Back to History Current Version

Evaluation of Living Goods/BRAC entrepreneurial CHW model in Uganda - Phase II

Last registered on May 04, 2020

Pre-Trial

Trial Information

General Information

Title
Evaluation of Living Goods/BRAC entrepreneurial CHW model in Uganda - Phase II
RCT ID
AEARCTR-0002392
Initial registration date
October 10, 2017

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
October 10, 2017, 3:13 PM EDT

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

Last updated
May 04, 2020, 11:10 AM EDT

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

Locations

Region

Primary Investigator

Affiliation
Stockholm School of Economics

Other Primary Investigator(s)

PI Affiliation
Makerere University
PI Affiliation
Trinity Colleg Dublin
PI Affiliation
IIES, Stockholm University

Additional Trial Information

Status
On going
Start date
2016-01-25
End date
2021-01-01
Secondary IDs
Abstract
We study the scaling up of an incentivized Community Health Worker program aiming at improving primary healthcare provision and reducing child mortality in rural areas of Uganda. The community health promoters (CHP) program is implemented in Uganda by two Non-Governmental Organizations – Living Goods (LG) and BRAC. One of the key innovations of this program is that, unlike most volunteer-based community health worker programs, it provides a set of financial incentives for the health workers. More specifically, there are two different categories of financial incentives. First, CHPs make profits by selling a range of health-related products to community members while carrying out their standard activities as community health workers. Second, they receive additional performance-based remuneration based on a set of key health activities that they perform, which include sick child assessment, registration and support of pregnant women, and visits to newborns in the first week of life.

The CHP program is organized into geographically based branches, and managed by branch managers and supervised by the two NGOs. The CHPs are selected through a competitive process among female community members aged 18 to 45 who applied for the position in each village and who possessed basic writing and math skills. Eligible candidates receive 3 weeks of health and business training. At the end of the training, candidates need to pass a skills test in order to be equipped as an active CHP. The NGOs provide an initial set of products to all newly recruited CHPs, together with a uniform, a mobile phone, and a set of training materials and visual aids to use during household visits. CHPs also attend a one-day training each month to review and refresh key health and business topics.

The CHPs tasks mirror the standard Community Health Workers tasks (conduct home visits, educate households on essential health behaviors, provide basic medical advice, referring the more severe cases to the closest health center), but on top of that, as mentioned above, they also sell preventive and curative health products. The product line they have at disposal includes prevention goods (e.g. insecticide treated bednets, water purification tablets, and vitamins), curative treatments (e.g., oral rehydration salts, zinc, and ACTs), as well as other health-related commodities (e.g. diapers, hand soap, fortified food) and durables with health benefits (e.g. improved cook stoves, solar lights, and water filters). These products are sold by the CHP at a discount. The retail price is determined by the NGOs head office with a target of keeping prices for preventive and curative products about 20% lower than the prevailing local market prices. The CHPs in turn purchase these products directly from Living Goods or BRAC branches at wholesale prices between 30-50% below market prices and therefore earn an income on each product sold. Thus, the CHPs operated as micro-entrepreneurs with financial incentives to meet household demand. The broad product mix has three potential benefits: (i) driving up total sales and income for the CHPs; (ii) enabling the NGOs to cross-subsidize prices (dropping prices on essential health products and increasing the margins on other products); (iii) motivating CHPs to be out visiting households regularly by including high-velocity items (such as soap and fortified foods) in the product mix. The business training received by the CHPs stresses the importance of building up a customer-base by providing free services like health education, referrals, and newborn visits. As described above, the income deriving from the micro-entrepreneurial activity is then further increased through performance-based incentives, designed by the NGOs to further encourage key health activities such as household visiting, sick child assessment, registration and support of pregnant women, and visits to newborns in the first week of life. Since 2013, Living Goods and BRAC also equip the CHPs with smartphones that includes a rich mobile health application. The application helps guide the CHW through workflows, keep track of their stock, serve as a client management system, and prioritize certain activities based on timeliness (e.g. pregnancy follow-up) or household risk. Overall, this allows monitoring the CHPs’ activity, while collecting real-time health data from the field.

