The Consistency of Hypercapnic Respiratory Failure Case Definitions in Electronic Health Record Data

Background: Several studies suggest a high prevalence and disease burden associated with hypercapnic respiratory failure of any cause in emergency department (ED) and inpatient settings. However, these studies use different case definitions. The consistency and validity of these case definitions have not been assessed. Research Questions: Do the various case definitions from health record-based studies of adults with hypercapnic respiratory failure identify the same, or similar, patients? Methods: We identified case definitions for hypercapnic respiratory failure from the peer-reviewed literature. We constructed an emulation data set enriched for the presence of hypercapnic respiratory failure from TriNetX (TriNetX, LLC), which aggregates health records from 76 hospitals in the United States. Adult ED and inpatient encounters occurring in 2022 were eligible for inclusion. We assessed consistency among case definitions by applying them to the emulation data set on the day of ED or hospital admission. We calculated Cohen's κ value and compared the patient characteristics and short-term outcomes for the cohorts defined by each case definition. We evaluated the accuracy of hypercapnic respiratory failure diagnosis codes for capturing hypercapnia demonstrated on first-day arterial blood gas samples. Results: Ten case definitions for hypercapnic respiratory failure were identified and were amenable to emulation. The emulation data set contained 515,286 encounters. Case definitions showed limited agreement (median κ = 0.35; interquartile range, 0.21-0.56) for hypercapnic respiratory failure identification. Key demographic characteristics, comorbidities, and outcomes such as receiving ventilatory support (range, 29%-100%) and 60-day mortality (range, 12%-28%) varied substantially among the resulting cohorts. Diagnosis codes were insensitive for capturing first-day arterial blood gas-demonstrated hypercapnia (sensitivity, 23.5%; 95% CI, 23.1%-23.8%). Interpretation: Case definitions from studies of hypercapnic respiratory failure identify patients with widely different attributes and outcomes, thereby hampering interpretation of associated study findings. Standardization of case definitions is necessary to improve the rigor and generalizability of research on hypercapnic respiratory failure.