduction of overall symptom scores Effective: 25% to 75% reduction of overall symptom scores Ineffective: less than 25% reduction of overall symptom scores Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random number table was used
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Unlikely to be blinded (open trial)
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline and as a treatment outcome.

Lin 2013.

Methods	Study design: three‐arm RCT Duration: January 2010 to August 2012 Follow‐up: 4 and 8 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 54 (10) years in all participants Gender (M/F): 67/53 Major eligibility criteria: 1) aged 40 to 65 years 2) type 2 diabetes based on the diagnostic criteria in the 2010 Guidelines for Prevention and Treatment of Type‐2 Diabetes by Chinese Medical Association 3) diabetic gastroparesis diagnosed by the Diagnostic Criteria for Internal Medicine Diseases (China) 4) a diabetes diagnosis of at least 5 years 5) no evidence of underlying gastric diseases confirmed by upper GI endoscopy and ultrasound
Interventions	Manual acupuncture once daily (group A, n = 40) versus manual acupuncture twice daily (group B, n = 40) versus domperidone 10 mg three times a day (group C, n = 40) for 4 weeks
Outcomes	Response rate Gastric emptying rate Measurements: at 4 and 8 weeks (data measured at 8 weeks contributed to the analyses on the short‐term effects)
Funding	No information
Conflicts of interest	No information
Notes	Criteria for symptoms 'improved' (3‐point scales) Cardinal and related symptoms (bloating, nausea/vomiting, belching, anorexia, epigastric pain) were scored as 3 (severe), 2 (moderate), 1 (mild), or 0 (none). The sum of each symptom score consisted of the overall symptom scores. The percentage change in overall symptom scores before and after treatment was used to determine ratings of symptoms 'improved'. Markedly effective: more than 75% of complete improvement of clinical symptoms Effective: 20% to 75% improvement of clinical symptoms Ineffective: less than 25% of clinical symptoms Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective). Group A and B were combined into one group ('acupuncture') for pairwise comparison between acupuncture and gastrokinetic drug.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random number table was used
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline and as a treatment outcome.

Liu 2001.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 30 days
Participants	Country: China Condition: diabetic gastroparesis Age (mean (range)): 56.8 (52 to 74) in the acupuncture group; 54.2 (50 to 73) in the control group Gender (M/F): 28/16 Major eligibility criteria: 1) diabetes defined by the 1985 WHO criteria, 2) cardinal symptoms, 3) delayed gastric emptying confirmed by the barium meal study (gastric emptying less than 60% at 30 minutes), and 4) reduced GI motility
Interventions	Manual acupuncture once daily plus domperidone 10 mg three times daily (n = 25) versus the same medication alone (n = 19) for 4 weeks
Outcomes	Proportion of people whose symptoms 'improved' Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Moxibustion was used for 22% of the stimulated acupuncture points (2 of 9). Criteria for symptoms 'improved' (3‐point scales) Markedly effective: complete recovery of bloating symptoms, ability to consume a normal diet, gastric wave frequency equal to or greater than 3 times per minute, or equal to or greater than 60% of gastric emptying at 30 minutes Effective: significant improvement of bloating symptoms, partially increased amount of food intake, frequency of gastric wave equal to or greater than 2 times per minute, or less than 60% of gastric emptying at 30 minutes Ineffective: no significant improvement of bloating symptoms, gastric wave frequency less than 2 times per minute, or less than 60% gastric emptying at 30 minutes Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No randomisation method was reported.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline.

Liu 2009.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 4 weeks
Participants	Country: China Condition: mild diabetic gastroparesis Age (mean (SD)): 59 (10) among all the participants Gender (M/F): 48/42 Major eligibility criteria: 1) diabetes defined by the WHO 1999 criteria and 2) cardinal symptoms
Interventions	Manual acupuncture once daily versus domperidone 10 mg three times a day for 40 days
Outcomes	Response rate Measurements: at 6 weeks (results contributed to the short‐term effect estimates)
Funding	No information
Conflicts of interest	No information
Notes	Criteria for symptoms 'improved' (3‐point scales) Recovery: complete recovery of clinical symptoms and normalization of gastric emptying measured by a barium meal test Effective: complete recovery of clinical symptoms and delayed gastric emptying measured by a barium meal test Ineffective: no significant improvement of clinical symptoms and delayed gastric emptying measured by a barium meal test Ratings of symptoms 'improved' were dichotomised into 'improved' (combining recovery and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information on the method of random sequence generation
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Unlikely to be blinded (open trial)
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Unclear risk	Delayed gastric emptying was objectively measured at baseline in some of participants, although there was no description of the objective measurement of delayed gastric emptying in the eligibility criteria. Delayed gastric emptying was confirmed by the barium meal study in the sixty‐four percent of participants (31 of 45 participants in the acupuncture group and 27 of 45 in the control group) at baseline (bias due to misclassification of gastroparesis in 36% of participants whose delayed gastric emptying was not confirmed).

Liu 2011.

Methods	Study design: parallel‐group RCT Duration: October 2005 to October 2010 Follow‐up: 4 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 51.4 (2.1) in the acupuncture group; 51.6 (2.2) in the control group Gender (M/F): 52/44 Major eligibility criteria: 1) diabetes defined by the WHO 1999 criteria, 2) a diabetes diagnosis of at least 1 year, 3) persistent cardinal symptoms for at least 4 weeks, and 4) no evidence of underlying gastric diseases confirmed by upper GI endoscopy or radiography
Interventions	Manual and electroacupuncture three times daily plus domperidone 10 mg three times daily (n = 48) versus the same medication alone (n = 48) for 4 weeks
Outcomes	Proportion of people whose symptoms 'improved' Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (4‐point scales) Recovery: complete reduction of symptoms or change in gastric motility or gastric emptying time to pre‐disease status Markedly effective: significant reduction of symptoms, increased appetite, residual bloating, increased gastric motility or reduced gastric emptying time, and radiographic contrast exams showing no gastric dilation and reduction of retained food Effective: reduction of symptoms, intermittent presence of early satiety and constipation, improved gastric motility, reduced gastric emptying time, radiographic contrast exams showing mild gastric dilation, and presence of a small amount of retained food Ineffective: no change in symptoms and no change in gastric motility or gastric emptying time Ratings of symptoms 'improved' were dichotomised into 'improved' (combining recovery, markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No randomisation method was reported.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Unclear risk	No description of the objective measurement of delayed gastric emptying was provided.

Liu 2012.

