Background:
Myocardial infarction (MI) is a key end point in cardiovascular medicine. Similarly to other parts of the world, German health insurance claims, coded for reimbursement, are increasingly used for observational research, however, their validity has not yet been confirmed by gold-standard, that is electronic health record (HER) chart review.
Objectives:
To assess the validity of using medical claims for identifying patients with MI, documented utilizing the International Classification of Diseases, 10th Revision (ICD-10) German Modification (GM) coding system.
Methods:
This study utilized EHR-linked claims data from a German university hospital between 2019 to 2024. Claims for acute MI included ICD codes I21.0, I21.1, I21.2, I21.3, I21.4, I21.9, I22.0, I22.1, I22.8, I22.9, and for old MI ICD code I25.2. Linkage to corresponding EHRs allowed for claims validation by identifying the MI diagnosis through chart review. This process was carried out by two independent researchers following predefined guidelines. Inter-rater reliability (IRR) was assessed to ensure consistency. Diagnoses from charts were categorized into acute MI (further classified into ST-elevation MI (STEMI), non-ST-elevation MI (NSTEMI)) and old MI. Performance of the codes in correctly identifying MI was reported as positive predictive value (PPV). Sensitivity analyses included restriction based on key demographic and temporal factors (admission period, sex, and age at admission).
Results:
Among 8,533 patients identified with a MI code, a random sample of 470 cases with an acute MI claim and 230 with an old MI claim were reviewed against their EHR. The mean age at admission across the cohort was 66.8 years, with males comprising 82.4% of the study population. IRR for the MI status was 98.3% for acute MI and 97.6% for old MI with Cohen’s kappa indicating moderate and substantial agreements respectively. The PPV for acute MI in any position was 94.0% [(95%-CI) 91.5-96.0], with STEMI at 89.8% [86.0-92.9] and NSTEMI at 80.6% [73.1-86.7]. In the primary position, the PPV for acute MI was 97.3% [95.2-98.6], with STEMI at 90.6% [86.8-93.7] and NSTEMI at 86.3% [78.0-92.3]. In the secondary position, the PPV for acute MI was 78.2% [68.0-86.3], with STEMI at 83.3% [65.3-94.3] and NSTEMI at 67.3% [53.3-79.3]. The PPV for old MI was 93.0% [89.0-96.0] in secondary position, as none of the EHRs were coded for an old MI in primary position. Sensitivity analysis showed results consistent with the primary findings across all subgroups.
Conclusions:
To our knowledge, this is the first study ascertaining the validity of ICD-10-GM coded claims for MI via gold-standard chart review in EHRs. Overall, a very high PPV for MI diagnosis was observed. These findings suggest that claims identifying MI are a reliable source for observational studies.
