Feasibility and short-term comparative outcomes of a digital-assisted community-based electrical stimulation rehabilitation program for postmenopausal women with pelvic floor dysfunction: a non-randomized controlled study Authors: Qinyuan Yang, Qingyang Hu, Weiwei Zhou, Qi Chu, Jianlin Zhuang, Jiabei He, Yan Liang Published: 2026 ABSTRACT AbstractAimThis study aimed to assess the feasibility and short-term comparative outcomes of a community-based BES rehabilitation program assisted by a digital service package for postmenopausal women with PFD, and to explore whether pelvic floor muscle strength outcomes were compatible with a prespecified exploratory margin compared with tertiary hospital care.MethodsWe conducted a non-randomized controlled study involving 200 community-dwelling postmenopausal women with PFD. Participants in the intervention group received community-based BES plus a digital service package, while contemporaneous controls received standard BES care at a tertiary hospital. Follow-up lasted 1 month. Because treatment allocation was non-randomized, doubly robust estimation and multivariable regression were used to reduce measured confounding; findings should therefore be interpreted as exploratory rather than causal.ResultsThe observed rate differences in improvement of Type I and Type II pelvic floor muscle strength were 0.008 and 0.054, respectively. The lower bounds of the 98.75% confidence intervals remained above the prespecified exploratory margin of − 0.20, suggesting that the observed differences were compatible with the prespecified exploratory margin in this exploratory analysis. While no significant between-group difference was found for self-perceived symptom severity, the intervention group was associated with higher self-reported adherence (β = 4.604, P < 0.001), health knowledge awareness (β = 24.419, P < 0.001), and quality of life (OR = 5.292, P = 0.023). Feasibility metrics were favorable, with a mean System Usability Scale score of 81.40. However, given the non-randomized design and the prespecified margin, these findings should be interpreted cautiously.ConclusionsThis pragmatic, digital-assisted community-based rehabilitation program appears feasible and was associated with favorable short-term outcomes in postmenopausal women with PFD. Compared with tertiary hospital care, the integrated community-based program showed pelvic floor muscle strength outcomes that were compatible with the prespecified exploratory margin, while also being associated with higher self-reported adherence, health knowledge, and quality of life. However, because of the non-randomized design, retrospective registration, short follow-up, and potential residual confounding, these findings should be considered preliminary and hypothesis-generating rather than confirmatory evidence of non-inferiority, equivalence, or causal effects.RegistrationChinese Clinical Trial Registry (ChiCTR2400088675); Registered on 23 August 2024, retrospectively registered.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12905-026-04610-z.Keywords: Digital service intervention, Pelvic floor dysfunction, Postmenopausal women, Community rehabilitation Background Pelvic floor dysfunction (PFD) is a critical public health issue for postmenopausal women, characterized by high prevalence, multifaceted physical and psychosocial burdens, and systemic healthcare gaps. Epidemiological studies indicate that PFD affects approximately 37.7% of women in community settings [1], with prevalence increasing with age, menopause, and parity. Postmenopausal women are particularly vulnerable due to hormonal changes and cumulative obstetric injuries. For instance, menopause-associated pelvic floor muscle weakening (AOR = 4.03, 95% CI: 2.20–8.27) [1] and prior heavy labor (AOR = 3.21, 95% CI: 1.86–5.72) [1] are significant risk factors. Symptoms such as urinary incontinence, pelvic organ prolapse, and chronic pelvic pain impair quality of life, with urinary incontinence affecting 60% of community-dwelling women [2] and chronic pelvic pain impacting 15.3% of women in China [3]. The psychosocial consequences are equally profound. Women with PFD exhibit high rates of depression (26.8% for chronic pelvic pain, 26.3% for urinary incontinence) and anxiety (46.9% for urinary incontinence) [4], exacerbated by stigma and low health-seeking behaviors. For example, only 29% of women with urinary incontinence seek care [5], often due to shame, lack of awareness, or inadequate access to services [6, 7, 8]. Rural and underserved populations face additional barriers, such as limited healthcare infrastructure [6, 9] and cultural misconceptions [10]. The economic burden of PFD is substantial, driven by long-term management costs and the high price of invasive treatments that often follow delayed clinical intervention [11]. These factors underscore the urgent need for targeted, community-based strategies to address PFD in postmenopausal women. Current interventions for PFD, such as pelvic floor muscle training (PFMT) and biofeedback with electrical stimulation (BES), demonstrate efficacy but are limited by accessibility and population-specific gaps. PFMT, including Kegel exercises, improves muscle strength and reduces symptoms in perimenopausal women [12], with combined PFMT and yoga showing superior outcomes [12]. Similarly, BES enhances neuromuscular coordination and adherence when delivered in clinical settings [13]. However, most studies focus on postpartum women [7, 9, 14], neglecting postmenopausal populations. Additionally, interventions predominantly rely on tertiary healthcare facilities or specialized hospitals [15, 16], which are inaccessible to rural and low-income communities [6, 17]. Community-based studies highlight low PFMT awareness (12.14% adherence in postpartum Ethiopian women) [9] and poor knowledge of pelvic floor anatomy [18], suggesting that clinic-centric models fail to address systemic barriers. Digital solutions, such as telehealth and mobile apps, are emerging [19–22], but their integration with clinical therapies like BES remains underexplored. For example, mobile apps could enhance PFMT adherence [19], and a fully remote digital program improved mental health in postmenopausal women [22], yet no studies combine in-clinic BES with digital follow-up for perimenopausal women. This gap limits scalability and sustainability, particularly in resource-limited settings. To address these gaps, a digital service package was developed, integrating in-clinic BES with remote support via mobile