Impact of lipid-based nutrient supplementation (LNS) on children’s diet adequacy in Western Uganda

Lipid-based nutrient supplements (LNS) can help treat undernutrition; however, the dietary adequacy of children supplemented with LNS, and household utilisation patterns are not well understood. We assessed diet adequacy and the quality of complementary foods by conducting a diet assessment of 128 Ugandan children, ages 6-59 months, who participated in a 10-week programme for children with moderate acute malnutrition (MAM, defined as weight-for-age z-score<-2). Caregivers were given a weekly ration of 650kcalday-1 (126gday-1) of a peanut/soy LNS. Two 24-h dietary recalls were administered per child. LNS was offered to 86% of targeted children at least once. Among non-breastfed children, over 90% met their estimated average requirement (EAR) cut-points for all examined nutrients. Over 90% of breastfed children met EAR cut-points for nutrient density for most nutrients, except for zinc where 11.7% met cut-points. A lower proportion of both breastfed and non-breastfed children met adjusted EARs for the specific nutritional needs of MAM. Fewer than 20% of breastfed children met EAR nutrient-density guidelines for MAM for zinc, vitamin C, vitamin A and folate. Underweight status, the presence of a father in the child's home, and higher programme attendance were all associated with greater odds of feeding LNS to targeted children. Children in this community-based supplemental feeding programme who received a locally produced LNS exhibited substantial micronutrient deficiencies given the special dietary needs of this population. These results can help inform programme strategies to improve LNS targeting, and highlight potential nutrient inadequacies for consumers of LNS in community-based settings. Bundibugyo is one of four districts in Uganda's Western Region. At the time of the study, Bundibugyo was the only district in the region with no paved roads or electricity. With the Democratic Republic of Congo to the west and the Rwenzori mountains to the east, the district is geographically isolated from the rest of Uganda. The Bakonjo and Babwisi are the two predominant people groups in the 290 000‐person district, which includes 52 500 (18%) children less than 5 years (UBOS and Macro International, Inc. 2012). The stunting prevalence (height‐for‐age z‐score < −2) of 43.9% is the second highest in the country, and well above the national prevalence of 33.4% (Jilcott et al. 2007; UBOS and Macro International, Inc. 2012). The total fertility rate of 6.4 is higher than the national mean of 6.2. Uganda as a whole has the highest total fertility rate in East and Southern Africa, and one of the highest birth rates worldwide (UBOS & Macro International, Inc. 2011). Children are regularly recruited for BBB programme enrolment at local schools and churches and at routine outpatient weighing at health centres. All caregivers who enrolled in the 10‐week programme between June 2008 and June 2009 were eligible for participation in the study (n = 138). Study personnel were present during each weekly programme session for the duration of the study. Beginning at the fourth week of each programme cycle, a random sample of the caregivers in attendance at the programme was recruited to provide 24‐h dietary recall information for their enrolled child. Caregivers who were absent from the programme were recruited at their homes using information provided in the programme register. Recalls were conducted through week 9 of the program, with the goal of complete recruitment into the study of all mothers who contributed to and participated in the programme. Verbal informed consent was obtained prior to the dietary assessment. Demographic surveys and dietary recalls were conducted in either Lubwisi or Lukonjo, the two primary local languages, depending on the preference of the caregiver. After agreeing to participate, caregivers were interviewed following the daily program, or at a later day within the same week. About 70% of recalls were administered on weekdays in order to provide a proportional assessment of weekday and weekend dietary patterns. However, child diets do not vary much according to the day of the week and are generally very monotonous. Follow‐up recalls were scheduled at the conclusion of the first recall, and were conducted on non‐consecutive days within a 10‐day period to approximate usual food consumption, and to account for within‐person variation. This method of scheduling recalls allowed us to obtain dietary information for children on most days of the week in order to provide variation in the elapsed time since LNS distribution. Three study personnel trained in diet assessment methods (S.I., B.C., K.M.) performed all dietary recall measurements. Information collected during dietary recall included the name and time of each meal, and the ingredients and preparation method for each food item. Morbidity was assessed simply through a single question on whether the child's diet was typical diet on the day of recall. We included dietary recalls from children who were ill, but this affected only 2% (five total) recalls. Next, the portion size offered and the amount consumed by the child of interest were estimated using standard local utensils (e.g. tablespoon, 800‐mL plate, 500‐mL cup). Cups and plates were marked with fraction lines that divided the utensil into fifths to assist caregivers in estimating portion sizes. De‐identified dietary recalls and demographic surveys were scanned and entered into an electronic database for analysis. Demographic data were collected during the first recall with each caregiver using a modified questionnaire from the Uganda Demographic and Health Survey (UBOS & Macro International, Inc. 2007). Demographic questions included caregiver parity; the number of children living at the compound; and the education, presence, occupation, and marital status of the index child's parents. Socio‐economic