Do features of randomized controlled trials of pelvic floor muscle training for postprostatectomy urinary incontinence differentiate successful from unsuccessful patient outcomes? A systematic review with a series of meta-analyses.

Abstract

Efficacy of pelvic floor muscle training (PFMT) for postprostatectomy incontinence (PPI) differs between randomized clinical trials (RCT). This might be explained by variation in content/delivery of PFMT (eg, biofeedback, muscles targeted, and time of commencement of training). This review investigated whether outcome of meta-analysis differs based on presence or not of specific RCT features. PubMed, CINAHL, EMBASE, Physiotherapy Evidence Database (PEDro), and Cochrane were searched for RCTs that investigated efficacy of PFMT on PPI. PFMT details and outcomes were extracted. Methodological quality and comprehensiveness of PFMT reporting was assessed using the PEDro scale and Consensus on Reporting Template, respectively. Effect size and 95% confidence intervals were calculated for incontinence rate at 3 months. To identify features that influenced efficacy, separate meta-analyses were performed for studies with and without specific features. Twenty-two trials were included, and 15 were used for meta-analysis of 3-month outcomes. Overall, meta-analysis showed reduced incontinence with PFMT (risk ratio [RR] = 0.85; I2 = 55%; P = .005). Meta-analyses showed better outcomes for PFMT than control groups when the study included preoperative PFMT (RR = 0.76; I2 = 30%; P = .006), biofeedback (RR = 0.73; I2 = 58%; P = .006), instruction to contract around the urethra (RR = 0.9; I2 = 43%; P = .009), a control group without PFMT instruction (RR = 0.85; I2 = 69%; P = .05), inclusion of all men irrespective of continence status (RR = 0.84; I2 = 30%; P = .003) and continence defined as no leakage (RR = 0.85; I2 = 48%; P = .05). Preoperative PFMT, biofeedback, urethral instructions, no PFMT instruction for controls, inclusion of all men, and continence defined as no leakage are features associated with successful patient outcomes. Future studies should consider these features in design of interventions and pooling data for meta-analysis.