apps, telehealth consultations, and AI-driven progress tracking [23]. Existing digital interventions show promise but lack comprehensiveness. For example, mobile applications reduced symptom severity [20, 21], and social media-based campaigns enhanced health literacy in Latinas [10]. However, these approaches often exclude structured clinical components like BES, which are critical for severe PFD [16]. Our model uniquely combines clinic-based BES sessions with real-time digital monitoring, personalized exercise plans, and psychosocial support forums. With the widespread adoption of BES devices, pelvic floor rehabilitation incorporating BES has become increasingly common in community health centers in China. However, potential disparities in therapeutic outcomes may persist when compared to tertiary hospitals or specialized hospitals, primarily due to differences in staffing, professional expertise, and technical proficiency. These factors could limit the effectiveness of traditional rehabilitation approaches, even when the same equipment is utilized. To address this gap, we hypothesize that integrating BES with a comprehensive digital service package in the community might support rehabilitation outcomes that are comparable to tertiary hospital care and compatible with a prespecified exploratory margin in short-term pelvic floor muscle strength outcomes. If supported in exploratory analyses, this model could represent a potentially scalable approach for comprehensive rehabilitation services, leveraging digital technology to compensate for the resource limitations in community settings. Such an approach may improve accessibility and support the quality of care for a broader population. Thus, the primary objective of this study was to assess the feasibility of an integrated community-based BES rehabilitation model assisted by a digital service package, and to explore whether short-term pelvic floor muscle strength outcomes in the community setting were compatible with a prespecified exploratory margin, when compared with tertiary hospital care in the same region. Secondary objectives were to examine whether the integrated program was associated with differences in adherence, health knowledge, and quality of life. Methods Study design This was a pragmatic, non-randomized controlled study. Random allocation was not implemented because the digital intervention was delivered within a community care setting and was considered susceptible to contamination of participants receiving usual care. In addition, participants in the intervention group voluntarily adopted the digital service package, which may have introduced selection bias. To improve transparency, we clarify that the primary and secondary outcomes, as well as the statistical analysis plan, were finalized before any outcome data analysis was undertaken. The study was performed in a tertiary hospital and a community health service center in Changning District, Shanghai, China between March 2024 and November 2024. The intervention was performed by the primary care team within the community health service center, for 4 consecutive weeks. The ordinary health program was maintained in the tertiary hospital as usual. The trial was intended to be registered prospectively. An initial registration application was submitted to an international clinical trial registry before the start of the study. However, because the study was conducted in China, the registry required that registration be transferred to the Chinese Clinical Trial Registry. As a result, the study was officially registered at ChiCTR (ChiCTR2400088675) on 23 August 2024, which occurred after the initiation of participant recruitment, leading to retrospective registration. Importantly, all prespecified outcomes and the statistical analysis plan were finalized prior to any data analysis and were followed as planned. We now explicitly state that retrospective registration reflects administrative transfer of registry platform rather than post hoc outcome selection; nevertheless, the retrospective registration status remains an important limitation and has been acknowledged as such. Participants The participants were postmenopausal women with PFD. We consecutively enrolled postmenopausal women with PFD at Xianxia Community Health Service Center in Shanghai Changning District as the intervention group, while contemporaneous postmenopausal patients with PFD receiving standard care at Shanghai Tongren Hospital served as controls. Only eligible patients who voluntarily agreed to use the digital service package were enrolled into the intervention group, whereas contemporaneous eligible patients receiving standard tertiary outpatient care served as controls. Given the established association between postmenopausal estrogen deficiency and PFD risk [24], and previous study has indicated that postmenopausal women aged ≥ 50 years demonstrate a higher prevalence of complex pelvic floor issues, such as mixed urinary incontinence [25], we selected this age threshold as an inclusion criterion. Additionally, eligible participants need to meet the following inclusion criteria: (1) documented history of childbirth; (2) clinical diagnosis of PFD [26]; (3) absence of severe visual or auditory impairments; (4) no psychiatric history; (5) unimpaired cognitive function; and (6) proficiency in using WeChat for daily communication. Patients with concurrent vaginal bleeding, lower urinary tract infections, tumors, traumatic injuries, severe pelvic pain, chronic debilitating diseases or severe cardiopulmonary conditions, or received pelvic floor surgery within the preceding six months were excluded. The participant flow is presented in Figure S1. Of the 203 participants initially assessed, 3 were excluded from the final analysis: 1 in the intervention group (did not complete the treatment course) and 2 in the control group (incomplete questionnaire data). All 200 participants included in the final analysis (100 per group) had complete data for primary outcomes; therefore, complete-case analysis was performed without imputation. The sample size was calculated using a two-sample non-inferiority proportion formula, with the improvement rate of pelvic floor muscle strength as the primary outcome. We established a one-sided level of 0.025. With type I and type II muscle strength improvement rates serving as co-primary endpoints, we applied Bonferroni correction =0.025/2 = 0.0125. Based on the expected improvement rate in the control group from historical data, we assumed a control-group event rate