Population-Based Measures from Administrative Data to Guide Efforts to Examine and Improve the Quality of Hospice Care (S711)

Abstract

•Identify seven population based measures that can be calculated for Medicare claims data.•Understand how certain characteristics of hospices correspond to being an outlier on multiple measures listed in Objective 1. Medicare requires hospices to submit claims containing information that could be used to create population based measures of processes and outcomes to guide efforts to monitor hospices and improve care delivery. Identify hospices that are outliers in the provision of hospice services along seven dimensions: (1) provision of General Inpatient and Continuous Home Care days, (2) visits by nurse or social worker at the end of life, (3) long length of stays in hospice, (4) rates and patterns of hospice live discharge, (4) selective enrollment of nursing home and ALF patients, (6) infrequent hospice visits by nursing staff over the entire course of the episode, and (7) average time spent by nurse during a visit. Population based measures are calculated using 100% of Medicare hospice claims from FFY 2014. For each measure, hospices are grouped into deciles. We count how times a hospice is in the highest decile for each measure and use that count to determine outliers. 460 out of 3,453 hospices scored in the highest decile for 3 or more measures. Those hospices were in southern census region (56% vs. 38%, p <.001) and were for-profit (75% vs 55% for not-for-profit, p<.001). A multivariate ordinal logit model showed newer, smaller, for-profit hospice programs were more likely to be above the thresholds for 3 or more measures compared to larger, older, not-for-profit programs (AOR 2.52 - 95% CI: 1.86-3.42). Potential markers of poor performance can be calculated using hospice claims. Ownership and location are correlated with poor performance across multiple measures.