Is behavioural activation effective in the treatment of depression in young people? A systematic review and meta‐analysis

Purpose Depression is currently the leading cause of illness and disability in young people. Evidence suggests that behavioural activation (BA) is an effective treatment for depression in adults but less research focuses on its application with young people. This review therefore examined whether BA is effective in the treatment of depression in young people. Methods A systematic review (International Prospective Register of Systematic Reviews reference: CRD42015020453), following Preferred Reporting Items for Systematic Reviews and Meta‐Analyses guidelines, was conducted to examine studies that had explored behavioural interventions for young people with depression. The electronic databases searched included the Cochrane Library, EMBASE, MEDLINE, CINAHL Plus, PsychINFO, and Scopus. A meta‐analysis employing a generic inverse variance, random‐effects model was conducted on the included randomized controlled trials (RCTs) to examine whether there were overall effects of BA on the Children's Depression Rating Scale – Revised. Results Ten studies met inclusion criteria: three RCTs and seven within‐participant designs (total n = 170). The review showed that BA may be effective in the treatment of depression in young people. The Cochrane risk of bias tool and the Moncrieff scale used to assess the quality of the included studies revealed a variety of limitations within each. Conclusions Despite demonstrating that BA may be effective in the treatment of depression in young people, the review indicated a number of methodological problems in the included studies meaning that the results and conclusions should be treated with caution. Furthermore, the paucity of studies in this area highlights the need for further research. Practitioner points Currently BA is included within National Institute for Health and Clinical Excellence (NICE, 2009) guidelines as an evidence‐based treatment for depression in adults with extensive research supporting its effectiveness. It is important to investigate whether it may also be effective in treating young people. Included studies reported reductions in depression scores across a range of measures following BA. BA may be an effective treatment of depression in young people.

provides some evidence that BA may be effective when delivered in a low intensity form (e.g., guided self-help).
Although treatment recommendations and guidelines for individuals experiencing depression differ between adults and young people, given the extensive research supporting the use of BA with adults it is important to investigate whether it may also be effective in treating young people.
There has been less research on the use of BA with children and young people. Research in this area has generally been in the form of case series with small sample sizes, for example Chu, Colognori, Weissman, and Bannon (2009;n = 5); Wallis, Roeger, Milan, Walmsley, and Alison (2012;n = 5). Both of these reviews provided support for the use of BA in the treatment of young people with depression and/or anxiety, finding high treatment satisfaction and clinical benefits including symptom reductions. Similar results were obtained in a pilot, uncontrolled study of the use of BA for treating depressed young people in rural Australia (Jacob, Keeley, Ritschel, & Craighead, 2013), with all participants (n = 5) showing reduced levels of depressive symptoms between baseline and completion (at 6 months).
Given the paucity of research in this area and the lack of any published systematic review, an examination of BA for use with children and young people is required and timely (Chartier & Provencher, 2013). Young people experiencing depression may be treated more effectively using computerized therapies which have increased availability and accessibility , less stigma and are presented in a format attractive to many young people compared to traditional face-to-face therapies. Thus far, much research has focused on the delivery of CBT in this form (e.g., Abeles et al., 2009;Spence, Holmes, March, & Lipp, 2006;Spence et al., 2008;Wright et al., 2014), with computerized CBT (CCBT) representing an alternative form of therapy delivery that can be administered at a lower cost than traditional treatments (Merry et al., 2012). Although adult research has also often focused upon CCBT for depression, research into computerized BA has started to emerge. In a review by Spates et al. (2016), five web-based BA interventions were identified all of which have demonstrated relative success in initial pilot trials.
Research is therefore required to establish whether BA in a computerized format has been used with young people experiencing depression also and, if so, whether it is an effective treatment.
This review sought to investigate (1) whether BA is effective in the treatment of depression in young people and, if so, (2) whether it can be effectively delivered in a computerized form.

Methods
This review was completed with reference to the guidelines reported in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Liberati et al., 2009). The review protocol was registered on the International Prospective Register of Systematic Reviews (PROSPERO), an online database for systematic review protocols (reference: CRD42015020453).