of 0.70, the non-inferiority margin was set at = − 0.2. An exploratory margin of −0.20 was prespecified prior to data analysis for this proof-of-concept study. This exploratory margin was selected considering: (1) the absence of an established clinically validated margin for community-based pelvic floor rehabilitation; (2) the pragmatic and exploratory nature of this study; and (3) precedent in the pelvic floor literature where margins have varied considerably depending on context—for example, Gupta et al. used a 30% margin in a randomized non-inferiority trial of sacral neuromodulation for women with pelvic floor disorders [27]. We acknowledge that this margin is relatively wide, and the findings should be interpreted as hypothesis-generating rather than confirmatory. The sample size was calculated to be 200 (100 per group). Intervention The intervention group received community-based rehabilitation supported by the digital service package for health. Specifically, on the basis of conventional pelvic floor biofeedback electrical stimulation therapy, patients were provided with a digital health service package as an adjuvant intervention. After patients signed the informed consent form and completed the baseline pelvic floor electrophysiological function test, the family doctor in the community health service center administered conventional pelvic floor biofeedback electrical stimulation therapy using a pelvic floor rehabilitation instrument. A single session of pelvic floor biofeedback electrical stimulation therapy lasted 20 min. Upon completion of the electrical stimulation, the patients were instructed to contract and relax their pelvic floor muscles regularly according to the graphics displayed on the monitor to perform biofeedback training. Pelvic floor biofeedback electrical stimulation therapy was delivered 2–3 times per week, with one month defined as one treatment course. After the patients finished one full course of treatment, they underwent a follow-up pelvic floor electrophysiological function test and completed a questionnaire assessment of pelvic floor rehabilitation efficacy, and relevant data were collected accordingly. The digital service package was provided to patients when they received biofeedback electrical stimulation therapy for the first time. The digital service package included five core components: (1) evidence-based health education content, (2) diversified training materials (e.g., video-based Kegel exercise tutorials), (3) WeChat-based group chat for peer support and interaction with healthcare providers, (4) automated text or voice reminders (sent 3 times per week), and (5) AI-driven chatbot providing instant responses to common queries, with human administrator backup for complex queries (addressed within 48 h). The package was delivered over 4 weeks via the WeChat platform. Detailed intervention specifications are provided in Supplementary Table S1. Engagement was monitored qualitatively through weekly review of WeChat group chat logs by the study coordinator, but standardized quantitative metrics (e.g., login frequency, module completion rates, chatbot usage counts, response times) were not prospectively collected. This limits our ability to examine dose–response relationships between engagement levels and rehabilitation outcomes, which is acknowledged as a limitation. Full development and preliminary evaluation of the digital package have been reported previously [23]. This digital package can deliver pelvic floor health education, pelvic floor muscle training skills, and professional supervision, feedback and support from medical staff by providing educational videos, video-based tutorials, real-time communication via group chat platforms, and automated reminders and personalization through online chatbot [23]. It enables postmenopausal women with PFD to complete their rehabilitation in home-based settings while receiving real-time feedback and support from healthcare providers [23]. To ensure intervention fidelity, all automated reminders and educational messages were sent according to a prespecified schedule. The content of the digital service package (educational videos, follow-along exercise videos, and chatbot responses) was standardized and identical for all participants. Adherence to the intervention protocol was monitored by a study coordinator who reviewed weekly the WeChat group chat logs and the frequency of keyword-based information retrieval from the chatbot. Our study did not restrict treatment protocols at Xianxia Community Health Service Center, with therapeutic decisions jointly determined by the doctors and the patients. Only patients meeting inclusion criteria who voluntarily adopted the digital service package for health were enrolled as the intervention group. The intervention duration was set at one month. Control participants underwent conventional biofeedback-assisted electrical stimulation therapy and usual care (e.g. patient education) in tertiary care outpatient settings. Participants in the control group received usual care from tertiary hospitals, which consisted of routine face‑to‑face consultations during clinic visits. At these visits, they were provided with standard verbal education on pelvic floor dysfunction and basic instructions for pelvic floor muscle training (PFMT). No structured printed materials (e.g., brochures, self‑management guides) were given, and no home‑based guidance was offered. The recommended PFMT frequency was only verbally mentioned (typically “perform exercises several times a day”), without any scheduled reminders or supervision. There were no follow‑up contacts outside the clinic (e.g., telephone calls, text messages, or digital messages), no remote monitoring, and no tailored feedback. The entire control period lasted the same duration as the intervention group (1 month). Thus, the control group received no ongoing support, no home reinforcement, and no interaction beyond the initial clinic visit, which contrasts with the multi‑component, remotely supported digital service package delivered to the intervention group. Certified clinicians at the participating hospital conducted baseline pelvic floor muscle assessments for all eligible candidates prior to treatment initiation. Outcome measurements Outcome parameters of each patient were assessed at baseline and one-month post-intervention to evaluate the efficacy of digital service package for health assistance in pelvic floor rehabilitation. Primary outcome Our study’s main outcomes of patients were improvement rates