Impact Evaluation of Community-Driven Nutrition Interventions in Kailahun District, Sierra Leone

2018-11-19

2020-10-12

Household module:
Handwashing and household water purification practice
Access to handwashing in
Child mortality

Women module:
Antenatal and postnatal care - institutional delivery, skilled birth attendance, and newborn care

Under-5 module
Child feeding practices
Minimum dietary practices
Immunization rates
Care provided for diarrhea and fever

Anthropometry module
Under-five children physical measurements (height & weight)

Both treatment and control groups center around a selected 81 healthcare facilities in the Kailhun district with higher-than-average stunting rates. 41 of these 81 facilities were further randomly selected as the treatment group to receive a comprehensive package of community-based nutrition interventions coordinated largely by the MSGs and CHWs. The remaining 40 facilities and surrounding areas were randomly selected to act as a control.

The Sierra Leone Multi-Indicator Cluster Survey Round 6 (MICS6) from Statistics Sierra Leone and UNICEF is used as the baseline, and IPA has been contracted by the World Bank to conduct an endline surrounding all 81 facilities. As part of this endline, surveys will be conducted within each health facility with administrators and CHWs, within households, and within communities. In addition, qualitative focus group discussions (FGDs) and in-depth interviews (IDIs) will take place with CHWs and households, stratified by gender.

Household Surveys.
About 33 households with at least one child under age 5 will be randomly selected in the catchment area of the 81 facilities for 2,662 households in total. Sampling for the household survey will take place in two stages. First, the households to be sampled per community will be divided across each of the villages in the catchment area using a population-proportionate method. Within each village, households will then be selected using a “random walk” method which involves a guide-assisted mapping exercise, randomly selected starting point, and household selection pattern which varies based on the village layout (i.e., whether households are clustered or along a linear path). In the absence of census data, this two-stage sampling method ensures selected households will be representative of the overall catchment areas. Enumerators will collect basic information from each contacted household through the random walk process, taking on average no more than 5 minutes per household to assess household eligibility, basic demographic, and re-contact information. Heads of households who satisfy the eligibility criteria of having at least one child under 5 will then be administered the full household survey questionnaire. In addition, these children under 5 will have their height and weight measured by their caretaker.

Community Survey.
A community survey will cover 81 community leaders in the 81 sample clusters and will take 60-90 minutes to implement. The survey will be collected through interviews with one community leader or key respondent in each community identified upon community entry.

Focus Group Discussions (FGD)
Sixteen FGDs will be conducted from a random sampling, synchronizing with the selection of households for the household survey. Each FGD will have a maximum of 6 persons. There are 10 categories of the target populations for FDGs as shown in the table above and include women with children under 5 who belong to MSGs or are male/female CHWs who interact with MSGs.

In-Depth Interviews (IDI).
Twenty-four IDIs will be conducted from a random sampling, synchronizing with the selection of households for the household survey as the above. There are 10 categories of the target populations for IDIs as shown in the table above, though like the FGDs, the respondents here are women with children under 5 who belong to MSGs or are male/female CHWs who interact with MSGs.

Health Facility Survey.
One lead administrator or manager per facility will be surveyed for a total of 81 in total. It may be the case that this administrator is unable to answer specific questions, and in those instances, subject matter experts, such as accountants or lab technicians may be sought out to answer these questions. One community healthcare worker per facility will also be surveyed, for a total of 81 in total. The selection of these CHWs will be aided by the facility administrator. Facility entry will be supported and guided by our partners at the Ministry of Health and Kailahun District Medical Management Team.

Protocols Within Communities
Compound Selection

Before starting data collection, the team leader needs to randomly select compounds for the enumerator using the following procedure:

For each EA, the team leader should work with the EA maps, village leader, or other significant community stakeholders to identify the boundaries of the community. If possible, find a local guide who can accompany you. He/she can introduce you to the families and help you find the households and the community boundaries.

