The effectiveness of depression management for improving HIV care outcomes in Malawi: protocol for a quasi-experimental study

Our evaluation study has two objectives. The primary objective of this evaluation is to assess the impact of a pragmatic, scalable mental health treatment program on HIV care outcomes including retention in care and viral suppression. The secondary objective of this study is to evaluate the impact of the mental health treatment program on mental health outcomes, including depression remission and depression response.

Specifically, our study is designed to evaluate the following hypotheses:

The study employs a multiple baseline evaluation design in two clinics to evaluate the impact of the integrated depression treatment program on HIV outcomes (Fig. 1). The reasons for the choice of the study design has been described published elsewhere [39].

Study activities will last 27 months. This period includes abstracting depression screening data on patients starting ART over a 20-month period and abstracting additional treatment and outcome data for an additional 7 months to identify follow-up viral load, appointment attendance, and depressive symptom outcomes. An additional 3 months is anticipated to complete data abstraction, analyses and dissemination activities.

We are abstracting data on HIV and mental health outcomes for patients newly initiating ART who screen positive for depression (Table 1).

Depressive severity is being measured using the Chichewa version (vernacular language for Malawi) of the Patient Health Questionnaire (PHQ)-9. The score of 5–9 on the PHQ-9 is considered mild depression while that of 10 and above is moderate to severe depression [40]. Providers screen all patients for depression at the time of ART initiation by administering the PHQ-9 rather than self administration because of the low literacy levels among the patients. Patients who score ≥ 5 at ART initiation are re-assessed with the PHQ-9 by the HIV provider at each subsequent ART visit. This information is captured on a form called the Mental Health Mastercard that is kept with the patient’s HIV file.

Depression Remission will be defined as a PHQ-9 score < 5 at 6 months.

Retention in HIV care will be defined as being “on time” to all scheduled ART return visits in the first 6 months on treatment. “On time” will be defined as no more than 30 days late; a secondary, tighter definition will define “on time” as no more than 14 days late At ART initiation and each subsequent ART appointment, providers give patients a follow-up appointment date and a sufficient supply of ART. This information is recorded in the patient’s chart, called an HIV Mastercard, and in the Electronic Medical Record (EMR). However, generally for the first 6 months of care, new ART patients receive a 30-day supply of ART and a return appointment date in 30 days. Most often, patients need to attend monthly ART refill appointments for their first 6 months of care in order to maintain ART supply through 6 months.

Consistent ART supply will be defined as never going more than 5 days without ART in the first 6 months, calculated from the days’ supply of ART dispensed at each appointment in the first 6 months and the time between appointments.

Viral Suppression will be defined as a viral load < 1000 copies/mL at 6 months. A viral load threshold of 1000 copies/ml was chosen as this is the threshold used in the resource-limited Malawian health care system to guide decisions about treatment failure and second-line alternatives. Viral load testing is performed at Bwaila Hospital in Lilongwe using Abbott m2000 RealTime HIV-1 assay instruments with a lower limit of detection of 40 copies/mL. The results of the viral load test are recorded in the EMR.

We integrated depression screening and treatment into ART care into two clinics in Lilongwe, Malawi using a multiple baseline design in two phases: a screening phase and a treatment phase.

During the screening phase, HIV care providers screened patients for depression and monitored their depressive symptoms using the Patient Health Questionnaire-9 (PHQ-9). HIV care providers managed patients identified with depression according to existing care pathways, which included: 1) counselling by the HIV care provider; 2) referral to an on-site or off-site psychiatric nurse or other mental health specialist; or 3) in acute cases, transport to the outpatient psychiatric unit at the nearby district hospital.

At the launch of the treatment phase, HIV care providers were trained in algorithm-guided antidepressant management [38]; additionally, clinic-based lay health workers completed training in the Friendship Bench problem-solving therapy counselling protocol [38]. During this phase, providers are continuing to screen patients for depression. The providers manage cases of depression identified during the treatment phase with either antidepressants or problem-solving therapy, and providers monitor their depression treatment response at follow-up visits.

We are conducting the evaluation study at two public-sector primary care HIV clinics in Lilongwe, Malawi: Area 18 and Area 25 health clinics. These clinics are managed by Lilongwe District Health Office. More detailed description of the study setting has been published elsewhere.

We are abstracting screening data for all consenting adult (≥18 years old) patients newly initiating ART and screened for depression at the time of ART initiation at the study sites during the study period. Additional depression treatment and outcome data are being abstracted for all patients who screen positive for mild, moderate, or severe depressive symptoms (PHQ-9 total score ≥ 5) [56, 57].

