Assessment of Perinatal Pelvic Floor Dysfunction Symptoms and Risk Factors: Measurement Properties of the International Consultation on Incontinence—Perinatal Pelvic Health Self‐Assessment Questionnaire (ICIQ‐PPHSAQ)

ABSTRACT

Objective

To assess the measurement properties of the International Consultation on Incontinence—Perinatal Pelvic Health Self‐Assessment Questionnaire (ICIQ‐PPHSAQ) and to identify symptom subscales for scoring.

Design, Setting and Population

Two cohorts of perinatal women (antenatal and postnatal) at nine NHS trusts in England completed ICIQ‐PPHSAQ at baseline (16–19 weeks' gestation for the antenatal cohort and 6–9 weeks postnatal for the postnatal cohort) and at up to three further timepoints, with follow‐up extending to 9 months postnatal.

Methods

Descriptive analyses and exploratory factor analyses (EFAs) identified symptom subscales from which subscale scores were calculated. Internal consistency was evaluated using Cronbach's alpha (α). Test–retest reliability was assessed using an interval of 1 week after baseline, using weighted Kappa (κ) for individual ordinal items and the intraclass correlation coefficient (ICC) for subscale scores. Construct validity and known‐groups validity were assessed by comparison with the Patient Global Impression of Severity (PGI‐S). Responsiveness was assessed by examining the mean differences between baseline and each timepoint.

Results

A total of 162 antenatal women who were 16–19 weeks gestation, and 173 women 6–9 weeks postnatal were recruited. Almost all the items performed acceptably in the test–retest analyses and completion rates were high in all the scored items (< 5% missing data). EFAs derived nine symptom subscales; seven of which had acceptable overall reliability indicators, except the vaginal symptom and vaginal changes subscales. Known‐groups validity was demonstrated for all nine subscales, and six out of nine showed responsiveness to change (p < 0.05) over the study time‐period.

Conclusion

The study provides evidence supporting the validity and reliability of ICIQ‐PPHSAQ, with responsiveness demonstrated across several domains. The instrument can be used to assess perinatal pelvic floor dysfunction risk factors and symptoms in English perinatal pelvic health services.

1. Introduction

Pregnancy and childbirth are well‐recognised risk factors for pelvic floor dysfunction (PFD), a term that encompasses a range of conditions that include urinary and bowel symptoms, pelvic organ prolapse, sexual dysfunction and chronic pelvic pain [1]. Some of these symptoms may initially have relatively little impact but can be underreported due to embarrassment, a belief that they will resolve spontaneously, and the assumption that they are a natural consequence of childbirth [2, 3]. If unaddressed, the physical and emotional impact and often long‐term persistence of PFD symptoms can be significant [4, 5, 6]. It has been found that of women who experience urinary incontinence 3 months postnatal, 76% still report persistent symptoms 12 years later [5] and 19% of women experience anal incontinence symptoms 1 year after birth [7].

As part of the NHS England Maternity Transformation Programme, Perinatal Pelvic Health Services (PPHS) were established in 2021 throughout England to improve the prevention, identification, and access to treatment for pelvic health problems [8]. A key element of this initiative was the development of a validated patient‐reported tool to enable women to monitor themselves throughout their maternity journey, allow early identification of those women with problems, and inform referral to pelvic health clinics, where precise assessment and management would occur. Many tools for the identification of pelvic floor dysfunction exist, although they vary in terms of their comprehensiveness and specificity to the perinatal population [9]. Notably, although initially developed for a general gynaecological patient population, the Australian Pelvic Floor Questionnaire (APFQ) has been widely used, translated into multiple languages, and recently validated in an Irish obstetric population [10, 11]. An adaptation of the APFQ in German also includes the assessment of risk factors to provide a more integrated evaluation of perinatal outcomes [12]. However, given the intended context of use within English national maternity services, it was considered important that items included in the measure were initially derived from a perinatal population within this healthcare system. The International Consultation on Incontinence—Perinatal Pelvic Health Self‐Assessment Questionnaire (ICIQ‐PPHSAQ) incorporated content from existing high‐quality tools [13, 14, 15], and was developed through co‐development with women from the English maternity pathway and clinical experts [16]. The aim of this study was to identify symptom subscales of the ICIQ‐PPHSAQ, and to evaluate its measurement (psychometric) properties including its validity, reliability and responsiveness.

