Impact of Plasmodium falciparum malaria and intermittent preventive treatment of malaria in pregnancy on the risk of malaria in infants: a systematic review

MEDLINE, EMBASE, Global Health, and Malaria in Pregnancy (MiP) Library [31] databases were searched from inception to 22 May 2018. All review authors participated in developing the search strategy and AK conducted the search. MEDLINE, EMBASE and Global Health databases were searched via the Ovid interface using Medical Subject Headings (MeSH) of key search terms. The MiP library [31] was searched using key search terms, including malaria, Plasmodium falciparum, falciparum malaria, pregnancy, placenta, maternal, foetal, prenatal, utero, new-born, infant, child, offspring, and preschool. The search was limited to journal articles reporting human studies and published in the English language. Additional studies were identified by scrutinising reference lists of studies that met the inclusion criteria and relevant review articles. A bibliography of included studies was shared with other experts in the field of MiP to assess whether all the relevant articles had been retrieved.

Lists of titles and abstracts of retrieved articles were exported to Endnote and duplicates were removed. Retrieved titles and abstracts were assessed for eligibility at the level of titles and abstracts by the corresponding author. Full articles and abstracts of potentially relevant articles and those where there was uncertainty about whether to include or exclude the article were retrieved and assessed for eligibility by two independent reviewers (AK and DC). Where there was disagreement, it was resolved by discussing with the rest of the authors.

Data were extracted by one reviewer (AK) and verified by a second reviewer (DC). A data extraction matrix in an Excel spreadsheet was developed and piloted prior to data extraction. The Excel spreadsheet included the following variables: author and year of publication, country where the study was done, study period (dates of fieldwork), transmission intensity as measured by entomological inoculation rate, study design (considered RCT if IPTp was randomized, otherwise defined as an observational study), length of follow-up, follow-up schedule, study objectives, study population, sample size, long-lasting insecticide treated net (ITN) coverage among study participants, whether IPTp was given, type and frequency of IPTp regimen, whether maternal peripheral malaria parasitaemia was measured, timing of measurement of maternal peripheral malaria parasitaemia and how it was measured, PM detection, PM case definition, when and how clinical malaria or parasitaemia were detected in the infant, study outcomes, proportion of infants born to mothers with peripheral malaria parasitaemia or PM, adjustment for potential confounding factors, losses to follow-up, proportion of infants with outcomes of interest, results including effect sizes with confidence intervals and p-values, and study strengths and limitations. Disagreements between the two reviewers were resolved by discussion. Unresolved disagreements were settled by a third reviewer (either SS or GD or SR). Corresponding authors of included studies were contacted by email for any missing or unclear information. Corresponding authors who did not respond to the first email contact were contacted two more times and if they did not respond at the third contact, they were not contacted any further.

The risk of bias in individual studies was assessed by two reviewers using the Newcastle–Ottawa quality assessment scale for cohort studies [32], and the Cochrane Collaboration tool for randomized controlled trials [33]. Details of how the risk of bias was assessed in cohort studies were published in the systematic review protocol (PROSPERO Registration number: CRD42018088869). Studies were rated on the following categories: selection, comparability and outcome. For the selection category, studies were rated on the following items: representativeness of the exposed group (whether the exposed cohort was truly representative of the average from the community), selection of the non-exposed group, measurement of the exposure (if the exposed and the unexposed were from the same community), and demonstration that the outcome was not present at the start of follow-up. For the comparability category, studies were assessed based on whether the study adjusted for malaria transmission or controlled for malaria prevention during pregnancy using IPTp or ITNs. In the outcome category, studies were rated on the following items: ascertainment of outcome, whether follow-up was long enough for outcomes to occur (follow-up of at least 6 months was considered adequate), and completeness of follow-up (proportion of infants lost to follow, and whether characteristics of those lost to follow-up were reported). Each item in the selection and outcome categories was awarded a maximum of one point. The comparability category was awarded a maximum of two points. Studies were awarded a maximum of nine points. Studies that had a score of nine were rated as having medium risk of bias while those with a score of less than nine were rated as high risk.

