Introduction: The prevalence of Diabetes Mellitus (DM) is on the rise the world over. About 30% of DM patients suffer from Depression. Depression in DM patients is associated with adverse outcomes including poor medication adherence, poor glycaemic control, and early death. In resource constrained sub-Saharan Africa (SSA) clinics where patient volumes are high and staff shortages rife, peer support has been suggested as a means of delivering psychosocial care for persons with chronic illnesses in order to improve patient's outcomes. However, little has been done to examine the efficacy of peer support on clinical outcomes.

Project aims: The main study objective will be developing a peer support model of depression care for patients with DM and testing its efficacy on clinical outcomes.

Methods: This study will employ both qualitative and quantitative measures. First, the investigators will present the peer support model to health workers within the DM clinic, and ask them about the feasibility of using such a model for DM patients with depression. The investigators will then identify 10 DM patients with major depression and initiate them on antidepressants. Once the patients are in clinical remission, the investigators will interview them to assess their perceptions about the feasibility of using peer support for DM patients newly diagnosed with depression. The investigators will also interview health care workers and hospital administrators to assess their perception about using peer support within the clinics, and potential barriers that need to be addressed before implementation of the model.

Based on the data from the qualitative interviews, the investigators will refine and adapt the peer support model, and then train 10 DM patients who have received antidepressants and are in clinical remission to deliver peer support to newly diagnosed patients with depression. Newly diagnosed depressed patients will be randomly assigned to receive either antidepressants plus peer support (n=65) or antidepressants alone (n=65). Study participants will be followed for 48 weeks and assessed for, glycaemic control, depression severity, mental illness stigma, depression treatment uptake and adherence.

Result: the investigators anticipate that the findings about the efficacy of peer support on DM and depression outcomes will be useful in generating data about effect sizes necessary for calculating a sample size for a cluster randomized trial (CRT).

Methods: This will be a pragmatic randomized control trial with a 1:1 allocation ratio between cases and controls.

Study Setting: This project will be implemented at the DM clinic of Nsambya Hospital in Kampala City. The clinic, which is staffed by three nurses, a medical officer and a specialist physician, is open once a week. The nurses provide most of the DM care including eliciting symptoms, providing medication refills, as well as referring complicated cases to the medical officer (MO) and specialist physician. The medical officer and specialist physician manage referrals and complications of DM.

Identification of depressed patients to participate in a peer support feasibility study: In the first four weeks of the study, all clinic attendees will be screened for depression by a trained lay health care worker(TLHW) using the PHQ-2 at triage. Screen positive cases (PHQ-2≥3) will be sent to the MO who will conduct a depression diagnostic interview using the Mini International Neuropsychiatric Interview (MINI) till 10 depressed patients are identified. All depressed patients will be initiated on either Imipramine 75mg/day or Fluoxetine 20mg/day, and then reassessed at week 4 and week 6 for side effects and symptom remission using the Hamilton Depression Rating scale (HAM-D).

Conducting qualitative interviews to assess feasibility of peer support: When the 10 patients are in remission (˜ 6-8 weeks) the investigators will conduct qualitative interviews with them, hospital administrators (n=2), and medical personel at the DM clinic (n=5). During the interview, the investigators will assess the participant's perceptions about the feasibility and acceptability of engaging experienced DM patients to provide peer support to patients newly diagnosed with depression. Informed consent will be sought from all participants.

Interview procedure: Dr Elialilia Okello will conduct the qualitative interviews. The participants will be asked about whom in the DM clinic would be a suitable candidate to provide peer support to newly diagnosed depressed patients. The peer support model will be presented to participants so that they can internalize/appreciate the concept, Interview guide: The interviews will be based on a guide with detailed standard probes to ensure that key questions are addressed, and to allow comparisons across individuals and categories of respondents. Questions will be open-ended to allow not only the exploration of new leads but also generation of rich narratives. All interviews will be tape recorded, translated in English (in cases where Luganda, the local language is used) and transcribed verbatim.

Qualitative data analysis: Data will be explored to identify key themes and relationships between themes and analyzed using Atlas.ti. The results from the qualitative interviews will help the investigators refine the peer support model before using it.

After the depressed patients are in full clinical remission, the investigators will ask clinic staff to nominate from among them atleast 5 potential candidates to become peer support buddies. This number should be adequate for the investigators to pilot test atleast 5 interview groups for the peer support process. Self-nominations will also be encouraged after the patients on treatment are made aware of the eligibility criteria. The eligibility criteria will be patients who are; (a) at least 18 years of age; (b) speak Luganda; (c) registered at the clinic for at least 6 months; (d) successful adherer, defined by staff as one who consistently keeps appointments and self-reports being able to adhere to treatment; (e) not currently suffering from any physical or co-morbid mental illness, or using substances of abuse ; (f) socially skilled as defined (albeit subjectively) by staff; (g), and have received antidepressants for at least 10 weeks (and in clinical remission for at least 4 weeks). The investigators will screen more clinic attendees to identify potential buddies if the number identified using the method above is less than 5. Patients who are depressed and are ineligible to become peer support buddies will be followed up for the duration of the study as controls.

Buddy training: A 2-day training workshop of the peer support "buddies" will be conducted by the PI using the buddy training manual developed by Simoni. The workshop will include didactic content, and role-plays. Buddies will be taught strategies geared to address the four components of peer support (self-care, adherence, stigma, and social support). Buddies will meet in a group for supervision with the PI to seek for clarity in areas that were missed out earlier before they can be assigned newly diagnosed patients. Once trained, the buddies will be assigned no more than 5 patient peers at any one time.