We focused on two populations of interest. For the primary population of interest, all patients with depressive symptoms (PHQ-9 score ≥ 5), the original sample size was determined to provide 80% power to detect an 11 percentage point pre-post difference in the primary outcome. For the secondary (smaller) population of interest, all patients with moderate to severe depression (PHQ-9 score ≥ 10), the original sample size would provide the same power to detect an 18 percentage point pre-post difference. To achieve this power, we would require 645 patients with any depressive symptoms and 225 patients with moderate to severe depressive symptoms.

Based upon the historical rate of new ART patients (122 per month between the two clinics) and the expected prevalence of any depressive symptoms (35%) and moderate to severe depressive symptoms (12%) based on prior literature [58] [7], we originally planned for a 15-month data collection period to accrue the targeted sample. During the course of data collection, the rate of new ART patients was higher than expected but the prevalence of depressive symptoms was lower than expected. After extending the recruitment period for an additional 5 months, the accrual of new ART patients concluded with a final sample of 2084 individuals. The prevalence of mild to severe depression was 24.1% (n = 502) and the prevalence of moderate to severe depression was 6.3% (n = 131). This sample has 80% power to detect a slightly larger difference in the primary outcome of 12 percentage points (e.g. 60% vs. 72%) or a risk ratio of 1.2 for those with mild to severe depression. For those with moderate to severe depression, the sample has 80% power to detect an improvement in the primary outcome of 22 percentage points (e.g. 60% vs. 82%) or a risk ratio of 1.4.

Data on viral loads, HIV appointment attendance, and depressive symptoms is abstracted from electronic and paper medical records and the depression treatment registry at the sites. ART appointment data including the date of the appointment, the expected return date, and medication prescribed is recorded in the EMR, on a paper patient file (called an HIV Mastercard), or both. The results of the viral load tests are recorded in the EMR. Data on patients’ depressive symptoms and depression treatment are only recorded on paper in an additional patient filed attached to the ART mastercard (called a mental health mastercard). The community health workers responsible for administering the Friendship Bench problem-solving therapy also maintain paper records (called a Counselling Mastercard) that capture engagement with the PST sessions. Cost information was collected using a time-audit approach to estimate costs of delivering the program, including the opportunity cost of training participation and the time providers and supervisors spend to deliver and supervise the program. Time-audit data was captured through a combination of direct observation of medical provider and counsellor encounters supplemented with self-reporting by counsellors.

The primary program evaluation question is whether the depression treatment program results in improved HIV and mental health outcomes for patients with HIV and depression relative to the standard of care prior to program implementation. The main analysis will follow an “intent-to-treat” approach, classifying patients according to screening vs. treatment phase (unexposed vs. exposed to the program) without regard to actual treatment received. In our main analysis, we will define our primary outcome, retention in HIV care with viral suppression 6 months after starting ART, as never being more than 30 days late to any scheduled return visit in the first 6 months on ART and having a viral load < 1000 copies/mL at 6 months; patients not retained or not suppressed will be coded as a failure. (Table 2) This composite outcome was chosen as the primary outcome because it is the definition of ART clinical success in the Malawian ART treatment system; those not retained and those retained but not suppressed are both clinical failures. This main analysis will estimate the probability of retention and viral suppression 6 months after starting ART using a generalized linear model with binomial error distribution, adjusting for calendar time and potential confounders that are imbalanced between arms at baseline, potentially including age, sex, clinic, PHQ-9 score, and presence of suicidal ideation. We will model these associations using separate generalized linear models among two samples: (a) those with mild, moderate, or severe depression (PHQ-9 ≥ 5) and (b) those with moderate to severe depression (PHQ-9 ≥ 10).

In secondary analyses, we will consider ART care retention through 6 months (binary), number of ART appointments attended in the first 6 months (count), maintaining consistent ART supply through 6 months (binary), and viral suppression at 6 months among those retained (binary). Further, secondary analyses will take an “as treated” approach, comparing patients who receive treatment in the treatment phase to all patients during the screening phase; and an “as adequately treated” approach, comparing patients who receive adequate treatment (≥3 consecutive months of protocol-concordant antidepressant prescriptions or ≥ 4 counseling sessions) during the treatment phase to all patients during the screening phase. Data are not available on antidepressant adherence. (Table 3).

The screening phase began at Clinic A in April 2017 and Clinic B in May 2017. The treatment phase was integrated into Clinic A after 7 months, in November 2017, and into Clinic B after 11 months, in April 2018. Identification of new patients concluded at both clinics in November 2018. Follow-up data abstraction is expected to continue through June 2019.