constructs included building materials of the home, participation in a small business, primary means of food acquisition and whether the household cultivated cash crops. Access to health services and sanitation was measured by caregiver estimates of the distance in meters to the nearest clean water source and health centre. Information on the frequency of visits to health facilities was not collected. Trained health centre staff assessed children's anthropometric measurements. Weights were measured to the nearest 0.1 kg with a hanging Salter scale. Length was measured to the nearest 0.1 cm using a length board for children 6–24 months or who were shorter than 65 cm. Height for children >24 months or who were taller than 65 cm was measured using fixed measuring tapes with a headboard. Scales and length boards were calibrated twice monthly during the course of the programme. The University of North Carolina at Chapel Hill Institutional Review Board (Study # 08–1100) and the Bundibugyo District Health Office approved all study protocol. The BBB programme is managed by World Harvest Mission in the Bundibugyo District of Uganda and operates in two health centres and enrols 50 underweight [weight‐for‐age z‐score (WAZ) < −2] children ages 6–59 months per 10‐week cycle. Deworming medicine is offered at 5‐week intervals during the programme cycle. Anthelminthic treatment has been found to improve appetite when administered quarterly (Stoltzfus et al. 2004). As this population was well covered for de‐worming, the proximity of anthelminthic treatment to the day of dietary recalls was unlikely to influence the level of dietary intake. At each weekly visit, caregivers receive child growth monitoring and promotion, a weekly supply of LNS, children's multivitamin with iron, and nutrition education. Education is delivered by community health workers and health centre staff to emphasise: (1) the impact of early nutrition on school performance later in life; (2) healthy antenatal nutrition; (3) the importance of breastfeeding; (4) healthy complementary feeding practices; (5) using an attentive, responsive child feeding style; (6) feeding children during and after illness; and (7) hygiene practices. Caregivers receive 650 kcal day−1 of a peanut‐ and soy‐based LNS. This high dosage of LNS, particularly for children <2 years, was standard across age strata to simplify distribution logistics. If consumed without wastage or leakage, the daily LNS dose of 650 kcal day−1 would provide 63.5% of the mean total daily energy needs for children under 2 years, and 96.6% for children 1 year of age (FNB & IOM 2005). The LNS formula is fortified with dried moringa leaf powder, but no additional micronutrients. Therefore, children were also given a weekly supply of a children's daily multivitamin with iron at each programme session. The multivitamin was a standard children's chewable vitamin with iron. Parents were instructed to feed children half a tablet per day. Multivitamins contained vitamins A, C, D, E, B6, B12, thiamin, riboflavin, niacin, folic acid, pantothenic acid, sodium and iron. Adherence to the multivitamin was not formally assessed; however, anecdotes from programme staff indicated that adherence might have been low. Further, as the goal of our study was to examine the impact of LNS‐based supplementation on dietary adequacy, the micronutrient contribution from multivitamins was not considered in our analysis. Caregivers were instructed to feed the LNS to the enrolled child only, either directly without cooking or as a thick porridge, at each feeding episode (Jilcott et al. 2010). Although caregivers were not instructed with a specific dosage, the amount of LNS used in demonstration preparations and feedings was 2–4 Tb. LNS was made locally using hand‐powered grinders to prepare two products: (1) roasted groundnut (peanut) paste fortified with dried Moringa oleifera leaf powder (440 g); and (2) roasted soy flour (440 g). Each product was distributed in a plastic bag and was mixed by caregivers into a single bag at the conclusion of each programme session. The resulting product resembles thick peanut butter. To examine the composition of the LNS, three separate samples were analysed at the Department of Food Science and Technology, Makerere University, Uganda in June 2008, for moisture content (oven method), proximate composition (crude fat, crude protein, dietary fiber and ash; AOAC, 1999 method), energy (bomb calorimeter method), vitamin C and A (AOAC 1999 method) and aflatoxin content (VICAM fluorometer method) (Jilcott et al. 2010). The product was found to be free from aflatoxin. The 128 g day−1 ration provides 100% of the estimated average requirement (EAR) for vitamin A, 45% for vitamin C, 120% for folate, 34% for calcium, and 150% for zinc, for children 12–36 months (FNB & IOM 1997, 1998a, 1998b, 2000a, 2000b, 2005). Per 100 g, the LNS product contained 532 kcal. Table 1 summarises the nutrient composition of the LNS product. Nutrient composition of the lipid‐based nutrient supplement, per 100 g World Health Organization (WHO) Indicators for the assessment of IYCF were used to measure whether children ages 6–24 months were fed a ‘minimum acceptable diet’. Breastfed children met this criterion if they were fed four or more food groups and the minimum number of times per day (≥2 for children 6–8 months, ≥3 for children ages 9–23 months). Non‐breastfed children met this criterion if they were fed: (1) four or more food groups (apart from milk); (2) milk or milk‐based products two or more times daily; and (3) four or more times per day (WHO 2008). Breastfeeding status was assessed through self‐report. As the focus of the study was to determine nutrient adequacy of children during supplementation with LNS, the rationale for not breastfeeding was not explored in the present study. A food group was counted if a child consumed at least 1 g from any of the following seven groups: (1) grains, roots, and tubers, including porridge or