A first evaluation of the impact of the CHP program began in 2010 (Björkman Nyqvist et al, 2019). The evaluation was based on a cluster-randomized controlled trial that involved 214 villages in 10 districts across Uganda. The villages were stratified by geographical zones and 115 villages were randomly assigned to the treatment group, where the CHP program started operating in January 2011, while 99 villages were assigned to the control group. The evaluation was based on an endline survey collected at the end of 2013, which covered 7,018 households and 11,563 children under-5 that lived in the same village throughout the trial. The study found that over the three years the CHP program reduced under-5 mortality rate by 27% (adjusted rate ratio 0.73, 95% CI 0.58-0.93) in the treatment compared to the control arm. The effects were of similar order of magnitude for infant mortality (adjusted rate ratio 0.67, 95% CI 0.51-0.87) and neonatal mortality (adjusted rate ratio 0.73, 95% CI 0.55-0.98).

Following the first study, the program has been massively scaled up across Uganda. The study presented in this submission takes advantage of the scaling up of the program to investigate the following two key questions: 1) Can the reduction in child mortality observed in the “proof-of-concept” study be sustained when the program is scaled-up? 2) 2) What is the impact of scaling up an incentivized community health worker program on existing health service providers?

This new study involves the same main actors of the first one: program implementers, data collection agency , and funding agency. This helps ensuring that the design, the management, and the implementation of the research program remains the same as in the first study. There are, however, also few important differences: the new study will measure treatment effects over a longer time period , it relies on a much larger sample (500 villages and more than 12,500 households), it exploits a much richer set of data, including survey data from other providers in the community, and it relies on a panel of households identified at baseline, rather than on a cross-section.

The results from this evaluation will allow for better understanding the long term effect of this innovative program for community health delivery, once it operates at scale, and will moreover inform the Ugandan authorities about the added value of a different CHW cadre in which health workers receive financial incentives. The evidence can also be used to inform other programs and guide other international organizations, governments and stakeholders to design effective programs for improving child health in areas underserved by the official health system.

Registration Citation

Citation
Awor, Phyllis et al. 2020. "Evaluation of Living Goods/BRAC entrepreneurial CHW model in Uganda - Phase II ." AEA RCT Registry. May 04. https://doi.org/10.1257/rct.2392-2.0
Former Citation
Awor, Phyllis et al. 2020. "Evaluation of Living Goods/BRAC entrepreneurial CHW model in Uganda - Phase II ." AEA RCT Registry. May 04. https://www.socialscienceregistry.org/trials/2392/history/67477
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details

Interventions

Intervention(s)
The CHP program is organized into geographically based branches, and managed by branch managers and supervised by the two NGOs. The CHPs are selected through a competitive process among female community members aged 18 to 45 who applied for the position in each village and who possessed basic writing and math skills. Eligible candidates receive 3 weeks of health and business training. At the end of the training, candidates need to pass a skills test in order to be equipped as an active CHP. The NGOs provide an initial set of products to all newly recruited CHPs, together with a uniform, a mobile phone, and a set of training materials and visual aids to use during household visits. CHPs also attend a one-day training each month to review and refresh key health and business topics.

