Abstract
Video-electroencephalogram (EEG) monitoring in the epilepsy monitoring unit (EMU) is essential for managing epilepsy and seizure mimics. Evaluation of care in the EMU would benefit from a validated code set capable of identifying EMU admissions from administrative databases comprised of large, diverse cohorts. We assessed the ability of code-based queries to parse EMU admissions from administrative billing records in a large academic medical center over a four-year period, 2016–2019. We applied prespecified queries for admissions coded as follows: 1) elective, 2) receiving video-EEG monitoring, and 3) including diagnoses typically required by major US healthcare payers for EMU admission. Sensitivity (Sn), specificity (Sp), and predictive value positive/negative (PVP, PVN) were determined. Two approaches were highly effective. Incorporating epilepsy, seizure, or seizure mimic codes as the admitting diagnosis (assigned at admission; Sn 96.3%, Sp 100.0%, PVP 98.3%, and PVN 100.0%) or the principal diagnosis (assigned after discharge; Sn 94.9%, Sp 100.0%, PVP 98.8%, and PVN 100.0%) identified elective adult EMU admissions with comparable reliability (p = 0.096). The addition of surgical procedure codes further separated EMU admissions for intracranial EEG monitoring. When applied to larger, more comprehensive datasets, these code-based queries should enhance our understanding of EMU utilization and access to care on a scalable basis.
•Code-based queries identify elective EMU admissions from a hospital billing dataset•Inclusion of diagnoses required by payers for EMU admission increased accuracy•Use of these queries may facilitate monitoring of EMU utilization in large cohorts.