of pelvic floor muscle strength (Type I and II). Muscle fatigue parameters (Type I and Type II) were assessed using a PHENIX USB 2 device (manufacturer: ELECTRONIC CONCEPT LIGNON INNOVATION, France). A combined pelvic floor electromyography (EMG) probe and pressure probe were used. The recording protocol was as follows: participants were positioned in supine. For rapid (Type II) fiber assessment, patients performed five brief maximal contractions, each lasting 1–2 s, with a 10‑second rest interval between contractions. For slow‑twitch (Type I) fiber endurance testing, patients were instructed to sustain a contraction for 10 s, with a requirement of stable maintenance for at least 5 s to be considered valid. Rest intervals between sustained contractions were either 10 s or equal to the contraction duration. A longer rest period of 10 min was given between different test phases to avoid muscle fatigue. Fatigue percentages were derived from raw EMG signals as follows: after averaging multiple contraction cycles, normalized values were calculated using a standardized manual cursor‑based point‑to‑point method. These fatigue indices were thus computed from processed EMG data rather than being device‑generated built‑in parameters. To minimize assessment bias, the pelvic floor muscle assessments were performed by independent clinicians who were blinded to the group assignments and not involved in recruitment, allocation, or intervention delivery. Assessments were conducted in a separate room, and participants were instructed not to disclose their treatment setting. Assessors received standardized training in the Modified Oxford Scale procedure, which was delivered by a professional association and was identical for both community-based assessors and those from the tertiary hospital. Formal inter-rater reliability was not assessed during this study, which is acknowledged as a limitation. Pelvic floor muscle strength was evaluated via digital vaginal palpation using the Modified Oxford Scale, a validated six-point system (0–5) ranging from no discernible contraction (Grade 0) to a strong contraction with powerful resistance against the examiner’s fingers (Grade 5) [28]. A clinically meaningful treatment was defined as 2 or more grades improvement on muscle strength scales after intervention. The improvement rate was expressed as the percentage of participants achieving the meaningful treatment threshold among all participants. Secondary outcomes We also evaluated other electrophysiological parameters of pelvic floor function, including Type I muscle fiber fatigue, Type II muscle fiber fatigue, and vaginal dynamic pressure values. These parameters were measured using pelvic floor rehabilitation devices. Type I and Type II muscle fiber fatigue were expressed as percentages, while vaginal dynamic pressure was recorded in cmH₂O. Participants’ adherence was evaluated using a validated pelvic floor exercise adherence questionnaire [29]. The questionnaire consisted of three items: (1) average daily time spent on PFMT, with five response options (“no exercise”, “<5 minutes”, “5–10 minutes”, “10–20 minutes”, “>20 minutes”) scored from 1 to 5; (2) daily repetition frequency, with five response options (“no exercise”, “<3 times”, “3–5 times”, “5–10 times”, “>10 times”) scored from 1 to 5; and (3) degree of conformity of self-performed Kegel exercises with the standard technique, measured on a visual analog scale from 0 (“not at all conforming”) to 10 (“completely conforming”). The total adherence score was the sum of the three item scores, ranging from 2 to 20, with higher scores indicating better adherence. Urinary incontinence symptom severity was assessed through the International Consultation on Incontinence Questionnaire-Short Form (ICIQ-SF) [30]. Participants’ knowledge of PFD was assessed using a 22-item pelvic floor dysfunction cognitive questionnaire developed by our research team in early stage. The 22-item pelvic floor dysfunction knowledge questionnaire was self-developed by the research team based on clinical experience and a literature review of pelvic floor health education content. Items were reviewed by a panel of three clinical experts for content validity and pilot-tested with 20 women with PFD for clarity. Each item is scored as correct (1) or incorrect (0), with total scores ranging from 0 to 22; higher scores indicate better knowledge. However, formal psychometric validation (e.g., factor analysis, test-retest reliability) was not performed, and knowledge-related findings should be interpreted cautiously. Quality of life was assessed using the Pelvic Floor Impact Questionnaire-7 (PFIQ-7) [31], a validated brief instrument for evaluating pelvic disorder-related life impact. Furthermore, at the end of the intervention, participants of intervention group were invited to complete additional online questionnaires assessing the feasibility of the digital service package for health. Usability was measured using the System Usability Scale (SUS) [32]. We assessed acceptability via the Net Promoter Score (NPS) [33]. Assessment of ease-of-use was through After-Scenario Questionnaire (ASQ) [34]. Covariates Participant sociodemographic variables were captured at enrollment. The general sociodemographic data of two groups included age, marital status, educational attainment, employment status, monthly income, medical insurance. Other covariates reported in previous studies were also included, such as, body mass index (BMI) [35], number of pregnancies and deliveries [36], and family support [37]. Data analysis Data was analysed using Stata 17 and R 4.4.3. Given the non-randomized study design, doubly robust estimation was used to reduce potential confounding bias when comparing pelvic floor muscle strength improvement rates between groups. First, a propensity score model was constructed using logistic regression to estimate each participant’s probability of receiving the intervention. Baseline covariates included age, BMI, gravidity, parity, delivery mode, marital status, health insurance type, educational level, monthly income, family supervision, and baseline Type I and Type II pelvic floor muscle strength. Average treatment effect (ATE) weights were then generated using inverse probability of treatment weighting (IPTW) based on the propensity scores to improve balance in measured baseline covariates between groups and to create a weighted pseudo-population. Covariate balance before and after weighting was assessed using standardized mean differences (SMDs) and