Effectiveness and costs associated with a lay counselor-delivered, brief problem-solving mental health intervention for adolescents in urban, low-income schools in India: 12-month outcomes of a randomized controlled trial.

BACKGROUND: Psychosocial interventions for adolescent mental health problems are effective, but evidence on their longer-term outcomes is scarce, especially in low-resource settings. We report on the 12-month sustained effectiveness and costs of scaling up a lay counselor-delivered, transdiagnostic problem-solving intervention for common adolescent mental health problems in low-income schools in New Delhi, India. METHODS AND
FINDINGS: Participants in the original trial were 250 school-going adolescents (mean [M] age = 15.61 years, standard deviation [SD] = 1.68), including 174 (69.6%) who identified as male. Participants were recruited from 6 government schools over a period of 4 months (August 20 to December 14, 2018) and were selected on the basis of elevated mental health symptoms and distress/functional impairment. A 2-arm, randomized controlled trial design was used to examine the effectiveness of a lay counselor-delivered, problem-solving intervention (4 to 5 sessions over 3 weeks) with supporting printed booklets (intervention arm) in comparison with problem solving delivered via printed booklets alone (control arm), at the original endpoints of 6 and 12 weeks. The protocol was modified, as per the recommendation of the Trial Steering Committee, to include a post hoc extension of the follow-up period to 12 months. Primary outcomes were adolescent-reported psychosocial problems (Youth Top Problems [YTP]) and mental health symptoms (Strengths and Difficulties Questionnaire [SDQ] Total Difficulties scale). Other self-reported outcomes included SDQ subscales, perceived stress, well-being, and remission. The sustained effects of the intervention were estimated at the 12-month endpoint and over 12 months (the latter assumed a constant effect across 3 follow-up points) using a linear mixed model for repeated measures and involving complete case analysis. Sensitivity analyses examined the effect of missing data using multiple imputations. Costs were estimated for delivering the intervention during the trial and from modeling a scale-up scenario, using a retrospective ingredients approach. Out of the 250 original trial participants, 176 (70.4%) adolescents participated in the 12-month follow-up assessment. One adverse event was identified during follow-up and deemed unrelated to the intervention. Evidence was found for intervention effects on both SDQ Total Difficulties and YTP at 12 months (YTP: adjusted mean difference [AMD] = -0.75, 95% confidence interval [CI] = -1.47, -0.03, p = 0.04; SDQ Total Difficulties: AMD = -1.73, 95% CI = -3.47, 0.02, p = 0.05), with stronger effects over 12 months (YTP: AMD = -0.98, 95% CI = -1.51, -0.45, p < 0.001; SDQ Total Difficulties: AMD = -1.23, 95% CI = -2.37, -0.09; p = 0.03). There was also evidence for intervention effects on internalizing symptoms, impairment, perceived stress, and well-being over 12 months. The intervention effect was stable for most outcomes on sensitivity analyses adjusting for missing data; however, for SDQ Total Difficulties and impairment, the effect was slightly attenuated. The per-student cost of delivering the intervention during the trial was $3 United States dollars (USD; or $158 USD per case) and for scaling up the intervention in the modeled scenario was $4 USD (or $23 USD per case). The scaling up cost accounted for 0.4% of the per-student school budget in New Delhi. The main limitations of the study's methodology were the lack of sample size calculations powered for 12-month follow-up and the absence of cost-effectiveness analyses using the primary outcomes.
CONCLUSIONS: In this study, we observed that a lay counselor-delivered, brief transdiagnostic problem-solving intervention had sustained effects on psychosocial problems and mental health symptoms over the 12-month follow-up period. Scaling up this resource-efficient intervention is an affordable policy goal for improving adolescents' access to mental health care in low-resource settings. The findings need to be interpreted with caution, as this study was a post hoc extension, and thus, the sample size calculations did not take into account the relatively high attrition rate observed during the long-term follow-up. TRIAL REGISTRATION: ClinicalTrials.gov NCT03630471.

Introduction

Psychosocial interventions have been advocated to address the growing burden of adolescent mental health problems globally [1,2]. Although there is a large body of research on the short-term outcomes of psychosocial interventions for adolescents with common mental health problems, evidence for their sustained effectiveness is relatively scarce. A systematic review of youth psychotherapy trials for internalizing and externalizing problems found that only one-third of adolescent-focused trials (52 out of 155) reported longer-term follow-up outcomes, averaging 11 months in duration. The pooled effect size in these follow-up studies was 0.28, indicating a small sustained effect of psychotherapies across diverse problem types [3]. Examination of trial characteristics indicated that most intervention protocols were time intensive (16.5 sessions on average) and involved delivery by specialists. Relatively few studies (n = 13) examined interventions targeting comorbid presentations, which represent the majority of “real-world” case mix [3]. Moreover, no longer-term follow-up studies were identified from low- and middle-income countries (LMICs). These regions account for 90% of the global adolescent population [4], yet command a small fraction of mental health service resources and associated research infrastructure worldwide.

The PRemIum for aDolEscents (PRIDE) research program was conceived to meet the need for contextually sensitive, evidence-based interventions that target common adolescent mental health problems in India and LMICs more generally. The PRIDE intervention model is situated in secondary schools and built around a transdiagnostic stepped care architecture, which comprises a brief problem-solving intervention (“Step 1”) and a higher-intensity personalized psychological treatment (“Step 2”) for adolescents who do not respond to the first-line intervention [5,6]. The first step was evaluated in a randomized controlled trial in New Delhi, India, where lay counselors functioned as the delivery agent in the intervention arm [7]. When compared against a control condition consisting of printed problem-solving booklets without counselor input, there was evidence of an intervention effect on one of the 2 primary outcomes (self-reported psychosocial problems) at 6 weeks and 12 weeks, but not on the other primary outcome (self-reported mental health symptoms). However, symptom trajectories suggested that differences between the intervention and control arms were widening at 12 weeks. It was speculated that a longer follow-up period might reveal a distal effect on mental health, as newly learned coping skills were practiced and consolidated over time [7]. The current study, therefore, examined trajectories of trial outcomes for both arms over 12 months following randomization. We also estimated the incremental costs of setting up and delivering the intervention during the trial and used these estimates to enumerate the costs of scaling up the intervention across public schools in New Delhi.

Methods

Study design and participants

The original trial protocol (ClinicalTrials.gov NCT03630471) was restricted to short-term endpoints at 6 and 12 weeks. Based on recommendations of the Trial Steering Committee in a meeting on May 20, 2019, a protocol modification was made to conduct an additional follow-up at 12 months after randomization. Ethical approval for the protocol modification was obtained from the Institutional Review Boards of Harvard Medical School (sponsor) and Sangath (implementing organization in India). The protocol and analysis plan for the 12-month follow-up study have been provided as a Supporting information file (S1 Protocol). This study is reported as per the Consolidated Standards of Reporting Trials (CONSORT) guideline (S1 Checklist).

Detailed descriptions of participant recruitment, trial design, and conduct are available elsewhere [8]. The original recruitment was conducted from August 20 to December 4, 2018 in 6 government-run schools (3 all-boys schools, 2 all-girls schools, and 1 coeducational school), catering to low-income communities in New Delhi. Referrals into the trial were generated through whole-school and classroom-level sensitization activities that were intended to (i) raise awareness of available mental health support, and (ii) address factors such as low mental health literacy and confidentiality concerns that might otherwise limit the demand for school mental health services. The participants were adolescents in grades 9 to 12 who scored at or above the locally validated “borderline” cutoff scores of 19 for boys and 20 for girls on the Strengths and Difficulties Questionnaire (SDQ) Total Difficulties scale [9,10], reported an “abnormal” score of 2 or more on the SDQ Impact scale [9], and indicated persistent problems of more than 1 month on the SDQ Chronicity index. Adolescents were excluded if they needed urgent medical attention from a specialist, were receiving another mental health intervention, had taken part in previous PRIDE studies, demonstrated receptive/expressive language difficulties that affected their ability to participate fully in trial procedures, or declined consent for research participation.

The estimated sample size for the originally stipulated primary endpoint assumed loss to follow-up of 15% at 6 weeks (based on piloting) and a 1:1 allocation ratio to provide over 90% power to detect an effect size of 0.5 (and 80% power to detect an effect size of 0.4) at the α level of 0.025 for each of the 2 primary outcomes [8]. Based on these estimates, a total of 251 participants were initially enrolled, and 125 were randomly allocated to each arm after accounting for one participant in the intervention arm who withdrew consent after randomization (Fig 1). Participants were randomized using the randomization list generated by the senior trial statistician and concealed using sequentially numbered sealed envelopes [8]. Baseline demographic and clinical characteristic for the trial participants in the 2 arms are given in Table 1. For the 12-month follow-up, we approached all 250 participants who opted to be contacted about future research during the original consent procedure. One week prior to the 12-month follow-up date, attempts were made to establish telephone contact with the original trial participants in order to gauge their interest and availability for participation in the long-term follow-up study. Contact was also made with parents/guardians (“caregivers”) for adolescents aged less than 18 years. If agreed, a face-to-face appointment was scheduled at the individual’s home or in school, as per their preference. Written assent (or consent for individuals aged 18 years or older) was obtained from adolescents, followed by consent from a caregiver for adolescents aged less than 18 years. Those providing assent and consent completed the self-reported outcome measures in Hindi on a digital tablet device at a convenient location. In contrast to 6- and 12-week follow-up assessments, no follow-up data were collected from caregivers at 12 months. The 12-month follow-up assessments were completed between August 26, 2019 and January 2, 2020.

