Research Aims: In 2007, VA initiated comprehensive suicide analytics for patients receiving care in the VHA health system. This scientific investment enables rapid replication of scientific findings, to inform whether published non-VHA studies may generalize to the VHA patient population. Using administrative claims databases in Ontario, Canada, Mamdani and colleagues (2019) documented the odds of suicide among older patients receiving ARBs is 1.63 (95% CI, 1.33-2.00) times higher when compared to those receiving ACEIs. The study included 964 cases and 3856 controls. This important finding requires replication assessments, both as part of ongoing scientific evidence synthesis and to inform potential suicide prevention policies (Katz, 2019). To advance these goals, we examined these associations among Veterans receiving care in the VHA health system.
Methods: Replicating a case-control design, the study cohort included all suicide decedents, 2015-2017, with VHA inpatient or outpatient encounters in the year of death or the prior year and with either ACEI or ARB medications exposure in the 100 days prior to death. Analyses excluded individuals who received both medication types in this period. VHA records were drawn from the Corporate Data Warehouse and mortality was assessed per National Death Index search results included in the VA/DoD Mortality Data Repository. 5243 controls were matched to 1311 cases using a 4:1 ratio, matching on age, sex, and hypertension and diabetes diagnoses. Conditional logistic regression was then used to adjust for covariates. Following Mamdani et al., these included Charlson Comorbidity Index score, long-term residential care, drug use, psychiatrists and cardiology visits, diagnosis of alcohol use disorder, stroke chronic kidney disease, chronic liver disease, and coronary artery disease, all based on the preceding year. Income was not included, due to insufficient data. Two sensitivity analyses were conducted: 1) comparisons of results using only those covariates that differed significantly between case and control groups, and 2) age-restricting the VHA cohort to those 65 and older, to parallel Mamdani et al.
Results: ARBs were received by 21.6% of controls and 19.6% of cases, while ACEIs were received by 78.4% of controls and 80.4% of cases. The crude odds ratio for ARBs versus ACEIs was 0.885 (95% CI: 0.759, 1.031). Controlling for covariates, the adjusted odds ratio for ARBs was 0.905 (95% CI: 0.768, 1.066). Similar findings were observed in sensitivity analyses using only those covariates that differed significantly between groups, and when restricting the cohort to individuals ages 65 and older.
Conclusions: In contrast to recent observations from Ontario, Canada, study findings did not observe a significant difference in odds of suicide mortality by receipt of ARBs versus ACEIs. Although non-significant, the point estimate was also in the opposite direction.
Implications: Dynamics of suicide risk among patients treated for hypertension in Ontario, Canada, may not generalize to the population of Veterans receiving similar treatment in the VA health system. The study also documents the utility of established structures for suicide surveillance and analytics in health systems.
Katz IR. Concerns Raised by a Study of Suicide as an Adverse Drug Effect—Replicating Findings From Real-World Data. JAMA Netw Open. Published online October 16, 20192(10):e1913284. doi:10.1001/jamanetworkopen.2019.13284
Mamdani M, Gomes T, Greaves S, et al. Association Between Angiotensin-Converting Enzyme Inhibitors, Angiotensin Receptor Blockers, and Suicide. JAMA Netw Open. Published online October 16, 20192(10):e1913304. doi:10.1001/jamanetworkopen.2019.13304