Hospitals Likely Miscount True STEMIs, Analysis Suggests
With varying sensitivity and precision for identifying STEMI, a healthy skepticism of STEMI care data is warranted, say docs.
Just two-thirds of ST-segment elevation MIs from a single medical center included in different national databases were correctly adjudicated as a STEMI, according to the results of a new study.
The reliability of approaches used to identify STEMIs across the various national databases, including the National Cardiovascular Data Registry (NCDR) CathPCI Registry and the National Quality Measures Clearinghouse (NQMC), ranged substantially, with the NCDR being the most sensitive and precise for capturing clinical events.
The research, according to investigators, has important implications given that studies evaluating the clinical characteristics and outcomes of STEMI patients are often informed by these registries. If the databases aren’t accurately identifying STEMI cases, it calls into question the validity of some of the published research.
“It’s a really good test,” senior investigator Tim Henry, MD (The Christ Hospital, Cincinnati, OH), told TCTMD. “Every institution does this—they write down their STEMIs in multiple different ways. You report them in the NCDR CathPCI Registry, and of course there’s Medicare data, or the Nationwide Inpatient Survey, and all that’s published and taken as gospel, that this is what STEMI care looks like in the United States. Every year in major journals, we have articles published from those places, but it doesn’t reflect comprehensive STEMI care.”
Just 10% of STEMIs Captured by All Databases
The new analysis, which was published as a research letter February 19, 2021, in Circulation: Cardiovascular Quality and Outcomes, was intended to test the sensitivity, precision, and agreement between the different approaches to capturing STEMI cases. To do so, the researchers identified all patients from Cedars-Sinai Medical Center in Los Angeles, CA, diagnosed with STEMI in 2016 by one of four approaches. These included the International Classification of Diseases, Tenth Revision (ICD-10) coding and definitions included as part of the NQMC, the NCDR CathPCI Registry, or the internal STEMI registry at Cedars-Sinai Medical Center. All STEMIs captured by one of the four approaches were then blindly adjudicated by investigators using the fourth Universal Definition of MI.
Overall, 302 unique cases were identified in at least one of the four databases during the study period. Of these, 205 were adjudicated as confirmed STEMIs. The sensitivity of each approach for identifying STEMI ranged from 35.1% with the NQMC to 71.2% for the NCDR CathPCI Registry. The sensitivity of the internal Cedars-Sinai registry and ICD-10 coding for STEMI was 68.3% and 52.7%, respectively. The precision, or positive predictive value, of confirmed STEMIs in the databases ranged from 60.3% with ICD-10 coding to 90.7% with the NCDR. Agreement between the different approaches also varied, with the best, if moderate, agreement seen between the internal STEMI registry and NCDR.
“CathPCI was the most accurate, but you only get included in the registry if you have a PCI,” said Henry. “Well, every STEMI doesn’t go for PCI. Some go for CABG, some might go for medical therapy, while some might be treated with lytics and then have PCI. It doesn’t reflect all of the STEMIs. It excludes a clear percentage of people.”
Only 22 of the 205 definite STEMIs adjudicated were captured by all four approaches (ICD-10, NQMC, NCDR, and internal STEMI registry).
Henry noted there is no single national registry that includes all STEMI patients, including the high-risk subsets of patients with cardiac arrest and/or cardiogenic shock. For that reason, the Midwest STEMI Consortium (MSC), which includes four large, regional STEMI care hospitals (Iowa Heart Center, Minneapolis Heart Institute Foundation, Prairie Heart Institute, and The Christ Hospital), was established to develop a comprehensive, multicenter, prospective registry. To date, there are roughly 20,000 patients included in the registry.
“We take every single STEMI activation, even if they’re 90 years old, even if they die before getting to cath lab, if they have cardiac arrest or cardiogenic shock, and even if it turns out they don’t need PCI or it turns out to be a false-positive,” said Henry. “It reflects complete STEMI activation.”
As for the present study, Henry said it should serve as an eye-opener. Based on these findings, he advises physicians to have a healthy skepticism when interpreting STEMI data from observational studies, such as the NCDR or Medicare registries, because the definition of STEMI can vary from database to the next, as can the reliability of accurately capturing clinical outcomes.
Brown CL, Ebinger J, Bradley SM, et al. Reliability and validity of current approaches to identification of patients with ST-segment-elevation myocardial infarction. Circ Cardiovasc Qual Outcomes. 2021;14:e007228.
- Henry reports no relevant conflicts of interest.