Direct and Spillover Impacts of a Community-Level HIV/AIDS Program: Evidence from a Randomized Controlled Trial in Mozambique

Last registered on February 18, 2021

Pre-Trial

General Information

Title

RCT ID

AEARCTR-0003990

Initial registration date

March 08, 2019

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

March 08, 2019, 4:02 PM EST

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Last updated

February 18, 2021, 11:40 AM EST

Last updated is the most recent time when changes to the trial's registration were published.

Locations

Country

Mozambique

Region

Primary Investigator

Name

Dean Yang

Affiliation

University of Michigan

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Hang Yu

PI Affiliation

University of Michigan

Contact Investigator

PI Name

James Riddell

PI Affiliation

University of Michigan

Contact Investigator

PI Name

Arlete Mahumane

PI Affiliation

Beira Operational Research Center

Contact Investigator

Additional Trial Information

Status

Completed

Start date

2017-01-01

End date

2020-12-31

Keywords

Education, Health

Additional Keywords

JEL code(s)

I12, I21, O15

Secondary IDs

Abstract

We study a prominent effort to help families cope with HIV/AIDS: a U.S. government-funded program in Mozambique implementing an interrelated set of health and education interventions. We report the results of a randomized evaluation of this program. We hypothesize that the program raises HIV testing rates by reducing imperfect information (related to HIV) and by reducing HIV-related stigma concerns. Primary outcome variables are survey-reported and directly-observed rates of HIV testing. We also examine a range of secondary outcomes related to information and stigma mechanisms, and secondary outcome variables such as school attendance and overall household well-being. Causal identification exploits multilevel random assignment (of communities to program receipt, and of treatment-community households to strong encouragement for program participation), allowing separate identification of impacts on households of being in a program community (with take-up at the community-average rate), versus having been strongly encouraged to participate the program (with very high take-up). We also examine spillovers (based on geographic proximity or social network connections) from strongly encouraged to other households. Additional treatments randomly assigned by our research team allow insight into complementarities with interventions providing targeted information and financial incentives for HIV testing.

External Link(s)

Registration Citation

Citation

Mahumane, Arlete et al. 2021. "Direct and Spillover Impacts of a Community-Level HIV/AIDS Program: Evidence from a Randomized Controlled Trial in Mozambique." AEA RCT Registry. February 18. https://doi.org/10.1257/rct.3990-5.1

Former Citation

Sponsors & Partners

Experimental Details

Interventions

Intervention(s)

Intervention (Hidden)

Intervention Start Date

2017-01-01

Intervention End Date

2020-12-31

Primary Outcomes

Primary Outcomes (end points)

We consider the following outcome variable to be primary outcome of interest: HIV testing at the household level (binary outcome)

Primary Outcomes (explanation)

The outcome variable of primary interest is HIV testing at the household level. This will be a binary outcome indicating that the household either self-reports having had or is directly observed by our survey staff having an HIV test upon our recommendation. This outcome captures the combination of having already had an HIV test, as well as openness to recommendations for future testing, both of which may be influenced by exposure to the FCC program.
To be specific, the component variables of this outcome variable are:

- HIV testing (self-reported): an indicator that anyone in the household has been tested for HIV in the last 12 months. This is a household-level variable equal to 1 if at least one household member is reported to have had an HIV test in the last 12 months, and 0 otherwise.
- HIV testing (directly observed): an indicator that at least one of a household’s HIV testing coupons has been redeemed. This is a household-level variable equal to 1 if at least one of a household’s incentive coupons is presented at the local health clinic for the HIV testing incentive payment before the 14-day deadline, and 0 otherwise.

Our composite HIV testing outcome is therefore equal to 1 if HIV testing (self-reported) is equal to 1 or HIV testing (directly observed) is equal to 1, and 0 otherwise.

Secondary Outcomes

Secondary Outcomes (end points)

Secondary Outcomes (explanation)

Experimental Design

The study uses a three-stage randomized controlled trial (RCT) methodology.
Stage 1: communities were randomly assigned to treatment or control status (inclusion in or exclusion from the FCC program).
Stage 2: a subset of households within FCC treatment communities were randomly assigned to a strong encouragement to participate in FCC programs (“directly enrolled” households). These directly enrolled households receive a home visit by an FCC program community worker and are assessed for inclusion in various FCC subcomponent programs.
Stage 3: as part of the endline survey, households in all communities (treatment and control) are randomly assigned to treatments to provide information about HIV, provide information about HIV treatment (antiretroviral therapy, or ART), provide both HIV and ART information, reduce concerns about HIV-related stigma, and receive higher compensation for getting an HIV test.

Experimental Design Details

Randomization Method

Randomization done in office by a computer

Randomization Unit

Stage 1: communities (38 treatment, 38 control)
Stage 2: households within FCC treatment communities
Stage 3: households in all communities

Was the treatment clustered?

Yes

Experiment Characteristics

Sample size: planned number of clusters

For Randomization Stage 1, 76 communities as clusters (38 treatment, 38 control).
Randomization Stages 2 and 3 are at the household level, not clustered.

Sample size: planned number of observations

76 communities as clusters (38 treatment, 38 control). 4,560 households across all communities.

Sample size (or number of clusters) by treatment arms

Randomization Stage 1: 76 communities as clusters (38 treatment, 38 control). 60 households in each community are included in survey sample.
Randomization Stage 2: 35 households in each treatment community randomly assigned to direct enrollment in FCC program; 25 in each treatment community are controls (not assigned to direct enrollment).
Randomization Stage 3: equal share of households assigned to each of six treatment conditions (HIV information, ART information, both HIV and ART information, anti-stigma, high incentive for HIV testing, and control).

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Supporting Documents and Materials

IRB