QUEST2
Assistance to Quality Improvement Organizations (QIOs) in Home Health Care

Funded by: Centers for Medicare & Medicaid Services
Subcontract from: Delmarva Foundation for Medical Care
Project Period: 7/03 - 7/05

Abstract

Objective: To provide analyses, statistical assistance, and education/training support to the Delmarva Foundation for Medical Care in its role as the national lead Quality Improvement Organization (QIO) to train and assist home health agencies (HHAs) with outcome-based quality improvement (OBQI) using Outcome and Assessment Information Set (OASIS) data. For example, the Division of Health Care Policy and Research (HCPR) has developed the Resource Binder, which provides information for HHAs on the 11 publicly reported outcomes and strategies for outcome enhancement. An example analysis (described below) examines the effects of agency size on agency outcome measures and the ability of agencies to achieve statistically significant outcome improvements.

Methods: (1) Data from all Medicare-certified HHAs with at least 10 episodes in calendar year 2001 were used (N=6,178). Case mix and outcome differences by agency size categories were examined. Simulations were carried out to determine the extent of changes in outcome rates necessary to achieve statistically significant improvements for the different size categories. (2) Data from a five-state OBQI pilot project conducted from 2000 to 2002 were used to examine outcome enhancement performance as a function of sample size.

Results: (1) Case mix differences exist as a function of agency size, but the direction of the difference varies, sometimes indicating greater disability in the very small agencies (e.g., functional status) and sometimes indicating less disability (e.g., IADLs at start of care). The pattern of risk-adjusted outcome rates indicates consistently less favorable outcome rates for very small agencies compared to other agencies. Simulations showed that in some instances it would be impossible for agencies with small samples to show statistically significant change in outcome rates. (2) Target outcome enhancement activities for outcome measures with small sample sizes are less likely to show significant improvement than are outcome measures associated with larger sample sizes. A significant proportion of agencies with very small target outcome sample sizes are faced with "impossible to succeed situations" (in outcome enhancement) that do not exist for larger agencies.

Discussion: The findings for small agencies (and outcomes based on small sample sizes) indicate that demonstrating success under OBQI is difficult due to statistical reasons rather than the agencies' outcome enhancement performance. The restriction of outcome reports to only Medicare or Medicaid skilled care patients increases the number of agencies affected by small sample constraints. These differences highlight the varied circumstances and challenges QIOs face when facilitating OBQI efforts for a wide range of home health agencies.

Key Staff: Schlenker, Crisler, Powell, Hittle, Baillie, Goodrich, Richard

v