Prenatal anemia control and anemia in children aged 6–23 months in sub-Saharan Africa

It is unclear whether routine prenatal anemia control interventions can reduce anemia risk in young children. This study examines the associations between prenatal iron supplementation and/or deworming and anemia in children aged 6–23 months in sub-Saharan Africa (SSA). We analyzed data from Demographic and Health Surveys conducted between 2003 and 2014 in 25 SSA countries. The surveys collected data on prenatal iron supplementation and deworming and determined children’s hemoglobin levels through blood testing. We assessed the associations between prenatal iron supplementation and/or deworming and anemia using multinomial logistic regression. The study included 31,815 mother–child pairs: 25.0%, 41.4%, and 4.8% of children had mild, moderate, and severe anemia, respectively. Compared with children whose mothers did not take iron and deworming drugs prenatally, the risk of moderate/severe anemia was reduced among children whose mothers took only iron supplements for ≥6 months (odds ratio [OR]: 0.58; 95% confidence interval [CI]: 0.45–0.76); only deworming drugs (OR: 0.73; 95% CI: 0.56–0.93); deworming drugs plus iron for <6 months (OR: 0.79; 95% CI: 0.67–0.93); and deworming drugs plus iron for ≥6 months (OR: 0.77; 95% CI: 0.59–0.99). Prenatal use of only iron for <6 months was not associated with moderate/severe anemia. Prenatal iron and/or deworming drugs had no effect on mild anemia. Prenatal anemia control interventions are associated with reduced risk of moderate/severe anemia but not with mild anemia in young children in SSA. Iron supplements should be taken for ≥6 months or with deworming drugs prenatally to reduce moderate/severe anemia risk in children. This study utilized data from most recent Demographic and Health Surveys (DHS) from SSA countries. The DHS program represents the largest effort to obtain health and demographic data from nationally representative household surveys in low‐income and middle‐income countries. DHS collect data utilizing stratified multistage cluster sampling method. In the first stage, clusters (primary sampling units) are selected from a list of enumeration areas, using stratified random sampling, with regions constituting strata. The second stage involves systematically sampling households in selected clusters. All women aged 15–49 years who are present in the household on the night before the surveys are eligible for interview. Blood samples are collected from all the households or from a random subset of selected households based on considerations such as required sample size, financial costs, and logistics. DHS use household and women's questionnaires to collect data relevant to women and children. The household questionnaire is used to collect data on characteristics of the household's dwelling unit and to record height, weight, and hemoglobin (Hb) measurements of women aged 15–49 years and children. The women's questionnaire is used to collect data from all women aged 15–49 years on various issues including socio‐demographics, birth history, maternal and child health, and nutrition. Details about the DHS methodology and tools are available elsewhere (Rutstein and Rojas, 2006, The DHS Program, 1987–2016). We included in this study all SSA countries with a severe public health problem of anemia in children—defined as anemia prevalence of >40% in preschool‐aged children (WHO, 2008)—if their latest DHS were conducted after the year 2000, and the dataset contains information on Hb measurement, iron supplementation, and use of deworming drugs during pregnancy. Appendix 1 shows the list of included countries and the percentage of households selected for Hb measurement in each country. Appendix 2 shows the list of excluded countries, with reasons for exclusion. The study population for the present analysis is children aged 6–23 months and their mothers. The outcome variable was anemia, adjusted for altitude and categorized as none (Hb ≥11.0 g/dl), mild (Hb 10–10.9 g/dl), moderate (Hb 7.0–9.9 g/dl), and severe (Hb <7.0 g/dl; Sharman, 2000; WHO, 2011). In regression analyses, the last two categories were combined because the proportion of children with severe anemia was small (4.8%). The DHS program tests for anemia in a standardized way across surveys (Sharman, 2000): Blood specimens are collected from children aged 6–59 months using a microcuvette from a drop of blood taken from a finger or heel prick (for undernourished/skinny children), and Hb analysis is carried out on‐site using a battery‐operated portable HemoCue® analyzer (HemoCue, Ängelholm, Sweden), a highly valid method when compared to standard laboratory methods (Nkrumah et al., 2011). General procedures for collecting blood samples are available elsewhere (ICF International, 2012). Women with a live birth 5 years preceding the survey were asked whether, during the pregnancy of the most recent birth, they took iron tablets or syrup (with or without folic acid), the number of days they consumed the supplements, and whether they took