Predictive modeling and threshold scores for care seeking among women with urinary incontinence: The short forms of the Pelvic Floor Distress Inventory and Urogenital Distress Inventory

Aims To further the interpretability of the Pelvic Floor Distress Inventory (PFDI‐20) and Urogenital Distress Inventory (UDI‐6) by (i) evaluating the ability of these measures to distinguish between women with urinary incontinence who do and do not seek care, (ii) defining PFDI‐20 and UDI‐6 threshold scores above which women with urinary incontinence seek care, and (iii) developing a predictive model for incontinence care seeking. Methods An observational study was conducted with two groups of women with urinary incontinence: 256 who had not sought care and 90 seeking initial care at a tertiary center. Sample sizes were based upon the prevalence of care seeking for urinary incontinence and the number of potential predictors for care seeking. Wilcoxon rank‐sum tests, receiver operating characteristics, and multivariable logistic regression were use to achieve the study aims. Results Women with urinary incontinence who sought care had higher median PFDI‐20 and UDI‐6 scores compared to non‐care seekers (73.96 vs. 16.67, P  < 0.0001, and 41.67 vs. 8.33, P  < 0.0001). A PFDI‐20 score of 33.33 (83.33% sensitivity and 79.30% specificity) had very good discriminatory accuracy in distinguishing care and non‐care seekers (AUC 0.886 ± 0.019 [95%CI 0.8518, 0.9254] P  < 0.0001). A UDI‐6 score of 25.00 (83.33% sensitivity and 83.59% specificity) had excellent discriminatory accuracy in distinguishing care and non‐care seekers (AUC 0.9025 ± 0.0190 [95%CI 0.8653, 0.9398] P  < 0.0001). A multivariable predictive model accurately identified 82.4% of care and non‐care seekers. Conclusions A PFDI‐20 score of 33.33 and UDI‐6 score of 25.00 provide meaningful benchmarks for care seeking among women with urinary incontinence. Neurourol. Urodynam. 35:949–954, 2016 . © 2015 Wiley Periodicals, Inc.