We evaluated 19 trials conducted in 39,958 children and adolescents (ages 7 to 19 years) that examined the effects of primary care relevant tobacco use interventions on smoking initiation and/or cessation. Seven of these trials examined an intervention's effects on overall smoking prevalence, 10 trials reported an intervention's effects on smoking initiation among nonsmokers, and nine trials examined smoking cessation among smokers. Two of the nine cessation trials included the adjunctive use of bupropion to help smokers quit smoking. All of the studies varied widely in terms of methodological quality, sample size, and the types of interventions tested. None of the included trials assessed health outcomes (beyond tobacco use) in children and adolescents or examined subsequent rates of adult smoking. While we sought to include interventions that addressed all the forms of tobacco use, this body of evidence primarily included studies focused specifically on cigarette smoking.

Effects of Tobacco Use Interventions

A summary of evidence for benefits and harms of all interventions is presented in Table 11. Meta-analyses showed that behavior-based interventions reduced smoking initiation among nonsmoking youth, but failed to show a statistically significant effect of smoking cessation among children and adolescents who already smoked. The studies included were generally of fair methodological quality, with various threats to internal validity. While no factors were clearly related to effect size in the included trials, high variability in the interventions' approaches may have masked important relationships.

Table 11. Summary of Evidence for Benefits and Harms of Tobacco Use Interventions.

Table 11

Summary of Evidence for Benefits and Harms of Tobacco Use Interventions.

Meta-analysis was not statistically significant among the combined prevention and cessation trials. The absolute prevalence of smoking among 12- to 18-year-olds at 7- to 12-months followup ranged from approximately 6 percent to almost 48 percent, where the absolute difference between intervention and control groups was generally modest (i.e., 1% to 7% difference). While a few studies93,95,97 showed negative effects (i.e., where the intervention group smoked more than the control group at followup), none of these differences reached statistical significance. Longer-term effects (i.e., at 2 years) generally mirrored the results seen at 12 months. The effect of these combined trials appears to be largely influenced by preventing smoking initiation among nonsmokers rather than inducing current smokers to quit, though some cessation-specific trials show promise.

Prevention Interventions

Our review of 10 trials that examined the effectiveness of interventions aimed at preventing smoking uptake among nonsmoking children and adolescents found a statistically significant pooled intervention effect at around 1 year (range, 6 to 36 months followup); the percent of nonsmoking children and adolescents initiating smoking ranged from 2 percent to nearly 20 percent, with an average absolute difference between the treatment groups of 3 percent, in favor of the intervention (range, 8 percentage points in favor of the intervention group to 3 percentage points in favor of the control group). The variability in effects appears to be driven, in part, by how the trials determined smoking status. In the trial by Bauman and colleagues,100 for example, approximately 19 and 28.5 percent of the nonsmoking 12- to 14-year-olds in this sample were considered to have initiated smoking at 7- and 16-months followup, respectively (with nonsignificant differences between groups). In this study, youth were classified as smokers if they reported ever smoking “even a puff at followup (Table 6). On the other hand, in the trial by Curry and co-authors,11 only 2.4 and 2.3 percent of 10- to 12-year-olds in the intervention and control groups, respectively, were considered to have initiated smoking at 20-months followup (again, a nonsignificant difference between groups). However, in this trial, youth were considered to be smokers if they had smoked in the past 30 days at followup (i.e., current smoking). It may be that the measure of “ever smoking” (i.e., lifetime use) may be overly sensitive and offer little prognostic value in distinguishing “true” smokers.

A meta-analysis combining nine of the 10 prevention trials found a statistically significant pooled relative risk reduction of the intervention of 19 percent, or an NNT of 50. While two trials found more youth in the intervention group starting to smoke at 12- to 20-months followup,11,90 these individual results were not statistically significant.

Our findings are generally consistent with previous reviews examining the effects of provider-, family-, community-, and school-based prevention interventions.38,43,73,105 To date, most of the results have been mixed and the reported effects are relatively small. Overall, there is limited research examining the effects of primary care relevant interventions on the risk of smoking initiation among children and adolescents. Our review found a 19 percent reduced risk of starting to smoke among intervention participants versus control participants around 1-year followup. There is little evidence demonstrating the long-term effectiveness of such interventions.

