Skip to main content
Download PDF

Individual-, family- and school-based interventions to prevent multiple risk behaviours relating to alcohol, tobacco and drug use in young people aged 8-25 years: a systematic review and meta-analysis

BMC Public Health volume 22, Article number: 1111 (2022) Cite this article

Abstract

Background

Engagement in multiple substance use risk behaviours such as tobacco smoking, alcohol and drug use during adolescence can result in adverse health and social outcomes. The impact of interventions that address multiple substance use risk behaviours, and the differential impact of universal versus targeted approaches, is unclear given findings from systematic reviews have been mixed. Our objective was to assess effects of interventions targeting multiple substance use behaviours in adolescents.

Methods

Eight databases were searched to October 2019. Individual and cluster randomised controlled trials were included if they addressed two or more substance use behaviours in individuals aged 8-25 years. Data were pooled in random-effects meta-analyses, reported by intervention and setting. Quality of evidence was assessed using GRADE. Heterogeneity was assessed using between-study variance, τ2 and Ι2, and the p-value of between-study heterogeneity statistic Q. Sensitivity analyses were undertaken using the highest and lowest intra-cluster correlation coefficient (ICC).

Results

Of 66 included studies, most were universal (n=52) and school-based (n=41). We found moderate quality evidence that universal school-based interventions are likely to have little or no short-term benefit (up to 12 months) in relation to alcohol use (OR 0.94, 95% CI: 0.84, 1.04), tobacco use (OR 0.98, 95% CI: 0.83, 1.15), cannabis use (OR 1.06, 95% CI: 0.86, 1.31) and other illicit drug use (OR 1.09, 95% CI: 0.85, 1.39). For targeted school-level interventions, there was low quality evidence of no or a small short-term benefit: alcohol use (OR 0.90, 95% CI: 0.74-1.09), tobacco use (OR 0.86, 95% CI: 0.66, 1.11), cannabis use (OR 0.84, 95% CI: 0.66-1.07) and other illicit drug use (OR 0.79, 95% CI 0.62-1.02). There were too few family-level (n=4), individual-level (n=2) and combination level (n=5) studies to draw confident conclusions. Sensitivity analyses of ICC did not change results.

Conclusions

There is low to moderate quality evidence that universal and targeted school-level interventions have no or a small beneficial effect for preventing substance use multiple risk behaviours in adolescents. Higher quality trials and study reporting would allow better evidence syntheses, which is needed given small benefit of universal interventions can have high public health benefit.

Trial registration

Cochrane Database of Systematic Reviews 2014, Issue 11. Art. No.: CD011374. DOI: 10.1002/14651858.CD011374.

Peer Review reports

Introduction

Risk behaviours such as tobacco smoking, antisocial behaviour, hazardous alcohol consumption, physical inactivity and unprotected sexual intercourse are commonly initiated in adolescence [1]. These risk behaviours have also been shown to co-occur in adolescence, with engagement in one behaviour increasing the likelihood of engaging in others [2]. Multiple risk behaviour (MRB) refers to the engagement in two or more behaviours directly or indirectly associated with health, well-being and healthy development of personality [3]. There is evidence that MRB is associated with negative outcomes over and above the effects of the individual risk behaviours, including increased risk of premature mortality, morbidity, poor educational attainment, obesity and mental health problems later in life [4,5,6]. Given that adolescents make up a quarter of the world’s population, youth represents a critical time in the life course to address MRB for the improvement of health and wellbeing [7].

The co-occurrence of risk behaviours in adolescence is of particular importance to substance use behaviours. For instance, tobacco smokers are more likely to consume alcohol and vice versa, with cannabis users being more likely to engage in other drugs [8,9,10]. A UK study found that adolescents who engaged in hazardous alcohol use were six times more likely to engage in tobacco and drug use [11]. Prevalence estimates of concurrent alcohol, tobacco and drug use among young people range from 4% to 17% [10, 12,13,14,15,16]. Given that engagement in these substance use behaviours may continue from adolescence into early adulthood [17,18,19] MRB related to substance use is an important area of investigation due to the high public health costs of these behaviours and burden to society [20].

Recognising the co-occurrence of substance use risk behaviours and shared antecedents [21,22,23] has led to recommendations for interventions that address multiple, rather than single, risk behaviours in an efficient approach that can provide a cost-effective means of prevention [24, 25]. However, there are few examples of reviews that examine the effectiveness of interventions targeting substance related MRB. Typically, reviews focus on interventions that target behaviours such as alcohol use [26, 27], tobacco use [28] and illicit drug use [29] in isolation. One review assessed interventions targeting MRB and found that a large proportion of identified interventions targeted substance use behaviours [30]. The authors found prevention programmes to be effective, with small effect sizes but comparative to single risk behaviour interventions [30]. A more recent review assessed combined student- and parent-based programs and found that nine programmes had promise in preventing alcohol and other drug use among adolescents [31]. There has also been an overview of systematic reviews showing small but consistent positive effects of school-based prevention programmes targeting alcohol and drug misuse in 12-18 year olds [32]. Finally, other systematic reviews have assessed MRB including substance use behaviours, but either set out to undertake a narrative synthesis or due to heterogeneity of studies were unable to conduct quantitative meta-analyses [33, 34]. Our review differs from these studies as we include a wider variety of intervention settings, include any combination of substance use behaviours and assess interventions delivered at the individual-, family- and school-level.

Objectives

The primary objective of this review was to assess the effects of interventions at the individual-, family- and school-level that target multiple substance use risk behaviours (two or more from alcohol, tobacco, cannabis and other substance use) for the primary or secondary prevention of MRB and related harms in children and young people aged 8-25 years. It is the partner review to MacArthur et al. [24], which examined interventions that targeted two or more risk behaviours with at least one being a non-substance use risk behaviour (such as risky sexual behaviour or anti-social behaviour).

Methods

Search strategy and study selection

A peer-reviewed, published protocol is available for this review [35]. This review was originally part of a Cochrane registered review addressing a wide range of risk behaviours. However, due to the identification of many studies that addressed a combination of substance use risk behaviours without addressing other risk behaviours, we decided to narrow the scope of this review. Thus, this review focuses on interventions targeting a combination of at least two or more substance use risk behaviours only. Searches were conducted in three stages, reflecting initial and update searches. Initial searches were conducted prior to the reviews being separated [24, 35]. Seventeen electronic databases were searched, from inception to May 2012, including websites and databases for ongoing or unpublished studies, without language restriction. The reference lists of relevant articles were hand searched and experts in the field contacted to identify ongoing research. Following MacArthur et al [24], database selection was streamlined to eight databases, and search terms refined to improve specificity of update searches in 2018 and 2019. Further details on the approach and search strategies for Medline are provided in Additional file 1.

The title and abstracts of the first 500 records from the May 2012 searches were independently screened by two authors, and a further 10% were randomly selected and double-screened for quality assurance. Kappa statistics showed a high level of agreement (>0.8). The remainder of the title and abstract screening was conducted by a single reviewer. Full text screening was conducted independently by two reviewers. For the 2018 and 2019 update searches, after the reviews had been separated [35, 36], two independent reviewers screened all records for inclusion, at both title and full-text screening stages. Disagreements were resolved by discussion, or by a third reviewer if necessary.

Eligibility criteria

Studies were eligible for inclusion if they were randomised controlled trials (RCTs), including clustered RCTs, aimed at addressing at least two substance use behaviours of interest, including: tobacco smoking, alcohol consumption, cannabis use and other substance use e.g. illicit drugs, legal highs, solvents, aerosols, or inhalants. Eligible participants were children and young people aged 8-25 years of age. Interventions targeting parents, carers, peers, teachers and others were eligible if the intended impact of the intervention was to prevent or reduce engagement in MRBs among children and young people aged up to 18 years (at study baseline, but up to 25 years at follow-up). Studies were included if they had a minimum follow up period of six months from the start of the intervention. Interventions could be provided universally (without regard of the young person’s level of risk) or targeted towards young people or families thought to be high risk. Interventions could be delivered in a wide variety of individual-, family-, or school-settings, and across the prenatal, antenatal, nursery, preschool, primary and secondary school ages. Delivery could include a variety of methods and individuals such as trained staff members, nurses, teachers, parents, peers and the police or via electronic equipment. Eligible comparator interventions included usual practice (as defined by the study author), no intervention or placebo.

