Declines in AMI hospitalizations: Findings overall and by race change with different claims-based definitions

research objective. Recent literature reports sharp declines in acute myocardial infarction (AMI) incidence among Medicare beneficiaries, with more modest declines among African Americans.   These “traditional” estimates may be misleading because they rely on diagnostic codes found in administrative data and count only hospitalizations with a principal (first-listed) diagnosis of AMI, ignoring those with AMI diagnoses listed in other positions. We examined trends in AMI occurrence when AMI was identified using only the first-listed versus all diagnoses, and whether these different definitions affected estimates of hospitalizations similarly across population subgroups.

study design. Retrospective cohort study of trends in AMI hospitalizations between 2002 and 2011 using: 1) principal diagnosis only versus 2) any coded discharge diagnosis. Data Source: MedPAR (Medicare Provider Analysis and Review) data linked to Medicare Denominator files. Outcomes: AMI (ICD-9-CM code=410.xx, excluding 410.x2) hospitalization rates per 100,000 beneficiaries (herein, “occurrence”).

population studied: Fee-for-Service (FFS) Medicare beneficiaries aged 65 and older with an AMI hospitalization identified using alternate case definitions between 2002-2011.

principal results:  Using “traditional” definition 1, AMI occurrence declined by 37% between 2002 and 2011 (from 1,184 to 758); other AMI hospitalizations increased 2% (from 432 to 441). Among African Americans (8% of beneficiaries), definition-1 hospitalizations dropped only 21% (from 1,026 to 808) and other AMI hospitalizations increased 40% (from 444 to 623). Consequently, when counting all AMI hospitalizations (definition 2), AMI occurrence declined by 26% among all beneficiaries (from 1,616 to 1,199), by 28% among non-African Americans (from 1,629 to 1,179), and by 2.6% among African Americans (from 1,470 to 1,432). (All p<0.001.)

conclusions. During these 10 years, hospitalization rates for Medicare beneficiaries with a principal discharge diagnosis of AMI declined substantially, although the decline was less for African Americans. When counting all hospitalizations, overall declines in AMI occurrence were more modest, but still substantial (26%).  In contrast, there was almost no decline for African Americans.

implications for policy or practice. Increases in hospitalizations with AMI coded in non-principal positions may reflect temporal trends in coding practices in response to Medicare payment policies, quality-reporting initiatives, or other factors.  While some later-listed AMI diagnoses were likely consequences, not causes, of hospital admission, different definitions of what constitutes an AMI hospitalization can lead to very different perceptions of trends in cardiovascular health and how these trends vary by race.