Trends in and disparities for acute myocardial infarction: an analysis of Medicare claims data from 1992 to 2010

More than two decades of research has demonstrated significant disparities in the management of patients with acute myocardial infarction (AMI) based upon patient sex and race in both the United States (US) and Europe [1]-[5]. Reports from the US Institute of Medicine (IOM) and the European Union have called for providers and policymakers to confront and eliminate socioeconomic and racial and ethnic disparities in health care [6],[7]. The World Health Organization has described the social determinants of health that lead to health inequities [8]. The US Affordable Care Act (ACA) established a dedicated office to reduce disparities [9].

While reduction of race- and sex-based disparities has been an international priority for decades, relatively few longitudinal studies have assessed the success of disparity reduction efforts. Moreover, most of the available studies were relatively short in duration (that is, 5 to 10 years) and few focused comprehensively on AMI [1],[3]. In 2005, Jha et al. reported no decline in racial disparities for nine surgical procedures during the 1990s [10], but did not study disparities in AMI-associated hospitalization or mortality and this study has not been updated with contemporary data. The case fatality rate was consistently higher among Maori and Pacific Islands people than in Europeans in New Zealand for each age group and both sexes [5].

The overarching objective of our analysis was to comprehensively examine long-term trends in AMI disparities by patient race (black and white) and sex (male and female). In particular, we were interested in examining longitudinal differences in AMI-associated: 1) hospitalization rates; 2) utilization of revascularization procedures; and 3) mortality. In aggregate we felt that these three measures could comprehensively evaluate the full spectrum of AMI care from the underlying burden of cardiovascular disease (hospitalization rate), to processes of care (revascularization) and outcomes (mortality).

In a longitudinal analysis of US Medicare beneficiaries hospitalized with AMI between 1992 and 2010, we report a number of important trends in disparities. We found that AMI hospitalization rates declined substantially for all patient groups. We also found that differences in hospitalization rates in blacks and whites and men and women have narrowed over time, but disparities persist. Utilization of interventional procedures after AMI has increased over time, but disparities persist with both women and blacks being significantly less likely to receive most procedures. Lastly, 30-day mortality following an AMI has declined over time for all patient groups, but remains significantly higher for women compared to men. Several findings merit further discussion.

First, we found declines in overall hospitalization rates for AMI among all four of our study cohorts (black women, black men, white women, and white men). While there are several studies reporting changes over time in AMI mortality or quality of AMI care [17]-[19], there are fewer studies of long-term trends in AMI hospitalization rates, with somewhat contradictory results. Chen and Krumholz examined trends in AMI hospitalization among Medicare beneficiaries between 2002 and 2007 and noted an approximately 20% decline in hospitalization rates [20]; they also found evidence of higher AMI hospitalization rates among men compared to women and white men as compared to black men. Similarly, Talbott et al. reported a 20% decline in AMI hospitalization rates using the Centers for Disease Control (CDC) data [21]. Alternatively, data show declines in cardiovascular disease mortality over time without corresponding reductions in AMI hospitalization rates [22]. Our findings are particularly intriguing when viewed through the lens of increasingly sensitive diagnostic tests for AMI (that is, serum troponin) that would be expected to increase AMI diagnosis and hospitalization rates [23]. In a nationwide Danish study, the standardized incidence rate per 100,000 people decreased in the 25-year period, 1984 to 2008, by 37% for women (from 209 to 131) and by 48% for men (from 410 to 213) [24].

Our finding of differential hospitalization rates by race and sex is noteworthy and merits brief comment. It is possible that the lower hospitalization rates truly represent lower population-based AMI risk among blacks (compared to whites) and women (compared to men). However, an array of epidemiological data suggests a higher burden of cardiovascular disease among blacks making this scenario unlikely [25],[26]. An alternative and more ominous explanation is that the lower hospitalization rates reflect under-diagnosis of AMI in blacks and women in the pre-hospital setting. Differences in AMI hospitalization rates by race and sex remained statistically significant in 2009 to 2010 (P <0.001) despite narrowing of the differences between cohorts over time in analyses adjusted for age and year (Figure 1A). This implies that only some racial and sex differences were attributable to age and period effects, and some disparities still persisted, although they narrowed in adjusted analyses. Another study using the CDC data reported that the male-female ratio for AMI hospitalization rates remained constant at 2:1 from 2000 to 2008 [21].