Exclusion criteria

Interventions that addressed additional risk behaviours not classified as substance use were excluded as they were assessed in the companion Cochrane review [24]. Interventions delivered at a community- or population-level, such as media campaigns or policy, regulatory or legislative were excluded. Interventions with a follow-up period of less than six months (including the intervention period) were excluded. Interventions aimed at individuals with clinically diagnosed disorders or interventions where the majority of the adolescent participants were 18 years at baseline assessment were also excluded. Details of excluded and ongoing studies can be found in Additional file 2.

Data extraction

Data were extracted directly into a piloted form in Microsoft Excel. Details on study characteristics and primary and secondary outcome data extracted are in Table 1. Data were extracted by one reviewer and checked by a second with disputes resolved between them or by involving a third reviewer. Study authors were contacted for further information if there were missing or unclear data in the full text publications. No automation tools were used. The full list of extracted data is reported in the protocol [35].

Table 1 Data extracted on characteristics, primary and secondary outcomes
Full size table

Quality assessment

Risk of bias was assessed by two review authors using the Cochrane Risk of Bias 1.0 tool [37], with disagreements resolved through discussion and inclusion of a third reviewer if needed. Study quality was assessed by one reviewer using Grades of Recommendation, Assessment, Development and Evaluation (GRADE) criteria [38].

Data handling and preparation

Following data extraction, some studies could not be included in the synthesis because they either did not report substance use outcome data or reported a combined substance use outcome (such as a combined alcohol, tobacco and drug use score) which was not of interest for this review. A list of these studies can be found in Additional file 3 [39,40,41,42,43,44,45,46,47].

We anticipated outcome data would be inconsistently reported and summarised across studies [35]. The main data transformations and assumptions made are described here and in more detail in Additional file 4. To maximise the number of studies available for meta-analysis, we converted all outcome data to a log odds ratio scale [40, 41]. Where cluster RCTs had ignored clustering in their analyses, or where it was not clear, we inflated the SE to account for intra-cluster correlations and to avoid a unit-of-analysis error [37] (See Additional file 3). Where the number of participants in each study arm was unclear, but where total number of participants was provided and attrition was comparable between each arm, we assumed equal group sizes. In multi-arm trials, intervention groups were combined where they were deemed similar enough. The calculations and formulae used for combining were dependent on data type and are described in Additional file 4. Where studies did not provide sufficient information for meta-analysis, or data transformations were not possible, studies were omitted from analysis. A list of studies, with reason for omission, is provided in Additional file 3.

Substance-use outcomes

Although the interventions in this review address multiple substance use behaviours, the majority of studies report substance use outcomes separately (not as a composite MRB score). Therefore, as specified in the review protocol to allow for meta-analysis and to align with the partner review [24] we have reported the behaviour outcomes separately, and our intentions outlined in the protocol [35]. Where a study reported multiple measures per outcome, a pre-specified hierarchy was applied: [1] the ‘most harmful’ measure from a frequency and public health perspective (more recent and/or more regular use) and [2] the measure which contained more information were preferred for meta-analysis. For example, a recent, regular measure of drinking frequency (e.g. average number of drinks per week) was preferred to a longer-term binary measure of drinking (e.g. drunk alcohol in the past month). Where authors reported multiple ‘heavy use’ outcomes such as ‘binge drinking’, ‘heavy drinking’, ‘being drunk’ and ‘daily cigarette use’, the same hierarchy was applied.

After applying the above criteria, the outcome measures included are detailed in Table 2. Further detail on outcomes contributing to the meta-analyses is reported in Additional file 5.

Table 2 Included substance use outcome measures
Full size table

Data synthesis and subgroup analysis

We used the metafor package in R 4.0.2 [48], to conduct random effects meta-analyses allowing for between study variation in intervention effects (i.e. different studies could estimate different, yet related, intervention effects) [49]. Separate analyses were conducted by outcome and by short-term (closest to 12 months) and longer term (longest available) follow-up periods. Results are summarised as odds ratios (ORs) with 95% confidence intervals (CIs). Heterogeneity was evaluated using the between-study variance, τ2, and Ι2 measure, as well as the p-value of between-study heterogeneity statistic Q (based on the random effects model). We planned to investigate heterogeneity using subgroup analyses, and the full list of planned analyses is available in the protocol and Additional file 6. However, with the exception of “intervention type” (study setting and intervention focus) there were insufficient numbers of studies for meaningful subgroup analysis. Following MacArthur [24], and for ease of presentation, our results are presented on a single forest plot grouped by intervention type and setting.

Small-study effects and sensitivity analyses

Where there was a sufficient number of studies (minimum of 10), we used funnel plots to examine the study effect size against the sample size to look for publication bias or small-study effects [50]. For the study outcomes that required us to estimate an ICC, we carried out sensitivity analyses by using the highest and lowest reported ICC to calculate design effects and inflated SEs for each outcome. We ran meta-analyses for each group using these robust SEs to calculate best- and worst-case scenarios. Planned sensitivity analyses focusing on studies at low risk of bias were not conducted due to insufficient numbers (Additional file 6).

Results

Search results

From a total of 24,263 titles, we obtained 141 full text articles and included 66 studies in the systematic review (Fig. 1).

Fig. 1
figure 1

Flow diagram detailing the systematic review screening process. Shows the systematic review process included the number of studies that were retrieved at the different search points, the number of studies excluded at each stage and the final sample of intervention studies

Full size image

Study characteristics

A summary of study characteristics is provided in Additional file 7. Sixty-three studies were conducted in high-income countries (Australia n=4; Germany n=1; Netherlands n=1; Multi-country n=2; Spain n=3; Sweden n=1; Taiwan n=2; UK n=4; USA n=45) and three studies were conducted in middle-income countries (Thailand, Mexico, Brazil). Studies were either RCTs (n=13) or cluster RCTs (n=53). Most studies adopted a universal prevention approach (n=53), of which 41 were implemented in school settings, five in family settings, five in a combination of settings and two focussed on individuals. Thirteen studies were targeted at high-risk groups, for example, as determined by behaviour scores (indicated) or by ethnic or socioeconomic factors (selective). These targeted interventions were at the school level (n=8); at the family-level (n=1), at the individual level (n=1) and at a combination of the levels (n=3).

Studies that reported using a theoretical framework (n=45) varied in terms of type of theory, number of theories and combinations. The most commonly used theories were Social Influence Model (n=14), Social Learning Theory (n=11) and Health Belief Model (n=6). Most studies used a sociopsychological, cognitive or behavioural theory or model as the framework, with many taking inspiration from several models instead of strictly following one. There was also a range of activities used within the interventions. All studies used a curriculum to guide intervention activities (n=45), with reported details of the curriculum varying. Studies that reported what the curriculum involved mentioned elements such as interactive activities, workbooks, quizzes, teacher-led classes and demonstrations. Other key components (reported separately to the curriculum, but could have been embedded within it) included role play (n=33), homework (n=17), group discussion (n=27), parental involvement (n=13) and didactic teaching (n=16). Theoretical frameworks and components were spread across the different study types and levels.

