Introduction:
Public reporting of cardiovascular outcomes from registries has been useful in monitoring and improving quality of care. However, participation in registries such as National Cardiovascular Data Registry (NCDR) is voluntary and information could be biased due to self-selection of the participating hospitals thus limiting the external validity. Administrative databases usually do not include detailed information, but include all patients.
Hypothesis:
To compare in-hospital mortality following admission for acute myocardial infarction (AMI) between NCDR and a statewide database of all admissions to nonfederal hospitals in New Jersey (Myocardial Infarction Data Acquisition System, MIDAS) for the year 2011.
Methods:
We compared the proportion of in-hospital deaths for patients admitted for AMI between NCDR (published data) and MIDAS (unadjusted and adjusted for age, gender and cardiac catheterization (CATH) or percutaneous coronary intervention (PCI). MIDAS data were derived by direct standardization using the 2011 National Inpatient Sample (NIS) as the standard population.
Results:
The in-hospital mortality rates were lower in NCDR than MIDAS (4.56% vs. 6.37% crude, 5.92 adjusted). NCDR patients were younger and received more CATH (84.97% vs. 62.07%) PCI (64.17% vs. 41.67%). Adjustment for demographics and PCI explained 33% (1.36 vs. 1.81) of the mortality difference between NCDR and MIDAS. Similar findings were obtained when mortality was adjusted for demographics and CATH (40%, 1.81 vs. 1.29, of the mortality difference explained).
Conclusions:
Prospective registries and administrative databases provide complementary information. The significant differences in patient profiles, frequency of procedures and in-hospital mortality rates between NCDR and MIDAS imply that the care in NCDR hospitals is better. The magnitude of the differences warrants caution in generalizing the NCDR data to the entire nation.
Effect of gender according to age on in-hospital mortality in patients with acute myocardial infarction in the ACC-National Cardiovascular Data Registry