2. Material and Methods

Two participant cohorts of perinatal women were recruited at NHS trusts in England, between 28th September 2023 and 31st October 2023. Cohort 1 at baseline was antenatal women at 16–19 weeks gestation; cohort 2 were women who were 6–9 weeks postnatal. Recruiting antenatal and postnatal cohorts allowed sufficient data to be obtained at the key postnatal timepoint (6–9 weeks) for psychometric analyses, while also enabling examination of symptom changes across the maternity pathway.

ICIQ‐PPHSAQ comprises 3 questionnaires that are designed to be completed at different times throughout the maternity pathway. These questionnaires were administered electronically using the platform REDCap at three timepoints (T): T1, T2 and T3 [17]. For cohort 1, T1 corresponded to baseline (16–19 weeks' gestation), followed by T2 and T3 during pregnancy and the postnatal period. For cohort 2, baseline corresponded to 6–9 weeks postnatal (Figure 1). In addition, the Patient Global Impression of Severity (PGI‐S) was administered to support the assessment of construct validity. The wording of all questionnaire items and their descriptive statistics are provided in Tables S1–S3. The following instruments were administered:

1. ICIQ‐PPHSAQ Antenatal: 13 items assessing absence and/or severity of antenatal risk factors (e.g., obstetric history).

2. ICIQ‐PPHSAQ Postnatal: 10 items assessing the presence, absence and/or severity of postnatal risk factors (e.g., birth events).

3. ICIQ‐PPHSAQ Symptoms: 39 items over four sections comprising ‘bladder’ (12 items); ‘bowel’ (12 items); ‘vagina/prolapse’ (9 items); ‘sexual activity’ (6 items). Responses to the items are on a 4 or 5‐point symptom severity Likert scale (e.g., from “never” to “always”). Every item also has an associated ‘bother’ item asking ‘how much did this bother you’ on a scale of 0 to 10. Finally, after each section, there is an item assessing overall interference with everyday life on a scale of 0 to 10.

4. Patient Global Impression of Severity (PGI‐S) [18]: A single item was administered after each section of the ICIQ‐PPHSAQ symptoms: “Overall, please indicate the one number that best describes your symptoms now; “no symptoms” (1), “mild” (2), “moderate” (3), or “severe” (4).”

FIGURE 1.

Data collection recruitment and administration timepoints.

2.1. Statistical Analysis

All statistical analyses were carried out using R [19]. Descriptive statistics were used to summarise all sociodemographic and item‐level data at all administrations, with continuous variables presented as mean, standard deviation, and range (min‐max), and categorical variables as frequency and percentage (n [%]). Psychometric analyses and scale derivation were undertaken for the symptom questionnaire, while the antenatal and postnatal risk factor questionnaires were evaluated descriptively and through test–retest analyses only. Symptom items were considered as having a floor effect or ceiling effect if skewed to the minimum or maximum response respectively.

2.2. Subscale Identification and Item Reduction

Exploratory factor analysis (EFA) was used to identify sets of related symptom items within each section that could be grouped together to form subscales for scoring [20, 21]. Each section (bladder, bowel, vagina/prolapse, and sexual activity) was analysed separately. Polychoric correlations for ordinal data were used to determine whether the relationship between questions was strong enough to support EFA [22]. The factorability of the data was determined using a cut off based on a measure of sample adequacy (Kaiser‐Mayer‐Olkin statistic (KMO) ≥ 0.60), correlation adequacy (Bartlett's Test of Sphericity, p‐value < 0.05) [22], and sample size of at least 10 respondents per variable, recommended for EFA of ordered (Likert scale) variables [23]. To determine the number of factors for each section, parallel analysis was used with the principal factor solution and the review of scree plots. Models with the number of factors suggested by the above criteria were fitted using the principal axis factoring method and oblique rotation to allow for correlations between factors. In addition to examining for poor factor loadings (< 0.3), communalities (the proportion of each variable's variance explained by the factors), test–retest performance, and clinical utility were considered when determining whether items should be removed from the scoring or retained in the final tool. Multiple iterations of EFA were performed, with the final model chosen based on acceptable criteria of model interpretability. The final model's fit was assessed using Tucker Lewis Index of factoring reliability (TLI) and root mean square of the residuals (RMSEA) with values > 0.9 and < 0.08 respectively indicating a good fit [22]. The following reliability analyses used factor‐based summative scores for each of the identified subscales.

