The Consolidated Standards of Reporting Trials (CONSORT) checklist

The Consolidated Standards of Reporting Trials (CONSORT) 2010 statement [22] was used to guide the reporting of this cluster-randomized controlled trial (cRCT). The CONSORT checklist consists of a 25-item checklist that focuses on the key elements for reporting the design, conduct, analysis, and interpretation of RCTs. An extension to the CONSORT statement for reporting cRCTs [23] was also consulted. We have addressed all applicable items from the CONSORT checklist in the reporting of this cRCT (S1 File).

Study design and participants

A community-based, parallel-groups, two-arm, cRCT was conducted among women with symptomatic POP stages I–III in D-W HDSS. First, a baseline survey was conducted to determine the burden and risk factors of POP and to identify eligible cases [24]. Women aged over 18 years with at least one symptom of POP, confirmed as stages I–III using the standard POP-Q measure [25], were included in the study. The identified cases were then randomized by community clusters into intervention and control arms. Cluster randomization was preferred to minimize spillover of the intervention to the control arm and due to the feasibility of providing the intervention at group level (nearby health facility). The clusters were kebeles which are the smallest administrative units in Ethiopia, similar to neighborhoods or wards. They have a wide range of population sizes, from 500 to several thousand residents [26]. The surface area of kebeles varies depending on their location and population density, with those in densely populated urban areas being smaller compared to rural kebeles. Each kebele is further divided into smaller units called gots or villages. Kebele administrations are responsible for providing basic services, such as health, education, and infrastructure, to their residents. They also play a crucial role in population-based surveys and research studies in Ethiopia, often serving as the primary sampling units [27,28]. Kebeles are organized into larger administrative divisions known as woredas (Districts).

Sample size calculations were performed using Open-Epi version 3.1, based on the following assumptions: a 99% confidence level, 80% power, an exposed-to-unexposed ratio of 1, a pelvic organ prolapse symptom score (POP-SS) mean difference of 3.16 with a standard deviation of 4.78 in the intervention group, and a mean difference of 0.12 with a standard deviation of 3.86 in the control group over six months [15]. This calculation yielded a required sample size of 96, adjusted for an anticipated 16% refusal and loss to follow-up [15], resulting in an individual-based total sample size of 111.34. This sample size was utilized for all outcomes since we were unable to determine the effect of PFMT on quality of life as measured by the Prolapse Quality of Life Questionnaire. The decision to adjust the confidence level to 99% was based on the adjusted level of significance to 0.0125 than the usual P-value of 0.05 to account for multiple comparisons across various outcomes (POP-SS and three distinct domains of P-QoL).

To determine the minimum number of clusters required, the individual-based sample size was multiplied by the intra-cluster correlation (ICC) coefficient. Since there was no reported ICC from earlier studies, we used an ICC of 0.03, which was suggested as an estimated fairly typical value of the ICC [29,30]. This calculation indicated a need for 1.67 clusters per arm. To ensure adequate power [31,32], a total of eight clusters were selected randomly from the 12 available clusters in the D-W HDSS.

After adjusting for cluster effects with a variance inflation factor of 1.39, the total effective sample size increased to 155, resulting in an expected average of 19.4 participants per cluster. However, during enrollment, a slightly varying number of eligible cases was identified in each cluster. Excluding additional participants was deemed unethical, and alternative treatment options could not be provided due to the cRCT design. Furthermore, it was believed that including all eligible cases would be manageable and enhance the study’s power [29,30]. Consequently, all eligible cases were included, raising the total sample size from 155 to 187 participants. The exclusion criteria were: women who were diagnosed to have symptomatic POP stages I–III, but planned to leave the study area within 6 months (study period), severely ill women, women with psychiatry disorders, women with delivered myoma, or women who want to have prolapse surgery within 6 months.

