Abstract
OBJECTIVE: To identify baseline clinical and demographic characteristics associated with clinically important treatment responses in a randomized trial of nonsurgical therapies for fecal incontinence (FI). METHODS: Women (N = 296) with FI were randomized to loperamide or placebo- and manometry-assisted biofeedback exercises or educational pamphlet in a 2 × 2 factorial design. Treatment response was defined in 3 ways from baseline to 24 weeks: minimal clinically important difference (MID) of -5 points in St. Mark's score, ≥50% reduction in FI episodes, and combined St. Mark's MID and ≥50% reduction FI episodes. Multivariable logistic regression models included baseline characteristics and treatment groups with and without controlling for drug and exercise adherence. RESULTS: Treatment response defined by St. Mark's MID was associated with higher symptom severity (adjusted odds ratio [aOR] 1.20, 95% confidence interval [CI] 1.11-1.28) and being overweight vs normal/underweight (aOR 2.15, 95% CI 1.07-4.34); these predictors remained controlling for adherence. Fifty percent reduction in FI episodes was associated with the combined loperamide/biofeedback group compared with placebo/pamphlet (aOR 4.04, 95% CI 1.36-11.98), St. Mark's score in the placebo/pamphlet group (aOR 1.29, 95% CI 1.01-1.65), FI subtype of urge vs urge plus passive FI (aOR 2.39, 95% CI 1.09-5.25), and passive vs urge plus passive FI (aOR 3.26, 95% CI 1.48-7.17). Controlling for adherence, associations remained, except St. Mark's score. DISCUSSION: Higher severity of FI symptoms, being overweight, drug adherence, FI subtype, and combined biofeedback and medication treatment were associated with clinically important treatment responses. This information may assist in counseling patients, regarding efficacy and expectations of nonsurgical treatments of FI.