Lipophilic β-Blockers and Suicide in the Elderly
β-Adrenergic blockers (β-blockers) are among the most widely prescribed drugs in the world, with more than 191 million prescriptions dispensed in the United States in 2010 alone.1 They are commonly used to manage hypertension, heart failure, and arrhythmias and for the secondary prevention of myocardial infarction. Some β-blockers have been associated with adverse effects involving the central nervous system (CNS), including fatigue, hallucinations, poor sleep, and depression.2–4 This may reflect their passage across the blood-brain barrier, which is heavily influenced by lipophilicity. The entry of lipophilic β-blockers such as propranolol into the CNS is more than twice that of β-blockers of intermediate lipophilicity such as metoprolol,5,6 whereas hydrophilic β-blockers such as atenolol do not appreciably enter the CNS.5,6
A possible association between β-blockers and depression was first described in 1967 in patients taking propranolol.7 Some evidence suggests that depression is particularly common in patients receiving propranolol,8,9 whereas other studies dispute these findings.10,11 However, several case reports describe reversible depression after initiating propranolol or increasing the dose.12,13
Although an association between lipophilic β-blockers and suicide has been postulated, only 2 small studies have explored this phenomenon, reaching conflicting conclusions.14,15 Given the widespread use of β-blockers, any differential risk of suicide among agents could have significant public health implications. We examined the association between β-blocker lipophilicity and suicide in older patients.
We conducted a population-based case-control study of all Ontario residents 66 years or older between January 1, 1993, and December 31, 2012. These individuals had universal access to physician services, hospital care, and prescription drug coverage. The study was approved by the Research Ethics Board of Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada.
We linked several administrative databases in an anonymized fashion. We identified prescription drug records using the Ontario Drug Benefit Database, which contains the prescription claims of Ontarians 65 years or older. We defined case patients as those who died of suicide within 100 days of receiving a prescription for an oral β-blocker. The date of the suicide was considered the index date for all analyses. For each case, we selected up to 4 control subjects who did not die of suicide on or before the index date of the corresponding case, matching on age (within 1 year), sex, hospitalization for coronary artery disease, and diagnosis of hypertension in the preceding year. Control subjects were assigned the same index date as their corresponding case, and like cases received a single β-blocker in the preceding 100 days. We excluded cases that could not be matched to at least 1 control, as well as patients during their first year of eligibility for prescription drug coverage (aged 65 years) to avoid incomplete medication records.
We identified all prescriptions for oral β-blockers in the 100 days preceding the index date, categorizing the lipophilicity of each as high (propranolol and labetalol), intermediate (metoprolol, bisoprolol, carvedilol, acebutolol, timolol, and pindolol), or low (atenolol, nadolol, and sotalol) based on the octanol-water partition coefficient for each.2 We use the term “hydrophilic” to denote β-blockers of low lipophilicity. We used a 100-day exposure period because this is the maximum reimbursed duration of a prescription in Ontario, although most prescriptions are issued for shorter durations. To test the robustness of our conclusions and to reduce the possibility of exposure misclassification, we replicated our primary analysis using a 60-day exposure period.
To test the specificity of our findings, we also examined the association between β-blocker lipophilicity and death from lymphoma. Because there is no expected association between β-blocker lipophilicity and death from lymphoma, we reasoned that a null association in this analysis would enhance causal inference in our main analyses.