Randomization and masking

The study site, D-W HDSS, comprises eight kebeles from Dale district and four kebeles from Wonsho district. These were stratified into two strata based on their geographical location (Dale and Wonsho districts). Stratifying on location reduces stratum variability and helps to ensure that they are balanced across study arms [31,32]. To improve similarity across treatment and control groups, a balanced number of clusters from each stratum were assigned to each arm. First, all eligible clusters in each district were listed and assigned a unique identifier number. Then, a lottery method was used to select two of four clusters from the Wonsho district and randomly assigned to each arm. Similarly, six of eight clusters from Dale districts were selected and three clusters were assigned to each arm randomly by lottery method. Then, 102 households were selected from each cluster using computer-generated random numbers (based on the list of their house numbers) to identify eligible women. One woman (usually the head of the household) was recruited for the study.

To reduce the risk of spillover-effect of the intervention, the interventional and control clusters were separated by a buffer zone [31], and when it is impossible due to the limited number of clusters in the D-W HDSS, adjacent clusters were assigned to the same arm. Finally, all eligible study participants from each cluster identified during the baseline survey were included in the study (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Randomization and Sampling procedure used to assess the effect of pelvic floor muscle training on prolapse symptoms and quality of life in Sidama Region, Ethiopia, August 2024.

https://doi.org/10.1371/journal.pmed.1004468.g001

The enrollment period was open for 8 months from 13 March to 28 October 2023. Each participant was followed for 6 months after enrollment. The last participants were enrolled on October 28, 2023 so the final follow-up visit was on April 27, 2024. Allocation concealment was conducted by staff of D-W HDSS at Hawassa University. To ensure unbiased allocation, the lottery method was performed by an independent researcher who was not involved in the recruitment, intervention, or outcome assessment processes. Two staff members of Yirgalem Hamline Fistula Center, who performed pelvic examination to determine the POP stage, recruited eligible participants for intervention.

The participants in the intervention group were aware that they were receiving the PFMT program, but they were not aware of the control group. Similarly, the control group knew that they are receiving lifestyle counseling, but they are not aware of the intervention group. The same is true for the healthcare providers delivering the counseling. Moreover, the outcome assessors, who collected data at the end of intervention, were blinded to the participants’ treatment allocation.

Procedure

In Ethiopia, each kebele/cluster is served by its own health post, staffed by health extension workers. Additionally, groups of five or more kebeles are supported by a single health center, staffed by formally trained healthcare professionals, including midwives. For this study, the intervention was implemented at eight distinct health centers. The intervention was implemented at health centers due to their capacity to deliver the training and their geographical accessibility to participants, especially when compared to hospitals. In this study area, all health centers have at least three degree-holding midwives responsible for providing antenatal, delivery, and postnatal care. Each health center is equipped with various rooms, including delivery rooms, cervical cancer screening rooms, and outpatient rooms, which are suitable for conducting the Pelvic Organ Prolapse Quantification (POP-Q) [25] assessments and ensuring confidentiality during individual counseling sessions.

From each selected health facility, two female midwives were trained either on PFMT and how to counsel patients, or only on how to counsel for life style modification based on their pre-allocated health facility. The training that included education on theory, simulation, and video aids was given by a fellowship trained urogynaecologist. Then, the trained midwives administered the intervention to the women assigned in the intervention arms.

Women who were assigned to intervention received PFMT and prolapse lifestyle advice focusing on risk factors for POP such as constipation, chronic coughing, heavy lifting, and other medical conditions. The control group received only lifestyle modification. In the interventional group, the pelvic examiners link eligible women to the assigned midwife with the POP-Q findings. Then, at the initial visit, women were informed by midwives about the stage of POP they had and type of prolapse (anterior, posterior, or central) which was explained in a private room. Then, the advantage of pelvic floor muscle exercise, and how it works, the detail of the exercise and follow-up schedule were explained in detail at the group level in a private room for 40–60 min.