Kolmogorov–Smirnov statistics; an SMD < 0.1 was considered indicative of adequate balance. The covariate balance plot comparing standardized mean differences before and after weighting is shown in Fig. 1. Outcome models were subsequently fitted separately for the intervention and control groups using logistic regression. Doubly robust estimation was performed using the augmented inverse probability weighting (AIPW) estimator, which combines the IPTW weights with the outcome regression models to estimate adjusted rate differences and corresponding confidence intervals. Propensity score weighting analyses were conducted using the “WeightIt” package in R version 4.4.3. Secondary outcomes were analyzed using linear regression models with covariate adjustment for potential confounders. For secondary outcomes analyses, statistical significance was defined as <0.05 using two-tailed tests. Covariate balance before and after propensity score weighting The concurrent controlled intervention study employed a doubly robust estimation approach to evaluate the clinical efficacy of community-based therapy assisted by a digital service package for health in improving pelvic floor muscle strength among postmenopausal women with PFD [38]. We hypothesize that community-based rehabilitation using a digital service package for health achieves comparable outcomes of the improvement efficacy rate of pelvic floor muscle strength to standard tertiary hospital treatment for postmenopausal women with PFD in the same geographic region. Results Baseline characteristics The baseline demographic and clinical characteristics of the intervention group (n = 100) and the control group (n = 100) are summarized in Table 1. For continuous variables, age was compared using the Wilcoxon rank-sum test, while BMI was analyzed using the independent-samples t-test. Categorical variables—including gravidity, parity, delivery method, marital status, insurance type, education level, employment status, monthly income, and family supervision—were compared using the Chi-square test or Fisher’s exact test, as appropriate. Table 1Baseline characteristics of community-dwelling older females with PFD in intervention and control groupsVariableCategoryIntervention Group(n = 100)Control Group(n = 100) Age/67.00 (59.50,70.00)65.00 (58.00,69.00)0.028^BMI/24.41 ± 2.3524.12 ± 1.990.346#Gravidity124 (24.0%)23 (23.0%)0.001264 (64.0%)44 (54.0%)≥ 312 (12.0%)33 (23.0%)Parity136 (36.0%)42 (42.0%)0.070257 (57.0%)43 (43.0%)37 (7.0%)15 (15.0%)Delivery methodVaginal93 (93.0%)93 (93.0%)1.000Cesarean7 (7.0%)7 (7.0%)Marital statusMarried88 (88.0%)96 (96.0%)0.076&Divorced2 (2.0%)0 (0.0%)Widowed10 (10.0%)4 (4.0%)Insurance typeResident medical insurance2 (2.0%)1 (1.0%)1.000&Employee medical insurance98 (98.0%)99 (99.0%)Education levelIlliterate2 (2.0%)1 (1.0%)0.615&Primary school60 (60.0%)67 (67.0%)Junior high school27 (27.0%)20 (20.0%)Senior high/vocational school10 (10.0%)11 (11.0%)Associate degree0 (0.0%)1 (1.0%)Bachelor’s degree1 (1.0%)0 (0.0%)EmploymentstatusEmployed8 (8.0%)5 (5.0%)0.390Retired92 (92.0%)95 (95.0%)Monthly income2000 ~ 500092 (92.0%)95 (95.0%)0.373&5000 ~ 10,0008 (8.0%)4 (4.0%)> 10,0000 (0.0%)1 (1.0%)Family supervisionYes7 (7.0%)9 (9.0%)0.602No93 (93.0%)91 (91.0%)^Indicates Wilcoxon rank-sum test#Indicates independent-samples t test&Indicates Fisher’s exact test Baseline characteristics of community-dwelling older females with PFD in intervention and control groups Employment status ^Indicates Wilcoxon rank-sum test #Indicates independent-samples t test &Indicates Fisher’s exact test Statistical analysis indicated that the two groups were comparable across the majority of baseline characteristics (all P > 0.05). However, statistically significant differences were observed in age (P = 0.028) and gravidity (P = 0.001). To address these initial imbalances and mitigate potential confounding, inverse probability weighting and doubly robust estimation were employed for primary outcome evaluation. Furthermore, multivariable regression models were utilized for all secondary outcome analyses to ensure the robustness and comparability of the results. Primary outcome: pelvic floor muscle strength improvement Unadjusted comparisons of Type I and Type II pelvic floor muscle strength improvement rates are provided in Table S2, showing no statistically significant differences between groups. Given the non-randomized design and the potential for confounding, effect estimates were further evaluated using doubly robust estimation with inverse probability weighting. Propensity scores were estimated based on a comprehensive set of baseline covariates including Type I and II pelvic floor muscle strength (detailed in Table S3). The distribution of propensity scores and inverse probability weights are depicted in Figure S2. After weighting, the effective sample sizes were 85.32 for the intervention group and 88.99 for the control group. Covariate balance after adjustment was meticulously assessed using SMD and Kolmogorov-Smirnov statistics, which are detailed in Table S4. Fig. 1 presents the SMD of covariates before and after propensity score weighting. All covariates reached adequate balance after adjustment, with SMD values below the 0.1 threshold, confirming comparability between the intervention and control groups. Table 2 presents the adjusted improvement rates and effect estimates for Type I and Type II pelvic floor muscle strength. For Type I muscle strength, the estimated improvement rate was 90.6% (95% CI: 0.858, 0.953) in the intervention group and 89.8% (95% CI: 0.846, 0.950) in the control group, with an adjusted rate difference of 0.008. For Type II muscle strength, the adjusted improvement rates were 85.6% (95% CI: 0.796, 0.917) in the intervention group and 80.3% (95% CI: 0.734, 0.871) in the control group, with an adjusted rate difference of 0.054. Neither comparison showed a statistically significant between-group difference at the prespecified Bonferroni-adjusted alpha level of 0.0125 (Type I: P = 0.775; Type II: P = 0.168). Table 2Adjusted effect estimates for improvement in Type I and Type II pelvic floor muscle strength in the two groupsParametersType I Fiber Muscle StrengthType II Fiber Muscle StrengthEstimated value95%CI Estimated value95%CI Mean value of the intervention group0.906(0.858, 0.953)< 0.0010.856(0.796, 0.917)< 0.001Mean value of the control group0.898(0.846, 0.950)< 0.0010.803(0.734, 0.871)< 0.001Rate difference (RD)0.008(-0.047, 0.063)0.7750.054(-0.023, 0.130)0.168Relative risk (RR)1.009(0.950, 1.072)0.7751.067(0.972, 