Efficacy of pharmacotherapies for short-term smoking abstinance: A systematic review and meta-analysis

Background Smoking cessation has important immediate health benefits. The comparative short-term effectiveness of smoking cessation interventions is not well known. We aimed to determine the relative effectiveness of nicotine replacement therapy (NRT), bupropion and varenicline at 4 weeks post-target quit date. Methods We searched 10 electronic medical databases (inception to October 2008). We selected randomized clinical trials [RCTs] evaluating interventions for our primary outcome of abstinence from smoking at at-least 4 weeks post-target quit date, with biochemical confirmation. We conducted random-effects odds ratio (OR) meta-analysis and meta-regression. We compared treatment effects across interventions using head-to-head trials and calculated indirect comparisons. Results We combined a total of 101 trials evaluating delivery of NRT versus inert controls at approximately 4 weeks post-target quit date (total n = 31,321). The pooled overall OR is OR 2.05 (95% Confidence Interval [CI], 1.89-2.23, P =< 0.0001). We pooled data from 31 bupropion trials contributing a total n of 11,118 participants and found a pooled OR of 2.25 (95% CI, 1.94-2.62, P =< 0.0001). We evaluated 9 varenicline trials compared to placebo. Our pooled estimate for cessation at 4 weeks post-target quit date found a pooled OR of 3.16 (95% CI, 2.55-3.91, P =< 0.0001). Two trials evaluated head to head comparisons of varenicline and bupropion and found a pooled estimate of OR 1.86 (95% CI, 1.49-2.33, P =< 0.0001 at 4 weeks post-target quit date. Indirect comparisons were: NRT and bupropion, OR, 1.09, 95% CI, 0.93-1.31, P = 0.28; varenicline and NRT, OR 1.56, 95% CI, 1.23-1.96, P = 0.0002; and, varenicline and bupropion, OR 1.40, 95% CI, 1.08-1.85, P = 0.01. Conclusion Pharmacotherapeutic interventions are effective for increasing smoking abstinence rates in the short-term.


Introduction
Smoking remains the leading cause of preventable death in the world. [1] Smoking cessation is associated with important benefits at the individual and societal levels. Given the prevalence of smoking, considerable efforts have been directed toward developing interventions to assist smokers in quitting. However, smoking cessation interventions have had heterogeneous successes. [2] Smoking cessation is necessary to reduce future morbidity and mortality, however many patients have difficulty discontinuing.
Both psychosocial and pharmaceutical interventions have been evaluated for their success in achieving smoking discontinuation. [3,4] Drug therapies are now licensed in North America and Europe to promote smoking cessation. The most commonly evaluated of these has been nicotine replacement therapy [NRT]. [5,6] More recently, attention has focused on the use of anti-depressant therapy and specifically the agent bupropion [7]. A new intervention approved in 2006, varenicline, targets nicotine receptors to reduce craving and pleasure sensations. Recent guidelines and evaluations call for combining therapies to provide optimal patient management. [3,8] We,. [9] and others,. [10][11][12][13] have previously reported on the efficacy of these interventions for longer-term cessation (3-12 months) durations. No systematic review has yet evaluated short-term quit rates from available therapies. Guidelines for smoking cessation programmes consider quitting 4-weeks post-planned quit date as a successful short-term cessation.
[14] Short-term smoking abstinence is especially important in patients requiring immediate behaviour changes, such as those with recent cardiovascular events. [15] or undergoing surgery. [16] We conducted a meta-analysis of Randomized Clinical Trials [RCTs] to identify the effectiveness of the various pharmacological interventions in improving abstinence rates at 4weeks and 6 months.

Eligibility Criteria
Our primary outcome of interest was smoking abstinence at approximately 4 weeks post-target quit date (TQD). Our secondary outcomes were short-term smoking abstinence defined as 6 months after initiating treatment or closest available data to that time point, within one month. We included any RCT of NRT of any delivery method, bupropion or varenicline. We included only RCTs of at least 4 weeks duration with biochemical confirmation of smoking abstinence because of the likelihood of abstinence over-reporting. While methods of assessing smoking abstinence vary from study to study, the most common method is self-report. However, this can have false cessation rates as high as 30%. [17]False reporting is most likely to occur in a trial setting or in assessing smoking status after a medical event. Laboratory tests are often used to verify smoking status, especially in clinical trials. Methods of biological verification include serum and saliva thiocyanate (SCN), expired carbon monoxide (CO), plasma, saliva and urinary cotinine and plasma and urinary nicotine. Each of these have various strengths and weaknesses. [18] Studies had to report smoking abstinence as either sustained abstinence at the time periods or pointprevalence of abstinence. When both outcomes were available, we considered sustained abstinence to be a superior clinical marker of abstinence. We excluded dose ranging studies, non-RCTs, post-hoc analyses, maintenance therapy, and studies that reported outcomes as selfreport.

