Highlight
- Clinical audit across eight U.S. hospitals found only 39% of type 1 MI (T1MI) and 72% of type 2 MI (T2MI) administrative codes accurately reflected true diagnoses.
- Nearly half of patients coded as T1MI had T2MI, while over a quarter of T2MI codes actually represented myocardial injury—not MI.
- Hospitals with cardiac catheterization labs had significantly lower misclassification rates.
- Findings have major implications for epidemiological surveillance, health policy, quality metrics, and clinical care pathways.
Background
Myocardial infarction (MI) remains a leading cause of morbidity and mortality worldwide. Contemporary definitions, notably the Fourth Universal Definition of MI, distinguish type 1 MI (T1MI: acute atherothrombotic event due to plaque rupture/erosion) from type 2 MI (T2MI: ischemia due to supply-demand imbalance without acute plaque disruption) and from myocardial injury (elevated cardiac troponin without overt ischemia). Accurate classification is crucial for clinical management, epidemiological tracking, payment models, and public health reporting. However, the reliability of administrative coding (particularly ICD-10) in reflecting these clinical nuances has not been rigorously validated. Given the reliance on coded data for research, hospital benchmarking, and policy decisions, any coding inaccuracies may have far-reaching consequences.
Study Overview and Methodological Design
Martinez et al. performed a retrospective, cross-sectional audit of administrative codes assigned for T1MI and T2MI across eight hospitals in the Mass General Brigham system from October 2017 to May 2024. A randomly selected cohort of 700 inpatient encounters (350 coded as T1MI, 350 as T2MI) included patients aged 65 years and older. Clinical adjudication of MI subtype was performed using chart review and the Fourth Universal Definition of MI, focusing on evidence of acute coronary thrombosis/erosion versus supply-demand mismatch or isolated myocardial injury. A second, blinded physician review was conducted on a subset (146 challenging and 146 nonchallenging cases) to assess the reproducibility of adjudications. The study also compared misclassification rates between hospitals with and without cardiac catheterization laboratories.
Key Findings
The audit revealed substantial misclassification:
- Among T1MI-coded patients, only 39% (n=138) had true T1MI upon clinical review. Nearly 45% (n=159) had T2MI, and 10% (n=35) had isolated myocardial injury.
- Among T2MI-coded patients, 72% (n=251) had true T2MI, but 26% (n=91) had myocardial injury, and 1% (n=4) had T1MI.
- Physician adjudication reproducibility was high: 94% agreement in nonchallenging and 86% in challenging cases.
- Hospitals with cardiac catheterization labs had lower misclassification rates for T1MI coding (43% vs 58%; P = .0298).
These findings confirm and extend earlier reports of administrative code inaccuracy for MI phenotypes (DeFilippis AP et al., JAMA 2019; Sandoval Y et al., Circulation 2021).
Mechanistic Insights and Pathophysiological Context
The distinction between T1MI and T2MI is not merely semantic but reflects fundamentally different pathophysiologies and therapeutic implications. T1MI results from acute, typically thrombotic, coronary artery occlusion—a setting where antithrombotic therapy and urgent revascularization are indicated. T2MI, by contrast, is secondary to a mismatch between myocardial oxygen supply and demand (e.g., sepsis, anemia, tachyarrhythmia) without acute plaque rupture. Myocardial injury may arise from non-ischemic causes (e.g., myocarditis, renal failure), and its management does not primarily involve targeting coronary disease. The observed coding inaccuracies likely stem from overlapping clinical presentations (e.g., troponin elevation with chest pain), limitations in documentation, and the pressure of coding for reimbursement, especially in hospitals lacking interventional cardiology resources or advanced diagnostics.
Clinical Implications
The misclassification of MI subtypes in administrative data has immediate and far-reaching consequences:
- Under- or overestimation of true MI incidence in population health statistics and registries.
- Potential misdirection of quality improvement initiatives, risk-adjusted benchmarking, and public reporting.
- Risk of inappropriate or delayed therapy for individual patients (e.g., unnecessary anticoagulation or omission of guideline-based secondary prevention).
- Distorted resource allocation, payment models, and research funding streams predicated on flawed case ascertainment.
For example, an older adult admitted with sepsis and elevated troponin may be miscoded as T1MI, prompting unwarranted invasive intervention or antithrombotic therapy with associated bleeding risk. Conversely, a true T1MI miscoded as T2MI or injury may miss timely reperfusion.
Limitations and Controversies
This study is limited by its focus on older adults (≥65 years) within a single, large academic health system; generalizability to other populations and care settings should be confirmed. Despite robust adjudication, some residual subjectivity remains, particularly in ambiguous cases where the clinical presentation and biomarker profile overlap. The study did not assess the impact of coder training, electronic health record prompts, or the potential for improvement with updated ICD coding definitions. Finally, it is unclear to what extent misclassification is driven by documentation quality versus coder interpretation.
Expert Commentary or Guideline Positioning
Current guidelines (Thygesen K et al., Circulation 2018; ACC/AHA Chest Pain Guidelines 2021) emphasize the importance of distinguishing MI types for both management and reporting. Dr. James Januzzi, a co-author, notes: “The accuracy of administrative data has been assumed in countless studies and policy frameworks. These findings compel us to revisit how we collect, code, and interpret MI diagnoses—especially as they guide resource allocation, reimbursement, and quality measurement.”
Conclusion
Substantial misclassification exists in the administrative coding of myocardial infarction subtypes, particularly for T1MI, with nearly half of codes failing to reflect the true diagnosis. This has significant ramifications for clinical care, health policy, and research. Interventions to improve diagnostic coding—such as enhanced education, clinical decision support, and integration of structured diagnostic criteria—are urgently needed. Further studies should explore scalable solutions and validate these findings across diverse healthcare systems.
References
- Martinez A, Etiwy M, Januzzi JL Jr, et al. Misclassification of Myocardial Infarction Subtypes in Administrative Claims: Implications for Clinical Epidemiology and Quality Assessment. J Am Coll Cardiol. 2025;86(4):284-286. doi:10.1016/j.jacc.2025.05.036 IF: 22.3 Q1 .
- Thygesen K, Alpert JS, Jaffe AS, et al. Fourth Universal Definition of Myocardial Infarction. Circulation. 2018;138(20):e618-e651.
- DeFilippis AP, Chapman AR, Mills NL. Assessment and Classification of Myocardial Injury and Infarction—A Systematic Review and Meta-Analysis. JAMA. 2019;321(11):1098-1099.
- Sandoval Y, Thygesen K. Myocardial Infarction Type 2 and Myocardial Injury. Circulation. 2021;144(6):495-498.
- ACC/AHA Guideline for the Evaluation and Diagnosis of Chest Pain. J Am Coll Cardiol. 2021;78:e187-e285.