Cessation Interventions

Our review failed to find statistically significant effects of either behavior-based or behavior-plus-medication-based smoking cessation interventions among child and adolescent smokers. A pooled meta-analysis of seven behavior-based trials found the interventions had no effect when compared with controls at 6- to 12-months followup (RR, 0.96 [95% CI, 0.90 to 1.02]). Absolute quit rates ranged from 7 percent to over 40 percent in the intervention groups and from 3 to 37.5 percent in the control groups in the behavior-based trials (Table 9). The largest difference between the intervention and control groups was seen in the fair-quality trial by Colby, which found that 23.5 percent of daily smokers in the intervention group versus 2.9 percent of daily smokers in the control group reported 7-day abstinence at 6-months followup, although this difference was reduced and deemed nonsignificant using biochemical confirmation (9% vs. 2% in the intervention and control groups, respectively).87 While two trials79,95 found more youth in the intervention groups still smoked at 12-months followup, the differences were not statistically significant.

Previous reviews on tobacco cessation interventions for children and adolescents39,60 have generally found more positive effects of interventions than we found in this review. These reviews included cessation trials conducted in a variety of settings, including complex school- and community-based programs. In our review, smoking cessation rates in both intervention and control groups were generally higher than previous reports (Table 9). The review by Sussman and colleagues,60 for example, found average quit rates of 9 percent among intervention participants and 6 percent among control participants.

The lack of effect seen across the cessation trials may reflect the limited number of studies that targeted regular, established smokers or presented stratified data to examine the effects among these youth. Hollis and colleagues, for example, found strong cessation effects at 1 and 2 years for self-identified “smokers,” but no effect on self-described “experimenters” at baseline.91 It was beyond our resources to request unpublished results that may have stratified participants by the quantity or frequency with which they smoked. In several of the included studies, an adolescent would have been considered a smoker at baseline if they reported smoking only one cigarette in the 30 days prior to the intervention. After the intervention, that same adolescent may again have reported that they smoked one cigarette in the past 30 days. The cessation effect of the intervention on that adolescent, compared with a similar adolescent in the control group, would have been null. However, if you asked those adolescents (as was done in the Hollis study) if they considered themselves to be “smokers” at baseline, those who said “no” might have impacted a cessation analysis based on their self-identified smoking status. Smoking acquisition is complex, and complicates the interpretation of cessation trials in youth. Given the opportunity, these youth may have described themselves as someone who tries smoking now and again or someone who only smokes in social situations. As such, the participants may have felt the messages of “quitting smoking” did not apply to them, because they do not feel that they are true smokers. The intervention strategies and messages for these so-called experimenters and the measures for capturing any change may have to be much more sensitive to detect true cessation. A logical next step would be to replicate the few studies that have targeted established smokers87 or those that tailor their interventions according to youths' stages of acquisition and/or cessation and stratify study results as such.91 For instance, although the recent good-quality trial by Pbert and colleagues80 included smokers if they smoked at least once during the previous 30 days, youth in this study were smoking an average of nearly seven cigarettes a day (slightly lower than the average of 10 per day that adolescents in the Colby trial were smoking87). Overall, this study found no effect among all youth. Examining the effects of this trial according to the amount that youth smoked, however, may have led to different findings.

Our review included only two studies that explored the adjunctive use of medication to assist smokers in quitting. One study tested bupropion as an adjunct to NRT and one evaluated bupropion alone at two different dosages—the standard adult dose of 300 mg or a single daily dose of 150 mg. This evidence suggested that bupropion alone (in addition to a behavior-based intervention) was not effective in getting youth smokers to quit smoking at 6-months followup, although medication compliance was generally low. Both trials included relatively intense behavior-based interventions for both the intervention and control groups. In most cases, these behavioral interventions were more intense and of greater duration than many behavior-only interventions.

