The protocol for this systematic review and meta-analysis was registered in PROSPERO (No. CRD42023425087; https://www.crd.york.ac.uk/PROSPERO/), and followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Page et al., 2021).

Two researchers (STY and WKY) independently searched PubMed, EMBASE, Cochrane Library, Web of Science, Chinese National Knowledge Infrastructure (CNKI), SinoMed, Chinese Science and Technique Journals Database (VIP), Wanfang Database, and two clinical trial registries (ClinicalTrials.gov and the Chinese Clinical Trial Registry) from study inception to May 9, 2023. The language restriction was English and Chinese. The search terms were “Constipation” “Dyschezia” “Colonic Inertia” “Astriction” “Stroke” “Cerebral Infarction” “Cerebral Hemorrhage” “Cerebrovascular Disorders” “Post stroke constipation” “Acupuncture” “Electroacupuncture” “Dry Needling” “Moxibustion” “Integrated Chinese and western medicine” “Complementary Therapies” and related terms. We also consulted citations from relevant systematic reviews. Details of the search strategies were shown in Supplementary File S1.

We excluded the following studies: (1) studies with other Chinese medicine treatments, such as Chinese herbs, massage, acupoint injection, auricular acupressure, scraping, cupping, and catgut embedding therapy, and trials that compared different acupuncture types; (2) studies with incomplete or incorrect data; (3) studies involving patients with disorders of consciousness, cognitive impairment, or serious diseases of the heart, liver, and kidney hematopoietic system; (4) studies without full-text availability.

Two researchers (WKY and LLL) independently screened the titles, abstracts, and full texts of the retrieved studies for eligibility and independently extracted the data of the final included literature. Disagreements were resolved by mutual negotiation or by consultation with a third researcher (STY). The following information was extracted: authors, publication year, general information, participants’ characteristics, details of interventions (type of acupuncture, acupoints, frequency, duration of treatment, retention time of acupuncture), and outcomes.

Two researchers (ZL and ZM) independently assessed the methodological quality of the included studies using the Cochrane Risk of Bias Tool 2.0 (RoB 2.0) (Sterne et al., 2019), which contains six aspects: randomization, deviations from the intended interventions, missing outcome data, measurement of the outcome, selective outcome reporting, and overall bias. Each aspect was evaluated as “low risk of bias,” “some concerns,” or “high risk of bias.” Disagreements were resolved by mutual negotiation or by consultation with a third researcher (YSY).

Two researchers (YMY and WJ) independently evaluated the reporting quality of interventions in each study with the Revised Standards for Reporting Interventions in Clinical Trials of Acupuncture (STRICTA) (MacPherson et al., 2010). The revised STRICTA consists of six items: acupuncture rationale, details of needling, treatment regimen, other components of treatment, practitioner background, and control intervention. Disagreements were resolved by mutual negotiation or by consultation with a third researcher (STY).

All statistical analyses were conducted using Review Manager 5.4 and Stata 15.1 software. Dichotomous outcomes were expressed as the risk ratio (RR) with 95% confidence interval (CI), while continuous outcomes were expressed as standardized mean difference (SMD) with 95% CI (Murad et al., 2019). Heterogeneity was assessed by the χ² test and the I² statistic (Higgins et al., 2019). The fixed-effect model was used in cases with low heterogeneity (p > 0.1, I² < 50%), and the random-effect model was applied in cases with substantial heterogeneity (p ≤ 0.1, I² ≥ 50%) (Higgins et al., 2019; Murad et al., 2019).

Sensitivity analysis was conducted by excluding individual studies to investigate the stability of the results. Subgroup analysis was conducted to investigate the potential causes of heterogeneity with four prespecified aspects: (1) treatment duration (<2 weeks, ≥2 weeks); (2) acupuncture frequency (<1 time/day, ≥1 time/day); (3) needle retention time (<30 min, ≥30 min); (4) types of control interventions (osmotic laxatives, stimulant laxatives, gastrointestinal prokinetic drugs, secretagogues). In addition, a funnel plot and Egger’s test were applied to evaluate publication bias when the number of included studies was more than 10, and the trim-and-fill method was used to explore whether publication bias impacted the results. Descriptive analysis was performed if the data were not suitable for meta-analysis.

