| Key question | Intervention | Number of included studies and participants* | Summary of findings | Consistency and precision | Other limitations | Strength of evidence† | Applicability |
|---|---|---|---|---|---|---|---|
| 1: Health outcomes | Combined pharm and behavioral | 0 | NA | NA | NA | Insufficient | NA |
| Pharm | 0 | NA | NA | NA | Insufficient | NA | |
| Behavioral | 1 review (1 RCT, n=1445) | One trial found favorable effects on all-cause and coronary disease mortality and lung cancer incidence and mortality 20 years following an intensive behavioral intervention, although results were not statistically significant. | NA | Only one review reported the results of one intervention in men. Within that trial, the rate of smoking among control group participants declined steadily over the followup period, narrowing the intervention effect. | Low evidence of potential benefit | One trial conducted among civil servant men aged 40–59 years in the UK with high risk of cardiorespiratory disease. Intervention took place in the 1970’s. | |
| Electronic cigarettes | 0 RCTs | NA | NA | NA | Insufficient | NA | |
| 2: Cessation outcomes | Combined pharm and behavioral | 1 review (53 RCTs, n=25,375) | Combined pharmacotherapy and behavioral interventions increased smoking quit rates by 68–98% compared with no or minimal treatment (RR 1.83 [95% CI, 1.68 to 1.98]) at 6 months or more followup. | Reasonably consistent Reasonably precise | May be risk of bias due to lack of blinding of participants. | High evidence of benefit‡ | Treatment effects appear to be comparable in a range of populations, settings and types of interventions and in smokers with and without other co-morbidities. The literature almost exclusively addressed treatment for cigarette smoking, as opposed to the use of other forms of tobacco, so results may not be generalizable to all forms of tobacco. |
| Pharm | 5 reviews (336 RCTs, n>159,000) | NRT, bupropion, and varenicline significantly increased the chances of quitting smoking compared with placebo or no medication. Reviews suggested that NRT might increase smoking abstinence at 6 months or longer by 49–61% (RR 1.55 [95% CI, 1.49 to 1.61]); bupropion by 49–76% (RR 1.62 [95% CI, 1.49 to 1.76]); and varenicline by 106–143% (RR 2.24 [95% CI, 2.06 to 2.43]). Absolute quit differences averaged 6.4% for NRT; 8.2% for bupropion, and 14.5% for varenicline. Using a combination of NRT products increased quitting more than the use of a single NRT product (RR 1.25 [95% CI, 1.15 to 1.36]). Direct comparisons between drugs suggested that varenicline may be superior to NRT and bupropion in achieving smoking abstinence at 6 months or longer.‖ | Reasonably consistent Reasonably precise | Possibility of publication bias but unlikely that the presence of additional studies with lower relative risks would alter the findings given large number of studies and consistency in findings for each type of drug. | High evidence of benefit§ | Treatment effects appear to be comparable in a range of populations, settings and types of interventions and in smokers with and without other co-morbidities. The literature almost exclusively addressed treatment for cigarette smoking, as opposed to the use of other forms of tobacco, so results may not be generalizable to all forms of tobacco. | |
| Behavioral | 20 reviews (830 RCTs, n>500,000) | Health provider advice and counseling, individual counseling, group-based interventions, telephone counseling, mobile phone-based interventions, tailored and interactive internet-based interventions, and incentives showed significant increased smoking cessation at 6 or more months relative to controls (15% to 88%). For example, for physician advice versus minimal controls or usual care: RR 1.76 (95% CI, 1.58 to 1.96). Providing more intense adjunctive behavioral support to smokers receiving pharmacotherapy may increase cessation by 8–22% (RR 1.15 [95% CI, 1.08 to 1.22]). Evidence on the use of motivational interviewing, decision aids, print-based, nontailored self-help materials, real-time video counseling, biomedical risk assessment, exercise, complementary and alternative therapies, and system-level interventions was limited and not definitive in the effects on cessation. | Reasonably consistent Reasonably precise | Individual trials may be represented in more than one review and/or meta-analysis. Indication of possible publication bias for evidence related to motivational interviewing and acupuncture. Fixed-effects models were used in nearly all meta-analyses. | Moderate to High evidence of benefit¶ | Treatment effects appear to be comparable in a range of populations, settings and types of interventions and in smokers with and without other co-morbidities. The literature almost exclusively addressed treatment for cigarette smoking, as opposed to the use of other forms of tobacco, so results may not be generalizable to all forms of tobacco. | |
| Relapse prevention | 1 review (77 RCTs, n=67,285) | Analyses of behavioral interventions among abstainers did not detect an effect in both studies of assisted abstainers (RR 0.99 [95% CI, 0.87 to 1.13]; I2=56%; k=10; n=5408) and unaided abstainers (RR 1.06 [95% CI, 0.96 to 1.16]; I2=1%; k=5; n=3561) from the general population. There was some evidence that extending varenicline could be beneficial in preventing relapse, but it was only reported by two studies. NRT was found to help in unassisted abstainers, but no difference was seen among those who achieved abstinence with NRT. None of the six studies that examined the use of bupropion to prevent relapse found a statistically significant effect. | Inconsistent Imprecise | Highly variable study designs and included interventions. | Moderate evidence of no benefit of behavioral interventions on relapse prevention Moderate evidence of benefit of varenicline on relapse prevention; low evidence of no benefit of bupropion or NRT on relapse prevention | Studies were highly heterogenous and may not be applicable to the general adult population. | |
