Objective: To examine the distribution of risk and the correlation between risks in a home care population with regard to several important adverse outcomes.
Background: Researchers and policy makers have long recognized the heterogeneity of home care populations. Most research in this area focuses on identifying predictors of adverse outcomes. The degree of the heterogeneity of risks is much more poorly understood. Yet understanding the degree of risk heterogeneity at the population level is important because it has implications for the extent to which the level of care should vary among recipients.
Study setting: Patients enrolled in the Arizona Health Care Cost Containment System (AHCCCS) program, between the December 1992 and April 1998.
Outcome measures: Estimating the risk for nursing home placement, hospitalization, death, and functional decline.
Methods: Estimating discrete time hazard models. From these models the predicted risk for each outcome is estimated and the distribution and correlation of predicted risks is examined. Model fit is assessed through split sample techniques and by examining the ratio of predicted to actual outcomes for selected sub-groups.
Results: The estimates reveal a wide variation in predicted risk. The ratio of predicted risk at the 90th percentile relative to the 10th percentile ranges from 4.99 for nursing home admission to 6.65 for hospitalization. The distributions of predicted risks are all skewed, particularly the distributions for death and nursing home admission. Predicted nursing home risk is highly correlated with the predicted risk for death (rho = 0.71). The predicted risk for hospitalization is not strongly correlated with the predicted risk for either death or nursing home admission.
Conclusion: The wide variation in risk among home care patients suggests that efficient allocation of resources would require variation in spending and targeting of services based on patient characteristics. Greater research regarding the effectiveness of home care for different sub-populations is called for.