A systematic narrative synthesis of the included studies was conducted. A meta-analysis was not conducted because studies had substantial variation in the definition of malaria exposure during pregnancy, type of IPTp given, length of follow-up, and determination of primary outcome. Results were summarized using tables and in text. Tables with summary descriptions of study design, MiP exposure measurement, follow-up time, outcome measurement and results were generated. Studies were grouped in clusters according to the type of exposure (maternal peripheral malaria during pregnancy, or PM or IPTp), and outcome of interest (incidence of malaria in infants, prevalence of malaria parasitaemia, time to first malaria parasitaemia or first episode of malaria).

Overall, 2084 titles and abstracts were identified and retrieved from searches of electronic databases. An additional two records were identified from searching lists and contacting experts in the field (Fig. 1). Of the 2086 records, 461 duplicates were removed, and 1625 records were screened. Of these, 1582 records were excluded after review of the title and abstract, and 29 were excluded for various reasons after reviewing full text articles. Only 14 articles were deemed to be eligible for inclusion in the systematic review.

All 14 studies were conducted in sub-Saharan Africa (Table 1); 10 prospective cohort studies and four RCTs. Of the 10 cohort studies, IPTp-SP was given in seven studies, and IPTp was not given in two [17, 34] because they were conducted before implementation of the WHO recommendation on IPTp. Malaria transmission data in form of entomological inoculation rate (number of infectious mosquito bites per person per year) was reported in half of the included studies; 20.5 [35], 35 [18], 38 [16], 50 [17], 257 [34], and 400 [19, 21]. The duration of follow-up of infants ranged from 1 to 5 years. In 10 studies, infants were followed up from birth to 1 year of age [16, 18‐21, 34‐38]. Maternal malaria exposure was determined by (1) microscopy (N = 5) [18, 36, 37, 39] or DNA PCR of maternal blood [40], (2) microscopy of placental blood only (N = 6) [17, 18, 21, 34, 35, 37], or (3) histology of placental tissue (N = 5) [16, 19, 36, 40, 41]. Malaria infection during infancy was measured as malaria parasitaemia (prevalence of parasitaemia, N = 7; time to first parasitaemia, N = 4), and clinical malaria (incidence of clinical malaria, N = 5; time to first clinical malaria, N = 2). Malaria parasitaemia was assessed during weekly [35], fortnightly [21], or monthly [18, 37] routine visits. Clinical malaria was assessed by active surveillance in three studies [21, 35, 36] and by passive surveillance in 11 of the studies [16‐20, 34, 37‐41]. ITN use in infants was not reported in half of the studies. In five studies, ITN use was reported as 100% [20, 41], 51% [21], 73% [19], and 95% [18]. In one study, 66% of infants were from families that owned at least an ITN at enrolment [35] while in another study, all infants did not have ITNs [34]. The median number of infants born to mothers with maternal parasitaemia was 224 (range 28–236). The median number of infants born to mothers with PM detected by microscopy or histology was 59 (range 15–445). Out of 11 studies which assessed association between PM and the risk of malaria in infants, there were three studies with ≥ 100 infants born to mothers with PM.