Intervention duration varied across studies. The most common intervention duration was <3 months (n=28), with the remainder lasting 3-6 months (n=10), 6-12 months (n=4), 12-24 months (n=10) and more than 24 months (n=6). Few studies explicitly reported intervention length and for eight (12%) it could not be determined at all. Studies targeted either two (n=16), three (n=28) or four (n=22) substance use behaviours concurrently. Thirty five studies reported the number of randomised individuals, either in total or by intervention/control arm. Total sample size ranged from 70 to 19,529 individuals. Most studies stated the control group received no intervention or usual practice (n=49). Eleven studies stated the control group received an alternative intervention. Six studies did not describe what the control group received. Most studies were two-armed studies (intervention/control) (n=41), with the remainder either not stating the number of arms or being three and four armed studies including comparing multiple interventions.

Participant age at baseline was inconsistently reported and a variety of summaries were used. For example, mean age (with and without standard deviation), age range, proportions, approximate age ‘average’, school grade or school grade range. Participant age at baseline could not be determined in 10 studies. Of those studies reporting an age summary (n=56), we cautiously estimated that most addressed early adolescents between 11 and 14 years (n=36) with the remainder (n=20) including older adolescents (up to 18 years). No studies indicated participants were under 10 years at baseline. Most studies reported gender of participants at baseline (n=60). Four studies only included female participants, the remainder were mixed with the majority an approximate equal split of male and female, with two studies having over 60% female participants [39, 51] and two studies having 31-33% female participants [52, 53]. Thirty studies reported a baseline socioeconomic measure, with a wide variety of summaries including free school meal eligibility, parental occupational status and household income at either an individual or school level. Forty studies reported ethnicity at baseline. Two studies were targeted at one ethnic minority group [54, 55]. For the remainder of studies there were varying levels of detail around ethnicity (e.g. only reporting the majority ethnic group) and whether reported at individual or school level.

Quality of evidence

Here we report the risk of bias assessments for the domains of randomised sequence generation and allocation concealment only, with results for all domains reported in the final column of Additional file 7. Six studies were at low risk of bias for both domains, with a further seven at low risk for just randomised sequence generation. Forty-five studies were at unclear risk of bias for both randomised sequence generation and allocation concealment. The remaining studies were at high risk of bias for at least one domain. Using GRADE criteria, the quality of the evidence was rated as being of moderate or low certainty for most outcomes. This quality of evidence was primarily due to risk of performance bias and unexplained heterogeneity between studies.

Meta-analyses of substance-use outcomes

Short- and long-term effects of interventions on each substance-use outcome were meta-analysed separately. Heterogeneity across all subgroups was substantial due to differences in population targeted and intervention type (I2 ranged from 0% to 71%, for short-term effects and 97% to 47%, for long-term effects). Therefore, we analysed pre-planned subgroups according to population targeted (universal and targeted) and intervention type (individual, family, school and combination). For full transparency we also report results from fixed effect analyses and the overall pooled summary effects for both random and fixed effects models in Additional file 8.

Short-term intervention effects (one year post-intervention)

Alcohol use and binge drinking

Twenty-seven studies contributed to the analyses for alcohol use (Fig. 2). There was moderate quality evidence that universal, school-based interventions have no or a small beneficial effect in relation to short-term alcohol use (OR 0.94, [95% CI: 0.84 to 1.04; 15 studies; 31,803 participants; I2=20%). There was moderate-quality evidence, that universal, family-level interventions (OR 0.69, [95% CI: 0.54 to 0.89]; 2 studies; 1999 participants; I2=39%) and universal, individual-level interventions (OR 0.75, [95% CI: 0.59 to 0.95]; 2 studies; 1024 participants; I2=7%) were effective for preventing short-term alcohol use.

Fig. 2
figure 2

Forest plot for short term alcohol use (<12 months) by intervention level. The plot shows the data meta-analysed in intervention-level subgroups (school, family, individual, combination and universal or targeted) for the short-term (as close to 12 months post-intervention end as possible) alcohol use outcome. Red boxes represent point estimate (odds ratio) and horizontal lines the 95% confidence intervals. Black diamonds are pooled estimates for each subgroup. Estimates above 1 (on the right-hand side of the y-axis labelled ‘favours control’) suggests a lower alcohol use in the control group compared to the intervention group, while estimates below 1 (on the left-hand side of the y-axis labelled ‘favours intervention’ suggests lower alcohol use in the intervention group compared to the control group. The overall pooled estimate was not calculated as the subgroups differ by design and a decision was made a priori to report subgroups separately

Full size image

For targeted school-based interventions, there was low quality evidence of no or a small beneficial effect of the intervention in the short term (OR 0.90, [95% CI: 0.74 to 1.09]; 4 studies; 2103 participants; I2=0%). Other intervention and setting groups contained only one study each but are reported in Fig. 2 to enable a visual comparison.

There was moderate quality evidence from six studies for a short-term benefit in relation to heavy alcohol use (binge drinking) from school-based universal interventions (OR 0.70, [95% CI: 0.50 to 0.99]; 6 studies; 23,907 participants; I2=80%). There was substantial statistical heterogeneity, largely driven by two studies [56] and [57], Fig. 3. It was not possible to meta-analyse other intervention and setting groups, however individual studies are reported where possible in Fig. 3.

Fig. 3
figure 3

Forest plot for short term heavy alcohol use (<12 months) by intervention level.The plot shows the data meta-analysed in intervention-level subgroups (school, family, individual, combination and universal or targeted) for the short-term (as close to 12 months post-intervention end as possible) heavy alcohol use outcome. Red boxes represent point estimate (odds ratio) and horizontal lines the 95% confidence intervals. Black diamonds are pooled estimates for each subgroup. Estimates above 1 (on the right-hand side of the y-axis labelled ‘favours control’) suggests a lower heavy alcohol use in the control group compared to the intervention group, while estimates below 1 (on the left-hand side of the y-axis labelled ‘favours intervention’ suggests lower heavy alcohol use in the intervention group compared to the control group. The overall pooled estimate was not calculated as the subgroups differ by design and a decision was made a priori to report subgroups separately

Full size image

Tobacco use and heavy use

Twenty-four studies contributed to the analyses for short term tobacco use (Fig. 4). There was moderate quality evidence that universal school based MRB interventions are likely to have no short-term benefit in relation to tobacco use (OR 0.98, [95% CI: 0.83 to 1.15]; 13 studies; 21,762 participants; I2=0%). The quality of evidence was rated as low for universal, individual-level (OR 0.82, 95% CI: 0.55 to 1.23]; 2 studies; 1024 participants; I2=60%) and universal, combined-setting interventions (OR 0.85, [95% CI: 0.46 to 1.59]; 2 studies; 2202 participants; I2=0%), and findings are consistent with both a beneficial, null and harmful effect of the intervention in the short term.

Fig. 4
figure 4

Forest plot for short term tobacco use(<12 months) by intervention level. The plot shows the data meta-analysed in intervention-level subgroups (school, family, individual, combination and universal or targeted) for the short-term (as close to 12 months post-intervention end as possible) tobacco use outcome. Red boxes represent point estimate (odds ratio) and horizontal lines the 95% confidence intervals. Black diamonds are pooled estimates for each subgroup. Estimates above 1 (on the right-hand side of the y-axis labelled ‘favours control’) suggests a lower tobacco use in the control group compared to the intervention group, while estimates below 1 (on the left-hand side of the y-axis labelled ‘favours intervention’ suggests lower tobacco use in the intervention group compared to the control group. The overall pooled estimate was not calculated as the subgroups differ by design and a decision was made a priori to report subgroups separately

Full size image

For targeted school-level interventions, there was moderate quality evidence from four studies of no, or a small beneficial effect for short-term tobacco use (OR 0.86, [95% CI: 0.66 to 1.11]; 4 studies; 2103 participants; I2=14%). There was only one study each in universal, family-level and targeted, family-level and combination groups, so these could not be meta-analysed but are shown on the forest plot. Not enough studies were available to be able to evaluate heavy tobacco use.