They were trained to sit or lie comfortably, train pelvic muscles with 10 times 10-s maximum holds, and 10 fast contractions [15,33]. They were instructed to repeat this at least three times per day (morning, afternoon, and night) [15,34]. Women were taught how to precontract the pelvic floor muscles against increases in intra-abdominal pressure (Knack exercise). They were trained to exercise 10 long squeezes and lift contractions, each held for 10 s, allowing time between the squeeze to relax and lower the muscle for up to 5 s. The long squeezes were followed by 10 short squeezes. Women were advised to try to feel the releasing and relaxing of the squeeze after 10 short squeezes, and finally, to do a squeeze with a light cough for 10 times. The long squeezes were for 10 s each and the 10 quick squeezes for 1–2 s each.

To help participants correctly identify and contract their pelvic floor muscles, the participants in the intervention group received PFMT with internal digital biofeedback administered by the trained midwives. During the individual training sessions, midwives used various verbal cues and analogies. For example, participants were instructed to “imagine you are trying to stop the flow of urine or hold back gas. Squeeze as though you are trying to slow down or stop your urine mid-stream” [35]. This relevant analogy helped participants visualize the specific action they needed to perform. Throughout the exercises, midwives provided descriptive feedback to guide participants. They would say things like, “As you squeeze, focus on lifting the muscles around your vagina and anus upwards. You should feel a tightening and lifting sensation” [33,35,36]. This encouraged participants to be aware of the specific muscles being engaged during the contractions. We translated the Lifestyle Advice Leaflet from the POPPY trial [15] Into Sidaamu Afoo, the local language, to enhance understanding and effectiveness during counseling sessions.

They were appointed to attend five visits (at weeks 0, 2, 6, 10, and 15) at a nearby health facility. The appointment frequency was adapted from the United Kingdom National Health Service (NHS) [37]. Initial appointments were scheduled closely together to reinforce correct exercise techniques and ensure participants fully understood all the advice provided. Subsequently, appointments were spaced further apart to promote independent home exercise. A 15-week intensive training was recommended to get adequate muscle strength [38]. The change in prolapse symptoms and HRQoL were analyzed at 6 months.

A lifestyle counseling was given for both arms focusing on the following points [15,36].

Outcomes

Our primary outcomes were POP symptoms and HRQoL. The POP-SS was measured by the Sidaamu Afoo version of the POP-SS questionnaire [39]. The psychometric properties of the questionnaire have been evaluated in the same setting and found to have acceptable reliability and validity. It is a patient-reported-outcome measure that contains seven items with five Likert-scale ranging from 0 to 4, with a possible total score of 0–28 (the higher score representing worse symptom) [40]. HRQoL was measured by using the prolapse quality of life (P-QoL) questionnaire translated from the Amharic version [41]. While the original version has 20 items with nine domains [42], in a previous work designed to adapt the P-QoL locally (to Amharic version), the dimensions have been condensed into three: physical function (including General health perception, prolapse impact, role, physical and social limitation, and severity measures); psychological function (including emotional disturbances and sleep/energy disturbances); and personal relationships. Each domain is related to a particular aspect of QoL, and scores in each domain range from 0 to 100, the higher score indicating poor quality of life. This tool has previously been validated in a similar context (northern Ethiopia).

The secondary outcome, women’s perceived change In prolapse, was evaluated using the patient global impression of change index (PGI-I), a seven-point Likert scale (1 =  very much better to 7 =  very much worse) [43]. In this study, patients were classified as improved if they scored 1, 2, or 3 on the PGI-I scale, no change if they scored 4, and worse if they scored 5, 6 or 7.

Intervention adherence was measured in terms of attendance at appointments [15] and the amount of exercise women reported at follow-up time. Women’s adherence to training protocol is considered adequate when they complete at least 80% of the training sessions given by midwives [44].

The details of the baseline data collection procedures and tools have been published elsewhere [24]. All tools were translated from English and Amharic into the local language (Sidaamu Afoo).

