Manual Therapy in Primary Dysmenorrhea: A Systematic Review and Meta-Analysis.

Primary dysmenorrhea (PD) is defined as painful menstruation in the absence of pelvic pathology and is characterized by recurrent, crampy, lower abdominal pain during menstruation, affecting 50–90% of women. 1 Among them, 15–50% of women with moderate to severe symptoms find their daily activities, including work, school, or other activities, disrupted. 2 The adverse effects of PD extend beyond physical discomfort, affecting performance by reducing attention and productivity. 3 , 4 The guidelines recommend the use of non-steroidal anti-inflammatory drugs (NSAIDs) as the primary therapeutic approach for PD. 5 Nevertheless, a subset of individuals fails to relieve their pain with NSAIDs and, instead, may experience undesirable side effects such as nausea, dyspepsia, headaches, or insomnia. 6 The non-pharmacological and non-invasive therapies are endorsed to manage dysmenorrhea symptoms, including acupuncture, biofeedback, heat treatments, transcutaneous electrical nerve stimulation, exercises, and manual therapy. 7 , 8 Manual therapy, a specialized branch of physical therapy, employing skilled hands-on techniques to assess, diagnose, and treat musculoskeletal and neuromuscular conditions. This therapeutic approach encompasses a variety of hands-on techniques, including massage, acupressure, holographic therapy, spinal manipulative therapy and more. The utilization of manual treatment for PD remains controversial. Manual therapy, either in isolation or in combination with other treatment modalities, has demonstrated potential advantages for overall health with a low reported risk of adverse effects. 9 In contrast, in 2006, a Cochrane systematic review indicated inadequate evidence to substantiate the efficacy of manual therapy in alleviating pain for women with dysmenorrhea. 10 Given these divergent viewpoints, this research was to evaluate the efficacy and safety of manual therapy for PD, utilizing updated data and comparing its effects with no treatment, placebo or first-line treatment.

This meta-analysis adhered to the guidelines outlined in the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines ( Table S1 ). 11 The study protocol was registered in the PROSPERO database ( https://www.crd.york.ac.uk/PROSPERO/ ; registration number CRD42023443160) and fully conducted in this research. A thorough search was conducted across multiple databases, including PubMed, EMBASE, MEDLINE, Cochrane Central Register of Controlled Clinical Trials, CNKI, Wanfang, SinoMed, and VIP, covering the period from inception to July 2023. The search strategy employed a combination of medical subject headings (MeSH) terms with free words such as dysmenorrhea, menstrual pain, massage, manipulation, manual therapy, and was adjusted based on the specific requirements of each database ( Table S2 ). The inclusion criteria included female patients of reproductive age who were experiencing PD. PD was defined as cyclic pelvic pain occurring during menstruation without the presence of gynecological pathologies such as endometriosis, adenomyosis, or uterine myoma. Furthermore, patients with secondary dysmenorrhea or significant medical conditions were excluded. All randomized controlled trials (RCTs) assessing pain intensity and associated outcomes were systematically included to evaluate the efficacy of manual therapy in females with PD. The treatment duration spanned a minimum of 2 menstrual cycles. In this research, the intervention group comprised various forms of manual therapies, while combinations of manual therapy with other interventions were excluded. The control group encompassed sham or placebo treatments, no treatment, and oral NSAIDs. The primary outcome measures were centered on a direct assessment of pain intensity or severity, utilizing assessment tools such as the Visual Analogue Scale (VAS) or Numeric Rating Scale (NRS) and McGill Pain Questionnaire (MPQ). The secondary outcomes included the Moos Menstrual Distress (MDQ) questionnaire, adverse events, and the effective rate. The effective rate was determined by applying specific evaluation criteria, including the improvement in clinical symptoms or other measurable outcomes. Participants were classified as “cured”, “significantly improved”, “improved”, or “non-responders” following therapy. The effective rate was computed using the following formula: \documentclass[12pt]{minimal} \usepackage{wasysym} \usepackage[substack]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage[mathscr]{eucal} \usepackage{mathrsfs} \DeclareFontFamily{T1}{linotext}{} \DeclareFontShape{T1}{linotext}{m}{n} {linotext }{} \DeclareSymbolFont{linotext}{T1}{linotext}{m}{n} \DeclareSymbolFontAlphabet{\mathLINOTEXT}{linotext} \begin{document} ${\mathrm{Effective\ Rate = }}\left({{\mathrm{N1 + N2 + N3}}} \right){\mathrm{/N}}$\end{document} , where N, N1, N2, and N3 represent the total sample size and the number of patients who were cured, significantly improved, and improved, respectively. 