Cost and Outcome of Behavioural Activation versus Cognitive Behavioural Therapy for Depression (COBRA): a randomised, controlled, non-inferiority trial

Summary Background Depression is a common, debilitating, and costly disorder. Many patients request psychological therapy, but the best-evidenced therapy—cognitive behavioural therapy (CBT)—is complex and costly. A simpler therapy—behavioural activation (BA)—might be as effective and cheaper than is CBT. We aimed to establish the clinical efficacy and cost-effectiveness of BA compared with CBT for adults with depression. Methods In this randomised, controlled, non-inferiority trial, we recruited adults aged 18 years or older meeting Diagnostic and Statistical Manual of Mental Disorders IV criteria for major depressive disorder from primary care and psychological therapy services in Devon, Durham, and Leeds (UK). We excluded people who were receiving psychological therapy, were alcohol or drug dependent, were acutely suicidal or had attempted suicide in the previous 2 months, or were cognitively impaired, or who had bipolar disorder or psychosis or psychotic symptoms. We randomly assigned participants (1:1) remotely using computer-generated allocation (minimisation used; stratified by depression severity [Patient Health Questionnaire 9 (PHQ-9) score of <19 vs ≥19], antidepressant use, and recruitment site) to BA from junior mental health workers or CBT from psychological therapists. Randomisation done at the Peninsula Clinical Trials Unit was concealed from investigators. Treatment was given open label, but outcome assessors were masked. The primary outcome was depression symptoms according to the PHQ-9 at 12 months. We analysed all those who were randomly allocated and had complete data (modified intention to treat [mITT]) and also all those who were randomly allocated, had complete data, and received at least eight treatment sessions (per protocol [PP]). We analysed safety in the mITT population. The non-inferiority margin was 1·9 PHQ-9 points. This trial is registered with the ISCRTN registry, number ISRCTN27473954. Findings Between Sept 26, 2012, and April 3, 2014, we randomly allocated 221 (50%) participants to BA and 219 (50%) to CBT. 175 (79%) participants were assessable for the primary outcome in the mITT population in the BA group compared with 189 (86%) in the CBT group, whereas 135 (61%) were assessable in the PP population in the BA group compared with 151 (69%) in the CBT group. BA was non-inferior to CBT (mITT: CBT 8·4 PHQ-9 points [SD 7·5], BA 8·4 PHQ-9 points [7·0], mean difference 0·1 PHQ-9 points [95% CI −1·3 to 1·5], p=0·89; PP: CBT 7·9 PHQ-9 points [7·3]; BA 7·8 [6·5], mean difference 0·0 PHQ-9 points [–1·5 to 1·6], p=0·99). Two (1%) non-trial-related deaths (one [1%] multidrug toxicity in the BA group and one [1%] cancer in the CBT group) and 15 depression-related, but not treatment-related, serious adverse events (three in the BA group and 12 in the CBT group) occurred in three [2%] participants in the BA group (two [1%] patients who overdosed and one [1%] who self-harmed) and eight (4%) participants in the CBT group (seven [4%] who overdosed and one [1%] who self-harmed). Interpretation We found that BA, a simpler psychological treatment than CBT, can be delivered by junior mental health workers with less intensive and costly training, with no lesser effect than CBT. Effective psychological therapy for depression can be delivered without the need for costly and highly trained professionals. Funding National Institute for Health Research.