Methods	Study design: parallel‐group RCT Duration: July 2006 to July 2009 Follow‐up: 20 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 47.3 (8.7) in the acupuncture group; 49.1 (9.3) in the control group Gender (M/F): 33/37 Major eligibility criteria: 1) diabetes defined by the 1999 WHO criteria, 2) cardinal symptoms persisting over two months, 3) delayed gastric emptying confirmed by a barium meal study, and 4) no evidence of underlying gastric diseases confirmed by upper GI endoscopy
Interventions	Manual acupuncture once daily plus mosapride 30 mg daily (n = 35) versus the same medication alone (n = 35) for 4 weeks In both groups, mosapride 30 mg daily was administered for up to 20 weeks of follow‐up visits.
Outcomes	Proportion of people whose symptoms 'improved' Control of blood glucose (haemoglobin A1c, mean blood glucose, mean amplitude of glycaemic excursions, hypoglycaemic events) Adverse events Measurements: at 4 and 20 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (3‐point scales) Markedly effective: complete reduction of gastroparetic symptoms including nausea, vomiting, bloating, and anorexia Effective: partial reduction of gastroparetic symptoms Ineffective: no change or worsened gastroparetic symptoms Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	No randomisation method was reported, but additional information was obtained after author contact. "computer‐generated random number table was used".
Allocation concealment (selection bias)	Low risk	Not reported in the article. By author contact: concealed opaque envelope method was used.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline.

Mo 2005.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 4 weeks (28 days)
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 51.7 (5.5) in the acupuncture group; 55.9 (7.1) in the control group Gender (M/F): 44/37 Major eligibility criteria: 1) type 2 diabetes defined by the American Diabetes Association 1997 guidelines, 2) a diabetes diagnosis of at least 5 years, 3) cardinal symptoms, 4) delayed gastric emptying confirmed by a barium meal study (more than 6 hours of gastric emptying), and 5) no evidence of underlying gastric diseases confirmed by upper GI endoscopy or radiography
Interventions	Manual acupuncture once daily plus domperidone 10 mg three times daily (n = 41) versus the same medication alone (n = 40) for 28 days
Outcomes	Proportion of people whose symptoms 'improved' Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (3‐point scales) Markedly effective: complete reduction of symptoms and normalised gastric motility measured by upper gastrointestinal barium radiographic study Effective: significant improvement of symptoms and improved gastric motility measured by upper gastrointestinal barium radiographic study Ineffective: no improvement of symptoms and gastric motility measured by upper gastrointestinal barium radiographic study Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No randomisation method was reported.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	Not reported
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline.

Mu 2010.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 4 weeks
Participants	Country: China Condition: postsurgical gastroparesis Age (mean (range)): 62 (46 to 80) in the treatment group; 62 (45 to 79) in the control group Gender (M/F): 22/19 Major eligibility criteria: 1) no mechanical gastric outflow obstruction and 2) cardinal symptoms present after the completion of gastric drainage or after the initiation of semi‐fluid diets
Interventions	Manual and electroacupuncture once daily with routine care (n = 24) versus routine care alone (n = 17) for 2 to 4 weeks
Outcomes	Proportion of people whose symptoms 'improved' at 2, 3 and 4 weeks Measurements: at 2, 3 and 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	The number of treatment sessions was not reported. Time‐to‐event outcomes were used to define the proportion of people whose symptoms 'improved'. These data were not meta‐analysed with other types of proportion of people whose symptoms 'improved' in other studies. Criteria for symptoms 'improved': patients with no bloating, no problems passing flatus and defecation, and no bloating after completion of gastric decompression
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No randomisation method was reported
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available
Other bias	Unclear risk	No description of the objective measurement of delayed gastric emptying was provided.

Qi 2013.

Methods	Study design: three‐arm RCT Duration: June 2010 to June 2012 Follow‐up: 8 weeks and 1 year
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 46.8 (9.1) years in the mosapride alone group; 47.6 (9.5) years in the acupuncture alone group; 49.6 (8.3) in the acupuncture combined with mosapride group Gender (M/F): 97/85 Major eligibility criteria: 1) type 2 diabetes based on the diagnostic criteria in the 2010 Guidelines for Prevention and Treatment of Type‐2 Diabetes by Chinese Medical Association, 2) diabetic gastroparesis diagnosed by the Diagnostic Criteria for Internal Medicine Diseases (China), 3) chronic history of diabetes, 4) presenting cardinal symptoms (e.g. bloating or nausea) or gastrolithiasis, 5) delayed gastric emptying confirmed by a barium meal study (more than 4 hours of gastric emptying), and 6) no evidence of underlying gastric diseases confirmed by upper GI endoscopy or radiography
Interventions	Manual acupuncture once daily (n = 54) versus mosapride 5 mg 3 times daily (n = 62) versus manual acupuncture combined with the same mosapride prescription (n = 66) for 8 weeks
Outcomes	Proportion of people whose symptoms 'improved' Measurements: at 8 weeks and 1 years
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (3‐point scales) Markedly effective: complete reduction of symptoms and normalised gastric motility measured by upper gastrointestinal barium radiographic study. Effective: significant improvement of symptoms but delayed gastric motility measured by upper gastrointestinal barium radiographic study. Ineffective: no improvement of symptoms and gastric motility measured by upper gastrointestinal barium radiographic study. Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed (at 8 weeks measurement). (1‐year follow‐up outcomes were measured in 146 (80%) of participants).
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline.

Shi 2012.

Methods	Study design: two‐arm parallel RCT Duration: October 2010 to June 2012 Follow‐up: 8 and 12 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 60.7 (8.1) years in the real acupuncture group; 58.6 (8.2) years in the sham acupuncture group Gender (M/F): 30/30 Major eligibility criteria: 1) aged 40 to 75 years, 2) diabetes presenting less than 7.0% of HbA1c and currently using insulin injections, 3) gastroparesis defined by delayed gastric emptying confirmed by a barium meal study (more than 6 hours of gastric emptying), 4) at least one cardinal symptoms among early satiety, nausea, vomiting, abdominal bloating or belching, 5) symptoms presented at least 3 months ago. 6) no evidence of underlying gastric diseases confirmed by upper GI endoscopy
Interventions	Manual acupuncture once daily (i.e. 5 times per week) (n = 30) versus sham acupuncture (insertion of acupuncture needles on non‐acupuncture points with the same manipulation method) once daily (i.e. 5 times per week) (n = 30) for 2 weeks. Whether gastrokinetics or other medication as concomitant treatments were used was not reported.
Outcomes	Gastroparesis Cardinal Symptom Index (GCSI) scores Proportion of people whose symptoms 'improved' Number of participants with delayed gastric emptying Serum gastrin level Serum motilin level Measurements: at 2 weeks (all outcomes) and 12 weeks (GCSI scores)
Funding	Research fund by the Chengdu University of Traditional Chinese Medicine
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (percentage reduction of the GCSI scores before and after treatments were used to determine ratings of symptoms 'improved') Markedly effective: more than 75% of complete improvement of clinical symptoms Effective: 20% to 75% improvement of clinical symptoms Ineffective: less than 25% of clinical symptoms Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective). Only GCSI score measured at 12 weeks contributed to the analysis.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated random number table was used
Allocation concealment (selection bias)	Low risk	Opaque sealed envelope method was used
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Participants were blinded to the allocated group, although whether the blinding was successful remains unclear (blinding credibility test was not performed). The practitioner performing real and sham acupuncture could not be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Trial authors reported that outcome assessors were blinded
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed both at 2 weeks and 12 weeks.
Selective reporting (reporting bias)	Low risk	The study protocol was not available. However, primary and secondary outcomes as well as measurement timepoint were defined in the methods section. All defined outcomes were reported in the results section for 2 weeks (for all outcomes) or at 12 weeks measurement (for the Gastroparesis Cardinal Symptom Index (GCSI) scores alone).
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline.