Content of peer support training: Buddies will be taught to strive toward the following goals with their assigned peers: (a) explain the purpose of the contacts as educational and supportive, aimed at helping the peer to adhere to both antidepressants and hypoglycaemic medications; (b) learn the peer's regimen and make suggestions on how to follow it, borrowing from their own experiences; (c) encourage peers to follow treatment regimens by praising them for adherence and by expressing confidence in their ability to adhere; (d) allow the peer to express worry, anxiety, and concerns so these can be dealt with, (e) be warm and friendly, (f) and refer frequently to themselves and their success with their own regimen to present a guide or model which the peer can follow. The peers and buddies will use the above goals to set achievable targets for each meeting. Buddies will be specifically instructed not to give any medical advice to the peers, but rather encourage the peers to discuss with their clinicians any questions of a medical nature.

Sample size and power calculation for the peer support study objective: The sample size for the RCT component of the study was calculated to compare the primary outcomes as measured by HbA1c ≤ 7 and HAM-D ≤7 between the intervention and control arms at 48 weeks after initiation of antidepressant treatment. To calculate our sample size, the investigators used findings from an RCT by Safren et al,[61]. Using the standard deviations of the means from this study [61], the investigators estimated that 65 patients per arm (intervention vs. control) are required to have 80% chance to detect, as significant at the 5% level, a 1.4 [SD = 2] mean difference in the HbA1c levels and HAM-D depression scores, taking into account a 20% loss to follow-up at 48 weeks.

Peer support process: Participants in the intervention arm will receive peer support every two weeks for 12 weeks, then monthly 'booster' sessions at weeks 16, 20, 24 and 28 after commencing treatment. However, the results from our qualitative interviews will guide the investigators about the feasibility of the number of meetings.

Peer support will be provided to a group of patients rather than a one-to one session. A number of barriers (transport availability, distance from the clinic, work schedules) may make it difficult to pre-determine when the meetings will take place, but the investigators will ensure that the inclusion criteria for peers into the program includes their willingness to attend all sessions. Buddies will keep ongoing records of all their contacts with their peer clients as an assessment of the integrity of the intervention across buddies, across peers, and over time.

Monitoring fidelity and supervision: In the first 4 weeks of implementation of the peer support intervention (17 weeks into the study), all sessions will be audio recorded to assess adherence to manual guidelines. The PI will then listen to the recordings, and then advise the buddies accordingly. The peers will complete an anonymous evaluation questionnaire following the first two sessions; information from these questionnaires will be used to improve the quality of peer support. A buddy log will be designed to capture the following information; a) which days the meetings occurred, b) duration of the meeting, c) members present, d) issues discussed, and e) any challenges encountered. The investigators will have a master's degree clinical psychologist sit in these meetings and also collect this data from each of the buddies at the end of the meeting, and relay this data to the PI. Over the entire duration of the intervention, the PI will also attend the sessions on a rotator basis; the PI will have the chance of attending sessions from each of the groups.

The study participants from both arms will be followed up for 48 weeks after initiation on antidepressant treatment. The patients in both arms will be reviewed by the MO at weeks 2 and 4 to address any challenges due to side effects, which are critical in preventing early discontinuation of treatment. At each visit, the research assistant will use the HAM-D to assess depression symptom severity, and the standardized side effect screening instrument the antidepressant side effect check-list (ASEC) [63] to assess tolerance to medications and any side effects. The HbA1c test will be performed at baseline, week 24 and 48. At these visits, the research assistant will also collect data from the participants. Research data will be collected at baseline, then weeks 12, 24, and 48 after initiation of patients into treatment.

Quantitative data analysis: A trial profile shall be used to detail recruitment and follow-up of patients. Analyses will be conducted to compare the groups at baseline, weeks 12, 24, and 48 to assess short- and long-term effects of the peer support intervention. The investigators shall analyze the data using the intention-to-treat approach. The dependent variables (HbA1c and HAM-D) will be presented as continuous variables. Predictor outcomes including adherence, mental illness stigma levels, social support and self-care will be presented as continuous and categorical variables. Baseline characteristics of the intervention and control arms will be compared at the 5% level to assess if successful randomization was achieved. Students t-tests (if continuous and normally distributed), Mann-Whitney U-test (if continuous but non-normally distributed), and Chi-squared tests (categorical) will be used. Data on potential confounders and effect modifiers, including variables that fail to achieve successful randomization (e.g. sociodemographic parameters) shall be used to control for confounding and effect modification. The investigators shall conduct both within-subjects (time) and between-subject (group) analyses. Between-subject analysis at 48 weeks will be used to assess the direct effect of the intervention by determining if there is a significant difference between the mean HAM-D and HbA1c scores in the intervention and control arms. Within-subject analysis will be performed among patients in the intervention arm by applying the Generalized Estimating Equations (GEE) method on repeated measures data. To examine the role of social support and mental health stigma as mediating factors in the relationship between depression and both treatment adherence and levels of glycaemic control outcomes, three simple regression analyses each between i) the predictor (social support and mental illness stigma) and outcome (HbA1c and HAM-D scores), ii) predictor and mediator, and iii) mediator and outcome, will be conducted followed by a multiple regression analysis.

Inclusion Criteria:

- All Patients with Diabetes Melitus who have been accessing care at the clinic for atleast 6 months

Exclusion Criteria:

• Patients with overt psychiatric or physical illness for which emergency medical care is needed

Random sequence generation will be done by an independent biostatistician