The baseline data about sociodemographic characteristics and prolapse symptoms were collected through face-to-face interviews with women at their homes by eight first-degree nurses using electronic data collection forms prepared using Kobo tool box. The women were subsequently invited for pelvic examination in a nearby health center to confirm their prolapse status and degree of prolapse by using the POP-Q [25]. The POP-Q was performed by staff members of Yirgalem Hamlin Fistula Center who have special training on Fistula and POP identification. Then, women with symptomatic POP stages I–III were interviewed about how much the prolapse had affected their quality of life using P-QoL tool. The detailed POP-Q procedures and data quality assurance method were published elsewhere [24].

At 3 and 6 months of intervention, four first-degree nurses went to the house of the study participants and collected the data through face-to-face interviews. The data collection tool included the POP-SS and P-QoL questionnaires used at baseline, PGI-I change, and questions used to assess their adherence to training protocol [44].

Statistical analysis

All data analyses accounted for the complex sampling design, which included stratification by district (Dale and Wonsho), clustering by kebele, and weighting for unequal probability of selection. The weighting variable was the product of cluster and household selection weights, normalized by dividing each unit’s survey weight by the unweighted average of the survey weights. Categorical variables were summarized as frequencies and percentages, while continuous variables were summarized using means and standard deviations or medians and interquartile ranges.

The analysis was based on the intention-to-treat principle. In practice, all participants adhered to their allocated treatment and completed the follow-up as scheduled. Although the intervention was randomly assigned at the cluster level, outcomes were evaluated at the individual level. An Individual-level analysis was performed using a mixed-effects generalized linear model (MEGLM), incorporating random effects to account for clustering by health centers and survey weights to adjust for unequal selection probabilities.

To complement the primary analysis, a cluster-level analysis was conducted following Hayes and Moulton’s framework. Cluster-level means were calculated for key outcomes (mean change in POP-SS, physical domain, psychological domain, and personal relationship) and analyzed using Mann–Whitney U test to compare intervention and control clusters (S5 File).

To compare the primary outcomes at 6 months, we fitted a multilevel MEGLM to the change from baseline in POP-SS and each domain of HRQoL at three and 6 months. The model included fixed effects for treatment group, time, and other relevant covariates (stage of prolapse, duration of prolapse symptoms, residence, parity, age, level of education, history of prolonged labor, wealth index, constipation, and chronic cough), as well as a random intercept for patients within centers and a random slope for time within patients. The relevance of the covariates was determined based on prior research, clinical aspect, and their potential impact on the intervention and outcomes (prolapse symptoms and quality of life). MEGLM was chosen for its flexibility and robustness needed to accurately model the complex and varied data structures often found in cRCT leading to more reliable and interpretable results [45].

The difference between intervention and control group was estimated by the mean change from baseline at 6 months. Bonferroni correction for multiple measurements [46] was used to adjust the level of significance to 0.0125 as we measured the effect of PFMT on four outcomes (POP-SS and three domains of HRQoL). Then a mean change difference between the intervention and control arms was presented with 99% confidence interval and P-values. This adjustment is crucial for maintaining statistical rigor and minimizing the risk of Type I errors in the analysis [46].

To evaluate model fit, we initially fitted a null model for each outcome variable (POP-SS and the three domains of Prolapse Quality of Life (P-QoL)), incorporating random effects for both cluster and participant ID. We calculated the intra-cluster correlation coefficient (ICC) to determine the proportion of variance in the outcome attributable to the clustering structure, which was found to be 4% and less in this study. Subsequently, we fitted models that included the intervention variable to the outcome variables individually. This step allowed us to assess the impact of the intervention on each outcome. We evaluated model fitness using the Pseudo Log-Likelihood values for model fitness and Adjusted Wald Test for significance of predictors. Higher (less negative) Pseudo Log-Likelihood values indicate better model fit and significant Wald test indicates the importance of the predictors [47]. This approach allowed us to identify the most appropriate models for each outcome. Residual analysis was done to assess homoscedasticity and normality of residuals by using the histogram. To check for effect modifiers, we employed statistical tests to assess interactions between PFMT and potential effect modifiers. This approach allowed us to determine whether the effect of PFMT varies across different subgroups. The data set on which the analysis based was provided as a S2 File.

