Please confirm that your email address is correct, so you can successfully receive this alert.
Dr. Blow has served as a consultant to the Hazelden Foundation. The other authors report no financial relationships with commercial interests.
The parent study was funded by VA grant IIR 10-176-3 (to Dr. Zivin) and VA Health Services Research and Development Services grants CD2 07-206-1 (to Dr. Zivin), CDA 10-036-1 (to Dr. Pfeiffer), and CDA 09-204 (to Dr. Bohnert). No funding was received for writing this commentary.
From the VA Serious Mental Illness Treatment Resource and Evaluation Center, Ann Arbor, Mich.; the VA Center for Clinical Management Research, Ann Arbor; and the Department of Psychiatry and Division of Cardiovascular Diseases, University of Michigan Medical School, Ann Arbor.
Address correspondence to Dr. Zivin (email@example.com).
Copyright © 2014 by the American Psychiatric Association
Citalopram, which was approved by the U.S. Food and Drug Administration (FDA) in 1998, has been widely regarded as a first-line agent in the treatment of depression because of its minimal drug interactions, its demonstrated safety in older adults and medically ill patients, its low cost, and its relative effectiveness (1). These attributes led to the selection of citalopram as the initial treatment for the National Institute of Mental Health-funded Sequenced Treatment Alternatives to Relieve Depression (STAR*D) trial, in which over half of the patients received more than 40 mg/day of the medication, and there was no clear signal for cardiac adverse events (1, 2).
The FDA warning for citalopram thus appeared to be at odds with previous research findings, and it recommended a substantial change in routine clinical practice. We therefore sought to examine whether the relationship between citalopram dosage and adverse events was consistent with the FDA warnings under real-world conditions. In our study (3), we assessed whether among patients treated for depression in the Veterans Health Administration, the largest national integrated health care system in the country, we could find an increased incidence of abnormal heart rhythms (including torsade de pointes) or mortality associated with high-dosage citalopram. Our study of more than 600,000 patients taking citalopram found no elevated risks for either ventricular arrhythmias or all-cause, cardiac, or noncardiac mortality associated with daily doses above 40 mg compared with lower doses. Instead, higher doses were associated with fewer adverse outcomes. Our study is not alone in finding no significant association between treatment with citalopram and cardiac risks. Recent research has confirmed that QT prolongation with the use of citalopram and other antidepressants has been modest (4).
Our findings do not substitute for or invalidate the data presented in the FDA warnings, although the study cited as the basis for the warnings has not been published in the peer-reviewed scientific literature. It is unknown how many patients in that study reached clinically significant QT prolongation >500 ms and how many postmarketing reports of QT prolongation and torsade de pointes the FDA has received. Thus, it is not possible at this time to fully understand the relative merits and weaknesses of these data relative to other evidence for informing clinical practice. The warnings may have better served the clinical and research communities by placing the new data within the context of the numerous studies from more than a decade, including epidemiological studies of routine clinical practice and randomized trials, indicating relative safety and effectiveness of citalopram.
There are limitations to the methods used in epidemiological studies, as we discussed in our study (3) and as highlighted by Bird et al. (5). The central limitations are possible selection bias and unmeasured confounding. Bird et al. raise concerns regarding potentially insufficient controls for comorbidity, despite our utilization of a continuous comorbidity index. In response, we tested our comorbidity adjustment by using each individual disorder (with the exception of depression [N=29]) in the Elixhauser comorbidity index (6) as individual covariates. We detected no significant changes in the direction or magnitude of our findings. We recognize that as with any study, it is not possible to account for all potential confounders.
It is also possible that we did not find negative health outcomes associated with higher daily doses of citalopram because prescribing clinicians took into account potential risks and prescribed lower doses to patients at greater risk of arrhythmia. Under this scenario, in which clinicians were already prescribing citalopram in such a way as to nullify any detectable increase in cardiac risk with daily doses above 40 mg, the added value of the FDA warnings is not completely obvious. To address concerns that our findings might be explained by differences among patients titrated to high-dosage citalopram and those initiating and remaining on low-dosage citalopram, we reanalyzed our data including only the first observed dose period for each patient. Higher dosage was still not associated with higher rates of any of the outcomes examined. Previous research demonstrated that antidepressant use decreased and suicide rates increased after FDA warnings were issued regarding potential suicide risk among pediatric patients using selective serotonin reuptake inhibitors (SSRIs) (7). In other words, warnings themselves have the potential to do substantial harm by decreasing appropriate use (8).
Bird et al. also criticize our study for using outcome definitions that include but are not specific to torsade de pointes and its related mortality. We note that the outcome used as the basis for the FDA warnings, QT prolongation, also has its limitations, as summarized by Habib and Gan (9). The QT interval is only a surrogate measure for torsade de pointes, and there are challenges that remain with its measurement. For example, there is no consensus on the way to correct the QT interval (QTc) to account for heart rate changes, and the threshold for defining a normal QT interval depends on this correction. These issues with estimating QTc are spurring ongoing research into alternative measures of abnormal cardiac repolarization.
The true incidence of drug-related torsade de pointes is unknown. Bird et al. acknowledge that torsade events are exceedingly rare and that including other forms of arrhythmia or cardiac mortality in our outcome measures removes the possibility of detecting the presumed proarrhythmic effect of high-dosage citalopram. The FDA Adverse Event Reporting System is used to identify cases of torsade de pointes associated with medication use, yet information used from this system is insufficient to detect whether risks are associated with a particular medication and dosage (for example, many patients are taking multiple medications). Several other SSRIs had the same number of adverse event reports associated with them as citalopram, yet they are not subject to the same warning (10). Even though torsade de pointes is the most common reason for medications receiving FDA warnings, only 10 of 140 medications with known torsade risks have received FDA warnings or have been removed from the market (11). Thus, it appears that FDA actions are inconsistent regarding how to weigh and manage a medication’s risk of torsade de pointes relative to the benefits associated with its use.
Ultimately, it is up to clinicians and their patients to weigh these potential risks with the potential benefits. Clearly, there is a perceived clinical benefit to citalopram dosages above 40 mg/day; 12% of prescriptions in our study exceeded this threshold. Even if a randomized controlled trial did not find greater efficacy for 60 mg compared with 40 mg of citalopram, randomized controlled trials are limited in their generalizability to real-world patients and practices (12). No study methodology is perfect. Patients, providers, and health systems deserve to have all available information on which to base their health care decisions. We believe that FDA warnings should incorporate studies across scientific disciplines assessing medication risks, along with the potential risks and benefits of complying with the warnings themselves.
Download citation file:
Web of Science® Times Cited: 2