Starting from the right end of the community, the team leader numbers each compound/ dwelling unit in sequential order. As much as possible the team leader must move in a pattern that is easy to predict. The team leader must only label compounds that are currently occupied and are used for residential purposes. Public facilities or buildings that are used for commercial purposes only such as supermarkets, shops or hotels should not be numbered.

The team leader should number each of the compounds with chalk and indicate the direction in which the team leader will move next. Numbering for the first compound in EA must start with the EA shortcode followed by “-” and then the compound number. eg. SIE-001 and continue by increasing the number by one until the last compound is chalked. For instance, if the last compound is 123, then the number on the compound must be SIE-123.

On the tablet, open the “WB Nutrition Project - Randomization” form on SurveyCTO. Fill out the form by specifying the appropriate EA details. Enter the total number of compounds/dwellings in the EA.
The survey form will display a list of numbers in random order. This random order is the order in which the compounds should be enumerated.

The team leader will assign the numbers to the enumerators who will then locate the compounds that were chalked with the number assigned. Enumerators will then proceed with the household selection protocol.

The team leader must keep track of the number of households that are interviewed per compound. ie. households that had at least 1 child under 5 years and consented to surveys.
If the total number of household interviewed in EA are 33 and above: The team has completed its interviews in the EA.
If the total number of households interviewed in the EA is less than 33: The team leader must continue to assign compounds in the specified random number until the number of eligible households interviewed in the EA is 33 households or the team has fully exhausted all the compounds in the EA.





Household Selection
Before starting any data collection within a compound, the enumerator must identify the various households within the compound they have been assigned.
The surveyor must find an adult member of the compound and administer the screening form to that household member.

With the help of this adult compound member, the enumerator will identify and number each household within the compound. For e.g. Household 1 & 2. If there is only 1 household within the compound, then that household automatically assumes the number 1.

In the household screening form, the surveyor will capture the name and gender of each household head in the order in which they were numbered earlier.

For each household, the surveyor will verify from the adult compound member if there is at least 1 child under 5 years in the household and fill out the form accordingly.
If there is at least one child under 5 years of age in the household: The surveyor will Administer a separate household survey to each household with at least 1 child under 5 years. The surveyor will then Administer other surveys to households once the household survey is complete.
If there is no child under 5 years of age in the household: The surveyor will move to the next compound assigned and repeat the household selection criteria or report back to the team leader to be re-assigned.

IDI RESPONDENTS
Please note the following for the IDI:
IDIs will only be conducted in 24 pre-selected EAs.
Each EA selected will be pre-assigned criteria for identifying the IDI respondent.
Only one IDI will be conducted in each pre-selected EA.

Category - Randomization Protocol
Women who have a child or children under 5 and belong to an MSG as a member
In-field randomization of MSG Lead Mother from a list of MSG lead mothers provided by the PHU
In-field randomization of MSG Members from a list of MSG Members provided by selected MSG lead Mother

Women who have a child or children under 5 and belong to an MSG as a Lead Mother
In-field randomization of MSG Lead Mother from a list of MSG lead mothers provided by the PHU

Women who have a child or children under 5 and not belong to an MSG (either an MSG doesn’t exist in her village or she chooses not to join)
In-field Randomization using Quantitative Sample
Female CHWs who interact with MSGs in their catchment communities In-field Randomization using List of CHWs from PHU

Male CHWs who interact with MSGs in their catchment communities
In-field Randomization using List of CHWs from PHU

Category 1: Women who have a child or children under 5 and belong to an MSG as a member
To identify the IDI respondent, the IDI enumerator will rely on a list of MSG members that will be provided by the lead mother. The lead mother will also be randomly selected from the list of lead mothers in the PHU.
The IDI enumerator contacts the PHU for a list of Lead Mothers in the EA. If the EA serves multiple villages, then the IDI enumerator will only request names of MSG Lead mothers from the village in which the PHU facility is located.

