Development of Claims‐Based Measures of Unplanned Acute Care with Superior Power for Assessing the Effectiveness of Interventions Following Acute Care

Objective

To develop outcome measures that are more sensitive than current measures for evaluating primary or transitional care after hospitalizations, emergency department (ED) visits, or observation stays.

Data Sources

Medicare claims data from January 1, 2015, to October 31, 2017, for 1,261,707 Medicare fee‐for‐service beneficiaries served by (a) primary care practices participating in Track 1 of the Comprehensive Primary Care Plus (CPC+) initiative, and (b) their matched comparison practices.

Study Design

Given the poor statistical power in many studies to detect effects on readmissions, we developed two novel claims‐based measures of unplanned acute care (UAC) following an index acute care event. The first measure assesses the proportion of hospitalizations followed by an unplanned readmission, ED visit, or observation stay within 30 days of discharge; the second assesses the proportion of ED visits and observation stays followed by a hospitalization, ED visit, or observation stay within 30 days. We calculate minimum detectable effects (MDEs) for both measures and for a conventional measure of 30‐day unplanned readmissions, using CPC+ data.

Principal Findings

Repeat UAC events are common among Medicare beneficiaries served by the CPC+ practices. In 2017, 22% of discharges and 21% of ED visits and observation stays had a UAC event within 30 days. Readmissions were the most common UAC event following discharge, whereas ED visits were most common following index ED visits or observation stays. MDEs are 25%‐40% lower for the new measures than for the standard 30‐day readmissions measure, indicating better statistical power to detect impacts of primary or transitional care interventions.

Conclusions

This study introduces two new claims‐based measures to assess quality of care during a patient's vulnerable period following acute care. The new measures complement existing measures, covering a broader range of UAC events than the standard 30‐day readmissions measure, and yielding greater statistical power.