Wang 2003.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 4 weeks (25 days)
Participants	Country: China Condition: diabetic gastroparesis Age (range): 38 to 65 in the acupuncture group; 39 to 68 in control group 1; 40 to 65 in control group 2 Gender (M/F): 43/37 Major eligibility criteria: 1) type 2 diabetes defined by the WHO 1997 criteria, 2) cardinal symptoms, 3) delayed gastric emptying confirmed by a barium meal study (more than 4 hours of half gastric emptying time or more than 6 hours of gastric emptying), and 4) no evidence of underlying gastric diseases confirmed by upper GI endoscopy or radiography
Interventions	Manual acupuncture once daily for 20 days with a five‐day interval (n = 35) versus domperidone 10 mg three times daily (n = 25) versus no additional treatment (n = 20) for 25 days All participants in three groups received routine management of glucose control including diet modification, exercise, oral antidiabetic medications or insulin injection.
Outcomes	Proportion of people whose symptoms 'improved' Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (4‐point scales) Recovery: complete reduction of symptoms, percentage of reduction of aggregated symptom scores equal to 100%, and normalised gastric motility or gastric emptying time Markedly effective: significant reduction of symptoms, percentage of reduction of aggregated symptom scores equal to or greater than 75%, and significant increase in gastric motility or significant reduction of gastric emptying time Effective: reduction of symptoms, percentage of reduction of aggregated symptom scores equal to or greater than 25% and less than 75%, and increase in gastric motility or reduction of gastric emptying time Ineffective: no change in symptoms, percentage of reduction of aggregated symptom scores less than 25%, and no change in gastric motility or gastric emptying time Ratings of symptoms 'improved' were dichotomised into 'improved' (combining recovery, markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No randomisation method was reported.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline.

Wang 2007.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 4 weeks (25 days)
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 57.6 (6.6) in the acupuncture group; 58.0 (8.0) in the control group Gender (M/F): 31/29 Major eligibility criteria: 1) aged 40 to 75 years, 2) type 2 diabetes defined by the WHO 1999 criteria, 3) a diabetes diagnosis of at least 5 years, 4) cardinal symptoms, 5) delayed gastric emptying confirmed by a barium meal study (more than 4 hours of gastric emptying time), 6) gastric emptying rate equal to or less than 50%, and 7) no evidence of underlying gastric disease
Interventions	Manual acupuncture once daily (n = 30) versus mosapride 5 mg three times a day (n = 30) for 30 days
Outcomes	Symptoms score Proportion of people whose symptoms 'improved' Gastric emptying rate Adverse events Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (4‐point scales) Symptoms (bloating, epigastric pain, anorexia, belching, abnormal defecation) were scored as 4 (very severe), 3 (severe), 2 (moderate), 1 (mild), or 0 (none). The sum of each symptom score consisted of the overall symptom scores. The percentage change in overall symptom scores before and after treatment was used to determine ratings of symptoms 'improved'. Markedly effective: equal to or greater than 76% reduction of overall symptom scores Effective: 33% to less than 76% reduction of overall symptom scores Ineffective: equal to or less than 33% reduction of overall symptom scores Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of random sequence generation not reported
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Unlikely to be blinded (open trial)
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	95% of participants completed the study and provided data. The number of dropouts in the acupuncture group (n = 1) and control group (n = 2) were similar.
Selective reporting (reporting bias)	Unclear risk	No study protocol available
Other bias	Low risk	Delayed gastric emptying objectively measured at baseline and as a treatment outcome

Wang 2009.

Methods	Study design: parallel‐group RCT Duration: 2006 to 2008 Follow‐up: 4 weeks (30 days)
Participants	Country: China Condition: diabetic gastroparesis Age: not reported Gender (M/F): 40/36 Major eligibility criteria: 1) type 2 diabetes, 2) TCM diagnoses of diabetes, abdominal fullness, or nausea/vomiting, 3) cardinal symptoms, 4) delayed gastric emptying measured by ultrasonography (more than 6 hours of gastric emptying), 5) reduced gastric motility measured by ultrasonography, and 6) no evidence of underlying gastric diseases confirmed by ultrasonography
Interventions	Manual acupuncture once daily (n = 40) versus domperidone 10 mg three times (n = 36) for 30 days
Outcomes	Proportion of people whose symptoms 'improved' Number of improved patients with respect to each cardinal symptom (early satiety, bloating, anorexia, nausea, vomiting, loss of appetite, reduced body mass) Number of patients with improved GI motility Number of patients with improved gastric emptying Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (3‐point scales) (TCM clinical research guidelines) Markedly effective: complete improvement of clinical symptoms, ability to consume a normal diet, and normalised gastric motility or gastric emptying time Effective: improvement of major clinical symptoms, bloating and loss of appetite after overeating, improved gastric motility measured by gastric ultrasonography, and gastric emptying time less than 6 hours Ineffective: no significant improvement of clinical symptoms, no change in gastric motility measured by gastric ultrasonography, and gastric emptying time more than 6 hours Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No randomisation method was reported.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	Study protocol was not available
Other bias	Unclear risk	Delayed gastric emptying was objectively measured at baseline and as a treatment outcome. The study did not report any criteria for diabetes (bias related to misclassification of diabetic gastroparesis).