Information sources and study identification
The following electronic databases were searched between July and August 2015: Cochrane Library, EMBASE, MEDLINE, CINAHL Plus (EBSCO), PsychINFO, Scopus, and the ISCRTN registry. To cover peer review and grey literature sources, the Health Management Information Consortium, NHS evidence, Open Grey, the Networked Digital Library of Theses and Dissertations, Web of Science Conference Proceedings, and ZETOC were searched with the search simplified accordingly. The reference lists of all included studies were examined and forward citation searching carried out in Google Scholar. No restrictions on publication status or language were imposed.
Titles and abstracts were screened independently by the primary researcher alongside a second reviewer to enhance the reliability of included studies. If any disagreements occurred between the two researchers, the two met to discuss these. In the event that a decision could not be reached between the two, a third researcher was consulted and asked to screen the disagreed paper(s) and make an overall decision regarding selection.
The search strategy used was based on three main constructs: behavioural interventions (including BA, behavioural therapy, behavioural interventions, selfmonitoring, and activity scheduling), depression (including depressive disorder, depressive, depression, and depressed) and young people (including adolescents, children, teen, youth, juvenile, pre-pubescent, and student). See Appendix for the search strategy used.

Eligibility criteria
The population of interest was young people aged 18 years and below. Studies employing a population that crossed the age of 18 were included if a minimum of 90% of the sample was under 18. Trial participants had to be experiencing depression or depressive symptoms as established by a validated screening measure or diagnosis based on a structured clinical interview conducted to internationally recognized standards (e.g., International Classification of Disease, Diagnostic Statistical Manual).
For inclusion, interventions had to be based upon either operant conditioning principles or comprise techniques fundamental to behavioural treatments of depression (activity scheduling, self-monitoring, goal setting). Interventions based on third-wave CBT principles (e.g., acceptance and commitment therapy) were excluded.
No restrictions were placed on comparator or control group types to avoid excluding any relevant studies reporting on BA. Studies that did not employ a control group were also included.
The primary outcome measure was levels of depression/depressive symptoms as measured by validated assessments. Assessments could include self-report measures and clinician or researcher administrated ratings. Additional outcomes included levels of anxiety symptoms (measured by validated assessments), cost-effectiveness, quality of life, and school attendance.
Randomized controlled trial (RCT) and pre-/post-study designs presenting relevant outcomes in a useable format were included in the review.
No restrictions were placed on intervention duration, delivery settings (e.g., community, health care, educational), or delivery mode (e.g., computerized, face-toface), the timings of the measurement of the outcome measures nor upon sample sizes or sampling methodologies.

Data extraction
Information extracted (using a pre-piloted proforma) from included studies comprised study characteristics (study name, author(s), year of publication/production (if unpublished), study location, and setting), study design, study populations (basic demographics of participants, depression diagnosis methods), intervention details and comparators (intervention type, comparator, duration of the intervention, number of sessions), and relevant outcome data for effect size calculations (depression severity, unit of measurement).

Quality assessments
The methodological quality of RCTs was formally assessed using the Cochrane risk of bias tool (Higgins et al., 2011), a general tool used to assess risk in any RCT and the Moncrieff scale (Moncrieff, Churchill, Drummon, & McGuire, 2001), specifically designed to assess the quality of controlled studies examining interventions for depressive and nonpsychotic symptoms. The Cochrane risk of bias tool categorises risk as 'high', 'low', or 'unclear' (where insufficient information is supplied to assess level of risk), whilst the Moncrieff scale awards scores (0-2) based on a study's success at addressing 23 risk items with higher scores representing higher levels of quality.
The inclusion of both scales allowed comparisons to be made between a general quality assessment tool and one specifically designed for use within the proposed field. The Moncrieff scale was also used to assess the quality of within-participant design studies.

Data synthesis and statistical analysis
A summary of the outcome measures of all included studies is provided alongside a forest plot providing a graphical display of the study outcomes of the RCTs. Owing to the limited number of RCTs, only one meta-analysis was conducted for the Children's Depression Rating Scale -Revised (CDRS-R; Poznanski & Mokros, 1996). Studies were pooled using the generic inverse variance method with a random-effects model. All analyses were undertaken in stata version 13 (Stata Corporation, 2013). Statistical heterogeneity was assessed using the I 2 statistic with a value of 25% being regarded as low, 50% as moderate, and 75% as high (Higgins, Thompson, Deeks, & Altman, 2003). Publication bias could not be examined due to insufficient numbers of included studies.
Five of the ten studies also collected outcome data relating to anxiety, whilst only one reported on quality of life. None of the other secondary outcomes of interest, costeffectiveness, and school attendance was reported in any of the studies. Table 1 provides a summary of the descriptive characteristics of the included studies.

