The primary objective is to evaluate the efficacy of an 8-week standardized Tuina therapy plus usual care (UC) compared to UC alone in reducing fatigue severity in patients with CFS. Secondary objectives include assessing effects on the clinical response rate, sleep quality, mental health, quality of life, and clinical global impression, as well as evaluating the safety of the intervention and the long-term sustainability of therapeutic effects up to 24 weeks after treatment.

This study uses a prospective, multicenter, randomized, open-label, assessor-blinded, parallel-group design. The trial will be conducted at 6 clinical centers across China, with Shuguang Hospital affiliated to Shanghai University of Traditional Chinese Medicine serving as the leading center. The protocol has been developed in accordance with the SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) 2025 statement (Checklist 1) [17]. Table 1 outlines the major time points of this trial.

The study timeline is divided into three phases:

In this multicenter trial, patients with CFS will be recruited from outpatient clinics at 6 participating hospitals through advertisements on posters, WeChat, and hospital websites. Specialized physicians will confirm the diagnoses based on the 1994 Centers for Disease Control and Prevention (CDC) criteria, using comprehensive medical history taking, physical examinations, and necessary laboratory tests to exclude other organic or psychiatric causes of fatigue. To promote recruitment and retention, the treatment and relevant clinical assessments involved in the study will be offered free of charge to eligible candidates. Written informed consent will be obtained from all participants before the screening. Eligibility criteria are presented in a 2-column format in Textbox 1.

Eligible participants will be randomized in a 1:1 ratio to either the intervention group (Tuina plus UC) or the control group (UC) via a central web-based randomization system (Interactive Web Response System [IWRS]). To ensure balance between groups across the 6 centers, the minimization method will be used. Stratification factors, based on baseline data collected at the end of the run-in phase, include:

The randomization sequence is generated by the IWRS and remains concealed from all investigators, therapists, and participants until the moment of assignment. Researchers will obtain the group assignment only after confirming eligibility and completing baseline assessments.

Because Tuina is a manual therapy, neither participants nor therapists can be blinded to group allocation. This open-label design may introduce performance and expectation bias. Moreover, most efficacy outcomes, including the 11-item Chalder Fatigue Questionnaire (CFQ-11), Pittsburgh Sleep Quality Index (PSQI), HADS, SF-36, and EQ-5D-3L, are patient-reported; therefore, assessor blinding provides limited protection against bias for these self-reported outcomes. This limitation is explicitly acknowledged and will be mitigated by standardized participant instructions, equal assessment schedules, electronic or paper questionnaire completion before clinical discussion, and separation of treatment providers from outcome assessors.

Outcome assessors responsible for collecting data will be blinded to group allocation and will not participate in treatment. The CGI will be assessed by clinicians who are blinded to allocation whenever feasible. Independent statisticians will analyze data using masked group codes until the primary analysis is finalized. Participants will be instructed not to disclose treatment details to assessors during follow-up visits.

Tuina will be delivered by licensed practitioners with at least 5 years of clinical experience. Before trial initiation, all practitioners will complete centralized training consisting of 4 two-hour sessions within 1 week. Training will include theoretical instruction, video demonstration, supervised hands-on practice, calibration of manipulation frequency and force, and review of the trial standard operating procedure (SOP). Practitioners must pass a competency assessment before treating trial participants.

Symptoms of anxiety and depression will be evaluated using the HADS. This scale consists of 14 items, divided into two subscales: Anxiety (HADS-A, 7 items) and Depression (HADS-D, 7 items). Each item is scored on a scale of 0 to 3. The scores for each subscale range from 0 to 21, with higher scores indicating more severe symptoms of anxiety or depression.