Wu 2015.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 30 days
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 52 (11) in the acupuncture group and 54 (10) in the control group Gender (M/F): 66/54 Major eligibility criteria: 1) diabetes and diabetic gastroparesis as defined by the 1997 American Diabetes Association Diagnostic Criteria, 2) a diabetes diagnosis of at least 5 years, 3) cardinal symptoms, 4) delayed gastric emptying confirmed by a barium meal study (more than 6 hours of gastric emptying), 5) decreased gastric motility, and 6) no evidence of underlying gastric diseases confirmed by upper GI endoscopy and ultrasound
Interventions	Manual acupuncture once daily plus domperidone 10 mg three times a day versus the same medication for 28 days
Outcomes	Proportion of people whose symptoms 'improved' Measurements: at 4 weeks
Funding	No information
Conflicts of interest	No information
Notes	Criteria for symptoms 'improved' (4‐point scales) (Guiding Principles of Clinical Research on Traditional Chinese Medicine). Symptoms (nausea and vomiting, bloating, fatigue, anorexia, and early satiety) were scored as 3 (severe), 2 (moderate), 1 (mild), or 0 (none). The sum of each symptom score consisted of the overall symptom scores. The percentage change in the overall symptom scores before and after treatment was used to determine the ratings of symptoms 'improved'. Recovery: equal to or greater than 95% reduction of overall symptom scores Markedly effective: equal to or greater than 70% reduction of overall symptom scores Effective: equal to or greater than 30% reduction of overall symptom scores Ineffective: less than 30% reduction of overall symptom scores Ratings of symptoms 'improved' were dichotomised into 'improved' (combining recovery, markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information on the method of random sequence generation
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Unlikely to be blinded (open trial)
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline.

Xu 2012.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 4 weeks (3 weeks of treatment and 1 week of post‐treatment follow‐up)
Participants	Country: China Condition: diabetic gastroparesis Age (range): 44 to 75 in the acupuncture group; 36 to 76 in the control group Gender (M/F): 25/25 Major eligibility criteria: no criteria were reported for the study population.
Interventions	Manual and electroacupuncture once daily plus cisapride 10 mg three times daily (n = 25) versus the same medication alone (n = 25) for 3 weeks
Outcomes	Proportion of people whose symptoms 'improved' Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria of symptoms 'improved' (4‐point scales) Complete recovery: complete reduction of symptoms, normalised gastric emptying, and improvements sustained up to the 3‐month follow‐up Markedly effective: complete reduction of symptoms, normalised gastric emptying, and improvements not sustained up to the 3‐month follow‐up Effective: partial reduction of symptoms and partial improvement of gastric emptying Ineffective: no change of or worsened symptoms at the 3‐week post‐treatment assessment and no change in gastric emptying Ratings of symptoms 'improved' were dichotomised into 'improved' (combining complete recovery, markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A random number table was used.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Unclear risk	No description on the objective measurement of delayed gastric emptying was provided. The study did not report any criteria for diabetes (bias related to misclassification of diabetic gastroparesis)

Xu 2016.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 45 days
Participants	Country: China Condition: diabetic gastroparesis Age (range): 35 to 70 years Gender (M/F): 42/40 Major eligibility criteria: 1) diabetes defined by the WHO 1999 criteria, 2) at least 5 years of diagnosed diabetes, 3) cardinal symptoms, 4) delayed gastric emptying confirmed by a barium meal study (more than 6 hours of gastric emptying), and 5) no evidence of underlying gastric diseases confirmed by upper GI endoscopy or radiography
Interventions	Manual acupuncture once daily plus usual care (management of blood pressure, control of glucose and cholesterol levels, exercise, and dietary modification to improve gastric motility) versus usual care alone for 45 days
Outcomes	Proportion of people whose symptoms 'improved' Measurements: at 6 weeks (results contributed to the short‐term effect estimates)
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (3‐point scales) (TCM clinical research guidelines) Recovery: complete improvement of clinical symptoms, normalised gastric motility or less than 4 hours of gastric emptying time measured by upper gastrointestinal barium radiographic study. Improved: significant improvement of clinical symptoms and improved gastric motility or 4 to 6 hours of gastric emptying time measured by upper gastrointestinal barium radiographic study. Not improved: no significant improvement of clinical symptoms and more than 6 hours of gastric emptying time measured by upper gastrointestinal barium radiographic study Ratings of symptoms 'improved' were dichotomised into 'improved' (combining recovery and improved) and 'not improved' (not improved).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information on the method of random sequence generation
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Unlikely to be blinded (open trial)
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline.

Xue 2014.

Methods	Study design: parallel‐group RCT Duration: November 2011 to November 2013 Follow‐up: 30 days
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 52.9 (9.5) years in the acupuncture group; 51.9 (8.7) years in the mosapride group Gender (M/F): 33/27 Major eligibility criteria: 1) aged 30 to 75, 2) type 2 diabetes by the WHO 1999 criteria, 3) diabetic gastroparesis diagnosed by the Diagnostic Criteria for Internal Medicine Diseases (China), 4) diabetes diagnosed at least 5 years ago, 5) impaired gastric motility at least 2 months ago, 6) no evidence of underlying gastric diseases
Interventions	Manual acupuncture once daily (n = 30) versus mosapride 10 mg three times daily (n = 30) for 30 days
Outcomes	Overall symptom scores Proportion of people whose symptoms 'improved' Fasting glucose level Measurements: at 4 weeks (results contributed to the short‐term effect estimates)
Funding	Funding source: not reported
Conflicts of interest	Not reported
Notes	Criteria for symptoms 'improved' (4‐point scales) (Guiding Principles of Clinical Research on Traditional Chinese Medicine). Symptoms (nausea and vomiting, bloating, fatigue, anorexia, and early satiety) were scored as 3 (severe), 2 (moderate), 1 (mild), or 0 (none). The sum of each symptom score consisted of the overall symptom scores. The percentage change in the overall symptom scores before and after treatment was used to determine the ratings of symptoms 'improved'. Recovery: equal to or greater than 80% reduction of overall symptom scores Markedly effective: 50% to 79% reduction of overall symptom scores Effective: 25% to 49% reduction of overall symptom scores Ineffective: less than 25% reduction of overall symptom scores Ratings of symptoms 'improved' were dichotomised into 'improved' (combining recovery, markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random number table was used
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Unclear risk	Delayed gastric emptying was not measured at baseline.

Yang 2013.

Methods	Study design: parallel‐group RCT Duration: May 2009 to March 2012 Follow‐up: 4 weeks (30 days)
Participants	Country: China Condition: diabetic gastroparesis Age (range): 37 to 86 in the acupuncture group; 40 to 85 in the control group Gender (M/F): 30/40 Major eligibility criteria: 1) a diabetes diagnosis of at least 5 years, 2) cardinal symptoms persisting over one month, 3) delayed gastric emptying confirmed by a barium meal study (more than 6 hours of gastric emptying), and 4) no evidence of underlying gastric diseases confirmed by upper GI endoscopy or sonography
Interventions	Manual acupuncture once daily (n = 35) versus mosapride 5 mg three times daily (n = 35) for 30 days
Outcomes	Overall symptom scores Proportion of people whose symptoms 'improved' Changes in serum motiline and HbA1c levels Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declarations of interest: not reported
Notes	Upper limit of overall symptom scores were not clearly reported (lower limit of range is zero). Scoring criteria was also not provided. HIgher scores reflect worse outcome. Criteria for symptoms 'improved' (3‐point scales) Severity of symptoms was categorised into four grades: 1) grade 0; no symptoms; 2) grade 1; mild symptoms; 3) grade 2: moderate symptoms not affecting working ability; 4) grade 3: severe symptoms affecting working ability Markedly effective: improvement of symptoms with a decrease in grade equal to or greater than 2 from baseline or achieving grade 0 Effective: improvement in symptoms with a decrease in grade 1 from baseline Ineffective: no improvement or worsened symptoms Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No randomisation method was reported.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Unclear risk	Delayed gastric emptying was objectively measured at baseline. The study did not report any criteria for diabetes (bias related to misclassification of diabetic gastroparesis).