The GRADE (Grading of Recommendations Assessment, Development and Evaluatio) system (Liu, 2022) was used to rank the quality of evidence in five downgrading domains: risk of bias, inconsistency, indirectness, imprecision, and publication bias. The quality of evidence was classified into four grades: high, moderate, low, or very low.

A total of 2,966 publications were retrieved from the eight databases and two clinical trial registries, and 1,413 duplicate publications were eliminated. After a review of the titles and abstracts, 1,409 publications were excluded, leaving 144 publications for secondary assessment. After reading the full text, 114 studies were eliminated (reasons for exclusion are shown in Supplementary Table S1), leaving 30 studies for inclusion (Zhou and Wang, 2001; Li and Song, 2005; Liu et al., 2008; Wang et al., 2008, 2019, 2021; Zhang et al., 2008, 2015; Cao and Sun, 2009; Shi, 2009; Zhu and Chen, 2010; Tian and Wang, 2012; Man, 2014; Yuan et al., 2014, 2021; Ma and Li, 2015; Song and Liu, 2015; Xie et al., 2016; Zhang, 2016; Gao et al., 2017; Li et al., 2018; Liu Q. et al., 2018; Liu Z. et al., 2018; Yang, 2018; Guan, 2019; Luo, 2019; Lu et al., 2020; Wu et al., 2020; Liu and Wang, 2022; Tian, 2022; Zhong et al., 2022; Figure 1).

Thirty RCTs were included in the systematic review and meta-analysis. The 30 RCTs enrolled a total of 2,220 participants (n = 1,125 and n = 1,095 from the intervention and control groups, respectively), with sample sizes ranging from 40 to 110. All trials were single-center RCTs conducted in China and published in Chinese from 2001 to 2022. Twenty-seven studies (Zhou and Wang, 2001; Li and Song, 2005; Liu et al., 2008; Wang et al., 2008, 2019, 2021; Zhang et al., 2008, 2015; Cao and Sun, 2009; Shi, 2009; Tian and Wang, 2012; Man, 2014; Yuan et al., 2014; Ma and Li, 2015; Song and Liu, 2015; Xie et al., 2016; Zhang, 2016; Gao et al., 2017; Li et al., 2018; Liu Q. et al., 2018; Liu Z. et al., 2018; Yang, 2018; Guan, 2019; Luo, 2019; Wu et al., 2020; Liu and Wang, 2022; Tian, 2022; Zhong et al., 2022) were two-armed, while three studies (Zhu and Chen, 2010; Lu et al., 2020; Yuan et al., 2021) were three-armed. Of the three-arm studies, two (Zhu and Chen, 2010; Yuan et al., 2021) compared low stimulus vs. high stimulus vs. CT, and one (Lu et al., 2020) compared Tianshu (ST25) vs. Zhigou (SJ6) vs. CT.

Eleven studies compared MA with CT, six trials compared EA with CT, three trials compared WA with CT, one trial compared moxibustion vs. CT, seven trials compared MA plus CT vs. CT, and two trials compared moxibustion plus CT vs. CT. Control interventions included osmotic laxatives (polyethylene glycol, lactulose), stimulant laxatives (senna, bisacodyl, phenolphthalein), and gastrointestinal prokinetic drugs (cisapride). Table 1 displays the characteristics of the 30 studies.

