2006 HSR&D National Meeting Abstract
3090 — Race or Region? Examining Race Disparities in AMI Mortality in VHA
Lowy E. (Northwest Center for Outcomes Research in Older Adults)
Greiner G (Northwest Center for Outcomes Research in Older Adults)
Sales AE (Northwest Center for Outcomes Research in Older Adults)
Maynard C (Northwest Center for Outcomes Research in Older Adults)
Fihn SD (Northwest Center for Outcomes Research in Older Adults)
We describe a risk adjustment model for 30 day mortality in veterans hospitalized with acute myocardial infarction (AMI) in VA medical centers. We present a reporting format for comparing quality at a regional level. We present evidence of geographic/ethnic disparities that are revealed through the risk-adjustment/quality measurement process.
Data come from the FY2004 External Peer Review Program retrospective review of all AMI patients discharged from VA medical centers, supplemented with administrative data from Austin. We excluded veterans with AMI onset after hospital admission, and those transferred from non-VA facilities. 6029 veterans admitted for AMI during a twelve month period were included in the logistic regression to estimate risk-adjusted mortality rates at the VISN level, including VISN identifiers to examine geographic trends. VISN level ethnicity includes AMI patients only; in continuing analysis, we will include ethnicity calculated over all enrolled veterans in each VISN.
Significant predictors of 30 day AMI mortality include age, body mass index, history of cancer or congestive heart failure, beta blockers before admission, prolonged chest pain, systolic blood pressure, and heart rate on admission, first troponin measurement, and peak serum creatinine. The model has a c-statistic of 0.78.
Using this model, analysis at the VISN level shows a trend in which more ethnically diverse VISNs appear to have higher mortality rates. Logistic regression of 30 day mortality on percent non-white AMI patients within a VISN, including patient-level ethnicity, and controlling for the risk adjusters yields: an OR for an increase of 10% in the non-white population of 1.13 (p=.05), OR for being black of 0.69 (p<.05), OR for race unknown of 0.80 (P<.05).
The effective risk-adjustment model that we developed through collaboration with OQP revealed geographic/ethnic disparities which may be addressed in part by further quality measurement/improvement initiatives.
Veterans with AMI who are black or of unknown race have significantly lower 30 day mortality than their white neighbors, a finding in stark contrast to most of the literature on health disparities. Regional differences in care, however, work to opposite effect, so that within the VA as a whole, there is no correlation between ethnicity and 30 day mortality after AMI.