Providing lipid-based nutrient supplement during pregnancy does not reduce the risk of maternal P falciparum parasitaemia and reproductive tract infections: a randomised controlled trial

The study methods have been described in detail elsewhere [24]. Briefly, this study was a sub-study of the iLiNS-DYAD-M trial, an outcome assessor-blinded randomised controlled trial that provided antenatal nutrient supplementation to improve pregnancy outcomes and child growth (trial registration: www.​clinicaltrials.​gov, trial identification NCT01239693). We recruited the study participants from antenatal clinics at four health facilities in Mangochi District, southern Malawi. Pregnant women were eligible to participate in the study if they were at least 15 years old, had ultrasound-confirmed pregnancy of <20 gestation weeks (gw), had no chronic medical condition or allergies and no evident pregnancy complications. The catchment population was rural to semi-urban and subsisted mainly on fishing and farming. Maize was the staple food in this population, which faces seasonal food insecurity with a lean period just before harvest time. Malaria transmission was holoendemic in this area.

The enrolled women were randomised into three intervention groups receiving either one tablet of IFA, the standard antenatal nutrient supplement in Malawi; one tablet of MMN which contained 18 micronutrients (including iron and folic acid) or one 20 g sachet of SQ-LNS containing the same 18 micronutrients and four additional minerals, EFAs (linoleic acid and α-linolenic acid), protein, fat and 1.2 g of sucrose daily from enrolment to delivery [24]. The dose of iron was higher for women in the IFA group (60 mg) than for those in the MMN and SQ-LNS groups (20 mg). The lower iron dose in SQ-LNS was chosen so as not to greatly exceed the recommended daily intake of iron for lactating women [28], and because a dose of 20 mg/day has in some conditions been sufficient for preventing iron deficiency anaemia in pregnancy [20, 29]. The doses of the rest of the micronutrients in the MMN and SQ-LNS supplements were at least the recommended dietary allowance (RDA) for pregnancy, or the maximum amount that could be included in the supplement considering technical constraints [20]. The exact ingredients and dosing of the study supplements are shown in Additional file 1: Table S1. The participants also received two doses of IPT-SP according to the Malawi national guidelines at the time, which were concordant with the World Health Organisation’s recommendations [30].

The outcomes reported here were pre-specified secondary outcomes measures. These were peripheral P. falciparum parasitaemia; trichomoniasis; vaginal candidiasis; and bacterial UTI. We diagnosed P. falciparum parasitaemia by rapid diagnostic testing (RDT) at 32 gw, by polymerase chain reaction (PCR) at 36 gw and by RDT and PCR at delivery. RDT at 32 gw was assessed from finger prick blood samples whereas PCR at 36 gw and RDT and PCR at delivery were assessed from venous blood samples. Study nurses performed RDT according to the manufacturer’s instructions using Clearview® Malaria Combo (British Biocell International Ltd., Dundee, UK) which detects the antigens histidine-rich protein 2 and Plasmodium-specific aldolase. We conducted PCR testing for P. falciparum on dried blood spot samples collected from whole blood, which were tested in a real-time PCR assay targeting the P. falciparum lactate dehydrogenase (pfldh) gene [31].

RTIs and UTIs were assessed at one week after delivery using samples from vaginal swabs and mid-stream urine, respectively. The study participant was lying in the dorsal lithotomy position for the collection of the vaginal samples. After digitally separating the labia, a study nurse advanced a cotton tipped swab through the vaginal introitus towards the posterior fornix and rotated the sample back and forth three times before pulling it out. Then the vaginal samples were sent to the laboratory where a laboratory technician processed them immediately by smearing the vaginal discharge from the swab on a microscope slide and performed wet mount microscopy after adding saline solution. We defined trichomoniasis as the visualisation of motile protozoa and candidiasis as the observation of budding yeasts, pseudohyphae or hyphal forms on microscopy. The study nurse performed urinalysis immediately after urine collection using Multistix ® 10 SG reagent strip (Siemens Medical Solutions USA, Inc., Pennsylvania, USA) and UTI was indicated by the presence of nitrite in a freshly voided urine sample [32].

A sample size of 1400 was originally calculated in accordance with the main objectives of the iLiNS-DYAD-M trial. This sample size provided 80% power and 95% confidence to detect an approximately 30% reduction (among the women who received SQ-LNS, as compared to the control women) in the prevalence of maternal RTIs (from 25 to 18.3%) or P. falciparum parasitaemia (from 15 to 9.5%). We conducted statistical analyses using Stata 12.1 (StataCorp, College Station, USA), according to an analysis plan written and published before the intervention code was opened [24]. The analysis was based on the principle of intention-to-treat.