Health-related quality of life will be assessed using the SF-36 [25]. This instrument measures 8 domains of health status: physical functioning (PF), role limitations due to physical health (RP), bodily pain (BP), general health (GH), vitality (VT), social functioning (SF), role limitations due to emotional problems (RE), and mental health (MH). For each domain, item scores are coded, summed, and transformed onto a scale from 0 (worst possible health state) to 100 (best possible health state). This study will specifically focus on the PF and BP subscales to assess functional improvement. For the PF subscale used as a stratification factor, the 10 PF items will be summed as a raw score ranging from 10 to 30 and transformed using the formula PF = (raw score − 10)/20 × 100; therefore, possible PF scores occur in 5-point increments.

Additionally, the EQ-5D-3L will be used to evaluate health utility. It covers 5 dimensions: mobility, self-care, usual activities, pain or discomfort, and anxiety or depression. Each dimension has 3 levels: no problems, some problems, and extreme problems. The responses are converted into a single utility index using the Chinese population-based value set [26]. The scale also includes a EuroQol Visual Analogue Scale (EQ-VAS), where participants rate their overall health on a vertical scale from 0 (worst imaginable health) to 100 (best imaginable health).

The Clinical Global Impression (CGI) scale will be used to assess the clinician’s global view of the patient’s illness and improvement [27]. It includes two subscales: Severity of Illness (CGI-S), rated on a 7-point scale from 1 (normal, not at all ill) to 7 (among the most extremely ill patients), and Global Improvement (CGI-I), rated from 1 (very much improved) to 7 (very much worse).

Adverse events (AEs) in this trial are defined as any untoward medical occurrence in a participant, regardless of its causal relationship with the treatment. Serious adverse events (SAEs) refer to events that result in death, are life-threatening, require hospitalization or prolonged hospitalization, or cause significant disability. AEs will be comprehensively monitored and recorded for both the intervention and control groups throughout the study. Specifically, monitoring will focus on intervention-related events such as subcutaneous hemorrhage (bruising), persistent local pain, skin abrasion, or fainting in the Tuina group, while simultaneously assessing potential side effects of concomitant medications or symptom fluctuation in the UC group. The number and type of AEs will be calculated, and patients will receive appropriate intervention if necessary. Any SAEs must be reported to the principal investigator and the Ethics Committee within 24 hours, and affected participants may be withdrawn from the study to ensure safety.

The sample size was determined based on the primary outcome, namely the change in CFQ-11 total score from baseline to week 8. The variability estimate was derived from individual-level data from an investigator-initiated, single-center pilot study conducted before the current multicenter protocol was finalized. In that pilot study, the mean reduction in CFQ-11 total score was 10.29 (SD 6.02) in the Tuina group and 5.36 (SD 6.39) in the control group. Because the primary end point of the present trial is change from baseline, the SDs of the baseline-to-posttreatment change scores were used for sample size planning. These values differ from the posttreatment group-level SDs reported in the pilot study table, which described the distribution of CFQ-11 scores at the posttreatment visit rather than the distribution of change scores. The pooled SD was calculated using the following formula:

With n₁=n₂=55, SD₁=6.02, and SD₂=6.39, the pooled SD was as follows:

The expected between-group difference was set at 3.0 points, corresponding to the prespecified clinically useful improvement threshold for CFQ-11 [23]. This conservative choice was made because the pilot study was single-center and relatively small, and pilot-based treatment effect estimates may be unstable or overestimate the effect expected in a larger multicenter trial. Using the MCID as the target effect ensures that the trial is powered to detect a clinically meaningful difference. To detect a 3.0-point between-group difference with a pooled SD of 6.21, 90% power, and a 2-sided significance level of 0.05, 92 participants per group are required. Allowing for a 20% dropout rate, the target sample size is 115 participants per group, for a total of 230 participants. The calculation was performed using PASS software.