Fig 1

CONSORT diagram for trial participants.

CONSORT, Consolidated Standards of Reporting Trials.

Table 1

Baseline characteristics of participants.

	Control arm (n = 125)	Intervention arm (n = 125)
Age (in years), M (SD)	15.59 (1.68)	15.63 (1.68)
Sex, n (%)
Females	38 (30.4%)	38 (30.4%)
Males	87 (69.6%)	87 (69.6%)
School grade, n (%)
9th class	52 (41.6%)	58 (46.4%)
10th class	36 (28.8%)	31 (24.8%)
11th class	8 (6.4%)	8 (6.4%)
12th class	29 (23.2%)	28 (22.4%)
Primary caregiver age (in years)2, M (SD)	38.30 (8.16)	35.47 (8.81)
Primary caregiver education2, n (%)
No formal education	25 (21.2%)	25 (21.2%)
Completed primary	3 (2.5%)	2 (1.7%)
Completed secondary school	56 (47.5%)	57 (48.3%)
Completed higher secondary and above	24 (20.3%)	27 (22.9%)
Not available	10 (8.5%)	7 (5.9%)
Primary caregiver occupation2, n (%)
Not employed outside home	45 (38.1%)	46 (39.0%)
Manual	59 (50.0%)	52 (44.1%)
Clerical	9 (7.6%)	5 (4.2%)
Professional	3 (2.5%)	4 (3.4%)
Other	2 (1.7%)	11 (9.3%)
SDQ Total Difficulties scale3, M (SD)	23.12 (3.01)	23.22 (3.31)
SDQ Impact scale3, M (SD)	5.20 (2.37)	5.38 (2.41)
SDQ Internalizing symptoms subscale3, M (SD)	11.96 (2.45)	12.09 (2.49)
SDQ Externalizing symptoms subscale3, M (SD)	11.16 (2.40)	11.14 (2.37)
SDQ Chronicity index3, n (%)
1 to 5 months	41 (33.1%)	35 (28.0%)
6 to 12 months	17 (13.7%)	22 (17.6%)
Over 1 year	66 (53.2%)	68 (54.4%)
YTP score, M (SD)	7.35 (2.06)	7.24 (2.24)
PSS-4 score, M (SD)	9.22 (2.47)	9.04 (2.49)
SWEMWBS score, M (SD)	20.92 (5.28)	20.50 (4.82)

1 Denominator for each cell is number of participants in that arm, until otherwise specified.

2 Primary caregiver data not collected for 14 participants (7 in each arm) as index adolescent aged above 18 years; denominator for each row under this head is 118.

Baseline SDQ details were missing for one participant in the control arm; see Michelson and colleagues [7].

M, mean; PSS-4, Perceived Stress Scale 4-item version; SD, standard deviation; SDQ, Strengths and Difficulties Questionnaire; SWEMWBS, Short Warwick–Edinburgh Mental Well-Being Scale; YTP, Youth Top Problems.

Intervention and control arms

Participants in the intervention arm received 4 to 5 individual face-to-face sessions of a problem-solving intervention, spread over 3 weeks, and delivered on school premises. The frequency of sessions was flexible according to the preference and availability of individual participants. Participants attended sessions during a free period at school or else were excused from a scheduled class. Problem solving was introduced and practiced using a 3-step heuristic (P-O-D, referring to “problem,” “options,” and “do it”), and participants were supported to put this into practice for one or more prioritized problems. The final session focused on consolidating and discussing use of problem-solving skills in different situations. Additionally, participants received a set of 3 illustrated booklets. These booklets contained contextually appropriate stories to explain the problem-solving steps and also included suggestions for home-based practice exercises. The booklets were provided sequentially over 3 sessions, and a poster summarizing the problem-solving steps was distributed in the final session. The intervention was delivered by 8 Hindi speaking, college graduates, who had no formal clinical qualifications or prior experience of delivering a psychological treatment. These counselors were employed by the implementing organization (Sangath) and trained through a combination of office-based learning and supervised field practice. Weekly supervision was provided in a peer group format, which was moderated by a rotating counselor and overseen by a masters- or doctoral-level psychologist. Audio-taped sessions (1 or 2 per week) were reviewed and rated on a number of key competencies with corrective feedback provided as appropriate; further details are described elsewhere [7].

Process indicators from the intervention arm indicated that a mean (M) of 4.06 (standard deviation [SD] = 1.6) sessions were delivered per participant. The average session duration was 23.27 minutes (SD = 4.3). A total of 98 (78.4%) participants completed the intervention (i.e., attended 4 or more sessions) [7]. Reasons for noncompletion for the 27 participants were documented by counselors using a checklist with prespecified categories. These included rapid resolution of the identified problems (n = 13, 48.1%), competing demands at school (n = 6, 22.2%), persistent absence from school (n = 4, 14.8%), and unspecified (n = 4, 14.8%) (Fig 1).

Control arm participants received the same printed materials as intervention arm participants, but these were handed out by a researcher. The researcher introduced the materials using a brief standardized script, lasting approximately 90 seconds, which explained their purpose and instructions for use. No further guidance or support was provided.

The original trial protocol [8] included a provision for adolescents to receive a more intensive intervention from a psychologist if they continued to experience symptoms or impairment (defined as scoring above thresholds on the SDQ Total Difficulties and/or SDQ Impact scales) at the originally stipulated 12-week endpoint. A total of 16 out of 119 adolescents with persistent difficulties took this up: 9 (16.1%) participants from the intervention arm and 7 (11.1%) participants from the control arm. These participants were not excluded from the 12-month follow-up nor from the complete case analysis. None of the other participants reported receiving any additional mental health intervention.

Outcome measures

A detailed description of the adolescent-reported outcome measures is available in the original trial protocol [8]. Two primary outcomes were assessed using self-report measures at 12 months after randomization: (i) mental health symptoms measured using the SDQ Total Difficulties score (range: 0 to 40) [9], which is derived by summing 20 items covering both internalizing and externalizing symptoms; and (ii) psychosocial problems, measured using the Youth Top Problems (YTP) score (range: 0 to 10) [11], which is an idiographic measure for which an overall score is derived by averaging individual ratings for up to 3 prioritized problems nominated by the respondent. Higher scores indicated greater severity of symptoms and psychosocial problems, respectively. As indicated in Michelson and colleagues [7], 2 primary outcomes were included in the trial as we considered that problem solving would lead to reductions in both prioritized problems (an person-centered outcome) and mental health symptoms (a standardized outcome). Self-reported outcomes were also collected for (i) distress/functional impairment (SDQ Impact scale, range: 0 to 10) [9]; (ii) internalizing symptoms (SDQ Internalizing symptoms subscale, range: 0 to 20) [9]; (iii) externalizing symptoms (SDQ Externalizing symptoms subscale, range: 0 to 20) [9]; (iv) perceived stress (Perceived Stress Scale 4-item version [PSS-4], range: 0 to 16) [12]; (v) well-being (Short Warwick–Edinburgh Mental Well-being Scale [SWEMWBS], range: 7 to 35) [13]; and (vi) proportion of remitted cases (assessed using the “crossing clinical threshold method” [14] and defined as scoring below eligibility thresholds on the SDQ Total Difficulties scale and SDQ Impact scale [8]). Additionally, we used a 4-point Likert scale to collect data on adolescents’ self-reported use of problem-solving materials (“In the past year, how often have you used the booklets received during the program?”) and problem-solving skills in the preceding 12 months (“In the past year, how often have you used the skills learned during the program?”).

Costing framework

A retrospective ingredients approach [15] was used to develop an intervention costing framework based on information obtained from the provider perspective after completing the trial. This included retrospectively mapping various activities, resources, and time allocation used in setting up and implementing the intervention during the trial, with costing based on records provided by the implementing organization, Sangath. Activities included training and recruitment of counselors, referral generation activities in schools, delivery of sessions, and counselors’ supervision. Resources covered the materials used for delivering the intervention (booklets, posters, referral forms, manuals, session and supervision record forms, voice recorders, “drop boxes” for collecting paper-based referrals, and laptops and projectors for classroom sensitization), as well as resources used for setting up consulting areas in schools (chairs, tables, and screens). Interviews with staff from Sangath were conducted to assess the percentage of staff time devoted to trial activities.