Cannabis and other illicit drug use

A total of 18 studies contributed to the analyses for cannabis use (Fig. 5). There was moderate quality evidence from universal, school-based interventions, to suggest little or no short-term benefit in relation to cannabis use (OR 1.06, [95% CI: 0.86 to 1.31]; 10 studies; 18,234 participants; I2=0%). There was high heterogeneity arising from two conflicting studies of universal, individual-based interventions (OR 0.83 [95% CI: 0.36 to 1.88]; 2 studies, 1024 participants; I2=90%). For targeted, school-level interventions, there was low quality evidence to support a short-term benefit for cannabis use (OR 0.84, [95% CI: 0.66-1.07]; 4 studies, 2103 participants; I2=0%). There was only one study in the targeted, individual-based and family-based (and none in the remaining groups), so these could not be meta-analysed but are included on the forest plot for visual comparison.

Fig. 5
figure 5

Forest plot for short term cannabis use (<12 months) by intervention level.The plot shows the data meta-analysed in intervention-level subgroups (school, family, individual, combination and universal or targeted) for the short-term (as close to 12 months post-intervention end as possible) cannabis use outcome. Red boxes represent point estimate (odds ratio) and horizontal lines the 95% confidence intervals. Black diamonds are pooled estimates for each subgroup. . Estimates above 1 (on the right-hand side of the y-axis labelled ‘favours control’) suggests a lower cannabis use in the control group compared to the intervention group, while estimates below 1 (on the left-hand side of the y-axis labelled ‘favours intervention’ suggests lower cannabis use in the intervention group compared to the control group. The overall pooled estimate was not calculated as the subgroups differ by design and a decision was made a priori to report subgroups separately

Full size image

Seventeen studies were included in the forest plot for other illicit drug use (Fig. 6). There was moderate quality evidence to suggest that universal, school-based interventions had little or no short-term benefit for other illicit drug use (OR 1.09, [95% CI: 0.85 to 1.39]; 11 studies; 21,239 participants; I2=0%). There is a suggestion of a benefit of universal, individual-level interventions in relation to other illicit drug use, however, this is based on only two studies judged to provide very low certainty evidence (OR 0.68, [95% CI: 0.40 to 1.15]; 2 studies; n=1024 participants; I2=76%). Two targeted, school-level studies also provided low quality evidence of a potential short-term benefit for other illicit drugs (OR 0.79, [95% CI: 0.62 to 1.02]; 2 studies; 1615 participants; I2=0%). There was only one study in universal, combination and targeted, family-level subgroups (and no studies in the remaining subgroups) so these could not be meta-analysed but are included on the forest plot.

Fig. 6
figure 6

Forest plot for short term illicit drug use(<12 months) by intervention level.The plot shows the data meta-analysed in intervention-level subgroups (school, family, individual, combination and universal or targeted) for the short-term (as close to 12 months post-intervention end as possible) illicit drug use outcome. Red boxes represent point estimate (odds ratio) and horizontal lines the 95% confidence intervals. Black diamonds are pooled estimates for each subgroup. Estimates above 1 (on the right-hand side of the y-axis labelled ‘favours control’) suggests a lower illicit drug use in the control group compared to the intervention group, while estimates below 1 (on the left-hand side of the y-axis labelled ‘favours intervention’ suggests lower illicit use in the intervention group compared to the control group. The overall pooled estimate was not calculated as the subgroups differ by design and a decision was made a priori to report subgroups separately

Full size image

Longer-term intervention effects (longest follow-up available over 1 year)

Fewer studies included longer term follow-ups and statistical heterogeneity within groups was greater than observed for short-term outcomes. Pooled results suggest there was no benefit, on average, for universal, school-based interventions in relation to alcohol use (OR 1.01, 95% CI: 0.84 to 1.23]; n=6 studies; n=28,840 participants,; I2=65%) or tobacco use (OR 1.00, [95% CI 0.75 to 1.35]; n=5 studies, n=28,030 participants; I2=65%), and potential benefit in relation to cannabis use (OR 0.68, [95% CI 0.42 to 1.09]; studies; n=23,228 participants; I2=45%). Only two studies reported universal, school-based interventions that measured longer-term binge drinking, with suggestion of a beneficial effect (OR 0.77, [95% CI: 0.47 to 1.27]; 2 studies; 9529 participants; I2=82%). Only two studies [58, 59] investigated targeted, school-based interventions and there was an absence of evidence of effectiveness, due to wide confidence intervals, across all substance-use outcomes (see Additional file 9 for complete statistics). Other subgroups with single studies (or no studies) could not be meta-analysed but are included on the forest plots reported in Additional file 9.

Sensitivity analyses and assessment of small study effects

Sensitivity analyses exploring the impact of larger or smaller imputed ICCs did not alter the findings reported above (Additional file 8). Funnel plots are reported for universal, school-based interventions for short-term alcohol use (n=15), tobacco use (n=13) and illicit drug use (n=11) outcomes. There was no evidence to suggest possible small study or publication bias, however, we note the small number of studies included (Additional file 10).

Secondary outcomes

Secondary outcomes were sparsely and inconsistently reported, and meta-analysis was not possible. Additional file 11 contains a table showing all the relevant secondary outcomes reported for each study, as defined by study author.

Discussion

Overall, we found insufficient evidence to support the effectiveness of multi-substance use interventions to prevent substance use-specific behaviours. In most analyses, we observed the point estimates were in the direction of a small beneficial effect, but with most confidence intervals crossing null.

Universal, school-level interventions were the most common intervention type included (n=41 across all substance use outcomes). However, there was evidence of no or only very small benefit of effectiveness on substance use outcomes, whether measured at either short- or long-term follow-up. There was some evidence of benefit for universal, family- and individual-level interventions, however with only two studies contributing to the meta-analysis of each substance use outcome, and the considerable between study heterogeneity observed in the universal, individual-level analyses, caution should be exercised when drawing conclusions here.

Targeted, school-level interventions were the next most reported intervention type (n=8 across all substance use outcomes). However, there was also inconclusive evidence of an effect across each substance use outcome, at both short- and longer-term follow-ups. There were not enough individual-, family- or combination- interventions on targeted populations to be able to evaluate these groups.

This review contributes to the mixed evidence base observed from other reviews examining the impact of multi-substance use interventions. For instance, our findings support a meta-analysis evaluating universal school-based interventions (n=53), which showed mixed evidence of effectiveness in relation to alcohol and drug use outcomes [60]. An overview of 21 systematic reviews, found a small but consistent positive effect of school-based prevention programmes for alcohol and drug use [32]. Further, a systematic review of universal school-based interventions (n=13) that combined student and parent based programmes, found there was potential efficacy of these types of interventions in preventing and reducing alcohol and drug use among adolescents [31]. Finally, in our companion Cochrane review assessing a wider range of risk behaviours [24], colleagues found that universal school-based programmes targeting substance use in combination with other MRB were more beneficial than we found in the present review, in relation to tobacco use and alcohol use, and that such interventions may be effective for illicit drug use [24]. Although the direction of intervention effects in our analyses is consistent with those observed in the Cochrane review, our summary effect estimates are smaller, with confidence intervals often spanning no intervention effect. The meta-analytic approach taken is similar across both reviews, however the present review includes almost twice the number of studies and three times as many participants. An unanswered question, in either review, is whether interventions that address prevention of non-substance and substance-use behaviours concurrently have greater benefit for individual outcomes. In both reviews, MRB outcomes were analyzed separately, ignoring the correlation that may be present between outcomes within studies and at a population level [61]. Future work, exploiting advances in multivariate meta-analysis, might address whether interventions which prevent one MRB (e.g. smoking) are effective for preventing another MRB (e.g. anti-social behaviour).