2.3. Test–Retest Reliability

Test–retest reliability was assessed using an interval of 1 week after baseline for each cohort, in light of rapid changes that may occur during pregnancy or postnatal. Test–retest analyses were also undertaken for the antenatal and postnatal risk factor items to assess the stability of participant responses over a short interval. The weighted Kappa coefficient with squared weights was calculated for each item [24]. Intraclass correlation coefficients (ICC) with 95% confidence intervals were calculated for each subscale [25]. Values of the Kappa statistic (κ) were interpreted as follows: 0 = poor, 0.01–0.20 = slight, 0.21–0.40 = fair, 0.41–0.6 = moderate, 0.61–0.8 = substantial, and 0.81–1 = almost perfect agreement [26]. The ICC ranges from 0 to 1 with the generally accepted criterion for adequate reliability as ≥ 0.7, and excellent as > 0.9 [21].

2.4. Internal Consistency

Cronbach's alpha (α) was calculated for each subscale. Higher values indicate that the items are more likely to measure the same characteristic, or underlying construct. The broad interpretation of the α value is > 0.6–0.7 being questionable, > 0.7–0.9 being acceptable [21].

2.5. Construct Validity (Known‐Groups Comparison)

The distribution of factor scores across categories defined by the PGI‐S was examined using the Kruskal‐Wallis test. A p‐value < 0.05 was considered statistically significant [27].

2.6. Responsiveness

Distributions of the subscale scores were compared for cohort 1 at baseline, 14 weeks later (T2), 30 weeks later (T3), and for cohort 2 at baseline, 18 weeks later (T2), and 30 weeks later (T3). For each cohort, mean differences of the scores between baseline and each timepoint were calculated. The standardised response mean for each mean difference was calculated to assess internal responsiveness of the factor‐derived constructs. Effect sizes of 0.2, 0.5 and 0.80 or greater have been proposed to represent small, moderate and large degrees of responsiveness respectively [28].

2.7. Ethics Statement

This study was conducted in accordance with the ethical standards of the national research committees and with the Declaration of Helsinki. The protocol and amendments were approved by an independent Research Ethics Committee. Reference 23/LO/0531.

3. Results

3.1. Study Population

A total of 162 antenatal women were recruited (16–19 weeks gestation) in cohort 1, and 173 women in cohort 2 (6–9 weeks postnatal) across nine geographically diverse NHS trusts in England. For the EFA and reliability analysis, cohort 1 and cohort 2 participants were combined at their common timepoint: 6–9 weeks postnatal (n = 285). This timepoint corresponds to the 3rd administration for cohort 1 (30 weeks after baseline) and to the baseline questionnaire for cohort 2 and was selected because it provided the largest combined sample and represented an early postnatal stage at which a broad range of pelvic floor symptoms was expected. Median age was 30–34 years, and respondents were White British (87%), Asian British (7.4%), Black British (2.5%) or other (3.1%).

3.2. Item Descriptive Statistics

At the combined common timepoint (Tables S1–S3), missing data for the ICIQ‐PPHSAQ items in all three questionnaires was < 5%. Exceptions to this in the symptom questionnaire were ‘staining of the underwear/use of pads’ (6.7%) and the overall interference on everyday life items in the ‘sexual activity’ and ‘vaginal’ sections (19% and 9% respectively) (Tables S1–S3). Missing data for the ‘associated bother’ questions were also < 5%, once responses indicating no experience of the condition (e.g., ‘Never’) were recoded to ‘0’.

