Abstract
The incidence of hypertensive (HTN) crises in older adults has more than doubled over the past two decades. ICD-10 introduced codes for HTN urgency and emergency, but their validity has not been evaluated. We analyzed a 20% national sample of fee-for-service Medicare beneficiaries aged ≥66 years (2016-2019). Medical records were reviewed for 220 randomly sampled emergency department visits or hospitalizations with a primary ICD-10 HTN crisis code (100 HTN urgency [I16.0], 100 HTN emergency [I16.1] and 20 unspecified [I16.9]). Physicians abstracted and adjudicated cases. Two gold standards (GS) were used: (1) treating physician diagnosis and (2) adjudicated diagnosis based on BP (SBP ≥180 mmHg or DBP ≥110 mmHg), with or without end-organ damage. PPVs and 95% confidence intervals were estimated. Among 220 charts sampled from 53,100 HTN crisis cases (mean age 73 years; 33% male; 66% White), PPV for HTN emergency (I16.1) was 91% (95% CI, 85%-96%) using the treating physician GS and 78% (95% CI, 71%-86%) using the adjudicated GS. PPV for HTN urgency (I16.0) was 97% (95% CI, 93%-99%) and 81% (95% CI, 72%-88%), respectively. Overall PPV for any HTN crisis (I16.X) using the adjudicated GS was 95% (95% CI, 91%-97%). Including unspecified codes reduced PPV to 86% (95% CI, 79%-92%) and 78% (95% CI, 71-86%). ICD-10 HTN crisis codes demonstrate high PPV, supporting their use in epidemiologic and health services research.