A multihospital, single health system validation of International Classification of Diseases, 10th Revision, Clinical Modification coding for status epilepticus in the United States

Status epilepticus (SE) is a common, life-threatening neurologic emergency. Understanding population-level outcomes after SE requires a validated case definition, yet International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) codes of SE have not been well-validated in US populations since adoption in 2015. We aimed to determine whether the ICD-10-CM code-based definitions accurately identify SE in the in-hospital setting. The population included all ages (excluding neonates) admitted to a Mount Sinai Health System (MSHS) intensive care unit (ICU) in 2019. A data collection form was developed, tested, and used by trained reviewers. Every admission in a random month (November) was reviewed to determine whether all SE cases had at least one code for epilepsy, seizure, or convulsion followed by all charts with an ICD-10-CM diagnosis code for seizure/epilepsy/convulsion in 2019. Chart review data were linked to MSHS electronic medical record data. Sensitivity (Sn), specificity (Sp), negative predictive value (NPV), and positive predictive value (PPV) with 95% confidence intervals (CIs) and Youden index were calculated for ICD-10 coding of SE (G40.xx1 or G40.xx3, as G41 was not adopted in the United States). MSHS had 13 694 ICU admissions in 2019, for which 1851 charts were reviewed and of which 173 were admissions with definite SE. The ICD-10-CM case definition (G40.xx1) has an Sn of 68.7% (95% CI = 61.5-75.8) and Sp of 92.6% (95% CI = 91.4-93.8). PPV was 47.4% (95% CI = 41-53.8), and NPV was 96.8% (95% CI = 95.9-97.6). Youden index was 61.3%. ICD-10-CM coding for SE has high specificity but limited sensitivity. These findings align with SE prevalence studies showing a decrease in prevalence with the change from ICD-9-CM to ICD-10-CM, which may be related to the United States' unique adoption of ICD-10-CM, which did not include the standalone SE code (G41). Our findings emphasize the importance of revision and improvement of coding practices to best represent the prevalence of SE, and of consideration when planning for the next iteration of ICD coding.