Yuan 2004.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 4 weeks (30 days)
Participants	Country: China Condition: diabetic gastroparesis Age (mean (range)): 40 (21 to 80) Gender (M/F): 55/20 Major eligibility criteria: 1) diabetes defined by the 1999 WHO criteria, 2) cardinal symptoms, 3) delayed gastric emptying confirmed by a barium meal study, and 4) reduced GI motility
Interventions	Manual acupuncture once daily (n = 50) versus cisapride (unknown dose) three times daily (n = 25) for 30 days
Outcomes	Proportion of people whose symptoms 'improved' Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declarations of interest: not reported
Notes	Outcomes were not included in the meta‐analysis due to the erroneous number of participants. Criteria for symptoms 'improved' (3‐point scales) Markedly effective: complete improvement of symptoms (bloating, nausea, vomiting, and belching), normalised gastric emptying Effective: partial improvement of symptoms, delayed gastric emptying, and reduced gastric motility Ineffective: no improvement of symptoms, delayed gastric emptying, and reduced gastric motility Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No randomisation method was reported.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	The number of participants who completed the outcome measure (n = 127) was not the same as the randomised number (n = 75).
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline.

Zeng 2006.

Methods	Study design: parallel‐group RCT Duration: February 2003 to February 2006 Follow‐up: 4 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 53.0 (12.0) in the acupuncture group; 51.0 (15.0) in the control group Gender (M/F): 32/48 Major eligibility criteria: 1) type 2 diabetes defined by the WHO 1985 criteria, 2) a diabetes diagnosis of at least 5 years, 3) cardinal symptoms, 4) delayed gastric emptying confirmed by a barium meal study (more than 6 hours of gastric emptying), and 5) no evidence of underlying gastric diseases confirmed by upper GI endoscopy or radiography
Interventions	Different manual acupuncture stimulations (strong, medium, and weak stimulation) plus glucose control medication once daily (n = 20; n = 20; n = 20) versus the same medication (glipizide 5 mg three times daily and metformin 250 mg two times daily) (n = 20) for 4 weeks
Outcomes	Proportion of people whose symptoms 'improved' Gastric motility measured by electrogastrography Measurements: at 4 weeks
Funding	Funding source: public grant from the Chinese government
Conflicts of interest	Declaration of interests: not reported
Notes	The patients in the control group were equally divided into three acupuncture groups for the meta‐analyses. Criteria for symptoms 'improved' (4‐point scales) Severity of symptoms and signs were categorised into four grades: complete recovery: complete reduction of symptoms; markedly effective: improvement of symptoms and signs with changes in grade equal to or greater than 2; effective: improvement of symptoms and signs with changes to grade 1; ineffective: no change of or worsened symptoms and signs. Ratings of symptoms 'improved' were dichotomised into 'improved' (combining complete recovery, markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A random number table was used.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline and as a treatment outcome.

Zeng 2008.

Methods	Study design: parallel‐group RCT Duration: March 2005 to August 2007 Follow‐up: 4 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 53.0 (12.0) in the acupuncture group; 51.0 (15.0) in the control group Gender M/F): 23/37 Major eligibility criteria: 1) type 2 diabetes defined by the WHO 1999 criteria, 2) a diabetes diagnosis of at least 5 years, 3) cardinal symptoms for at least 4 weeks, 4) delayed gastric emptying confirmed by a barium meal study, and 5) no evidence of underlying gastric diseases confirmed by upper GI endoscopy or radiography
Interventions	Manual acupuncture once daily plus domperidone 10 mg three times daily (n = 30) versus the same medication alone (n = 30) for 4 weeks
Outcomes	Overall symptom scores Proportion of people whose symptoms 'improved' Measurements: at 4 weeks
Funding	Funding source: public grant from the Chinese government
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (3‐point scales) Five symptoms (bloating, belching, anorexia, nausea/vomiting, and epigastric pain) were assessed using 4‐point scales (0 to 3; none (0), mild (1), moderate (2), and severe (3)). The sum of each symptom score consisted of the overall symptom scores. The percentage change in overall symptom scores before and after treatment was used to determine ratings of symptoms 'improved'. Markedly effective: percentage change in overall symptom scores equal to or greater than 75% Effective: percentage change in overall symptom scores equal to or greater than 25% and less than 75% Ineffective: percentage change in overall symptom scores less than 25% Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A random number table was used.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline.

Zhang 2007.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 5 weeks (34 days)
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 47.26 (5.13) in the acupuncture group; 48.31 (6.57) in the control group Gender (M/F): 37/35 Major eligibility criteria: 1) type 2 diabetes defined by the WHO 1999 criteria, 2) a diabetes diagnosis of at least 5 years, 3) cardinal symptoms, 4) delayed gastric emptying confirmed by a barium meal study (more than 6 hours of gastric emptying), 5) no evidence of underlying gastric diseases confirmed by upper GI endoscopy or radiography, and 6) reduced gastric motility measured by electrogastrography
Interventions	Manual acupuncture twice daily plus domperidone 10 mg three times daily (n = 36) versus the same medication alone (n = 36) for 34 days
Outcomes	Proportion of people whose symptoms 'improved' Cardinal symptom scores (bloating, nausea/vomiting, belching, loss of appetite, epigastric pain, diarrhoea/constipation) Adverse events Measurements: at 5 weeks (contributed to the short‐term effect estimates)
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (3‐point scales) Six symptoms (bloating, belching, loss of appetite, nausea/vomiting, epigastric pain, and diarrhoea/constipation) were assessed using 4‐point scales (0 to 3; none (0), mild (1), moderate (2), and severe (3)). The sum of each symptom score consisted of the overall symptom scores. The percentage change in overall symptom scores before and after treatments was used to determine ratings of symptoms 'improved'. Markedly effective: percentage change in overall symptom scores equal to or greater than 75% Effective: percentage change in overall symptom scores equal to or greater than 25% and less than 75% Ineffective: percentage change in overall symptom scores less than 25% Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A random number table was used.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline.

