The nutritional status data used forthis study were obtained from food security and nutrition analysis unit(FSNAU). Food Securityand Nutrition Unit (FSNAU) is a unit in World Food Programme of United Nations(UNFAO), which was set up in 1994 to provide evidence-based analysis of Somalifood, nutrition and livelihood security to enable both short-term emergencyresponses and long- term strategic planning. Therefore, in partnership withUNICEF, FSNAU has been conducting bi-annual seasonal nutrition assessmentsurveys since 2001.
Our study focuses on survey data ranging from 2007 -2010Figure 1. Within this period, FSNAU, in partnership with United NationsChildren’s Emergency Fund (UNICEF), conducted bi-annual seasonal nutritionassessment surveys using standard methods, indicators and tools for datacollection26,28. In each survey, a stratified multi-stage cluster sampling design wasadopted where the sampling frame of a selected district was based on the fourlivelihood definitions (pastoral, agro-pastoral, riverine and fishing) withinwhich 30 communities and 30 households within each village were selected usingsystematic random sampling methodand the urban population wereclearly defined and considered separately. Vulnerable groups that could not beclassified in any of the livelihood such as IDPs were surveyed separately. Respectivesamples sizes (number of households and number of children) were calculatedusing the Epiinfo/Ena 2008 software after considering the population size,estimated prevalence and desired precision. A list of all villages andpopulation within each of the assessed livelihoods served as a sampling frameand was used to construct cumulative population for the assessment area. Villageswere then randomly selected from these villages with the chance of any villagebeing selected being proportional to the size of its population.
This is calledsampling with “probability proportional to population size (PPS)”. Selectionof households within the village was done using systematic random sampling,preferably from a list of eligible names or a map of households. Where thesewere not available, the number of households in the village was estimated fromthe population figures (the total population divided by the mean householdsize). Starting from a random household, every nth household was selected andall eligible children (aged 6-59) in that household measured (Figure SI 1).Retrospective mortality data was collected from all the households in each villagefrom each livelihood, including even those that did not have children aged 6-59months.
At the individual child level, age, gender, weight, height, mid-upperarm circumference (MUAC), vitamin A supplementation in the last six months,diarrhoea, acute respiratory infections (ARI) and febrile illness in the twoweeks before the survey, and Polio and Measles vaccination history werecollected. At the household level, information recorded included the householdsize and age structure, gender of the household head, and access to differenttypes of foods in the last 24 hours. Data on falciparum malaria infection in children aged 5-59 months werecollected in sub-sets of villages at the request of UNICEF29–31. The data used in this study weretherefore a subset of the whole survey dataset with information on both thechildhood malnutrition and malaria.
We considered two outcome measurementsto describe the anthropometric indicators of malnutrition, lowweight-for-height (wasting) and low-MUAC, which detect different sets of childrenas malnourished. Wasting is traditionally the main indicator in communitysurveys. Although MUAC is a better predictor of mortality32, few studies have examined associationsbetween MUAC and specific pathogens.
A child was defined as wasted whens/he was below -2 Z scores for weight-for-height, according to World HealthOrganization (WHO) 2006 standards33. A child with MUAC below 125mm wasclassified as having low-MUAC. Malaria parasitaemia was determined usingParacheck Pf™ (Orchid Biomedical Systems, Goa, India) in a subset of the samplein every FSNAU surveys during this period33. A child was regarded as malaria infectedwhen s/he had a positive Paracheck Pf™ test result, regardless of any clinicalsymptoms. A detailed search were undertaken toestablish a set of spatial coordinates for each village in Somalia using the villagenames in the data. The location of village was verified by using Google Earth(Google, Seattle, USA) and other online databases to visually inspect whetherthe coordinates matched evidence of human settlement. Those settlements forwhich no reliable source of the coordinates was obtained were excluded from theanalysis. Environmental dataA set of four plausible environmentalcovariates, together with wasting, low-MUAC and malaria in children wereincluded in the analysis18,34.
These were rainfall, enhancedvegetation index (EVI), mean temperature, and urbanization. Rainfall and meantemperature were derived from the monthly average grid surfaces obtained fromWorldClim database35. The EVI values were derived from theMODerate-resolution Imaging Spectroradiometer (MODIS) sensor imagery for period2007-2010 while the urbanization information was obtained from Global RuralUrban Mapping Project (GRUMP)36,37. All the environmental covariateswere extracted from 1 x 1 km spatial resolution grids. Rainfall, temperatureand EVI were summarized to compute seasonal averages corresponding to the timeof survey.
Statistical methodsTheoverall aim of this study was to model the ecological comorbidity of wastingand low-MUAC with malaria parasitaemia among children aged 6-59 month inSomalia from 2007 – 2010. To achieve this, we implemented the Bayesiangeostatistical shared component model through stochastic partial differentialequation (SPDE) approach in integrated nested Laplace approximations (INLA)using R-INLA library in R project version 3.2.3. 22–24,38,39.
Therefore, we modelled twounderlying spatial risks common to: (1) wasting and malaria and (2) low-MUAC andmalaria at child level. The relative risk of each condition depends on a latentspatial component shared by each pair and a condition-specific component aftercontrolling for environmental covariates40,41. The household survey andenvironmental predictors of malnutrition and malaria were controlled atindividual, household and village level. Finally, to determineif the risks were correlated, we performed a significant test by looking at the2.5% and 97.5% quantiles of each element of the random effect using thequintile correction (QC) method as implemented by Bolin and Lindgren 201242.
Further, the empirical correlation betweenthe conditions were explored using correlation plots.