Eighteen studies (Zhang et al., 2008, 2015; Cao and Sun, 2009; Shi, 2009; Man, 2014; Yuan et al., 2014, 2021; Xie et al., 2016; Gao et al., 2017; Li et al., 2018; Liu Q. et al., 2018; Liu Z. et al., 2018; Yang, 2018; Guan, 2019; Lu et al., 2020; Wang et al., 2021; Liu and Wang, 2022; Tian, 2022; Zhong et al., 2022) used MA, six studies (Zhou and Wang, 2001; Liu et al., 2008; Wang et al., 2008, 2019; Tian and Wang, 2012; Luo, 2019) used EA, three studies (Li and Song, 2005; Song and Liu, 2015; Zhang, 2016) used WA, and three studies (Zhu and Chen, 2010; Ma and Li, 2015; Wu et al., 2020) used moxibustion. The frequency of treatment ranged from 5 to 7 sessions per week for a 1–7 week duration of treatment. Retention times ranged from 15 to 120 min. Twenty-five studies (Zhou and Wang, 2001; Li and Song, 2005; Liu et al., 2008; Wang et al., 2008, 2019; Zhang et al., 2008, 2015; Cao and Sun, 2009; Shi, 2009; Zhu and Chen, 2010; Tian and Wang, 2012; Yuan et al., 2014, 2021; Ma and Li, 2015; Xie et al., 2016; Zhang, 2016; Gao et al., 2017; Li et al., 2018; Liu Q. et al., 2018; Liu Z. et al., 2018; Yang, 2018; Luo, 2019; Lu et al., 2020; Wu et al., 2020; Liu and Wang, 2022; Zhong et al., 2022) applied fixed treatment protocols, three studies (Man, 2014; Song and Liu, 2015; Wang et al., 2021) applied individualized treatment protocols (fixed acupoints combined with acupoints based on symptoms and syndrome differentiation of Chinese medicine), and two (Guan, 2019; Tian, 2022) reported only representative acupoints. The most common acupoints were Tianshu (ST25), Zusanli (ST36), Qihai (RN6), Zhigou (SJ6), Zhongwan (RN12), Taichong (LR3), Shangjuxu (ST37), Neiguan (PC6), Danzhong (RN17), and Guanyuan (RN4). Frequency ranking of acupoints is shown in Supplementary Table S2.

The reporting quality of trial treatment protocols was evaluated by STRICTA with 17 items. As shown in Supplementary Figure S1, nearly all trials reported item 1a (acupuncture rationale), item 1b (reasoning for acupuncture treatment), item 3a (number of treatment sessions), item 3b (frequency and duration of treatment sessions), and item 6b (precise description of the control group). No studies mentioned item 4b (setting and context of treatment) and item 5 (acupuncturist’s background). The STRICTA checklist is provided in Supplementary Table S3.

Regarding randomization, 16 studies (Wang et al., 2008, 2019, 2021; Zhang et al., 2008, 2015; Shi, 2009; Tian and Wang, 2012; Man, 2014; Yuan et al., 2014, 2021; Ma and Li, 2015; Gao et al., 2017; Luo, 2019; Lu et al., 2020; Tian, 2022; Zhong et al., 2022) provided a sufficient randomized sequence generation process. In addition, one study (Ma and Li, 2015) used consecutively numbered, opaque, sealed envelopes for allocation concealment and was evaluated as low risk. The remaining studies did not provide specific details of allocation concealment, and therefore we evaluated the risk of bias as unclear. Regarding deviations from the intended interventions, due to the specificity of acupuncture therapy, one study (Gao et al., 2017) implemented blinding of therapists, and the remaining 29 studies did not report the implementation of blinding. With respect to missing outcome data, one study (Ma and Li, 2015) reported missing visits but did not perform an intention-to-treat approach, which may affect the true outcome. The remaining 29 studies had no missing data, so we evaluated the risk of bias as low. Regarding outcome measurement, one study (Gao et al., 2017) implemented blinding of outcome assessors, and the remaining 29 studies did not mention the implementation of blinding. Regarding selective outcome reporting, the expected outcomes of all studies are fully reported. The risk of bias for all trials is shown in Figure 2.