NRT is another treatment approach for which we did not find any eligible studies, despite reports that approximately 17 percent of pediatricians have prescribed NRT to their adolescent patients.68 We reviewed one article that examined the effects of the use of a nicotine patch on adolescent smokers.106 This study, however, did not meet our inclusion criteria because outcomes were reported at less than 6 months. In this study, 100 adolescents ages 13 to 19 years who smoked at least 10 cigarettes per day for at least 6 months and were motivated to quit smoking were randomized to receive an active nicotine patch or a placebo patch. Both groups received 10–15 minutes of individual cognitive behavioral counseling during each visit over the course of 10 weeks of the intervention. Both 7-day and 30-day point prevalence rates revealed no statistically significant differences between treatment groups at 10-weeks followup.

Similarly, we reviewed two articles that examined the effects of acupuncture or acupressure for the treatment of smoking cessation among adolescents that did not meet our eligibility criteria because of the short followup time (i.e., less than 6 months).107,108 Neither study found a statistically significant effect of acupuncture on smoking cessation among adolescents at 4-weeks and 3-months followup. Among adults, there is no consistent evidence that acupuncture is more effective than sham acupuncture on smoking cessation in the short- (less than 6 months) or long-term (6- to 12-months followup).109

Effectiveness of Specific Prevention and Cessation Intervention Strategies

The interventions included in this review were very heterogeneous in their focus (e.g., prevention, cessation, or both), intensity, primary mode of contact (e.g., face-to-face, print, telephone), level of family involvement, and time spent interacting with a health care provider. Of those interventions that included interaction with a health care provider, the least provider-intensive strategy consisted of brief advice (i.e., 30–60 seconds) during a routine office visit. Several trials included one to six booster sessions or telephone calls with other study staff or trained health counselors within the 6 to 12 months following the intervention period. Only three studies included provider advice during subsequent health care visits, although the extent to which this followup actually occurred in practice was minimal. One study showed a dose-response relationship between the amount of smoking advice from orthodontic staff (through the use of written “prescriptions”) and the percent of youth initiating smoking. Youth who received four or more advice prescriptions over 2 years were more likely to remain smokefree than youth who received zero to three messages. However, more open or compliant youths may have been the ones to receive prolonged advice. The three nonU.S. studies were all of very minimal intensity: one study included one brief advice message from dental providers, while the other two consisted of a series of mailed print materials to participants' homes over 9 weeks to 12 months.

We did not find a clear association between including parents or families in the intervention and the interventions' effects on preventing smoking initiation or cessation. We did not include literature that examined the effects of primary care interventions designed to decrease tobacco use among parents as a secondary strategy for reducing youth tobacco use or exposure to environmental (secondhand) tobacco smoke. However, parental smoking can have a significant impact on youth smoking initiation;30,110,111 children and adolescents who are exposed to smokers in their household are three times more likely to initiate smoking themselves.29 Therefore, encouraging and assisting parents to quit smoking may be another important strategy to preventing adolescent smoking.

Primary care well-child visits and other ongoing pediatric care may provide an ideal setting to intervene directly with parents, particularly given that, for many parents, their encounters with primary care may be limited to the visits they make with their children. For parents without their own primary care provider, their child's doctor may be the only access they have for ongoing smoking cessation counseling, including medication advice and prescriptions. A recent review by Rosen and colleagues112 included 18 trials that focused on parental smoking cessation that took place in hospitals, pediatric clinical settings, well-baby clinics, and homes. Quit rates averaged 23.1 percent in the intervention group and 18.4 percent in the control group, resulting in a 4 percent absolute difference between parental quit rates in the intervention and control groups. A good example of this approach is a study conducted by Curry and colleagues113 that randomized 303 low-income women to a smoking-cessation intervention or usual care as they accompanied their children to a pediatric clinic visit. During the clinic visit, women received a motivational message from the child's clinician (usually lasting 1–5 minutes), a guide to smoking cessation, a 10-minute motivational interview with a nurse or study interventionist, and up to three outreach counseling telephone calls during the 3 months following the visit. At the 12-month followup, 7-day abstinence rates were 13.5 percent among the intervention group compared with 6.9 percent in the control group. This resulted in a statistically significant adjusted OR of 2.77 (95% CI, 1.24 to 6.60). While outcomes related to children's uptake of smoking or quit attempts are not included in these evaluations, it is plausible that establishing abstinence among parents could have measurable impacts on the rate at which youth experiment with and transition to regular smoking.