Second, we found a marked increase in the utilization of revascularization among all patient cohorts over time. These results are consistent with prior studies [22]. Interestingly, while disparities in revascularization have been well documented [27], relatively few studies have examined whether disparities in revascularization have declined over time [28]. Peterson et al., using data from the National Registry of Myocardial Infarction (NRMI) found evidence that disparities by race and sex in the use of revascularization procedures did not decline between 1990 and 2006 [18]. Alternatively, Jha and colleagues found evidence that disparities for certain revascularization procedures declined among certain patient populations between 1992 and 2001 (for example, PCI in black and white women) but other disparities persisted (for example, CABG in black and white men) [10]. Our results extend prior studies by examining a more comprehensive list of revascularization procedures over a longer time period. Our unadjusted analyses showed minimal reductions in disparities for most procedures and that disparities between men and women remain particularly large. Interestingly, in analyses that adjusted for patient comorbidity and clustering of patients within hospitals [see Additional file 1], racial disparities appeared small although disparities by patient sex remained significant. The differences between our unadjusted and adjusted results are consistent with prior studies that have suggested that many disparities can be explained by differences in patients’ acuity and comorbidity and also the quality and practice patterns of the physicians and hospitals where different patient groups seek care [2],[18],[29],[30].

Third, our findings with regards to mortality are important and extend prior work. We found that racial disparities in AMI 30-day mortality grew relatively larger in the early part of the 21^st century, but have declined over the last five years. Alternatively, disparities by patient sex (higher mortality for both black and white women when compared to their male counterparts) have declined little [31]; higher mortality has been reported in younger (<50 years), but not older women compared to men of the same age after AMI [32]. Studies have suggested that sex-based disparities in mortality can be attributed in part to differences in the receipt of evidence-based treatments and interventions, but also perhaps to differences in AMI presentation, severity and age-of-onset [33]. Irrespective of the exact cause, the sex-based differences in AMI mortality are striking, similar to a previous population-based Finnish study [34]. In this study including all age groups comparing 1994 to 1996 to 2000 to 2002 in two registers, the 28-day case fatality after AMI declined in men from 46.5% to 41.0% in the FINAMI register and from 54.7% to 50.1% in the Finnish National Cardiovascular Disease Register (CVDR), respectively. The corresponding numbers in women decreased from 53.5% to 48.6% in FINAMI and from 58.9% to 54.6% in CVDR [34]. Another recent study found that mortality from ischemic heart disease increased across all ages in adult women whereas the corresponding increase was blunted at older ages in men [35], a potential explanation of sex differences in AMI mortality in our data as well. A Dutch study of 14,434 patients in coronary units from 1985 to 2008 showed no sex difference in adjusted 30-day mortality [36], as well as a decline in mortality in both men and women over a 24-year period. The differences in country setting, health care delivery system, patient comorbidity load and single center versus national data between this previous study and our study may explain the differences in findings. Our mortality models were based on models developed by the US Centers for Medicare and Medicaid Services and adjusted for hospital clustering and, therefore, our confidence in these findings is high.

Our study has several limitations that warrant mention. First, our analysis was limited to fee-for-service Medicare beneficiaries and should be extended to other populations with caution. In particular, it is uncertain whether an evaluation of a younger non-Medicare population would yield similar results. Second, our analysis focused exclusively on AMI and we cannot comment on long-term trends in disparities for other conditions. Third, our analyses could have been impacted by changes in coding over time related to either AMI diagnosis or procedures. However, there is a long history of using Medicare data in the evaluation of AMI and we are unaware of any coding changes that would have directly impacted our results. Fourth, Medicare data may have race misclassification since this was not self-identified prospectively; however, this would bias our findings towards null and, therefore, our findings are conservative. In addition, this is a general limitation of studies using Medicare data. The impact of the exclusion of 2006 is unclear; however we have a long-enough observation period, and a single year is unlikely to have meaningfully changed the findings. Fifth, the time-trends analyses for where the procedures were done are unadjusted (index hospitalization, transfer hospitalization, or post-discharge within 30 days); it is anticipated that a reduction in the length of AMI stay over the study period likely impacted these rates, and, therefore, these should be interpreted with caution. Lastly, we are unable to account for patient preferences, which play a role in utilization of these procedures or physician preferences, or the clinical information about the appropriateness of the interventional procedures. This is an important aspect of disparities that needs to be understood; however, this is likely on the causal pathway and, therefore, not a confounder of the estimates presented in this study.

In summary, over the past 18 years we observed declines in hospitalization rates and mortality and increases in revascularization for AMI in the US Medicare population. We found that disparities by sex and also by race in hospitalization rates, receipt of revascularization procedures and mortality have persisted over time. Our study provides evidence of the continued challenge of insuring equal care to all patient groups.

Study concept and design: JAS, XL and PC. Acquisition of data: XL and SI. Analysis and interpretation of data: JAS, XL, SI and PC. Drafting of the manuscript: SI and PC. Critical revision of the manuscript for important intellectual content: JAS, XL, SI and PC. Statistical analysis: XL. Obtained funding: PC. Administrative, technical, or material support: SI and PC. Study supervision: SI and PC. PC had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the analysis. All statistical analyses were performed at the University of Iowa by XL with supervision by PC. All authors read and approved the final manuscript.