Placental Malaria is associated with reduced early life weight development of affected children independent of low birth weight

Background: Infection with Plasmodium falciparum during pregnancy contributes substantially to the disease burden in both mothers and offspring. Placental malaria may lead to intrauterine growth restriction or preterm delivery resulting in low birth weight (LBW), which, in general, is associated with increased infant morbidity and mortality. However, little is known about the possible direct impact of the specific disease processes occurring in PM on longer term outcomes such as subsequent retarded growth development independent of LBW. Methods. In an existing West-African cohort, 783 healthy infants with a birth weight of at least 2,000 g were followed up during their first year of life. The aim of the study was to investigate if Plasmodium falciparum infection of the placenta, assessed by placental histology, has an impact on several anthropometric parameters, measured at birth and after three, six and 12 months using generalized estimating equations models adjusting for moderate low birth weight. Results. Independent of LBW, first to third born infants who were exposed to either past, chronic or acute placental malaria during pregnancy had significantly lower weight-for-age (-0.43, 95% CI: -0.80;-0.07), weight-for-length (-0.47, 95% CI: -0.84; -0.10) and BMI-for-age z-scores (-0.57, 95% CI: -0.84; -0.10) compared to infants born to mothers who were not diagnosed with placental malaria (p = 0.019, 0.013, and 0.012, respectively). Interestingly, the longitudinal data on histology-based diagnosis of PM also document a sharp decline of PM prevalence in the Sukuta cohort from 16.5% in 2002 to 5.4% in 2004. Conclusions. It was demonstrated that PM has a negative impact on the infant’s subsequent weight development that is independent of LBW, suggesting that the longer term effects of PM have been underestimated, even in areas where malaria transmission is declining. © 2010 Walther et al; licensee BioMed Central Ltd. Data were collected from children born between January 2002 and July 2005 as part of an ongoing open cohort study at the maternity ward of the Health Centre in Sukuta, a semi-urban Gambian village 30 km south of the capital Banjul, where malaria transmission varies considerably between season, with highest incidence during the wet season and immediately afterwards (‘malaria season’: August – December). The main study, as well as the analysis strategy of the data presented here, was approved by the Joint Gambian Government/MRC Ethics Committee. The newborns delivered at the maternity ward were enrolled after informed consent of the mother was obtained. The purpose of the main cohort was to study immune responses of infants to vaccines and infections. Since overt morbidity, as well as risk factors associated with LBW, are known to increase susceptibility to infectious diseases and thus bias any immunological responses [32-35] only healthy singleton children were eligible for recruitment into the cohort. Newborns below a birth weight of 2,000 g were excluded from the study. A survey among Gambian pregnant women in 2000-01 [36] estimated the prevalence of HIV to be 1.0% (CI: 0.9-2.4%) in Serekunda, an urban settlement approximately 10 km from Sukuta. Based on this estimate, HIV+ cases are unlikely to have significantly impacted on the results of this study, and following ethical guidance HIV testing was not performed. Demographic, anthropometric and clinical data from 783 mother/child pairs were used to investigate the association of maternal PM and height and weight in the first year of life of the offspring. Basic demographic and socio-economic data, such as, ethnicity, age, and parity of the mother, education of the parents and, as a measure for overcrowding, number of persons sleeping in one bedroom, as well as information on malaria treatment and bednet use were collected in face-to-face interviews by trained fieldworkers using questionnaires shortly after birth. Infants were followed up monthly over a 12-month period for morbidity and anthropometric measurements. At each follow up visit trained nurses recorded the number of illness episodes during the previous month reported by the mother or guardian of the child and administered vaccines according to the recommended Gambian Expanded Programme of Immunization schedule. Maternal height and weight were recorded six months after delivery. PM status was defined by placental histology. After birth a placental biopsy was taken and embedded in paraffin and stained for histological analysis of PM status [29]. Cases were classified according to Ismail [37] as active infection (presence of infected red blood cells in the intervillous space of the placenta) and past infection (no parasites, but haemozoin deposition in macrophages). Active infection was further subdivided into acute infections (only parasites and minimal haemozoin deposition) and chronic infections (parasites and significant haemozoin deposition). Childrens’ weight and length measurements at three, six and 12 months of age (+/- 15 days) were used to calculate four age/gender-standardized indicators: ‘weight-for-age’ (wfa), ‘length-for-age’ (lfa), ‘weight-for-length’ (wfl) and ‘body-mass-index-for-age’ (BMIfa) based on the WHO “Child Growth Standards, 2006” using the Stata macro ‘igrowup.ado'[38]. All data were double entered into an Access database (Microsoft), validated and checked for range and consistency. Prior to investigating if PM is associated with infant’s size and nutritional status, univariable analyses were used to assess whether known PM risk factors (young maternal age, low parity, low socio-economic status, no bednet use, and seasonality of malaria transmission) [6] were present in this study population. A logistic regression model was fitted including all risk factors with p-values less than 0.2 in univariable analysis, which were kept in the model at a significance level p ≤ 0.05. The final model including parity, PM season and year, and duration of schooling of the mother was used to calculate adjusted ORs and corresponding p-values of being PM+ comparing the individual exposure groups to a specified baseline group. The hypothesis that exposure to PM would be associated with lower values for anthropometric indicators was tested at each of the three time points three, six and 12 months using univariable analyses first. Additional mother and child characteristics such as maternal age, height, BMI, parity, duration of schooling, living in crowded housing conditions as well as gender of the child, moderate low birth weight of 2,000 g – 2,499 g, current age of the child, month and year of birth, and whether the child was born in the ‘hungry’ season (during the months with intensive rain fall from July to October) [39,40] were investigated as risk factors for reduced indicators. Univariable analyses were conducted to examine the association of each factor with the four anthropometric indicators described above at three, six and 12 months of age. Cross-sectional times series models using generalized estimating equations (GEE) to fit the parameters of the models with exchangeable correlation structure were then constructed for each anthropometric indicator. The final models account for multiple weight and length measurements at three, six and 12 months of age and confounding factors such as sex, moderate low birth weight of 2,000 g – 2.499 g, infant’s age, birth month and year, parity and education (duration of schooling) of the mother. An interaction term was included for PM exposure status of the child and parity to investigate if the association of PM and size of the infant varied between parity groups. Since the association of PM and the growth indicators wfa, wfl and BMIfa was not significantly different for two or three pregnancies, but for more than four pregnancies compared to the first pregnancy, data was split into two groups according to 1-3 and 4 and more pregnancies. Maternal age and living in crowded housing conditions lacked a significant association with z-scores and was therefore not included in the final models. Mother’s height and BMI were not considered, since 61% of the data were missing. No difference was seen for anthropometric measurements taken during or after the ‘hungry’ season; thus “hungry season” was not included in the models. The resulting models were used to assess the effect of PM exposure on each anthropometric indicator independent of birth weight. Whether maternal exposure to PM was associated with being underweight, wasted or stunted (wfa, wfl or lfa z-scores<-2, respectively) was investigated using logistic regression. Odds ratios for the group of wasted or underweight infants and a second group of stunted infants were calculated choosing all the remaining well-nourished infants as the baseline group.