Introduction
Clinical depression is a common and debilitating mental health disorder, being the second largest cause of global disability. 1 Globally, the eff ect of depression on aggregate economic output is predicted to be US$5·36 trillion between 2011 and 2030. 2 Reduction of these substantial costs is a key objective for low-income, middle-income, and high-income countries alike.
Antidepressant medication and cognitive behavioural therapy (CBT) have the most clinical evidence. However, although antidepressant medications are cheap, their use is limited by side-eff ects, poor patient adherence, and discontinuation relapse risk. CBT is as eff ective as are antidepressants 3 and provides longterm protection against relapse, but it is complex and its eff ectiveness is dependent on the skills of psychological therapists, who are expensive to train and employ. For low-income and middle-income countries in particular, the need for an extensive professional infrastructure of such therapists limits access to CBT.
Globally, health services require eff ective, easily implemented, and cost-eff ective psychological treatments for depression that can be delivered by less specialist health workers than are needed at present to close a treatment gap that can be as much as 80-90% in some low-income countries. 4 One potential alternative, behavioural activation (BA), is a simple psychological treatment for depression. It might be easy and quick to train junior mental health workers (MHWs) in BA who have no professional training in psychological therapies. 5 However, this method is only appropriate if BA delivered in this way is as eff ective as and more cost-eff ective than is CBT.
Although BA compares favourably with CBT in systematic reviews, 6,7 the existing evidence is insuffi ciently robust to establish comparability. 8 Authors of a Cochrane review 7 called for more quality studies than have been done so far and the UK National Institute for Health and Care Excellence (NICE) regarded the international evidence as insuffi cient to recommend BA for fi rst-line treatment in clinical guidelines, 8 instead recommending a large noninferiority study: "to establish whether behavioural activation is an eff ective alternative to CBT". 8 Given these recommendations, we hypothesised that BA is non-inferior to CBT for depression treatment response in adults with depression and that BA is cost-eff ective compared with CBT.

Study design and participants
In this randomised, controlled, open-label, non-inferiority trial (the Cost and Outcome of Behavioural Activation versus Cognitive Behaviour Therapy for Depression [COBRA] trial), we recruited participants from primary care and psychological therapy services in Devon, Durham, and Leeds (UK). Eligible participants were adults aged 18 years or older who met diagnostic criteria for major depressive disorder assessed by researchers using a standard clinical interview (Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition [SCID] 9 ). We excluded people at interview who were receiving psychological therapy, were alcohol or drug dependent, were acutely suicidal or had attempted suicide in the previous 2 months, or were cognitively impaired, or who had bipolar disorder or psychosis or psychotic symptoms.
We recruited participants by searching the electronic case records of general practices and psychological therapy services for patients with depression, identifying potential participants from depression classifi cation codes. Practices or services contacted patients to seek permission for researcher contact. The research team interviewed those that responded, provided detailed information on the study, took informed written consent, and assessed people for eligibility. The UK South West Research Ethics Committee gave national approval for the study (NRES/07/ H1208/60). The protocol has been published previously. 10

Randomisation and masking
After eligibility was established, consent agreed, and baseline data collected, we randomly allocated participants

Research in context
Evidence before this study Authors of published systematic reviews, including a Cochrane review, have commented on the limitations of existing evidence for the eff ectiveness of behavioural activation (BA) for depression compared with cognitive behavioural therapy (CBT) and the scarcity of cost-eff ectiveness data, with the existing evidence insuffi ciently robust to establish comparability. Authors of the Cochrane review called for studies that improve the quality of evidence. Our pretrial evidence took published review fi ndings from the UK National Institute for Health and Care Excellence (NICE), who reported no diff erence in treatment outcome between BA and CBT immediately after treatment (Hedges' g 0·139 [95% CI −0·400 to 0·122]; p=0·296) and subsequent follow-up (0·135 [−0·456 to 0·186]; p=0·409). The authors of NICE's review regarded the existing international evidence as insuffi cient to recommend BA for fi rst-line treatment in clinical guidelines for depression.

Added value of this study
This trial addresses these research recommendations and is, to our knowledge, the only high-quality, fully powered non-inferiority and cost-eff ectiveness study addressing both the eff ects and costs of BA compared with CBT for depression. When we combine the data from our study with data from other international studies in the meta-analysis done by NICE, our data reduce the 95% CIs around the eff ect size for depression symptoms immediately after treatment (Hedges' g 0·054 [95% CI −0·214 to 0·107]; p=0·514) and at follow-up (0·059 [−0·234 to 0·115]; p=0·503) and unequivocally show both non-inferiority of BA compared with CBT and that BA is more cost-effective than is CBT against commonly applied decision maker willingness-to-pay thresholds.