12 Two researchers (LX and WYH) independently conducted searches, integrating the results using NoteExpress software to eliminate duplicates and identify potentially eligible articles through a review of titles and abstracts. The full texts of the selected articles were subsequently assessed against predetermined inclusion and exclusion criteria. Data were extracted in a standardized manner, with any discrepancies resolved through consultation with a senior reviewer (YCH). The data collection form included recorded information such as the first author, publication year, sample size, age, course of PD, intervention regimen, treatment duration, and follow-up. Two reviewers (LX and SYN) employed Cochrane Handbook of Systematic Reviews to assess the potential for bias in each included RCT. 13 This assessment encompassed seven items categorized by the risk of bias, with each item being assigned to one of three risk categories: low risk, unknown risk, and high risk. The evidence quality was then rated as good, moderate, low, and very low using the grading of recommended assessment development and evaluation tool (GRADEpro GDT). 14 If any of the following conditions were met: 1) risk of bias; 2) inconsistency; 3) indirectness; 4) imprecision; and 5) publication bias, the GRADE grading downgraded quality of each study from excellent. To gauge the clinical significance of each treatment effect, a comparison was made with the minimum clinically important difference (MCID): 1.3 points for pain intensity on NRS, 15 10 points for MDQ, 16 and a 19.9% improvement for the effect rate. 12 The outcomes were classified into short-term (2–3 months), intermediate (3–12 months), and long-term (>12 months). In cases where multiple time frames fell within the same category, data closest to 3 months (short-term), 6 months (intermediate), and 12 months (long-term) were considered. Meta-analysis was performed using the RevMan 5.3. Heterogeneity was assessed through I-square (I 2 ) statistics and Cochrane Q test. 17 I 2 0.01 indicated a low heterogeneity. When possible, a random-effects model was adopted for meta-analysis. In cases where substantial heterogeneity was observed, sources of heterogeneity were explored through subgroup or sensitivity analysis. For dichotomous variables, odds ratios (ORs) with 95% confidence intervals (CIs) were utilized, and other binary data were converted into OR values. Continuous variables were analyzed using the Hedges’ g method and expressed as standardized mean difference (SMD) or mean difference (MD) with 95% CIs. The findings were descriptively synthesized if there was only one relevant study or when the data were inappropriate for quantitative synthesis. To assess publication bias, a funnel plot was employed when more than ten papers were available for pooling.

A total of 6068 citations were retrieved initially. After excluding 2367 duplicate studies, a further 3603 studies were excluded based on the screening of titles and abstracts, as they did not align with the study purpose. Subsequently, the full texts of 98 studies were carefully reviewed, and 65 were found not to meet the inclusion criteria. Finally, 32 RCTs that met the specified criteria were included in the analysis. All studies were published between January 1999 and 2019, with 18 studies published in Chinese 18–35 and 14 in English. 36–49 The PRISMA flowchart of the screening process is shown in Figure 1 . Figure 1 PRISMA flowchart. Adapted from Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. J Clin Epidemiol . 2021;134:178–189. Creative Commons. 11 PRISMA flowchart. Adapted from Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. J Clin Epidemiol . 2021;134:178–189. Creative Commons. 11 The study population comprised a total of 2566 participants who were randomly assigned to one of the 32 comparator groups. Among these, 18 studies were conducted in the Chinese mainland, 18–35 5 in Iran, 36 , 37 , 39 , 40 , 47 2 in Germany, 45 , 46 3 in Taiwan, 41–43 and 1 each in HongKong, 48 America, 44 Italy, 38 and Turkey. 