Study endpoints
Our primary endpoint was the 4-week post-TQD. This is variably reported in studies over years of publications. National committees require data on the 4-week post-TQD and each group of trials of intervention deals with this endpoint differently. Newer studies typically report this as the last 4-weeks of treatment as pharmacotherapy is begun prior to TQD. Where this specific endpoint is reported, we extracted data on 4-week post-TQD. Where not reported, we extracted data on 4 weeks post-intervention. Our secondary endpoint, 6-months post intervention is typically reported as 6 months post-treatment, but may also be reported as 6 months post TQD. Where reported specifically, we extracted data on 6-month post-TQD.
[27] Searches were not limited by language, sex or age.

Study selection
Two investigators (EM, PW) working independently, in duplicate, scanned all abstracts and obtained the full text reports of records, that indicated or suggested that the study was a RCT evaluating a smoking abstinence therapy on the outcomes of interest. After obtaining full reports of the candidate trials (either in full peer-reviewed publica-tion or press article) the same reviewers independently assessed eligibility from full text papers.

Data collection
Two reviewers (PW, EM) conducted data extraction independently using a standardized pre-piloted form. Reviewers collected information about the smoking intervention tested, the population studied (age, sex, underlying conditions), treatment dosages and dosing schedules, the treatment effect at 4 weeks post-TQD and at 6 months postintervention, the specific measurement of abstinence (sustained or point-prevalence), and the chemical confirmation methods. Study evaluation included general methodological quality features including allocation concealment, sequence generation, blinding status, intentionto-treat, and appropriate descriptions of loss to follow-up. We entered the data into an electronic database such that duplicate entries existed for each study; when the two entries did not match, we resolved differences through discussion and consensus.

Data analysis
In order to assess inter-rater reliability on inclusion of articles, we calculated the Phi statistic, which provides a measure of inter-observer agreement independent of chance.
[28] We calculated the Odds Ratios [OR] and appropriate 95% Confidence Intervals [CIs] of outcomes according to the number of events of abstinence reported in the original studies or sub-studies. Odds Ratios are the preferred effect measure in smoking cessation trials. In circumstances of zero outcome events in one arm of a trial, we added 1 to each arm, as suggested by Sheehe.
[29] We first pooled studies of all NRT interventions versus all controls using the DerSimonian-Laird random effects method,.
[30] which recognizes and anchors studies as a sample of all potential studies, and incorporates an additional between-study component to the estimate of variability.
[31] We calculated the I 2 statistic for each analysis as a measure of the proportion of the overall variation that is attributable to between-study heterogeneity.
[32] Forest plots are displayed for each primary analysis, showing individual study effect measures with 95% CIs, and the overall DerSimmonian-Laird pooled estimate. We then conducted a meta-regression analysis on the NRT studies with predictors of heterogeneity including the following covariates: placebo control; reporting of sequence generation; reporting of allocation concealment; use of gum or patch; and, method of chemical confirmation of abstinence. We additionally conducted separate pooled analyses of NRT versus placebo, gum versus control and patch versus control. We conducted all analyses at 4 weeks and also at 6 months post-TQD. For bupropion trials, we pooled all bupropion trials versus all controls and conducted a meta-regression analysis using the following cov-ariates: placebo control; reporting of sequence generation; reporting of allocation concealment; method of chemical confirmation of abstinence; and plans to quit. We conducted separate meta-regression analyses and calculated the relevant ORs for the covariates as the exponent of the coefficient.
[33] We additionally pooled all placebo-controlled trials and evaluated effect sizes at 4 weeks and at 6 months post-TQD. For head-to-head trials of bupropion versus NRT, we conducted pooled random-effects analyses at 4 weeks and at 6 months post-TQD. For varenicline trials, we conducted pooled random-effects analyses of varenicline versus placebo and for head-to-head trials of varenicline versus bupropion or NRT at 4 weeks year and at 6 months. post-TQD. Head-to-head trials provide the strongest inferences regarding intervention superiority.
[34] However, with so few head-to-head trials of varenicline versus NRT, we conducted indirect comparisons of these interventions versus placebo using methods described by Bucher et al. [35] This method maintains the randomization from each trial and compares the summary estimates of pooled interventions with CIs. Analyses were conducted using StatsDirect (version 2.5.2, http:// www.statsdirect.com) and Comprehensive Meta-analysis (version 2, http://www.meta-analysis.com).