Statistical analysis

Effectiveness analysis

The statistical analysis plan was finalized before analysis (included in S1 Protocol). Analyses were done with Stata version 15.1. Baseline characteristics were split by follow-up status (i.e., completed or lost to follow-up [LTFU]). These were summarized using descriptive statistics and compared using t tests or chi-squared tests, as appropriate. Serious adverse events (SAEs) were reported as the number of individuals that incurred SAEs in each arm, using criteria described in the original trial protocol [8]. Primary outcome analyses at 12-month were conducted using complete case analysis. Analyses were adjusted for the baseline value of each outcome measure, school, and variables associated with missingness at any time point (class, age, week of enrollment, and baseline YTP M score). The intervention effects on the SDQ and YTP were analyzed using a linear mixed model for repeated measures. This included data from 6 weeks, 12 weeks, and 12 months, with time by group interaction terms to estimate the intervention effect at each time point including at the 12-month endpoint [16]. In addition, for secondary outcome analyses, the intervention effects on the SDQ, YTP, and other self-reported measures were analyzed over 12 months using similar repeated measures analysis. This analysis used data from all 3 follow-up time points (i.e., at 6 weeks, 12 weeks, and 12 months), assuming a constant effect across the time points. Thus, analysis “over 12 months” represents the estimated effect size at each time point and, more generally, throughout the 12 months follow-up period. The assumption of a constant intervention effect over time was tested using a likelihood ratio test comparing this model with one allowing the effect to vary at the 3 time points. Analogous methods using logistic mixed effects regression were used for binary outcomes. Sensitivity analysis examined the effect of missing data using multiple imputations with a linear imputation model (truncated at 0) adjusting for factors associated with missingness for the imputation (age, class, parents’ gender, baseline SDQ Total Difficulties and YTP scores, and arm allocation) with 50 imputed datasets. Sensitivity analysis was also used to examine the effect following exclusion of the 16 participants who had received an additional intervention after 12 weeks. Adjusted mean difference (AMD)/adjusted odds ratios (AORs), adjusted effect size (Cohen’s d), 95% confidence intervals (CIs), and p-values were reported for primary and secondary outcomes.

An exploratory, prespecified effect moderation analysis tested for heterogeneity of intervention effects by the following: baseline chronicity of mental health difficulties, baseline severity of mental health difficulties, YTP typology, and SDQ caseness profile [8]. Post hoc, SDQ caseness profile was removed from moderation analysis, due to the small number of observations (less than 6) in certain cells. In addition, exploratory mediation analyses were conducted to examine whether a priori factor of perceived stress, and 2 process indicators (use of problem-solving materials and use of problem-solving skills) mediated the effects of the intervention on the primary outcomes in the longer term. The 3 variables were examined in separate mediation models using the generalized structural equation models with bootstrapped CIs and the causal steps outlined by Baron and Kenny [17]. A dose–response effect in the intervention arm was examined using mixed effects regression models to assess differences in each of the primary outcomes according to the frequency of session attendance (adjusted as for the primary analyses).

Cost analysis

An incremental cost analysis was conducted for setting up and implementing the intervention during the trial. Incremental costs excluded costs that were incurred in both control and intervention arms (i.e., expenses related to printing booklets and referral generation activities). Costs related to evaluation activities and costs incurred during the schools’ summer vacation period were additionally excluded. All included costs, including those related to personnel and capital, were allocated based on proportion of use required for setup and implementation. Capital costs were annualized over their expected useful life and discounted at 3%. Costs incurred in Indian rupees were converted to United States dollars (USD) using the annual average exchange rate from the Bank of India and then inflated to 2020 USD using the available figures from the US Labor Department.

Using the same costing framework, we modeled costs of scaling up the lay counselor–delivered problem-solving intervention as a standard implementation program to 20 underresourced schools in New Delhi for an academic year. In modeling the requirements of the hypothetical scale-up, we used findings from the original trial report [7], as related to the number of working days, types of program activities (sensitization, screening, and intervention), the time required for each activity, and time needed for training and supervision. Additionally, we assumed that each counselor would work full time and engage exclusively in activities related to the problem-solving intervention over the duration of one academic year (not counting holidays). We also assumed that each counselor’s available time in school would be fully occupied by referred cases (i.e., there would be no shortfall in demand). Counselor time was costed using the actual cost of employing counselors as per government pay grades [18]. Supervisor time was costed using pay grades of the implementing organization Sangath, which was assumed to provide technical oversight in the scale-up scenario. Detailed assumptions for scale-up are provided in S1 Text of the Supporting information section. Costs related to the development of booklets and training materials were excluded as these had already been developed with the intention of being free to use. Indirect costs, such as space and furniture, were assumed to be already present within schools and were therefore excluded from the scale-up scenario.

Results

Out of 250 original trial participants, 221 (88.4%) were successfully contacted, and 177 (70.8%) consented to take part in the 12-month assessments. One participant was excluded after consent due to poor health restricting their involvement. Reasons for nonparticipation among the remaining 44 adolescents were lack of interest (n = 34, 77.2%), competing time demands at school (n = 4, 9.1%), consent declined by parents (n = 3, 6.8%), and unspecified (n = 3, 6.8%) (Fig 1). The M duration of follow-up was 12.30 months (SD = 0.28), which was on an average 11.56 months (SD = 0.29) after the planned termination of the intervention for those who were “completers” (i.e., attended 4 or 5 sessions). The LTFU rate was higher among participants in the intervention arm (44 out of 125 [35.2%]) than the control arm (30 out of 125 [24.0%]) (p = 0.07). LTFU was also higher among participants from grade 12 (25 out of 57 [43.9%]) compared to grade 11 or lower (49 out of 193 [25.4%]) (p = 0.06). Participants who were LTFU had higher YTP scores at baseline (M = 7.74, SD = 1.94) than those who completed follow-up (M = 7.11, SD = 2.20) (p = 0.03) (S1 Table). One SAE was reported involving a life-threatening event experienced by a research participant. This SAE was determined by the Data and Safety Monitoring Board to be unrelated to the intervention, and the participant’s data were retained in the analysis.

Effectiveness

At the 12-month endpoint, participants in the intervention arm scored, on average, 1.73 points lower on the SDQ Total Difficulties scale than those in the control arm (AMD = −1.73, 95% CI = −3.47, 0.02; d = 0.28, 95% CI = 0.13, 0.43; p = 0.05). The average difference in YTP score between the 2 arms at the 12-month endpoint was 0.75 points, in the direction favoring the intervention arm (AMD = −0.75, 95% CI = −1.47, −0.03; d = 0.27, 95% CI = 0.12, 0.42; p = 0.04). Assuming a constant intervention effect across 3 follow-up time points, there was strong evidence of an intervention effect on both SDQ Total Difficulties and YTP scores over 12 months (SDQ Total Difficulties: AMD = −1.23, 95% CI = −2.37, −0.09; d = 0.21, 95% CI = 0.05, 0.36; p = 0.03; YTP: AMD = −0.98; 95% CI = −1.51, −0.45; d = 0.34, 95% CI = 0.19, 0.50; p < 0.001) (Table 2). Although the intervention effects on both primary outcomes increased slightly across the time points (Fig 2), there was no statistical evidence of an interaction between group and time (SDQ Total Difficulties: p = 0.72; YTP: p = 0.51).

Table 2

Primary and secondary outcomes for 12 months follow-up.

	At 12 months		Over 12 months
Outcomes	Control arm(n = 95)1M (SD)	Intervention arm1(n = 81)M (SD)	Control arm2(n = 339)M (SD)	Intervention arm2(n = 325)M (SD)	AMD/AOR (95% CI)	Adjusted effect size (95% CI)	p-value
SDQ Total Difficulties scale	14.14 (6.04)	13.05 (6.07)	16.70 (5.99)	15.78 (5.94)	At 12 months: −1.73 (−3.47, 0.02)	0.28 (0.13, 0.43)	0.05
					Over 12 months: −1.23 (−2.37, −0.09)	0.21 (0.05, 0.36)	0.03
YTP	2.91 (2.65)	2.22 (2.33)	3.84 (2.88)	2.88 (2.63)	At 12 months: −0.75 (−1.47, −0.03)	0.27 (0.12, 0.42)	0.04
					Over 12 months: −0.98 (−1.51, −0.45)	0.34 (0.19, 0.50)	<0.001
SDQ Impact scale	1.18 (1.82)	0.90 (1.77)	1.87 (2.61)	1.43 (2.21)	Over 12 months: −0.51 (−0.93, −0.09)	0.21 (0.06, 0.36)	0.023
SDQ Internalizing subscale	6.91 (3.45)	6.25 (3.31)	8.22 (3.59)	7.72 (3.35)	Over 12 months: −0.76 (−1.42, −0.10)	0.22 (0.06, 0.37)	0.03
SDQ Externalizing subscale	7.23 (3.44)	6.80 (3.17)	8.48 (3.30)	8.06 (3.26)	Over 12 months: −0.47 (−1.09, 0.14)	0.14 (0.01, 0.30)	0.13
PSS-4	6.68 (2.45)	6.23 (2.39)	7.04 (2.59)	6.52 (2.47)	Over 12 months: −0.54 (−1.00, −0.08)	0.21 (0.06, 0.36)	0.02
SWEMWBS	25.06 (6.19)	26.33 (6.61)	23.74 (5.88)	24.69 (6.32)	Over 12 months: 1.16 (−0.07, 2.38)	0.19 (0.04, 0.34)	0.06
Remitted, n/N (%)	57/95 (60.64%)	57/81 (70.37%)	NA	NA	At 12 months: 1.47 (0.73, 2.96)	NA	0.28

1 This analysis uses outcome data from the 12-month endpoint only.

2 This analysis uses outcome data from all 3 follow-up time points (i.e., at 6 weeks, 12 weeks, and 12 months), assuming a constant effect across the points.