The CHPs tasks mirror the standard Community Health Workers tasks (conduct home visits, educate households on essential health behaviors, provide basic medical advice, referring the more severe cases to the closest health center), but on top of that, as mentioned above, they also sell preventive and curative health products. The product line they have at disposal includes prevention goods (e.g. insecticide treated bednets, water purification tablets, and vitamins), curative treatments (e.g., oral rehydration salts, zinc, and ACTs), as well as other health-related commodities (e.g. diapers, hand soap, fortified food) and durables with health benefits (e.g. improved cook stoves, solar lights, and water filters). These products are sold by the CHP at a discount. The retail price is determined by the NGOs head office with a target of keeping prices for preventive and curative products about 20% lower than the prevailing local market prices. The CHPs in turn purchase these products directly from Living Goods or BRAC branches at wholesale prices between 30-50% below market prices and therefore earn an income on each product sold. Thus, the CHPs operated as micro-entrepreneurs with financial incentives to meet household demand. The broad product mix has three potential benefits: (i) driving up total sales and income for the CHPs; (ii) enabling the NGOs to cross-subsidize prices (dropping prices on essential health products and increasing the margins on other products); (iii) motivating CHPs to be out visiting households regularly by including high-velocity items (such as soap and fortified foods) in the product mix. The business training received by the CHPs stresses the importance of building up a customer-base by providing free services like health education, referrals, and newborn visits. As described above, the income deriving from the micro-entrepreneurial activity is then further increased through performance-based incentives, designed by the NGOs to further encourage key health activities such as household visiting, sick child assessment, registration and support of pregnant women, and visits to newborns in the first week of life. Since 2013, Living Goods and BRAC also equip the CHPs with smartphones that includes a rich mobile health application. The application helps guide the CHW through workflows, keep track of their stock, serve as a client management system, and prioritize certain activities based on timeliness (e.g. pregnancy follow-up) or household risk. Overall, this allows monitoring the CHPs’ activity, while collecting real-time health data from the field.
Intervention (Hidden)
Intervention Start Date
2016-05-09
Intervention End Date
2021-01-01

Primary Outcomes

Primary Outcomes (end points)
Question 1) Child mortality (Under-5, Infant, Neonatal)
Question 2) Number of drug shops and number of active CHWs operating in the study villages; quality of the drugs sold in the drug shops; level of interaction and type of activities carried out by the CHWs operating in the study villages
Primary Outcomes (explanation)
To assess the impact of the scaled-up program, in relation to the first research question, the primary outcome of interest is under-5 mortality. We will compute mortality at the cluster level using information contained in the household survey. The survey records: 1) detailed birth information on all children under five living in the households at the time of the survey; 2) detailed birth and death information on all children that died under the age of five during the study period.
At endline, for each child, we will define the number of month of exposure to the risk of death during the trial period, defined as the difference between the birth date of the child, or the start date of the trial if the child was born before that date, and the date that the child turned five years if that occurred during the trial period, or the date of the endline household survey if the child was less than five years old at that time, or the date of the death of the child. Under-five mortality will then be calculated as number of under-five deaths over the trial period per 1,000 child-years of exposure to the risk of dying under the age of five. We will also compute infant mortality as number of deaths during the trial period arising within the first year of life per 1,000 infant-years of exposure, with infant-years of exposure calculated in a similar way as the child-years of exposure described above. Finally, we will compute neonatal mortality as the number of deaths during the trial period within the first month of life per 1,000 births. All three measures will be defined at the village level.


To study how other health actors react to the scaling-up of the CHP program, in relation to the second research question, we will look at both the extensive and the intensive margins. We will start by studying the impact on the extensive margin. Here, we will focus on the number of drug shops and the number of active CHWs operating in the study villages at endline. These outcomes will be defined at the village level. We will identify as active CHW any CHW that carried out any CHW-related activity over the six months preceding the survey. Next, we will look at the intensive margin. Here, we will focus on the quality of the drugs sold in the drug shops, as well as the level of interaction and type of activities carried out by the CHWs operating in the study villages. In these case the outcomes will be defined at drug sample and CHW level, respectively.

Secondary Outcomes

Secondary Outcomes (end points)
Note: The following list of questions should be considered only indicative, as we are going to revise and edit the survey tools during piloting phase.

