I have a broad question (Does varenicline increase the risk of adverse cardiovascular events?) and a focused question (Is the Peto odds ratio the appropriate statistic for a specific meta-analysis? Would the results be different using a Mantel-Haenszel approach?)

Varenicline (Chantix) is a medication used to assist smoking cessation. 

In Circulation 2010 Jan 19;121(2):221 a randomized trial with 714 patients with stable cardiovascular disease found varenicline to increase continuous abstinence rates and did not find a statistically significant increase in adverse cardiovascular events. The proportion of participants with an adjudicated cardiovascular event was 7.1% in the varenicline group and 5.7% in the placebo group (difference, 1.4; 95% CI, −2.3 to 5.0) in an analysis of 703 patients. Events reviewed included nonfatal or fatal MI, hospital admission for chest pain, hospitalization for angina pectoris, need for coronary revascularization, resuscitated cardiac arrest, hospitalization for congestive heart failure, fatal or nonfatal stroke or transient ischemic attack, new diagnosis of or admission for a procedure to treat peripheral vascular disease, and death resulting from any cause

On June 16, 2011 the FDA issued a warning (label change) because of the increase in cardiovascular events and required the manufacturer to conduct a meta-analysis of randomized trials. This warning was based on 3 specific adjudicated cardiovascular events in this trial (nonfatal MI in 2% vs. 0.9%, need for coronary revascularization in 2.3% vs. 0.9%) and new diagnosis or procedure for peripheral vascular disease in 1.4% vs. 0.9%) without statistical significance, but recognizing inadequate statistical power to analyze these endpoints. (http://www.fda.gov/Drugs/DrugSafety/ucm259161.htm)

On July 4, 2011 an early release publication in CMAJ (http://www.cmaj.ca/content/early/2011/07/04/cmaj.110218.full.pdf+html) reports a systematic review and meta-analysis of serious adverse cardiovascular events in randomized trials of varenicline.    The data in this meta-analysis included:

a.       The Circulation 2010 trial with cardiovascular events in 25/355 varenicline vs. 20/359 placebo patients (using ITT approach) with Peto odds ratio 1.28 (95% CI 0.7-2.34)

b.      One trial with statistically significant increase with 4/603 vs. 0/607 events (Peto odds ratio 7.48, 95% CI 1.05-53.2)

c.       9 trials with non-significant results tipping toward increased risk (often with 1-2 events vs. 0 events)

d.      2 trial with equal event rates in both groups (1 or 2 events per group)

e.      1 trial with non-significant result tipped toward decreased risk (0/214 vs. 1/218) – Peto odds ratio 0.14, 95% CI 0-6.95)

From the Cochrane Handbook regarding Peto odds ratio method:

The approximation used in the computation of the log odds ratio works well when intervention effects are small (odds ratios are close to one), events are not particularly common and the studies have similar numbers in experimental and control groups. In other situations it has been shown to give biased answers. As these criteria are not always fulfilled, Peto’s method is not recommended as a default approach for meta-analysis.

Are the assumptions for the Peto odds ratio met in this meta-analysis?

Intervention effects are small when considering absolute effects (because event rates were low in all but 1 trial) but do odds ratios of 0.14 and 7.44 fit this assumption (odds ratios close to one)?

8 of 14 trials had similar numbers in experimental and control groups, but 6 trials had substantially more patients in experimental than control group.

Data never “fit perfectly” and Peto odds ratio may very well be the most appropriate approach given the low event rates in most trials.  I’m just asking hoping someone on the list who does this type of meta-analysis can comment.   If Peto odds ratio were not used, would other approaches give statistically significant results with this data (found in Figure 2)?

Brian S. Alper, MD, MSPH
Editor-in-Chief, DynaMed
Medical Director, EBSCO Publishing
[log in to unmask]
10 Estes Street
Ipswich, MA  01938-0231
Phone 800-356-6500 or 978-356-6500 ext. 2749
Cell 978-804-8719
Fax 978-356-7026
www.ebscohost.com/dynamed

 "It only takes a pebble to start an avalanche."