Notably, none of these reviews included all studies by intervention level and population, with the exception of the companion Cochrane review [24]. Our review contributes to the evidence base here as, in public health terms, there are unique assets and challenges to universal and targeted approaches [62]. We know from the Rose hypothesis [63] that while targeted approaches may have greater individual benefit to those in the high risk group, as ‘risk’ is normally distributed, universal approaches are likely to be more effective on a population level. Further, Rose’s prevention paradox [63] means these universal approaches are likely to show little individual benefit. Adopting the prevention paradox means that despite the effect being small for universal interventions, this may still be the preferable health prevention approach. Having conducted separate meta-analyses for intervention and setting means we can begin to assess what types of intervention are likely to be effective for adolescent substance use behaviours. The present review contributes to an emerging evidence base that universal interventions hold the greatest promise for prevention of MRBs [24, 64]. However, the effect sizes observed across all analyses in our review were small and, on balance, between-study heterogeneity was moderate, especially for the longer term timepoints. As such, we were unable to pinpoint specific intervention features that might be of interest. Using advanced synthesis methods [65], further work to formally evaluate the MRB intervention characteristics associated with larger effect sizes is needed. For example, a component network meta-analysis approach could be used to understand which intervention elements could be driving the effects of universal interventions [66, 67].

Most of the studies included in the review did not report their findings by characteristics known to be associated with equity or wider determinants of health. Using the studies included in the present review and the companion Cochrane review, we undertook a subgroup analysis to investigate whether intervention effects differed by socioeconomic status of participants [68]. However, there were too few studies that measured the appropriate demographic characteristics and therefore the study lacked power. There is a need for routine reporting of equity data so that stronger evidence of effect by SES and race/ethnicity can be demonstrated and that interventions can be evaluated for their impact on health inequalities [68, 69]. Moreover, given most studies were conducted in high income countries, predominantly the USA, there is a need for further research in different cultural, geographic and socioeconomic contexts. For instance, there is a distinct lack of studies conducted in the Middle East, Africa and South America, which is the case for both adolescent and adult populations [70]. This is of particular relevance to adolescent substance use behaviours, given that the majority of the world’s young people aged 10-24 years live in low- and middle-income countries, and there are high rates of tobacco-, drug- and alcohol-use within these regions [71,72,73,74]. It is critical we consider the global context related to adolescent health and wellbeing to develop and assess context-appropriate multiple risk behaviour interventions in areas with the greatest need.

Strengths and Limitations

This systematic review is among the first to quantitatively assess interventions targeting multiple substance use behaviours in young people. This review involved a comprehensive search and review process, across an extensive number of databases. We included universal and targeted populations and a wide range of interventions, delivered at individual-, family-, school- and combination-levels, yet analysed separately by population and setting to reflect the heterogeneity in design.

The findings are limited by the large proportion of studies that were classified as unclear or high risk of bias due to the lack of blinding and lack of clarity surrounding allocation concealment or reporting. Within some of our intervention subgroups (family targeted, and universal and targeted individual), there were too few studies with useable data and therefore findings are restricted. We were also unable to meta-analyse the results by other potentially meaningful distinctions aside from intervention level. For example, it was often unclear whether studies were adopting a purely prevention approach (to young people uninitiated to any substance use behaviours), a harm reduction approach (to those already engaging in substance use behaviours), or both. With greater consistency of reporting across individual studies, future reviews will be able to interrogate the results more comprehensively and compare different approaches.

Our review defined young people within a wide age range of 8-25 years in an attempt to capture the adolescent period, with some allowance for interventions that started earlier but ran into adolescence and studies that followed up participants into their mid-twenties. This broad age range presents a challenge in interpreting our data as young, middle and older adolescents have been viewed as occupying different developmental and social life stages [75]. Older and younger adolescents will have very different experiences with substance use behaviours as well as differing levels of independence, peer influence and access to adult privileges (e.g. driving, legal alcohol consumption) [75, 76], suggesting they may respond to interventions differently. Although the majority of included studies focused on adolescents aged 11-14 years, our meta-analyses include some older and younger adolescents. With more detailed reporting of demographic characteristics in the individual studies, in future work we would be able to separate the studies into more focused age groups.

The majority of included studies had a cluster randomisation design, however a large proportion did not transparently report adjustment for clustering in their analyses. Not accounting for correlation within clusters can lead to inflated and over-precise estimates of effect. Our retrospective estimation of ICC and use of robust standard errors for meta-analyses is likely to have provided more conservative precision estimates, which in turn will have contributed to wider confidence intervals. We note however, that the sensitivity analyses did not change our conclusions. Our most recent search was undertaken in October 2019, so there may be relevant studies missing from our analysis. The COVID-19 pandemic has affected capacity to update the searches.

Between-study heterogeneity was substantial in some analyses, but surprisingly low in others. The challenges we experienced relating to heterogeneity are common when reviewing public health interventions of this nature, which highlights the debate surrounding whether RCTs and systematic reviews are the most appropriate methods for measuring the impact of complex public health interventions, particularly if multiple components are involved [77]. There are also limitations to using GRADE to assess certainty of evidence from complex public health interventions, which have been debated [78,79,80]. We would not expect complex interventions to have a rating of 'high' certainty, due to the non-standardised nature of such interventions and heterogeneity [78].

Conclusion

Our systematic review and meta-analysis suggests universal and targeted interventions of different types of setting have no or small benefit in preventing multiple risk behaviours relating to alcohol, tobacco and drug use in young people aged 8-25 years. However, small benefits of universal interventions in trials can have high public health benefit, so universal interventions warrant further investigation. The heterogeneity within some subgroups suggests future work could explore combinations of interventions and components that may be more effective than others. Methodologically robust trials and improved and more consistent reporting within individual studies are needed to allow for better evidence syntheses.

Availability of data and material

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Abbreviations

MRB:

Multiple Risk Behaviour

PRISMA:

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

N:

Number

OR:

Odds Ratio

CI:

Confidence Interval

RCT:

Randomised Controlled Trial

GRADE:

Grades of Recommendation, Assessment, Development and Evaluation

SMD:

Standard Mean Difference

SE:

Standard Error

logOR:

log Odds Ratio

ICC:

Intracluster Correlation Coefficient

References

  1. Patton GC, Sawyer SM, Santelli JS, Ross DA, Afifi R, Allen NB, et al. Our future: a Lancet commission on adolescent health and wellbeing. Lancet. 2016;387(10036):2423–78.

    Article  Google Scholar 

  2. Brooks FM, Magnusson J, Spencer N, Morgan A. Adolescent multiple risk behaviour: an asset approach to the role of family, school and community. J Public Health. 2012;34(suppl 1):i48–56.

    Article  PubMed  Google Scholar 

  3. Hurrelmann K, Richter M. Risk behaviour in adolescence: the relationship between developmental and health problems. J Public Health. 2006;14(1):20–8.

    Article  Google Scholar 

  4. Campbell R, Wright C, Hickman M, Kipping RR, Smith M, Pouliou T, et al. Multiple risk behaviour in adolescence is associated with substantial adverse health and social outcomes in early adulthood: Findings from a prospective birth cohort study. Prev Med. 2020;138:106157.

    Article  PubMed  PubMed Central  Google Scholar 

  5. Wright C, Kipping R, Hickman M, Campbell R, Heron J. Effect of multiple risk behaviours in adolescence on educational attainment at age 16 years: a UK birth cohort study. BMJ Open. 2018;8(7):e020182.

    Article  PubMed  PubMed Central  Google Scholar 

  6. Biglan A. Helping adolescents at risk: Prevention of multiple problem behaviors: Guilford Press. 2004.

    Google Scholar 

  7. Patton GC, Coffey C, Cappa C, Currie D, Riley L, Gore F, et al. Health of the world’s adolescents: a synthesis of internationally comparable data. Lancet. 2012;379(9826):1665–75.

