Presentation Authors: Sophie Sohval, Shirly Solouki, Nitya Abraham*, Bronx, NY
Introduction: A recent study demonstrated that a validated model was more accurate in predicting de novo stress urinary incontinence (SUI) than both preoperative stress testing and expert prediction. However, this model was validated using a population that was 85% White. The study objective was to evaluate if this model correctly predicted de novo SUI after pelvic organ prolapse (POP) surgery in a diverse population.
Methods: This is a retrospective review of women without SUI who underwent POP surgery with or without a prophylactic incontinence procedure from January 2014 to January 2017. Charts were reviewed for demographic/clinical information and patient report of SUI up to 12 months after surgery. Patient characteristics were entered into the risk calculator and predicted risk was compared to actual outcome.
Results: 95 women without SUI underwent POP surgery during the inclusion period. 39 (48.2%) were Hispanic, 18 (22.2%) Black, 13 (16.1%) White, and 11 (13.6%) Other. 14 women developed de novo SUI (14.7%). Women with de novo SUI had a significantly higher BMI, smoking rate, and POPQ Aa point. The overwhelming majority of patients who developed de novo SUI were of Hispanic background (85.7%) There was no difference in the mean predicted percentage risk of de novo SUI after POP surgery without a concomitant incontinence procedure in women who developed SUI (36.2 95%CI 31.2-41.3) versus women who did not develop SUI postoperatively (33.0 95%CI 31.4-34.6). The majority of patients (85.7%) developed SUI within 6 months after surgery. Only 12 women underwent a prophylactic incontinence surgery at time of POP repair, of whom one patient developed de novo SUI, precluding comparison of these two groups.
Conclusions: A risk calculator predicting de novo SUI after POP surgery without an incontinence procedure validated using a primarily White population did not correctly predict de novo SUI in this diverse population. The racial/ethnic composition of data sets used to create predictive models may affect its application in certain patient populations. Additional studies in diverse populations are needed to confirm these findings.