Impact of risk adjustment on cross-sectional variations and short-term stability of the nursing home outcome measure of urinary/fecal incontinence

The "Nursing Home Compare" report card publishes and regularly updates a set of outcome-based quality measures derived from resident health assessments. This study focused on one published measure (urinary/fecal incontinence rate of a nursing home), and tested whether multivariable risk adjustment for resident characteristics can improve (1) the validity of the measure for cross-sectional comparisons of nursing home quality, and (2) the short-term stability of the measure (ie, stability over 3 months). Our sample from the 2005 MDS included over 600,000 long-term care residents in 10,437 facilities. We estimated logistic regression models to predict the outcome rate for each resident based on his/her demographics and physical and mental health functions. We also tested the impact of "case mix" variation (ie, facility variation in patient characteristics) on both currently published and multivariable risk-adjusted measures. We found that at least 50% of the cross-sectional variation of the published measure was explained by that of facility case mix (R-square>=0.50), and over 25% of the short-term variation of the published measure was due to that of case mix (R-square>0.25). In contrast, the cross-sectional and short-term longitudinal variations of the risk-adjusted measure were much less susceptible to case mix variations (R-square<0.10). We conclude that multivariable risk adjustment through statistical regression can minimize the overall impact of case mix, and make the measure more likely to reflect a facility's cross-sectional standing and short-term stability in continence care practice (or quality). The risk-adjusted rate is more appropriate than unadjusted rate for use as a report card measure.