641: Does risk adjustment for patient characteristics affect assessment of hospital quality performance?

Abstract

affect assessment of hospital quality performance? William Grobman For the Eunice Kennedy Shriver National Institute of Health and Human Development, Maternal-Fetal Units Network, Bethesda, MD OBJECTIVE: The objective of this analysis was to determine whether risk adjustment, using detailed patient data, affects assessment of hospital performance regarding quality measures. STUDY DESIGN: Data were obtained by trained abstractors, with ongoing data edits and audits, from maternal and neonatal charts of all deliveries on 365 randomly selected days at 25 hospitals over a threeyear period. A priori outcome measures, rigorously predefined, were severe postpartum hemorrhage (PPH), maternal peripartum infection, perineal trauma (3rd or 4th degree laceration) at spontaneous vaginal delivery (SVD), and a composite adverse neonatal outcome. Observed rates of each outcome were calculated for each hospital, and hospitals were ranked from 1 (lowest frequency) to 25 (highest frequency). Patient-specific factors associated with each outcome were determined, and used to estimate each hospitals expected outcome rate. The observed/expected ratio (O/E) was calculated for each outcome (e.g. a ratio 1 indicates that the observed outcome rate was less frequent than expected on the basis of the hospitals patient characteristics), and each hospital was re-ranked according to this ratio. Descriptive statistics were performed, and Kendall’s test used to assess the concordance between the ranks based on the observed rate and O/E ratio for each outcome measure. RESULTS: Data were collected on 115,502 women. For all outcomes, the observed rank and O/E ratio (risk adjusted) rank were statistically similar, although there was change in rank of individual hospitals, as illustrated by the Figure for perineal trauma. The range of the difference in ranks as well as the median difference in ranks are presented in the Table. CONCLUSION: Among 25 participating hospitals, risk-adjustment using detailed patient data resulted in a similar overall ranking of hospitals for the chosen outcome measures. However, individual hospitals did change rank after risk adjustment. This movement may have relevance in the assessment of an individual hospital’s performance regarding specific quality measures.