Does anthropometric status at 6 months predict the over-dispersion of malaria infections in children aged 6-18 months? A prospective cohort study

Background: In malaria-endemic settings, a small proportion of children suffer repeated malaria infections, contributing to most of the malaria cases, yet underlying factors are not fully understood. This study was aimed to determine whether undernutrition predicts this over-dispersion of malaria infections in children aged 6-18 months in settings of high malaria and undernutrition prevalence. Methods: Prospective cohort study, conducted in Mangochi, Malawi. Six-months-old infants were enrolled and had length-for-age z-scores (LAZ), weight-for-age z-scores (WAZ), and weight-for-length z-scores (WLZ) assessed. Data were collected for ‘presumed’, clinical, and rapid diagnostic test (RDT)-confirmed malaria until 18 months. Malaria microscopy was done at 6 and 18 months. Negative binomial regression was used for malaria incidence and modified Poisson regression for malaria prevalence. Results: Of the 2723 children enrolled, 2561 (94%) had anthropometry and malaria data. The mean (standard deviation [SD]) of LAZ, WAZ, and WLZ at 6 months were − 1.4 (1.1), − 0.7 (1.2), and 0.3 (1.1), respectively. The mean (SD) incidences of ‘presumed’, clinical, and RDT-confirmed malaria from 6 to 18 months were: 1.1 (1.6), 0.4 (0.8), and 1.3 (2.0) episodes/year, respectively. Prevalence of malaria parasitaemia was 4.8% at 6 months and 9.6% at 18 months. Higher WLZ at 6 months was associated with lower prevalence of malaria parasitaemia at 18 months (prevalence ratio [PR] = 0.80, 95% confidence interval [CI] 0.67 to 0.94, p = 0.007), but not with incidences of ‘presumed’ malaria (incidence rate ratio [IRR] = 0.97, 95% CI 0.92 to 1.02, p = 0.190), clinical malaria (IRR = 1.03, 95% CI 0.94 to 1.12, p = 0.571), RDT-confirmed malaria (IRR = 1.00, 95% CI 0.94 to 1.06, p = 0.950). LAZ and WAZ at 6 months were not associated with malaria outcomes. Household assets, maternal education, and food insecurity were significantly associated with malaria. There were significant variations in hospital-diagnosed malaria by study site. Conclusion: In children aged 6-18 months living in malaria-endemic settings, LAZ, WAZ, and WLZ do not predict malaria incidence. However, WLZ may be associated with prevalence of malaria. Socio-economic and micro-geographic factors may explain the variations in malaria, but these require further study. The iLiNS-DOSE and iLiNS-DYAD-M studies were conducted in four facilities: one public district hospital (Mangochi), one mission hospital (Malindi), and two rural public health centres (Lungwena and Namwera) in Mangochi District, Southern Malawi. The total catchment population of 180,000 largely subsisted on farming and fishing. Mangochi site is low-lying at an altitude of ~ 485 m above sea level, but traversed by the Shire River (the largest river in Malawi). Two of the study sites (Lungwena and Malindi) are also low-lying with the similar altitude along the eastern shore of Lake Malawi. In contrast, Namwera lies at the top of Namwera Hills, bordering Mozambique, at an altitude of ~ 900 m above sea level and is far from the large water bodies. Namwera experienced higher rainfall and cooler temperatures than the other three study sites [18, 19]. In Malawian children aged < 5 years, the prevalence of malaria (by microscopy), diarrhoea and acute respiratory infections were 24.3%, 22% and 5%, respectively, with seasonal fluctuations [2, 20]. The sub-tropical climate comprising a warm, wet season from November to April, a cool, dry winter season from May to August, and a hot, dry season from September to October [21] is favourable for the Anopheles mosquitoes which transmit Plasmodium parasites. Plasmodium falciparum is the most dominant and causes about 98% of all malaria infections in Malawi. Malaria transmission occurs throughout the year with highest transmission rates occurring between October and April (rainy season), mainly in low-lying and high temperature areas. The data for this analysis were taken from the iLiNS-DOSE and iLiNS-DYAD-M studies—two large community-based randomized controlled trials conducted in rural Malawi. In the iLiNS-DOSE study, 6-months old children were randomly allocated to one of five intervention groups provided with different doses or formulations of LNS or to a control group that did not receive LNS during the 12-month study period, between November 2009 and May 2012. In the iLiNS-DYAD-M study, pregnant women < 20 weeks’ gestation were randomly allocated to one of three groups to receive iron and folic acid (IFA), multiple micronutrients (MMN) or a small-quantity (20 g) of LNS daily. After delivery, women in the IFA group received placebo tablets, while MMN and LNS supplementation was continued up to 6 months postpartum. Children of mothers in the LNS group also received LNS 10 g twice daily from 6 to 18 months. This study was conducted from February 2011 to April 2015. Details of design, randomization and enrolment for the two studies can be found in the main outcome papers [15, 16]. In both studies, research assistants visited the children’s homes every week from age 6 to 18 months to interview the guardians about the child’s health in the previous 7 days using a structured questionnaire. The information was complemented by a picture calendar filled out by the guardians daily to aid memory of their children’s morbidity status. These were done to minimise problems of recall associated with community morbidity assessments [22, 23]. The use of maternal interviews to collect data on child morbidity has been validated in previous studies [24, 25]. The research assistants referred all cases of fever to the nearby health facility for a malaria rapid diagnostic test (RDT) and treatment with artemether/lumefantrine, the nationally recommended anti-malarial drug. The children were followed throughout the year, covering periods of both high and low malaria transmission. Facility health workers were trained to collect data on clinical diagnosis, RDT, and/or malaria microscopy results whenever the child was treated at the health facility. In addition, all children had malaria microscopy tests done during the scheduled study clinic visits at age 6 months and 18 months. Blood smears were obtained from all children at the time of the blood sampling for biochemical assessments, stained with 2% Giemsa for 30 min. All thick slides were reviewed by two microscopists using