The distribution of responses was skewed towards the least severe response options in almost all the symptom questionnaire items, indicating the presence of floor effects. However, the 4th or 5th response option (highest severity) was selected by respondents for all items (Tables S1–S3). There were five symptom items that had < 10% of participants responding with any other response option than ‘not at all’/‘never’. These were flagged for possible removal alongside other psychometric and clinical utility considerations. The mean bother scores were also generally low and ranged from 0.1 to 3.4 (out of 10).

3.3. Exploratory Factor Analysis and Item Reduction

The initial EFA for the ‘bladder’ symptoms section suggested a two‐factor solution with an acceptable sample adequacy KMO of 0.81. The item ‘Do you have a burning feeling when you urinate?’ had low polychoric correlations and cross‐loaded onto multiple factors and was removed following further consideration of clinical utility. The final two‐factor solution had no cross‐loadings of items across factors and all factor loadings > 0.40 (Table S4). The Tucker‐Lewis Index was 0.93 and RMSEA index 0.072.

For the ‘bowel’ section the initial EFA suggested a two‐ or three‐factor solution with an acceptable sample adequacy KMO of 0.81. The decision to remove the item ‘Do you experience any staining of your underwear or need to wear pads because of your bowels?’ was made based on poor loading (< 0.3) and communalities, alongside its very low prevalence. The item ‘Are you having any bowel problems such as leakage of faeces, wind or constipation?’ loaded onto multiple factors and was highly correlated with two other items, so it was removed. The item ‘Do you notice blood in your stool?’ also had poor loading and communalities, so it was removed from the scoring, but retained in the final tool due to its clinical utility as a ‘red flag’ symptom. The final accepted solution (n = 268) was 3‐factor, with a Tucker‐Lewis Index of 0.99 and RMSEA index of 0.026 (Table S5).

The initial EFA for the ‘vaginal’ section suggested a two, or three‐factor solution with an acceptable sample adequacy KMO of 0.64. The removal ‘Are you aware of dragging sensation in?’ ‘your lower abdomen?’ and ‘Are you aware of soreness in your vagina?’ was explored, due to weak loading onto factors; however, the decision to retain these items in the final tool was due to their clinical relevance. The fit measures were still acceptable for the final two‐factor solution: Tucker‐Lewis Index = 0.86 and RMSEA index = 0.077 (Table S6).

The ‘sexual activity’ section suggested a two‐factor solution with a marginally unacceptable sample adequacy KMO of 0.50, however the correlation adequacy suggested factor analysis was still useful with the data. The item ‘Have you ever had a sexual experience that you consider to be harmful?’ showed poor loading and communalities so was removed from the scoring, but retained as a standalone item due to its potential for clinical utility. The final two‐factor solution showed reasonable fit measures of Tucker‐Lewis Index = 0.90 and RMSEA = 0.084 (Table S7).

3.4. Test–Retest Reliability

The weighted kappa coefficients provided evidence of substantial agreement or better (κ ≥ 0.61) for all items, except six items in the symptom questionnaire that had moderate agreement (κ = 0.41–0.61). Two items were flagged with ‘fair’ agreement (κ = 0.21–0.40): ‘Since the birth of your baby, have you had a dragging feeling or bulge of the vagina?’ and ‘Are you able to control watery or loose stool leaking from your back passage?’.

A total of 5/9 factor‐based subscales ICCs within the symptom questionnaire were adequate or excellent (≥ 0.7) (Table 1). Two ‘bowel’ section subscales had slightly lower reliability at 0.68, the ‘vaginal changes’ subscale was 0.69, and the ‘Problems with sex’ subscale was 0.65.

TABLE 1.

Descriptive statistics and psychometric properties of the subscale scores of the ICIQ‐PPHSAQ symptom questionnaire (6–9 weeks postnatal n = 285).

