The importance of public health, poverty reduction programs and women’s empowerment in the reduction of child stunting in rural areas of Moramanga and Morondava, Madagascar

Background: Malnutrition accounts for 45% of mortality in children under five years old, despite a global mobilization against chronic malnutrition. In Madagascar, the most recent data show that the prevalence of stunting in children under five years old is still around 47.4%. This study aimed to identify the determinants of stunting in children in rural areas of Moramanga and Morondava districts to target the main areas for intervention. Methods: A case-control study was conducted in children aged from 6 to 59.9 months, in 2014–2015. We measured the height and weight of mothers and children and collected data on child, mother and household characteristics. One stool specimen was collected from each child for intestinal parasite identification. We used a multivariate logistic regression model to identify the determinants of stunting using backwards stepwise methods. Results: We included 894 and 932 children in Moramanga and in Morondava respectively. Stunting was highly prevalent in both areas, being 52.8% and 40.0% for Moramanga and Morondava, respectively. Stunting was most associated with a specific age period (12mo to 35mo) in the two study sites. Infection with Trichuris trichiura (aOR: 2.4, 95% CI: 1.1–5.3) and those belonging to poorer households (aOR: 2.3, 95% CI: 1.6–3.4) were the major risk factors in Moramanga. In Morondava, children whose mother had activities outside the household (aOR: 1.7, 95% CI: 1.2–2.5) and those perceived to be small at birth (aOR: 1.6, 95% CI: 1.1–2.1) were more likely to be stunted, whereas adequate birth spacing (24months) appeared protective (aOR: 0.4, 95% CI: 0.3–0.7). Conclusion: Interventions that could improve children’s growth in these two areas include poverty reduction, women’s empowerment, public health programmes focusing on WASH and increasing acceptability, and increased coverage and quality of child/maternal health services. We conducted this study in two rural areas of Madagascar: the Health and Demographic Surveillance Site (HDSS) of Moramanga and the municipality of Bemanonga, district of Morondava. The two selected areas belong to 2 regions with different nutritional profiles: Alaotra-Mangoro region is an area with high stunting prevalence whereas Menabe region is located in an area with average stunting prevalence. We assumed these differences could be related to regional differences in stunting determinants. The HDSS of Moramanga is located in the central eastern part of Madagascar and is composed of three municipalities, 30 villages and 60,000 inhabitants. These three municipalities have a total of 9 operational health facilities (1 hospital, 8 health centers). The municipality of Bemanonga is located in the southwestern part of Madagascar. This study was implemented in 13 of the 34 villages in Bemanonga. The 13 villages had an estimated total population of 13,700 and had two health centers. The participants of this community-based case-control study were children aged 6 to 59.9 months and their mothers or caregivers. Sample size was calculated to detect an odds ratio of 1.5 for the association between stunting and intestinal parasite carriage, with a 95% confidence interval, a statistical power of 80% and an expected prevalence of parasitic infection of 10% in non-malnourished children. The total sample size was 1620 children in both study areas (405 cases and 405 controls at each study site). Firstly, a screening phase was conducted by measuring weight, height/length of all children under 5 years. A simple random sampling of non-malnourished and stunted children was later performed to form the case and control groups. Cases were children aged 6–59.9 months with stunting, defined as a height-for-age z-score<-2SD of the median height of the WHO reference population. Controls were children aged 6–59.9 months without stunting nor wasting. If mothers had more than one eligible child, only the first randomly selected child was considered. Children under 5 years with disabilities preventing anthropometric measurements were not included in the screening phase. Those with both stunting and wasting and those not accompanied by their own mothers or caregivers during the survey were not included in the case-control study. Locally recruited fieldworkers were trained to perform the anthropometric measurements to identify the source populations of cases and controls and to conduct interviews with the mothers. Data were collected from January to July 2014 in Moramanga and from August to November 2014 in Morondava. All eligible children were measured during the screening phase in both sites. Weight was measured to the nearest 100 g using a hanging scale (Salter®) for children weighing ≤ 25 kg and with a weighing scale (Gima®) for those weighing > 25 kg, those for whom using a hanging scale was difficult (agitated, did not want to be separated from their mothers, etc.) and for mothers. Length/height was measured to the nearest 0.1 cm in a recumbent position for children < 24 months and in a standing position for those who were older using collapsible length/height boards. All scales were frequently calibrated, before and during the survey. We interviewed mothers/caretakers in Malagasy using a structured questionnaire. The following information was collected: maternal characteristics (education, occupation, dietary patterns, pregnancy history, antenatal information, marital status), household characteristics (housing characteristics, water source, sanitation, goods, possession of land and livestock, feeding practices) and child characteristics (date of birth, sex, childhood illness symptoms two weeks before the survey, immunization status, child diet, place of delivery, vitamin A capsule intake and use of de-worming medication within 