3 Estimated using robust standard errors due to heteroskedasticity.

AMD, adjusted mean difference; AOR, adjusted odds ratio; CI, confidence interval; M, mean; PSS-4, Perceived Stress Scale 4-item version; SDQ, Strengths and Difficulties Questionnaire; SWEMWBS, Short Warwick–Edinburgh Mental Well-Being Scale; YTP, Youth Top Problems.

Fig 2

Primary outcomes over time according to arm.

(a) M YTP score. Error bars indicate 95% CIs. (b) M SDQ Total Difficulties score. Error bars indicate 95% CIs. CI, confidence interval; M, Mean; SDQ, Strengths and Difficulties Questionnaire; YTP, Youth Top Problems.

There was evidence of a beneficial intervention effect on other secondary outcomes over 12 months. These included SDQ Impact scores (AMD = −0.51, 95% CI = −0.93, −0.09; d = 0.21, 95% CI = 0.06, 0.36; p = 0.02), SDQ Internalizing symptoms subscale score (AMD = −0.76, 95% CI = −1.42, −0.10; d = 0.22, 95% CI = 0.06, 0.37; p = 0.03), and PSS-4 score (AMD = −0.54, 95% CI = −1.00, −0.08; d = 0.21, 95% CI = 0.06, 0.36; p = 0.02). A modest effect was observed on the SWEMWBS score, but this did not reach statistical significance (AMD = 1.16, 95% CI = −0.07, 2.38; d = 0.19, 95% CI = 0.04, 0.34; p = 0.06). There was no evidence for an intervention effect on the SDQ Externalizing symptoms subscale score (AMD = −0.47, 95% CI = −1.09, 0.14; d = 0.14, 95% CI = 0.01, 0.30; p = 0.13) and on remission over 12 months (AMD = 1.47, 95% CI = 0.73, 2.96; p = 0.28) (Table 2). The intervention effect on most outcomes was relatively stable across sensitivity analyses, except for the SDQ Total Difficulties and SDQ Impact scores. The effect on these 2 outcomes no longer reached statistical significance in the sensitivity analysis adjusting for missing data, and the effect was slightly attenuated compared with the primary analysis (S2 Table).

Moderator, mediators, and process indicators

For both primary outcomes, a stronger intervention effect was found for those scoring within the borderline range on the SDQ Total Difficulties scale at baseline, compared with those scoring above the “abnormal” threshold at baseline (SDQ Total Difficulties, borderline range: AMD = −4.27, 95% CI = −7.62, −0.92; SDQ Total Difficulties, abnormal range: AMD = −0.53, 95% CI = −2.55, 1.50; p for interaction = 0.06; YTP, borderline range: AMD = −1.79, 95% CI = −3.20, −0.38; YTP, abnormal range: AMD = −0.37, 95% CI = −1.21, 0.48; p for interaction = 0.09). However, these differences did not reach statistical significance. There was no evidence for moderation effects by any of the other prespecified variables (baseline chronicity of mental health difficulties, YTP typology, and SDQ caseness profile) on primary outcomes (S3 and S4 Tables).

There was no evidence that perceived stress, use of problem-solving materials or skills at the 12-week endpoint mediated the intervention effect on primary outcomes at 12 months (S5 and S6 Tables).

For the participants in the intervention arm, a dose–response analysis indicated that at 12 months, participants who completed the counselor-led problem-solving intervention had an SDQ Total Difficulties score that was 5.20 points higher than those who discontinued the intervention early (95% CI = 1.72, 8.68, p = 0.003). However, intervention completers and noncompleters did not differ on YTP score at 12 months (S7 Table).

Incremental cost and modeled scale-up cost

Total incremental economic costs for setting up and delivering the counselor-led problem-solving intervention in 6 schools over a period of 22 weeks was $19,742 USD, as compared to the costs of providing booklets alone for the same period. These incremental costs were $3,290 USD per school with per-student incremental costs of $3 USD for a period of 4 months (based on an M of 1,060 students in the targeted classes from each school). The per-case cost was $158 USD when analyzed by number of participants (n = 125) in the trial intervention arm. Personnel accounted for 93% of the total costs, whereas capital costs (office and school equipment) and material costs (printed materials and stationery supplies) accounted for 4% and 3% of the total costs, respectively (Table 3).

Table 3

Costs for delivering the counselor-led problem-solving intervention in the trial and in a scale-up scenario.

	Incremental costs for intervention delivery in 6 schools during the trial (in USD)(Total size of targeted school population = 6,356)	Economic costs for scaling up intervention delivery in 20 schools (in USD)(Total size of targeted school population = 21,200)
Total cost	19,742	85,920
Total setup cost	3,638 (18%)	4,886 (6%)
Total implementation cost	16,104 (82%)	81,034 (94%)
Component-wise costs1
Staff cost	18,275 (93%)	63,520 (74%)
Materials cost	658 (3%)	21,236 (25%)
Capital cost	809 (4%)	1,164 (1%)
School-wise costs
Per-school cost2	3,290	4,296
Per-student cost3	3	4
Per-case cost4	158	23

1 Component-wise breakdown of total cost into staff, material (e.g., printed sheets), and capital costs (e.g., furniture).

2 Indicates cost for each school to set up and implement the program in an academic year; calculated by dividing the total cost by the number of schools enrolled in the program.

3 Indicates per capita cost; calculated by dividing the per-school cost by the average number of students in each school.

4 Indicates cost for each treated case calculated by dividing the per-school cost by the average number of cases treated in each school.

USD, United States dollar.

For the modeled scale-up scenario, our analysis indicated a total cost of $85,920 USD to scale up the counselor-led program in 20 government-run schools in New Delhi over one academic year (39 weeks). This corresponds to a cost of $4,296 USD per school for an academic year and a per-student cost of $4 USD per academic year (Table 3). In the context of the per student per annum expenditure of $890 USD incurred by the local government [19], the per-student cost estimated for scaling up the counselor-led problem-solving intervention is about 0.4% of the budgetary allocation per student. In estimating the maximum caseload that could theoretically be managed by each school counselor within this time frame, we divided the total working hours available to deliver the intervention by the average number of sessions provided per participant in the main trial. We estimate that in each school, a counselor would be able to provide the intervention for up to 189 eligible students, affording a larger caseload than was possible for each counselor in the trial. This corresponds to 3,780 cases across 20 schools in an academic year with a per-case cost of $23 USD (S1 Text). The higher caseload in the scale-up model led to a higher material costs, accounting for 25% of the total cost, compared to 3% of the total cost during the trial.

Discussion

We found that a brief problem-solving intervention, delivered by lay counselors and supported by printed booklets, effectively improved adolescents’ self-reported psychosocial problems and mental health symptoms at 12 months following randomization, when compared with problem-solving booklets alone. Compared with previously reported postintervention outcomes at 6 and 12 weeks [7], our findings indicated small but sustained intervention effects on psychosocial problems and mental health symptoms over 12 months. However, the intervention effect on mental health symptoms was borderline, which attenuated further in the sensitivity analysis that adjusted for missing data. There was also evidence for longer-term effects on internalizing symptoms, distress/functional impairment, perceived stress, and well-being. Similar to the results obtained at 6 and 12 weeks, we found no evidence of an intervention effect on remission at 12 months. There were modest effects of preintervention mental health symptom severity on both primary outcomes, whereby those scoring above caseness thresholds were more likely to have poorer mental health outcomes and more severe psychosocial problems at 12 months. The observed intervention effect at 12 months was not mediated by any of the hypothesized variables, i.e., perceived stress, use of problem-solving materials, and problem-solving skills.