The primary outcome will be analyzed using MMRM based on the FAS. The model will include treatment group, visit, center, treatment-by-visit interaction, and baseline stratification factors (depression and physical function) as fixed effects, with the baseline CFQ-11 score a covariate. Within-participant correlations among repeated measurements will be modeled using an unstructured covariance matrix. If the model does not converge, alternative covariance structures such as first-order autoregressive or compound symmetry will be selected according to Akaike information criterion and convergence diagnostics. Restricted maximum likelihood estimation will be used. The primary contrast will be the adjusted between-group mean difference at week 8, with a 95% CI and 2-sided P value.

For secondary continuous outcomes, between-group differences in change from baseline will be compared at each scheduled postbaseline time point using 2-tailed Student t test when data are approximately normally distributed or the Wilcoxon rank-sum test when distributional assumptions are not met. This time point–specific approach is prespecified because secondary outcomes are supportive and exploratory end points, the trial is powered for the primary CFQ-11 outcome rather than for multiple longitudinal secondary scales, and the main clinical interpretation of secondary efficacy will focus on the end-of-treatment assessment. Week 4 will be used to describe early treatment response, week 8 will be defined as the main secondary efficacy time point, and weeks 20 and 32 will be used to assess durability and will be interpreted as supportive follow-up contrasts. For categorical secondary outcomes, the χ² test or Fisher exact test will be used, as appropriate. The clinical response rate at week 8 will be compared between the groups using logistic regression adjusted for baseline CFQ-11 score, center, and baseline stratification factors. Ordinal CGI outcomes will be analyzed using proportional-odds models when the proportional-odds assumption is met; otherwise, nonparametric methods will be used. AE rates will be summarized descriptively and compared using the χ² test or Fisher exact test when appropriate.

To address multiplicity among secondary efficacy outcomes, P values for the prespecified week 8 secondary efficacy outcomes will be adjusted using the Benjamini-Hochberg false discovery rate procedure at 5%. Week 4, week 20, and week 32 contrasts will not be included in this multiplicity family; these follow-up contrasts will be reported with effect estimates and 95% CIs and interpreted as exploratory supportive evidence of early response or durability rather than as confirmatory efficacy tests.

All missing data will be documented with reasons whenever possible and summarized by group and visit. No last-observation-carried-forward approach will be used. The primary MMRM analysis handles missing repeated outcome data through likelihood-based estimation under the missing-at-random assumption, using all available observed data. Multiple imputation by chained equations will be performed as a sensitivity analysis for the primary outcome and key secondary continuous outcomes when postbaseline data are missing. The imputation model will include treatment group, center, baseline stratification factors, baseline outcome values, observed postbaseline outcomes, demographic variables, and relevant UC or concomitant-treatment variables. At least 50 imputed datasets will be generated, and estimates will be combined using Rubin rules [28-30]. For the clinical response rate, response status will be derived from imputed CFQ-11 values in the MI sensitivity analysis; an additional conservative sensitivity analysis will classify participants with missing week 8 CFQ-11 data as nonresponders. If overall missingness exceeds 20% or differs between groups by more than 10 percentage points, additional sensitivity analyses using delta-adjusted pattern-mixture methods will be considered to assess departures from the missing-at-random assumption. Additionally, the robustness of the primary outcome will be assessed using the PPS population as a sensitivity analysis.

All statistical analyses will be performed using SAS version 9.4 (SAS Institute). Statistical significance is defined as a 2-sided P<.05.

Ethical approval for this study was obtained from the Ethics Committee of Shuguang Hospital affiliated to Shanghai University of Traditional Chinese Medicine, China (approval number 2025-1860-200-02), and the study will be conducted in accordance with the Declaration of Helsinki. All participants will be fully informed of the study objectives, procedures, potential risks, and anticipated benefits, and written informed consent will be obtained prior to enrollment.

Participant confidentiality will be strictly maintained throughout the study. Access to identifiable data will be restricted to authorized members of the research team, and all data will be anonymized for analysis. Study findings will be disseminated through peer-reviewed publications, and no identifiable personal information or images will be included in any published materials. Data will be collected and managed using a secure, web-based electronic data capture (EDC) system. To ensure data integrity, the system incorporates automated logic checks, range validations, and a complete audit trail of all data modifications. All participant data will be deidentified using unique study identification numbers to protect confidentiality. Regular data backups will be performed, and access to the database will be strictly restricted to authorized personnel with individual passwords. Patients and the public were not formally involved in the design, conduct, reporting, or dissemination plans of this study.