The plot of acupuncture vs. CT on total responder rate was visibly asymmetric (Figure 13A), and Egger’s test revealed potential publication bias (Egger’s test p = 0.001) (Figure 13B). We conducted trim-and-fill test analysis to assess the effect of publication bias on the interpretation of the results, and we found that this publication bias did not affect the estimates, although several RCTs showing negative findings remained unpublished (Supplementary Table S4).

The certainty of evidence for the meta-analysis was evaluated by GRADE. The quality of evidence ranged from very low to high (Table 2). The primary reasons for downgrading were inconsistency (high heterogeneity) and imprecision (small sample size).

In this study, a total of 30 RCTs involving 2,220 (1,125/1,095) patients were included to systematically evaluate the efficacy and safety of acupuncture treatments for PSC. We found that both acupuncture and acupuncture plus CT offer significant benefits over CT on metrics of total responder rate, serum SP levels, time to first BM, and serum VIP levels. Acupuncture was also superior to CT in reducing constipation symptom scores, but the level of evidence was very low to high because of potential publication bias and heterogeneity. Acupuncture operates by multiple mechanisms.

Acupuncture operates by multiple mechanisms. Clinical studies (Xiong et al., 2014; Chen et al., 2022) demonstrate that acupuncture regulates autonomic function by increasing vagal activity and inhibiting sympathetic activity, thereby affecting the central nervous system and reflex pathways involved in defecation. The treatment rebalances inhibitory and excitatory gastrointestinal hormone levels to promote colonic motility, shorten colonic transit time, and improve constipation symptoms. Xu et al. (2020) observed that in patients with functional constipation, compared with mosapride & sham EA group and mosapride control group, EA significantly increased the weekly spontaneous bowel movements, improved stool consistency, and reduced the intensity of defecating difficulty. Meanwhile, it also ameliorated the quality-of-life scores, and there were no serious adverse events during the course of the study. The potential mechanisms of acupuncture stimulation in promoting gastrointestinal function were studied more deeply in animal experiments. Studies have confirmed that acupuncture not only regulates the level of hormones related to intestinal motility (Zhu et al., 2016; Jang et al., 2017), but also improves the morphologic structure of colonic smooth muscle (He et al., 2023), and rebalances the gut microbiota (Xu et al., 2023). Wang et al. (2023) found that acupuncture could improve enteric glial cells autophagy by inhibiting PI3K/AKT/mTOR signaling. Moreover, acupuncture could restore gastrointestinal basic electrical rhythm by increasing the number of interstitial cells of Cajal (He et al., 2023).

Notably, the two interventions differed in the degree of improvements in BSS scores and AEs. Regarding BSS scores, acupuncture offered no significant advantage over CT, while acupuncture plus CT was superior to CT, which is consistent with previous findings (Zheng et al., 2018; Wang et al., 2020; Liu et al., 2021) and may result from direct action on the intestines that changes stool consistency to facilitate defecation (Chang et al., 2022, 2023). In contrast, acupuncture operates through a series of neurological and endocrine mechanisms to improve gastrointestinal tract function, and we found that acupuncture synergizes well with CT to effectively improve stool consistency, even though short-term treatment with acupuncture alone does not offer relief. Regarding safety, acupuncture alone was superior to CT, but there was no significant difference when comparing acupuncture plus CT vs. CT. It is clear that acupuncture does not cause serious AEs as a monotherapy or an adjunctive therapy.

We evaluated the effects of treatment duration, acupuncture frequency, types of control interventions, and needle retention time on total responder rates, which are relevant to clinical practice. We found that acupuncture significantly improves total responder rate with all studied treatment frequencies, control interventions, and defined needle retention times, and is effective as a monotherapy or an adjunctive therapy. Notably, though, acupuncture therapy was not found effective with uncertain needle retention time. Moreover, subgroup meta-analysis revealed that treatment durations of >2 weeks may be more effective than shorter treatment regiments at improving total responder rate. Furthermore, we evaluated acupuncture treatment regimens using STRICTA and summarized commonly used acupoints, providing scientific guidance for clinical practice and for the design and implementation of clinical studies. A total of 51 acupoints with frequencies ranging from 1 to 23 were used; the most commonly used acupoints for PSC were Tianshu (ST25), Zusanli (ST36), Qihai (RN6), and Zhigou (SJ6).

