Objective: To investigate which patient and health-service factors are predictive of outcome after surgery for stress incontinence.
Patients and methods: In all, 232 women were assessed before and 3, 6 and 12 months after surgery for stress incontinence. Twelve patient characteristics and seven health-service factors were considered. Successful outcome was assessed in four ways: avoidance of complications after discharge, reduction in symptom severity. reduction in the social impact of stress incontinence, and improvement in activities of daily living (ADL). Relationships between possible predictors and outcome were explored using logistic regression models.
Results: Complications were less likely (r2=0.11) in women over 50 years of age, with no comorbidity and no urgency/urge incontinence, operated on by a gynaecologist. Reduction in symptom severity was more likely (r2=0.21) in younger women (< or = 50 years) with little limitation in their ADL, who suffered from severe or very severe stress incontinence symptoms but no urgency or urge incontinence, and who underwent preoperative urodynamic tests. A reduction in symptom impact was more likely (r2 = 0.15) in women in whom symptoms were having a moderate or severe impact, who were not grossly overweight, and who underwent a colposuspension or needle suspension performed by a gynaecologist. The only factor associated with improvement in ADL was the preoperative ADL score, i.e. the more limited a woman, the greater the likelihood of improvement (r2=0.15). The study provided evidence to support several of the widely held views about surgical success. Good outcomes were more likely if there is no urge incontinence, there is no or only mild comorbidity, no or only slight obesity, urodynamic investigations are conducted, and the surgeon is a gynaecologist. In contrast, there was no evidence that outcomes are worse if patients have undergone previous surgery for incontinence. Outcomes were also not related to grade of surgeon, surgeon's workload, or teaching status of the hospital.
Conclusion: The relative importance of different factors depends on the outcome of interest, i.e. impairment (complications), disability (symptom severity) or handicap (symptom impact and ADL). These findings might help to improve clinical practice by indicating those characteristics that are associated with a greater likelihood of a good outcome. The results could also aid evaluation and audit of surgery, having identified the main prognostic factors that need to be considered when comparing the outcome of different procedures, surgeons or hospitals. Before doing so, the predictive models described in this paper need to be tested using an independent dataset.