49 The age of participants ranged from 14 to 41 years, and the duration of symptoms varied from a mean of 3 months to 17 years. Across the 32 studies, manual therapy consisted various techniques, with 15 studies focusing on massage, 10 on acupressure, 2 on foot reflexology, 2 on holographic therapy, 2 on spinal manipulative therapy, and 1 on neuromuscular therapy. In the control group, 8 studies employed no treatment, 3 used sham therapy, and 21 involved oral medications. Details are presented in Table 1 . Table 1 Summary of Included Trials Basic Information Study Country Study Design Number of Participants Age, Mean (SD)/Range Intervention Control Intervention Control Intervention Control Atrian 2013 36 Iran RCT 27 32 21.40 (2.45) 21.95 (3.01) Acupressure No treatment Azima 2015 37 Iran RCT 34 34 21.41 (0.95) 21.08 (1.21) Massage No treatment Barassi 2018 38 Italy RCT 30 30 31.56 (7.44) Neuromuscular therapy NSAIDs Bazarganipour 2017 39 Iran RCT 30 30 22 (1.71) 21.76 (1.73) Acupressure Placebo Behbahani 2016 40 Iran RCT 40 40 20.02 (1.44) 20.37 (1.54) Acupressure NSAIDs Chen 2004 41 Taiwan RCT 35 34 17.78 (1.43) Acupressure No treatment Chen 2010 42 Taiwan RCT 36 35 16.75 (1.36) 16.77 (1.19) Acupressure No treatment Chen 2011 18 China RCT 30 30 22.47 (4.58) 23.70 (4.56) Massage NSAIDs Chen 2015 43 Taiwan RCT 65 64 18.75 (1.74) 18.73 (0.63) Acupressure No treatment Chen 2016 19 China RCT 25 25 20.44 (1.54) 20.44 (1.8) Massage No treatment Guo 2015 20 China RCT 34 34 14–30 14–30 Acupressure NSAIDs Hondras 1999 44 America RCT 69 69 31.1 29.7 Spinal manipulative therapy Placebo Hu 2015 21 China RCT 40 40 22.61 (1.63) 22.83 (1.72) Massage NSAIDs Jia 2011 22 China RCT 16 16 21–27 21–27 Holographic therapy No treatment Lin 2010 23 China RCT 20 20 14–41 Massage NSAIDs Lin 2017 24 China RCT 20 20 17.45 (1.13) Massage NSAIDs Lin 2018 25 China RCT 20 20 20.2 (1.43) 20.93 (1.53) Massage NSAIDs Liu 2012 26 China RCT 55 55 22.46 (7.21) 23.51 (6.98) Massage NSAIDs Ru 2015 27 China RCT 50 50 26.62 (2.14) 26.77 (2.21) Massage NSAIDs Susanne 2018 45 Germany RCT 111 110 24.4 (3.3) 23.7 (3.9) Acupressure No treatment Tang 2012 28 China RCT 30 30 21.41 (1.58) 21.39 (1.03) Spinal manipulative therapy NSAIDs Vagedes 2019 46 Germany RCT 23 17 30.22 (7.72) 26.65 (8.4) Massage NSAIDs Valiani 2010 47 Iran RCT 32 36 21.6 (1.79) Foot reflexology NSAIDs Wang 2003 29 China RCT 62 30 17–35 15–32 Massage NSAIDs Wong 2010 48 Hong Kong RCT 19 21 22 21.57 Acupressure No treatment Wu 2017 30 China RCT 12 12 22.41 (1.56) 22.6 (1.72) Massage No treatment Xu 2008 31 China RCT 62 47 20.3 Holographic therapy NSAIDs Yilmaz 2019 49 Turkey RCT 29 29 20.3 19.7 Foot reflexology Placebo Zhao 2015 32 China RCT 30 30 22 (3.6) 21 (4.1) Massage NSAIDs Zhou 2016 33 China RCT 34 30 26.2 (2.59) 25.3 (3.1) Massage NSAIDs Zhu 2012 34 China RCT 50 46 – – Massage NSAIDs Zhu 2015 35 China RCT 125 125 18–27 18–27 Acupressure NSAIDs Summary of Included Trials Basic Information In terms of outcome reporting, the VAS was the most frequently utilized, with 15 studies 18 , 24 , 28 , 30 , 33 , 36–38 , 41–44 , 47–49 providing relevant data. Following closely, 14 studies 19 , 20 , 22–24 , 26 , 27 , 29–35 reported effect rates, all originating from China. Additionally, 4 studies, 22 , 25 , 26 , 30 reported pain scores, 4 41–43 , 48 reported outcomes using the MDQ, 3 21 , 45 , 46 mentioned outcomes using the NRS, and 2 40 , 47 demonstrated outcomes using the MPQ. Adverse events were documented in two studies. 28 , 45 The duration of most treatments lasted 2 to 4 menstrual cycles, with 6 studies having follow-ups of up to 6 months and one study extending to 12 months. Details are presented in Table 2 . Table 2 Summary of Included Trials’ Outcome Measures and Follow-Up Period Study Outcomes Course of Disease, Range/Mean±SD Intervention Period (Menstrual Cycle) Follow-Up Period (Menstrual Cycle) Intervention Control Atrian 2013 36 VAS – 3 3 Azima 2015 37 VAS, Anxiety Level – 3 3 Barassi 2018 38 VAS, Number of days with pain – 4 4 Bazarganipour 2017 39 Wong-Baker faces pain scale, SF12 – 3 3 Behbahani 2016 40 MPQ – 2 2 Chen 2004 41 VAS, VASA, MDQ – 3 6 Chen 2010 42 VAS, VASA, MDQ – 3 6 Chen 2011 18 VAS, PGF2a, UHD, PGE2 4.46±3.44 year 5.51±3.38 year 3 3 Chen 2015 43 VAS, MDQ – 4 12 Chen 2016 19 Traditional Chinese Medicine Symptom Score, Effective Rate – 