Quality assessments
Randomized controlled trials On the Cochrane risk of bias tool (Table 2), all three RCTs demonstrated low risk regarding the reporting of other sources of bias, incomplete outcome data, and blinding outcome assessors. Level of bias was unclear where no, or insufficient, information regarding a particular domain (allocation concealment (Chu et al., 2016;Stark, 1985), random sequence generation; Stark, 1985) was provided. In addition, with no prepublished protocols, it was hard to determine whether all pre-specified outcomes had been reported in each of the three RCTs and therefore the level of bias due to selective reporting was unclear. As the three RCTs involved the delivery of BA neither participants nor personnel could be blinded to treatment allocation. However, a lack of blinding meant that all three RCTs demonstrated a high risk of bias on this domain.
On the Moncrieff scale (Table 3), all three RCTs attained maximum quality scores for method of allocation, providing clear descriptions of treatment, using clear diagnostic criteria and the recording of exclusion criteria. However, on a number of the domains examined, although bias was minimized, it was still present (e.g., in conducting appropriate statistical analyses and assessment of compliance to treatments). High risk of bias was reported where studies provided no or insufficient information about power calculations, concealment of allocation (Chu et al., 2016;Stark, 1985), sample sizes, and declarations of interest (Stark, 1985). Only the study by Chu et al. (2016) attained a maximum score for reporting side effects, with neither of the other two RCTs discussing this.

Within-participant designs
Only one domain on the Moncrieff scale (Table 4) attained a maximum quality score on all seven included studies. This related to the outcome measures used with all studies using validated and reliable instruments. However, none of the studies conducted power calculations, blinded assessors, employed adequate sample sizes, or reported any information about the side effects of treatment. As a result, every study was awarded zero suggesting the presence of bias on each of these domains. For the remainder of bias domains on the Moncrieff scale, most studies attained a score of one or two suggesting measures had been taken to minimize the level of bias. However, a number of studies Notes. Depression measures: CDRS-R, Children's Depression Rating Scale -Revised (Poznanski & Mokros, 1996); SMFQ, Short Mood and Feelings Questionnaire (Angold, Costello, Messer, & Pickles, 1995); CES-D, Center for Epidemiologic Studies Depression Scale (Radloff, 1977); CDI, Children's Depression Inventory (Kovacs, 2011); CDS, Children's Depression Scale (Reynolds, 1980); BDI-II, Beck Depression Inventory (Beck, Steer, & Brown, 1996); K-SADS, Kiddie Schedule for Affective Disorders (Kaufman et al., 1997); MINI-KID, Mini International Neuropsychiatric Interview for Children and Adolescents (Sheehan et al., 2010). Anxiety measures: MASC, Multi-Dimensional Anxiety Scale for Children (March, Parker, Sullivan, Stallings, & Conners, 1997); SCARED, Screen for Anxiety Related Emotional Disorders ; RCMAS, Revised Children's Manifest Anxiety Scale (Reynolds & Richmond, 1978); ADIS-IV, Anxiety Disorders Interview Schedule for DSM-IV Child Interview (Silverman & Albano, 1996). Quality of life measures: FQOLS, The Family Quality of Life Scale-Family Interactions Subscale (Hoffman, Marquis, Poston, Summers, & Turnbull, 2006). RCT, randomized controlled trials. Notes. Maximum total score is 56; higher scores denote lower bias.
Wallis et al.
Wallis et al.
Weersing et al. Notes. Maximum total score is 56; higher scores denote lower bias.
Through not employing a randomized methodology, two items on the Moncrieff scale (methods of allocation and concealment of allocation) were not applicable to the withinparticipant designs.

BA effectiveness and depression
The effectiveness of BA is reported separately for each of the two types of study designs included in the review (RCTs, within-participant designs). Owing to the limited number of RCTs included, only one meta-analysis could be conducted. However, a forest plot depicting depression outcomes on all measures employed within each of the three RCTs is provided.