Zhang 2013.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 3 months
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 47.62 (5.13) in the acupuncture group; 46.96 (6.12) in the control group Gender (M/F): 99/95 Major eligibility criteria: 1) aged 25 to 60 years, 2) diagnosis of diabetic gastroparesis according to the following criteria: 1) a diabetes diagnosis of at least 5 years, 2) cardinal symptoms, 3) delayed gastric emptying confirmed by a barium meal study (more than 6 hours of gastric emptying), 4) no evidence of underlying gastric diseases confirmed by upper GI endoscopy or radiography, 5) reduced gastric motility measured by electrogastrography, and 6) congestion, swelling, and erosion of gastric body and antral mucosa by upper GI endoscopy
Interventions	Manual acupuncture once daily plus domperidone 10 mg three times daily (n = 98) versus the same medication alone (n = 96) for 3 weeks
Outcomes	Proportion of people whose symptoms 'improved' Cardinal symptom scores (bloating, nausea/vomiting, belching, loss of appetite, epigastric pain, diarrhoea/constipation) Adverse events Measurements: at 3 and 12 weeks
Funding	Funding source: public funds (Shanghai Municipal Science and Technology Commission of Chinese medicine modernization project)
Conflicts of interest	Declaration of interests: no conflicts of interest were declared.
Notes	Results were based on participants who completed the study follow‐up (n = 194) Criteria for symptoms 'improved' Clinical symptoms (were scored on a four‐point scale as 4 (very severe), 3 (severe), 2 (moderate), 1 (mild), and 0 (none)). The sum of each symptom score consisted of the overall symptom scores. The percentage change in overall symptom scores before and after treatment was used to determine ratings of symptoms 'improved'. Markedly effective: complete reduction of upper gastric symptoms, normal findings for upper GI endoscopy, and percentage change in overall symptom scores equal or greater than 75% Effective: reduction of upper gastric symptoms, improvement based on findings of upper GI endoscopy, and percentage change in overall symptom scores equal to or greater than 25% Ineffective: no improvement in upper gastric symptoms, no change in findings from upper GI endoscopy, and percentage change in overall symptom scores less than 25% Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random number table was used.
Allocation concealment (selection bias)	Low risk	Sequence of random allocation was concealed (the exact method of concealment was not reported).
Blinding of participants and personnel (performance bias) All outcomes	High risk	Unlikely to be blinded (open trial)
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	Almost all participants were analysed. The number and reasons for dropouts in each group were clearly reported, showing that the exclusion of participants who dropped out would not affect the robustness of the results.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Unclear risk	Delayed gastric emptying was objectively measured at baseline. The study did not report any criteria for diabetes (bias related to misclassification of diabetic gastroparesis).

Zhang 2014.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 4 weeks (28 days)
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 59.82 (8.13) in the acupuncture group; 58.57 (8.76) in the control group Gender (M/F): 44/36 Major eligibility criteria: 1) diabetic gastroparesis defined by Western medicine criteria, 2) a diabetes diagnosis of at least 3 years, 3) impaired gastrokinetics for at least four weeks, 4) delayed gastric emptying confirmed by a barium meal study (more than 4 hours of gastric emptying), 5) reduced gastric motility measured by electrogastrography, and 6) gastroparesis diagnosed by TCM
Interventions	Electroacupuncture once daily plus omeprazole 20 mg once a day (n = 40) versus the same medication alone (n = 40) for 28 days
Outcomes	Proportion of people whose symptoms 'improved' Cardinal symptom scores (bloating, nausea/vomiting, belching, loss of appetite, epigastric pain, diarrhoea/constipation) Level of serum motiline, gastrin, and fasting glucose Quantity of residual barium at 4 hours after meal intake Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for Psymptoms 'improved' (3‐point scores) Ratings of symptoms 'improved' were based on the change in clinical symptoms (upper abdominal bloating, tenderness, belching, amount of dietary intake, and abnormal defecation). Recovery: complete reduction of clinical symptoms Effective: symptom reduction of at least one degree on the scale (the scale was not explicitly defined) Ineffective: no improvement of clinical symptoms Ratings of symptoms 'improved' were dichotomised into 'improved' (combining recovery and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random number table was used
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Unable to blind
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline and as a treatment outcome.

Zhao 2004.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 4 weeks (30 days)
Participants	Country: China Condition: diabetic gastroparesis Age (mean (range)): 45 (34 to 76) in the acupuncture group; 43.5 (37 to 71) in the control group Gender (M/F): 33/35 Major eligibility criteria: 1) a diabetes diagnosis of at least 5 years, 2) cardinal symptoms, 3) delayed gastric emptying confirmed by a barium meal study (more than 6 hours of gastric emptying), and 4) no evidence of underlying gastric diseases confirmed by upper GI endoscopy or radiography
Interventions	Manual acupuncture twice daily (n = 34) versus cimetidine 800 mg once daily (n = 34) for 30 days
Outcomes	Improvement of symptoms Proportion of people whose symptoms 'improved' Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (4‐point scales) Markedly effective: complete recovery of gastrointestinal symptoms Effective: partial recovery of gastrointestinal symptoms Ineffective: no change in gastrointestinal symptoms Worsened: aggravation of symptoms Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective and worsened).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A drawing of lots was used as a randomisation method.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	The study protocol was not available.
Other bias	Unclear risk	Delayed gastric emptying was objectively measured at baseline. The study did not report any criteria for diabetes (bias related to misclassification of diabetic gastroparesis).

Zhao 2011.

Methods	Study design: parallel‐group RCT Duration: not reported Follow‐up: 4 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 52.0 (11.0) in the acupuncture group; 55.0 (14.0) in the control group Gender (M/F): 25/35 Major eligibility criteria: 1) type 2 diabetes defined by the WHO 1999 criteria, 2) cardinal symptoms persisting more than one month, 3) delayed gastric emptying confirmed by a barium meal study (more than 6 hours of gastric emptying), and 4) no evidence of underlying gastric diseases confirmed by upper GI endoscopy or sonography
Interventions	Manual and electroacupuncture once daily with diet modification plus domperidone 10 mg three times daily (n = 30) versus diet modification and the same medication (n = 30) for 4 weeks
Outcomes	Proportion of people whose symptoms 'improved' Measurements: at 4 weeks
Funding	Funding source: not reported
Conflicts of interest	Declaration of interests: not reported
Notes	Criteria for symptoms 'improved' (3‐point scales) Markedly effective: complete reduction of cardinal symptoms; complete recovery of gastric emptying rate Effective: partial reduction of cardinal symptoms; incomplete but equal to or greater than 25% recovery of the gastric emptying rate compared with baseline Ineffective: no change in or worsened symptoms and signs; less than 25% recovery of gastric emptying rate compared with baseline Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No randomisation methods were reported.
Allocation concealment (selection bias)	Unclear risk	No information
Blinding of participants and personnel (performance bias) All outcomes	High risk	Open trial
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	No information
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participants were analysed.
Selective reporting (reporting bias)	Unclear risk	No study protocol was available.
Other bias	Low risk	Delayed gastric emptying was objectively measured at baseline.