Most of the existing evidence for the longer-term effectiveness of youth mental health interventions is based on relatively lengthy (16.5 sessions on average), disorder-specific treatment protocols [3]. It is noteworthy that the longer-term effect of this brief, transdiagnostic, problem-solving intervention on mental health symptoms, although small, is comparable to the long-term outcomes achieved by more intensive interventions, mostly delivered by specialists [3,20]. Further, the effect on mental health symptoms was accompanied by a comparable effect on the YTP (an idiographic measure of problems’ meaningful to a young person), as well as significant improvements in domains of functional impairment and perceived stress. These long-lasting effects across multiple domains strengthen the evidence base for brief transdiagnostic interventions [21] and indicate the potential for small but meaningful and sustained impact in low-resourced public health contexts [22]. The study also adds to the growing evidence that adequately trained and supervised nonspecialist providers may play an important role in delivering effective psychological interventions for common mental health problems, and this strategy can help to overcome supply side barriers involved in scaling up evidence-based mental health care in low-resource settings [23,24]. However, it is not possible to discern the specific mechanism of the intervention effect in the absence of a fully powered mediation analysis. Qualitative exit interviews with trial participants will be the focus of a future trial report and may shed more light on the potential role of problem-solving skills as an active ingredient in driving therapeutic change, as well as considering the respective contributions of booklets and counselors in the development of problem-solving capacity.

Our dose–response analysis indicated that participants who completed at least 4 intervention sessions reported relatively higher mental health symptoms at 12 months compared with early dropouts from the intervention. A possible explanation for this counterintuitive finding may be related to the observation that early problem resolution was the single most common reason for dropping out from the intervention. In other words, a substantial proportion of participants in the intervention arm appeared to improve rapidly in the space of 1 or 2 weeks, and therefore, opted out of further face-to-face sessions, whereas those with persisting problems continued with the intervention. This raises the prospect of a tailored delivery schedule, where an optimal number of sessions for each participant would be decided through shared decision-making [25]. Another potential area for tailoring is suggested by the observed modest effect of baseline mental health symptom severity on intervention outcomes. This could be taken as support for a stratified care model, where those participants with the most severe symptoms would step up to a more intensive intervention directly, bypassing the low-intensity problem-solving step. Our results require replication in a larger sample before stronger recommendations can be made regarding the utility of progressive or stratified versions of stepped care. Future research should also examine the feasibility and resource implications of dynamic, data-driven clinical decision-making in interventions delivered by nonspecialist providers.

While there is a growing evidence base on the effectiveness of adolescent-focused psychosocial interventions, there is still very little information on the economic costs of effective interventions [26]. Our economic evaluation indicated that a brief, lay counselor–delivered problem-solving intervention could be scaled up at a cost of $4 USD per student (or $23 USD per case), which is a small fraction (0.4%) of the per-student budgetary allocation in government run-schools in New Delhi, India [19]. By hiring full-time school counselors and supporting the costs of their training and supervision, school authorities could provide coverage for adolescents who might not otherwise receive services due to barriers such as transportation, cost, or perceived stigma [27]. In doing so, it is important that the school authorities carefully plan the allocation of counselors’ time, as the personnel cost was the biggest cost driver in the per-student cost estimate. Whereas a relatively large school might require more than 1 counselor, several smaller schools could conceivably share 1 full-time counselor to ensure more efficient use of resources. Given these cost variabilities, the per-student costs reported in this study should be considered indicative.

We acknowledge a number of limitations to the study. First, the 12-month follow-up was a post hoc addition to the originally scheduled follow-up assessments at 6 and 12 weeks. Consequently, the sample size calculation did not take into account the numbers expected for the 12-month assessments. The observed LTFU rate at the 12-month point (29.4%) was higher than assumed for the original endpoints. That said, the 12-month LTFU was similar to the attrition rate at long-term follow-up in other randomized trials of interventions for youth anxiety and depression [28,29]. To reduce risk of bias due to attrition while using complete case analysis, variables associated with LTFU were adjusted in these analyses. Further, sensitivity analysis with missing data showed intervention effects were similar to the primary analysis for most variables except SDQ Total Difficulties and Impact scores at 12 months, where the effects were weaker. Second, long-term outcomes may have been affected by participation in other interventions during the extended follow-up period. That said, sensitivity analysis found no difference in effectiveness when removing those participants who were stepped up to a more intensive intervention due to nonresponse at 12 weeks. Third, our analysis was not powered to examine the effects of mediators and moderators. Future research should examine the moderating and mediating roles of baseline characteristics and process variables, which, in turn, can guide the use of stratified models of stepped care and protocols for tailoring intervention dosage. Fourth, the study was not powered to detect an intervention effect on the binary outcome of remission. This may be the reason that we observed effects on most continuous outcomes, yet there was no effect on remission based on crossing prespecified clinical thresholds on the SDQ Total Difficulties and Impact scales. Finally, we did not carry out cost-effectiveness analyses using the primary outcomes, as it would have been difficult to interpret per-unit change for primary outcomes in the absence of a measure of quality-adjusted life years (QALYs), which is most often used in other economic evaluations. Future research could examine statistical mapping algorithms using the responses of condition-specific instruments for the estimation of QALYs. Sensitivity analyses are also needed that modify the assumptions of the delivery parameters in the scale-up scenario, in order to test the robustness of the observed results.

Notwithstanding these limitations, the study demonstrated that a brief problem-solving intervention delivered by lay counselors had durable effects on mental health symptoms, psychosocial problems, and other self-reported outcomes over 12 months. The external validity of findings is strengthened by including adolescents with diverse mental health presentations that reflect real-world case mix. The trial was conducted in government-run schools in low-income, urban areas, which are widespread in India and other LMICs, strengthening the generalizability of findings. In our estimation of costs for scaling up, we have used data on the actual cost of employing personnel and printing materials, which forms the bulk of the cost of providing the intervention. Further research utilizing more sophisticated economic modeling techniques is required to examine possible costs under different scenarios for large-scale implementation.

In conclusion, our findings indicate that a brief problem-solving intervention delivered by nonspecialist school counselors and supported by printed booklets had sustained effects over 12 months and represented good value for money. In view of these longer-term effects, problem-solving delivered by lay counselors can be considered a leading candidate for a low-cost, transdiagnostic intervention. Future evaluations are needed to examine whether incremental benefits can be achieved by supplementing a first-line problem-solving intervention with a more intensive treatment as part of a stepped care model.

Protocol and analysis plan for the 12-month follow-up.

(PDF)

Click here for additional data file.

CONSORT checklist.

(DOC)

Click here for additional data file.

Assumptions guiding cost estimates for the modeled scale-up of the counselor-led problem-solving intervention in 20 schools for one academic year.

(DOCX)

Click here for additional data file.

Baseline characteristics of participants who completed follow-up (n = 174) and those lost to follow-up (n = 76) at 12 months.

(DOCX)

Click here for additional data file.

Sensitivity analysis using imputed data for primary and secondary outcomes.

(DOCX)

Click here for additional data file.

Primary outcome (SDQ) by potential effect modifiers at 12 months.

SDQ, Strengths and Difficulties Questionnaire.

(DOCX)

Click here for additional data file.

Primary outcome (YTP) by potential effect modifiers at 12 months.

YTP, Youth Top Problems.

(DOCX)

Click here for additional data file.

Mediation effect of perceived stress, use of problem-solving materials, and problem-solving skills on SDQ Total Difficulties score at 12 months.

SDQ, Strengths and Difficulties Questionnaire.

(DOCX)

Click here for additional data file.

Mediation effect of perceived stress, use of printed materials, and problem-solving skills on YTP at 12 months.

YTP, Youth Top Problems.

(DOCX)

Click here for additional data file.

Dose–response effect on primary outcomes at 12 months for intervention completers and noncompleters.

(DOCX)

Click here for additional data file.

20 Oct 2020

Dear Dr Patel,

Thank you for submitting your manuscript entitled "Sustained effectiveness and costs of a lay counsellor-delivered, brief problem-solving intervention for adolescent mental health problems in urban, low-income schools in India: 12-month outcomes of a randomized control trial" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff as well as by the Special Issue guest editors and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by .

Login to Editorial Manager here: https://www.editorialmanager.com/pmedicine

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Artur A. Arikainen,

Associate Editor

PLOS Medicine

1 Mar 2021

Dear Dr. Patel,

Thank you very much for submitting your manuscript "Sustained effectiveness and costs of a lay counsellor-delivered, brief problem-solving intervention for adolescent mental health problems in urban, low-income schools in India: 12-month outcomes of a randomized control trial" (PMEDICINE-D-20-05026R1) for consideration in PLOS Medicine’s Special Issue on Global Child Health.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to three independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. In particular, reviewer 3 has raised critical points pertaining to the study design and analyses. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by Mar 22 2021 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see http://journals.plos.org/plosmedicine/s/submission-guidelines#loc-methods.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Caitlin Moyer, PhD

Associate Editor

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

1. Data availability statement: Please revise the data statement. PLOS Medicine requires that the de-identified data underlying the specific results in a published article be made available, without restrictions on access, in a public repository or as Supporting Information at the time of article publication, provided it is legal and ethical to do so. Please see the policy at

http://journals.plos.org/plosmedicine/s/data-availability

and FAQs at

http://journals.plos.org/plosmedicine/s/data-availability#loc-faqs-for-data-policy

Specifically, access to data cannot be dependent on requests to one of the authors of the study. Please update the statement, noting how data underlying the study will be made available.

2. Competing Interests: Please add this statement to the manuscript's Competing Interests: "VP is an Academic Editor on PLOS Medicine's editorial board."