1) HH interactions with CHWs in general and Living Goods/BRAC CHP in particular
- HH visited by any CHWs/CHPs over the previous 30 days
- HH received any health service from the CHWs/CHPs (health products / education / diagnosis / referral / maternal care / follow-up visit)
- HH knows how to contact the CHWs/CHPs in the village


2) Health services
a. Household received follow-up health visits by any health staff…
- …following health-related problems with children under-5 (malaria, diarrhea, pneumonia) to specifically find out about child’s recovery
- …during pregnancy to monitor pregnancy
- …after delivery to check the mother and child health (If so, was the visit performed during the first week of life?)
b. Household received referrals to a health facility due to health-related problems with children under-5, or pregnancy
c. Pregnant woman received counselling and health recommendations…
- …on where to deliver
- …on medicines to take (Folic Acid, Iron and/or Vitamins / Malaria Prophylaxis / Deworming medicine)
- …on newborn feeding practices

3) Health Outcomes
a. Anthropometric measures for children under-5
- Height-for-age (expressed in z-scores)
- Weight-for-height (expressed in z-scores)
- MUAC-for-age (expressed in z-scores)
b. Malaria, diarrhea and pneumonia prevalence among children under-5
- Child fell sick with malaria in the previous 3 months
- Child fell sick with diarrhea in the previous 3 months
- Child fell sick with pneumonia in the previous 3 months
c. Share of miscarriages and stillbirths during the study period
d. Unmet need for family planning and unwanted pregnancies

4) Health Knowledge
a. Respondent knowledge concerning causes and treatment of malaria, diarrhea, and pneumonia
- Respondent believes mosquito bites are the only cause of malaria
- Respondent believes one can make environmental changes to prevent malaria
- Respondent knows Zinc can be used to treat diarrhea
- Respondent knows diarrhea can be transmitted by drinking un boiled/untreated water?
b. Respondent knowledge concerning nutrition and breastfeeding practices
- Respondent knows about vitamins & added nutrients
- Respondent knows colostrum is healthy
- Share of correct answers on a short case study presented to the respondent, which compares health evolution of two children treated differently in terms of nutrition and breastfeeding.
c. Respondent knowledge concerning family planning
- Respondent knows about family planning methods (share)

5) Household Health Behavior
a. Household standard prevention and treatment practices for diarrhea, malaria, and pneumonia
- Respondent washes hands with soap most of the time
- HH treats malaria with ACT drugs
- HH treats pneumonia with antibiotic
- HH treats diarrhea with ORS and Zinc
b. Household food consumption habits
- Child has varied diet (based on number of different food categories consumed the previous date, obtained from a detailed food consumption section)
c. Ante-natal and post-natal care practices, including breast-feeding practices
- Women sought ANC at least 4 times
- Woman fed newborn within 1hr of birth
- Woman fed baby non-breastmilk fluids after 6 months
- Woman during pregnancy took Folic Acid / Iron and/or Vitamins / Malaria Prophylaxis / Deworming medicine
- Woman took Vitamin A and/or folic acid in first two months after delivery
- Woman gave birth outside a health facility
- Woman devised a birth plan

6) Community Health Workers knowledge and activity
a. Level of satisfaction and confidence of health workers
- First principal component from on a set of questions related to satisfaction (e.g. “I am satisfied with the community thanks and recognition I receive for my work”)
- Self-reported level of confidence that the CHW provides correct advise and/or treatment services for the community
- Revenues as health worker
b. Level of CHW turnover (village level variable)
- Measure constructed using information provided by CHWs on when they joined the organization and whether / how often the organization changed workers
c. Knowledge of health workers concerning malaria, diarrhea, and nutrition
- CHW believes mosquito bites are the only cause of malaria
- CHW believes one can make environmental changes to prevent malaria
- CHW believes combination ORS & Zinc can treat diarrhea
- CHW knows correct signs/symptoms of Pneumonia and Malaria
- Share of correct answers on a short case study presented to the respondent, which compares health evolution of two children treated differently in terms of nutrition and breastfeeding.
- CHW knows which food contains more protein value than others
- CHW knows danger signs during pregnancy
d. Level of self-reported activity of the health workers
- Days worked as CHW if last 30 days
- Number of Activities in last 30 days (Pregnant women visited; Newborn babies visited; Children < 5 years visited; Family planning visits; People tested for malaria; People treated for malaria; Patients referred to health center)
- Number of Health forums / health education campaigns held