1.171)0.172odds ratio (OR)1.094(0.592, 2.023)0.7741.466(0.847, 2.537)0.172 Adjusted effect estimates for improvement in Type I and Type II pelvic floor muscle strength in the two groups Figure 2 presents the exploratory comparison of Type I and Type II pelvic floor muscle strength improvement rates against the prespecified exploratory margin. For Type I muscle strength, the adjusted rate difference was 0.008, with a lower bound of − 0.062 for the 98.75% confidence interval. For Type II muscle strength, the adjusted rate difference was 0.054, with a lower bound of − 0.044 for the 98.75% confidence interval. For both outcomes, the lower bounds of the 98.75% confidence intervals were above the prespecified exploratory margin of − 0.20, suggesting that the observed differences were compatible with this margin in the present exploratory analysis. However, given the non-randomized design, these findings should be interpreted cautiously and should not be considered confirmatory evidence of non-inferiority, equivalence, or causal effects. Forest plot of exploratory comparison of Type I and Type II pelvic floor muscle strength improvement rates against the prespecified margin Secondary outcomes Figure 3 presents the multivariable linear regression estimates for six key secondary outcomes, adjusted for demographic and baseline characteristics (details in Table S5). For Type I muscle fiber fatigue, the regression coefficient (β) for the intervention group was 0.121 (95% CI: −0.050 to 0.293, P = 0.165; Table S6), and for Type II muscle fiber fatigue, β was 0.149 (95% CI: −0.016 to 0.313, P = 0.076; Table S7), indicating no significant differences between groups. By contrast, dynamic vaginal pressure was significantly higher in the intervention group (β = 3.347, 95% CI: 0.834 to 5.860, P = 0.009; Table S8). For self-perceived severity of incontinence symptoms, the ICIQ-SF score was lower in the intervention group (β = −0.308, 95% CI: −0.681 to 0.065, P = 0.105; Table S9), though not statistically significant. In contrast, self-reported adherence (β = 4.604, 95% CI: 3.505 to 5.703, P < 0.001; Table S10) and health knowledge awareness (β = 24.419, 95% CI: 20.338 to 28.501, P < 0.001; Table S11) were significantly higher in the intervention group. Regarding safety, no intervention-related adverse events were reported in either the intervention or the control group during the study period. Multivariable linear regression estimates of factors associated with pelvic floor muscle strength improvement. Note: All models were adjusted for age, BMI, gravidity, parity, mode of delivery, marital status, type of health insurance, educational level, monthly income, and family supervision For quality of life, based on PFIQ-7 scores, most participants reported 0 after treatment, leading to a markedly skewed distribution. Accordingly, scores were dichotomized, with 0 indicating that PFD continued to affect quality of life and 1 indicating no impact. Logistic regression analyses demonstrated that participants in the intervention group were more likely to report no PFD impact compared with controls (OR = 5.292, 95% CI: 1.258 to 22.262, P = 0.023; Table S12). Regarding feasibility, usability reflected by the SUS yielded a mean score of 81.40 (SD = 5.57), corresponding to a grade of “B” and indicating favorable user perceptions. Acceptability, reflected by the Net Promoter Score, showed that 51.0% of participants were promoters, 30.0% were passives, and 19.0% were detractors, resulting in a net score of + 32%, suggesting good overall acceptance. Ease of Use, according to After-Scenario Questionnaire results, showed a mean total score of 5.67 (SD = 0.45). Among After-Scenario Questionnaire subdimensions, task completion efficiency scored highest (mean = 5.74, SD = 0.76), while satisfaction with provided support information scored lowest (mean = 5.49, SD = 1.13), highlighting potential areas for improvement in patient-facing educational content. No adverse events or unintended effects were observed or reported by the participants in either group during the study period. Discussion In this pragmatic, non-randomized study, the integrated community-based BES program assisted by a digital service package was feasible and showed promising short-term comparative outcomes. The primary outcome results were compatible with the prespecified exploratory margin, and the intervention group also reported better adherence, health knowledge, and quality of life. However, these findings should be interpreted as preliminary because the study was non-randomized, the intervention was multicomponent, and follow-up was limited to 1 month. Importantly, the comparison in this study was between an integrated community-based program and tertiary hospital usual care, rather than between the digital package alone and no digital support. Therefore, the observed between-group differences cannot be attributed solely to the digital components, as the intervention also differed in supervision intensity, reminders, and patterns of interaction with healthcare staff. There are several plausible mechanisms that may help explain the exploratory findings related to the prespecified exploratory margin of the community-based intervention. Unlike tertiary hospital care, where in-person guidance is often time-limited and its quality may be variable across different clinicians [11], the integrated intervention package included standardized follow-along videos provided consistent, step-by-step visual instruction [23]. This standardization may help mitigate the “provider expertise gap” between tertiary and community settings, potentially enabling that patients can master correct PFMT techniques through repeated, self-paced practice. Romero-Franco et al. [19] similarly demonstrated that digital tools providing multimedia visual guidance can effectively support autonomous rehabilitation [20]. Furthermore, the intervention employed a WeChat Group Chat that enabled continuous interaction and professional supervision between healthcare providers and patients [22]. Such persistent clinical support and timely feedback may be critical in maintaining training quality and enhancing patients’ confidence in the intervention. This aligns with the findings of Lin et al. [39], who demonstrated that professional guidance delivered via video consultations and telephone coaching is a feasible and effective alternative to face-to-face therapy. However, because the intervention was multicomponent, the independent contribution of each element cannot be isolated