Implications of all the available evidence
Junior mental health workers with no professional training in psychological therapies can deliver behavioural activation, a simple psychological treatment, with no lesser eff ect than CBT has and at less cost. Eff ective psychological therapy for depression can be delivered without the need for costly and highly trained professionals.
(1:1) to BA or CBT using computer-generated allocation, stratifi ed by depression severity according to the Patient Health Questionnaire 9 (PHQ-9) 11 (<19 vs ≥19), antidepressant use (taking antidepressants or not), and recruitment site (Devon, Durham, or Leeds). A computerbased system allocated the fi rst 20 participants to each group on a truly random basis. For subsequent participants, allocation was minimised to maximise the likelihood of balance in stratifi cation variables across the two study groups. The registered Peninsula Clinical Trials Unit (Plymouth University, Plymouth, UK) allocated participants remotely after baseline data entry to ensure allocation concealment. Treatment was given open label, but outcome assessors were masked to participants' allocations. Concealment was ensured by use of an externally administered password-protected trial website with retention of a stochastic element to the minimisation algorithm. We recorded instances when outcome assessors were unmasked during interviews if participants informed them of their allocation.

Procedures
We developed our clinical protocols in line with published treatment protocols, 12,13 including those from our own trials, 14,15 advice from international collaborators, and NICE recommendations 8 for duration and frequency of BA and CBT. Full-time National Health Service (NHS) MHWs and therapists worked half of their working week for COBRA (with the other half worked as normal) and followed written manuals to deliver a maximum of 20 sessions over 16 weeks, with the option of four additional booster sessions if the patients wanted them. 8 Treatment included core and supplementary techniques appropriate to the BA or CBT protocol to be used as clinically indicated; for example, behavioural or cognitive strategies for management of anxiety. All core components of both treatments were delivered by session eight, which we considered to represent a minimally suffi cient dose of therapy (appendix). Sessions were face to face, lasting for 60 min. BA and CBT experts on the trial team trained MHWs and therapists for 5 days in either BA or CBT. MHWs and therapists were assessed for competence at the end of training with use of standardised quality criteria instruments consistent with the relevant treatment: either the Quality of Behavioral Activation Scale (Dimidjian S, University of Colorado, personal communication) or the Revised Cognitive Therapy Scale for CBT. 16 Further training was given if competency was not demonstrated. MHWs and therapists received 60 min of clinical supervision fortnightly from NHS psychological therapists clinically experienced in BA or CBT, overseen by trial team experts.
Junior MHWs-graduates trained to deliver guided self-help interventions, but with neither professional mental health qualifi cations nor formal training in psychological therapies-delivered an individually tailored programme re-engaging participants with positive environmental stimuli and developing depression management strategies. Participants were encouraged to increase their contact with individually specifi ed positive situations and reduce their avoidance of other situations. Specifi c BA techniques included identifi cation of depressed behaviours, analysis of the triggers and consequences of depressed behaviours, monitoring of activities, development of alternative goal-orientated behaviours, scheduling of activities, and development of alternative behavioural responses to rumination.
Professional or equivalently qualifi ed psychotherapists, accredited as CBT therapists with the British Association of Behavioural and Cognitive Psychotherapy, with a postgraduate diploma in CBT, delivered a personalised treatment programme based on an assessment of how participants' beliefs lead to emotional distress and ineff ectual coping. Participants used cognitive and behavioural exercises to specifi cally test the accuracy of those beliefs by identifying and modifying negative thoughts and beliefs that give rise to them. Specifi c techniques included participants monitoring moods and activities, planning of exercises to test negative beliefs, and thought records to identify and examine the accuracy of negative automatic thoughts and underlying beliefs. We did follow-up assessments 6 months, 12 months, and 18 months after randomisation.
We assessed the quality of and adherence to treatment using audiotapes and written records of therapy sessions. Independent experts in both treatments rated a random (with use of a computer-generated random number sequence) sample of tapes, stratifi ed by therapist, therapy session, and intervention, for competence using the Revised Cognitive Therapy Scale 16 for CBT (range 0-72) and the Quality of Behavioral Activation Scale (Dimidjian S, University of Colorado, personal communication) for BA (range 0-96). All therapists recorded the specifi c therapeutic techniques that they had used for each session on a checklist.