Harms of Prevention and Cessation Interventions

There were no explicit harms reported in any of the behavior-based trials of prevention or cessation. Some trials reported higher absolute prevalence of smoking in the intervention than the control groups after completing the intervention, but none were statistically significant. Possible harms related to the use of bupropion include increased risk of high blood pressure, increased heart rate, nausea, throat symptoms, sleep disturbance, headache, and cough. The extent to which participants experienced these side effects appears to be mixed in the literature.

Assessment of Youth Tobacco Use

Distinguishing between children and adolescents who are “potential” or susceptible smokers, experimenters, and regular or established users is often difficult. The continuum of smoking acquisition consists of several stages: 1) not open to smoking; 2) open to smoking, when youth think about smoking but do not engage in any smoking behavior; 3) experimentation, which may include trying a puff of a cigarette or inconsistent, yet repeated smoking; 4) nondaily smoking, when youth smoke only in certain situations, such as at parties or with certain friends; and 5) established smoking, when youth smoke every day or almost every day.10 However, this behavioral acquisition sequence may not closely mimic the development of nicotine addiction.13, 114 Recent work emphasizes that 50 percent of youth who ever try smoking eventually become addicted and that smoking frequency is correlated with, but not predictive of addictive symptoms.13 Nonetheless, most research has used smoking behaviors to categorize adolescent smokers rather than addiction, so we focus on it here.

Youth who are at different stages along the behavior continuum may be at different levels of risk for becoming established smokers and thus, require different intervention approaches. Low- and moderate-risk children might include youth that have never smoked and are not open to smoking (low risk) or are open to smoking in the future (moderate risk). High-risk youth might experiment with smoking or smoke occasionally, often depending on the social context. Understanding the different stages along the progression to regular, established smoking and subsequent levels of nicotine dependence is critical in identifying at-risk youth and tailoring intervention messages.

As seen in this review, there are several different definitions of a smoker used in this body of research, and how studies operationalized these definitions varied greatly. Such variation makes it difficult to make concrete comparisons and to generalize the results. Smoking status at a given moment in time depends on the complex interaction of previous experiments, starts, and quits. From a clinical and public health standpoint, the measure of lifetime or “ever” smoking, even a single puff, may not be a meaningful endpoint and may never lead to regular use. Future research should consider using measures that reflect more regular use (e.g., smoking in the past 30 days), or the frequency or quantity smoked.

In terms of prevention interventions, it is unclear if measures of “ever” smoking only one or two puffs is a meaningful measure of true smoking “initiation,” as opposed to experimentation or a trial behavior. Including measures of self-reported susceptibility and/or stage of acquisition may help further delineate the various stages that many youth, particularly younger children, are in.12, 115 Identifying children and adolescents who are at greatest risk for smoking may help clinicians target them for more intensive prevention. In addition, there is likely a need for ongoing assessment to ensure that any counseling intervention is not merely deferring smoking initiation, but rather, strengthening or establishing a solid resolve not to experiment with or start regular smoking. Again, this may be particularly true for younger children. Youth who initiate smoking when they are younger (e.g., age 12 years) are more likely to go on to be daily smokers in later adolescence than those who initiate or experiment with smoking during older adolescence.116

One of the critical issues for smoking cessation research in children and adolescents is how baseline smoking and subsequent quitting are defined and verified. In our review, how each study defined a smoker at baseline varied from “regular or occasional” use, at least 1 day or one cigarette in the previous 30 days, at least one cigarette a week for the past 30 days, daily use for the past 30 days, to currently smoking 10 or more cigarettes a day and had done so for 6 or more months (for the medication trials). Our review was more inclusive than at least one previous review on smoking cessation in terms of the criteria used for defining baseline smokers. In the Cochrane review by Grimshaw and colleagues,39 for example, a regular smoker was defined as a young person who smokes an average of at least one cigarette per week, and had done so for at least 6 months. Their review excluded cessation trials that targeted young people who did not meet this smoking threshold. Our review included cessation studies or cessation outcomes that involved youth smokers, no matter how that was defined, which is a similar approach to that taken by Sussman and colleagues in their most recent review.60,61