7) Drugs availability
a. Number of drug stores that opened (closed down) during the study period (village level variable)
b. Drug store provided the appropriate medicine to treat the disease (malaria and pneumonia)
c. Price of the drugs sold by the store
Secondary Outcomes (explanation)
Concerning the first research question, the secondary outcomes will serve to investigate the following secondary hypotheses:
1.1) The program increased the chances that a household interacts with and benefits from services provided by the CHPs;
1.2) The program increased the overall amount and quality of health services received by households;
1.3) On top of the impact on child mortality, the program improved additional health outcomes, related to family planning, pregnancy, newborn and child health;
1.4) The program improved the basic health knowledge of the households;
1.5) The program improved the health behavior (both preventive and curative) of the households;
Concerning the second research question, the secondary outcomes will serve to investigate the following secondary hypotheses:
2.1) The program increased the (average) satisfaction, motivation, and confidence of the CHWs operating in the village;
2.2) The program lowered the turnover of the CHWs in the village;
2.3) The program increased the (average) health knowledge of the CHWs operating in the village;
2.4) The program increased the amount of (self-reported) activities of the CHWs operating in the village;
2.5) The program impacted the supply of drugs in the community, by reducing the number of drug stores operating in the local markets, raising the quality of their service, and lowering the price of the drugs;

Experimental Design

Experimental Design
The study is a stratified cluster randomized controlled trial embedded in the scale up of the CHP program. 500 trial clusters, organized into 15 zones which span 13 districts in four regions of Uganda, will participate in the trial. Each trial cluster is one village (LC1). Each zone contains villages that are assigned to one Living Goods or BRAC branch. Within each of the 15 zones, the villages are randomly divided into an intervention group and a control group (randomized block design), using balanced randomization. None of the villages included in the study has been exposed to the CHP program before. CHPs will be assigned to each village in the treatment group, while no CHP will be assigned to villages in the control group.
Experimental Design Details
Randomization Method
Random number generator on a computer.
Randomization Unit
The unit of randomization in the study is a village. Within each of the 15 study zones, the villages are randomly divided into an intervention group and a control group (randomized block design), using balanced randomization.
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
500 villages
Sample size: planned number of observations
12,500 households
Sample size (or number of clusters) by treatment arms
The sample is equally divided to 250 treatment and 250 control villages. From each of these villages, 25 households will participate in the survey (a total of 6,250 households in the treatment and 6,250 in the control villages).
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
The sample size was designed to detect a reduction in under-5 mortality (primary outcome of interest), defined as number of under-5 deaths per 1,000 child-years of exposure to the risk of death under the age of 5. We used data from the control group in the proof-of-concept study conducted by the research team in similar settings (Björkman Nyqvist et al, 2019) to obtain the relevant inputs for the computation. A total sample of 500 clusters (250 per study arm) and 25 households per cluster at baseline (12,500 households in total) allows us to detect a reduction in child mortality of 20% or larger, at the 5% significance level with 80% power, assuming between-cluster coefficient of variation equal to 0.43 and attrition rate of 16% (or 4 households per cluster). Under the same assumptions, this design and sample size also has 80% power to detect a 21% reduction in infant mortality and a 25% reduction in neonatal mortality at the 0.05 significant level.
IRB

Institutional Review Boards (IRBs)

IRB Name
IPA Institutional Review Board
IRB Approval Date
2015-10-15
IRB Approval Number
13774
IRB Name
Mildmay Uganda Research Ethics Committee
IRB Approval Date
2015-09-21
IRB Approval Number
0109-2015
IRB Name
Uganda National Council for Science and Technology
IRB Approval Date
2015-10-15
IRB Approval Number
SS 3938
Analysis Plan

Analysis Plan Documents

pre-analysis+plan.pdf

MD5: c6d920542f73132cf1e019ef2f6129e7

SHA1: 0645f7c5336116c96981f83fdab78bcc17814223

Uploaded At: May 04, 2020

Post-Trial

Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials