Menu
Menu
Menu
Menu
Background: Little is known about how to scale up care for depression in settings where non-physician lay workers constitute the bulk of frontline providers. We aimed to compare a stepped-care intervention package for depression with usual care enhanced by use of the WHO Mental Health Gap Action Programme intervention guide (mhGAP-IG). Methods: We did a cluster-randomised trial in primary care clinics in Ibadan, Nigeria. Eligible clinics were those with adequate staffing to provide various 24-h clinical services and with regular physician supervision. Clinics (clusters), anonymised and stratified by local government area, were randomly allocated (1:1) with a computer-generated random number sequence to one of two groups: an intervention group in which patients received a stepped-care intervention (eight sessions of individual problem-solving therapy, with an extra two to four sessions if needed) plus enhanced usual care, and a control group in which patients received enhanced usual care only. Patients from enrolled clinics could participate if they were aged 18 years or older, not pregnant, and had moderate to severe depression (scoring ≥11 on the nine-item patient health questionnaire [PHQ-9]). The primary outcome was the proportion of patients with remission of depression at 12 months (a score of ≤6 on the PHQ-9, with assessors masked to group allocation) in the intention-to-treat population. This trial is registered with the International Standard Randomised Controlled Trials Number registry (ISRCTN46754188) and is completed. Findings: 35 of 97 clinics approached were eligible and agreed to participate, of which 18 were allocated to the intervention group and 17 to the control group. 1178 patients (631 [54%] in the intervention group and 547 [46%] in the control group) were recruited between Dec 2, 2013, and June 29, 2015, among whom 976 (83%) were female and baseline mean PHQ-9 score was 13·7 (SD 2·6). Of the 562 (89%) patients in the intervention group and 473 (86%) in the control group who completed 12-month follow-up, similar proportions in each group had remission of depression (425 [76%] in the intervention group vs 366 [77%] in the control group; adjusted odds ratio 1·0 [95% CI 0·70–1·40]). At 12 months, 17 (3%) deaths, one (<1%) psychotic illness, and one (<1%) case of bipolar disorder in the intervention group, and 16 deaths (3%) and one (<1%) case of bipolar disorder in the control group were recorded. No adverse events were judged to be related to the study procedures. Interpretation: For patients with moderate to severe depression receiving care from non-physician primary health-care workers in Nigeria, a stepped-care, problem-solving intervention combined with enhanced usual care is similarly effective to enhanced usual care alone. Enhancing usual care with mhGAP-IG might provide simple and affordable approach to scaling up depression care in sub-Saharan Africa. Funding: UK Medical Research Council.
As described in the protocol,18 the study was a two-arm, parallel, cluster-randomised, controlled clinical trial conducted in primary care clinics in the city of Ibadan, a large metropolis in the southwest of Nigeria. Among all the primary health-care centres within the city's 11 local government areas (five urban and six rural), those that had a full complement of primary health-care workers (ie, adequate staffing to provide a broad range of 24-h clinical services and with regular physician supervision) were assessed for eligibility. Clinics mainly focused on maternal and child health care in the perinatal period were excluded. The units of randomisation (clusters) were primary care clinics, and the units of analysis were individual participants. Eligible clinics that provided consent to participate were randomly allocated to one of two study groups: a stepped care plus enhanced usual care group (intervention) or an enhanced usual care only group (control). We recruited consecutive attendees at the enrolled clinics who had a score of 11 or higher on the nine-item patient health questionnaire (PHQ-9),20 which has been previously validated in our setting.18, 21 Other eligibility criteria were ability to speak the study language (Yoruba), age 18 years or older, not being pregnant or breastfeeding, not needing immediate medical attention, and meeting the criteria for a diagnosis of DSM-IV major depression assessed with the short form of the composite international diagnostic interview.17, 22 Full criteria are described in the protocol.18 The study was approved by the University of Ibadan and University College Hospital ethics committee and was monitored by an independent trial steering committee. All participants provided written (or witnessed, if illiterate) informed consent. Following recruitment of the initial group of participating clinics, anonymised codes for each clinic were provided by the research team in Ibadan to the study statistician (AAM), who generated the allocation sequence and carried out the random allocation. Primary health-care centres were stratified by local government area and randomly allocated in a 1:1 ratio to the intervention group or the control group. For each local government area, a single balanced block equal to stratum size was generated with use of computer-generated random numbers to ensure balanced allocation to treatment groups. We aimed to recruit 90 participants from each of the 16 clinics initially randomised; however, when recruitment was slower than anticipated, we randomised an additional 19 clinics. Allocation of these additional clinics followed the same procedure. Outcome assessors were blinded to patients' group allocations, were not involved with screening or recruitment of trial participants in the clinic, and were randomly assigned to participants from any clinic in either group of the study. Data were collected and stored electronically using android tablets and downloaded to a secure server located in the research office. Data were kept anonymously using codes to