Predictors of hysterectomy in women with common pelvic problems: A uterine survival analysis

BACKGROUND: Comprising over 600,000 patients per year, hysterectomy is the most common nonobstetrical operation performed in US women. Little is known about the natural history of the noncancerous uterine conditions leading to hysterectomy. We followed a prospective cohort of women with common pelvic problems to determine whether simple clinical characteristics could predict a subsequent hysterectomy. STUDY DESIGN: We recruited 762 women seeking care for abnormal uterine bleeding, chronic pelvic pain, or symptomatic uterine fibroids and ascertained their hysterectomy status during 4 years of surveillance. We collected baseline patient-reported sociodemographic and clinical data and fit Cox models to predict the effects of covariates on hysterectomy across patient age. RESULTS: There were 99 hysterectomies, resulting in 0.044 hysterectomies per person-year of observation and a 13.5% cumulative hysterectomy rate. Hysterectomy was independently predicted by multiple pelvic symptoms or symptomatic fibroids (hazard ratio [HR], 1.97; 95% CI, 1.18-3.28), previous use of a gonadotropin-releasing hormone agonist (HR, 2.54; 95% CI, 1.53-4.24), and an absence of symptom resolution (HR, 2.24; 95% CI, 1.46-3.44). Survival curves plotted for subgroups with combinations of these predictors showed an escalating risk of hysterectomy with each additional risk factor. Predicted hysterectomy rates ranged from 20%, if all 3 predictors were absent, to 95%, if all 3 were present. CONCLUSIONS: For women with common pelvic problems, three easily measured clinical characteristics (symptom combination, degree of resolution, and earlier use of a gonadotropin-releasing hormone agonist) predict the likelihood of subsequent hysterectomy and can be used to inform counseling about the likely success of alternative treatments. (J Am Coll Surg 2007;204:633-641. (C) 2007 by the American College of Surgeons).