A local-ingredients-based supplement is an alternative to corn-soy blends plus for treating moderate acute malnutrition among children aged 6 to 59 months: A randomized controlled non-inferiority trial in Wolaita, Southern Ethiopia

Background Globally, moderate acute malnutrition (MAM) affects approximately 5% of children below five years of age. MAM is a persistent public health problem in Ethiopia. The current approach in Ethiopia for managing MAM is a supplementary feeding program; however, this is only provided to chronically food-insecure areas. The objective of the study was to compare a local-ingredients-based supplement (LIBS) with the standard corn-soy blend plus (CSB+) in treating MAM among children aged 6 to 59 months to test the hypothesis that the recovery rate achieved with LIBS will not be more than 7% worse than that achieved with CSB+. Methods and findings We used an individual randomized controlled non-inferiority trial design with two arms, involving 324 children with MAM aged 6 to 59 months in Wolaita, Southern Ethiopia. One hundred and sixty-two children were randomly assigned to each of the two arms. In the first arm, 125.2 g of LIBS with 8 ml of refined deodorized and cholesterol-free sunflower oil/day was provided. In the second arm, 150 g of CSB+ with 16 ml of refined deodorized and cholesterol-free sunflower oil/day was provided. Each child was provided with a daily ration of either LIBS or CSB+ for 12 weeks. Both intention-to-treat (ITT) and per-protocol (PP) analyses were done. ITT and PP analyses showed non-inferiority of LIBS compared with CSB+ for recovery rate [ITT risk difference = 4.9% (95% CI: -4.70, 14.50); PP risk difference = 3.7% (95% CI: –5.91, 13.31)]; average weight gain [ITT risk difference = 0.10 g (95% CI: -0.33 g, 0.53 g); PP risk difference = 0.04 g (95% CI: -0.38 g, 0.47 g)]; and recovery time [ITT risk difference = -2.64 days (95% CI: -8.40 days, 3.13 days); PP difference -2.17 days (95% CI: -7.97 days, 3.64 days]. Non-inferiority in MUAC gain and length/height gain was also observed in the LIBS group compared with the CSB+ group. Conclusions LIBS can be used as an alternative to the standard CSB+ for the treatment of MAM. Thus, the potential of scaling up the use of LIBS should be promoted. This study was approved by the Hawassa University College of Medicine and Health Sciences Institutional Review Board (IRB/024/10) and Regional Committees for Medical and Health Research Ethics in Norway (2018/69/REK vest). The approval obtained from the institutional review boards covered all sites included in the study. The purpose of the study and methods of data collection, confidentiality, and voluntary participation were explained to the mothers of children who were invited to sign an informed consent form. Verbal (getting thumb marks after reading the information) and written informed consent were obtained from all caregivers of children who met enrollment criteria before the recruitment of their children into the study. All interviews and intervention procedures were conducted in privacy. This trial was registered at Pan-African Clinical Trial Registration as PACTR201809662822990. The authors confirmed that all ongoing and related trials for these food supplements were registered. We conducted the study in Damot Pulassa district in the Wolaita zone in the South-Western part of Ethiopia, where there is a high level of food insecurity and child malnutrition. Damot Pulassa is characterized by having fragmented farm and land ownership. Damot Pulassa has the highest population density in Ethiopia reaching up to 600 persons per square kilometer in some Kebeles (smallest administrative units) of it. The discrepancy between population and land balance has by far continued to be the primary cause of endemic food insecurity in the area [21]. According to a study protocol of this trial, Damot Pulassa district is characterized as maize and root crop livelihood area as these are the main crops cultivated in the district [5]. This district has five health centers and 23 health posts (one per kebele) that are led by health extension workers. These health facilities deliver nutrition-linked services like counseling on feeding practices, screening for nutritional status of young children, and nutritional management of acutely malnourished children. Damot Pulassa district (Woreda) was purposely selected based on a consideration of the high level of food insecurity, high level of child malnutrition, and access to transportation. All the kebeles in the district had similar admission loads to the management program of MAM, for practical reasons, six out of the 23 kebeles were randomly selected