Outcomes
The primary outcome was self-reported depression severity (PHQ-9 score 11 ) at 12 months. Secondary outcomes were PHQ-9 score at 6 months and 18 months and Diagnostic and Statistical Manual of Mental Disorders IV major depressive and anxiety disorder status and number of depression-free days between follow-ups (SCID), 9 anxiety (Generalized Anxiety Disorder 7), 17 and health-related quality of life (36-Item Short Form Survey) 18 at 6 months, 12 months, and 18 months. For adverse events, we recorded deaths from whatever cause and all self-harm and suicide attempts. The independent Data Management Committee reviewed all adverse events and made relevant trial conduct recommendations.

Statistical analysis
Previous research has suggested that non-inferiority margins should be half of the mean controlled eff ect size See Online for appendix from historical trials. 19 Accordingly, we estimated the non-inferiority margin for the primary outcome using meta-analysis data from trials of BA 14 for which BA was superior to controls by a mean of 0·7 SD units (95% CI 0·39-1) or 3·8 PHQ-9 score units (2·1-5·4). Therefore, our non-inferiority margin was 1·9 PHQ-9 points (ie, 0·5 × 3·8). We infl ated our sample size by 20% for participant follow-up attrition. We planned to recruit   220 participants per arm to detect a between-group noninferiority margin of 1·9 PHQ-9 points with a one-sided 2·5% α. Furthermore, although fi ndings from a trial 20 of CBT have shown little eff ect of outcome clustering by therapists, the presence of a small therapist clustering eff ect (ie, an intracluster correlation coeffi cient of 0·01) would still provide the same power. We did all analyses using a statistical analysis plan prepared in the fi rst 6 months of the trial, agreed with the Trial Management Group, Trial Steering Committee, and Data Management Committee. We assessed betweengroup equivalence of baseline characteristics and outcomes descriptively and did a descriptive analysis of baseline characteristics by recruitment method.
We compared observed primary and secondary outcomes between groups 12 months after randomisation using linear regression models adjusted for baseline outcome values and stratifi cation variables. We did modifi ed intention-to-treat (mITT) and per-protocol (PP) analyses, as security of inference depends on both PP and intention-to-treat analyses showing non-inferiority. 21 PP analysis provides some protection for any theoretical increase in the risk of type I error (erroneously concluding non-inferiority). Our mITT population comprises all patients according to and included in random allocation with complete data. We defi ned the PP population as participants meeting the mITT defi nition and receiving at least eight treatment sessions (representing a minimally suffi cient dose of therapy). We analysed safety in the mITT population. We did sensitivity analyses for our primary outcome and for diff erent defi nitions of PP (eight, 12, 16, and 20 treatment sessions) to check security of inference of non-inferiority.
We accepted non-inferiority of BA to CBT (in a 0·025 level test) if the lower bound of the two-sided 95% CI (equivalent to the upper bound of one-sided 97·5% CI) was within the non-inferiority margin of -1·9 PHQ-9 points. We checked for non-equivalence of the primary outcome at all follow-up points using the same approach.
We did secondary analyses to compare groups at followup across 6 months, 12 months, and 18 months using hierarchical linear regression. To ease clinical interpretation, we calculated proportions of recovery (participants with PHQ-9 scores of ≤9) and response (50% reduction from baseline PHQ-9 scores). We ran sensitivity analyses to assess the likely eff ect of missing data using multiple imputation models. We did imputation by treatment group using chained equations to create 20 complete datasets under the assumption that data were missing at random. 22 Imputation models included covariates as defi ned for the primary analysis model and auxiliary variables that were predictive of outcomes. After analysis, we combined the eff ect estimates from the imputed datasets using Rubin's rule. 23 For economic analyses, we took the UK NHS and personal social services perspective consistent with the NICE reference case, 24 also examining a wide societal perspective, adding productivity losses due to time off work in a sensitivity analysis. We collected participants' use of BA and CBT from clinical records, with additional resource information (eg, training, supervision, and other non-face-toface activities) from therapists and trainers. We used the Adult Service Use Schedule to measure other health and social care services used, including psychotropic medications. We measured productivity losses using the absenteeism and presenteeism questions from the BA (n=221) CBT (n= 219) All (n=440)

