(See: Zawertailo L, Dragonetti R, Bondy SJ, Victor JC, Selby P. Reach and effectiveness of mailed nicotine replacement therapy for smokers: 6-month outcomes in a naturalistic exploratory study. Tobacco Control 2013;22:e4.)
The study compares six-month quit rates between two groups of smokers: (1) callers to an Ontario quitline which provided eligible smokers with free NRT for five months, in addition to telephonic counseling and referral services; and (2) a similar sample of participants in the Ontario Tobacco Survey who did not receive free NRT, matched by smoking behavior and motivation to quit.
The results are summarized as follows:
"Of those with complete follow-up data, the percentage reporting abstinence after 6 months in the treatment cohort was 21.4%, relative to 11.6% in the no-intervention cohort (rate ratio of 1.84; 95% CI 1.79 to 1.89), with the 30-day point prevalence of 17.8% and 9.8% for the intervention and no-intervention cohorts, respectively (rate ratio 1.81; CI 1.75 to 1.87)."
The paper concludes as follows:
"Provision of free NRT by mail following a brief telephone intervention is an effective strategy to reach and assist a large number of smokers making a quit attempt. ... This was a high-reach cost-effective public health measure and as such should be an integral part of comprehensive tobacco control programmes."
The Rest of the Story
This paper is a dream-come-true for the pharmaceutical industry. In it, the authors recommend that all comprehensive tobacco control programs purchase five months worth of NRT for all quitline callers. The financial implications of this advice are immense. If this advice were followed, NRT would become the main feature of smoking cessation programs throughout the world and sales of NRT would skyrocket.
There is nothing wrong with this, assuming that the study conclusion was valid and that the study did indeed demonstrate that the provision of NRT resulted in a six-month quit rate of 18%, nearly doubling the baseline quit rate of about 10% for similarly motivated smokers who do not receive free NRT.
The rest of the story, however, is that the study conclusion is not valid, the six-month quit rate was not 18%, and the NRT intervention was not demonstrated to be effective.
Because the authors reported (and based their conclusions on) the wrong data.
The study reports the quit rate among the smokers who were successfully followed up for six months. But we know from previous literature that smokers who are lost to follow-up are much more likely to be non-quitters. Thus, the proper number to report is that based on an "intent-to-treat" analysis, where subjects who are lost to follow-up are assumed to be non-quitters. In this study, the loss to follow-up was substantial (58%). Thus, failure to use an intent-to-treat analysis heavily biases the results.
Even in the heavily-biased 2008 Clinical Practice Guideline update, intent-to-treat analyses were exclusively relied upon in evaluating the effectiveness of NRT. The intent-to-treat analysis is also used in a new clinical trial which will be released later today that evaluates the effectiveness of electronic cigarettes. In fact, intent-to-treat analyses are the basis for all comparisons of the effectiveness of different smoking cessation medications and approaches.
So what are the actual results of this study, under an intent-to-treat analysis?
It turns out that under an intent-to-treat analysis, the 30-day point prevalence of quitting at six months in the free NRT group was only 9%, not 18%. This compares to a quit rate of just below 9% in the comparison group, resulting in a relative risk of quitting of 1.03 (not statistically significant) associated with the intervention. Thus, the truth is that this study fails to provide evidence that the provision of free NRT increased the smoking cessation rate. And it certainly fails to provide evidence that provision of free NRT will yield a cessation rate of 18% at six months.
The bottom line: in an intent-to-treat analysis (i.e., the correct analysis), this study documents only a 9% quit rate at six months for free NRT and finds that this intervention is not effective.
So why, then, do the authors choose to report the wrong results in the abstract and to base their conclusion on results from only 42% of the sample? Why do they choose to veer away from the standard reporting mechanism in hundreds of NRT clinical trials and to rely on a measure that is invalid on its face?
Unfortunately, the presence of significant financial conflicts of interest with Big Pharma create the appearance that the authors are being influenced (albeit subconsciously, perhaps) by these conflicts, which may have introduced investigator bias. In other words, the conflict of interest has the appearance of having biased the interpretation and reporting of the study findings.
What are the conflicts of interest? As reported in the paper (but apparently, not available unless you purchase the article):
- "L.Z. has received speaker honoraria from Pfizer Canada Inc. ... ."
- "P.S. has received speaker's honoraria from Schering Canada, Johnson & Johnson Consumer Health Care Canada, Pfizer Canada Inc., Pfizer Inc. Sanofi-Synthelabo Canada, GlaxoSmithKline Canada Inc., Genpharm Canada, Prempharm Canada, NABI Pharmaceuticals, and consulting fees from Schering Canada, Johnson & Johnson Consumer Health Care Canada, Pfizer Canada Inc., Pfizer Inc., Sanofi-Synthelabo Canada, GlaxoSmithKline Canada Inc., Genpharm Canada, Prempharm Canada, NABI Pharmaceuticals, ... and AstraZeneca Canada Inc."
- "R.D. has received speaker honoraia from Pfizer Inc."
(Also see John Polito's response to the article. The response was cleverly entitled "Free NRT by mail an intent to treat defeat.)