Randomized controlled trials
Two of the included RCTs measured depression outcomes using two continuous measures of depression (Chu et al., 2016 (Stark, 1985: CDRS-R, Children's Depression Inventory [CDI], Children's Depression Scale [CDS]). A forest plot was produced to provide a graphical display of these study outcomes (Figure 2). During data extraction, it was noted that in one of the studies (Chu et al., 2016), there was a baseline imbalance in depression severity scores (intervention 42.57 [SE 5.08], wait list 46.00 [SE 3.95]) and an error in the reporting of the standard error for the depression severity scores post-intervention (wait list SE = 0; intervention SE = 5.36). Rather than exclude this study from the analysis, it was decided to use the reported estimates and standard errors from the adjusted models (including adjustments for baseline scores) to ensure inappropriately large differences were not attributed to the intervention.
All three RCTs measured depression using the CDRS-R and demonstrated reductions in depression scores using this measure. McCauley et al. (2015) reported that mean CDRS-R scores between pre-treatment and end of treatment reduced from 57.6 (SD: 11.8) to 40.18 (SD: 13.9) for those receiving BA in comparison with a reduction from 57.8 (SD: 8.3) to 45.05 (SD: 14.2) for those receiving treatment as usual. At end of treatment, 76% of participants randomized to BA scored forty or below on the CDRS-R, indicating a depression diagnosis to be either 'possible' or 'unlikely' in comparison with 42% of those receiving treatment as usual. These pre-treatment to end-of-treatment outcomes fell within the 95% confidence interval suggesting reliability in the change scores. In the study by Chu et al. (2016), CDRS-R depression scores reduced from 42.6 (SD: 5.08) to 37.67 (SD: 5.36) from pre-treatment to post-treatment in the BA group, whilst scores increased from 46.0 (SD: 3.95; pre-wait list) to 57.0 (SD: 0.00; post-wait list) in the control group.
Owing to the small sample size employed within this study, statistical analyses were not performed, and therefore, the significance of these results cannot be inferred. Finally, Stark (1985) reported reductions in mean CDRS-R depression scores for those receiving BA to be from 33.50 (SD: 10.27) at pre-treatment to 24.02 (SD: 6.01) at end of treatment and then to 24.28 (SD: 4.68) at follow-up. For those receiving self-control therapy, mean CDRS-R scores reduced from 37.22 (SD: 8.36) at pre-treatment to 22.90 (SD: 4.36) at end of treatment and to 20.69 (SD: 3.45) at follow-up. Reductions in the wait list group reduced from 27.57 (SD: 3.51) at pre-treatment to 27.24 (SD: 5.74) at end of treatment and to 22.60 (SD: 5.03) at follow-up. However, the results of an ANCOVA test demonstrated that the difference between groups on the CDRS-R at end of treatment was not significant (p < .30).
As the included RCTs were judged to be sufficiently similar, a meta-analysis was conducted for the CDRS-R. The effect of BA on CDRS-R depression scores was moderate with a pooled mean difference of À4.17 (95% CI: À8.25, À0.09; Figure 3). This demonstrates a statistically significant difference in CDRS-R scores in favour of BA. The I 2 statistic was 0% (p = .926) suggesting no statistical heterogeneity was present (Higgins et al., 2003).
In relation to the other depression measures, in the study by Stark (1985), mean depression scores as measured by the CDI reduced in the BA group from 22.40 (SD: 8.47) at pre-treatment to 9.11 (SD: 8.32) at end of treatment and to 7.43 (SD: 7.23) at follow-up. In the self-control therapy, mean CDI scores reduced from 21.60 (SD: 5.48) at pretreatment to 8.09 (SD: 6.65) at end of treatment and to 5.36 (SD: 5.04) at follow-up, whilst of an ANCOVA test demonstrated that these between-group differences were statistically significant (p < .01) at the end of treatment. Thus, those receiving treatment had a greater reduction in depression scores, at this point than those in the wait list control group.
In the same study (Stark, 1985), mean depression scores as measured by the CDS also reduced in the BA group from 71.10 (SD: 10.38) at pre-treatment to 55.24 (SD: 12.18) at end of treatment and to 50.03 (SD: 13.23) at follow-up. In the self-control therapy group, CDS scores reduced from 72.40 (SD: 10.31) at pre-treatment to 50.29 (SD: 8.63) at end of treatment and then further reduced to 46.46 (SD: 8.31) at followup. For the wait list group, mean CDS scores reduced from 66.00 (SD: 18.80) at pretreatment to 62.61 (SD: 7.14) at end of treatment and then to 48.20 (SD: 13.29) at follow-up. These differences, however, fell short of conventional levels of statistical significance (p < .07).
On the CES-D in the study by Chu et al. (2016), rates of depression reduced in the BA group from 21.00 (SD: 2.15) to 16.38 (SD: 2.30) compared to a reduction from 20.22 (SD: 2.73) to 19.07 (SD: 3.15) in the wait list group. However, once again statistical tests were not performed due to a small sample size and lack of control.
The final continuous measure used within the three RCTs was the SMFQ. In the BA group, mean scores reduced from 16.1 (SD: 6.1) at pre-treatment to 6.3 (SD: 7.4) at end of treatment in comparison with a reduction from 15.6 (SD: 6.2) at pre-treatment to 6.5 (SD: 6.5) at end of treatment in the treatment as usual group. These differences in groups were, however, not significant (p = .53). Results demonstrated that 77% of BA participants no longer met diagnostic criteria for depression at end of treatment compared to 25% in the treatment as usual group.
In the study by Riley and Gaynor (2014), all participants received non-directive therapy (NDT) and, if not demonstrating improved depression ratings, subsequently received BA. Of those who received BA, 57% demonstrated a clinically significant change on both the CDRS-R and the CDI at the end of treatment. There were also significant differences on both measures from post-NDT to post-BA (CDRS-R: M = 41.57 [11.79]: Z = À2.37, p = .02; CDI: M = 16.29 [10.24]: Z = À2.37, p = .02).
All participants in the study by Douleh (2013)  In all within-participant designs employing a case series methodology, reductions in depression scores were evident following BA. Jacob et al. (2013) and Wallis et al. (2012) reported reduced scores on the BDI-II for all participants from baseline to trial completion with the participants in the latter of these two studies attaining depressive scores in the 'normal' range on this measure. Jacob et al. (2013) also reported reductions in depressive scores on the CDRS-R and the K-SADS with 2/3 participants no longer meeting the criteria of a depressive disorder following BA. Similarly, Ritschel et al. (2011) reported significant reductions in depressive scores as measured by both the BDI-II and the CDRS-R with 66% of participants being in the 'normal' range following treatment completion and thus similar to a non-clinical sample. In the study by Weersing et al. (2008), both participants demonstrated a decrease in depression scores on the CDI from baseline to six-month follow-up, whilst Chu et al. (2009) reported significant reductions in depression scores on the CES-D for 2/5 participants. All reported results for the within-participant design studies can be seen in Table 5.