Trial characteristics
Method of recruitment Health and Work Performance Questionnaire. 25 We calculated eff ectiveness in terms of quality-adjusted lifeyears (QALYs) using the EuroQol-5D-3L measure of health-related quality of life. 26 We assigned health states from the EuroQol-5D-3L measure a utility score using responses from a representative sample of adults in the UK. 27 We calculated QALYs as the area under the curve defi ned by the utility values at baseline and each followup, assuming that utility score changes over time followed a linear path. We compared the costs and cost-eff ectiveness of treatments at 18 months to capture the economic eff ect of events like relapse with unit costs from the 2013-14 fi nancial year. 28,29 We discounted costs and QALYs in year 2 at 3·5%. 24 We used complete case analysis with missing data explored in a sensitivity analysis using multiple imputation with chained equations. We calculated the cost of each treatment using a microcosting (bottom-up) approach. 30 31 weighted to account for time spent on non-patient-facing activities. We applied nationally applicable unit costs for other health and social care services.
We assessed cost-eff ectiveness in terms of QALYs using the net benefi t approach. 32 We analysed diff erences in mean cost per participant at 18 months using parametric t tests, with the validity of results confi rmed using biascorrected, non-parametric bootstrapping. 33 We calculated incremental cost-eff ectiveness ratios and constructed costeff ectiveness planes using 1000 bootstrapped resamples from regression models of total cost and outcome by treatment group. We used these bootstrapped replications to calculate the probability that each of the treatments is the optimal choice for diff erent values a decision maker is willing to pay for a unit improvement in outcome, representing uncertainty around the cost and eff ectiveness estimates, with cost-eff ectiveness acceptability curves illustrating the probability that BA is cost-eff ective compared with CBT, dependent on willingness to pay per QALY. 34 We controlled for stratifi cation variables and baseline values of the variables of interest, truncating data to exclude infl uential outliers-ie, cases with total costs in the 99th percentile that make a signifi cant diff erence to the results. We did all analyses using Stata version 14.1. This trial is registered with the ISCRTN registry, number ISRCTN27473954.

Role of the funding source
The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. DAR, RST, FCW, SB, SR, and BB had full access to all the data in the study and DAR had fi nal responsibility for the decision to submit for publication.