In the 2006 Sussman review, among the 48 studies included, the average level of baseline smoking (the sum of the averages across studies divided by 48) was 10.44 cigarettes per day, with a range of 0.11 to 18.44 cigarettes per day. In our review, of the six behavior-based cessation trials, four79,80,87,99 presented the frequency (e.g., mean number of days or percent smoking daily) or quantity (e.g., mean number of cigarettes) smoked. For instance, adolescent smokers in the cessation trial by Colby and colleagues87 reported smoking an average of 10.5 cigarettes a day and on 6.6 days of the week. In the recent cessation trial by Pbert and co-authors,80 adolescents were included if they reported any smoking in the past 30 days. While this inclusion criteria may have allowed adolescents who are not yet regular, established smokers to participate in the study, on average, youth in this study were smoking almost seven cigarettes a day and on nearly 27 of the past 30 days. Two of the combined trials79,99 also reported frequency, quantity, and/or recency of smoking. In these studies, nearly all past-30-day smokers smoked daily99 and almost half smoked the day of the survey.79 In the trial by Hollis and colleagues,91 76.2 percent of adolescents who had smoked in the past 30 days considered themselves “smokers” rather than “experimenters.” As stated earlier, a large, positive effect of the intervention was found at 1 and 2 years among those self-described smokers and not among those who considered themselves experimenters. These examples demonstrate that while our criteria for including baseline smokers may have captured youth who are not generally established smokers, several of the studies also examined the quantity or frequency with which youth smoked and demonstrated relatively regular smoking among youth. As stated earlier, examining the intervention effect among youth who smoke at various levels is an obvious next step in this area of research.

In our review, studies used various definitions of smoking cessation. The most common outcome measure used by studies in this review was 30-day point prevalence, which is the recommended measure of cessation for youth trials.90 None of our studies used a measure of continuous cessation from the point of intervention.20 Point prevalence abstinence was used in the majority of studies and ranged from cessation for 7 to 30 days at followup. Because youth often engage in smoking patterns that are highly variable on a day-to-day basis, standard adult measures of abstinence, such as 7-day point prevalence, may not discriminate true quitters from temporary abstainers, which would inflate the true smokefree rate. One trial included the use of a bogus pipeline to increase the validity of youths' self-reports and five studies (the five studies exclusively designed as cessation trials) included biochemical measures (e.g., expired CO and saliva cotinine) to verify youths' self-reports or to analyze as secondary measures of quitting.

Applicability

Of the 19 trials included in this review, five of the interventions were conducted in a U.S. primary care setting79,80,91,98 or dental practice.81,95 The majority of these studies included relatively brief face-to-face interaction with a health care provider, such as a 30- to 60-second advice message to encourage adolescents to quit smoking or not to start smoking. In addition, the brief advice from clinicians was supported with subsequent face-to-face or telephone counseling sessions with other trained study staff and print materials and/or the use of an interactive computer program. One intervention included a provider-delivered component based on the 5A model.79 The intervention incorporated a patient-centered approach in which the providers asked about smoking, advised cessation or continued abstinence, and referred the patient to a peer counselor to develop a personalized strategy for cessation or maintained abstinence. This study found relatively large, statistically significant intervention effects on overall smoking prevalence, smoking initiation, and smoking cessation at 6 months; however, none of these effects were statistically significant at 1 year. The most provider-intensive of the primary care interventions98 showed no significant effects on smoking or smokeless tobacco use among children (average age, 11 years) at 12-, 24-, or 36-months followup.

Only one trial conducted in primary care91 found a statistically significant intervention effect on the overall prevalence of smoking, initiation among nonsmokers, and cessation among smokers at 12 months. In this sample, 77.2 percent of youth in the intervention group were smokefree at 12 months versus 72.8 percent of youth in the control group. This effect remained significant at 2 years (72.8% of the intervention group vs. 68.6% of the control group were smokefree). Among baseline nonsmokers, only 9.2 percent of the intervention group compared with 12.1 percent of the control group initiated smoking at 1 year (although, this effect attenuated at 2 years). Among adolescents who had smoked one or more cigarettes during the previous 30 days at baseline, the intervention produced significant effects at both the 1- and 2-year assessments. The intervention, however, had no effect on the small subgroup of self-described baseline experimenters.