This protocol describes a multicenter randomized trial designed to test the hypothesis that 8 weeks of standardized Tuina therapy plus UC will produce a clinically meaningful reduction in fatigue severity compared with UC alone among patients with CFS. We also hypothesize that Tuina will increase the clinical response rate, improve sleep, mood symptoms, physical function, bodily pain, and health-related quality of life, and show acceptable safety. The 24-week posttreatment follow-up will allow the assessment of whether any treatment benefits are sustained beyond the active intervention period.

CFS creates a substantial burden on individuals and health care systems. As highlighted by the 2021 NICE guidelines, standard nonpharmacological interventions, such as GET, are no longer recommended due to safety concerns, while CBT is primarily supportive rather than curative [1,8,9]. This underscores the need for additional nonpharmacological approaches that are acceptable to patients and unlikely to exacerbate postexertional symptoms. Previous studies suggest that Tuina and related manual therapies may improve fatigue and accompanying symptoms [11-14,31], but the evidence remains limited by small samples, single-center recruitment, heterogeneous intervention protocols, and short-term follow-up [15,16]. The present trial addresses these gaps by using a larger multicenter design, a standardized treatment manual, centralized randomization, prespecified statistical methods, and longer follow-up.

The study has several strengths. First, to our knowledge, this is the first large-scale, multicenter randomized controlled trial to investigate the efficacy and safety of Tuina therapy for CFS. The study design incorporates several methodological strengths to ensure rigor. Second, building upon our preliminary single-center pilot work, we have optimized the intervention protocol. We extended the treatment duration from 4 weeks to 8 weeks and the session length to 30 minutes. This dose optimization aligns with the chronic nature of CFS and aims to induce more stable, long-term therapeutic effects. A major challenge in manual therapy trials is operator variability [32]. Third, we have implemented a strict quality control system, including a unified SOP, centralized practitioner training, systematic treatment logs, and regular fidelity monitoring. This ensures that the intervention is reproducible across different clinical centers and allows cointerventions and concomitant medications to be described and explored in sensitivity analyses. Fourth, this trial includes a 24-week posttreatment follow-up, enabling evaluation of durability.

Several limitations should be acknowledged. First, because Tuina is a manual therapy, participant and therapist blinding is infeasible [33,34]. This open-label design may introduce performance or expectation bias. To mitigate this, we have implemented a strict assessor-blinded design, where outcome evaluators and statisticians are unaware of group allocations. Second, the primary outcome (CFQ-11) and key secondary outcomes are patient-reported measures, which are subjective. However, these scales are validated and widely accepted in CFS research, and we have included the CGI to provide a clinician’s perspective. Third, the study population is limited to patients meeting the CDC 1994 criteria; future research may need to explore efficacy across different diagnostic definitions (eg, Institute of Medicine or NICE criteria) to enhance generalizability.

If Tuina is shown to be effective and safe, it could serve as an adjunctive nonpharmacological option for CFS management. The standardized protocol may support practitioner training and facilitate implementation in clinical settings. Future research should examine pragmatic delivery models, cost-effectiveness, objective activity or physiological outcomes, mechanisms of action, and effectiveness across alternative CFS/ME diagnostic criteria. Trial findings will be disseminated through peer-reviewed publications, conference presentations, trial registry updates, and sharing of the standardized Tuina protocol where appropriate.

This multicenter randomized controlled trial will evaluate the efficacy, durability, and safety of Tuina therapy plus UC for patients with CFS. The findings are expected to provide clinically relevant evidence for whether Tuina can be integrated into CFS management as a scalable nonpharmacological intervention.