from the present data. A notable observation in this study was the higher self-reported adherence to PFMT in the intervention group. This finding should be interpreted with caution, as self-reported adherence may be subject to recall bias and social desirability bias. The integrated intervention incorporated automated reminders sent at regular intervals, which may help reduce forgetfulness. This aligns with the conclusions of Wadensten et al., who suggest that reminder notifications, as part of eHealth tools, may offer possibilities for promoting adherence [21]. In addition, the intervention employed a WeChat group chat to host incentivized activities, which may help maintain participant engagement and motivation during the training process. Digital health interventions that incorporate interactive or game-like components have been shown to improve motivation for physical activity [40]. Moreover, the inclusion of standardized, follow-along videos may have provided a practical solution to the lack of confidence and technique-related uncertainty that often hinder independent practice [41], thereby potentially fostering a more supportive environment for behavioral change. Nevertheless, the specific components driving the observed adherence difference cannot be determined from this study. The intervention group also demonstrated a greater improvement in health knowledge awareness regarding PFD. However, it should be noted that the knowledge questionnaire was self-developed and lacked full psychometric validation; therefore, these findings should be interpreted cautiously. A survey by Akkoç et al. revealed a widespread lack of knowledge regarding pelvic floor disorders among community-dwelling women [42]. To address this knowledge gap, our intervention was designed to deliver a structured and multi-modal repository of educational resources. Participants were able to access various forms of educational materials, including short videos and infographics, which could be reviewed repeatedly at their convenience. A recent systematic review and meta-analysis by Morgado et al. found that multimodal educational strategies can facilitate knowledge acquisition by presenting information through both visual and auditory channels [43]. These findings may help explain the observed improvement in knowledge scores in our intervention group. Furthermore, the intervention showed a more pronounced improvement in quality of life. Previous research has shown that pelvic floor disorders can substantially impair women’s quality of life by affecting daily activities, emotional well-being, and social participation [44]. In addition, the severity of urinary incontinence symptoms is closely associated with poorer health-related quality of life among affected women [45]. In our study, although both groups reported reduced self-perceived severity of incontinence symptoms after treatment, no significant between-group difference was observed. Nevertheless, the intervention group demonstrated a significantly greater improvement in quality of life, suggesting that the integrated intervention may be associated with additional psychosocial benefits beyond symptom relief. One possible explanation is that the group-chat platform provided social and emotional support through its group-chat platform, which facilitated open communication and connection among patients and healthcare providers. As Toye and Barker have highlighted, women with urinary incontinence often experience stigma leading to concealment of symptoms and reluctance to seek help [46]. By offering a safe and supportive online space, the intervention may have reduced feelings of isolation and self-stigma, thereby potentially enhancing patients’ psychological well-being and overall quality of life. However, because psychosocial variables such as social support or stigma were not directly measured in this study, this interpretation remains speculative and further research is needed to clarify the specific mechanisms through which such interventions may influence quality-of-life outcomes. Beyond its clinical and psychosocial associations, the digital service package also demonstrated acceptable feasibility in terms of usability, acceptability, and ease of use. The mean SUS score of 81.40 corresponds to a grade of “B” suggesting that this package was well-received by community-dwelling postmenopausal women. This positive outcome is likely attributable to a design that proactively addressed known barriers to technology adoption among older adults. Existing research has highlighted that age-related factors—such as cognitive load, visual limitations, and digital literacy—can influence the usability of digital health technologies among older populations [47]. Our digital package was designed to counteract these specific challenges by employing age-friendly features such as large fonts, clear images, and video-based content delivered by experienced gynecologists. Regarding ease of use, participants reported a mean After-Scenario Questionnaire score of 5.67, indicating that most users could efficiently complete training tasks under digital guidance. The highest subdimension score—task completion efficiency—suggests that the digital service package was time-efficient and convenient for home-based rehabilitation. However, the relatively lower satisfaction with the support information indicates that the current knowledge library may not fully meet the diverse and complex needs of community-dwelling postmenopausal women with PFD. Future iterations could consider integrating artificial intelligence—based features to offer dynamic, on-demand answers. The NPS of + 32% further reflected a favorable level of user acceptance. Prior evidence has shown that trust, perceived usefulness, and perceived ease of use are important determinants of users’ behavioral intention to adopt digital health services [48]. In this study, the favorable usability and ease of use of the digital package likely contributed to participants’ willingness to recommend it. Furthermore, the instructional videos featured familiar community healthcare professionals, and online support was provided directly by the participants’ own gynecologists. This design likely enhanced users’ trust in the intervention. Taken together, these findings suggest that the digital service package is a feasible intervention for community-dwelling postmenopausal women with PFD. The findings related to the prespecified exploratory margin should be interpreted cautiously. Although the observed confidence interval