drugs for intestinal worms. We used this information to define the exposure variable, referred to as prenatal anemia control intervention, as follows: none (no iron supplements and no deworming drugs); only iron supplements for <6 months; only iron supplements for ≥6 months; only deworming; deworming plus use of iron supplements for <6 months; and deworming plus use of iron supplements for ≥6 months. Due to missing data on the number of days iron was consumed, we created another category of iron for unknown period ± deworming. Women who did not take iron supplements and deworming drugs constituted the reference group in all comparisons. These included child, maternal, household, and contextual factors. Child's factors included age, sex, wasted, consumed iron/vitamin A rich foods in past 24 hr, and had pneumonia, diarrhea, and fever in the preceding 2 weeks. Maternal factors included mother's age at childbirth, years of education, parity, delivered by cesarean section, and body mass index (weight in kilograms per height in square meter). Household factors included wealth index quintile and type of cooking fuel. Wealth index quintiles were derived through factor analysis of ownership of household assets, access to public utilities, and type of housing material (Rutstein and Rojas, 2006). The first of the obtained factor scores was used to represent the wealth index (Rutstein and Rojas, 2006). Contextual factors included survey year, residence (urban/rural), and country. Details of the definitions of potential confounders are available in the footnotes of Table 1. Characteristics of 31,815 children aged 6–23 months by prenatal anemia control intervention in 25 sub‐Saharan African countries BMI <12.0 or >60.0. Pneumonia is defined as having symptoms of acute respiratory infection characterized by cough accompanied by short, rapid breathing and/or by difficult breathing, which were chest related. The symptoms were reported by the child’s caretaker. We created a dataset of eligible children by merging DHS datasets and excluding children aged <6 months or ≥24 months. Because the analysis involved combining data from different surveys, we first de‐normalized weights in each dataset by dividing the individual standard weight by the survey sampling fraction. Throughout the analyses, we used survey analysis functions to account for the primary sampling unit (cluster), strata, and sample weights. Accounting for clustering in the sample avoids underestimation of variability in the estimates by adjusting standard errors and confidence intervals (CIs), and weighting the data ensures representativeness. We tabulated the exposure variable against the outcome and the potential confounding variables. All the presented frequencies are unweighted, but the percentages are weighted to account for the study design. We used multiple imputation using chained equations to impute missing covariates' data (StataCorp, 2013). We created 10 imputed datasets using an imputation model that included all the covariates listed previously plus mother's anemia status, births in the past 5 years, place of delivery, stunting (child), strata, and country. We calculated unadjusted and adjusted odds ratios (ORs) and 95% CIs for the association between the exposure and the outcome using multinomial logistic regression. All the effect estimates and standard errors from imputed datasets were automatically combined using Rubin's rules (Rubin, 2008). We considered variables that are associated with anemia in children based on previous studies (Kyu et al., 2010; Mishra and Retherford, 2007; Li et al., 2015) or those associated with both the exposure and the outcome, at p < .2 threshold, in our unadjusted results, to be potential confounders. Because the variable “country” met the later criterion, the multivariate model was a fixed‐effects model that accounted for between‐country differences. We considered child's birth weight, stunting, and mother's anemia status at the time of the survey to be on the causal pathway between the exposure and the outcome and did not adjust for these variables. Mother's and child's age and wealth index quintiles were entered in the model as continuous variables. Because anemia risk in children increases with child's age (Crawley, 2004), we assessed for interaction between prenatal anemia control and child's age (6–13 months and 14–23 months). We present two‐sided p values of ORs from Wald's tests; p values of ≤.05 are considered to be statistically significant. All analyses were performed in Stata/MP version 13.1 (StataCorp, College Station, USA). During the surveys, informed consent was obtained for oral interviews and for biomarker measurements. The results of Hb measurement of children were given to their parents, both verbally and in writing, and parents of children with Hb <7 g/dl were instructed to take their children to health facilities for follow‐up care. This study was exempted from ethical review by the ethical review board of Kyoto University Graduate School of Medicine because it is based on de‐identified open‐source datasets.