The Far North Region is one of the 10 regions of Cameroon, and is divided into six administrative departments. The project was implemented across four contiguous communes out of the seven communes within the department of Mayo-Kani, where the government identified gaps in coverage with prevention and treatment services. The study was conducted in the commune of Kaélé, which has 14 health areas, a total population of 25,559, and is served by about 70 community health workers (CHW). Each health area encompasses several villages or towns and health facilities (hospitals, clinics, health posts). Agriculture is the economic mainstay, with most people cultivating cereals, mainly sorghum.

The study used a pre-post quasi-experimental design to implement a longitudinal cohort study, following and assessing a randomly selected subset of children enrolled in the FVP every 2 weeks from their enrollment in the program until recovery and exit. Children were enrolled in the study with a confirmed MUAC measure ≥ 115 and ≤ 125 mm, and were considered recovered with a MUAC measure > 125 mm. Since previously implemented standard treatment of ready-to-use supplementary foods in the region was no more available, it was not possible to include a comparison group. The study took place between January and August 2020.

Eligible children were identified through a community census and mass screening, during which MUAC of all the children aged 6–59 months in each village or town of the included health districts was measured by the responsible CHW. Children identified as moderately or severely wasted were referred to the nearest health facility, where a public health nurse did further assessments to confirm the diagnosis. Children confirmed as being severely wasted were treated at health facilities following government protocol. Children confirmed with moderate wasting were admitted by health workers into the food voucher program. Children from a randomly selected household were enrolled in the research study following informed consent by the primary caretaker. In households with more than one eligible child, the youngest child was selected for the study. Children were excluded from the study if severely wasted, severely ill, or if the caretakers refuse to give their approval although these children still received the food voucher packages from the FVP program.

Once their child was enrolled in the food voucher program by a health facility nurse, caretakers received a paper voucher with a value of 8000 FCFA (~ US$15) they could redeem for a standardized basket of food items from a designated local vendor (Fig. 1). The content of the basket was defined using NutVal software to provide sufficient calories and micronutrients to supplement the usual diet of children and support recovery from moderate wasting, and to provide a small surplus in anticipation of some sharing of the foods within the household (Table 1). The nurse provided instructions to caretakers on the shop or vendor where the voucher was to be redeemed, the composition of the food basket to be received, and to add the ingredients to supplement, not replace, the child’s usual meals. Caretakers also participated in discussion sessions on essential nutrition and hygiene actions, as well as cooking demonstrations of potential recipes to prepare with the voucher foods. These messages were reinforced during bi-weekly home visits by the local CHW to monitor appropriate use of vouchers. Caretakers were instructed to return to the health center with the child every 2 weeks for a health examination and to receive the next voucher.

A sample of food basket to be redeemed by an eligible caretaker in Mourbaré district

During the program design phase, information gathered through market assessments in the target areas verified that food basket ingredients were readily available in local markets and included a list of local vendors willing and eligible to participate in the program. Eligibility criteria for vendors included their willingness to sign a contract of participation, stock the food items to be redeemed by the voucher and being current with payment of government taxes (including having a tax ID number). The vendors received reimbursement for the vouchers they redeemed every 2 weeks. Monitoring data of voucher redemption and utilization for feeding of the enrolled child indicated that 95% of all vouchers were indeed claimed with vendors.

Enumerators received 3 days of training on project objectives, questionnaire administration, and anthropometric measurement. Two structured electronic questionnaires (baseline and bi-weekly follow-up) were pretested in French and local dialects with 33 caregivers in neighboring health districts. Each caretaker was explained the nature and objectives of the study and possible risks associated with their participation and provided informed consent before enrollment. Consent was obtained at each follow-up visit as well. Caretakers were allowed to have their questions answered or to refuse any part of the study procedure, or question.

A research team visited the homes of each enrolled child just before their receipt of the first voucher, and every 2 weeks for the duration of the child’s treatment. The baseline survey covered household demographic and socioeconomic characteristics gender-disaggregated asset ownership, agricultural production, food security, health knowledge, child’s recent illness and treatment, and infant and young child feeding knowledge and practices. A 24-h dietary recall was administered by the research team to the caretaker of each enrolled child before entry into the voucher program and at each bi-weekly follow-up visit. Data were collected using the ODK application (www.getodk.org) installed on smartphones and stored on the ONA server (www.ona.io). Local translators were employed to translate the interview from French into the appropriate local language.

The children enrolled in the program were measured every 2 weeks until their exit from the program. In addition to MUAC, three replicates of weight and height/length were recorded at each visit and the mean value calculated. Weight was measured using an electronic scale in kilogram to the nearest decimal fraction with the child standing with both feet in the center of the scale. Children unable or unwilling to stand were weighed in the caretaker’s arms, and the caretaker’s weight was subtracted. All values were reported to the nearest 0.1 kg. Height measurements were reported in centimeters to the nearest decimal fraction using a stadiometer. Children ≥ 2 years and/or ≥ 85 cm were measured for standing height; children < 2 years and/or < 85 cm were measured in the prone position. Date of birth of children was recorded from health card, immunization card or birth certificate to the nearest 0.1 month.

Children were defined as having recovered when their MUAC was measured as > 125 mm any time of six bi-weekly visits. Those who completed 3 months of treatment without recovery, and those deteriorating during the study period into severe wasting with or without bilateral pitting edema were referred for standard treatment according to the national protocol; specifically, referral to out-patient clinics for further assessment and treatment for SAM. Those who did not return to the health facility for two consecutive treatments were considered by the study as defaulted, although CHW continued to try to reach them to ensure they received appropriate care.

Based on the total population of 25,199 and 8.3% prevalence of moderate wasting (SMART, 2018) in Kaélé, Far North Region [15], a minimum sample size of n = 456 children ensured that a one-sided test with a significance level of 0.05 had 0.90 power to detect a mean difference of 3 mm increase in MUAC by 6 weeks of FVP, assuming a common standard deviation of 15 mm, a Pearson correlation coefficient of 0.01, 2,540 as finite population correction, and a non-response rate of 25% [16]. For the selection of children, a sampling frame was used consisting of the list of confirmed MAM children from the 14 health areas of Kaélé stratified by age and gender, with selection probability proportional to the size of each health area. Although children were also examined biweekly at health centers to review their nutritional status and provide the next voucher per the evaluation, those data were not included in the analyses presented here.

Proportions, means (standard deviations), and medians (interquartile ranges) were calculated for key baseline variables for households, mothers, and children. Time to recovery was evaluated with a univariate Cox proportional regression hazard model with associations quantified using hazard ratio (HR) with 95% confidence interval (CI). Covariates significant at ≤ 0.20 in a univariate regression were included and tested in a multivariate Cox proportional hazard model. The trend for MUAC, including its determinants, was studied at univariate level with simple linear fixed and random (mixed) effects models and at multivariate level with multiple linear fixed and random (mixed) effects models, the random variable being the visit. The variables assessed included child anthropometry (MUAC in mm, height in cm, and weight in kg), and demographic and socio-economic characteristics of the caretaker and households. Two-sided p values < 0.05 were considered statistically significant in the multivariate models. Data analysis was performed using Stata 16 (StataCorp, 2019) and IBM-SPSS 26 (IBM Corp, 2019).