The Urinary Incontinence Treatment Network (UITN) performed two large randomized comparative effectiveness trials studying surgical treatment of SUI in women. The first trial, Stress Incontinence Surgical Treatment Efficacy Trial (SISTEr), randomized 655 subjects to either Burch colposuspension or autologous rectus fascial sling in treatment of SUI. The second trial, Trial of Midurethral Slings (TOMUS), randomized 597 subjects to polypropylene midurethral slings placed either in the retropubic or transobturator approach. Primary outcomes for SISTEr have been published17
and will be available for TOMUS in the summer, 2009. Design papers are published for both trials18, 19
. This paper represents the analyses of the preoperative data collected from these two trials. World Health Organization definitions of BMI were used to define weight groups: obese, ≥30 kg/m2
, overweight, 25kg/m2
, and healthy weight, <25 kg/m2
Demographic variables reported included age, race/ethnicity, education, marital status, and occupational score. Continuous clinical variables included height, weight, BMI, specific parameters from the pelvic organ prolapse quantification examination (POPQ); the most prolapsed portion of the anterior vaginal wall [Ba]; the most prolapsed portion of the posterior vaginal wall [Bp]; and the genital hiatus [gh]), Q-tip test (delta angle), mean muscle strength (Brink) scores, 24-hour pad weight, incontinence episode frequency (IEF) from a 3-day bladder diary and general patient health score. Categorical clinical variables included prior UI surgery, prior prolapse surgery, prior hysterectomy, menopausal status, hormone replacement use (HRT), diabetes, and smoking status. Subjective measures included the Urogenital Distress Inventory (UDI), Incontinence Impact Questionnaire (IIQ), and the Medical, Epidemiologic and Social Aspects of Aging Questionnaire (MESA). Subjective categorical variables included responses to questions about physical accommodation, character of urine stream and fecal incontinence. Continuous urodynamic (UDS) variables included, valsalva leak point pressure (VLPP), intravesical pressure (Pves), intra-abdominal pressure (Pabd), bladder volume at first desire, bladder volume at strong desire, maximal cystometric capacity, and pressure-flow data (maximum flow rate [Qmax], Pves at Qmax, Pabd at Qmax, time to Qmax). The only categorical urodynamic variable was pressure-flow voiding pattern (normal or abnormal).
Analyses were carried out in parallel for the SISTEr and TOMUS subjects as the trials had different inclusion and exclusion criteria representing different populations. Continuous variables were summarized by mean and standard deviation (SD). Distributions of continuous measures were assessed for normality. Although the distribution of some measures were moderately skewed, we elected to conduct and report analyses in the natural scales for ease of interpretation. To investigate the bivariate relationships of demographic, clinical and UDS variables with BMI category, one-way analysis of variance (ANOVA) was used for continuous measures and cross-classification and Chi square test or Fisher’s Exact test for categorical measures as appropriate. In order to assess multi-colinearity among the multiple measures of incontinence a preliminary principal components analysis (PCA) was computed20
. The PCA indicated that there were 3 independent dimensions of stress incontinence. One dimension was weighted most heavily by the subjective measures composed of the MESA stress score, UDI stress and IIQ total scores. The second dimension was most heavily weighted by the objective measures of pad weight and mean incontinence episodes/day. The third dimension was weighted by the objective urodynamic measures of composed of VLPP and MUCP (latter in TOMUS only). Based on this analysis we selected independent measures of incontinence for further analysis to reduce the number of redundant hypothesis testing. Within each dimension we selected a single
measure to represent that aspect of incontinence, except in the subjective dimension as we wanted to explore both subjective symptom distress and symptom impact. Thus, we report the association of weight category with one objective measure of UI severity (IEF), two subjective measures of UI severity (UDI total score and IIQ), and one urodynamic parameter of UI severity (VLPP)21
. To further understand the associations of weight category with severity, we computed an analysis of covariance (ANCOVA) of each severity and impact measure on weight category controlling for clinically important variables and those significantly associated with weight in bivariate analysis.
Analyses were performed using SAS version 9.2 (SAS Institute, Inc. Cary, NC). Because of the large number of hypothesis tests, we defined statistical significance by p=0.01.