The other included studies that found statistically significant effects on overall smoking prevalence, initiation, and cessation that were not conducted in a primary care setting may still be applicable to primary care, as they primarily included the use of mailed print materials to participants' homes. Neither of the trials that included the use of bupropion recruited participants from or took place in a health care setting.

Importantly, although a number of interventions included face-to-face interaction with a health care provider, treatment participants were only moderately more likely than control subjects to report that their clinician discussed tobacco during the visit in several cases.11,79,91,98 In the trial by Hollis and colleagues,91 for example, 41 percent of the intervention participants reported that their clinician talked with them about tobacco use versus 28 percent of the control participants. Similarly, Pbert and colleagues (2008) found that participants in the intervention group reported that their provider only spent approximately a minute and a half more discussing smoking than participants in the usual care condition (4.3 vs. 2.9 minutes).79 In the trial by Stevens and colleagues, the rates of discussion about alcohol and tobacco reported by youth were not significantly different between the intervention and attention control groups after 1 year.98 Such modest differences could reflect the poor ability children and adolescents may have to recall what was discussed by their provider, the extent to which providers discuss tobacco and other related substance use issues as part of usual care, or the salience of the specific messages discussed in the intervention conditions.

Limitations in the Body of Evidence

Most of the studies reviewed included a number of threats to internal validity, including inadequate or unclear randomization procedures, uncertain or no allocation concealment and blinding of outcome assessors, and relatively high attrition. In addition, several studies did not report baseline values for all youth randomized or by treatment group to allow us to evaluate baseline comparability. Participation in the interventions and compliance varied, with some trials demonstrating very low adherence. For example, in an intervention consisting of four mailed booklets followed by counseling phone calls, only 61.8 percent of parents completed all four sessions.50 In another study held in dental clinics, only 70 percent of adolescents actually received the planned face-to-face counseling (primarily because of missed appointments) and only one third of those participants received a planned followup phone call.54

There were inconsistent definitions and measurement of baseline smoking status, prevalence, initiation, and abstinence. In addition, there was limited use of biochemical validation of self-reported smoking status. As demonstrated in one trial,87 several of those self-reported “abstinent” smokers were reclassified as smokers based on their biomarker data. Analysis of the biomarker data showed a nonsignificant effect, underscoring the need for more research on the use of biochemical measures among children and adolescents. However, it was not clear from this trial if these results reflected only those participants who completed biochemical verification or if it also included those lost to followup and those for whom biochemical data were not obtained (who were subsequently recoded as smokers at followup).

Very few of the included studies evaluated other forms of tobacco use beyond cigarette smoking. However, other tobacco products, including smokeless tobacco and newer products such as bidis, kreteks, or use of a hookah (i.e., waterpipe) are highly available in the U.S. market. These other tobacco products are increasingly being promoted as cigarette alternatives, with claims of being potentially less harmful. While this report aimed to examine interventions to prevent tobacco use in general, the majority of the included trials focused on cigarette smoking.

We were unable to include two studies85,98 in our meta-analyses due to the limited data presented. The study by Stevens and colleagues98 was highly applicable to primary care, as it took place in 12 pediatric clinics serving a diverse population. The intervention took place over 3 years and included materials and messages for both children (including both baseline smokers and nonsmokers) and parents. After adjustment for important characteristics, the authors found a nonsignificant effect of the intervention on ever smoking and ever using smokeless tobacco at 12, 24, and 36 months. In the trial by Ausems conducted in the Netherlands, 156 elementary schools were randomized into one of four conditions: 1) in-school (curriculum-based), 2) out-of-school (three tailored letters mailed to participants' homes), 3) in-school and out-of-school, or 4) control.85 We only included the out-of-school condition (vs. control). Among the baseline “never smokers,” the authors reported that 10.4 percent (95% CI, 6.8 to 14.0) of children in the out-of-school condition versus 18.1 percent (95% CI, 12.5 to 23. 7) of the control condition participants initiated smoking at 6-months followup.