identify individuals. These datasets did not contain the allocation status of the participants which was kept as a separate file and was available only to the trial statistician. In the Nigerian setting, front-line primary care providers consist of nurses, community health officers, and community health extension workers, each of whom has 2–3 years of post-secondary school professional training (ie, an average of 14 years' education). Supervision and support to all the primary health-care centres in each local government area (typically eight to ten centres) is provided by a general practitioner, acting as primary health-care coordinator, who runs outpatient clinics, provides clinical supervisions on a scheduled regular basis across the clinics, responds to clinical emergency calls, and has administrative management duties. At each of the participating clinics, two front-line primary care providers, of any cadre (ie, nurse, community health officer, or community health extension worker), were selected and trained to provide treatment appropriate to the study group. Providers in the intervention group and the control group delivered usual care enhanced with the mhGAP-IG,23 in which specifications for the treatment of depression consist of simple psychosocial approaches, including psychoeducation and counselling to address stressors and activate social networks, and pharmacotherapy when necessary. Primary health-care workers in the control group received a 2-day top-up training session in the use of mhGAP-IG. Primary health-care workers in the intervention group were also trained to deliver a structured psychological intervention consisting of behavioural activation (activity scheduling) and problem-solving therapy, previously culturally adapted and pilot tested by us.17, 18 These providers received 6 days of training on problem-solving therapy and on use of the mhGAP-IG to identify and treat depression, which included didactic lectures, clinical demonstrations, role plays on the delivery of the manualised intervention, procedures for support and supervision by the general practitioner through mobile phones, and how to monitor patients on antidepressant medication. Trained providers had to meet a predefined benchmark for post-training evaluation to be able to participate in the trial. Of the 39 potential providers trained, three did not achieve the set of competency standards and were excluded from the study. A few months into the trial, two trained research supervisors (coordinated by BDO) did a fidelity assessment through direct observation and rating of a total of 205 randomly selected sessions (around six sessions per health worker), using a checklist that consisted of key items of the intervention procedures. Items were rated as 0 (poor or not done), 1 (fair or partially done), or 2 (good or well done). Consecutive patients at participating primary health-care centres were screened on the PHQ-9 by trained research assistants. Each consenting participant who screened positive (scored ≥11 on the PHQ-9) was provided with their PHQ-9 score and referred to one of the workers providing trial treatment in the clinic. In the intervention group, in the first step, primary health-care workers use participants' PHQ-9 screen scores to determine treatment options: those with scores of 11–14 were offered eight sessions of psychological intervention delivered by primary health-care workers, and those with scores of 15 or more at baseline were assessed for additional antidepressant medication in consultation with the supervising general practitioner. Therapy sessions were done in person and individually. During the first session, participants were offered psychoeducation in which information about the symptoms of depression, possible causes, and treatments were discussed. After providing reassurance about the treatability of their condition, the structure of the subsequent sessions was discussed and negotiated. The following five sessions dealt with identified problems, difficulties, and stressors, with the therapist working with the patient to explore potential feasible solutions, including how to harness the assistance of supportive social networks. In the final two sessions, both therapist and patient worked together to integrate the experiences of the previous sessions, draw concrete lessons, and use these to prepare for the future. After the first eight sessions, each participant was reassessed with the PHQ-9, this time by the primary health-care worker. Those with a PHQ-9 score of 11 or more, or greater than 50% of baseline score, proceeded to step two. Step two consisted of additional therapy sessions or a combination of therapy and medication following a review by the supervising physician. All participants who did not improve after step two had their cases discussed with a psychiatrist in the third (final) step. In the intervention group, all supervision and consultations with doctors were provided on as-needed basis and through mobile phone contact, except when a face-to-face review was deemed necessary and feasible. The components and tasks for each treatment session and the clinical decisions and steps were detailed in manuals and charts provided to the primary health-care worker and the primary care physicians. When medication was required, the first-line antidepressant was amitriptyline, which non-physician primary care providers in Nigeria are authorised to prescribe. The trial manual stipulated that, when antidepressant medications were prescribed, the frontline clinician consults with the general practitioner either face-to-face or via mobile telephone to receive appropriate advice on dosing and monitoring. Other antidepressants could be prescribed by the general practitioners for patients who did not improve or had other contraindications to the use of tricyclic antidepressants. Any emergent medication side-effects were reviewed in consultation with the general practitioner. Participants in the control group received enhanced usual care