Subscale (possible score range)	Observed score range	Mean score (SD)	Test–retest items, Kappa (range)	Test–retest, ICC (95% CI)	Internal consistency Cronbach's α (95% CI)
Bladder leakage (0–16)	(0, 15)	2.4 (3.1)	0.74–0.82	0.80 (0.62, 0.90) a	0.84 (0.79–0.88)
Urinary habit (0–20)	(0, 14)	2.4 (2.1)	0.54–0.86	0.90 (0.81, 0.95)	0.66 (0.54–0.75)
Bowel symptoms (0–20)	(0, 14)	5.3 (3.1)	0.67–0.82	0.78 (0.60, 0.88)	0.72 (0.67–0.77)
Bowel leakage (0–12)	(0, 12)	1.6 (2.1)	0.25–0.68	0.68 (0.39, 0.84)	0.63 (0.55–0.70)
Bowel urgency (0–7)	(0, 6)	1.5 (1.5)	0.70–0.76	0.68 (0.22, 0.87)	0.68 (0.59–0.74)
Vaginal symptoms (0–12)	(0, 11)	1.8 (2.0)	0.64–0.75	0.80 (0.61, 0.89)	0.58 (0.49–0.66)
Vaginal changes (0–14)	(0, 11)	1.1 (1.6)	0.45–0.70	0.69 (0.31, 0.86)	0.58 (0.49–0.65)
Problems with sex (0–7)	(0, 4)	0.2 (0.6)	0.57–0.70	0.65 (0, 0.91) b	0.70 (0.59–0.78)
Leakage with sex (0–6)	(0, 6)	1.5 (1.7)	0.68–0.81	0.80 (0.56, 0.91)	0.73 (0.63–0.8)

Note: Kappa values are shown as the range across individual items within each subscale. ICC values represent the test–retest reliability of the summed subscale score and are presented with 95% confidence intervals (CI).

^a Data for ‘Do you have a sudden need to rush to the toilet to pass urine?’, in the bladder leakage’ subscale was not available for cohort 2 at the combined time point, so it was excluded from the ICC calculation.

^b Small n (degrees of freedom = 9), so the estimate is unstable.

3.5. Internal Consistency

The internal consistency (Cronbach's α) for each of the subscale scores was acceptable or good (≥ 0.7–0.9) for the ‘Bladder leakage’, ‘Bowel symptoms’, ‘Problems with sex’, and ‘Leakage with sex’ subscales. ‘Urinary habit’ and Bowel leakage’ were questionable (α ≥ 0.6–0.7), while the vaginal symptom subscales were poor (0.58) (Table 1).

3.6. Construct Validity (Known‐Groups)

Standardised factor scores differed significantly across PGI‐S severity categories for all subscales (p < 0.01) (Table 2).

TABLE 2.

Known groups validity by categories defined by the PGI‐S.

Subscale (range)	Factor score by PGI‐S category (median, [IQR])
	No symptoms a	Mild a	Moderate/severe a
Bladder leakage (0–16)	−0.73 (0.31)	0.43 (0.93)	2.31 (0.86)
Urinary habit (0–20)	−0.57 (0.38)	0.13 (0.94)	1.8 (1.55)
Bowel symptoms (0–20)	−0.6 (0.96)	0.42 (0.88)	1 (0.94)
Bowel leakage (0–12)	−0.38 (0.25)	−0.24 (0.56)	0.15 (1.02)
Bowel urgency (0–7)	−0.51 (0.98)	0.11 (1.31)	0.39 (1.66)
Vaginal symptoms (0–12)	−0.44 (0.74)	0.77 (1.28)	0.48 (1)
Vaginal changes (0–14)	−0.45 (0.51)	0.5 (1.19)	1.26 (1.62)
Problems with sex (0–7)	−0.75 (0.45)	0.49 (1.03)	1.79 (1.04)
Leakage with sex (0–6)	−0.25 (0)	−0.21 (0.26)	−0.18 (0.03)

^a PGI‐S Bladder; no symptoms n = 154, mild n = 109, moderate/severe n = 20. PGI‐S bowel; no symptoms n = 139, mild n = 106, moderate/severe n = 38. PGI‐S‐vaginal; no symptoms n = 210, mild n = 60, moderate/severe n = 12. PGI‐S‐sexual activity; no symptoms n = 206, mild n = 56, moderate/severe n = 14.

