Defining cause of death in a contemporary cohort with ANCA-associated vasculitis (AAV): A comparison of electronic health record and death certificate data.

Objectives: Patients with antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) face excess mortality compared with the general population. Mortality in clinical epidemiology research is often examined using death certificate diagnosis codes; however, the sensitivity of such codes in AAV is unknown.

Methods: We performed a retrospective cohort study using the Mass General Brigham AAV Cohort, including patients with AAV who died between 2002 and 2019. Causes of death were determined by electronic health record (EHR) review (reference gold standard) and via cause of death diagnosis codes on death certificates. We calculated the sensitivity of death certificate diagnosis codes for AAV.

Results: Of 684 patients in the registry, 184 died, 92 (52 %) of whom had adequate EHR data available determine cause of death and 72 (40 %) of whom had both EHR and death certificate data available. Death due to AAV, infection, cardiovascular disease, and cancer occurred in 8 %, 29 %, 5 %, and 18 %, respectively, when ascertained by manual review, as opposed to 0 %, 11 %, 25 %, and 21 %, as determined by death certificates. The sensitivity of AAV diagnosis codes for AAV was 16.6 % (95 % CI: 10.5, 22.6) among all patients with death certificate data available.

Conclusion: In a contemporary cohort of patients with AAV, infection was the most common cause of death, while death due to AAV itself was rare. We found a high degree of discordance between causes of death determined by manual review and death certificate diagnosis codes. Mortality research on AAV should include linkage to medical records data to reduce potential bias.