Identifying Renal Dysfunction in Stroke Patients Using Diagnostic Codes in the Taiwan National Health Insurance Research Database

Stroke patients with renal dysfunction have higher risks of mortality and poor outcomes than those without (1). The National Health Insurance Research Database (NHIRD) of Taiwan has been used for studying stroke patients with end-stage renal disease (2). Because the NHIRD has no laboratory data, diagnostic codes must be used to identify patients with renal dysfunction. However, the validity of renal dysfunction diagnoses in the NHIRD has not been determined. We aimed to validate the renal diagnostic codes in the NHIRD. NHIRD claims data were extracted for stroke patients hospitalized in a medical center and matched with the stroke registry. Relevant diagnostic codes (Supporting Information Table S1) in any outpatient or inpatient claims one-year before index admission were retrieved. We determined the validity of diagnostic codes in renal dysfunction and severe renal dysfunction, using estimated glomerular filtration rates <60 and <30 ml/ min/1·73 m derived from the Taiwan Stroke Registry as reference standards, respectively. Matched data for 2103 stroke patients were analyzed (Supporting Information Fig. S1). Of the 2103 stroke patients, the age (mean ± standard deviation) was 67 ± 13 years, and 60% were men. The validity of diagnostic codes in severe renal dysfunction was better than in renal dysfunction (Table 1). The NHIRD may be used in the future to study stroke patients with severe renal dysfunction, a subpopulation particularly vulnerable for poor clinical outcomes (3) but usually excluded from clinical trials.