for the study. Children aged 6 to 59 months with MAM, living in selected kebeles of Damot Pulassa district were included in the study. Children were excluded if they had SAM (based on WHO 2009 child growth standards), and/or if they had bilateral pitting edema [22], or had any illness or other medical complications that prevented the children from safely consuming supplementary food. The trained research teams and HEWs assessed the children for SAM. Medical complications of children were assessed by the senior pediatric nurse working in the below five years clinic. Children were also excluded if they were simultaneously involved in another supplementary feeding program. Recruited children were identified as MAM according to their MUAC values, and not only according to the values of weight-for-height/length z-scores. The respondents were mothers or caregivers of selected children. This study was a randomized, controlled, non-inferiority trial that assessed the efficacy of 125.2 g of LIBS with 8 ml of refined deodorized and cholesterol-free sunflower oil/day (the intervention), compared with conventional treatment, which is CSB+ in the amount of 150 g of CSB+/day with 16 ml of refined deodorized and cholesterol free-sunflower oil (the control), in treating MAM for 12 weeks. The primary outcome was recovery rate: percentage of children who attained a MUAC ≥12.5 cm and/or WHZ ≥ -2 without bipedal edema at the end of 12 weeks. The secondary outcomes were the mean recovery time (the duration within 12 weeks in which the child recovered from MAM) and average weight gain. Children who progressed to SAM during the study or persisted as moderately malnourished at the end of the 12-week follow-up were considered to have failed management for MAM. The selection of outcome measures was based on similar studies [15, 20, 23]. There were two research teams, each team consisted of one supervisor, six data collectors, and eighteen food distributors (selected females living in the study area). The supervisors and data collectors were trained by the principal investigator for seven days before the start of study. The training covered the objectives of the study, data collection systems, questions found in the questionnaire, interview techniques, and anthropometric measurements. Training in anthropometric measurement techniques, periodic standardization, and daily calibration of equipment was done based on WHO recommendations for anthropometric measurement protocols [24]. Before study start, the female food distributors were trained for five days on a daily supplement distribution, cooking of supplementary foods such as porridge and feeding of a child. When the child was a twin, we provided an additional amount of supplementary food to the caregiver to ensure that the enrolled child was fed with a full portion. When two children with MAM were found in the same household, we provided similar supplements for both children but only the randomly selected child was taken as a study subject. When the recruited child was not at home during the time of follow-up, data collectors and food distributers revisited such households until they found the child. Supervisors oversaw activities daily. The food distribution process, feeding techniques, and use of the provided food supplements were monitored among randomly selected households on a twice-monthly basis. We calculated the sample size to investigate if LIBS was not inferior to CSB+ in terms of the recovery rate among children with MAM aged 6 to 59 months. The sample size was calculated based on 80% power of the test, 7% margin of non-inferiority, and assuming a recovery rate with CSB+ of 67%. Considering the above assumptions, a total of 324 children (162 subjects per study group), allowing for a 10% withdrawal rate were required to be sure that the lower limit of a two-sided 95% confidence interval (CI) was above the -7% margin of non-inferiority. The anticipated 10% dropout rate was used based on the observed dropout rate reported in two studies [11, 24]. The non-inferiority margin (-7%) was set depending on the previous studies that showed the comparator group (children who received CSB) had a recovery rate from MAM of 67% [19] and the recovery rate from MAM without any treatment was 54% [18], hence the difference is 13%. The non-inferiority margin was specified considering that the LIBS group has at least a 7% (the average of the difference of two proportions) higher recovery rate than in a group with no supplementary food (placebo). This was based on the recommendations of selecting a non-inferiority margin [25] For this sample size calculation, the PharmaSchool sample size calculator