Identifying Patient Readmissions: Are Our Data Sources Misleading?

Abstract

BackgroundThe accuracy of data is vital to identifying hospitalization outcomes for clinical trials. Patient attrition and recall bias affects the validity of patient-reported outcomes, and the growing prevalence of Medicare Advantage (MA) could mean Fee-for-Service (FFS) claims are less reliable for ascertaining hospital utilization. Statewide health information exchanges (HIEs) may be a more complete data source but have not been frequently used for research. DesignSecondary analysis comparing identification of readmissions using 3 different acquisition approaches. SettingRandomized controlled trial of heart failure (HF) disease management in 37 skilled nursing facilities (SNFs). ParticipantsPatients with HF discharged from the hospital to SNF. MeasuresReadmissions up to 60 days post-SNF admission collected by patient self-report, recorded by nursing home (NH) staff during the SNF stay, or recorded in the state HIE. ResultsAmong 657 participants (mean age 79 ± 10 years, 49% with FFS), 295 unique readmissions within 60 days of SNF admission were identified. These readmissions occurred among 221 patients. Twenty percent of all readmissions were found using only patient self-report, 28% were only recorded by NH staff during the SNF stay, and 52% were identified only using the HIE. The readmission rate (first readmission only) based only on patient self-report and direct observation was 18% rather than 34% with the addition of the enhanced HIE method. Conclusions and implicationsMore than one-quarter (34%) of HF patients were rehospitalized within 60 days post SNF admission. Use of a statewide HIE resulted in identifying an additional 153 admissions, 52% of all the readmissions seen in this study. Without use of an HIE, nearly half of readmissions would have been missed as a result of incomplete patient self-report or loss to follow-up. Thus, HIEs serve as an important resource for researchers to ensure accurate outcomes data.