| Electronic cigarettes | 5 RCTs (n=3117) | Two trials (n=2008) found statistically significantly greater rates of smoking abstinence in those using e-cigarettes containing nicotine (with or without the co-use of NRT) compared with NRT alone or NRT plus non-nicotine e-cigarettes at 6- to 12-months followup, although continued use of e-cigs remained high after the treatment phase. Another trial (n=300) found a borderline statistically significant higher quit rate among those receiving nicotine-containing e-cigs (11%) vs no nicotine e-cigs (4%) at 12 months (p=0.04), but 27 percent of those who quit smoking continued to use e-cigarettes at 1 year. The remaining two trials found no statistically significant difference in biochemically verified abstinence at 6 months between those receiving e-cigs vs nicotine patch or placebo e-cig (n=807). | Inconsistent Imprecise | Limited statistical power to detect differences and differential loss to followup in all five trials (22–50%). Wide variance of nicotine concentrations in e-cig interventions (7.8mg vs. 18mg) | Insufficient | All five trials took place outside of the US, Korea, in New Zealand, Italy, and the UK. Two trials used older models of e-cigs of which is no longer available. One trial conducted among smokers not wanting to quit. | |
| 3: Harms | Combined pharm and behavioral | 0 reviews# | NA | NA | NA | Moderate evidence of no harms | NA |
| Pharm | 15 reviews# | NRT, bupropion, and varenicline were not associated with an increased risk in major CV or neuropsychiatric adverse events. NRT was associated with a higher rate of any CV adverse events largely driven by low-risk events, typically bradycardia and arrhythmia. There was no evidence of a difference in harms associated with medications for those with versus without severe mental illness. | Reasonably consistent Reasonably precise | Many trials that report cessation effectiveness do no report AEs, particularly CV- or neuropsychiatric-specific AEs. AEs typically measured through passive reporting and therefore susceptible to underreporting. | Moderate evidence of no serious harms | Findings appear applicable across populations and settings, including among patients with severe mental illness. | |
| Behavioral | 3 reviews# | There was no evidence that behavioral tobacco cessation interventions are associated with serious adverse events. | NA | Very few reviews assessed AEs related to behavioral interventions. | Moderate evidence of no harms** | Limited evidence on harms limits applicability. | |
| Electronic cigarettes | 9 RCTs (n=3942) | No trials reported serious AEs in either the intervention or control groups related to product use and no significant differences in the frequency of AEs among study groups. Coughing, nausea, throat irritation and sleep disruption were the most commonly reported side effects of e-cig use. | Reasonably consistent Imprecise | Limited statistical power to detect differences and differential loss to followup in all three trials (22–39%). One study did not report methods for AEs reporting. | Insufficient | The two US trials had the lowest enrollment (<100 participants). Two trials used older models of e-cigs, one of which is no longer available, one trial used an e-cig that is not available in US markets, and one trial used a prototype. |
Number of included studies reflects the number of systematic reviews designated as primary evidence for that body of evidence as well as the summed total number of included studies and observations from each review.
For our review-of-reviews method, we adopted the strength of the overall body of evidence assigned within the primary systematic review. In most cases, these grades were based on the Grading of Recommendations Assessment, Development and Evaluation (GRADE) working group definitions which consider study limitations, consistency of effect, imprecision, indirectness and publication bias. Where strength of evidence grades were not available, we adapted the EPC approach to assign an overall strength of evidence grade based on consensus discussions involving at least two reviewers.78
Some evidence of asymmetry in a funnel plot; excess of small trials detecting larger effects. However, in a sensitivity analysis, removing smaller studies did not markedly decrease the pooled estimate.
Sensitivity analysis including only those studies judged to be a low risk of bis did not impact the pooled results for any comparison; for NRT and bupropion, the funnel plots showed some evidence of asymmetry. However, given the large number of trials in these reviews, this does not suggest the results would be altered significantly were smaller studies with lower RRs included.
Evidence from existing systematic reviews as well as the EAGLES trial indicate that adult smokers randomized to varenicline have a statistically significant higher likelihood of quitting smoking at 6 months compared with those randomized to NRT or bupropion. In the EAGLES trial (n=8144) 21.8% of smokers randomized to varenicline quit smoking at 6 months compared with 15.7% randomized to NRT (OR 1.52 [95% 1.29 to 1.78]) and 16.2% randomized to bupropion (OR 1.45 [95% CI, 1.24 to 1.70]).161
Quality of the evidence differs for each specific type of intervention, but generally reflects moderate to high certainty grades. Most common reasons for downgraded the quality of evidence were unexplained statistical heterogeneity, several studies with high or unclear risk of bias, or inconsistency in the evidence base.
Total number of studies and observations not estimated
Despite the relatively limited number of reviews that reported harms related to interventions, we are moderately confident that there are no serious harms related to combined pharmacotherapy and behavioral counseling interventions or behavioral counseling alone for tobacco cessation.
Abbreviations: AE = adverse event; CI = confidence interval; CV = cardiovascular; e-cig = electronic cigarette; mg = milligrams; NA = not applicable; NRT = nicotine replacement therapy; pharm = pharmacotherapy; RCT = randomized controlled trial; RR = risk ratio; UK = United Kingdom; US = United States
From: Chapter 4, Discussion
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.