Social, dietary and clinical correlates of oedema in children with severe acute malnutrition: A cross-sectional study

Background: Severe acute malnutrition is a serious public health problem, and a challenge to clinicians. Why some children with malnutrition develop oedema (kwashiorkor) is not well understood. The objective of this study was to investigate socio-demographic, dietary and clinical correlates of oedema, in children hospitalised with severe acute malnutrition. Methods: We recruited children with severe acute malnutrition admitted to Mulago Hospital, Uganda. Data was collected using questionnaires, clinical examination and measurement of blood haemoglobin, plasma c-reactive protein and α1-acid glycoprotein. Correlates of oedema were identified using multiple logistic regression analysis. Results: Of 120 children included, 77 (64%) presented with oedematous malnutrition. Oedematous children were slightly older (17.7 vs. 15.0 months, p = 0.006). After adjustment for age and sex, oedematous children were less likely to be breastfed (odds ratio (OR): 0.19, 95%-confidence interval (CI): 0.06; 0.59), to be HIV-infected (OR: 0.10, CI: 0.03; 0.41), to report cough (OR: 0.33, CI: 0.13; 0.82) and fever (OR: 0.22, CI: 0.09; 0.51), and to have axillary temperature > 37.5°C (OR: 0.28 CI: 0.11; 0.68). Household dietary diversity score was lower in children with oedema (OR: 0.58, CI: 0.40; 85). No association was found with plasma levels of acute phase proteins, household food insecurity or birth weight. Conclusion: Children with oedematous malnutrition were less likely to be breastfed, less likely to have HIV infection and had fewer symptoms of other infections. Dietary diversity was lower in households of children who presented with oedema. Future research may confirm whether a causal relationship exists between these factors and nutritional oedema. This cross-sectional study is based on baseline data from a cohort study of children admitted for in-hospital treatment of SAM, between October 2012 and February 2013. Mwanamugimu Nutrition Unit, Mulago Hospital, Uganda, is the main national rehabilitation centre for children with complicated SAM. Patients were treated following the Ugandan National Protocol for Integrated Management of Acute Malnutrition [14] using therapeutic diets, F75 and F100, and empiric parenteral antibiotics, followed by outpatient treatment with ready-to-use therapeutic food. At the time of the study, the outpatient clinic was only working one day per week, so referred children were not routinely assessed with appetite tests to determine whether they should receive in-patient or out-patient treatment. However, all admissions came through the hospitals acute care unit, meaning that all children had been evaluated sick enough to require hospital admission. All biological mothers were offered counselling and testing for HIV, following World Health Organization (WHO) guidelines [15]. If the mother was HIV-infected or absent, the child was tested. Children were eligible if they were admitted on weekdays for treatment of SAM, defined as either weight-for-height z-score < −3, using the WHO Growth Standard, or mid-upper arm circumference (MUAC) < 11.5 cm, or bilateral pedal pitting oedema. Children had to be 6–59 months old, live near the hospital, and their parent or guardian had to give informed consent. Children were excluded if they had significant disability (like cerebral palsy); shock or severe respiratory distress requiring resuscitation at admission; haemoglobin < 4 g/dl, or a body weight < 4.5 kg. Caretakers were asked about the household where the child had lived during the two months preceding admission, about the child’s breast-feeding history and symptoms present. Caretakers were asked to state which one of their child’s symptoms they perceived as most severe. If the mother was present, she was asked to recall the child’s birth weight. Dietary data was collected by asking about whether seven types of high-quality foods were served in the household during the last two weeks (Additional file 1: Figure S1). The foods were locally available, but likely in limited amounts in resource-poor households. From this, a simple dietary diversity score (DDS) was calculated, as the sum of different food types served. Food insecurity was evaluated using the validated Household Food Insecurity Access Scale (HFIAS) [16]. The scale consists of nine questions regarding perceived food insecurity, each with four frequency options, from “never” to “more than ten times in the past four weeks”. A HFIAS score from 0–27 was calculated, and households classified as having no, mild, moderate, or severe food insecurity. On admission, oedema was diagnosed according to guidelines [17], and axillary temperature was measured. Anthropometric measurements were done by one nurse, trained in anthropometry, assisted by the child’s caretaker. Measurements were done in triplicate, and the average of three measurements was used. Length was measured using an infant length board and MUAC using measuring tape, both to the nearest 1 mm. Body weight was measured daily to the nearest 100 g using a digital scale. Anthropometric z-scores were computed in Stata using the command “zscore06” based on the 2006 WHO Growth Standards [18]. The lowest weight recorded during admission was used to compute z-scores for all children, to use the weight free from oedema. Maternal weight was measured using a digital scale with a precision of 100 g and height measured with a precision of 1 mm using a wall-mounted stadiometer. We did not by default remove extreme anthropometric values, but it was checked that they were likely to be real, by being present on different days, and by being in accordance with other measurements (e.g. those with extreme z-scores also had very low MUAC). To assess appetite, it was noted whether the child consumed all of the first served F75, or not. On admission, haemoglobin was measured in venous blood collected in heparinized Vacutainer tubes, using HemoCue (Hb 201+, Ängelholm, Sweden). Plasma was obtained from a Vacutainer with citrate (Cell-Preparation Tube, Becton Dickinson, USA), stored at -80C°, until shipped to Denmark on dry ice, where plasma (P-) level of C-reactive protein (CRP) and α1-acid glycoprotein (AGP) was measured at University of Copenhagen, Department of Nutrition, Exercise and Sports, with ABX Pentra 400 (Horiba, France). The study was approved by Makerere University School of Medicine’s Research Ethics Committee, and Uganda National Council of Science and Technology. A consultative approval was obtained from the Danish National Board of Research Ethics. Parents or guardians willing to participate signed a written informed consent form, after oral and written information in English and Luganda. Participation in the study did not affect the medical and nutritional treatment, which was similar to that given to admitted children not included in the study. Data was double entered into EpiData (Odense, Denmark) and analysed using Stata 12 (StataCorp LP, College station, USA). To test differences in means between groups, t-tests were used for normally distributed variable, and for others (age, CRP, DDS and HFIAS score), Mann–Whitney rank-sum-tests were used. Chi-square tests were used to test for differences in proportions between groups. For analysis of CRP, children were classified as having CRP levels above or below 10 mg/L. The relationship between CRP and body temperature was examined with linear regression using the log-transformed CRP values. Correlates of oedema were identified using logistic regression, unadjusted and adjusted for sex and age in months, and results presented as odds ratios (OR) with 95% confidence intervals (CI). Additionally, background and clinical characteristics were analysed adjusted for age, sex, HIV-status, and breastfeeding status. Dietary factors and food insecurity were also analysed adjusting for age, sex, breastfeeding status and for month of inclusion, to account for seasonality.