GI: gastrointestinal; M/F: male/female; RCT: randomised controlled trial; SD: standard deviation; TCM: traditional Chinese medicine; WHO: World Health Organisation

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Bai 2001	Outcomes were measured at less than 4 weeks.
Garcia 2015	Acupuncture used for arthralgias were used as placebo control; however we thought this was not different type of acupuncture rather than a placebo control.
Han 2001	Outcomes were measured at less than 4 weeks.
Hong 2010	Quasi‐randomised trial
Li 2005	Outcomes were measured at less than 4 weeks.
Li 2006	Outcomes were measured at less than 4 weeks.
Li 2008	Outcomes were measured at less than 4 weeks.
Ren 2003	Outcomes were measured at less than 4 weeks.
Shi 2014	Duplication of Shi 2012
Sun 2010a	Outcomes were measured at less than 4 weeks.
Sun 2010b	Outcomes were measured at less than 4 weeks.
Wang 2004	Moxibustion was used for more than 25% of the total acupuncture points.
Wang 2008	Outcomes were measured at less than 4 weeks. The total number of acupuncture sessions was less than six (four sessions).
Wang 2009b	Moxibustion was used for three of seven acupuncture points (42%). This study was part of Wu 2008.
Wang 2010b	Chiropractic treatment was used with the acupuncture treatments.
Wang 2010c	A herbal injection was used.
Wu 2008	Moxibustion was used for three of seven acupuncture points (42%).
Xue 2006	Outcomes were measured at less than 4 weeks.
Yu 2009	Outcomes were measured at less than 4 weeks.
Zhang 1994	Patients with impaired gastric motility measured by electrogastrography but not defined as those with gastroparesis were included.
Zhang 2007a	Quasi‐randomised trial
Zhang 2009	Outcomes were measured at less than 4 weeks.
Zheng 2010	Moxibustion was used for more than 25% of the total acupuncture points.
Zho 2011	Outcomes were measured at less than 4 weeks.
Zhou 2005	Outcomes were measured at less than 4 weeks.
Zhuang 2005	Outcomes were measured at less than 4 weeks.

Characteristics of studies awaiting assessment [ordered by study ID]

Li 2017.

Methods	Study design: parallel‐group RCT Duration: November 2014 to August 2016 Follow‐up: 4 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 49.85 (3.21) in the treatment group; 49.76 (3.18) in the control group Gender (M/F): 42/24 Major eligibility criteria: 1) diabetic gastroparesis with spleen‐stomach qi‐deficiency (by TCM clinical research guideline), 2) no major comorbidity (central nervous system, liver, kidney or cardiac diseases), 3) no psychological disease, 4) no other etiologies that affect gastric motility 5) pervious history of gastrointestinal surgery
Interventions	Manual acupuncture once daily plus domperidone 30 mg/day, antidiabetic medication and nutritional and lifestyle education (n = 33) versus domperidone 30 mg/day, antidiabetic medication and nutritional and lifestyle education without acupuncture (n = 33) for 4 weeks
Outcomes	Proportion of people whose symptoms 'improved' Glucose measurement (fasting and two‐hour postprandial blood glucose) Gastric hormones (gastrin, somatostatin and motilin) Adverse events (lethargy, mild headache or dry mouth) Measurements: at 4 weeks
Notes	This study was identified via a late search and awaiting further translation required for evaluation. Criteria for symptoms 'improved' (3‐point scales) Markedly effective: significant or complete reduction of cardinal symptoms; complete recovery of gastric emptying rate Effective: partial reduction of cardinal symptoms; partial, incomplete recovery of the gastric emptying rate compared with baseline Ineffective: little change in symptoms as well as delayed gastric emptying rate compared with baseline Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).

Wang 2017.

Methods	Study design: quasi‐RCT (Participants were allocated into the treatment and the control group according to the hospital admission sequence) Duration: January 2015 to December 2016 Follow‐up: 4 weeks (30 days)
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 37.52 (4.41) in the treatment group; 39.01 (4.99) in the control group Gender (M/F): 24/18 Major eligibility criteria: 1) diabetic gastroparesis (other criteria was not reported)
Interventions	Manual acupuncture once daily (n = 21) versus mosapride 15mg/day (n = 21) for 4 weeks. Routine antidiabetic care was provided to both groups
Outcomes	Proportion of people whose symptoms 'improved' Glucose measurement (fasting and two‐hour postprandial blood glucose) Gastric emptying rate (measured by a barium meal study) Measurements: at 4 weeks (post‐treatment)
Notes	This study was identified via a late search and awaiting further translation required for evaluation. Criteria for symptoms 'improved' (3‐point scales) Markedly effective: complete reduction of clinical symptoms; complete recovery of gastric emptying rate (no residual barium fraction at 4 hours after meal) Effective: partial reduction of clinical symptoms; recovery of the gastric emptying rate compared with baseline (no residual barium fraction when measured at 4 to 6 hours after meal) Ineffective: little change in symptoms as well as delayed gastric emptying rate compared with baseline (residual barium fraction observed at more than 6 hours after meal) Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).

Yu 2017.

Methods	Study design: parallel‐group RCT Duration: January 2014 to May 2016 Follow‐up: 8 weeks
Participants	Country: China Condition: diabetic gastroparesis Age (mean (SD)): 68.5 (6.5) in the treatment group; 67.9 (7.1) in the control group Gender (M/F): 40/32 Major eligibility criteria: 1) diabetic gastroparesis defined by China guideline for type 2 diabetes 2007 criteria, 2) a diabetes diagnosis of at least 3 years, 3) cardinal symptoms, 4) delayed gastric emptying confirmed by a barium meal study (more than 6 hours of gastric emptying), 5) no history of hepatobiliary or gastrointestinal diseases/surgery
Interventions	Manual acupuncture once daily plus mosapride 15 mg/day, antidiabetic medication and nutritional and exercise intervention (n = 36) versus mosapride 15 mg/day, antidiabetic medication and nutritional and exercise intervention without acupuncture (n = 36) for 8 weeks
Outcomes	Proportion of people whose symptoms 'improved' Serum level of vasoactive intestinal peptide Serum level of nitric oxide Measurements: at 8 weeks
Notes	This study was identified via a late search and awaiting further translation required for evaluation. Criteria for symptoms 'improved' (3‐point scales) Markedly effective: significant reduction of cardinal symptoms compared with baseline; complete recovery of gastric emptying rate Effective: partial reduction of cardinal symptoms; partial, incomplete recovery of the gastric emptying rate compared with baseline Ineffective: little change in symptoms; sustained or worsened delayed gastric emptying rate compared with baseline Ratings of symptoms 'improved' were dichotomised into 'improved' (combining markedly effective and effective) and 'not improved' (ineffective).

Characteristics of ongoing studies [ordered by study ID]

ChiCTR‐TRC‐13003461.