BA effectiveness and other outcomes
Besides depression, this review also sought to examine the effectiveness of BA and several additional outcomes of interest. These were levels of anxiety symptoms, costeffectiveness, quality of life, and school attendance.

Randomized controlled trials
All three RCTs examined the effectiveness of BA in the treatment of anxiety. Chu et al. (2016) : 4.56]) from pre-to post-treatment as measured by the Screen for Anxiety Related Emotional Disorders (SCARED). Once again owing to the small sample size employed within this study, statistical analyses were not performed, and therefore, the significance of these results cannot be inferred. Finally, using the Revised Children's Manifest Anxiety Scale, Stark (1985) reported statistically significant reductions in anxiety from pre-to post-testing in those receiving either BA or self-control therapy (p < .01) and no improvement for those in the wait list condition. Individuals who received self-control therapy demonstrated the highest reductions in anxiety at post-testing.
None of the included RCTs reported on the cost-effectiveness of BA nor on its impact on quality of life or school attendance.

Within-participant designs
Of the seven within-participant design studies, two reported on the effectiveness of BA for treating anxiety (Chu et al., 2009;Weersing et al., 2008). Based on the SCARED, both participants in the study by Weersing et al. (2008) demonstrated reduced anxiety scores following BA. For one participant, anxiety scores reduced across all time points following treatment completion; however, for the other, anxiety scores had increased again by sixmonth follow-up. In the study by Chu et al. (2009), two of the five included participants had reduced anxiety scores, as measured by the MASC, following BA. For the remaining three, one saw an increase in their anxiety score, for one it remained the same, whilst the other withdrew from treatment and did not complete follow-up measures.
None of the included within-participant designs reported on the cost-effectiveness of BA nor on its impact on school attendance.