Because of the high heterogeneity of meta-analysis results, we performed sensitivity analysis and subgroup analysis to explore and eliminate sources of heterogeneity. We found that heterogeneity in total responder rates resulted from the inconsistency of treatment duration, while the heterogeneity in serum VIP levels resulted from differences in therapeutic acupoints. The source of heterogeneity in the time to first BM resulted from the higher risk of overall bias in the included studies, and the source of heterogeneity in the BSS score resulted from differences in diagnostic criteria. Some outcome indicators were still highly heterogeneous after our corrections, meaning that additional studies are needed to determine if acupuncture plus CT can improve serum VIP levels and if acupuncture can increase serum SP levels.

Compared to previous studies, our study has several strengths. First, we found that treatment duration may influence efficacy, thus providing a reference for clinical practice and clinical research. Second, we have adopted updated and objective constipation-related indicators such as time to first BM and serum gastrointestinal peptide levels to evaluate the effects of acupuncture, further supporting the use of acupuncture in treating PSC. Third, we enhanced the credibility of our results by using sensitivity analyses and subgroup analyses to explore the sources of heterogeneity, the robustness of the results, and the effects of some characteristics on the efficacy of acupuncture. Fourth, the STRICTA checklist and GRADE were used to assess clinical trial reporting quality and evidence quality, respectively. Finally, to minimize heterogeneity in the included studies, we limited the types of control interventions used and we excluded co-treatment with herbal formulas, Chinese patent medicines, and acupressure.

Some limitations in this work should be noted. First, there are no standardized diagnostic criteria for PSC, and the different Rome diagnostic criteria (e.g., Rome II vs. Rome IV) may affect the assessment of acupuncture efficacy. Second, all included studies were conducted in China, and the conclusions of our study must be verified in patients of other races. Third, the strength of our conclusions may be limited by small sample sizes, poor methodological quality (e.g., some of the included studies did not describe the randomization methodology and allocation concealment in detail), and the potential risk of bias. Fourth, most of the included studies do not report complete acupuncture details according to the STRICTA checklist, which may lead to bias in the interpretation of results. Fifth, Differences in gender and age of patients in RCTs, as well as in interventions in the control group, may lead to a risk of bias that reduces the credibility of the findings, so we should be cautious about the conclusions. Finally, we were unable to compare acupuncture vs. sham acupuncture, and the long-term effects of acupuncture, because of the limited number of studies and follow-up data.

Our meta-analysis suggests that acupuncture has great potential for the treatment of PSC and deserves further exploration. First, we found that 2 weeks may be the minimum effective treatment duration for acupuncture efficacy, though more research is needed to verify the robustness and scientific validity of this conclusion, as high-quality evidence is limited. Second, there are still few PSC trials that use sham acupuncture as a control treatment to eliminate the non-specific effects of acupuncture; future clinical studies should use sham acupuncture as a control. Third, the quality of future studies would be improved by standardizing study protocols, unifying diagnostic criteria, and improving the implementation of randomization, allocation concealment, and blinding. In addition, more scientific and objective outcome indicators such as complete spontaneous BMs and recurrence rates should be selected, and follow-up times should be extended. Finally, clinical studies should strictly follow the CONSORT 2010 statement (Schulz et al., 2010) and STRICTA (MacPherson et al., 2010) to improve reporting quality, especially regarding details of acupuncture treatments that are critical for the accuracy of results. More reliable guidance for clinical practice would be provided by large, high-quality, multicenter, double-blinded RCTs that comprehensively assess the efficacy and safety of acupuncture and identify efficient and rational treatment protocols.