• While a great deal of time and study has been devoted developing a clinically valid and reliable public home health performance reporting system, the system is still relatively new.  The CMS rollout of the Home Health Compare website was in October 2003.  Other national public reporting efforts – within and beyond health care – have experienced a period of instability wherein opportunities for improving data quality were recognized and acted upon.
• Home health performance measures reflect resident outcomes and therefore require rigorous risk adjustment.  While the quality measures are risk-adjusted to account for the fact that some agencies treat sicker patients, underlying differences in patient conditions or in the types of care provided by the agencies may not be fully accounted for by risk adjustment strategies.  The best achievable rates for these measures is not well understood at this time.
• The data presented on Home Health Compare comes from information collected by Medicare and Medicaid-certified home health agencies. This information is called the Outcome and Assessment Information Set (OASIS). OASIS is a group of elements that represent core items of a comprehensive assessment for an adult home care patient.  All of these data are reported by the home health agencies themselves and not formally audited to ensure accuracy.
• The OASIS data displayed on the website are updated monthly and represent a rolling 12 months of data but have a two-three month data lag time. Recent improvements made by the home health agency will not be evident in the rankings.
• Because of differences in the number of patients served by agencies the clinical conditions of the patients they care for, the number of patients used to compute reported performance rates varies by agency and measure.  This impacts the precision with which performance can be ranked.

Analytic considerations:

• Performance on the measure “Percentage of patients who had to be admitted to the hospital” contributed half to each agency’s national ranking.  The remaining measures were given equal weight in the analysis regardless of the eligible number of patients for each agency or of any assessment of the relative importance of the measure.
• While percentile scores, by definition, always range from 0-100, reported performance on a given measure may cluster very tightly.
• A single composite performance score or ranking can mask either strong or weak performance in specific measures.