2 2 Guo 2015 20 Effective Rate 6 month-5 year 3 3 Hondras 1999 44 VAS, PGF2a - 4 4 Hu 2015 21 NRS 5.63±2.51 year 6.21 ±1.89 year 3 3 Jia 2011 22 Effective Rate, Pain score, UHD 2–6 year 3 3 Lin 2010 23 Effective Rate 0.5–5 year 3 6 Lin 2017 24 VAS, Effective Rate, Clinical Symptom Scoring for Dysmenorrhea 3.75±0.67 year 2 5 Lin 2018 25 Pain score, Hormone levels, emotional factors 3.7±0.62 year 3 3 Liu 2012 26 Effective Rate, Pain score 0.5–10 year 3 3 Ru 2015 27 Effective Rate 7.26±3.15 month 7.13±3.04 month 3 3 Susanne 2018 45 NRS, Number of days with pain - 3 6 Tang 2012 28 VAS 63.89±30.36 month 68.32±15.58 month 3 3 Vagedes 2019 46 NRS, SF12 – 3 3 Valiani 2010 47 VAS, MPQ, PRI – 3 3 Wang 2003 29 Effective Rate 3 month-5 year 3 3 Wong 2010 48 VAS, MDQ – 3 3 Wu 2017 30 VAS, Effective Rate, Pain score, PPI, Traditional Chinese Medicine Symptom Score 6.33±2.05 year 6.25±1.76 year 3 3 Xu 2008 31 Effective Rate 0.7–4.3 year 3 3 Yilmaz 2019 49 VAS – 2 4 Zhao 2015 32 Effective Rate 3±2.4 year 3±2.4 year 3 3 Zhou 2016 33 VAS, Effective Rate 2.7±2.03 year 2.6±1.66 year 3 3 Zhu 2012 34 Effective Rate – 3 3 Zhu 2015 35 Effective Rate 2–8 year 3 3 Abbreviations : VASA, Visual Analog Scale for Anxiety; SF12, Short Form 12-Item Health Survey; PRI, Pain Rating Index; PPI, Pain Pressure Threshold; PGF2a, Prostaglandin F2 alpha; UHD, Uterine hemodynamics; PGE2, Prostaglandin E2. Summary of Included Trials’ Outcome Measures and Follow-Up Period Abbreviations : VASA, Visual Analog Scale for Anxiety; SF12, Short Form 12-Item Health Survey; PRI, Pain Rating Index; PPI, Pain Pressure Threshold; PGF2a, Prostaglandin F2 alpha; UHD, Uterine hemodynamics; PGE2, Prostaglandin E2. Figures 2 and 3 illustrated the risk of bias observed in the included studies. Notably, in 4 studies, 27 , 30 , 33 , 36 the randomization methods were considered to pose a high risk of bias, whereas the randomization methods were deemed appropriate in 13 studies. 18 , 23–25 , 37 , 39 , 40 , 43–47 , 49 Additionally, 15 studies 19–22 , 26 , 28 , 29 , 31 , 32 , 34 , 35 , 38 , 41 , 42 , 48 did not provide specific details regarding the generation of random sequences. Allocation concealment was implemented in 4 studies. 36 , 43 , 45 , 49 Except for 1 study 36 reporting blinding of participants, the remaining studies 18–20 , 22–27 , 31 , 33–35 , 44–46 , 21 , 28–30 , 32 , 37–43 , 48 , 49 , 47 did not furnish information on blinding. Notably, details of missing data were not reported in any of the enrolled studies, suggesting an absence of selective reporting bias. With the exception of 1 study 28 with uncertainty, all other studies 18–27 , 31 , 33–46 , 29 , 30 , 32 , 47–49 did not exhibit any other biases. Figure 2 Risk of bias graph: review of the authors’ judgments regarding each risk of bias item presented as percentages across all included studies. Figure 3 Risk of bias summary: review of the authors’ judgments regarding each risk of bias item in each included study. Risk of bias graph: review of the authors’ judgments regarding each risk of bias item presented as percentages across all included studies. Risk of bias summary: review of the authors’ judgments regarding each risk of bias item in each included study. To synthesize pain scores, a meta-analysis was conducted, incorporating 8 studies. 30 , 36 , 37 , 41–43 , 45 , 48 The pooled data exhibited significant heterogeneity (I 2 =57%), as depicted in Figure 4 . In the short term (n=191, MD=1.30, 95% CI: 0.24~2.37), manual therapy was preferred over no treatment by subgroup analysis; however, this effect was not seen in the intermediate term (n=361, MD=0.91, 95% CI: −0.18~1.99) or long term (n=129, MD=0.33, 95% CI: −0.33~1.09). Another subgroup analyses revealed that both acupressure (n=589, MD=0.61, 95% CI: 0.22~1.00) and massage (n=92, MD=2.48, 95% CI: 0.92~3.99) were superior to no treatment group, as depicted in Figure 5 . Figure 4 Overall and different follow-up times subgroup forest plot of weighted mean difference (95% CI) for pain intensity for manual therapy versus no treatment. Figure 5 Overall and various manual therapies subgroup forest plot of weighted mean difference (95% CI) for pain intensity for manual therapy versus no treatment. Overall and different follow-up times subgroup forest plot of weighted mean difference (95% CI) for pain intensity for manual therapy versus no treatment. Overall and various manual therapies subgroup forest plot of weighted mean difference (95% CI) for pain intensity for manual therapy versus no treatment. 