Toh S, Baker MA, Brown JS, Kornegay C, Platt R; Mini-Sentinel Investigators. Rapid assessment of cardiovascular risk among users of smoking cessation drugs within the US Food and Drug Administration’s Mini-Sentinel program. JAMA Intern Med. Published online March 25, 2013. doi:10.1001/jamainternmed.2013.3004.
eAppendix. Caveats of the smoking cessation drug rapid query
Toh S, Baker MA, Brown JS, Kornegay C, Platt R, Mini-Sentinel Investigators FT. Rapid Assessment of Cardiovascular Risk Among Users of Smoking Cessation Drugs Within the US Food and Drug Administration's Mini-Sentinel Program. JAMA Intern Med. 2013;173(9):817-819. doi:10.1001/jamainternmed.2013.3004
Author Affiliations: Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, Boston, Massachusetts (Drs Toh, Baker, Brown, and Platt); and Office of Surveillance and Epidemiology, Center for Drug Evaluation and Research, US Food and Drug Administration, Silver Spring, Maryland (Dr Kornegay).
In June 2011, the US Food and Drug Administration (FDA) issued a Drug Safety Communication indicating that varenicline tartrate, a drug prescribed for smoking cessation, may increase the risk of certain cardiovascular events in individuals with cardiovascular disease.1 The finding was based on the FDA's review of a randomized placebo-controlled trial of 714 smokers.2 In July 2011, the FDA requested that the Mini-Sentinel program perform a rapid safety assessment of the drug.
Mini-Sentinel is part of the FDA's Sentinel Initiative, a national system under development for monitoring medical product safety.3- 5 The program has created a distributed network of electronic health care databases with more than 125 million lives and 350 million person-years of longitudinal observation time.
In this rapid assessment, we identified individuals who filled a first prescription for varenicline or the comparator drug, bupropion hydrochloride, between January 1, 2006, and the most recent date available at each data partner as of July 5, 2011. We further restricted the cohort to individuals who were 20 years or older on the date of first varenicline or bupropion dispensing (the index date) and who met the following criteria during the preceding 180 days: (1) continuously enrolled in the health plan with medical and drug coverage; (2) had no dispensing of either drug; (3) had no diagnoses of acute myocardial infarction (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM ] code 410.xx), intermediate coronary syndrome or unstable angina (411.1), acute coronary occlusion without myocardial infarction (411.81), or coronary atherosclerosis (414.0x); and (4) had a diagnosis code for tobacco use disorder (305.1).
The outcome of interest was a composite of 3 cardiovascular end points (ICD-9-CM codes 410.xx, 411.1, and 411.81) recorded as the primary diagnosis in an inpatient or emergency department setting. We bridged gaps that were 7 or fewer days between prescription periods, which began with the dispensing date and lasted for the number of days supplied, and extended the treatment episode for 7 days after the end of the last dispensing. The analysis followed up patients from the index date until the occurrence of the outcome or until the end of the first treatment episode, health plan enrollment, or data availability.
We estimated the incidence and incidence rate of the outcome, the crude incidence rate ratio (IRR) comparing varenicline with bupropion, and the Mantel-Haenszel IRR stratified on age group, sex, and data partner (analysis 1). As a secondary analysis, we compared all initiators of varenicline therapy with initiators of therapy with a bupropion product specifically approved for smoking cessation, that is, Zyban and its generic equivalents (analysis 2). In addition, we analyzed all initiators of varenicline and bupropion therapy and did not restrict the analysis to patients with a tobacco use disorder code or those who used a bupropion product approved for smoking cessation (analysis 3). Analysis 3 included the largest number of users but was expected to be more confounded because patients using bupropion for indications other than smoking cessation (eg, depression) would be included and may not be smokers.
Data partners ran a centrally developed and tested analytic program and returned output that consisted entirely of aggregate counts and no protected health information. Mini-Sentinel is a public health surveillance activity that is not under the purview of institutional review boards.6
Analysis 1—the primary analysis—identified 89 519 eligible individuals initiating varenicline therapy and 113 378 eligible individuals initiating bupropion therapy who had a tobacco use disorder code (Table); the adjusted IRR was 1.02 (95% CI, 0.71-1.47). The results did not vary substantially by age group or sex (not shown). In analysis 2, the adjusted IRR was 0.98 (0.43-2.23) when we compared all initiators of varenicline therapy (n = 260 660) with initiators of therapy with bupropion products approved for smoking cessation (n = 11 203). The adjusted IRR was 1.52 (1.21-1.91) in analysis 3 when we compared all initiators of varenicline therapy (n = 260 660) with all initiators of bupropion therapy (n = 745 004).
