Identifying Patient Readmissions: Are Our Data Sources Misleading?

Area of Science:

• Health Services Research
• Clinical Informatics
• Data Science in Healthcare

Background:

• Accurate hospitalization outcome data is critical for clinical trials.
• Patient-reported outcomes are susceptible to attrition and recall bias.
• Increasing Medicare Advantage prevalence may reduce the reliability of Fee-for-Service claims for hospital utilization data.

Purpose of the Study:

• To compare the effectiveness of three data acquisition approaches for identifying hospital readmissions.
• To evaluate the utility of statewide health information exchanges (HIEs) as a research data source.

Main Methods:

• Secondary analysis of data from a randomized controlled trial of heart failure (HF) disease management in skilled nursing facilities (SNFs).
• Comparison of readmission identification using patient self-report, nursing home staff records, and state HIE data.
• Analysis of readmissions within 60 days post-SNF admission.

Main Results:

• A total of 295 unique readmissions within 60 days were identified among 657 participants.
• Patient self-report identified 20% of readmissions, nursing home staff records identified 28%, and HIE data identified 52%.
• The readmission rate increased from 18% (self-report/observation) to 34% with the addition of HIE data.

Conclusions:

• Over one-quarter (34%) of heart failure patients were rehospitalized within 60 days post-SNF admission.
• Statewide HIEs identified an additional 153 admissions (52% of total), highlighting their importance for comprehensive data.
• HIEs are essential for researchers to ensure accurate outcomes data, mitigating missed readmissions due to incomplete reporting or follow-up.