Malaria and Fetal Growth Alterations in the 3^rd Trimester of Pregnancy: A Longitudinal Ultrasound Study

Background: Pregnancy associated malaria is associated with decreased birth weight, but in-utero evaluation of fetal growth alterations is rarely performed. The objective of this study was to investigate malaria induced changes in fetal growth during the 3rd trimester using trans-abdominal ultrasound. Methods: An observational study of 876 pregnant women (398 primi- and secundigravidae and 478 multigravidae) was conducted in Tanzania. Fetal growth was monitored with ultrasound and screening for malaria was performed regularly. Birth weight and fetal weight were converted to z-scores, and fetal growth evaluated as fetal weight gain from the 26th week of pregnancy. Results: Malaria infection only affected birth weight and fetal growth among primi- and secundigravid women. Forty-eight of the 398 primi- and secundigravid women had malaria during pregnancy causing a reduction in the newborns z-score of -0.50 (95% CI: -0.86, -0.13, P = 0.008, multiple linear regression). Fifty-eight percent (28/48) of the primi- and secundigravidae had malaria in the first half of pregnancy, but an effect on fetal growth was observed in the 3rd trimester with an OR of 4.89 for the fetal growth rate belonging to the lowest 25% in the population (95%CI: 2.03-11.79, P<0.001, multiple logistic regression). At an individual level, among the primi- and secundigravidae, 27% experienced alterations of fetal growth immediately after exposure but only for a short interval, 27% only late in pregnancy, 16.2% persistently from exposure until the end of pregnancy, and 29.7% had no alterations of fetal growth. Conclusions: The effect of malaria infections was observed during the 3rd trimester, despite infections occurring much earlier in pregnancy, and different mechanisms might operate leading to different patterns of growth alterations. This study highlights the need for protection against malaria throughout pregnancy and the recognition that observed changes in fetal growth might be a consequence of an infection much earlier in pregnancy. © 2013 Schmiegelow et al. Ethical approval was granted by the Tanzania Medical Research Coordinating Committee (MRCC) (18th of April 2008, reference number NIMR7HQ/R.8a/Vol. IX/688). All procedures were conducted in accordance with the Declaration of Helsinki and Good Clinical and Laboratory Practices. All participants gave informed written consent. The study was conducted in Korogwe District, northeastern Tanzania. Malaria transmission is holoendemic, but has lately shown a remarkable decline [29]. Pregnant women attending the Reproductive and Child Health clinic of Korogwe District Hospital (KDH) or the Lwengera, Kerenge and Ngombezi Dispensaries were enrolled from September 2008 until March 2010 and followed throughout pregnancy. The last delivery occurred in October 2010. The study design has been reported previously [30], [31]. In short, inclusion criteria were: GA≤24 weeks determined by ultrasound, living in an accessible area of Korogwe District for >6 months, and willing to give birth at KDH. Women with preeclampsia and/or twins in the current pregnancy, and fetuses/newborns with severe malformations were excluded from the analysis, since these conditions can severely affect fetal growth [32], [33]. In analyses including BW, stillbirths were omitted. Sensitization campaigns were performed in the villages to reduce selection bias. After inclusion, women were booked for three antenatal visits (ANV) at a GA of 26 (ANV2), 30 (ANV3), and 36 (ANV4) weeks. If needed due to illness women attended extra clinic visits. Maternal age, anthropometric measures, obstetric history, educational level (≤primary school; ≥secondary school), HIV status, and hemoglobin level (Sysmex hematological analyzer®, Kobe, Japan) were documented. The project team performed all investigations except screening for maternal HIV infection which governmental nurses performed only three days a week. Women not completing follow-up were therefore less likely to have their HIV status determined. Transport was offered at the time of delivery and 77.2%, 5.3%, and 17.6% delivered at KDH, another health facility, and at home, respectively. Home visits were performed within one week of the bookings or the estimated date of delivery, if women failed to report at KDH. At all visits venous blood was collected and at delivery both venous and placental blood. Malaria was diagnosed using rapid diagnostic test (RDT) (Parascreen™ Zephyr Biomedicals, Goa, India, Paracheck Pf® Orchid Biomedical Systems, Goa, India or ParaHIT® Span diagnostics Ltd, Surat, India) [34]. Thick and thin blood smears were prepared and evaluated at the end of the study. Women were therefore treated based on the RDT results with Artemether-Lumefantrine (Coartem® Dispersible, Norvatis Corporation Suffern, New York, USA) or Quinine (Quinine sulfate coated tablets, ELYS chemical Industries Ltd, Nairobi, Kenya) (infections occurring in 1st trimester). Women with symptoms or sign of malaria had an immediate blood smear investigation in addition to the RDT. Infections requiring hospital admission were treated with Quinine (Quinine Dihydrochloride Injection BP, Healthcare PVT Ltd, Mumbai India). RDT positive women, who had been RDT positive within two weeks before the ANV, also had an immediate blood smear examination. Parasite HRP-2 antigen can circulate in the blood stream after clearance of the malaria parasites, and only women who were consistently blood smear positive were treated [35]. Blood smears were stained with Giemsa, and asexual parasites counted against 200 (500 