    Article  Google Scholar 

  8. Currie C, et al., eds. Social determinants of health and well-being among young people. Health Behaviour in School-aged Children (HBSC) study: international report from the 2009/2010 survey. Copenhagen: WHO Regional Oice for Europe; 2012 (Health Policy for Children and Adolescents, No. 6).

  9. Hale D, Viner R. Trends in the prevalence of multiple substance use in adolescents in England, 1998–2009. J Public Health. 2013;35(3):367–74.

    Article  PubMed  Google Scholar 

  10. Leatherdale ST, Ahmed R. Alcohol, marijuana, and tobacco use among Canadian youth: do we need more multi-substance prevention programming? J Prim Prev. 2010;31(3):99–108.

    Article  PubMed  Google Scholar 

  11. MacArthur G, Smith M, Melotti R, Heron J, Macleod J, Hickman M, et al. Patterns of alcohol use and multiple risk behaviour by gender during early and late adolescence: the ALSPAC cohort. J Public Health. 2012;34(suppl 1):i20–30.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Connell CM, Gilreath TD, Hansen NB. A multiprocess latent class analysis of the co-occurrence of substance use and sexual risk behavior among adolescents. J Stud Alcohol Drugs. 2009;70(6):943–51.

    Article  PubMed  PubMed Central  Google Scholar 

  13. Conway KP, Vullo GC, Nichter B, Wang J, Compton WM, Iannotti RJ, et al. Prevalence and patterns of polysubstance use in a nationally representative sample of 10th graders in the United States. J Adolesc Health. 2013;52(6):716–23.

    Article  Google Scholar 

  14. Dierker LC, Vesel F, Sledjeski EM, Costello D, Perrine N. Testing the dual pathway hypothesis to substance use in adolescence and young adulthood. Drug Alcohol Depend. 2007;87(1):83–93.

    Article  PubMed  Google Scholar 

  15. Fuller E, Hawkins V. Smoking, drinking and drug use among young people in England in 2011. London: Health and Social Care Information Centre; 2012.

    Google Scholar 

  16. McVie S, Bradshaw P. Adolescent smoking, drinking and drug use: University of Edinburgh, Centre for Law and Society. 2005.

    Google Scholar 

  17. Chassin L, Presson CC, Pitts SC, Sherman SJ. The natural history of cigarette smoking from adolescence to adulthood in a midwestern community sample: multiple trajectories and their psychosocial correlates. Health Psychol. 2000;19(3):223.

    Article  CAS  PubMed  Google Scholar 

  18. Chassin L, Pitts SC, Prost J. Binge drinking trajectories from adolescence to emerging adulthood in a high-risk sample: predictors and substance abuse outcomes. J Consult Clin Psychol. 2002;70(1):67.

    Article  PubMed  Google Scholar 

  19. Chen K, Kandel DB. The natural history of drug use from adolescence to the mid-thirties in a general population sample. Am J Public Health. 1995;85(1):41–7.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  20. Chisholm D, Doran C, Shibuya K, Rehm J. Comparative cost-effectiveness of policy instruments for reducing the global burden of alcohol, tobacco and illicit drug use. Drug Alcohol Rev. 2006;25(6):553–65.

    Article  PubMed  Google Scholar 

  21. Jackson C, Sweeting H, Haw S. Clustering of substance use and sexual risk behaviour in adolescence: analysis of two cohort studies. BMJ open. 2012;2:e000661.

    Article  PubMed  PubMed Central  Google Scholar 

  22. van Nieuwenhuijzen M, Junger M, Velderman MK, Wiefferink KH, Paulussen TWGM, Hox J, et al. Clustering of health-compromising behavior and delinquency in adolescents and adults in the Dutch population. Prev Med. 2009;48:572–8.

    Article  PubMed  Google Scholar 

  23. Biglan A, Brennan PA, Foster SL, Holder H. Helping Adolescents At Risk - Prevention of Multiple Problem Behaviors. London: The Guildofrd Press; 2004.

    Google Scholar 

  24. MacArthur G, Caldwell DM, Redmore J, Watkins SH, Kipping R, White J, et al. Individual‐, family‐, and school‐level interventions targeting multiple risk behaviours in young people. Cochrane Database Syst Rev. 2018;10(10):CD009927.

    PubMed  Google Scholar 

  25. Jackson C, Sweeting H, Haw S. Clustering of substance use and sexual risk behaviour in adolescence: analysis of two cohort studies. BMJ Open. 2012;2(1):e000661.

    Article  PubMed  PubMed Central  Google Scholar 

  26. Foxcroft DR, Tsertsvadze A. Universal alcohol misuse prevention programmes for children and adolescents: Cochrane systematic reviews. Perspect Public Health. 2012;132(3):128–34.

    Article  PubMed  Google Scholar 

  27. Gilligan C, Wolfenden L, Foxcroft DR, Williams AJ, Kingsland M, Hodder RK, et al. Family‐based prevention programmes for alcohol use in young people. Cochrane Database Syst Rev. 2019;3(3):CD012287.

    PubMed  Google Scholar 

  28. Carson KV, Brinn MP, Labiszewski NA, Esterman AJ, Chang AB, Smith BJ. Community interventions for preventing smoking in young people. Cochrane Database Syst Rev. 2011;6(7):CD001291.

    Google Scholar 

  29. Faggiano F, Minozzi S, Versino E, Buscemi D. Universal school‐based prevention for illicit drug use. Cochrane Database Syst Rev. 2014;2014(12):CD003020.

    PubMed Central  Google Scholar 

  30. Hale DR, Fitzgerald-Yau N, Viner RM. A systematic review of effective interventions for reducing multiple health risk behaviors in adolescence. AJPH. 2014;104(5):e19–41.

    Article  Google Scholar 

  31. Newton NC, Champion KE, Slade T, Chapman C, Stapinski L, Koning I, et al. A systematic review of combined student-and parent-based programs to prevent alcohol and other drug use among adolescents. Drug Alcohol Rev. 2017;36(3):337–51.

    Article  PubMed  Google Scholar 

  32. Emmers E, Bekkering GE, Hannes K. Prevention of alcohol and drug misuse in adolescents: An overview of systematic reviews. Nordic Stud Alcohol Drugs. 2015;32(2):183–98.

    Article  Google Scholar 

  33. Tremblay M, Baydala L, Khan M, Currie C, Morley K, Burkholder C, et al. Primary substance use prevention programs for children and youth: a systematic review. Pediatrics. 2020;146(3):e20192747.

    Article  PubMed  Google Scholar 

  34. Jackson C, Geddes R, Haw S, Frank J. Interventions to prevent substance use and risky sexual behaviour in young people: a systematic review. Addiction. 2012;107(4):733–47.

    Article  PubMed  Google Scholar 

  35. Hickman M, Caldwell DM, Busse H, MacArthur G, Faggiano F, Foxcroft DR, Kaner EFS, Macleod J, Patton G, White J, Campbell R. Individual-, family-, and school-level interventions for preventing multiple risk behaviours relating to alcohol, tobacco and drug use in individuals aged 8 to 25 years. Cochrane Database Sys Rev. 2014;(11):CD011374. https://doi.org/10.1002/14651858.CD011374.

  36. MacArthur G, Caldwell DM, Redmore J, Watkins SH, Kipping R, White J, Chittleborough C, Langford R, Er V, Lingam R, Pasch K, Gunnell D, Hickman M, Campbell R. Individual-, family-, and school-level interventions targeting multiple risk behaviours in young people. Cochrane Database Syst Rev. 2018;10(10):CD009927. https://doi.org/10.1002/14651858.CD009927.pub2.

  37. Higgins J. P, T, Green. S. Cochrane Handbook for Systematic Reviews of Interventions. Chichester: John Wiley and Sons; 2008.