3.7. Responsiveness

Six out of nine subscales demonstrated responsiveness to change by showing significant symptom score changes (p < 0.05) over the study time‐period (Table S8). In particular, ‘urinary habit’ (including items such as daytime and nighttime urinary frequency) showed an increase in severity during pregnancy (mean difference 1.3, effect size 0.7), before improving at 6–9 weeks postnatal (mean difference −2.6, effect size −1.0). ‘Bowel symptoms’ showed improvement postnatal at T2 and T3 (mean difference −1.2, effect size −0.5) and ‘Bowel urgency’ slightly worsened in the third trimester (mean difference −0.008, effect size −0.006). ‘Vaginal symptoms’ worsened during pregnancy at T2 (mean difference 0.3, effect size 0.2), and ‘vaginal changes’ showed a small improvement 9 months postnatal (mean difference −0.2, effect size −0.2). ‘Leakage with sex’ worsened during pregnancy (mean difference 0.3, effect size 0.2) and showed an improvement 9 months postnatal (mean difference −0.3, effect size −0.3). The ‘Bowel leakage’ subscale and ‘Problems with sex’ subscales did not reach statistical significance and ‘Bladder leakage’ was not assessed.

4. Discussion

4.1. Main Findings

This study provides evidence supporting the validity and reliability of the ICIQ‐PPHSAQ, with responsiveness demonstrated across several domains. A series of EFAs derived nine scored subscales for the symptom questionnaire, within the defined sections for ‘bladder’, ‘bowel’, ‘vagina/prolapse’ and ‘sexual activity’. Other unscored items were retained for their potential for clinical utility if they were considered ‘red flag’ items that may require a priority clinical action (e.g., ‘do you notice blood in your stool?’).

To assess the impact of symptoms comprehensively, it is also necessary to capture both the symptom and the extent to which it may affect their life [15]. ICIQ‐PPHSAQ is designed to do this through the associated bother questions, that are not included in the scoring but may be used to assess whether an individual is likely to request treatment for bothersome symptoms. The final questions at the end of each section are on a Scale of 0–10, and may also be used separately to give an indication of the overall interference on everyday life. The symptom questionnaire was reduced in length and finalised to an overall total of 35 items, plus associated bother items.

The antenatal and postnatal risk factor questionnaires performed well in test–retest analyses, and completion rates were high. These were not subject to full psychometric analysis and scale derivation, since there was no hypothesised underlying construct that would associate each of the very different items, so they were confirmed as 13 items and 10 items respectively.

4.2. Interpretation

Some symptoms such as stress urinary incontinence (51% at 6–9 weeks postnatal) were prevalent as expected [3, 5, 29]. However, most of the items had floor effects due to generally low reported symptom prevalence and severity. This was the expected symptom profile of this population [29, 30, 31]. Although the maximum total score was not achieved by any individual in any subscale, the observed score range for each of the subscales was much closer to the maximum possible subscale score, indicating that some individuals gave high severity responses to many of the items. Hence, we were not concerned by the observed floor effects being any indication of poor instrument sensitivity [32].

The test–retest analyses provided good evidence of the reliability of almost all of the individual items over the test period of 1 week, and that variability was not due to systematic differences among respondents [33]. The eight items that showed ‘moderate’ or ‘fair’ agreement were likely due to a low prevalence and imbalance on test–retest. Four out of nine subscales were marginally less than the acceptable ICC value of 0.7 (0.65–0.68). The 95% confidence intervals for several subscales were relatively wide, reflecting uncertainty in the reliability estimates and indicating that the intervals may include values consistent with lower reliability. This may be due to the inclusion of symptom items with low prevalence in the sample, and with greater day‐to day variability (e.g., faecal and coital incontinence).