Initial results were provided to the FDA 2 days after the analytic specifications were finalized, and the final results were provided within 3 weeks. A report with data aggregated across sites is available on the Mini-Sentinel website.7
Recent systematic reviews of randomized controlled trials provided conflicting results on the risk of adverse cardiovascular events associated with varenicline.8,9 This rapid assessment found no consistent evidence of increased cardiovascular risk during the first treatment episode of varenicline among individuals with no recent diagnosis of cardiovascular events when compared with bupropion use. The analysis included patients treated with these drugs in actual clinical practice, thereby providing evidence complementary to findings from randomized trials and other sources.
This rapid analysis was not intended to be a comprehensive evaluation of causality. The results should be interpreted in the context of the goal to provide safety information quickly, the limitations of electronic health data, and the limited analytic approaches available at this early stage of implementation of the rapid assessment program (see the eAppendix for discussion of the caveats of the analysis). However, the analysis demonstrates Mini-Sentinel's rapid response capability and illustrates how a large network of electronic health care databases can be used as part of a learning health care system to provide timely safety information about approved medical products.
Correspondence: Dr Toh, Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, 133 Brookline Ave, Sixth Floor, Boston, MA 02215 (email@example.com).
Published Online: March 25, 2013. doi:10.1001/jamainternmed.2013.3004
Author Contributions: Dr Toh had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Toh, Baker, Brown, Kornegay, and Platt. Acquisition of data: Toh, Baker, Brown, and Platt. Analysis and interpretation of data: Toh, Baker, Brown, Kornegay, and Baker. Drafting of the manuscript: Toh, Baker, and Platt. Critical revision of the manuscript for important intellectual content: Toh, Baker, Brown, Kornegay, and Platt. Statistical analysis: Toh and Baker. Obtained funding: Kornegay and Platt. Administrative, technical, and material support: Toh, Baker, Brown, and Platt. Study supervision: Toh, Brown, Kornegay, and Platt.
Mini-Sentinel Investigators: Site principal investigators (designated as PIs in this list) and lead analysts (all others) (for data partners) at the Mini-Sentinel collaborating institutions: Aetna, Blue Bell, Pennsylvania: Claire M. Spettell (PI) and Yihai Liu; America's Health Insurance Plans, Washington, DC: Barbara Lardy (PI); Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women's Hospital, Boston, Massachusetts: Sebastian Schneeweiss (PI); Division of General Internal Medicine, Brigham and Women's Hospital: David W. Bates (PI); Center for Health Care Quality, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio: Carole Lannon (PI); Department of Statistics, Columbia University, New York, New York: David Madigan (PI); Critical Path Institute, Tucson, Arizona: Klaus Romero (PI); Duke Clinical Research Institute, Duke University, Durham, North Carolina: Lesley H. Curtis (PI); HealthCore, Inc, Wilmington, Delaware: Gregory Daniel (past PI), Nandini Selvam (current PI), and Tom Puenpatom; Group Health Research Institute, Seattle, Washington: Denise Boudreau (PI) and Tyler Ross; Harvard Pilgrim Health Care Institute, Boston: Richard Platt (PI) and Yury Vilk; HealthPartners Research Foundation, Minneapolis, Minnesota: Pamala A. Pawloski (PI) and Lucas Ovans; Henry Ford Health System, Detroit, Michigan: Sharon H. Alford (PI) and Steve Anteau; Lovelace Clinic Foundation, Albuquerque, New Mexico: Margaret J. Gunter (PI) and Michael Schum; Marshfield Clinic Research Foundation, Marshfield, Wisconsin: Robert T. Greenlee (PI), Valerie McManus, and