a high-power microscope to determine the presence of malaria parasites. Discrepant readings were resolved by a third reviewer. Malaria RDT was also done at 6 months scheduled clinic visit using rapid test kit (Clearview Malarial Combo, Alere, South Africa). Anthropometric assessments were done at 6 months and 18 months. Study anthropometrists measured the infant’s length with a high-quality length board (Harpenden Infantometer; Holtain Limited) and recorded it to the nearest 1 mm. They weighed unclothed infants with electronic infant weighing scale (SECA 735; Seca GmbH & Co), recording to the nearest 10 g. The anthropometrists were trained and their measurement reliability was verified at the start of the study and at 6-months intervals thereafter with methods adapted from the procedures used in the WHO Multicentre Growth Reference Study [26]. The anthropometrists calibrated all equipment with standard weights and length rods daily. Study nurses collected 5–7 mL of blood by venepuncture using a 23-gauge needle into 7.5 mL evacuated, trace element-free polyethylene tubes containing lithium heparin (Sarstedt Monovette, NH4‐heparin, Sarstedt Inc., Newton, NC, USA). The blood tube was immediately inverted 10 times to mix the heparin anticoagulant with the blood to prevent clotting. A small aliquot of the whole blood was pipetted out and used to analyse haemoglobin (Hb) on the Hemocue 201+ system (Hemocue, Brea, CA, USA). The tube containing the remaining whole blood was then placed in an insulated cooler with ice packs and processed within 2 h of collection. Trained laboratory staff then aliquoted whole blood into microcuvettes and measured zinc protoporphyrin (ZPP) concentration from unwashed venous blood sample using a haematofluorometer (206D, AVIV Biomedical Inc., Lakewood, NJ, USA). At enrolment, research assistants interviewed the mothers to obtain household level information including assets, number of children, maternal education (years spent in school) and maternal age. Household food security was assessed using Household Food Insecurity Access Scale (HFIAS) [27]. Use of insecticide treated bed nets (ITNs) was assessed by asking the guardians the number of days that the child slept under a bed net during the week preceding the interview at 6 months. Site was defined based on the clinics where the two studies were conducted (i.e. Namwera, Mangochi, Malindi, and Lungwena). The primary outcome for this analysis was the incidence of ‘presumed’ malaria, derived from the weekly morbidity data. ‘Presumed’ malaria diagnosis was defined as history of fever either reported by the guardians or tympanic temperature ≥ 38 °C measured by the research assistants during the home visit. An episode of ‘presumed’ malaria was defined as the period starting from the day the child had malaria symptoms, preceded by at least 2 days of no symptoms or no data. The episode ended on the last day the child had malaria symptoms, if followed by at least 2 symptom-free days. Fever episodes accompanied by respiratory signs or diarrhoea were excluded from the malaria diagnosis. Secondary outcomes included (1) incidence of clinical malaria, taken from the malaria diagnosis made by the health worker in the absence of a diagnostic test whenever the child visited a health facility, (2) incidence of confirmed malaria, taken from the malaria diagnosis made by the health worker confirmed by a positive RDT, and (3) prevalence of malaria parasitaemia derived from the malaria microscopy results at age 18 months. Malaria incidence was calculated as total episodes of malaria/total child years at risk; malaria parasitaemia prevalence was calculated as the proportion with a positive malaria parasite slide. Length for age z-scores (LAZ), weight-for-age z-scores (WAZ), and weight for length z-scores (WLZ) were calculated from the anthropometry data using WHO growth standards [26]. Iron deficiency at age 6 months was defined as whole blood ZPP > 70 µmol/mol haem [28]. HFIAS z-scores were generated by summing the value of responses to nine questions regarding food insecurity: the higher the score, the higher degree of food insecurity in the last 4 weeks [27]. Household asset scores were defined as the principal components score based on baseline ownership of a set of assets and household quality: the higher the score, the better the living conditions. Number of children aged < 5 years was defined as number of children below the age of five who were part of the participant’s household at 6 months. All children who had malaria data at any point from age 6 to 18 months were included in the analysis. Negative binomial regression was used to assess the association of LAZ, WAZ, and WLZ (independent variables) with the incidence of malaria (dependent variable), and Poisson regression (with a robust variance estimator) [29] to assess the association of LAZ, WAZ, and WLZ with prevalence of malaria parasitaemia. To study the independent effect of various predictors, multivariate regression models were constructed that included potential predictors collected at 6 months. The following were potential predictors: household asset scores, maternal age (centred around the mean), and education, number of children aged < 5 years, HFIA score, whether the child received the study intervention (LNS) or not, adjusted for the additional control group [MMN] in iLiNS-DYAD, sex of the child, Hb concentration, iron status, month of birth, daily use of ITNs, distance to health facility and study site. Collinearity among the predictors was tested using the collin stata command. If the predictors were highly collinear (> 0.5), the one that was less strongly associated with the outcomes was dropped. The results are reported as incidence rate ratios [IRR] or prevalence ratio [PR] and their 95% confidence intervals (95% CI) at p = 0.05. Robust standard errors were computed to adjust for correlation of recurrent malaria episodes in a single child. Other potential predictors were considered including immunization status, markers of inflammation (C-reactive protein and alpha1-acid glycoprotein concentrations) at 6 months, maternal and child malaria immunity, and maternal HIV status. However, these variables were available only from a subsample of the two studies, hence were eventually dropped from the final models to maximize the sample size. Furthermore, these variables showed little effect on the model during sensitivity analysis. Stata version 14 (StataCorp, Texas, USA) was used for the main analyses.