How pooling fragmented healthcare encounter data affects hospital profiling

Area of Science:

• Health Services Research
• Health Informatics
• Cardiovascular Medicine

Background:

• Fragmented health records exist for patients with multiple healthcare payers, such as Medicare and the Veterans Health Administration (VA).
• The impact of utilizing more complete patient data on hospital profiling remains underexplored.

Purpose of the Study:

• To investigate how incorporating combined VA and Medicare data influences hospital profiling compared to using VA data alone.
• To assess the effect on risk-standardized mortality rates (RSMRs) and hospital outlier status.

Main Methods:

• Retrospective cohort study of 104 VA hospitals.
• Examined 30-day mortality post-acute myocardial infarction for veterans aged 66+ (2006-2010).
• Calculated RSMRs using VA-only versus combined VA/Medicare data, employing observed/expected (O/E) and predicted/expected (P/E) models.

Main Results:

• Thirty-day mortality was 14.8%.
• Adding Medicare data led to significant increases and decreases in RSMRs for approximately half of the hospitals studied.
• Hospital outlier designations changed for 4 hospitals using the O/E measure, but not with the P/E measure.

Conclusions:

• VA hospitals exhibit variability in patient Medicare utilization and data completeness within VA records.
• Combined VA/Medicare data provide a modestly different perspective on hospital profiling compared to VA data alone.