3 studies 19 , 22 , 30 reported that the short-term effective rate in the manual therapy group was significantly higher than that in the no treatment control group (n=106, OR=8.73, 95% CI: 3.03~25.19), with no observed heterogeneity (I 2 =0), as illustrated in Figure 6 . A subgroup analysis showed that, in terms of improving effective rate, massage was superior to no treatment (n=74, OR=11.15, 95% CI: 3.40~36.58), but holographic therapy was not as beneficial (n=32, OR=3.46, 95% CI: 0.32~37.47). Figure 6 Overall and various manual therapies subgroup forest plot of weighted OR (95% CI) for effective rate for manual therapy versus no treatment. Overall and various manual therapies subgroup forest plot of weighted OR (95% CI) for effective rate for manual therapy versus no treatment. Additionally, 4 studies 41–43 , 48 reported MDQ scores. The results indicated that acupressure can lead to a greater reduction in MDQ scores compared to no treatment (n=309, MD=3.84, 95% CI: 2.27~5.40), as depicted in Figure 7 . The subgroup analysis demonstrated that the benefit might extend from short-term, intermediate to long- term. Figure 7 Overall and different follow-up times subgroup forest plot of weighted mean difference (95% CI) for MDQ for manual therapy versus no treatment. Overall and different follow-up times subgroup forest plot of weighted mean difference (95% CI) for MDQ for manual therapy versus no treatment. 3 studies 39 , 44 , 49 comparing manual therapy to a placebo reported outcomes related to pain, suggested that the differences in the effects of manual therapy and the placebo on pain may not be statistically significant in short term (n=255, MD=0.01, 95% CI: −0.37~0.58), as depicted in Figure 8 . The interventions in these three studies were acupressure, spinal manipulative therapy, and foot reflexology. The meta-analysis results are consistent with the findings of the original studies but lack clinical significance. Figure 8 Forest plot of weighted mean difference (95% CI) for pain intensity for manual therapy versus place control. Forest plot of weighted mean difference (95% CI) for pain intensity for manual therapy versus place control. A total of 9 studies 18 , 21 , 24 , 25 , 28 , 33 , 38 , 46 , 47 were subjected to meta-analysis to synthesize pain intensity findings between manual therapy and NSAIDs. Due to substantial heterogeneity (I 2 =99%), subgroup analyses were performed based on different types of manual therapy. As illustrated in Figure 9 , 1 study 47 suggested that foot reflexology might yield greater pain relief than NSAIDs (n=68, MD=1.24, 95% CI: 0.41~2.07), although the difference did not attain statistical significance. 6 studies 18 , 21 , 24 , 25 , 33 , 46 indicated that massage therapy proved more effective in alleviating pain compared to NSAIDs (n=324, MD=3.85, 95% CI: 1.60~6.10). Furthermore, 1 study 28 reported that the spinal manipulative therapy could potentially offer superior pain relief compared to NSAIDs (n=55, MD=3.26, 95% CI: 2.21~4.40). Conversely, another study 38 found that Neuromuscular therapy demonstrated no remarkable difference in pain relief in comparison to NSAIDs (n=60, MD=−0.47, 95% CI: −1.35~0.41). Figure 9 Overall and various manual therapies subgroup forest plot of weighted mean difference (95% CI) for pain intensity for manual therapy versus NSAIDs. Overall and various manual therapies subgroup forest plot of weighted mean difference (95% CI) for pain intensity for manual therapy versus NSAIDs. Subgroup analysis based on different follow-up durations reveals that manual therapy, in both short-term (n=467, MD=2.48, 95% CI: 0.79~4.18) and intermediate-term (n=40, MD=7.18, 95% CI: 6.75~7.61), may provide more pain relief compared to the NSAIDs group, as depicted in Figure 10 . Figure 10 Overall and different follow-up times subgroup forest plot of weighted mean difference (95% CI) for pain intensity for manual therapy versus versus NSAIDs. Overall and different follow-up times subgroup forest plot of weighted mean difference (95% CI) for pain intensity for manual therapy versus versus NSAIDs. 