Trial profile

A total of 187 eligible women were randomized to the intervention group (n =  86) or the control group (101). All of them completed the follow-up and completed the questionnaire at 3 and 6 months. The age of participants ranged from 19 to 80 years with a median of 36 years (IQR: 31–50). All were parous, with a median of 4 childbirths (IQR: 4–6). The median hours of daily heavy lifting work were 4 h (IQR: 3–5). The most common stage of prolapse was stage II 120/187 (64.17%). The median duration of feeling prolapse symptom was 48 (30–60) months with the minimum of 5 months and maximum of 20 years. At baseline, a slightly higher number of women in menopause, family history of prolapse, and history of home birth were observed in the intervention group. The mean baseline POP-SS and quality of life domain scores were almost similar across the groups. Almost half 92 (49.2%) of the participants had prolapse of multiple vaginal compartments; whereby 161 (86.1%) had anterior prolapse, 113 (60.43%) had posterior prolapse and 28/187 (14.97%) had apical prolapse. Overall, while the intervention and control arms were balanced in terms of most characteristics at baseline, there were some important characteristics that showed significant imbalances. Detailed sociodemographic characteristics are presented in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Sociodemographic and clinical characteristics of trial participants at Baseline in Sidama Region, Ethiopia (August 2024) (n =  187).

https://doi.org/10.1371/journal.pmed.1004468.t001

Adherence and follow-up

All study participants adhered strictly to their allocated treatment and completed the follow-up as scheduled. No participants changed their allocation, missed follow-up appointments, or failed to complete the treatment regimen. Consequently, while our analysis framework was based on the intention-to-treat principle, the actual adherence and follow-up were perfect, ensuring that all participants were analyzed according to their initial allocation.

Prolapse symptoms and quality of life at 6 months

A summary at the cluster level reveals that all clusters experienced at least a minimum mean change of 1 point in the POP-SS score. Notably, a larger change was observed in the intervention clusters compared to the control clusters, suggesting a more significant impact of the PFMT intervention on participants in those areas (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Change in Pelvic Organ Prolapse Symptom Scores (POP-SS) and prolapse-quality of life domains by cluster after intervention, Sidama Region, Ethiopia, August 2024.

https://doi.org/10.1371/journal.pmed.1004468.t002

Women in the intervention group reported significant improvements in prolapse symptoms and across all domains of P-QoL at six months. In the unadjusted model, the most substantial improvement was observed in the psychological function component of the P-QoL, with a mean change difference of −11.18 (99% CI [−15.03, −7.32]; p < 0.001), indicating a large effect size. Moreover, a significantly higher number of women in the intervention group reported their condition as “better” after the intervention compared to the control group. In the control group, more than 27/101 (26%) of women became worse in 6 months (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Effect of pelvic floor muscle training on prolapse symptom and prolapse quality of life, Sidama Region, Ethiopia, August 2024.

https://doi.org/10.1371/journal.pmed.1004468.t003

Effect of covariates on pelvic organ prolapses and quality of life

We observed that women in the stages I and II prolapses experienced more marked improvements compared to stage III prolapse both in prolapse symptom and quality of life. Women with higher POP-SS scores, indicating more severe symptoms, experienced less improvement in the physical domain of quality of life compared to those with lower scores. The analysis showed that as the severity of prolapse symptoms increased, the associated improvements in the physical domain of the Prolapse Quality of Life (P-QoL) questionnaire were diminished. Moreover, after adjusting for the severity of prolapse symptoms and the stage of prolapse, the effect of PFMT on the physical domain of quality of life was slightly diminished but showed increased effects in the personal relationship and psychological domains (Table 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Adjusted effect of pelvic floor muscle training on prolapse symptoms and quality of life in Sidama Region, Ethiopia, August 2024.

https://doi.org/10.1371/journal.pmed.1004468.t004

Adverse events

No adverse events were observed.