Results
Between Sept 26, 2012, and April 3, 2014, we recruited 440 participants, randomly allocating 221 (50%) to the BA group and 219 (50%) to the CBT group (fi gure 1). 175 (79%) participants were assessable for the primary outcome in the mITT population in the BA group compared with 189 (86%) in the CBT group, whereas 135 (61%) were assessable in the PP population in the BA group compared with 151 (69%) in the CBT group. We noted no evidence of a diff erence in patient characteristics between recruitment methods (appendix). Patient-level and trial-level characteristics at baseline were well balanced between groups (table 1). PHQ-9 score at baseline was negatively skewed, with a high proportion of participants scoring towards the upper end of the distribution (data not shown), but scores were similar between groups (table 2).
Ten MHWs provided BA (median 22 participants each [IQR [19][20][21][22][23][24][25]) and 12 therapists provided CBT (21 [13][14][15][16][17][18][19][20][21][22][23]). MHWs had a mean of 18 months mental health experience (SD 11) and CBT therapists had a mean of 22 months post-CBT qualifi cation (24). We removed one CBT therapist from the trial in the early stages who did not meet acceptable competency. Participants received a mean of 11·5 BA sessions (7·8) or 12·5 CBT sessions (7·8   (table 2). The non-inferiority of BA to CBT was accepted for both the mITT and PP populations as the lower bound of the 95% CI (one-sided 97·5% CI) of the between-group mean diff erence lies within the non-inferiority margin of -1·9 PHQ-9 points (appendix). Although we initially planned to include therapist as a random-eff ects variable, given the low levels of observed clustering, we parsimoniously fi tted our models without therapist as a variable. We checked for no inference diff erence with and without inclusion of a random-eff ects therapist term. We ruled out superiority of CBT to BA as the lower bound of the 95% CI included zero for the mITT and PP populations. The inference of non-inferiority was robust to sensitivity analysis across diff erent PP defi nitions. We found no evidence of a signifi cant between-group treatment interaction across the mITT or PP populations with the primary outcome at 12 months as stratifi ed by depression severity, antidepressants use, and recruitment site (appendix).
We found that BA was not diff erent from CBT in anxiety (Generalized Anxiety Disorder 7), depression status, and depression-free days and anxiety diagnoses (SCID) for either the mITT or PP populations using observed or imputed data at 12 months (table 2). Because of substantial missing 36-Item Short Form Survey data at baseline, we analysed these data adjusted for stratifi cation variables only. We found no diff erence in numbers of participants with at least one anxiety diagnosis: BA 43 (28%) of 153; CBT 43 (27%) of 161 (mITT population; χ² 0·08; p=0·78).
Between 61% and 70% of mITT and PP participants in both groups met criteria for recovery from depression or response to treatment at 12 months, with no diff erences in the proportions of patients in each group who recovered or responded (  between the CBT and BA groups over the period of the trial, as indicated by a non-signifi cant time-by-treatment eff ect interaction, for both the mITT and PP populations (appendix). We found a small, negligible clustering of primary and secondary outcome scores at follow-up across therapists overall and within BA and CBT groups (intracluster correlation coeffi cient ≤0·04). Two (1%) non-trial-related deaths (one [1%] multidrug toxicity in the BA group and one [1%] cancer in the CBT group) and 15 depression-related, but not treatmentrelated, serious adverse events (three in the BA group and 12 in the CBT group) occurred in three [2%] participants in the BA group (two [1%] patients who overdosed and one [1%] who self-harmed) and eight (4%) participants in the CBT group (seven [4%] who overdosed and one [1%] who self-harmed). Of the 440 participants recruited, 76 (17%) had missing primary outcome data at 12 month follow-up. The proportion of missing PHQ-9 data was higher in the BA than in the CBT group (46 [21%] vs 30 [14%]; odds ratio 1·6 [95% CI 1·0-2·7]; p=0·05). Imputation of data for primary and secondary outcomes at 12 months showed that in accordance with the observed data analysis, no diff erence existed between groups (tables 2, 3), supporting our conclusion of noninferiority. The odds of missing PHQ-9 data were higher for patients with increased baseline severity of depression (PHQ ≥19, odds ratio 1·6 [95% CI 1·0-2·6]; p=0·05) and increasing age (in years) was associated with lower odds of missing PHQ-9 data (odds ratio 0·97 [0·96-0·99]; p=0·01). We found no evidence of an association between missingness and any other baseline characteristic (data not shown). Outcome assessors reported having been unmasked for 16 (4%) participants (fi ve [2%] in the BA group and 11 [5%] in the CBT group; due to participants informing assessors of their treatment allocation).
For economic analyses, at 18 months, full service use data was available for 159 (90%) of 176 participants in the BA group and 168 (93%) of 180 participants in the CBT group. We found a signifi cant diff erence in mean intervention costs between the two groups, but no diff erences in other categories of cost or in total cost (table 4). Mean health state utility scores according to EuroQoL-5D-3L were slightly higher in the BA group than in the CBT group across the entire follow-up period, with resultant QALYs also higher for BA, but the QALY diff erence was not signifi cant. Costs were lower and QALY outcomes better in the BA group than in the CBT group, generating an incremental cost-eff ectiveness ratio of -£6865. The scatterplot of bootstrapped cost and eff ectiveness pairs for BA versus CBT illustrates dominance of BA over CBT, with the point estimate and two-thirds of scatter points falling in the southeast quadrant of the cost-eff ectiveness plane, where BA replications are cheaper and more eff ective than are CBT ones (fi gure 2). The cost-eff ectiveness acceptability curve (appendix) showing the probability of BA being costeff ective compared with CBT does not fall below 75% and    24 to pay £20 000-30 000 per QALY. In all sensitivity analyses, including complementary therapies and productivity losses, as well as analyses taking narrow intervention and mental health service perspectives, BA was signifi cantly less costly than was CBT, so BA continues to have a higher probability of being cost-eff ective than does CBT at the NICE threshold (appendix). 24 Imputation of missing data increased the diff erence in total cost (BA £1841·67; CBT £2282·40; diff erence -£440·73 [95% CI -1007·71 to 126·26]; p=0·13), but reduced the diff erence in QALYs (BA 1·22; CBT 1·19; diff erence 0·03 [-0·06 to 0·11]; p=0·55), increasing the incremental cost-eff ectiveness ratio to -£16 951. The cost-eff ectiveness acceptability curve for the missing data analysis again supported the likelihood that BA is cost-eff ective compared with CBT (appendix).

