Risk Adjustment Methods for Hysterectomy Complications. Abstract, Executive Summary and Final Report

摘要

Risk adjustment methods on administrative data may not accurately predict adverse outcomes for conditions for conditions or procedures that are performed in younger, healthier populations. The authors used hospital discharge data from North Carolina from 1988-1994 to assess the ability of logistic regression models to predict surgical and medical complications after hysterectomy, a commonly performed procedure in younger women. Models for medical complications had better discrimination (c=0.714) then models for surgical (c=.630) In order to determine if uncoded factors explained some of the surgical complication risk, the authors then examined records of all women undergoing hysterectomy for uterine fibroids, the most frequent single indication for hysterectomy and one with substantial racial differences in incidence and outcome, at a single institution from 1992-1998. Although black women had a significantly higher crude risk for complications. The majority of this risk was attributable to differences in baseline characteristics specifically obesity, anemia, increasing uterine size, and adhesions. Oophorectomy and other concurrent procedures were also significant independent predictors of complications. Given the high incidence of fibroids, a risk adjustment model based on these factors could have substantial public health impact by allowing comparison of outcomes between providers if validated in prospective studies in different settings.