We performed tests for interaction between the intervention, maternal infection and pre-specified covariates using the likelihood ratio test. The pre-specified covariates were maternal age, maternal parity, gestational age at enrolment, socio-economic score, education (completed years at school) and body-mass-index (BMI) as continuous variables; HIV status, P. falciparum parasitaemia, anaemia, season at enrolment and study site as categorical variables. We performed stratified analyses for all covariates that had a statistically significant interaction (p < 0.1).

The study protocol was approved by the College of Medicine Research and Ethics Committee, University of Malawi and the Ethics Committee of Pirkanmaa Hospital District, Finland. The trial was performed according to Good Clinical Practice guidelines and the ethical standards of Helsinki Declaration. Study staff obtained informed consent from the study participants (or their legal guardian in the case of minors) prior to performing any study procedures. An independent data safety and monitoring board monitored the incidence of suspected serious adverse events and performed interim analyses for safety.

We recruited a total of 1391 pregnant women into the iLiNS-DYAD-M study between February 2011 and August 2012. The last delivery occurred in February 2013. The three groups were similar at baseline in terms of the participants’ average demographic and socioeconomic characteristics and nutritional and health status (p > 0.05) (Table 1). The overall rate of usage of LLINs was 70.9% and was similar across the groups (P = 0.536). We excluded 12 women from the analysis because they had twin pregnancies.

Follow up data for P. falciparum parasitaemia were available from 79.6% of the women at 32 gw, 76.9% at 36 gw, and from 81.4 and 78.7% at delivery by RDT and PCR respectively. Loss to follow up was similar between the groups (P = 0.805 at 32 gw, 0.638 at 36 gw, 0.180 and 0.393 for P. falciparum parasitaemia RDT and PCR at delivery). RTI and UTI data were collected from 86.6% of the women. Loss to follow up was similar between the groups (P = 0.958 for RTI and UTI after delivery) (Fig. 1). The participants who were excluded from the analysis tended to be younger (23.8 vs 25.4, P < 0.001); to be more educated (4.4 vs 3.9, P = 0.025); to have a slightly higher BMI (22.6 vs 22.0, P = 0.001); to have a higher wealth index (0.29 vs −0.09, P < 0.001); to be primiparous (31% vs 18%, P < 0.001) and to be anaemic (25% vs 19%, P = 0.014).

The overall prevalence of P. falciparum parasitaemia was 23.3% at enrolment (RDT), 10.7% at 32 gw (RDT); 9.0% at 36 gestation weeks (PCR); 8.3% by RDT at delivery and 20.2% by PCR at delivery. The prevalence of P. falciparum parasitaemia at 32 gw (RDT), 36 gw (PCR), delivery (RDT) and delivery (PCR) was 10.9, 7.6, 8.0 and 17.6%, respectively, in the LNS group, and was similar in the IFA and MMN groups (Table 2). Relative to the IFA group, the risk for P. falciparum parasitaemia.at 32 gw (RDT) was not significantly different in the SQ-LNS (RR 0.91, 95% CI 0.6- 1.36) or MMN (RR 0.79, 95% CI 0.52- 1.20) groups. Similarly, when compared to MMN, the risk for P. falciparum parasitaemia at 32 gw (RDT) in the LNS group was not significantly different (RR 1.14, 95% CI 0.74- 1.76). The risk for P. falciparum parasitaemia at the other time points was also similar in all 3 study groups (Table 2).

The overall prevalence of trichomoniasis, candidiasis and UTI after delivery was 10.5, 0.5 and 3.1% respectively. The prevalence of trichomoniasis, candidiasis and UTI was 8.3, 0.3, 3.2% respectively, in the LNS group, and was similar in the IFA and MMN groups (Table 3). Relative to the IFA group, the risk for trichomoniasis was not significantly different in the SQ-LNS (RR 0.71, 95% CI 0.47- 1.09) or MMN (RR 0.93, 95% CI 0.63- 1.37) groups. Similarly, when compared to MMN, the risk for trichomoniasis in the LNS group was not significantly different (RR 0.77, 95% CI 0.50- 1.18). The risk for UTI was also similar in all 3 study groups (Table 3). The risk ratio for candidiasis was not calculated because the prevalence of candidiasis was very low.

Tests of interaction with potential effect modifiers were not significant (P > 0.1) for maternal age, maternal parity, socio-economic score, HIV status, anaemia, season at enrolment and study site, but were significant for at least one outcome for gestational age at enrolment, maternal BMI, maternal education and baseline P. falciparum parasitaemia. Gestational age at enrolment modified the association between the intervention group and P. falciparum parasitaemia by RDT at 32 gw (P = 0.024); maternal BMI modified the association between the intervention group and trichomoniasis (P = 0.039): maternal education modified the association between the intervention group and P. falciparum by RDT at delivery (P = 0.048) and baseline P. falciparum parasitaemia modified the association between the intervention group and P. falciparum parasitaemia at delivery (by both RDT and PCR, P = 0.058 and 0.078 respectively). However, there were no statistically significant differences between the intervention groups in any of the strata (Additional file 2: Table S2).

