EP381 - Identifying distinct patient profiles in pelvic floor physiotherapy: a retrospective cluster analysis

Purpose/Background: Patients attending pelvic floor physical therapy (PT) clinics often present with diverse demographic characteristics and complex co-morbidities, which increase the time and resources required for effective management. As demand for PT rises and workforce shortages persist, healthcare systems face growing strain, resulting in delayed access, poorer outcomes, and reduced patient satisfaction. Missed appointments further compound these inefficiencies, limiting the capacity to deliver timely and effective care. Given this variability in patient needs and engagement, understanding the factors that influence PT attendance is critical for achieving value-based care. Examining how patient characteristics relate to PT may provide insight into these factors. Therefore, this study aimed to identify distinct patient subgroups defined by shared demographic and clinical features and to examine their association with PT utilization.

Methods/Interventions: Data from 4,482 adults referred to the Stanford Pelvic Health Center between November 2021 and January 2025 who attended at least one PT session were retrospectively analyzed. Latent class analysis, guided by the Bayesian Information Criterion, was used to identify patient subgroups based on age, sex, cardiometabolic burden, chronic pain, gastrointestinal symptoms, psychiatric co-morbidities, clinical history and investigations. Posterior probability-weighted Poisson regression models were fitted to evaluate differences in PT utilization across subgroups.

Results/Outcomes: Latent class analysis identified six patient classes (Figure 1). Classes 1 and 6 consisted of older adults (mean ages 71 and 67 years) with cardiometabolic disease but contrasting co-morbidity profiles. Class 1 showed high rates of hypertension (80%) and diabetes (34%) along with a high prevalence of chronic (95%), pelvic (89%), and back pain (77%), whereas Class 6 had a similar age distribution (70% ≥65 years) but moderate cardiometabolic burden (60% hypertension, 22% diabetes) and minimal pain (chronic 8%, pelvic 1%). Average PT utilization was comparable between these groups (3.76 and 3.86 visits). Classes 2 and 3 comprised younger adults (mean ages 44 and 56 years) characterized by chronic pain (69% and 94%) and psychological comorbidities including anxiety (49% and 80%) and depression (26% and 62%). These classes demonstrated significantly lower PT utilization than Class 1 (2.98 and 2.61 vs 3.76 visits; p < 0.01). Classes 4 and 5 comprised middle-aged adults (mean ages 43 and 51 years) with moderate rates of pelvic floor symptoms, including urinary incontinence (63% and 59%), but differing patterns of comorbidity. Class 4 showed low gastrointestinal (18%) and psychological (20%) comorbidity and PT utilization similar to the older groups (3.64 vs 3.76 visits; IRR 0.97). In contrast, Class 5 had a higher prevalence of gastrointestinal symptoms (76% constipation, 42% irritable bowel syndrome) and lower attendance (3.33 visits; IRR 0.89; p < 0.01).

Conclusion/Discussion: Using latent class analysis in a large diverse group of patients who were seen for PT in a tertiary care academic center, six patient phenotypes were identified, differing in age, comorbidity, and symptom burden. Older groups demonstrated the highest PT utilization, whereas younger patients with pain and psychological comorbidity attended fewer sessions. Aside from age, presence of psychological comorbidities, and chronic pain appears to influence PT use and may guide more individualized models of care.

Figure 1 - Class profiles of pelvic floor PT patients
Heatmap displaying conditional probabilities of variables across classes identified through latent class analysis. P(Yes) represents the probability that a given characteristic is present within each class, with darker shades indicating higher values.