3. Abstract: Methods and Findings: At line 47-48, please indicate you are reporting mean age of the participants. Please provide some demographic background on the adolescent participants, beyond age, such as the population and setting, years during which the study took place.

4. Abstract: Methods and Findings: The retrospective ingredients approach taken could be mentioned.

5. Abstract: Methods and Findings: For the YTP and SDQ main outcomes, please quantify the main results with both 95% CIs and p values.

6. Abstract: Conclusions: We suggest beginning the first sentence with “In this study, we observed…” or similar.

7. Abstract and Author Summary: Where you mention that effects were “moderated by the pre-intervention severity of mental health problems.” could you be more specific in terms of the moderation so the direction/relationship is clear.

8. Methods: While described elsewhere (in ref 8) it would be helpful if some details could be briefly mentioned, such as: briefly summary of details of recruitment and randomization, and background on the lay-counsellors, some details of a definition for the “remission” outcome.

9. Methods: Line 244: Thank you for your note that the statistical analysis plan is available upon request. Please include this as a supporting information file. In addition, you reference the published protocol- however this does not seem to include the updated 12 month time point. It would be helpful if you could please include the modified protocol as a supporting information file.

10. Results: Lines 355-356: Please indicate in the text that the effect on SWEMWBS score did not reach statistical significance.

11. Results: Lines 364-371: It seems as if there were no significant interactions found between SDQ or YTP outcomes and baseline SDQ. Please make this clear, and also present the evidence supporting that there were differences between borderline/abnormal range SDQ at baseline and both SDQ and YTP outcomes (direct comparison, in support of your statement, with the acknowledgement that there is no significant interaction effect).

12. Results: Line 370: Please re-iterate the pre-specified variables here, and please clarify that you are still discussing moderation effects on the primary outcomes.

13. CONSORT Checklist: Please complete the CONSORT checklist and ensure that all components of CONSORT are present in the manuscript. Please add the following statement, or similar, to the Methods: "This study is reported as per the Consolidated Standards of Reporting Trials (CONSORT) guideline (S1 Checklist)."

When completing the checklist, please use section and paragraph numbers, rather than page numbers.

14. Table 1 and Table 2: Please define all abbreviations used in the table within the legend.

15. S4 Table and S5 Table: Please provide the p-values for all the intervention effects (only provided for Baseline Severity- Borderline row of Table S4).

16. S8 Table: We would suggest using adherence or “adherers” or similar rather than using compliance/compliers.

Comments from the reviewers:

Reviewer #1: This trial makes a very significant and, to my knowledge, original contribution to the literature. It assesses the longer-term value of a problem solving intervention based in a school context and assesses its economic viability for scaling up. The authors are world leading experts in this area and the quality of the manuscript certainly reflects this standing in the field. The results are impressive, important, and worthy of worldwide dissemination.

I have some comments on the manuscript and suggestions for revisions/consideration.

Title: The comparator condition isn't mentioned in the title (as per PICOS), and I think it would be helpful to consider indicating it in some form, as the nature of the experimental comparison only becomes clear in the abstract.

p3, line 47: It may be helpful to state 'years old' after 15.61, so that it is clear what this statistic refers to.

I was surprised to read that there were two primary outcomes - my understanding of having a primary outcome is that it is, by definition, singular, such that the problems of multiple testing are reduced. This may be a common procedure but I was, as someone relatively familiar with trial design, unfamiliar with it.

It may be useful to state in the abstract that this was an intention to treat analysis.

From the abstract onwards (e.g., also on page 15), I was a bit confused by the difference between an analysis of outcomes AT 12 months and outcomes OVER 12 months. I assumed the latter was the difference from the baseline to the 12 month point, whereas the former was the raw comparison at 12 months, but this may be incorrect. Hence, it might be helpful to make this more explicit at some point in the paper, or use a more direct/explicit language (e.g., 'Change from baseline to 12 month assessment').

p12, line 72. It didn't seem correct to claim that the evidence was for a sustained effect 'during' the 12 month period, as very little was known from the 12 week test to the 12 month endpoint test. 'Over' would seem a more accurate description of the sustained effect.

The per-student cost is made more salient in the paper, and is the cost provided in the abstract, whereas the per-case cost seems to me a more common and meaningful indicator of the actual economic cost of the intervention. To say that it costs $4 per school pupil to provide this intervention, when most of these pupils are not, in fact, receiving the intervention, does not feel particularly meaningful or precise to me.

p.5, line 89. I was not familiar with term 'retrospective ingredients' approach. If this section is intended to be accessible to a non-specialist readership it may be advisable to use a more accessible or clearly descriptive term.

p9, line 166. It wasn't clear whose assent is being referred to here: adolescents or parents/carers.

p12, line 198. Where was the reason for noncompletion from? Was this self-report or otherwise?

p13, Although details of the primary outcome measures are given in the original paper, I think it would be helpful to describe just a bit more here, particularly as the format of the YTP is somewhat non-standard/idiographic.

p14, Can a little more be given about the nature of the self-report measure of PS materials and skills: how were the questions phrased here?

p20, I wonder if there is a need for the detail on the primary caregivers' education and occupation, or whether this might go into an appendices to shorten the paper somewhat.

p23, line 339-343: again, I struggled somewhat to understand what the statistic being given here was, as compared with the comparison of outcomes at 12 months. Is this the change from baseline? What does it mean 'SDQ = -1.23'? (I wonder if this is a typo and should read: 'SDQ: AMD ='

In the Discussion, I think more could have been said of the fact that neither the use of the booklet, or of problem solving skills, significantly mediated the outcomes. It states on page 31, line 427, that the findings strengthen the evidence base for the effects of problem solving, but without any evidence of a mediating effect, and given that the comparison condition was also problem solving (without counsellor input), it might be argued that the findings do not particular support the value of problem solving practices, per se but rather contact with a counsellor (for instance, as a consequence of relational common factors). Having said that, it is mentioned that the power was not sufficient for this mediation analysis and, if it is indeed the case that nothing can be gleaned from this null finding due to under-powering, I think it would be important to state that earlier on.

S8 Table. It might be the formatting, but I struggled to understand what the means were that were being presented here. Is the data definitely in the right boxes?

Reviewer #2: The study by Malik et al. provides evidence to address two substantial gaps in the literature - the sustained impact and cost of brief mental health interventions in LMIC adolescents. Among the strengths are the use of a "real-world" population and rigorous statistical assessment of outcomes and potential mediators and moderators. While the effect on mental health symptoms demonstrated in the manuscript is small, the authors include a fair reporting of these modest results and a clear discussion of important limitations. My concerns and suggestions to the authors are largely minor issues and points of clarification, outlined below.

Methods

Line 226-227: Are SDQ scores of 19 and 20 considered the clinical threshold here, as they were considered the cut-off scores for inclusion?

Lines 254-255 are not clear. The sentence describes adjustment for baseline measure, school, and missing FU, but the parenthetic (which seems to specify the particular variables adjusted for) includes age, class, and week of enrollment and do not include not missingness.

Discussion

How much variability is expected in cost of the intervention through other implementers? It would seem that other NGOs or governmental bodies would be needed to provide scale-up, at least beyond the 20 schools modeled. It would be helpful in considering the larger-scale impact if the authors could comment on how costs may vary in the paragraph starting at line 451.

Reviewer #3: This is an interesting RCT on the effectiveness and costs of a lay counsellor-delivered, brief problem-solving intervention for adolescent mental health problems in urban, low-income schools in

India. However, there are quite a few major issues needing attention.

1. Sample size. There is no sample size calculation in the paper which is inadequate. The original trial was designed for 6 and 12 weeks outcomes but modified to 12 months, why? and on what bases? However, as written in the paper "the sample size calculation did not take into account the numbers expected for the 12-month assessments", it means the current trial was not properly powered so would not be able to answer the research questions with confidence.

2. Loss of follow-ups. About 30% of loss of follow-ups is of a big concern. Not matter how data was imputed and sensitivity analyses were done, the reliability and robustness of the results are subject to scrutiny.

3. Adherence to intervention. In the intervention arm, there are 97 completed & 28 discontinued intervention, therefore only 97 are per protocol with around 22% of those in the treatment arm didn't complete the intervention. However, this issue was never discussed in the paper or addressed in the analyses.

4. Statistical analyses. The main results were shown in table 2 but very confusing. As shown in table 2, the control arm has 95 and intervention arm has 81 participants which shows this is complete case analysis rather than the 'intention to treat' analysis claimed by the authors. Also the section on 'over 12 months' with 339 controls and 325 interventions is very confusing as it's never specified in the stats analysis section or anywhere. It should be outcomes at 12 months only in table 2.

5. Interpretation of results. What we can see from table 2 main analysis on the 12-month outcomes and also sensitivity analysis from the S3 table, there were either no differences or borderline differences in both primary or secondary outcomes between intervention and control arms. Coupled with inadequate sample size, 30% loss of follow-ups and 22% non-adherence to protocol (intervention arm), there is no evidence with sufficient power to support authors' claim on the usefulness of the intervention.

