Predictors of Post-Prostatectomy Incontinence: A Multivariate Analysis
Brian F. Chapin, MD, Adam S. Feldman, MD, Sarah D. Psutka, Douglas M. Dahl, MD.
MGH, Boston, MA, USA.
BACKGROUND: Medical co-morbidities, disease pathology, previous treatment for obstructive uropathy, and peri-operative complications may play a role in continence outcomes after radical prostatectomy. Here, we wished to determine which of these variables may help to predict post-prostatectomy incontinence (PPI).
METHODS: Under IRB approved protocol, we performed a retrospective review of all patients with localized prostate cancer treated by laparoscopic radical prostatectomy between 2001 and 2005. Preoperative clinical data, operative technique and pathologic data were evaluated. Patients who had salvage prostatectomy, salvage radiation, androgen deprivation therapy and chemotherapy were excluded. Data analysis was limited to patients with minimum of 12 months follow-up (mean 32 months, range 12-69 months) for a total of 283 patients. Incontinence was patient-to-physician reported and described as any leakage requiring 1 or more pads per day. Age, BMI, PSA, clinical stage, history of TURP, preoperative erectile dysfunction (PreopEd), diabetes (Type I and II), hypertension, hyperlipidemia, coronary artery disease, post-operative urine leak and pathologic Gleason sum (Glsum) were analyzed in a univariate fashion using t-tests for continuous variables and chi-square tests for dichotomous and ordinal variables as predictors for incontinence. Univariate logistic regression analyses were used to assess all potential variables and then a multivariate logistic regression model was developed including all significant or clinically relevant covariates and confounders. The outcome of our model was incontinence.
RESULTS: The overall incontinence rate in the 283 patient cohort was 8.5% (24/283). There were 9 diabetic patients with ≥12 months follow-up. Of the diabetic patients 33% (3/9) had PPI. Among the variables analyzed on univariate analysis as predictors for PPI, a preoperative diagnoisis of diabetes was found to be the only significant variable (p= 0.0156). On multivariate analysis, when accounting for Age, BMI, Glsum, and PreopED, diabetes remained a significant predictor of post-operative incontinence (Odds Ratio [OR]: 7.94, [95% CI 1.45-38], p=0.010).
CONCLUSIONS: We demonstrate that patients with a preoperative diagnosis of diabetes incur a significantly higher risk of having post-prostatectomy incontinence defined as any leakage requiring protective pads. This should be considered in evaluation of the diabetic patient when considering options for treatment of prostate cancer.