Researchers Reanalyze Data from Previous Study Which Showed Cardiovascular Adverse Effects of Chantix and Find No Adverse Effect

A study published last week in BMJ presented a re-analysis of data from a previous meta-analysis of the potential adverse effects of the smoking cessation drug Chantix. The original study reported that the odds of serious cardiovascular events were 1.7 times higher among smokers treated with Chantix compared to those receiving placebo, using data from from 14 double-blind randomized controlled trials involving 8216 participants.

In the new study, the authors examined data from 22 randomized clinical trials of Chantix. The major differences in the two studies were:

1. The new study included 8 clinical trials in which there were no adverse cardiovascular events observed. Those studies were excluded from the original study.

2. The new study only included adverse cardiovascular events that occurred within one month of patients discontinuing Chantix. The original study included all observed cardiovascular effects in the clinical trials.

The authors failed to find any significant effect of Chantix on cardiovascular event risk. They conclude: "Our meta-analysis of treatment emergent, cardiovascular serious adverse events, with attention to bias and critical design issues, indicates that the risk of these events associated with varenicline use is small, and statistically and clinically insignificant."

The Rest of the Story

The authors of this study argue that the exclusion of trials in which there no adverse events leads to a bias away from the null hypothesis of no adverse cardiovascular events associated with Chantix. This is how they justify the inclusion of those trials. One could also argue, of course, that the inclusion of the trials with no observed adverse effects biases the results towards the null hypothesis (when one is using the absolute risk difference as the primary outcome measure). With no adverse effects observed, the risk difference has to be zero, because there can be no difference in absolute risk when the risk in both groups is zero.

The authors of this study also argue that the inclusion of adverse events occurring after the study drug was discontinued was inappropriate because the drug is no longer present in the person's system and thus can no longer have any effect. They write: "We conducted a systematic review and meta-analysis of treatment emergent, cardiovascular serious adverse events in all published, randomised controlled trials of varenicline for tobacco cessation. We defined these events as occurring during drug treatment or within 30 days of discontinuation. We chose a 30 day window because it was biologically relevant for detecting a drug toxicity effect, while still being conservative; the half life of varenicline is 24 hours, and any direct pharmacological effect should be gone within seven days."

However, one could easily argue that the exclusion of adverse cardiovascular events occurring more than 30 days after discontinuation of Chantix therapy is inappropriate because if the drug causes inflammatory processes that accelerate atherosclerosis or facilitate plaque formation, then the risk does not disappear immediately after the drug is withdrawn. For example, if studying the adverse cardiovascular effects of smoking, one would not exclude events occurring more than one month after smoking cessation.

So in a way, the exclusion of events occurring more than 30 days after discontinuation of Chantix therapy biases the results towards the null hypothesis and this procedure seems inappropriate for a study whose outcome is adverse cardiovascular events, where risk does not decrease immediately after discontinuation of an offending agent (such as tobacco smoke, hypertension, high cholesterol, or a pro-thrombotic or pro-atherogenic drug).

The point is that these authors are presenting a biased view of the relationship between Chantix and adverse cardiovascular events. It is almost as if they were not willing to accept the previous results and were searching for a way to negate the earlier findings. It may or may not be the case that they are correct, but the bias is apparent.

There would be no problem at all with that bias under normal circumstances. Every scientist looks at a problem with a particular bias, and that is generally a good thing. It ensures rigorous scientific consideration of hypotheses and study findings and generally enhances science.

However, the problem is that these are not "normal circumstances." In this case, one of the authors of the new study is a recipient of grant funding from Pfizer, the manufacturer of Chantix, and therefore has a significant financial conflict of interest. The bias present in the study therefore has the appearance of being a result of this conflict of interest, rather than as a result of the desire to ensure the rigor of reported scientific findings.

According to the study: "JJP [the lead author] is principal investigator on ... an Investigator Initiated Research award from Pfizer...".

The conflict of interest arises from this Pfizer funding. Because she receives funding from Pfizer, there is a financial conflict with her reporting the results of a meta-analysis of clinical trials on Chantix, because if she were to report an adverse effect of Chantix, this would obviously threaten any future funding she might receive from the pharmaceutical company.

I have serious concerns about a conflicted author making national recommendations like this about the safety of a drug, especially when other research has concluded otherwise, and when the FDA itself has forced the company to include a black box warning about the adverse cardiovascular effects of the drug. I believe that while it is fine for this author to conduct the study for which she has been funded by Pfizer, she should refrain from making national recommendations supporting the drug's use and minimizing its potential adverse effects, because those recommendations are by definition the results of a conflicted analysis, which is tarnished by a significant financial conflict of interest.