Discussion
We found that BA for depression is not inferior to CBT in terms of reduction of depression symptoms and is more cost-eff ective than is CBT against commonly applied decision maker willingness to pay thresholds. We observed our results using both mITT and PP analyses, using a conservative non-inferiority margin. Our economic analyses were driven by the lower costs of the MHWs who delivered BA compared with the more experienced psychological therapists who routinely deliver CBT. Our study results therefore substantiate the hypothesis that BA is as eff ective as is CBT and that its simplicity renders BA suitable for delivery by junior MHWs with no professional training in psychological therapies. 5 This trial is the largest trial of BA to date and is one of the largest trials of psychological treatments for depression. We followed up participants for 18 months and our economic analysis is one of few in this fi eld. Therapists and MHWs working in three diff erent routine UK care settings delivered treatment, providing evidence of potential generalisability. We assessed therapy quality using independent raters and ensured that treatment in both arms was delivered to the standard recommended guidelines. Our levels of attrition and outcome loss to follow-up were low at 12 months and 18 months, similar to other trials in this area, but are still a limitation. Although participants in the per-protocol population attended similar numbers of sessions to those in other CBT trials, 15 35% of participants chose to not even attend a minimal number of sessions, a problem well known to routine psychological therapy services. This pragmatic trial done in routine environments means that we were unable to quantify or control for the contribution of antidepressant medicines to outcomes. However, most participants who were taking medication had been doing so for a considerable time before entering the trial, making it unlikely that our results were driven by pharmacological treatment. Given the nature of the intervention and comparator, we could not mask patients or the mental health workers or therapists who were delivering the interventions to treatment allocation, but we used self-reported outcome measures and robust outcome assessor-masking procedures to reduce researcher unmasking to less than 5%. Missing data for the primary outcome measure was substantial. However, our between-group inferences were robust to data imputation.
Our fi ndings could have substantial implications for the scalability of psychological treatment for depression internationally 4 given the greater availability and ease with which a BA workforce could be trained than could a CBT workforce. For many years, CBT has been the foremost psychological therapy recommended by therapists, researchers, and policy makers. Our results challenge this dominance. Although more work needs to be done than has been done so far to fi nd ways to eff ectively treat the 20-23% of patients whose depression was unchanged by BA or CBT, our fi ndings suggest that BA should be a front-line treatment for depression, with substantial potential to improve reach and access to psychological therapy globally.
Our results in both groups compare favourably with a meta-analysis 3 of the eff ects of CBT that estimate proportions of patients with remissions of around 50%. Our cost-eff ectiveness analyses show the high probability that BA is cost-eff ective and aff ordable compared with CBT at standard willingness to pay thresholds. Our most striking fi nding is that BA leads to similar clinical outcomes for patients with depression, but at a fi nancial saving to clinical providers of 21% compared with the costs of provision of CBT, with no compensatory use of other health-care services by patients.
Driving these savings is the fact that BA can be delivered by inexperienced MHWs with no professional training in psychological therapies, with no lesser eff ect than that of more highly trained and experienced psychological   therapists giving patients CBT. Although many obstacles exist to successful dissemination in addition to training of MHWs, our fi ndings suggest that health services globally could reduce the need for costly professional training and infrastructure, reduce waiting times, and increase access to psychological therapies. 4 Our fi ndings have substantial implications given the increasing global pressure for cost containment across health systems in high-income countries and the need to develop accessible, scalable interventions in low-income and middle-income countries. Such countries might choose to investigate the training and employment of junior workers over expensive groups of psychological professionals. Our results, therefore, off er hope to many societies, cultures, and communities worldwide, rich and poor, struggling with the eff ect of depression on the health of their people and economies.

Contributors
DAR, DE, DM, RST, SB, PAF, SG, WK, HO'M, ERW, and KAW designed the study and were responsible for its conduct. SR, EF, and KF were responsible for study and data collection management. RST, SB, FCW, and BB did data analysis. SDH and NR provided expert advice on clinical and patient and public involvement. All authors contributed to writing and editing of the manuscript.