fortified baby food from grains, including matoke/plantains; (2) legumes and nuts; (3) dairy products, including milk other than breast milk, cheese, yogurt, or other milk products; (4) flesh foods such as meat, poultry, fish, shellfish, organ meats; (5) eggs; (6) vitamin A‐rich fruits or vegetables; and (7) other fruits and vegetables. To draw comparisons among age groups, a dietary diversity score (DDS) was calculated as the sum of foods consumed within these seven food groups, resulting in a 0 = low diversity; 7 = high diversity. The DDS was identical to that used in the WHO IYCF Indicator assessment (WHO 2008). Scores were based on seven food groups and thus range from 0 to 7. These scores are similar, but not identical to FAO DDSs, which range from 0 to 12 for households and 0 to 9 for women (FAO & WHO 2002; WHO 2008). Nutrient values were represented as the mean from the two 24‐h recalls, obtained on non‐consecutive days. Food items were assigned nutrient values from the Tanzania Food Consumption Table (FCT) (Lukmanji et al. 2008), Malawi FCT (Ferguson et al. 2004) and United States Department of Agriculture (USDA) Nutrient Database (USDA Agricultural Research Service 2009). Portion weights were estimated by multiplying the estimated portion size by the unit weight in grams (obtained from field measurements taken in February 2009, or from the USDA or Tanzania FCT, when available). Total daily nutrient intakes were calculated for selected nutrients that are of particular interest to child growth: energy, protein, vitamin C, vitamin A, vitamin B6, zinc, iron and folate. We conducted two dietary recalls on non‐consecutive days to approximate usual intake of children. We calculated correlation coefficients between the two daily nutrient values. The correlation for macronutrients was moderate for fat (0.50), total energy (0.33) and total protein (0.29). The correlation for micronutrients was high for iron (0.74), moderate for folate (0.41) and low for vitamin C (0.26). Vitamin A (0.24) and calcium (0.24) intakes were least correlated among observations. To estimate the nutrient contribution of LNS to the overall diet, the mean amount of selected nutrients consumed from LNS was divided by the mean total amount consumed of that nutrient over 2 days of recall. The correlation coefficient between total energy intake from LNS on both days of recall was 0.56. While this correlation indicates variation between the amount of LNS consumed over 2 days, the goal of our study was to estimate the average contribution that LNS made to children's overall diets, recognising that intake will vary on a daily basis for a variety of factors. Calculating the mean LNS consumption from non‐consecutive days in a community‐based setting allows us to minimise extreme consumption values that may result from an early depletion of food rations. Dietary adequacy was assessed separately for non‐breastfed and breastfed children, because the breast milk intake for breastfed children in our study population was unknown. For non‐breastfed children, the probability of adequate intake (AI) for each selected nutrient was calculated using the cut‐point method, which compared each child's nutrient consumption with his or her age‐specific EAR or AI, in the case of calcium (FNB & IOM 2000b; Murphy et al. 2006). Dietary reference intake and AI values were obtained from the Food and Nutrition Board and Institute of Medicine guidelines (FNB & IOM 1997, 1998a, 1998b, 2000a, 2000b, 2005). Absorbed calcium was estimated by multiplying the total grams of calcium consumed by an absorption factor of 0.32 for all food groups. Iron absorption was calculated using absorption factors of 0.06 and 0.11 from plant and animal sources, respectively (Working Group on Infant and Young Child Feeding Indicators 2006). For breastfed children whose breast milk intake amount was unknown, we assessed dietary adequacy by examining the nutrient density of complementary foods. Nutrient densities per 100 kcal were calculated as follows: The nutrient density of complementary foods was compared against the age‐specific nutrient densities guidelines that are specified by the WHO standards, including the specific requirements of children with moderate acute malnutrition (MAM) (FAO & WHO 2002; Dewey & Brown 2003; Golden 2009). All data analysis was conducted on Stata 12.0 (Stata Corp, College Station, TX, USA). Potential confounders in the regression models were selected based on findings from an earlier qualitative study (Ickes et al. 2012) and theoretical knowledge of factors that influence children's diets in developing countries (Black et al. 2008). Regression coefficients were considered statistically significant if the 95% confidence interval did not overlap with the null. Comparisons of mean nutrient intakes across age strata were made using a one‐way analysis of variance test. Differences were considered statistically significant if P < 0.05. A multivariate regression model was created to estimate the association of selected socio‐demographic factors with LNS consumption. Because of the substantial proportion of children who did not consume any LNS over the days of observation (14.1%), two separate regression models were used. First, we applied logistic regression to examine the factors associated with no LNS consumption by comparing non‐consumers with consumers among children who consumed a minimum of 1 g of LNS on either day. Second, we applied linear regression to examine the factors associated with the level of LNS consumption (coded continuously), conditional on consuming at least 1 g of LNS in either study day. As the factors associated with no daily consumption may differ markedly from the factors associated with varying levels of LNS consumption, we tested factors separately in each scenario. Both regression models controlled for age, gender, nutritional status (by WAZ score), presence of the father in the home, maternal education and primary means of food acquisition.