for non-inferiority trials was used. The trained data collectors, accompanied by health extension workers, visited all households found in selected kebeles (Waribira Golo, Bibiso Olola, Waribira Suke, Shanto, Tomtome, and Lera) in Damot Pulassa district with children aged 6 to 59 months, to assess children for eligibility by measuring MUAC. The recruitment was done from August 27 to September 20, 2018. Children with MUAC <12.5 cm were recorded and brought to the actual screening site with their mother/caregiver, where MUAC was re-measured, and weight and height or length were taken. In addition to the values of MUAC (i.e. between ≥11.5 cm and <12.5 cm), we used the WHZ or weight for length z-score for recruitment (i.e. WHZ: between −3 and −2 Z-scores). When the child was identified as moderately malnourished according to the MUAC and not according to the values of weight-for-height/length z-scores, we recruited them according to their MUAC values. Children aged 6 to 23 months were measured for weight and length, and children aged 24 to 59 months were measured for weight and height. Edematous malnutrition was also assessed using the bilateral pitting edema criterion. The screening for MAM was continued until the sample size was met. A computer-generated randomization list that contained codes was prepared using random allocation software. The allocation ratio was 1:1. Mothers selected an envelope containing coded numbers that matched with one of the two supplementary foods. A research assistant who was not otherwise involved with the study implemented the randomization process. After randomization, the investigator further classified children into sub-groups, with equal numbers of subjects based on their neighborhood. This was done to facilitate assigning one food distributor per sub-group, where food distributors simply had to access the households with a selected child. The food distribution process was organized by the research assistant who was aware of which number corresponded to which food. This person did not take part in the food distribution. This study was double-blinded, that are, caregivers, data collectors, and food distributors were blinded for the intervention. Both supplements were packed with the same plastic packs, similar in color and texture, and they were prepared and distributed in the same way. The sugar which was provided for the intervention group was blended with the flour before packing. Sunflower oil was distributed with the supplement for both intervention and control groups. Food distributors were assigned to support caregivers in cooking the porridge with 8 ml of oil for the intervention group and with 16 ml of oil for the CSB+ group. We distributed the oil with colored plastic cups. Before introducing the interviews, the study questionnaires were subjected to pilot-testing and were refined for clarity and correctness. The interviews were conducted with selected caregivers of children with MAM aged 6 to 59 months, for collecting baseline information such as socio-demographic and economic status, child’s age, dietary habits, breastfeeding practices, and history of child and maternal illness. The child’s weight was measured with a Seca weight scale to the nearest 0.1 kg. The data collectors ensured that the scale was positioned on a flat, firm surface, and weighing was done with light clothing. The length was measured to the nearest 0.1 cm, using a locally prepared wooden measuring board for children aged 6 to 23 months. For children aged 24 to 59 months, height was measured to the nearest 0.1 cm using a Seca height scale. Before height measurement, the data collectors ensured that the height board was on level ground and the child was barefoot; the collector kneeled to get to the level of the child and encouraged the caregiver to help. For length, data collectors measured the child lying down, being sure that the length board was placed on a flat and stable surface. MUAC was assessed by non-stretchable standard United Nations Children’s Fund (UNICEF) plastic tape measures. The measurement was taken halfway between the acromion and olecranon processes, with the measuring tape fitting comfortably, but without making a depression on the left upper arm. This was done twice for every child, using two different data collectors, and the average of the two measures was recorded to the nearest 0.1 cm. Participants in both groups were visited every week for consecutive 12weeks to collect anthropometric measurements using MUAC, and information on supplement use and morbidity. Every month, anthropometric data were collected with identical equipment as used at baseline. Bilateral pitting edema was