6. Table 1 should be on comparing two arms rather than on loss of follow-ups, and it's also not properly done. For categorical variables, the percentage should be done column-wise rather that row-wise so that can make sense.

7. Not able to comment on the costing models as need specialist on this part.

Any attachments provided with reviews can be seen via the following link:

[LINK]

5 Apr 2021

Submitted filename: point by point responses.docx

Click here for additional data file.

4 Jun 2021

Dear Dr. Patel,

Thank you very much for submitting your revised manuscript "Sustained effectiveness and costs of a lay counsellor-delivered, brief problem-solving intervention for adolescent mental health problems in urban, low-income schools in India: 12-month outcomes of a randomized control trial" (PMEDICINE-D-20-05026R2) for consideration in PLOS Medicine’s Special Issue: Global Child Health: From Birth to Adolescence and Beyond.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to the three original reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of the remaining points raised by the reviewers, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by Jun 25 2021 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Caitlin Moyer, Ph.D.

Associate Editor

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

1. How did the authors assess fidelity of lay counsellors to the intervention? This is important if lay counsellors are used and scale-up is recommended in a context where mental health resources for adolescents are limited and lay counsellors are needed for program delivery.

2. Please address the remaining points of the reviewers. Given Reviewer 3’s remaining concerns with the limitations of the study, please acknowledge these limitations as thoroughly as possible, and temper or qualify claims (in the Abstract and Discussion) in light of these limitations, where appropriate.

3. Response to reviewer comments point # 41: It would be helpful to have the table of baseline demographic and clinical characteristics for the two arms. Thank you for noting that this information for the two arms was presented in the primary trial paper (Michelson et al, 2020). Although it can be found elsewhere, it would be helpful to include here as well for ease of access.

4. Data availability statement: Thank you for providing link for accessing the de-identified data. If possible, please provide additional information, or a more direct link, for accessing the specific data included in this study.

5. Abstract Line 55-56: If helpful, you could clarify here where you describe that “A protocol modification was made…” that “the trial steering committee recommended a post-hoc extension of the follow-up period to 12 months” or similar language (to provide some context for the 12 month outcomes).

6. Abstract: Methods and Findings: Please include a sentence summarizing adverse events, if relevant. In the last sentence of the Abstract Methods and Findings section, please describe the main limitation(s) of the study's methodology.

7. Author summary: Why was this study done? For this section, it may be helpful to include an introductory point on the broader context for the study (beyond mentioning the previous report on the trial findings).

8. Results: Line 401-402: Please present the results in the text for the intervention effect on SDQ externalization symptom scores over 12 months, although you note this did not reach statistical significance.

9. Results: Line 404: We suggest revising to clarify that while some of the effects for SDQ total difficulties no longer reached statistical significance in the sensitivity analyses, the effect did not appear to be attenuated for the sensitivity analysis adjusted for missing data over 12 months.

10. References: Please ensure the use of the "Vancouver" style for reference formatting (including Journal title abbreviations), and see our website for other reference guidelines https://journals.plos.org/plosmedicine/s/submission-guidelines#loc-references

11. Figure 2a and 2b: Please mention in the legend that the error bars represent mean with SD, if accurate. Please also fully define abbreviations YTP and SDQ in the legend.

12. Table 2: Please note if values are reported as mean score with (SD), if accurate.

13. CONSORT checklist: Thank you for including the CONSORT checklist. Please remove references to page numbers, and instead use sections and paragraph numbers to refer to locations within the text. For “Funding” we suggest reporting the location as “Financial Disclosure statement” or similar.

Comments from the reviewers:

Reviewer #1: Thank you for the opportunity to review the revised submission for this important study, showing the longer term effects of a school-based problem solving intervention. I believe this is an important and robustly conducted and analysed study that is close to completion for publication.

I appreciated the authors' responses to the reviewers comments and their willingness to engage with a range of points and issues. As Reviewer 1, I have a few final recommendations for reconsideration by the authors, and one or two additional points.

Point 21: I continue to feel that the use of the term 'over' 12 months, as opposed to 'at' 12 months, is not as clearly defined as it could be across the MS. It may be that this is common terminology that I am not familiar with and, if so, I am happy to leave this point; but as an informed reader I was still left a bit uncertain what this meant particularly, for instance, on the data presented in Table 2. I think I now understand that the 'over 12 months' score is the predicted score based on the three timepoint, assuming linear change (as opposed to the exact score at 12 months). In the abstract this is explained as, 'the latter assumed a constant effect across three follow-up points'. That gets me closer to what is being described here, but it still doesn't quite tell me WHAT the score is (for instance, is it the predicted 12 months score based on the assumption of a constant effect, or the change over time assuming a constant effect?). From Table 2 it seemed to be the former, but 'over' just doesn't feel like the right word here. 'Predicted' or 'estimated' might be clearer?

Point 23: Although I can see the authors' point here, there seems something of an incongruity to me to be presenting all clinical change (in the abstract) on a per-case basis, but then the costs on a per-student basis. If the principal concern were to look at the whole school effects, surely the mental wellbeing outcomes should be per student as well. Perhaps a 'compromise' here would be to provide the per participant costs along with the per-student costs in the abstract: for instance in parenthesis after the USD4.

Line 209: It's just a little confusing that it talks about the people who delivered the intervention, and then subsequently the counsellors. I'm 99% certain, as a reader, that this is the same group, but perhaps just change 'The counsellors' to 'These counsellors' on line 201-211 to make clear it's the same people.

Line 245: It's good to have some explanation of why there are two primary outcomes. However, it does then beg the question of whether you looked to see any cross-lagged effects from changes in prioritized problems to effects on mental health symptoms. I couldn't see any of that analysis here. Is it something to suggest for future research? I guess that depends on how strongly you hold this hypothesis. If just a loose guess, it might make more sense on lines 244-246 to talk about these as two important outcomes rather than suggesting a causal relationship.

Table 2: Apologies if I missed this, but I wasn't sure why you had the 'at' and 'over' AMDs/ESs/ps for the two primaries, but then not for the secondaries. Do these need adding in too? If not, are the data given 'at' or 'over'?

Line 429 says that the per-case cost was USD 158, but this seems to reduce to USD18 at line 446. Is a 'per-student treated' cost the same as a 'per-case' cost (if so, please align terminology and, if not, please explain the former term). More importantly, I did struggle to understand how the cost when the intervention was scaled up was reduced by almost 90%. Are those calculations definitely right? (I checked S3 but that didn't really help explain the magnitude of reduction). Perhaps a few added words in the text here might help the reader to understand why the cost becomes so much less.

Table 3: The per-treated case cost is now 23 USD. Should that be 18, as above.

I really failed to understand the table here. Is there some way it can be made clearer? How is it that the Total cost in the top row is not the total of the subsequent costs? Why is the per-school cost greater in the scale up version than for the six schools, but the pre-treated case cost so much lower? As a reader (albeit not expert in economic analysis), this table, and the summary of it (lines 434-446) could do with some further work to really bring out the key findings and the logic of the analysis.

lines 559-562: It seemed strange to end the paper by referencing econd line interventions. No evidence was presented to support the claim that this would improve effects, and it seemed to take away from the value of the first-line intervention effects being reported here.

Reviewer #2: I am satisfied with the authors' responses to my initial comments. However, their clarification on the SDQ clinical threshold used and additional revisions regarding other reviewer comments have raised some additional major concerns.

Major Concerns:

I am either confused by the presentation or there is an error in the data entered in Table S8. How did completers have a baseline SDQ mean of 8? Wouldn't it have had to be higher than 19, considering that was the cutoff for inclusion? Is baseline at the end of the treatment and not the beginning of the trial? It looks like something may be wrong with the YTP cells as well, with completers having a mean score of 1.5 at baseline and 2.4 at follow-up while non-completers have baseline of 7.1 (much closer to the mean baseline YTP score in Table 1). Neither of these low baseline scores seem reasonable when looking at Figure 2 graph means and error bars. If these numbers are in fact correct, the authors need to clarify how the SDQ baseline mean was 8 (far below the inclusion cutoff) and why completers' baseline scores for both the SDQ and YTP were so low compared to non-completers. If they are incorrect, in addition to updating results text and the table, the authors need to change the discussion paragraph on dose-response to reflect the correct finding.

I think there is not enough attention paid to the fact that the average symptom reduction over time and the percent of remitters in the control and intervention are quite similar. Yes, there is a significantly higher reduction in the intervention, but the AMD for the SDQ total difficulties is <2. Considering 1) the SDQ total scores can range from 0-40, 2) 19/20 is the clinical cutoff, and 3) both the intervention arm and control arm seem to have an average symptom reduction of ~10 points over the 12 months (Table 2B), I think the authors need to comment on the clinical significance of such a small difference between arms. Moreover, while I appreciate the authors edits to remove emphasis on problem-solving and clarify the potential optimization by non-specialist providers as well as their mention that the study was not powered to detect mediators (lines 467-479), I think this needs to be more hashed out. Are the problem-solving booklets alone a good intervention? Considering the majority of costs came from personnel, the control condition may actually be a more cost-effective intervention. Is it possible that there was contamination from adolescents in the intervention group discussing their treatment with peers in the control group? I think there also may be a typo in lines 416-418 "There was no evidence that perceived stress, use of booklets of problem-solving skills at the 12-week endpoint mediated the intervention effect on primary outcomes at 12 months (S6-S7 Tables)."