With the exception of three trials (two by the same author),79,80,104 all of the included studies were published in 2007 or earlier. In recent years, there has been a substantial emphasis placed on tobacco-related legislation, environmental changes, and countermarketing. While these public health efforts are imperative in reducing tobacco use,45 continuing to reach children and adolescents on a more personal level through behavior-based interventions remains an important strategy.8 In addition, recently there has been a considerable discrepancy between funding for research on tobacco use and funding for research on the etiology, prevention, and treatment of obesity. Funding estimates from the National Institutes of Health during the previous 5 years (2007–2012) show that funding for obesity research was or is estimated to be nearly two to three times that for tobacco-related research.117 In 2012, over $800 million is expected to go toward obesity research, whereas only approximately $350 million will go toward tobacco research. Although the prevalence of youth tobacco use has experienced a stalled decline during this time period, nearly 4,000 children and adolescents initiate and experiment with tobacco products each day. As such, interventions designed to reduce the number of young children experimenting with and regularly using tobacco products must remain a priority.

Limitations in Our Approach

One limitation in our approach is that we combined studies that used different measures of smoking prevalence. That is, our meta-analyses combined studies that defined smoking status according to youths' lifetime use or current use, as defined by the last 30 or 90 days. This variability in outcomes (often described as clinical diversity) can lead to heterogeneity if the intervention effect was affected by the way in which the outcome was measured. However, we performed sensitivity analyses to evaluate this hypothesis (i.e., removing studies with inconsistent measures) and the results remained stable.

Among the combined trials and those focused on prevention, another potential limitation to our approach was in combining interventions that exclusively focused on cigarette smoking with those that targeted multiple behaviors (e.g., alcohol and other substance use, sexual behaviors, and other problem behaviors). These unrelated aims may have caused “noise” that masked the basic message to prevent smoking and may have led to null effects. In fact, two of the trials97,98 that included broader aims than reducing smoking saw the largest negative effect of the intervention on total smoking prevalence and in reducing smoking initiation. Because of the variability in intervention approaches and populations, as well as inconsistencies in measurement, meta-analysis results should be interpreted with caution.

As stated previously, we did not include interventions that were designed to decrease tobacco use among parents as a secondary strategy for reducing smoking or secondhand tobacco smoke exposure among youth. Similarly, we did not include interventions designed to restrict smoking in homes or cars as a strategy to reduce youths' exposure to or use of tobacco. However, research has shown that having a strict smokefree policy in the home is associated with fewer smoking youth than in households with unrestricted or partial policies (i.e., for only certain members of the household).29,118 More primary research is needed that includes a focus on parental smoking and smokefree policies to understand the effect they might have on youth tobacco use.

We did not identify any prevention or cessation trials that met our inclusion criteria that assessed health outcomes in children and adolescents or examined subsequent rates of adult smoking (KQ 1). Our review only included interventions that were conducted within a health care or comparable setting. However, trials in other settings (e.g., schools), and particularly those that span several years, have shown positive effects on regular smoking in young adulthood.119

Future Research

We have several recommendations for future research in tobacco use prevention and cessation among children and adolescents. In general, there are a small number of methodologically rigorous trials that examined the effectiveness of primary care relevant behavior-based interventions to prevent tobacco use and/or to help tobacco users quit. There are even fewer good-quality trials that evaluate the use of medication to aid adolescents in their cessation efforts. Unfortunately, there are also very few trials in press or in progress that may address these gaps in the literature (Appendix H). The need to replicate promising interventions and specific intervention components in well-controlled trials is significant. This research would include incorporating longer-term outcomes to examine the extent to which results hold over time, involving more diverse samples of children and adolescents, including those at various stages of risk, estimating intervention effects in real-world settings, and determining their feasibility and sustainability in a health care setting. While 30- to 60-second brief advice messages or counseling using the 5A model may be feasible in primary care settings, it is not clear whether the additional components that many of the trials included (e.g., in-person counseling following the provider encounter, tailored computer programming, and booster telephone calls and mailed print materials) could be easily replicated in a real-world setting unless other resources (e.g., centralized phone counselors) were employed. Similarly, understanding the important components of these interventions is also necessary, including determining whether specific behavioral theories or models produce more favorable outcomes and the extent to which the addition of family-focused or parent-delivered intervention components (including an emphasis on parental cessation and policies on smokefree homes and cars) might affect outcomes. Including comparative effectiveness trials of different behavioral- and medication-based interventions may also help define essential elements of effective interventions.

