Evaluating the Impact of a Community Health Worker Program in Neno, Malawi

Overview

This protocol concerns the implementation and evaluation of an intervention designed to realign the existing cadre of Community Health Workers (CHW) in Neno District, Malawi to better support the care needs of the clients they serve. The proposed intervention is a 'Household Model' where CHWs will be assigned to households, rather than HIV or TB specific patients, and will be trained to provide support for a wider range of conditions including HIV, hypertension, diabetes, and pediatric malnutrition. The new model is designed to improve retention in care for clients with chronic, non-communicable diseases, along with increased uptake of women's health services and treatment for pediatric malnutrition, while sustaining the high retention rates for clients in the HIV program. Eleven sites (health centres and hospitals) were arranged into six clusters by estimated size of the catchment area populations, with a population range of 11,680 to 26,260 and an average population of 20,400. The order in which the intervention will be rolled out across the sites will be randomized so that the intervention can be evaluated in a stepped-wedge cluster randomized controlled trial. These clusters were grouped based mostly on geographic location but also on catchment area sizes, in order to maximize feasibility of training for the CHW team and not overload CHW training sessions with too many trainees.

The objectives of the household model program are: 1. Timely case finding through education and screening for common, treatable conditions; 2. Linkage to care for symptomatic clients along with those qualifying through routine screening; 3. Ongoing support and accompaniment of patients in care, including adherence support, psychosocial support, and tracking of missed patient visits (NCDs, chronic care, Antenatal care, postnatal care); and 4. Health education for common health conditions and prevention and management of these conditions to optimize prevention, health services uptake, and health management behaviors in the household. All CHWs in Neno will be reassigned and trained in the Household model in a staggered rollout over two years. The maximum number of trainees per group is capped at 60 participants, with some trainings occurring with two groups of CHWs per catchment area. CHWs will receive a 4 to 5 day foundational training, followed by half-day refresher trainings each quarter. CHW training will be evaluated through the following tools: training attendance count; CHW knowledge assessment; CHW skill assessment; CHW refresher assessment; and overall through a training dashboard. The implementation of the new CHW model is designed so that it may be evaluated as a stepped wedge, cluster-randomized trial (SW-CRT). The stepped-wedge study design was selected for a number of reasons. First, the training of CHWs needs to be staggered due to training capacity constraints. Second, all sites in Neno will receive the intervention. And third, the stepped wedge RCT design permits estimation of the causal effects of the intervention. Eleven intervention sites were clustered into six groups based on population size such that each group had manageable number of CHWs to train. The order of implementation for these six sites was randomized by a third party. In the SW-CRT study design, each cluster crosses over from control to intervention group until all groups receive the intervention. The primary outcomes are: * HIV: % of enrolled clients with a visit to IC3 in the last 3m * NCDs * Hypertension: % of enrolled clients with a visit to IC3 in the last 3m * Asthma: % of enrolled clients with a visit to IC3 in the last 3m * Diabetes: % of enrolled clients with a visit to IC3 in the last 3m * Epilepsy: % of enrolled clients with a visit to IC3 in the last 3m * Mental Health: % of enrolled clients with a visit to IC3 in the last 3m * Malnutrition: % of children under five enrolled in care for moderate and severe pediatric malnutrition * Hypothesis: don't expect this to change because most cases are cured * Tuberculosis: % of total population diagnosed with new confirmed TB cases * Women's Health: o Family Planning: % WCBA on long-term family planning methods * Antenatal Care: * % women starting ANC within first trimester The secondary outcomes are: * HIV: * % clients initiated on ART in last year with visit in last 3m * % infants who attend 10w EID visit * % of population tested for HIV * Malnutrition: o % of children aged 6m-59 who were discharged as cured in SFP or OTP (cure rate) * Tuberculosis: * % TB cases completing treatment successfully (no loss to follow up or death) * Women's Health: * Family Planning: * % women of child bearing age receiving modern family planning methods * % women of child bearing age newly initiating family planning o Antenatal Care: * % expected pregnant women in ANC care * % number of women in cohort attending 4+ ANC visits * CHW retention o % of CHW retained during the entire intervention period Descriptive Statistics: * Measure of Facility Performance o % of facilities offering women's health services on a daily basis * % months with no facilities stocking out of RUTF (ready-to-use-food for malnutrition) * Average number of stock out days per month per facility for combination, adult TB medication. Stock outs are measured as when running balance on the facility's stock card is zero. * Average number of stock out days per month for non-communicable disease care at the two satellite pharmacies for several key drugs Outcomes Data To measure the outcomes listed above, we will collect data from: 1. Ministry of Health-monthly reports collected from each facility and entered into electronic database called DHIS2 2. Partners In Health Medical Record capturing patient-level HIV and NCD data. 3. Short, semi-structured qualitative interviews with purposively selected sample of CHW program recipients. The study is designed as a stepped wedge randomized controlled trial. However, unlike a typical trial of this type, data will be collected at the aggregate cluster level rather than from individuals within clusters. As such, we specify a model for the cluster-time cell means. In addition, the primary outcomes are proportions of people, therefore we will specify the model in logs and control for population size to transform to the whole real line and make a linear model appropriate.