Four out of nine subscales demonstrated good or acceptable internal consistency (α ≥ 0.7), with three further subscales slightly below conventional thresholds (α = 0.63–0.68). As Cronbach's α can be lower in scales with relatively few items, the polychoric correlations of the items were also assessed, along with the final measures of fit to ensure the factor acceptability. Given that other psychometric indicators were acceptable, these subscales are likely to be adequate for exploratory group‐level analyses [21]. The capability of all the subscale scores to detect differences in the severity of condition was also supported when compared with groups defined by the PGI‐S (demonstrating known groups validity), although the separation between PGI‐S categories was less distinct for some subscales, particularly bowel leakage and leakage with sex. Separately, it should be acknowledged that the vaginal symptom and vaginal changes subscales had α values below 0.6 (α = 0.58), indicating that these domains should be interpreted with caution, particularly if used for individual‐level interpretation [21]. Given the prevalence of symptoms was low for the items within these subscales, this may be due to vaginal prolapse symptoms not yet being symptomatic [31]. This has also been noted in other studies assessing postnatal prolapse symptoms [11].

In addition to being valid and reliable, a questionnaire that is intended for use in monitoring outcomes over time should be sensitive to change when it occurs. The instrument detected changes in subscale scores over the maternity pathway in six out of the nine subscales, with effect sizes generally small to moderate. For example, the increase in the ‘Urinary habit’ subscale score was consistent with the known increase in lower urinary tract symptoms prevalence during pregnancy; nocturia has been found to be almost four times more likely to be reported in the third trimester [34]. Despite lower reliability indicators, both the vaginal changes subscale and the vaginal symptoms subscale showed small responsiveness to change. However, despite having acceptable indicators of validity and reliability, the ‘Bowel‐leakage’ and ‘Problems with sex’ subscales did not show statistically significant changes in scores. This may be due to a low prevalence of the reported symptoms in the sample. As a result, changes in these symptoms may be difficult to detect, although they remain clinically important to assess when present.

The full exploration of the interpretation of the subscale scores across the maternity pathway requires further work beyond this study. This may include determining bandings of severity, thresholds for meaningful change, and minimally important differences, typically by using anchor and distribution based methods [35]. These will be important for the interpretation of change in scores over time. The predictive ability of the risk factor items is also yet to be determined. In the meantime, maternity services are encouraged to agree local protocols for categorisation and referral of responses based on expert clinical judgement, and in accordance with relevant national guidance. ICIQ‐PPHSAQ may be completed either electronically following migration to an existing electronic questionnaire administration platform, or by completion of the supplied paper version. ICIQ‐PPHSAQ is now advocated for national use in the PPHS, as outlined in the NHS England service specification [8]. It is recognised that the method of administration will need to be localised according to the availability of existing electronic infrastructure [36]. In addition, given the relatively low response rate of non‐white British and other groups of women, it is important to find solutions to any ethnic, cultural, literacy and language barriers to ensure the tool is equally accessible for completion by all. Further evaluation in more ethnically diverse populations and in non‐UK healthcare settings would also be valuable to confirm the wider applicability of the instrument and its scoring system.

4.3. Strengths and Limitations

Key strengths of the study include its geographically diverse NHS Trust sample, robust methods for evaluating measurement properties, examination of symptoms in a real‐world setting, and high response rates despite some anticipated longitudinal attrition. The achieved sample size was consistent with established recommendations for exploratory factor analysis, which typically advise recruiting 10–20 participants per variable [23, 37]. Statistically justifiable subscales were derived, representing an advance over other available questionnaires [11, 12], where the uni‐dimensionality of scales may be uncertain.

Several limitations merit discussion. Firstly, EFA and the main reliability analyses were undertaken using data from the combined 6–9 week postnatal assessment. While this provided the largest available dataset for deriving the symptom subscale structure, the factor structure was therefore established at a single stage of the maternity pathway. Although changes in subscale scores were examined across multiple timepoints through responsiveness analyses, formal evaluation of measurement invariance across antenatal and postnatal stages was beyond the scope of the present study. Further work using confirmatory approaches is therefore required to confirm the stability of the factor structure across the maternity pathway.