We investigated whether the provision of SQ-LNS during pregnancy affects the occurrence of P. falciparum parasitaemia and RTIs among women in a population with generally poor food security. Overall, we found no statistically significant differences in the prevalence of maternal P. falciparum parasitaemia and RTIs between the study intervention groups.

The probabilities of bias or random error in this study were minimised by the large sample size, randomised study design, blinding of outcome assessors and inclusion of several P. falciparum parasitemia measurements using sensitive detection methods (RDT and PCR) [31, 33]. The success of follow-up was not complete, due to loss of participants during the follow up or missed visits or biological samples, but the proportion of missing data was similar across the study groups and several adjusted analyses gave results similar to the main analysis; hence we believe that the missing data did not significantly bias our conclusions. The excluded participants were, however somewhat different from those included in the analysis in terms of some of their anthropometric, sociodemographic and health characteristics. For these reasons our results may not be fully representative of the target population. Furthermore, the methods that we used to detect vaginal candidiasis, trichomoniasis and bacterial UTI have low to moderate sensitivity [34‐36]. This may have led us to underestimate the burden of these conditions in our study population. Finally, the prevalence of each infection (except for P. falciparum parasitaemia at 36gw by PCR) was lower than what we had estimated in our sample size calculation, potentially affecting our power calculations. However, as argued by Feinstein and Concato [37], our point estimates and confidence intervals were consistently in the same direction and sufficiently precise to conclude that SQ-LNS supplementation did not influence the prevalence of maternal P. falciparum parasitemia or RTIs in rural Malawi, though we cannot conclusively rule out an effect on the prevalence of candidiasis and UTI.

The prevalence of P. falciparum parasitemia at delivery was much higher based on PCR testing compared to RDT among the study women. This finding was not unusual and it was probably due to low parasite density in the study population. The sensitivity of RDT tests varies with parasite density, with lower sensitivity at lower parasite density [38].

The prevalence of P. falciparum parasitaemia by RDT continued to decline from enrolment through 32 gestational weeks to delivery in our study population. This finding is unsurprising as similar results were obtained from women who received two doses of IPT-SP in a previous study conducted in the same study area [14] and from elsewhere in sub-Saharan Africa [39]. Assuming no loss to follow-up in our study, we would expect the prevalence of P. falciparum parasitaemia in pregnancy to drop in the period following enrolment into the study as the women were exposed to anti-malaria interventions such as IPT-SP and receipt of LLINs as part of the ante-natal care package.

Little is known about the effect of SQ-LNS on maternal health or morbidity. Other reports from the iLiNS-DYAD trial suggested that SQ-LNS and MMN may have increased the risk of maternal dental infections [40] on one hand but seemed to have had no effect on the development of maternal anti-malarial antibodies [41] on the other hand. Previous in vitro studies have suggested that EFAs and their derivatives such as arachidonic acid and docosahexaenoic acid may have toxic effects on P. falciparum [42], T. vaginalis [43] and C. albicans [27] via different mechanisms, hence our expectation that providing SQ-LNS which contained EFAs would reduce the prevalence of infections caused by these organisms. The apparent lack of benefit from the EFAs in this study could have happened because the study women were not deficient in EFAs to begin with. A recently published trial conducted in the same study area as our study found that the levels of breast milk arachidonic acid and docosahexaenoic acid were above average levels [44] contrary to our earlier assumptions based on previous studies in sub-Saharan Africa [45]. Alternatively, the absence of influence of SQ-LNS on maternal plasma composition of EFAs (Oaks et al., in press) could explain our observation of lack of impact.

A previous study that provided combined vitamin A and zinc supplementation to children in Ghana led to a reduction in the risk of clinical malaria episodes [46] possibly due to the synergistic effects of vitamin A and zinc on immune function [47]. Contrary to our expectation based on these previous studies, SQ-LNS and MMN which contained both vitamin A and zinc were not associated with a lower risk of P. falciparum parasitaemia. The lack of treatment response despite the provision of adequate doses of vitamin A and zinc in MMN and SQ-LNS and reasonable adherence to supplementation [24] could have been due to dietary factors inhibiting zinc absorption [48] or haemodilution in the latter half of pregnancy lowering levels of retinol binding protein resulting in functional vitamin A deficiency [49].