RESUMEN
PURPOSE: Epidemiological studies of primary subarachnoid hemorrhage (pSAH) frequently include population-based death registries for case finding. The positive predictive value of pSAH diagnoses in death registries is unknown. METHODS: This cross-sectional study identified all people in Ontario, Canada with pSAH listed as a cause of death between 2013 and 2017. pSAH was classified as "very likely" if diagnosis of pSAH was confirmed by autopsy, there was a previous hospitalization where pSAH probability exceeded 85% or death was preceded within a week by an emergency room visit where pSAH probability exceeded 25%. pSAH was classified as "very unlikely" if previous cerebrovascular imaging had never been done. Remaining cases were classified as "pSAH status unknown". RESULTS: 1,613 deaths attributed to pSAH were identified (mean 322/year). pSAH classification frequencies were as follows: very likely 528 (32.7%); very unlikely 433 (26.8%); and status unknown 652 (40.4%). CONCLUSION: We found that a quarter of pSAH cases in our province's death registry were very unlikely to be true pSAH while 40% had unknown veracity. These data should be considered when using death registries for pSAH case finding.