if parasite count was <10) leucocytes. One hundred thick film fields were read before a slide was declared negative. The malaria infection was considered symptomatic if the axillary temperature was above 37.5C. Intermittent preventive treatment for malaria (IPTp) with sulfadoxine-pyrimethamine (1500 mg/75 mg) (SULPHADAR®, Shelys Pharmaceutical Ltd., Dar es Salaam, Tanzania) was given as directly observed treatment; 1st dose in the 2nd trimester after a GA of 20 weeks and 2nd dose in the 3rd trimester, at least four weeks apart. If included early in pregnancy the 1st dose was given at an extra clinic visit at a GA of 20. If having received a 1st dose of IPTp before inclusion, but earlier than recommended by WHO (2nd trimester when quickening is felt [36]), the woman received a 2nd dose at week 20 and a 3rd dose in the 3rd trimester. According to the national program, voucher for procuring a bednet was provided and bednet use inquired. Trans-abdominal ultrasound examination was performed at inclusion, ANV2, ANV3, and ANV4 (Sonosite TITAN®, US High resolution ultrasound system, 5–2 MHz C60 abdominal probe, Bothell, Washington state, USA). At inclusion GA was estimated using crown-rump length (crown-rump length<75 mm) [37]) or head circumference of the fetus [38]. If the GA was <11 weeks a new GA estimation was performed at 11–16 weeks of gestation. At ANV2, 3, and 4 fetal weight (EFW) were estimated using the Hadlock algorithms based on head circumference, abdominal circumference and femur length [39]. The performance of the Hadlock algorithms in this population has been reported previously [31]. All investigations were performed by the author CS, assisted by a trained Tanzanian midwife. To account for dependency of growth velocity on initial weight, fetal growth was evaluated as the relative weight gain in g/week*kg in four fixed growth intervals of at least 14 days [28], [40] representing different time-points in pregnancy (ANV2-ANV3 (late 2nd trimester/); ANV3-ANV4 (early 3rd trimester); ANV4-Delivery (late 3rd trimester); ANV3-Delivery (entire 3rd trimester)) using the fetal weight (kg) in the beginning of each growth interval as the reference [27], [28], [41]. Relative weight gain was furthermore dichotomized as the lowest 25% or the highest 75%. At delivery BW, head and abdominal circumference, sex, and placental weight were documented. BW measured within 24 hours of delivery using a spring scale (Fazzini®, accuracy = 50 g, Vimodrone, Italy) (KDH, until July 2009 and all examinations at home) or a digital strain gauge scale (ADE®, accuracy = 10 g, Hamburg, Germany) (KDH, after July 2009) were considered eligible. To exclude variability due to GA and sex, EFW and BW were converted to z-scores using sex-specific standard weight charts developed from healthy pregnancies in the same cohort as a reference [31]. Changes in z-score over time were investigated to further evaluate alterations in fetal growth. Only offspring with an eligible BW and data from at least two growth intervals were evaluated. Primarily, the growth intervals ANV2-ANV3, ANV3-ANV4, ANV4-Delivery were used for evaluation. Alternative growth intervals (ANV2-ANV4, ANV3-Delivery) were used if a woman had incomplete attendance. Changes in z-scores (Δz) were calculated for the malaria negative group and used as a reference. Theoretically, a normally growing fetus will have a Δz close to zero, whereas a Δz>0 represents excess fetal weight gain and a Δz<0 insufficient weight gain. Evaluation of changes in z-scores can be affected by the fetal pulsatile growth pattern [42] and to limit false-positives the 25th centile for Δz was used as a cut-off to define insufficient weight gain. The effect of malaria was categorized in four groups: 1) Normal growth as Δz ≥25th centile in all growth intervals following a malaria infection, 2) Immediate effect as Δz <25th centile in the growth interval when the infection occurred or in the first growth interval after a malaria infection, followed by growth intervals where Δz ≥25th centile, 3) Late effect as a Δz<25th centile in a growth interval not immediately following the infection, and 4) Persistent effect as a Δz <25th centile immediately after a malaria infection and until the end of pregnancy. For analysis, women were considered to be malaria negative if they never had malaria based on RDT and microscopy results from the time they were enrolled. Women, who contracted malaria (RDT and/or microscopy positive), were considered to belong to the malaria group from when they were diagnosed until delivery. In growth intervals preceding the time of detection of the malaria infection they were excluded from the analysis. Hence, women contracting malaria between ANV2 and ANV3 were not included in analyses of the growth interval ANV2-ANV3, but considered malaria positive in all other growth intervals. All data were documented and validated using Microsoft Office Access 2003. Statistical analyses were performed in Stata 10 (Stata Corporation) using, when appropriate, Chi2 test, Fisher’s exact test, Mann Whitney ranksum, and Student t-test, and all with two-sided P-values. The effect of malaria on BW (as z-score) was investigated using multiple linear regression and on relative weight gain (dichotomized as lowest 25% and highest 75%) using multiple logistic regression. Crude and adjusted coefficients/odds ratios were calculated. Factors with a P<0.20 in univariate analysis were entered into the multivariate models. Using a step-wise backward elimination approach final models were obtained including variables with a P<0.10. A P<0.05 was considered significant. Final models only included women without missing values.