bounds were above the prespecified exploratory margin, the margin of − 0.20 was relatively wide and was adopted in the absence of an established, clinically validated margin for this specific community-based pelvic floor rehabilitation setting. The non-randomized design further limits the ability to exclude clinically meaningful differences between groups. Therefore, the present findings should be considered preliminary and hypothesis-generating rather than definitive evidence of non-inferiority, equivalence, or causal effects. The establishment of validated, clinically relevant non-inferiority margins for pelvic floor rehabilitation trials remains an area for future research. Our study has several strengths. A key strength is the study’s real-world community setting, which supports external validity—unlike controlled hospital-based trials, our findings may be more applicable to routine community healthcare practice [49]. Additionally, the use of doubly robust estimation, combining propensity score weighting and regression adjustment, helps reduce bias from measured confounders in this non-randomized design and represents a methodological advantage over conventional regression methods in observational studies [50]. However, as noted in the limitations, residual confounding from unmeasured variables cannot be excluded, and the findings should be interpreted as exploratory rather than confirmatory of causal effects. Several limitations should also be acknowledged. First, the non-randomized design and voluntary uptake of the digital service package may have introduced selection bias and residual confounding despite statistical adjustment. Women who chose to participate in the digital intervention may have differed from controls in motivation, treatment adherence, and digital literacy, each of which could independently influence rehabilitation outcomes. Second, the 1-month follow-up only captures short-term outcomes and does not allow assessment of durability. Third, the study compared a multicomponent community-based program with tertiary hospital usual care; therefore, the independent effect of the digital package cannot be isolated. Fourth, the prespecified non-inferiority margin was relatively wide and should be interpreted cautiously. Fifth, the Modified Oxford Scale is examiner-dependent, and formal inter-rater reliability was not assessed. Sixth, the pelvic floor knowledge questionnaire was self-developed and lacked full psychometric validation. In addition, standardized quantitative metrics of user engagement (e.g., login frequency, completed modules, chatbot usage counts, or response times) were not prospectively collected, limiting our ability to examine dose–response relationships between engagement levels and rehabilitation outcomes. Finally, the study did not include formal implementation, scalability, or cost-effectiveness evaluation. Conclusions This exploratory study provides preliminary evidence that integrating a digital health service package into community-based pelvic floor rehabilitation is feasible. The observed pelvic floor muscle strength outcomes in the intervention group were compatible with the prespecified exploratory margin when compared with tertiary hospital care, and the intervention was also associated with higher self-reported adherence, health knowledge awareness, and quality of life. However, given the non-randomized design, retrospective registration, short follow-up, the exploratory nature of the prespecified margin, and other limitations, these findings should be interpreted as preliminary and hypothesis-generating rather than as confirmatory evidence of non-inferiority, equivalence, or causal effects. Further well-designed randomized trials with clinically validated margins and longer follow-up are needed. Supplementary Information Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 4. Acknowledgements We would like to thank all the women who participated in this study for their time and cooperation. We also extend our gratitude to the clinical staff at the Xianxia Community Health Service Center and Shanghai Tongren Hospital for their support during the data collection process. Authors’ contributions QY and QH were major contributors in performing the investigation, formal analysis, data curation, visualization, and writing the original draft. WZ, QC, and JZ contributed to the investigation and resources. JZ also participated in data validation. JH and YL were responsible for the supervision and funding acquisition, and were involved in the writing—review and editing of the manuscript. JH also contributed to data validation , while YL was responsible for the conceptualization, methodology, and project administration. All authors read and approved the final manuscript. Funding This work was supported by the National Natural Science Foundation of China (grant number 72304071); the Fudan University School of Nursing Research Fund (grant number FNF202356);and the Shanghai Municipal Health Commission of Changning District (grant number 20234Y028). Data availability Due to restrictions on data management, the datasets analyzed during the current study are not publicly available but are available from the corresponding author upon reasonable request. Declarations Ethics approval and consent to participate This study was performed in accordance with the Declaration of Helsinki and received ethical approval from the Institutional Review Board of Shanghai Changning District Center for Disease Control and Prevention (Approval Number: 2023-18). The study was registered at the Chinese Clinical Trial Registry (ChiCTR2400088675) on 23 August 2024. Although registration occurred after the start of recruitment (retrospective registration), the study protocol, outcomes, and statistical analysis plan were finalized before any outcome data analysis was undertaken. Written informed consent to participate in the study was obtained from all participants prior to their inclusion in the study. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Qinyuan Yang and Qingyang Hu contributed equally to this work. Jiabei He and Yan Liang contributed equally to this work as co-corresponding authors. Contributor Information Jiabei He, Email: 843641501@qq.com. Yan Liang, Email: liangyan@fudan.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 4. Data Availability Statement Due to restrictions on data management, the datasets analyzed during the current study are not publicly available but are available from the corresponding author upon reasonable request.