The IDI enumerator will open the “Respondent Randomization Form” on SurveyCTO and with the help of the health worker input the names of all the MSG Lead mothers in the target village or EA.

The “Respondent Randomization Form” will randomly select and display the names of MSG Lead Mothers. Please note that this list is specified in the order in which the respondents will be contacted.
The IDI enumerator will contact the chosen Lead Mother to get a list of MSG members under her.

If the IDI enumerator is unable to interview the first MSG Lead Mother on the list, they will interview the second MSG Lead Mother on the list.

The IDI enumerator will continue on the “Respondent Randomization Form” on SurveyCTO and with the help of the MSG Lead Mother input the names of all the MSG Members in the target village or EA.

The “Respondent Randomization Form” will randomly select and display the names of 3 MSG Members. Please note that this list is specified in the order in which the respondents will be contacted.

If the IDI enumerator is unable to interview the first MSG member on the list, they will interview the second MSG Lead Mother on the list.

Category 2: Women who have a child or children under 5 and belong to an MSG as a Lead Mother
To identify the IDI respondent, the IDI enumerator will rely on a list of MSG Lead Mothers that will be provided by the PHU.
The IDI enumerator contacts the PHU for a list of Lead Mothers in the EA. If the EA serves multiple villages, then the IDI enumerator will only request names of MSG Lead mothers from the village in which the PHU facility is located.

The IDI enumerator will open the “Respondent Randomization Form” on SurveyCTO and with the help of the health worker input the names of all the MSG Lead mothers in the target village or EA.

The “Respondent Randomization Form” will randomly select and display the names of 2 MSG Lead Mothers. Please note that this list is specified in the order in which the respondents will be contacted.

The IDI enumerator will contact the chosen Lead Mother to get a list of MSG members under her.

If the IDI enumerator is unable to interview the first MSG Lead Mother on the list they will interview the second MSG Lead Mother on the list.

Category 3: Women who have a child or children under 5 and not belong to an MSG (either an MSG doesn’t exist in her village or she chooses not to join)
To identify the IDI respondent, the IDI enumerator will trail the household enumerators in the order of randomization and identify respondents using the following procedure.
The IDI enumerator goes to the first randomly chosen compound for the Quantitative interviews.

For each compound, the IDI respondent will verify that there is at least 1 compound member who meets the criterion that was assigned to the EA. ie. “Women who have a child or children under 5 and not belong to an MSG (either an MSG doesn’t in her village, or she chooses not to join)”. Then the IDI enumerator will identify all Women who have a child or children under 5 and not belong to an MSG in this compound.

If there is at least 1 qualified compound member:
The IDI enumerator will open the “Respondent Randomization Form” and enter the names and household numbers of each of the qualified compound members.
The “Respondent Randomization Form” will randomly select one of the pre-identified IDI respondents to be interviewed.
The IDI enumerator will move to the next preselected EA after completing the interview.
If the selected respondent declines to take part in the interview, then the IDI respondent will move to the next compound and repeat steps 1 - 3 to select a new respondent
If there is no qualified IDI respondent in that compound:
The IDI enumerator will move to the next compound and repeat steps 1 - 4 until a qualified respondent is identified for the IDI.
If all the compounds are exhausted without securing a respondent for the IDI Survey, the IDI enumerator will contact the project staff to be re-assigned a backup EA.

Category 4: Female CHWs who interact with MSGs in their catchment communities
To identify the IDI respondent, the IDI enumerator will rely on a list of CHWs that will be provided by the PHU.
The IDI enumerator contacts the PHU for a

Health Facilities

Yes

81 Health Facilities

Quantitative study: 33 household per health facility, 81 health facility staff, 162 community health workers CHW, 81 community stakeholders Qualitative study: 24 In-depth IDI interviews and 16 Focus group discussion FGDs

Health Facilities 41 Treatment and 40 Control