11 studies 20 , 23 , 24 , 26 , 27 , 29 , 31 , 33–35 , 49 reported that the manual therapy group exhibited a significantly higher effective rate to the NSAIDs (n=1029, OR=4.87, 95% CI: 3.29~7.20), showing no heterogeneity (I 2 =19%). When compared to NSAIDs, the subgroup analysis revealed that acupressure, massage, and a holographic treatment are more beneficial in increasing effective rate. These findings were summarized in Figure 11 . Subgroup analysis based on different follow-up durations indicates that manual therapy has an advantage over NSAIDs in improving effective rate, particularly in the short-term (n=949, OR=4.77, 95% CI: 3.17~7.17) and intermediate (n=80, OR=6.13, 95% CI: 1.57~23.98), as depicted in Figure 12 . Figure 11 Overall and various manual therapies subgroup forest plot of weighted OR (95% CI) for effective rate for manual therapy versus NSAIDs. Figure 12 Overall and different follow-up times subgroup forest plot of weighted OR (95% CI) for effective rate for manual therapy versus NSAIDs. Overall and various manual therapies subgroup forest plot of weighted OR (95% CI) for effective rate for manual therapy versus NSAIDs. Overall and different follow-up times subgroup forest plot of weighted OR (95% CI) for effective rate for manual therapy versus NSAIDs. The reported results from the single study 40 indicated a greater reduction in pain for acupressure compared to NSAIDs in short-term (n=80, MD=−6.92, 95% CI: −10.90~-2.94), which was proven by MPQ scores. Among the 32 studies analyzed, adverse events associated with the interventions were documented in only 2 studies. 1 study 45 highlighted soreness in the lower back region associated with spinal manipulative therapy. In contrast, another study 28 reported adverse events linked to self-acupressure, including bruises, deterioration, hand pain, pressure pain, menstrual cycle shifts, dizziness, nausea, leg pain, and finger tingling. The majority of these adverse events were considered minor, as they either resolved spontaneously or were treated effectively. To assess quality, GRADEpro GDT was employed, incorporating outcomes from the enrolled studies. The evidence was generally appraised as being of low or very low quality based on GRADE criteria, attributable to the elevated risk of bias and pronounced heterogeneity. Notably, self-rating scales employed as a primary outcome for pain assessment were considered low-quality evidence. The summarized findings of interventions are presented in Table 3 . Table 3 Summery of Finding Outcome No of Participants (Studies) Relative Effect (95% CI) Anticipated Absolute Effects (95% CI) Certainty Without Manual Therapy With Manual Therapy Difference Manual therapy compared to no treatment for PD Pain intensity № of participants: 629 (8 RCTs) – – – MD 0.81 higher (0.35 higher to 1.28 higher) ⨁◯◯◯ Very low a,b,c Effective rate № of participants: 106 (3 RCTs) OR 8.73 (3.03 to 25.19) 50.9% 90.1% (86.1 to 97.6) 39.1% more (24.9 more to 45.4 more) ⨁⨁◯◯ Low d MDQ № of participants: 309 (3 RCTs) – – – MD 3.84 higher (2.27 higher to 5.4 higher) ⨁◯◯◯ Very low a,d Manual therapy compared to placebo control for PD Pain intensity № of participants: 507 (9 RCTs) – – – MD 0.10 higher (0.37 lower to 0.58 higher) ⨁◯◯◯ Very low a,b,c Manual therapy compared to NSAIDs for PD Pain intensity № of participants: 507 (9 RCTs) – – – MD 3.01 higher (1.08 higher to 4.94 higher) ⨁◯◯◯ Very low a,b,c Effective rate № of participants: 1029 (11 RCTs) OR 4.87 (3.29 to 7.20) 73.3% 93.0% (90 to 95.2) 19.7% more (16.7 more to 21.9 more) ⨁⨁◯◯ Low a,c Notes : 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). 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. a Download one level for serious risk of bias: included studies did not conduct the blinding method, and unclear risk of bias in one or two domains. b Downgraded one level for serious inconsistent: interventions of included studies inconsistent, or the outcome indicators exist statistical heterogeneous. c Downgraded one level for wide confidence intervals. d Downgraded two level for serious imprecision: very small sample size. Abbreviations : CI, confidence interval; MD, mean difference; OR, odds ratio. Summery of Finding Notes : 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). 