6 months before the survey). The information collected on child diet were breastfeeding status, age at introduction of supplementary food and all food/beverages consumed by the child the day before the survey. For each randomly selected child, we collected approximately 10 g of fresh stool sample for parasite identification. Intestinal parasites, vegetative forms, cysts and eggs were detected by direct examination of smear on a clean glass slide after dilution of approximately 2 mg of stool in isotonic salt solution. Each slide was subsequently stained with Lugol’s Iodine solution and examined by conventional light microscopy by 2 certified laboratory technicians. Stool samples were tested for opportunistic parasites at the Institut Pasteur de Madagascar. Opportunistic parasites are parasites that take advantage of an opportunity not normally available such as host with a weakened immune system or an altered microbiota to infect the host. For this purpose, stool samples were conserved at room temperature in a 2.5% potassium dichromate solution until use. Microscopic diagnosis of oocysts of Cryptosporidium spp., Isospora belli and Cyclospora cayetanensis was performed using the modified Ziehl-Neelsen acid fast stain technique [17]. Microsporidia were detected by Trichrome stain [18,19]. For opportunistic parasites, diagnosis was confirmed by molecular analysis of positive samples detected by microscopy. DNA extraction from stool samples was performed using a modified QIAamp® DNA Stool Mini Kit protocol [20]. Briefly, before extraction, 200 μg of stool samples, conserved in potassium dichromate, were washed twice with PBS 1x (Phosphate buffer saline, Invitrogen). A sonication step, 30 minutes at 4°C, was subsequently performed, and 1.4 ml of ASL lysis buffer was added to the pretreated stool. This step was followed by incubating the preparation for 5 min at 95°C prior to DNA extraction, as recommended by the manufacturers. Cryptosporidium parvum, Entamoeba histolytica and Giardia lamblia were simultaneously detected by real time qPCR multiplex assay TaqMan StepOne-plus using primers and TaqMan probes following the protocol already described by Verweij et al [21,22]. Microsporidia species, i.e., Enterocytozoon bieneusi and Encephalitozoon spp., were also detected by real time qPCR assay as described [21]. Data were collected on electronic forms using a Microsoft Access database. We examined anthropometric data for completeness and plausibility using Emergency Nutrition Assessment software (ENA for Smart, Centers for Disease Control and Prevention). Weight and length/height of the children were converted into height-for-age z-scores and weight-for-height z-scores according to the 2006 WHO child growth standards [23] and we used R software version 3.0 for all subsequent analyses. After cleaning the data to address inconsistencies, descriptive analyses were performed. A wealth index was created using variables related to ownership of selected household assets, housing materials (floor, wall, and roof materials), access to utilities (electricity, safe water, latrine, bathroom, cooking location), family size and combustible used for cooking. We used principal component analysis for continuous variables and multiple correspondence analyses for categorical variables. We subsequently grouped the wealth index score into quartiles, with Quartile 1 representing the poorest segment of the population and Quartile 4, the wealthiest. Food consumed by children the day before the survey was used to calculate the dietary diversity score which is defined as the number of food groups consumed by the child the previous day. We considered seven food groups: (1) cereals, roots and tubers; (2) legumes and nuts; (3) milk and its derivatives; (4) meat products (meat, poultry, offal, and fish); (5) eggs; (6) fruits and vegetables; and (7) oils and fats. Low dietary diversity score was defined as consumption of ≤3 food groups for children < 36 mo and ≤4 food groups for children≥36 mo. The Body Mass Index (BMI) for mother was assessed by dividing the weight (kilogram) by the square of body height (meter); we used three categories for the analyses “thin”: BMI<18.5 kg/m2, “normal”: BMI between 18.5 and 25 kg/m2 and “overweight”: BMI>25 kg/m2. Household food consumption score was constructed according to recommendations from the World Food Program [24]. Six groups of foods were considered, including (1) cereals and tubers, (2) legumes and peas, (3) vegetables, (4) fruits, (5)meat-based foods, and (6) milk and its derivatives (data on the consumption of sugar, oil and condiments were not available). The calculation of the score was performed with the consumption frequency of these food groups the week before the survey and a standard weight according to the groups of the consumed foods. The obtained score was then classified into 3 categories: poor, limited and acceptable. We assessed candidate variables for the multivariate analysis using bivariate analysis; all explanatory variables with a p-value<0.2 were included in a backward stepwise logistic regression analysis. Odds ratios and 95% confidence intervals (95% CIs) were calculated to estimate the strength of the association between the occurrence of stunting and the potential explanatory variables. Interactions between explanatory variables were checked. We did not perform imputation of missing data, and in Morondava, we included observations with missing values as additional categories. Parents or children’s guardians were informed about the study and signed a letter of consent before inclusion. Children with severe acute malnutrition were referred to the nearest health center; those with any identified parasites were treated according to the national treatment guidelines. The protocol of the study was approved by the National Ethics Committee of the Ministry of public health of Madagascar (042-MSANP/CE, June 13th 2014).