Study inclusion
We identified 795 abstracts from our extensive searches. We excluded 532 as irrelevant to meeting our inclusion criteria. We obtained 263 full-text studies for screening. We further excluded 94 studies for reasons explained in figure 1 [See Additional File 1]. In total, we included data from 168 RCTs. Agreement was near perfect (φ => 0.9).  o 474 abstract screened for inclusion after searching with "nicotine" AND "smoking" AND "gum OR Patch OR spray OR inhalers OR Tablet OR lozenge" AND "random*" o 280 abstacts were obtained when using "bupropion" and "smoking" and "random" and "clinical trial" o 41 abstracts were obtained when using "varenicline" and "random" and "clinical trial"

Discussion
This study confirms the short-term effectiveness of all three smoking interventions compared to placebo. Our findings stand in line with outcomes evaluated over a longer period, up to one year, of these same interventions. [9,10] This finding should be of interest to clinicians, policy-makers and patients. As interventions to assist in smoking cessation are increasingly available, the combination of these interventions, along with sociobehavioural interventions, should be a research priority. [8] The definition of smoking abstinence and relapse are variable across studies. The most common time periods of  Random effects meta-analysis of sustained smoking abstinence with bupropion versus controls at 4-weeks post-TQD Figure 5 Random effects meta-analysis of sustained smoking abstinence with bupropion versus controls at 4-weeks post-TQD. For example, individuals with cardiovascular events can immediately benefit from smoking discontinuation because of improvements in several physiological variables including reduced myocardial oxygen demand, improved myocardial oxygen supply, reduced activation of the sympathetic system, reduced risk of arrhythmias and reduced acute thrombosis risk. These benefits could be particularly critical in the peri-event period when patients are at increased risk of complications or repeat events. Thus even if relapse occurs at a later stage, abstinence around the time of an event could prove beneficial.
When we previously evaluated varenicline to NRT and bupropion, we had data from only 4 trials. [9] This evaluation found that the addition of 7 trials continues to demonstrate elevated varenicline effects compared to NRT and bupropion. Further community effectiveness interventions will be required to ensure generalizability.
There are several strengths and limitations to consider when interpreting our analysis. Strengths of this review include the comprehensive search strategy that improved the likelihood of identifying all relevant studies. Duplicate extraction of data reduced the potential for bias in this component of the synthesis process. By limiting this review to randomized trials we ensured that the included studies would have reduced likelihood of systematic error and therefore have high internal validity. Our use of metaregression to identify sources of heterogeneity in the meta-analyses is a strength and demonstrated that several of the a priori chosen covariates were predictors of heterogeneity. To reduce patient-reporting bias, we included only studies that chemically confirmed the cessation of smoking at the specific time-points-this has been a weakness in previous reviews. [23] Limitations of this meta-analysis include the potential for publication bias, in particular the possibility that small negative studies would not be published. Publication bias on short-term effects is likely due to both author-initiated bias and journal-initiated bias against short-term evaluations. We included only published trials so it is possible that other trials have been conducted and never published. However, it is unlikely that the presence of these studies would have altered the findings of our analysis given the large number of studies included and the consistency with the longer-term evaluations (both 6 months and one year). [9,10] We limited our search to English language databases (although we would include non-English articles if identified) so the possibility of quality studies in other languages does exist. We used both direct and indirect comparisons to evaluate the relative effectiveness of agents. Head-to-head trials provide the strongest inferences regarding intervention superiority.
[34] In the presence of existing head-to-head trials of varenicline versus NRT.

Conclusion
In conclusion, our review demonstrates clear efficacy of smoking cessation pharmacotherapies in the short term and provides similar estimates of efficacy as longer term evaluations. [9,10] Given the benefits of smoking abstinence in both primary and secondary prevention of major morbidities, the use of these therapies in patients with active smoking related disease warrants further study. [15] Future research to evaluate the efficacy and safety of these interventions in combination and in patients with advanced diseases is warranted.

Funding
This study received unrestricted funding from Pfizer Ltd to evaluate anti-smoking agents. They had no role in the conduct, interpretation or writing of this manuscript.