Minor Comment

In sensitivity analyses adjusting for missingness, it appears that the SDQ impact score was not significantly different in the intervention vs. control over 12 months (Table S3). The authors should add to this sentence in addition to the mention of the weakened effect seen for the SDQ total difficulties score in the results and discussion sections.

Reviewer #3: Thanks authors for their effort to improve the manusript. However, I am not satisfied with the response and revision at all. The authors failed to address almost every single one of my comments/concerns. Just to reiterate,

there were either no differences or borderline differences in both primary or secondary outcomes between intervention and control arms. Coupled with inadequate sample size (not powered at all), 30% loss of follow-ups and 22% non-adherence to protocol (intervention arm), there is no evidence with sufficient power to support authors' claim on the usefulness of the intervention.

Key points:

1) The analyses are not 'intention to treat' at all. Please stop using it and acknowledge as a limitation carefully.

2) The study is not powered for the 12-month primary outcome at all. It looks more like an observational study with follow-ups.

3) For RCT, it's not acceptable without a baseline table 1.

Any attachments provided with reviews can be seen via the following link:

[LINK]

16 Jul 2021

Submitted filename: Editorcomment_July.docx

Click here for additional data file.

13 Aug 2021

Dear Dr. Patel,

Thank you very much for re-submitting your manuscript "Sustained effectiveness and costs of a lay counsellor-delivered, brief problem-solving intervention for adolescent mental health problems in urban, low-income schools in India: 12-month outcomes of a randomized controlled trial" (PMEDICINE-D-20-05026R3) for consideration in PLOS Medicine’s Special Issue: Global Child Health: From Birth to Adolescence and Beyond.

I have discussed the revised paper with my colleagues and the Special Issue Guest Editors, and it was also seen again by three reviewers. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We expect to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosmedicine@plos.org.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by Aug 19 2021 11:59PM.

Sincerely,

Caitlin Moyer, Ph.D.

Associate Editor

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

1. Title: We suggest revising the title to: “Effectiveness and costs associated with a lay counsellor delivered problem-solving mental health intervention for adolescents in urban, low-income schools in India: 12 month outcomes of a randomized controlled trial.” or please revise similarly with the aim of making the title as concise as possible.

2. Data availability statement: Thank you for providing a link to access the de-identified data. We request that you provide more specific information to facilitate access to the data. Please provide a more direct link to the data, and/or include the DOI or accession number for the dataset.

3. Response to reviewer 1: Please address the comment of the reviewer regarding the mediation analysis and lack of power.

4. Response to reviewer 3: Please balance the description of results and interpretation in response to Reviewer 3’s points regarding the loss to follow up at 12 months and the borderline significance of some findings (namely the SDQ Total Difficulties Scale comparison at 12 months, taken together with the results from the sensitivity analysis).

The editors ask that acknowledgement of this limitation be given more explicit emphasis throughout where implications of the results are considered, particularly in the Abstract and Discussion.

5. Abstract: Line 74: “The intervention effect was stable for most outcomes on sensitivity analyses; however, for SDQ Total Difficulties and impairment, the effect was slightly attenuated.” Please clarify the sensitivity analyses mentioned here (that this was the sensitivity analysis adjusting for missing data) rather than just saying the sensitivity analysis showed an attenuated effect.

6. Abstract: As the last sentence of the Abstract Methods and Findings section, please describe the main limitation(s) of the study's methodology.

7. Abstract: Line 82-83: In line with reviewer 3’s comments, we think it may be helpful to provide slightly more general detail on why caution is needed given the post-hoc extension of the trial. We suggest also mentioning the specific limitations as the last sentence of the Methods and Findings section of the abstract.

8. Discussion: Line 487-488: Referring back to Reviewer 3’s point, it may be helpful here to qualify the conclusions slightly, noting that the evidence supports an effect 12 months, but the effects are small particularly for SDQ Total Difficulties.

9. Tables: For the notes in the legend, we suggest not numbering the notes across tables, but instead numbering the notes beginning with “1” for each table.

10. Table 3: Please provide the definition for abbreviation (USD) in the legend.

11. References: Please use the "Vancouver" style for reference formatting, and see our website for other reference guidelines: https://journals.plos.org/plosmedicine/s/submission-guidelines#loc-references

Please double check formatting of all references, including journal title abbreviations. For example, in reference 2, The Lancet should be Lancet. In reference 5, Behaviour Research and Therapy should be Behav Res Ther. In reference 7, Lancet Child and Adolescent Health should be Lancet Child Adolesc Health. Please revise throughout the list.

12. Figure 1 Title and S2 Appendix: In the title of Figure 1 and for the title of S2_Appendix listed on page 41, please capitalize CONSORT.

13. S8 Table: In the file and title name for this table listed in the main text (page 41), please change “compliers” and “non-compliers” to completers. Thank you for making this change within the table file.

14. S3 Appendix: Please provide a clean version of the file (there is some mark-up in the reference to S1a, S1 b Tables). Please format the reference in this document similar to those in the main text (with the in text citation within square brackets rather than superscript, and the appropriate reference formatting).

Comments from Reviewers:

Reviewer #1: Thank you for the further work on this manuscript and the detailed response. It makes an excellent and much-needed contribution to our understanding of effective intervention for adolescent mental health. Congratulations on your work and your encouraging findings.

I have just a few final comments:

Abstract: The actual per-case (and per-student) costs need to be given in the abstract, as this is the principal finding from the study (USD158). If the estimated/scaled up per case/per-student costs are also given in the abstract (USD4/23), as present, it should be made clear that this is an 'estimate'.

line 512-514. It needs to be made clearer here that the mediation effects of specific mechanisms/processes were tested, and not found to significantly mediate. This seems to be the first time that the lack of power for this analysis is given, and it would be helpful if this was stated earlier in the methods. An indication of the specific power for this analysis may be helpful if possible.

Reviewer #2: I am satisfied with the authors' response and most recent revisions.

Reviewer #3: Thanks authors for their effort to improve the manuscript. The authors have addressed some of my concerns such as clarifying the 'intention to treat' and also adding the baseline Table 1. However, I am still not convinced/satisfied with the response to point 29 on sample size, power and statistical significance.

1) The authors said "Regarding the point about difference between arms and inadequate sample size: Table 2 (page 23-24) shows that there were significant intervention effects (p<0.05) on all primary and secondary outcomes except for the SDQ Externalizing subscale score (p=0.13) and SWEMWBS score (p=0.06)". However, out of 9 comparisons in table 2, there are 7 p-values ranging from 0.02 to 0.06 therefore means borderline significane (very close to 0.05) in 78% of all the results. Also the sensitivity analyses with missing data imputation didn't help. Basically the results in table 2 showed very weak and inconclusive evidence (if any) on the effectiveness of the intervention, which has never been acknowledged and discussed in the paper.

2) The authors said "we respectfully disagree that the study is not powered for the 12-month primary outcomes". However, it says in page 10 "The estimated sample size for the originally stipulated primary endpoint assumed loss to follow-up of 15% at six weeks (based on piloting) and a 1:1 allocation ratio", therefore clearly the sample size was calculated for outcome at 6 weeks rather than at 12 months. Now the primary analyses have changed to 12 months but the sample size was still from that at 6 weeks? Just to re-iterate, therefore, the study was not powered (sample size) for primary outcomes at 12 months. The sample size calculation at 6 weeks doesn't match the primary outcomes at 12 months. Due to the inadequte sample size, mostly borderline results and the 30% loss of follow-ups and 22% non-adherence to protocol (intervention arm), the reliability and believability of the results are subject to scrutiny.

Any attachments provided with reviews can be seen via the following link:

[LINK]

18 Aug 2021

Submitted filename: Editorscomment_Aug2021_rev.docx

Click here for additional data file.

20 Aug 2021

Dear Dr Patel,

On behalf of my colleagues and the Academic Editor, Kathryn Yount, I am pleased to inform you that we have agreed to publish your manuscript "Effectiveness and costs associated with a lay counsellor delivered, brief problem-solving mental health intervention for adolescents in urban, low-income schools in India: 12-month outcomes of a randomized controlled trial" (PMEDICINE-D-20-05026R4) in PLOS Medicine’s Special Issue: Global Child Health: From Birth to Adolescence and Beyond.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with medicinepress@plos.org. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for submitting to PLOS Medicine. We look forward to publishing your paper.

Sincerely,

Caitlin Moyer, Ph.D.

Associate Editor

PLOS Medicine