Trial name or title	Acupuncture and external treatment with traditional Chinese medicine of gastrointestinal cancer postoperative gastroparesis evaluation
Methods	Randomised controlled trial
Participants	Postoperative gastroparesis in patients with gastrointestinal cancer
Interventions	Group 1 (treatment group): acupuncture+external treatment with traditional Chinese medicine+foundation treatment (routine management) Group 2 (control group): metoclopramide tablets+foundation treatment (routine management)
Outcomes	Primary outcomes Remission rate of gastroparesis Time to remove stomach tube Symptom score Physical condition score
Starting date	July 2013
Contact information	Peiyu Cheng (cpy_00001@163.com), Yong Meixu (xymlfc@163.com)
Notes	Chinese Clinical Trial Registry: ChiCTR‐TRC‐13003461 Status: recruiting (accessed 20 December 2017) Date of registration: 30 July 2013 (WHO International Clinical Trials Registry Platform) Institution: Beijing Hospital of Traditional Chinese Medicine affiliated with Capital Medical University Country: China

NCT02452489.

Trial name or title	Different acupoint compatibility difference of the effect of treatment of diabetic gastroparesis
Methods	Three‐arm, parallel, randomised trial
Participants	Patients with diabetic gastroparesis
Interventions	Group 1: single point group (Zhongwan): patients treated with acupuncture with Zhongwan (RN12). Group 2: combination of He‐Mu points group: patients in the combination of He‐Mu points group who received acupuncture with Zusanli (ST36) and Zhongwan (RN12). Group 3: control group: patients in the control group who received acupuncture at the junction of the deltoid and biceps. (sham acupoint)
Outcomes	Primary outcome measures: Gastroparesis Symptoms Rating Scale (GCSI) (time frame: 8 weeks) Secondary outcome measures FBG (fasting blood glucose) (time frame: 8 weeks) PPG (postprandial blood sugar) (time frame: 8 weeks) Gastric emptying check by colour ultrasonography (time frame: 8 weeks) Gastric emptying check by X‐ray exam (time frame: 8 weeks) Short form 36 health survey (SF‐36) (time frame: 8 weeks) Self‐rating anxiety scale (SAS) (time frame: 1 week) Self‐rating depression scale (SDS) (time frame: 1 week)
Starting date	January 2015
Contact information	Tie Li, Ph.D. (zusanli_1999@126.com)
Notes	ClinicalTrials.gov Identifier:NCT02452489 Status: recruiting (accessed 20 December 2017) Date of registration: 20 May 2015 (WHO International Clinical Trials Registry Platform) Institution: The Affiliated Hospital of Changchun University of Chinese Medicine, Changchun, Jilin, China, 130117 Country: China

NCT02594397.

Trial name or title	Acupuncture for diabetic gastroparalysis: a randomised controlled trial (ADG‐RCT)
Methods	Three‐arm, parallel, randomised trial
Participants	Patients with diabetic gastroparesis
Interventions	Group 1: (experimental: acupoint combination 1) Acupoint combination 1 includes the specific acupoint ST36 (Zusanli) and a local acupoint CV12 (Zhongwan). Group 2: (active comparator: acupoint combination 2) Acupoint combination 2 includes the specific acupoint ST36 (Zusanli) and a distal acupoint PC6 (Neiguan). Group 3: (sham comparator: sham acupoint combination) Sham acupoint combination includes the specific acupoint ST36 (Zusanli) and a sham acupoint.
Outcomes	Primary outcome measures: the gastroparesis cardinal symptom index (GCSI) score (time frame: 4 weeks) Secondary outcome measures Gastric emptying time (time frame: 4 weeks) Gastrin concentrations (time frame: 4 weeks) Motilin concentrations (time frame: 4 weeks)
Starting date	December 2015
Contact information	Contact: Dr. Mailan Liu (445007305@qq.com), Ping Li (472198992@qq.com)
Notes	ClinicalTrials.gov Identifier: NCT02594397 Status: this study is not yet open for participant recruitment (accessed 20 December 2017) Date of registration: 31 October 2015 (WHO International Clinical Trials Registry Platform) Institution: the First Affiliated Hospital of Hunan University of Traditional Chinese Medicine Country: China

WHO: world health organization

Differences between protocol and review

Originally we expected that there would be an insufficient number of studies, so in the protocol we planned to include quasi‐randomised controlled trials (RCTs) as well as RCTs. However there were 32 randomised trials, and since quasi‐RCTs have a high risk of selection bias that could substantially impact the results, we decided to exclude quasi‐RCTs.

In our protocol we stated that we would use data from the first phase of randomised cross‐over trials to avoid any possible carry‐over effect (Manheimer 2010), unless more than 25% of the included studies were cross‐over trials, when we would have attempted to use the second period cross‐over data and an appropriate method of analysis, namely generic inverse variance (GIV). Given concerns about the risk of bias associated with using these methods, however, we decided at the review stage to consider only the first period of cross‐over studies.

Ordinal outcomes were dichotomised as "improved or not improved", while the classification was listed as "improved, stable, or worsened" in the protocol.

We specified outcomes for inclusion in 'Summary of findings' tables in our protocol. As none of the included studies provided data for the prespecified outcomes, we presented available data on the proportion of participants with 'improved' symptoms as defined by trial authors. We provided more detail on methods used to produce GRADE assessments in the review, as this is current practice.

As reported in the review, data were not available for some sensitivity and subgroup analyses. We performed and reported the main analyses by type of control interventions rather than as a subgroup analysis. We performed a post hoc subgroup analysis for the different types of gastrokinetic agents.

We changed the order of secondary outcomes, based on their importance.

We clarified in the Objectives section that 'the effects' of acupuncture meant both benefits and harms.

We followed the rough guide for interpretation of heterogeneity outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

Edzard Ernst was co‐author of the protocol, but did not participate at the review stage.

Contributions of authors

KHK, MSL, TYC, THK and Edzard Ernst (not on the author list for the review) drafted the protocol. KHK and TYC identified trials, checked the eligibility, and extracted the data. TYC contacted researchers to obtain additional information from the study investigators. KHK and THK entered data into Review Manager. KHK analysed the data and drafted the final review. All authors contributed to finalising the review. KHK will have the main responsibility for updating the review.

Sources of support

Internal sources

• National Research Foundation of Korean Government, Korea, South., Korea, South.

  This work was supported by a Grant to Korean Medical Science Research Center for Healthy Aging from the National Research Foundation of Korean Government (2014R1A5A2009936).

External sources

• No sources of support supplied

Declarations of interest

KHK: KHK regularly practices acupuncture for several conditions mainly including, but not limited to, chronic pain and other musculoskeletal conditions, in South Korea as a Korean Medicine Doctor (KMD).

MSL: none known

THK: none known

TYC: none known

New