Food and Drug Administration (FDA) warnings and medical association guidelines focus on potential health risks associated with medications, including psychotropic and smoking cessation medications, suggesting modifications for clinical practice. The VA Pharmacy Benefits Management Services (VA PBM) is responsible for initiating VA responses to FDA warnings. Despite VA PBM's use of several approaches to convey warnings to Veterans Integrated Service Networks (VISNs), VA facilities, and providers, much remains unknown about multiple contextual and organizational factors influencing responsiveness to these warnings. A better understanding of VA system and provider level responsiveness could lead to developing and sharing best practices and informing future strategies to improve warning adoption, thereby improving patient care and population safety.
Given that considerable VA facility-level variation exists in prescribing psychotropic and smoking cessation drugs, we sought to examine factors associated with higher and lower levels of responsiveness to warnings, and to use these insights for quality improvement within VA and elsewhere. We sought to understand variation in the adoption of warnings by VISNs, VA facilities, and providers, and the effects of varying levels of pharmacy integration on response to warnings. We used three examples of recent warnings: 1) sudden cardiac events associated with high doses citalopram [Celexa]; 2) impaired driving associated with zolpidem [Ambien]; and 3) change in behavior such as hostility, agitation, depressed mood, and suicidal thoughts or actions [Chantix].
We addressed the following three specific aims for Celexa and Ambien: Aim 1: To assess and describe PBM and VISN-level responsiveness and variation in responsiveness to warnings regarding psychotropic medications. Aim 2: To assess prescribing patterns before and after warnings regarding psychotropic medications. Aim 3: To understand specific strategies used by facilities and providers with high response to warnings and barriers encountered by facilities and providers with low response to warnings.
We addressed the following supplemental aim for Chantix: To assess prescribing patterns before and after warnings regarding smoking cessation medications using an interrupted time series analysis.
Using both qualitative (theory-driven semi-structured interviews) and quantitative methods (surveys and administrative data), we evaluated multiple factors that could influence facility and provider responsiveness to the Celexa and Ambien warnings. Aim 1 detailed the processes by which VHA developed prescribing policies and guidance, how it disseminated this information, and perceived barriers and facilitators to adoption across the system, based on semi-structured interviews with national VA PBM leaders and pharmacy leaders at each VISN (N=21). Aim 2 employed segmented regression techniques using administrative data pre- and post-warning to identify changes in psychotropic medication prescriptions and patient health monitoring in VHA by VISN and facility. Aim 3 detailed how facilities and providers changed practice in response to warnings and PBM dissemination efforts. We conducted interviews with facility leaders (pharmacy, primary care, mental health) in Aim 3A, and surveys with primary care providers (PCPs), and psychiatrists in Aim 3B, all of whom we intentionally sampled based on Aim 1 and 2 findings.
We also conducted an interrupted time series analysis to assess prescribing patterns before and after warnings regarding smoking cessation medications (Chantix).
Qualitative findings: Based on our interviews with VISN Pharmacy Executives (VPEs; Aim 1) we found that VISNs with the highest degree of clinical integration were the most responsive to FDA warnings. VISNs with the smallest relative change in high-dose citalopram had only a mid-range degree of clinical integration. We found that 5 out of the 16 VISNs we interviewed had both high and low performing facilities when looking at both qualitative and quantitative measures.
Our facility level interviews (Aim 3A) found that the most commonly mentioned mechanisms to alert and educate providers about medication issues included having pharmacists attend physician staff meetings and receiving emails directly from Pharmacy. Very few facility level interviewees could recall VISN-related activities related to the citalopram warning.
Our provider survey (Aim 3B) found that 71% recalled the citalopram warning and 65% recalled the zolpidem warning. The most commonly reported mechanism for staff to learn about FDA warnings were: from their P&T committee, emails from their Service Chief, emails from the FDA, staff meetings, and popular media.
Quantitative findings: High dose citalopram use decreased abruptly following the 2011 and 2012 FDA warnings. We found increases in the use of sertraline and other antidepressants during this time period, suggesting possible substitution of citalopram (Aim 2).
We found that higher than recommended dose zolpidem use decreased immediately after each warning and was 2% lower in 2014 among men. For women prescribed zolpidem, nearly half remained on higher than recommended dose following the drug safety warnings. We found that use of any dose zolpidem quadrupled after the 2007 warning. After the 2013 FDA warning, any dose zolpidem use declined, but remained at more than three times the use prior to the 2007 VHA warning for zolpidem (Aim 2).
Since its introduction in 2006, varenicline use increased rapidly and peaked in 2008, when ~1.5% of Veteran smokers used varenicline in VHA. Varenicline use subsequently declined to less than 0.3% of Veteran smokers in 2015. Since 2016, when the FDA removed the varenicline warning, use has been steadily rising (supplemental aim).
We shared our work with PBM leadership during an in-person meeting in August 2019. They can use our findings to better understand how pharmaceutical and therapeutic alerts are acted on in the system, in order to directly benefit VA practice and patients. We provided feedback to PBM that based on our experiences in this study, we suggest that similar research begin at the time a warning is released (rather than waiting for subsequent grant funding) to reduce recall limitations and challenges, and that longitudinal interviews should be conducted over the course of a warning.
- Ross EL, Vijan S, Miller EM, Valenstein M, Zivin K. The Cost-Effectiveness of Cognitive Behavioral Therapy Versus Second-Generation Antidepressants for Initial Treatment of Major Depressive Disorder in the United States: A Decision Analytic Model. Annals of internal medicine. 2019 Dec 3; 171(11):785-795.
- Bergmans RS, Mezuk B, Zivin K. Food Insecurity and Geriatric Hospitalization. International journal of environmental research and public health. 2019 Jun 28; 16(13).
- Kim HM, Gerlach LB, Van T, Yosef M, Conroy DA, Zivin K. Predictors of Long-Term and High-Dose Use of Zolpidem in Veterans. The Journal of clinical psychiatry. 2019 Feb 5; 80(2).
- Kim HM, Gerlach LB, Yosef M, Stano C, Conroy DA, Valenstein M, Pfeiffer PN, Sales AE, Zivin K. Responsiveness of Veterans Affairs Health Care System to Zolpidem Safety Warnings. Journal of clinical sleep medicine : JCSM : official publication of the American Academy of Sleep Medicine. 2018 Jul 15; 14(7):1135-1141.
- Gerlach LB, Kim HM, Yosef M, Kales HC, Henry J, Zivin K. Electrocardiogram Monitoring After the Food and Drug Administration Warnings for Citalopram: Unheeded Alerts?. Journal of the American Geriatrics Society. 2018 Aug 1; 66(8):1562-1566.
- Gerlach LB, Kim HM, Yosef M, Sales AE, Stano C, Kales HC, Zivin K. Assessing Responsiveness of Health Systems to Drug Safety Warnings. The American journal of geriatric psychiatry : official journal of the American Association for Geriatric Psychiatry. 2018 Apr 1; 26(4):476-483.