Secondly, it is acknowledged that convergent validity was not assessed by correlation with other questionnaires of known validity, although this was largely mitigated by the assessment of construct validity using known‐groups determined by the PGI‐S. Thirdly, it was not possible to evidence responsiveness of the ‘Bladder leakage’ subscale due to a late inclusion of the item for the assessment of ‘urgency’ in the data collection process. However, the individual items within this subscale and other psychometric properties were shown to be valid and reliable.

5. Conclusion

The study provides evidence supporting the validity and reliability of ICIQ‐PPHSAQ in a large sample in an English perinatal setting, with some evidence of responsiveness across several domains. The instrument can be used in perinatal pelvic health services to assess perinatal PFD risk factors and symptoms. The final instrument is available via the ICIQ website https://iciq.net.

Author Contributions

Conceptualisation: A.U., N.C., P.A. Methodology: A.U., N.C., P.A., K.A., J.N. Formal analysis: E.S., N.B. Investigation: A.F. Data curation: A.F., A.U. Patient and public involvement: J.E. Project administration: A.F. Supervision: A.U., N.C. Funding acquisition: A.U., N.C. (co‐leads), P.A. Writing – original draft: A.U. Writing – review and editing: All authors.

Funding

The National Health Service Commissioning Board (NHS England): Tender Reference C51688.

Ethics Statement

Ethical approval for this study was obtained from the Health Research Authority (HRA) and Health and Care Research Wales (HCRW) Research Ethics Committee (reference: 23/LO/0531; approved 16 August 2023).

Consent

All participants provided informed consent prior to participation.

Conflicts of Interest

Alan Uren and Anne Fee are employed by North Bristol NHS Trust, which hosts the ICIQ programme; the Trust may receive licensing income from use of the questionnaire. Nikki Cotterill is Chair of the ICIQ programme. Charles Knowles is Chief Medical Officer, co‐founder and shareholder of Amber Therapeutics Ltd, and Chief Medical Officer and shareholder of Enterika Ltd. He reports personal fees from Medtronic Inc, Exero Medical, Luna Therapeutics, Enteromed, and usMIMA, all outside the submitted work. All other authors declare no competing interests.

Supporting information

Table S1: Descriptive statistics of the PPHSAQ antenatal risk factor items (baseline cohort 1).

Table S2: Descriptive statistics of the PPHSAQ postnatal risk factor items and associated bother scores (baseline cohort 2).

Table S3: Descriptive statistics of the PPHSAQ symptom items and associated bother scores (combined cohort analysis set, 6 weeks postnatal).

Table S4: Final exploratory factor analyses of the PPHSAQ ‘bladder’ section (cohort 1 data 6–9 weeks postnatal n = 112*).

Table S5: Final exploratory factor analyses of the PPHSAQ ‘bowel’ section (combined cohort data n = 268).

Table S6: Final exploratory factor analyses of the PPHSAQ ‘vagina’ section (combined cohort data n = 206).

Table S7: Final exploratory factor analyses of the PPHSAQ ‘sexual activity’ section.

Table S8: Subscale scores, effect size and standardised response mean (± standard deviation) showing changes at the different interview time points for cohort 1: Baseline (n = 162), 14 weeks antenatal (T2, N = 115), 6–9 weeks postnatal (T3, N = 112), and cohort 2: Baseline (6–9 weeks postnatal; n = 173), 18 weeks postnatal (T2, n = 120), 30–33 weeks postnatal (T3, n = 120) (p‐value for t‐test for paired samples with Holm‐Bonferroni multiple comparison correction).

Uren A., Fee A., Abrams P., et al., “Assessment of Perinatal Pelvic Floor Dysfunction Symptoms and Risk Factors: Measurement Properties of the International Consultation on Incontinence—Perinatal Pelvic Health Self‐Assessment Questionnaire (ICIQ‐PPHSAQ),” BJOG: An International Journal of Obstetrics & Gynaecology 133, no. 10 (2026): 1876–1884, 10.1111/1471-0528.70236.

Data Availability Statement

The data supporting this study are not publicly available due to privacy or ethical restrictions.