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. a Download one level for serious risk of bias: included studies did not conduct the blinding method, and unclear risk of bias in one or two domains. b Downgraded one level for serious inconsistent: interventions of included studies inconsistent, or the outcome indicators exist statistical heterogeneous. c Downgraded one level for wide confidence intervals. d Downgraded two level for serious imprecision: very small sample size. Abbreviations : CI, confidence interval; MD, mean difference; OR, odds ratio. Sensitivity analyses demonstrated robustness by excluding those with a sample size less than 30, or those with unclear randomization procedures ( Table S3 ). As depicted in Figure 13 , the likelihood of publishing bias appeared to be minimal, as indicated by the study of the funnel plot, which did not reveal any significant asymmetry. However, it is essential to approach the interpretation of funnel plot results with caution, considering inherent heterogeneity in the meta-analysis and acknowledging other potential sources of bias. Figure 13 Funnel plot of effective rate for manual therapy versus NSAIDs. Funnel plot of effective rate for manual therapy versus NSAIDs.

The results in this work suggested that manual therapy might alleviate menstrual pain in short-term, surpassing both no treatment and NSAIDs. However, it is impossible to rule off an effective placebo effect during manual therapy. Manual therapy demonstrated only mild adverse events, indicating a relatively safe profile. It’s important to note that our study recommendations are constrained by limitations stemming from the low quality of the included RCTs. More rigorously designed trials are imperative to validate and confirm our findings.

We methodically examined and compiled the data from a diverse array of studies investigating manual therapy for PD in this meta-analysis. 32 studies included covered a wide variety of manual therapeutic techniques, such as massage, acupressure, foot reflexology, holographic therapy, spinal manipulative therapy, and neuromuscular therapy. By amalgamating data from these RCTs, our analysis provided a thorough overview of the effectiveness of manual therapy in addressing PD. Notably, the adverse events reported across the studies were predominantly minor in nature, with many resolving spontaneously or responding well to effective treatment. The evidence, rated as low to very low, indicates that manual therapy for dysmenorrhea may demonstrate superior efficacy compared to both no treatment and oral NSAIDs, resulting in a more substantial alleviation of pain. These results endorse the preference for manual treatment over no treatment in patients with PD. The treatment impact estimates exceed the MCID threshold of 20%, with an effect rate of 39.1%. Nonetheless, our meta-analyses revealed that the treatment impact estimates for quality of life (MDQ, 3.84 points on a 30-to-150-point scale) and pain (NRS, 0.81 points on a 0-to-10-point scale) were marginally below thresholds of MCID thresholds among patients with PD. These results bolstered the case for utilizing manual treatment rather than oral medication in PD patients. The decrease in pain intensity by 3.01 points (95% CI: 1.08~4.94) surpassed the MCID of 2 points, while the improvement in effective rate by 19.7% (16.7~21.9) is close the MCID of 19.9%. In studies comparing manual therapy with a placebo control (n=255, MD=0.10, 95% CI: −0.37~0.58), the results suggest that manual therapy did not exhibit superiority. Concerning this outcome, considerations arise on two fronts. Firstly, there is contemplation of the limited number of included studies and an insufficient sample size. Secondly, attention is drawn to the potential influence of the placebo effect. The placebo effect is defined as a genuine positive psychological or physiological response solely attributed to the knowledge of receiving a substance or undergoing a procedure, rather than the inherent efficacy of that substance or procedure. 50 This improvement is not initiated by the treatment itself but rather by the patient’s beliefs and expectations. Previous research underscores the significance of acknowledging the placebo effect in the context of physical therapy, especially when compared with placebos associated with pharmaceuticals or other intervention measures. 