Association of Diagnostic Coding with Trends in Acute Myocardial Infarction Incidence and Mortality Among Medicare Beneficiaries, 2002-2011

Context: Recent literature suggests sharp declines in both the incidence of and mortality following acute myocardial infarction (AMI) among Medicare beneficiaries. While these declines may be due to improving care, they may also – at least partly – be artifacts of diagnostic coding changes.  To evaluate the contribution of changed coding to these declines, we compare trends in AMI hospitalizations and subsequent mortality among elderly Medicare patients using two different case definitions:  one limited to patients with a principal discharge diagnosis of AMI, and another which includes patients with an AMI coded in the second hospital discharge diagnosis position.

Methods:   Data:  We use CMS’ 5% sample MedPAR (Medicare Provider Analysis and Review) data linked to its Medicare Denominator file.   Study Population:  Medicare beneficiaries aged 66 and older, continuously enrolled in Medicare Fee-for-Service (FFS) for the 12 months prior to an AMI hospitalization (discharge diagnosis of 410.xx) from 1999 through 2008 using two different case definitions: 1) the traditional identification method, based on principal diagnosis only and 2) using either of the first two listed diagnoses. Outcomes: AMI incidence and subsequent mortality within 30 days and one year of AMI discharge.  Analyses:  We describe trends in the age-sex-adjusted AMI hospitalization rates and age-sex-risk adjusted 30-day and 1-year mortality rates between 1999 and 2008, using each case definition. 

Results:   Using traditional case-finding for AMI, between 1999 and 2008 the age-sex adjusted incidence of hospitalization for AMI declined by fully 27.7%, from 1,086 to 785 per 100,000 beneficiaries; the age, sex and risk adjusted 30-day mortality rate declined by 20.4%, from 19.1% to 15.2%; and 365-day mortality rate declined 15.7%, from 36.4% to 30.7%. However, identifying AMIs from either the first- or second-listed diagnosis, AMI incidence declined only 18.6%, from 1,239 to 1,009 per 100,000; 30-day mortality rates declined 14.1%, from 20.6 to 17.7% and 365-day mortality declined 11.7%, from 38.5 to 34.0% between 1999 and 2008. (All p <0.001.)

Discussion:  From 1999 to 2008, hospitalization and mortality rates of Medicare beneficiaries with a principal discharge diagnosis of AMI (the definition commonly used in the literature) declined dramatically. When identifying an  AMI from a diagnosis in either the first or second position, the declines are less impressive.  Changes in coding practices appear to have affected reported declines in AMI incidence and mortality. The possibility that evolving Medicare performance assessment and reimbursement policies have influenced these coding changes merits investigation.