    Google Scholar 

  38. Guyatt GH, Oxman AD, Vist GE, Kunz R, Falck-Ytter Y, Alonso-Coello P, Schünemann HJ. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ. 2008;336(7650):924–6.

  39. Botvin GJ, Griffin KW. Life Skills Training: A competence enhancement approach to tobacco, alcohol, and drug abuse prevention. 2015.

    Google Scholar 

  40. Byrnes HF, Miller BA, Grube JW, Bourdeau B, Buller DB, Wang-Schweig M, et al. Prevention of alcohol use in older teens: A randomized trial of an online family prevention program. Psychol Addict Behav. 2019;33(1):1.

    Article  PubMed  PubMed Central  Google Scholar 

  41. Evers KE, Paiva AL, Johnson JL, Cummins CO, Prochaska JO, Prochaska JM, et al. Results of a transtheoretical model-based alcohol, tobacco and other drug intervention in middle schools. Addict Behav. 2012;37(9):1009–18.

    Article  PubMed  Google Scholar 

  42. Huang CM, Chien LY, Cheng CF, Guo JL. Integrating life skills into a theory-based drug-use prevention program: effectiveness among junior high students in Taiwan. J School Health. 2012;82(7):328–35.

    Article  PubMed  Google Scholar 

  43. Moskowitz JM, Malvin JH, Schaeffer GA, Schaps E. An experimental evaluation of a drug education course. J Drug Educ. 1984;14(1):9–22.

    Article  CAS  PubMed  Google Scholar 

  44. Schaps E, Moskowitz JM, Condon JW, Malvin JH. Process and outcome evaluation of a drug education course. J Drug Educ. 1982;12(4):353–64.

    Article  Google Scholar 

  45. Seal N. Preventing tobacco and drug use among Thai high school students through life skills training. Nurs Health Sci. 2006;8(3):164–8.

    Article  PubMed  Google Scholar 

  46. Skeer MR, Yantsides KE, Eliasziw M, Carlton-Smith AR, Tracy MR, Spirito A. Testing a brief substance misuse preventive intervention for parents of pre-adolescents: feasibility, acceptability, preliminary efficacy. J Child Family Stud. 2016;25(12):3739–48.

    Article  PubMed  PubMed Central  Google Scholar 

  47. Vogl LE, Newton NC, Champion KE, Teesson M. A universal harm-minimisation approach to preventing psychostimulant and cannabis use in adolescents: a cluster randomised controlled trial. Subst. Abuse Treat Prev Policy. 2014;9(1):1–14.

    Article  Google Scholar 

  48. Viechtbauer W. Conducting meta-analyses in R with the metafor package. Journal of Statistical Software. 2010;36(3):1–48.

  49. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7(3):177–88.

    Article  CAS  PubMed  Google Scholar 

  50. Sterne JAC, Sutton AJ, Ioannidis JPA, Terrin N, Jones DR, Lau J, et al. Recommendations for examining and interpreting funnel plot asymmetry in meta-analyses of randomised controlled trials. BMJ. 2011;343:d4002. https://doi.org/10.1136/bmj.d4002.

  51. Butzer B, LoRusso A, Shin SH, Khalsa SBS. Evaluation of yoga for preventing adolescent substance use risk factors in a middle school setting: a preliminary group-randomized controlled trial. J Youth Adolescence. 2017;46(3):603–32.

    Article  PubMed  Google Scholar 

  52. D’Amico EJ, Hunter SB, Miles JNV, Ewing BA, Osilla KC. A randomized controlled trial of a group motivational interviewing intervention for adolescents with a first time alcohol or drug offense. J Subst Abuse Treat. 2013;45(5):400–8.

    Article  PubMed  Google Scholar 

  53. McCambridge J, Slym RL, Strang J. Randomized controlled trial of motivational interviewing compared with drug information and advice for early intervention among young cannabis users. Addiction. 2008;103(11):1809–18.

    Article  PubMed  Google Scholar 

  54. Gilchrist LD, Schinke SP, Trimble JE, Cvetkovich GT. Skills enhancement to prevent substance abuse among American Indian adolescents. Int J Addict. 1987;22(9):869–79.

    Article  CAS  PubMed  Google Scholar 

  55. Fang L, Schinke SP, Cole KC. Preventing substance use among early Asian-American adolescent girls: initial evaluation of a web-based, mother-daughter program. J Adolescent Health. 2010;47(5):529–32.

    Article  Google Scholar 

  56. Hernández Serrano O, Espada Sánchez JP, Piqueras Rodríguez JA, Orgilés Amorós M, García-Fernández JM. Programa de prevención del consumo de drogas Saluda: evaluación de una nueva versión en adolescentes españoles. Health and Addictions/Salud y Drogas. 2013;13(2):135–44.

  57. Malmberg M, Kleinjan M, Overbeek G, Vermulst A, Monshouwer K, Lammers J, et al. Effectiveness of the ‘H ealthy S chool and D rugs’ prevention programme on adolescents’ substance use: a randomized clustered trial. Addiction. 2014;109(6):1031–40.

    Article  PubMed  Google Scholar 

  58. Sussman S, Dent CW, Stacy AW, Craig S. One-year outcomes of project towards no drug abuse. Prev Med. 1998;27(4):632–42.

    Article  CAS  PubMed  Google Scholar 

  59. Sussman S, Sun P, McCuller WJ, Dent CW. Project Towards No Drug Abuse: two-year outcomes of a trial that compares health educator delivery to self-instruction. Prev Med. 2003;37(2):155–62.

    Article  PubMed  Google Scholar 

  60. Agabio R, Trincas G, Floris F. A systematic review of school-based alcohol and other drug prevention programs. Clin Pract Epidemiol Ment Health. 2015;11(Suppl 1 M6):102.

    Article  PubMed  PubMed Central  Google Scholar 

  61. Riley RD, Jackson D, Salanti G, Burke DL, Price M, Kirkham J et al. Multivariate and network meta-analysis of multiple outcomes and multiple treatments: rationale, concepts, and examples. BMJ. 2017;358:j3932. https://doi.org/10.1136/bmj.j3932.

  62. Dodge KA. Annual Research Review: Universal and targeted strategies for assigning interventions to achieve population impact. J Child Psychol Psychiatry. 2020;61(3):255–67.

    Article  PubMed  Google Scholar 

  63. Rose G. Individuals and populations. The strategy of preventive medicine. Oxford: Oxford University Press; 1992.

    Google Scholar 

  64. Champion KE, Parmenter B, McGowan C, Spring B, Wafford QE, Gardner LA, et al. Effectiveness of school-based eHealth interventions to prevent multiple lifestyle risk behaviours among adolescents: a systematic review and meta-analysis. Lancet Digital Health. 2019;1(5):e206–21.

    Article  PubMed  Google Scholar 

  65. Smith EA, Cooper NJ, Sutton AJ, Abrams KR, Hubbard SJ. A review of the quantitative effectiveness evidence synthesis methods used in public health intervention guidelines. BMC Public Health. 2021;21(1):1–25.

    Article  Google Scholar 

  66. Welton NJ, Caldwell DM, Adamopoulos E, Vedhara K. Mixed treatment comparison meta-analysis of complex interventions: psychological interventions in coronary heart disease. Am J Epidemiol. 2009;169(9):1158–65.

    Article  PubMed  Google Scholar 

  67. Caldwell DM, Davies SR, Thorn J, Palmer JC, Caro P, Hetrick SE, et al. School-based interventions to prevent anxiety, depression and conduct disorders in children and young people: a systematic review, network meta-analysis and economic evaluation. Lancet Psychiatry. 2019;6(12):1011–20.

  68. Tinner L, Caldwell D, Hickman M, MacArthur GJ, Gottfredson D, Lana Perez A, et al. Examining subgroup effects by socioeconomic status of public health interventions targeting multiple risk behaviour in adolescence. BMC Public Health. 2018;18(1):1180.