Kerry Sparks; Meyers Primary Care Institute, Worcester, Massachusetts: Jerry H. Gurwitz (PI) and Daniel Peterson; Humana, Miami, Florida: Vinit Nair (PI) and Qianli Ma; Kaiser Permanente Colorado, Denver: Marsha A. Raebel (PI) and Carsie Nyirenda; Kaiser Permanente Georgia, Atlanta: Melissa G. Butler (PI) and Lee Cromwell; Kaiser Permanente Hawaii, Honolulu: Cynthia Nakasato (PI), John V. Parker, and Mark M. Schmidt; Kaiser Permanente MidAtlantic, Rockville, Maryland: Christine Bredfeldt (PI) and Veronica Song; Kaiser Permanente Northern California, Oakland: Lisa J. Herrinton (PI) and Jack Hamilton; Kaiser Permanente Northwest, Portland, Oregon: David H. Smith (PI) and Donald Bachman; Kaiser Permanente Southern California, Pasadena (currently not contributing data): Aniket A. Kawatkar (PI) and Julie W. Liu; OptumInsight, Waltham, Massachusetts (joined the Mini-Sentinel program after the smoking cessation drug rapid assessment was completed): K. Arnold Chan (PI) and Eva Ng; Outcome Sciences, Inc, Cambridge, Massachusetts: Daniel Campion (PI); Chronic Disease Management and Outcomes, Center for Health Services Research on Pharmacotherapy, Rutgers University, New Brunswick, New Jersey: Stephen Crystal (PI); Deep South Musculoskeletal Center for Education and Research on Therapeutics, University of Alabama at Birmingham: Kenneth G. Saag (PI); College of Pharmacy, University of Illinois at Chicago: Bruce L. Lambert (PI); Department of Epidemiology, College of Public Health, The University of Iowa, Iowa City: Elizabeth A. Chrischilles (PI); Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, University of Pennsylvania, Philadelphia: Sean Hennessy (PI); Departments of Pediatrics and Preventive Medicine, Vanderbilt University School of Medicine, Nashville, Tennessee: William O. Cooper (PI) and Tony Morrow; and Weill Cornell Medical College, New York: Alvin I. Mushlin (PI).
Conflict of Interest Disclosures: None reported.
Funding/Support: The Mini-Sentinel program is funded by the FDA through the Department of Health and Human Services contract HHSF223200910006.
Role of the Sponsor: The FDA initiated the smoking cessation drug rapid assessment and participated in the choice of exposure, outcome, and cohort definitions; the interpretation of the data; and the preparation and review of the manuscript. The FDA was not involved in the collection and management of the data. Other aspects of the analysis highlighted in this article were proposed and performed by the Mini-Sentinel Investigators in collaboration with the FDA.
Additional Contributions: Nicolas Beaulieu, MA, Melissa Robb, RN, Susan E. Andrade, ScD, Denise M. Boudreau, PhD, RPh, Melissa G. Butler, PharmD, PhD, Ryan Carnahan, PharmD, MS, Elizabeth A. Chrischilles, MS, PhD, William O. Cooper, MD, MPH, Lesley H. Curtis, PhD, Richard Elmore, MA, Sean Hennessy, PharmD, PhD, Lisa J. Herrinton, PhD, Aniket A. Kawatkar, PhD, Cheryl N. McMahill-Walraven, PhD, Vinit P. Nair, MS, RPh, Jennifer C. Nelson, PhD, Marsha A. Raebel, PharmD, Klaus Romero, MD, MS, Sebastian Schneeweiss, MD, ScD, Jasvinder A. Singh, MD, MPH, David H. Smith, RPh, PhD, Brian L. Strom, MD, MPH, Marcus D. Wilson, PharmD, and Raymond L. Woosley, MD, PhD, provided comments to an earlier draft of this manuscript. Lisa Trebino, MS, Ashley Wong, MS, and Tiffany S. Woodworth, MPH, from the Harvard Pilgrim Health Care Institute provided programming support; Roberta Constantine, RN, MBA, PhD, and Kimberly Lane, MPH, contributed managerial support; and Elizabeth Cavagnaro, BA, provided administrative support. Our FDA colleagues who helped develop this rapid response capability and this specific rapid assessment included Judy Racoosin, MD, MPH, Rachel Sherman, MD, MPH, Janet Woodcock, MD, Gerald Dal Pan, MD, MHS, Marsha Reichman, PhD, Judy Staffa, PhD, RPh, Celia Winchell, MD, Azadeh Shoaibi, MS, MHS, Carlos Bell, MPH, and Mitra Rocca, MS.