alone. The choice of intervention (either unstructured psychological treatment or medications as stipulated in the mhGAP-IG) was at the discretion of the primary health-care worker and no specification as to the number of sessions was made. Outcome assessments were done through face-to-face interviews at the respondents' homes. All outcome assessors were college-educated, experienced in the conduct of surveys, and received 2 weeks of training, including trial life interviews, inter-rater exercises, and regular field debriefing. Interviewers assessed patients' disability level using the WHO disability assessment schedule (WHODAS) 2.0,24 and quality of life using the WHO quality-of-life questionnaire (WHOQOL).25 The service utilisation questionnaire26 was used to collect resource-use data, including any inpatient care, consultations with health providers, use of drugs and laboratory tests, and time and travel costs associated with this service uptake. We obtained information on the financing sources for each of the categories to allow for an estimation of the extent of private, out-of-pocket expenditures incurred by study participants and their families. The unit costs or prices of these various resource inputs were derived through a costing analysis in some participating health facilities with use of data collection templates and protocols previously developed and applied by us.27 Yoruba versions of all the questionnaires were derived by standard protocols of iterative back-translations, as done in previous surveys, achieving good psychometric results.24, 28 To assess cost effectiveness at follow-up, changes in service costs were analysed with respect to changes on the PHQ-9 and WHODAS at the 6-month and 12-month follow-up visits. The service costs incurred in each group were collected with the service utilisation questionnaire. A set of unit costs and prices for all inpatient and outpatient service use, as well as the costs of the interventions, were generated using simplified costing templates and local data inputs. Cost-effectiveness acceptability curves were constructed for a unit improvement on PHQ-9 and WHODAS. The primary outcome was the proportion of patients who had remission of depression (predefined as PHQ-9 score of <6) at 12 months from entry into the trial. Secondary outcomes included depression symptoms as a continuous PHQ-9 score, assessed at 3 months, 6 months, 9 months, and 12 months, as well as level of disability (assessed with the WHODAS),24 quality of life (assessed with the WHOQOL),25 and health-care use (assessed with the service utilisation questionnaire),26 at 6 months and 12 months. All outcomes were assessed in all enrolled patients. Informed by the results of our pilot study,17 we sought to detect an absolute difference of 18 percentage points (59% in the intervention group and 41% in the control group; equivalent odds ratio [OR] 2·1) at 12 months. We assumed an intracluster correlation coefficient of 0·05, also based on pilot study data, and non-collection of the primary outcome for 20% of participants. The uninflated sample size required 131 participants per group for analysis to detect a difference of 59% versus 41% with 80% power and a two-sided α of 5%. We originally aimed to recruit 90 individuals per clinic. With 72 participants per cluster for analysis and an intracluster correlation coefficient of 0·05, the design effect was 4·55, giving a total number required for analysis of 1190. We therefore aimed to recruit 1440 individuals from 16 clinics. However, because participant recruitment was slower than anticipated, we recruited and randomised a further 12 clinics in March, 2014, and seven in November, 2014, giving a total of 35 clinics randomised in the study. Although the design effect was reduced with an increased number of smaller clinics, the total target sample remained 1440. The primary approach for comparative analyses was to analyse on an intention-to-treat basis at the individual level without imputation of missing data. We used descriptive statistics to assess balance between the study groups at baseline for both clinic and individual participant characteristics. To take appropriate account of the hierarchical nature of the data, we used multivariable mixed-effects regression models to estimate remission from depression at 12 months for the intervention group versus the control group. We present unadjusted and adjusted estimates. In both analyses, we accounted for clustering with use of a mixed-effects model with clinic and local government area as random effects. In the adjusted estimates, we also adjusted for baseline PHQ-9 score, age, and primary health-care centre, and for local government area as a stratification variable. In a sensitivity analysis, we imputed missing primary outcome data with multiple imputation. Similar analyses were repeated for secondary outcomes. For these secondary continuous outcomes, we estimated the difference in mean scores between the intervention and control groups. We also investigated whether between-group differences varied over time by using data from all follow-up visits in repeated-measures analyses. We investigated whether there was any differential effect of the intervention on the primary outcome according to baseline symptom severity (PHQ-9 score <16 vs ≥16) by including appropriate interaction terms in the primary regression model. Because the trial was powered to detect overall differences between groups rather than interactions of this kind, these results are to be interpreted with caution. For the cost and cost-effectiveness analyses, 95% CIs were derived through non-parametric bootstrapping techniques owing to the non-normal distribution of mean service costs per study participant (1000 resamples were run). All analyses were done in STATA (version 13.0) software. This study is registered with the International Standard Randomised Controlled Trials Number registry, number ISRCTN46754188. The funder of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.