Computerized BA
Although the ten studies included within this review provided information regarding the effectiveness of BA for young people with depression, none of the studies delivered BA in a computerized form. Therefore, the second objective of this review: whether BA can be effectively delivered in a computerized form, could not be investigated. Furthermore, none of the studies reported on the impact of BA on school attendance or investigated the cost-effectiveness of the treatmentthe remaining secondary outcomes specified.

Discussion
This systematic review aimed to assess the effectiveness of BA in the treatment of adolescents with depression and investigate whether BA delivered in a computerized form is effective within this treatment context. Across all included studies, regardless of methodology, reductions in depression were evident following BA. At the individual level, several of the findings were statistically significant, and when the RCT studies were combined within the meta-analysis, a statistically significant difference in CDRS-R scores from pre-to post-treatment was found in favour of BA.
The findings of this review provide some preliminary evidence that BA may be an effective treatment of depression in young people. Not only were reductions in depression scores reported following BA across studies, within the RCTs these reductions were greater in comparison with those randomized to a control group. These findings are similar to those previously reported in adult studies where BA has been found to be superior to control conditions (e.g., Ekers et al., 2008;Mazzuchelli et al., 2009).
Besides depression, several of the included studies examined the effectiveness of BA on two of the secondary outcome measuresanxiety and quality of life. Five of the included studies (Chu et al., 2016;McCauley et al., 2015;Stark, 1985;Chu et al., 2009;Weersing et al., 2008) measured anxiety scores with all reporting reductions following BA. In relation to quality of life, in the one study that reported on it (Riley & Gaynor, 2014), there was a significant increase found by the conclusion of BA. These findings provide preliminary evidence that BA may also be effective in reducing anxiety and increasing quality of life for young people experiencing depression.
The second aim of this systematic review was to investigate whether BA delivered in a computerized form is effective in the treatment of young people with depression. Despite research suggesting the effectiveness of computer-delivered therapies for young people (e.g., Merry et al., 2012;Stallard et al., 2010), none of the ten included studies delivered BA in a computerized form, and therefore, this aim could not be addressed. In addition, none examined the impact of BA on school attendance or investigated the costeffectiveness of the treatment.

Limitations
Although this review has generated important information relating to the effectiveness of BA in the treatment of depression in young people, a number of methodological limitations need to be considered when interpreting the findings.
Firstly, as only three RCTs were included within the review, only one meta-analysis could be completed. Although a pooled mean difference could be calculated and supports the effectiveness of BA for the treatment of depression, the results need to be interpreted with caution. The discrepancies noted with one of the included RCTs (Chu et al., 2016) and the subsequent adjustment made may have impacted upon the accuracy of the results reported. In addition, through the inclusion of only three studies in the meta-analysis, explorations of publication bias could not be conducted.
The methodological flaws identified in all included studies may also have impacted upon the results presented. For example, a number of studies (e.g., Chu et al., 2016;Stark, 1985) provided no information regarding allocation concealment which may have inflated the effect sizes in favour of positive results (Shulz & Grimes, 2002;Wood et al., 2008). As several studies (e.g., Chu et al., 2009;Jacob et al., 2013) did not conduct any statistical analyses, their findings could only be inferred. In addition, the sample sizes of the included studies were low with the maximum number recruited in any study being 60 (McCauley et al., 2015). It must be noted that this was the only study to employ power calculations. Therefore, it is unclear whether the other two RCTs or withinparticipant design studies recruited sufficient numbers of participants to identify an intervention effect.
In two of the studies, a transdiagnostic approach was taken (Chu et al., 2009;Weersing et al., 2008) whereby BA was used to treat both depression and anxiety. Although there is often comorbidity between anxiety and depression and so it was added within this review as a secondary outcome, the primary focus was on the effectiveness of BA for treating depression. Thus, it may be hard to distinguish between the elements of BA effective for treating depression and those for anxiety within these studies as the two are reported collectively.
Finally, no information was supplied in any of the studies about delivery of BA in a computerized form nor on the additional outcomes under review (school attendance, cost-effectiveness). These were deemed as important factors at inception of the review but unfortunately cannot be reported on.

Conclusions
This review was conducted to examine the effectiveness of BA for treating young people with depression. Ten studies (three RCTs and seven within-participant designs) met the inclusion criteria and were subsequently included. The results provided some initial evidence that BA may be an effective treatment of depression in young people.
A number of methodological constraints in the included studies mean that the results need to be interpreted with caution. Such constraints need to be addressed in any future research.