assessed by pressing for three seconds on the dorsum of the foot. Based on the follow-up measurement of anthropometry, children who developed SAM were sent to the SAM clinic. Subjects in the intervention group received a daily ration of 125.2 g of LIBS with 8 ml of refined deodorized and cholesterol-free sunflower oil for 12 weeks. The composition of LIBS was: 30 g of pumpkin seed, 25 g of peanut grain, 20 g of amaranth grain, 15 g of flaxseed, 10 g of emmer wheat, and 25.2 g of cane sugar with 8 ml of refined deodorized and cholesterol-free sunflower oil. This supplement (one serving) yielded 699 kcal, 22.6 g protein, 56.9 g carbohydrate, and 40.89 g fat. The cane sugar was added to LIBS in which the taste of LIBS made better and the amount of calories that should come from carbohydrates was improved but still lower than the level of carbohydrate found in the CSB+ (conventional food provided for children with MAM in the control group). Likewise, children in the control group received 150 g of CSB+/day with 16 ml of refined deodorized and cholesterol-free sunflower oil; this yielded 751 kcal, 21.25 g protein, 95 g carbohydrate, and 31.76 g fat daily for 12 weeks. The participants from both groups were served the supplements in the morning as breakfast (Table 1). Children aged 6–23 months and 24 to 59 months received a similar amount of food [24]. The food distributors who were trained for the preparation of porridge visited each household daily to provide the supplement and help the caregivers in the preparation of the porridge and feeding, to advise, assess, and resolve problems with feeding. The 150g of CSB+ was diluted with 600 ml water and cooked for 10 to 15 minutes whereas the 125.2g of LIBS was diluted with 400 ml water and cooked for 10 to 15 minutes. The food distributors checked and measured the amount of supplements consumed by the children using the local measuring cup with numbers. In case of not consuming all the prepared porridge, re-feeding was done. Abbreviations: LIBS: local-ingredients-based supplement; CSB+: corn-soy blend plus; kcal: kilocalorie; g: gram; and mg: milligram. Note: Nutrient values for the LIBS ration were calculated by using the United States Department of Agriculture (USDA) food composition database and NutriSurvey software. Nutrient values for the CSB+ were adapted from Amegovu KA, Ogwok P, Ochola S, Yiga P, Musalima HJ, Mutenyo E. Formulation of sorghum-peanut blend using linear programming for treatment of moderate acute malnutrition in Uganda. J Food Chem and Nutr. 2013; 1(2):67–77. The field supervisors controlled all data collection sheets for completeness. Data entry was done using Epi Data v. 3.1. (Odense, Denmark). Double data entry was done simultaneously to ensure data quality. According to the recommendations for analyzing and reporting equivalence and non-inferiority trials, both intention-to-treat (ITT) and per-protocol (PP) analyses were done, and the 95% CI level was used to interpret any differences [24]. The ITT analyses involved all patients who were randomly allocated, whereas the PP analysis excluded children who refused, transferred out of the management program, or were lost to follow-ups, but included children who died, or were discharged as cured or non-cured. Statistical analysis was done using SPSS v. 20 (IBM Chicago, IL, U.S.A.) and STATA 15 (StataCorp LLC) software. Comparisons of baseline and outcomes characteristics between CSB+ and LIBS were summarized as percentages and means (±SDs). Comparison of outcomes between the LIBS and CSB+ groups were made by using a chi-square test for categorical variables and generalized estimating equations (GEE) for continuous variables. Differences in the estimated proportion between the groups along with 95% CI were estimated to infer non-inferiority. Our trial is testing against a one-sided hypothesis, but the decision of non-inferiority is based on a two-sided 95% confidence interval, even if we deal with the lower limit of the interval. The recovery time between groups was predicted in Kaplan-Meier (log-rank) curves of survival analysis. Mean difference in weight gain, mean difference of MUAC gain, and mean difference in height/length gain were computed to describe the magnitude of the difference between the two groups. Rates of weight gain through the whole period of follow-up were estimated in g/kg body weight/day and compared between the study groups. These were calculated by dividing the weight gain (weight at exit minus weight at admission), expressed in grams, by the weight at admission (in kilograms), and the length of stay (in days). Anthropometric indexes were calculated using ENA for SMART 2011 software.