51–53 Subgroup analyses were conducted following significant heterogeneity in outcomes related to pain. In a subset characterized by a small sample size and a limited number of studies, conflicting results were observed. It is crucial to approach the interpretation of this specific subset with caution, considering its restricted representation and smaller study population, which could contribute to the observed discrepancies. We discerned a prevalent trend of subpar research quality in the scrutinized papers, as evidenced by our comprehensive analysis of bias risk and evidence quality. The primary factor contributing to this observation is the lack of transparent data in most studies regarding the generation of random sequences and participant assignment. Furthermore, the application of blinding procedures proves challenging due to the distinctive nature of manual therapy interventions; nevertheless, a body of research 36 has demonstrated the feasibility of blinding in this context. These methodological deficiencies underscore the imperative for enhanced reporting guidelines and innovative approaches to address the challenges associated with blinding and randomization in studies focusing on manual therapy interventions. Manual therapy provides a non-pharmacological option for addressing PD, catering to individuals who prefer non-drug interventions or seek complementary approaches alongside conventional medical treatments. The personalized nature of manual therapy allows for tailored interventions with therapists adjusting techniques based on pain severity, preferences, and overall health. The specific mechanisms underlying the efficacy of manual therapy in addressing PD are not fully understood. However, a study by Chen et al 54 revealed that massage therapy was beneficial for PD. Because it may relieve pain by enhancing uterine blood flow and regulating aberrant levels of prostaglandin F2a (PGF2a) and prostaglandin E2 (PGE2). Farzaneh et al 55 suggested that acupressure at the Sanyinjiao point (SP6) can be an effective, feasible, and cost-effective intervention for mitigating PD. The SP6 acupoint serves as the junction point of the liver, spleen, and kidney meridians. According to the traditional Chinese medicine, this intervention can strengthen the spleen function, resolve and expel dampness, and restore balance to Yin and blood, liver, and kidneys. 56 Furthermore, some individuals believe that manual therapy for dysmenorrhea is closely associated with reflex points. For instance, Mur et al 57 , 58 reported increased intestinal blood flow during stimulation of corresponding reflex points compared to subjects intervened with reflex stimulation on unrelated points. Similarly, Sudmeier et al 59 demonstrated increased renal blood flow, measured with Doppler sonography, with reflexology on foot points related to the kidney compared with individuals given placebo reflexology at points not associated with the kidney. Prior to this research, a systematic review 60 evaluated the effect of manual therapy for PD, incorporating four studies published in 2017. In contrast to this earlier review, our study involved a more comprehensive search strategy, encompassing a greater number of studies meeting the criteria. Meanwhile, different analytical methods were employed herein. Our research aimed to extensively explore various manual therapy modalities and considered the cyclical nature of dysmenorrhea. Additionally, we defined a minimum treatment duration of at least two months, acknowledging the periodicity of menstrual pain. The control group exclusively comprised placebos and NSAIDs. This design, with a placebo control, facilitates the evaluation of the specific effects of manual therapy. Additionally, the comparison of manual therapy with NSAIDs offers insights into the relative effectiveness and potential advantages of manual therapy over standard pharmacological interventions. This research was subjected to several limitations. Firstly, the included studies were of low quality, restricting the availability of high-quality evidence supporting the efficacy of manual therapy in treating dysmenorrhea. Secondly, various manual therapy modalities were not classified and quantified when assessing the overall effectiveness due to their diverse range. Lastly, despite conducting subgroup analyses to address significant heterogeneity observed in the meta-analysis, the issue persisted without resolution. In clinical practice, the findings underscored the potential efficacy of manual therapy in alleviating menstrual pain. Healthcare practitioners may contemplate incorporating manual therapy into their treatment protocols for individuals experiencing PD. From a research perspective, this research highlights the necessity for further investigating distinct manual therapy modalities and their comparative efficacy. Future research endeavors should strive to enhance the quality of evidence based on high-quality studies with rigorous methodologies. Additionally, exploring the mechanisms behind the effects of manual therapy on dysmenorrhea could contribute valuable insights to the field.