Expanding Access to Safe Water in Nigeria

20 communities in the treatment group will receive community demonstrations and redemption points will be set up, as follows:

(i) Community Demonstrations: Facilitators will collaborate with local leaders, WASH officials and health facility staff in conducting community demonstrations introducing the chlorine product (WaterGuard+) to communities in the treatment group. Local leaders in the community will organize the meeting, while facilitators in the country will demonstrate how to use the product, what the correct dosage is, and explain the importance of water treatment. They will also show how village water is contaminated with Ecoli. These meetings will be targeted at eligible women (pregnant women and women with children under the age of 5), however all members of the community will be invited to attend. At the meeting, each attendee will be given one bottle of WaterGuard+, which is enough to treat drinking water for a household of 5 members for about a month, and 1,000 naira for their time. Water Guard + is a 180g bottle of sodium dichloroisocyanurate (NaDCC) in granulated / powder form, produced and distributed by the Society for Family Health (SFH) in Nigeria.

(ii) Redemption points: Study staff will set up at least one chlorine redemption point for each community. The location of the redemption point will be determined in consultation with the local community during activity (i). The redemption points could be health facilities, local shops/dispensaries/kiosks, local pharmacies, house of the village leader, CHW, or health staff or other suitable sites. Households will be able to visit the location and redeem a month’s supply of a chlorine product. Distributors will use a redemption log to keep track of when participants redeem the chlorine product.

2025-10-11

2025-11-16

Treat drinking water with chlorine (tested in water sample)

Proportion of water samples with detectable chlorine in stored drinking water, measured up to 6 months of follow-up

1. Treat drinking water with chlorine (self-reported)
2. Presence of E.coli
3. Level of E.coli
4. Unsafe risk of Coliform
5. Heard about chlorine
6. Ever used chlorine
7. Last used chlorine

All secondary outcomes are measured up to 6 months of follow-up
1. Proportion of women who reported using chlorine in the previous two days
2. Proportion of water samples with detectable E.coli
3. Level of E-coli (in log10 MPN/100ml) in water samples
4. Proportion of water samples with unsafe coliform ( >100 colony forming units per 100mL)
5. Proportion of women who ever heard about chlorine
6. Proportion of women who ever used chlorine
7. Proportion of women who used chlorine yesterday or the day before

A list of 30 communities will be compiled from the following Local Government Areas (LGAs) in Kano, Nigeria: Kura, Dawakin Kudu, Kiru, and Rano. Criteria for selection will include, but is not limited to, distance, accessibility and buy-in from local leaders. Once a refined list has been compiled, we will randomly select 20 communities to receive the intervention (treatment) and 10 communities to function as a control (these 10 communities will then go on to also receive the intervention at the end of the data collection period).

Intervention: 20 communities in the treatment group will receive community demonstrations and redemption points will be set up (details described below).

Control group: 10 randomly selected control communities will initially receive no intervention. After three months, five of these control communities will receive an individual-level intervention, in which households are given chlorine vouchers and sensitized individually, rather than through community meetings. The other five control communities will continue to receive no intervention during the study period. This design will allow us to compare the effectiveness of group-based versus individual-level sensitization strategies in promoting chlorine use. The objective of this additional randomization within the control group is to test the hypothesis that individual-level randomization, often preferred to maximize power, can lead to underestimation of treatment effects in contexts where community-level encouragement is necessary.

Stratification
Within each ward (10) we randomly sampled 3 communities and allocated them in a ratio 2:1 for treatment and control.

Enrollment
We enroll pregnant women or women with children under the age of 5. We followed a random-walk procedure where enumerators start at the center of the community and split in different directions. The enumerator sampled every 5th household if the local leader reported that the population in his community was below 500 households, and every 8th household if the local leader reported that the population was above 500 households. With support from local leaders and Community Health Workers (CHWs), at baseline, we plan to enroll 10 women in each treatment community and 10 women in each control community for a total of 300 hhs (200 in treatment, 100 in control communities) to assess baseline rates of chlorine usage and other water practices.. At the endline we plan to enroll 10 women in each treatment community and 20 women in each control community for a total of 400 hhs (200 in treatment communities and 200 in control communities). We additionally asked women at baseline for contacts of 5 neighbors. Some of them may be enrolled at the endline.

Not available

Stratified randomization of communities was conducted using the R. The first step involved randomly selecting ten wards (two in Kiru and Rano and three in Kura and Dawakin Kudu), excluding wards which are inaccessible or too far (over three hours drive away). Then, we randomly selected 3 communities within each ward (excluding communities with less than 100 households and non-sedentary communities), two of which were randomly allocated to treatment and one to control. We also randomly selected back-up wards and communities where possible.

Communities

Yes

30 communities

At baseline: 300 women 10 women X 20 treatment communities = 200 women in treatment 10 women X 10 control communities = 100 women in control At endline: 400 women 10 women X 20 treatment communities = 200 women in treatment 20 women X 10 control communities = 200 women in control

At baseline: 30 communities (20 treatment, 10 control, 10 women per each) for a total of 300 women
At endline: 30 communities (20 treatment - 10 women in each community, 10 control - 20 women in each community) for a total of 400 women.

We conduct power calculations for the primary outcome — the proportion of water samples with detectable chlorine in stored drinking water at follow-up. Using a cluster-randomized design with 30 communities (20 treatment, 10 control) using upper bounds (to be conservative) from our pilot work in 14 communities in Kano so far, we assume the following: Baseline (control group) chlorine use rate: 5% Intra-cluster correlation (ICC): 0.05 10 women per treatment community, 20 women per control community Based on these parameters, comparing treatment and control communities, the study can detect an MDE of 12.77 percentage points in chlorine use at 5% significance level with 0.8 power. Turning to the within-community, individual-level randomization, with 100 households in each group (100 households individually randomized to receive vouchers in 5 control communities and 100 households individually randomized to remain pure control groups in the other 5 control communities), the study can detect an MDE of 12.4 percentage points in chlorine use.