One intervention strategy that may hold promise, particularly for smoking cessation, is the use of tailored computer-based programs and other electronic media channels.55,120,121 This strategy has been a key part of effective prevention and cessation interventions among both youth91 and adults.122 In these interventions, interactive programs are used to deliver highly tailored message about remaining abstinent or quitting according to the individual's risk, needs, and preferences (e.g., stage of acquisition or cessation, level of nicotine dependence, and self-identified barriers to remaining abstinent or quitting). If offered on a Web-based platform, clinicians could refer their patients to the program and then use their face-to-face time to check in and see what the youth had learned and/or applied and reinforce important messages.

In addition, there is a need for more studies that involve diverse samples of children and adolescents, including those of various racial/ethnic and socioeconomic backgrounds and at various stages of initiation and/or readiness to quit. Most of the studies in this review included fairly homogeneous samples, which limited our ability to determine whether the effects of the interventions varied by population subgroup. Disparities in tobacco use among children and adolescents in the United States exist along racial/ethnic, gender, and socioeconomic lines. Thus evaluating the effectiveness of interventions among different population subgroups should also be of high priority. As previously stated, there is also a need to recruit and/or stratify samples based on where participants fall on the behavioral continuum (i.e., susceptible, tried smoking, daily smoking). As shown, the few cessation interventions that found positive effects were among adolescents considered (either by the amount smoked or self-descriptors) to be more “established” smokers. However, given the large number of children and adolescents who have experimented with, yet not become regular smokers, there is a need to evaluate interventions and messages designed to reach this group.

We need more research examining the reliability and validity of self-reported measures and specific forms of biochemical verification among children and adolescents. Additionally, these measures should be standardized across intervention research. Future research should also consider including additional measures to evaluate the use of other forms of tobacco use beyond cigarette smoking to see the full effects on tobacco use and to make sure there are no substitution effects (e.g., quitting one type of tobacco, but starting another).

Finally, to facilitate systematic reviews and meta-analyses of both prevention and cessation studies, methodological and intervention details need to be reported as comprehensively as possible. The Youth Tobacco Cessation Collaborative evidence review panel emphasized the importance of reporting the following components in any published youth tobacco cessation study: 1) theoretical framework; 2) content and components; 3) intensity and duration; 4) site(s); 5) timing (e.g., time of day and year); 6) implementation (including intervention fidelity); 7) provider characteristics; 8) design; 9) inclusion/exclusion criteria; 10) sample size; 11) followup; 12) outcome measures; and 13) confirmation of self-report.54 These recommendations are applicable to both general tobacco reduction programs and prevention efforts.

Conclusions

Despite the substantial resources committed to reducing childhood and adolescent tobacco use over recent decades, approximately 10 percent of middle school students and nearly a quarter of high school students currently use tobacco in the United States. Consequently, child and adolescent tobacco users are a group at risk for the negative health outcomes associated with tobacco use, including becoming regular users as adults. Our findings suggest that primary care relevant interventions designed to reduce cigarette smoking among children and adolescents can have small, positive effects on smoking initiation among children and adolescents who have not yet become regular smokers. The evidence on the effectiveness of cessation interventions for youth who have experimented with cigarettes or are regular smokers is limited. Health care settings provide an opportunity to reach children and adolescents who are at risk of initiating tobacco use as well as those who have already begun experimenting with, or are regular users of, tobacco products. Ongoing policy and social changes associated with tobacco use will likely increase the pressure on youths to quit, in addition to health care clinicians providing counseling to remain abstinent and help them quit. Primary care interventions are an essential part of a comprehensive tobacco control program that complements broader school-based, community-based, media, and policy interventions.8,44