Key result
French PMSI ICD-10 codes accurately identify ventricular arrhythmias and cardiac arrest with ~96% PPV.
Why the study?
Assessing ventricular arrhythmias and cardiac arrest in electronic health databases requires prior validation of administrative ICD-10 diagnosis codes to ensure accuracy.
Observational (n=182)
No
Effect estimate: PPV 0.93 to 0.99 depending on code
ICD-10 administrative codes for ventricular arrhythmias and cardiac arrest used in the French hospital discharge database demonstrate excellent positive predictive value (>90%), validating their reliable use in pharmacoepidemiological research.
Supports reliable use of these codes in French pharmacoepidemiologic research on ventricular arrhythmias; leaves generalizability to other databases open.
Purpose: Drug-induced QT interval prolongation increases the risk of ventricular arrhythmias and cardiac arrest. Assessing these outcomes in electronic health databases and evaluating their association with drug exposure requires prior validation of administrative diagnostic codes. This study aimed to evaluate the positive predictive value (PPV) of International Classification of Diseases, 10 th revision (ICD-10) codes for ventricular arrhythmias and cardiac arrest in the French hospital discharge summary database ( Programme de Médicalisation des Systèmes d’Information (PMSI)). Methods: In a retrospective validation study, we examined PMSI data recorded at Amiens University Medical Center (Amiens, France). Hospital stays between September 2022 and August 2025 with ICD-10 codes for cardiac arrest (I46.0, I46.9), ventricular tachycardia (I47.2), or ventricular fibrillation/flutter (I49.0) were identified. Events were confirmed by electronic medical record review. PPVs with 95% confidence intervals (CI) were calculated. The study design prevented the estimation of sensitivity, specificity, and negative predictive values. Sensitivity analyses were stratified by year, and by type of care unit. Exposure to QT-prolonging medications prior to events was also assessed. Results: In a population of 182 patients (median age: 66 years; 75% males; in-hospital mortality rate: 45%), PPVs [95% CI] for all diagnosis codes were high: 0.98 [0.95; 1.00] for cardiac arrest with successful resuscitation, 0.96 [0.79; 1.00] for unspecified cardiac arrest, 0.99 [0.94; 1.00] for ventricular tachycardia, and 0.93 [0.87; 0.97] for ventricular fibrillation/flutter. Sensitivity analyses yielded consistent results. Overall, 49.7% of patients had been exposed to ≥ 1 QT-prolonging medications prior to a validated event, and 26.5% to ≥ 2 of these medications concomitantly. Conclusion: The high PPVs in PMSI data supports the use of the evaluated ICD-10 codes for outcome identification in pharmacoepidemiological studies of electronic health databases in general, and those investigating QT-prolonging medication use and the risk of serious cardiac events in particular.
No takes yet. Share an insight, caveat, or question.
Pluquet et al. (2026) conducted an observational in Ventricular arrhythmias and cardiac arrest (n=182). ICD-10 diagnosis codes (I46.0, I46.9, I47.2, I49.0) vs. Electronic medical record review was evaluated on Positive predictive value (PPV) of ICD-10 codes (PPV 0.93 to 0.99 depending on code). ICD-10 diagnosis codes for ventricular arrhythmias and cardiac arrest in the French PMSI database demonstrated high positive predictive values ranging from 0.93 to 0.99 when validated against medical records.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: