Testing Community-Led Outreach Clinics to Improve Health in Rural Areas

The HAC program establishes one-day integrated care outreach clinics in remote communities (communities that are greater than 5 km from the nearest health facility) every 1-2 months. Government healthcare workers are transported from the health facility to target communities to provide a range of health services. Local Community Associations in remote communities are trained to lead the program in collaboration with the local health facility and are also supported in financing the costs of the outreach clinics.

2026-07-01

2028-08-31

Health Utilization; Family Planning Method Usage

Health utilization is measured by (i) whether an individual received any healthcare in the past 30 days for all age groups as well as separately for those aged 0-5, 5–17, and 18–65, and (ii) family planning method usage during the study period by the primary respondent.

Health Facility Service Delivery, Health Outcomes, Schooling Outcomes, and Labor Market Outcomes

Service delivery is assessed at both the facility and individual levels to capture a comprehensive picture of the intervention's impact. At the individual level, it is measured through health utilization, health status, schooling, and labor market outcomes, focusing on villages located near health facilities that are not actively engaged in the HAC outreaches. At the facility level, service delivery is measured by key operational metrics, including the frequency of drug stockouts, adherence to facility operation schedules, and staffing levels among the participating health facilities.

Health outcomes are assessed by the number of days an individual was sick in the past 30 days, analyzed for all age groups as well as separately for those aged 5–17 and 18–60.

Schooling outcomes are measured by the number of school days missed in the past 30 days for individuals aged 5–17.

Labor market outcomes include the number of workdays missed due to usual work activities in the past 30 days for individuals across all age groups and whether individuals aged 18–60 are currently engaged in wage labor. These variables will be collected through multiple rounds of household surveys.

This study is a two-armed randomized control trial that will be conducted among 64 health facilities, 192 corresponding communities / villages (3 villages per health facility) that are located within each health facility’s catchment area, and an estimated 3840 eligible households (approximately 20 households per selected village). The study consists of a baseline survey with health facilities, villages, and households, followed by randomization of health facilities into the HAC intervention and control arms and the implementation of the two-year family planning intervention. The endline survey will be conducted around one year after the implementation begins.

We will select 64 HC-III or HC-IV health facilities (HFs), and we will identify 128 villages (2 villages per HF) that are located more than 5 km away from the 64 HFs and another 64 communities (1 community per HF) that are located adjacent to (within 5km of) the 64 HFs. From each village/community, we will select 20 households with women aged 18–35 who are non-pregnant and at least six months postpartum for baseline surveys.

Following a baseline survey with 3840 households and 64 HFs, we will randomize HFs, and their corresponding villages and households, into a treatment group or a control group:
• 32 HFs will be randomly selected as the control group. These HFs, and their corresponding villages and households, will receive no interaction with the HAC intervention, and no HAC outreach services will take place.
• 32 HFs will be randomly selected as the treatment group. Staff at these HFs will partner with HAC and will provide outreach services to households in the corresponding villages that are part of the HF's service catchment area..

We will then follow up with the surveyed households and health facility one year after the intervention rollout to: (i) measure impacts of the HAC intervention on short-term outcomes of interest, specifically contraception, (ii) identify any effects of the HAC intervention on the service delivery at health facilities, (iii) detect any unintended impacts of the HAC intervention on service utilization at HFs and the resulting spillover effects that these unintended effects may have had on non-target communities, and (iv) estimate the cost-effectiveness of the HAC.

Not available

Randomization done in office by a computer.

Our study will employ a clustered randomization design, stratified by district. Health facilities will be randomly assigned to either the treatment or control group.

Yes

64 health facilities

64 health facilities, 192 villages, and 3840 households.

Treatment group: 32 health facilities and their corresponding 64 villages and 1280 households within the catchment area.
Control group: 32 health facilities and their corresponding 64 villages and 1280 households within the catchment area.
Indirect analysis: 64 communities, and their corresponding 1280 households, that are located close to the health facilities.

We focus on the primary outcomes measured one year after the intervention rollout. In remote villages we can detect a 0.2 SD increase in current family planning use, or a 9.4 pp change, and a 0.235 SD increase in ever use during the intervention period, or a 11.3 pp increase. These estimates are reasonable, as our intervention directly delivers health care to an underserved population receiving outreaches for the first time. We remain well-powered to detect changes in months of usage among current users at a 70% take-up rate and unmet demand at 80% take up. Similarly, at a 90% take-up rate, we have approximately 80% power to detect a 0.239-0.26 SD change in overall healthcare utilization For our primary comparison, the study is powered to detect 0.215 SD changes on health care utilization in the last 30 days in remote villages. This corresponds to an increase of 9.4 percentage points over a baseline mean of 26%. For family planning outcomes, the study is powered to detect a 0.235 SD (or 11.449.46 pp) improvement in ever using a family planning method during the intervention period. Throughout the power calculations, we assume 80% power, 5% significance, and a 10% endline attrition. We assume intra-cluster correlation (ICC) and R-squared gains from covariates from existing data sources as well as a 0.27 improvement in R-squared by including covariates at health facility level. Our study may suffer from non-compliance at either an outreach level (i.e., the outreaches do not occur) or at a respondent level (i.e., the respondent does not take up contraception). We aggregate both forms of non-compliance into the “take-up rate” and analyze sensitivity of statistical power to various scenarios.