    Article  PubMed  PubMed Central  Google Scholar 

  69. Krieger N. Structural racism, health inequities, and the two-edged sword of data: structural problems require structural solutions. Front Public Health. 2021;9:301.

    Article  Google Scholar 

  70. King K, Meader N, Wright K, Graham HPC, Petticrew M, White M, et al. Characteristics of Interventions Targeting Multiple Lifestyle Risk Behaviours in Adult Populations: A Systematic Scoping Review. PLoS ONE. 2015;10(1):e0117015.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  71. Sawyer SM, Afifi RA, Bearinger LH, Blakemore SJ, Dick B, Ezeh AC, et al. Adolescence: a foundation for future health. Lancet. 2012;379:1630–40.

    Article  PubMed  Google Scholar 

  72. Lee Y-CA, Hashibe M. Tobacco, alcohol, and cancer in low and high income countries. Ann Global Health. 2014;80(5):378–83.

    Article  Google Scholar 

  73. Walls H, Cook S, Matzopoulos R, London L. Advancing alcohol research in low-income and middle-income countries: a global alcohol environment framework. BMJ Global Health. 2020;5(4):e001958.

    Article  PubMed  PubMed Central  Google Scholar 

  74. Peacock A, Leung J, Larney S, Colledge S, Hickman M, Rehm J, et al. Global statistics on alcohol, tobacco and illicit drug use: 2017 status report. Addiction. 2018;113(10):1905–26.

    Article  PubMed  Google Scholar 

  75. Irwin CE, Burg SJ, Cart CU. America’s adolescents: where have we been, where are we going? J Adolescent Health. 2002;31(6):91–121.

    Article  Google Scholar 

  76. Steinberg L, Monahan KC. Age differences in resistance to peer influence. Dev Psychol. 2007;43(6):1531.

    Article  PubMed  PubMed Central  Google Scholar 

  77. Bonell C, Fletcher A, Morton M, Lorenc T, Moore L. Realist randomised controlled trials: a new approach to evaluating complex public health interventions. Soc Sci Med. 2012;75(12):2299–306.

    Article  Google Scholar 

  78. Movsisyan A, Melendez-Torres GJ, Montgomery P. Outcomes in systematic reviews of complex interventions never reached “high” GRADE ratings when compared with those of simple interventions. J Clin Epidemiol. 2016;78:22–33.

    Article  PubMed  Google Scholar 

  79. Movsisyan A, Melendez-Torres GJ, Montgomery P. Users identified challenges in applying GRADE to complex interventions and suggested an extension to GRADE. J Clin Epidemiol. 2016;70:191–9.

    Article  PubMed  Google Scholar 

  80. Rehfuess EA, Akl EA. Current experience with applying the GRADE approach to public health interventions: an empirical study. BMC Public Health. 2013;13(1):1–3.

Download references

Acknowledgements

The work was undertaken with the support of The Centre for the Development and Evaluation of Complex Interventions for Public Health Improvement (DECIPHer), a UKCRC Public Health Research Centre of Excellence. Joint funding (MR/KO232331/1) from the British Heart Foundation, Cancer Research UK, Economic and Social Research Council, Medical Research Council, the Welsh Government and the Wellcome Trust, under the auspices of the UK Clinical Research Collaboration, is gratefully acknowledged. This work is part of a programme of work on multiple risk behaviours in young people being undertaken within DECIPHer.

Funding

The work was undertaken with the support of The Centre for the Development and Evaluation of Complex Interventions for Public Health Improvement (DECIPHer), a UKCRC Public Health Research Centre of Excellence. Joint funding (MR/KO232331/1) from the British Heart Foundation, Cancer Research UK, Economic and Social Research Council, Medical Research Council, the Welsh Government and the Wellcome Trust, under the auspices of the UK Clinical Research Collaboration, is gratefully acknowledged.

LT’s post-doctoral post is funded by Bristol, North Somerset and South Gloucestershire Clinical Commission Group (CCG) Research Capacity Funding via RC’s Senior Investigator Award.

JP is supported by a British Heart Foundation Accelerator Award (AA/18/7/34219) and works in a Unit that is supported by the University of Bristol and UK Medical Research Council (MC_UU_00011/6). The funders had no role in the research design or interpretation of results. Views expressed in this paper are those of the authors and not necessarily any funders.

GJM is funded by an NIHR post-doctoral fellowship award (PDF-2013-06-026). This report is independent research supported by the National Institute for Health Research (Post-Doctoral Fellowship, PDF-2013-06-026). The views expressed in this publication are those of the author(s) and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health. This work was primarily funded by the UK Clinical Research Collaboration.

Author information

Author notes

  1. Laura Tinner and Jennifer C Palmer contributed equally to this work.

Authors and Affiliations

  1. Bristol Medical School, Population Health Sciences, University of Bristol, 39 Whatley Road, Bristol, BS8 2PS, UK

    Laura Tinner, Jennifer C Palmer, Deborah M Caldwell, Georgie J MacArthur, Kaiseree Dias, Rebecca Langford, Linda Wittkop, Matthew Hickman & Rona Campbell

  2. MRC Integrative Epidemiology Unit at the University of Bristol, Bristol, UK

    Jennifer C Palmer & Sarah Holmes Watkins

  3. Department of Psychiatry, Columbia University Irving Medical Center, New York, New York, USA

    E. Caitlin Lloyd

  4. New York State Psychiatric Institute, New York, New York, USA

    E. Caitlin Lloyd

  5. York Teaching Hospitals NHS Foundation Trust, Wiggington Road, York, UK

    James Redmore

Authors
  1. Laura Tinner
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Jennifer C Palmer
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. E. Caitlin Lloyd
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Deborah M Caldwell
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Georgie J MacArthur
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Kaiseree Dias
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Rebecca Langford
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. James Redmore
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Linda Wittkop
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Sarah Holmes Watkins
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Matthew Hickman
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Rona Campbell
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

LT was involved in the data checking, creating tables, data interpretation and jointly wrote the first draft of the manuscript. JP undertook the meta-analyses, sensitivity analyses, data interpretation and jointly wrote the first draft of the manuscript. CL contributed to study screening, data extraction, and refining manuscript drafts. DC was involved in data checking, data analysis and interpretation, GRADE assessments and wrote sections of the first draft of the manuscript. RL was involved in the conceptual design of the study’s analysis plan and contributed to the writing of the manuscript. GJM contributed to the study protocol, literature search, screening of studies and data extraction. KD contributed to data extraction and commented on the manuscript. LW conducted full text screening and was one of two independent reviewers for 2018/19 updated searches. JR contributed to data abstraction and commented on the manuscript. SHW contributed to screening, data extraction and commented on the manuscript. MH and RC were involved in the conceptual design of the study and contributed to writing the manuscript. The author(s) read and approved the final manuscript.

Corresponding author

Correspondence to Laura Tinner.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

Authors declare no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1.

Search strategy.

Additional file 2.

Details of excluded and ongoing studies.

Additional file 3. 

Excluded studies from meta-analysis.

Additional file 4.

Details of data transformations.

Additional file 5.

Outcome measures.

Additional file 6.

Planned subgroup analysis.

Additional file 7.

 Table of characteristics.

Additional file 8.

Sensitivity analyses.

Additional file 9. 

Forest plots for long term substance use outcomes.

Additional file 10.

Funnel plots.

Additional file 11.

 Secondary outcomes.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Tinner, L., Palmer, J.C., Lloyd, E.C. et al. Individual-, family- and school-based interventions to prevent multiple risk behaviours relating to alcohol, tobacco and drug use in young people aged 8-25 years: a systematic review and meta-analysis. BMC Public Health 22, 1111 (2022). https://doi.org/10.1186/s12889-022-13072-5

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12889-022-13072-5

BMC Public Health

ISSN: 1471-2458

Contact us