The association between maternal parasitaemia and malaria risk in infancy was assessed in three cohort studies [18, 37, 39] and one RCT of IPTp-SP vs IPTp with chloroquine (CQ) vs CQ prophylaxis conducted in Malawi [40] (Table 2). In Malawi, infants born to mothers with malaria parasitaemia detected during the 2nd or 3rd trimester had higher odds of parasitaemia compared to infants born to mothers without parasitaemia during pregnancy, but the difference was not statistically significant [40]. In a cohort study conducted in Uganda, infants born to mothers with any parasitaemia detected by microscopy during pregnancy had a higher risk of parasitaemia during the first year of life compared to infants born to mothers without parasitaemia during pregnancy adjusted for gravidity, birth season, haemoglobin genotype, and residence (risk ratio [RR] 2.97; 95% confidence interval [CI] 1.37–6.42) [37]. In Benin [18], infants born to mothers having parasitaemia in the 3rd trimester had higher prevalence of parasitaemia (odds ratio [OR] 4.16; 95% CI 1.64–10.54), shorter time to first clinical malaria (hazard ratio [HR] 3.19; 95% CI 1.59–6.38) and shorter time to first parasitaemia (HR 2.95; 95% CI 1.58–5.5) compared to those born to mothers without parasitaemia in the 3rd trimester after adjusting for birth season and residence near the lake (Table 2). In the same study, maternal parasitaemia detected during 1st or 2nd trimester was not associated with an increased risk of malaria during infancy [18]. In another cohort study conducted in Uganda, infants born to mothers with malaria parasitaemia at enrolment or at delivery had a higher incidence of clinical malaria during the first 5 years of life compared to infants born to mothers without parasitaemia (HR 1.23; 95% CI 1.01–1.51) [39].

Six studies evaluating the association between PM detected by microscopy only and the risk of malaria in infants produced mixed results (Table 3). The prevalence of parasitaemia was higher in infants born to mothers with PM detected by microscopy compared to those born to mothers without PM (RR 10.42; 95% CI 2.64–41.10) in a cohort study conducted in Uganda [37] while the prevalence of parasitaemia and clinical malaria tended to be lower in Benin [18] though this was not statistically significant (OR 0.72; 95% CI 0.25–2.11). Compared to infants born to mothers without PM, infants born to mothers with PM detected by microscopy had a shorter time to first parasitaemia in Cameroon [34], Benin [35] and Tanzania [21], and a shorter time to first clinical malaria in Gabon (HR 2.1; 95% CI 1.2–3.7) [17]. The only study that adjusted for malaria exposure among other confounding factors showed an association between PM detected by microscopy and time to first parasitaemia but only among infants resident in houses with ITNs (HR 2.13; 95% CI 1.24–3.67) [35]. In another study conducted in Benin, there was no statistically significant association between PM detected by microscopy and time to first parasitaemia (OR 0.68; 95% CI 0.34–1.38) or time to first clinical malaria (HR 0.60; 95% CI 0.28–1.32) in infants [18].

Five studies evaluated associations between PM detected by histology and the risk of malaria in infancy (Table 4). Histology detected PM was associated with an increase in the odds of clinical malaria in Malawi (OR 3.9; 95% CI 1.2–13.0) [40], Tanzania (OR 4.79; 95% 2.21–10.38) [41], and Mozambique (OR 4.63; 95% CI 2.10–10.24) [16] while one study in Cameroon did not find a statistically significant association between histologically detected PM and prevalence of malaria in infants (OR 0.72; 95% CI 0.40–1.28) [36]. Histologically detected PM was associated with a higher incidence of malaria in Malawi (unadjusted incident rate ratio [IRR] 2.3; 95% CI 1.1–4.8) [40], but this was not observed in Ghana (IRR 0.86; 95% CI 0.54–1.37) [19].

Four studies evaluated the impact of IPTp on the risk of malaria in infants (Table 5). Of these, three were RCTs and one was an observational study where some women received IPTp-SP and others received no IPTp. There was no significant difference in the incidence of malaria among infants born to mothers randomized to IPTp-MQ vs IPTp-SP (IRR 0.95; 95% CI 0.81–1.13) in a multicentre RCT conducted in Benin, Gabon, Tanzania, and Mozambique [38] and among infants born to mothers randomized to intermittent screening and treatment of MiP (ISTp) with artemether–lumefantrine (AL) vs IPTp-SP (IRR 0.94; 95% CI 0.68–1.59) conducted in Ghana [19]. In a cohort study conducted in Ghana, the risk of malaria was higher in infants born to mothers who did not receive IPTp compared to infants born to mothers who received IPTp-SP, but this difference was also not statistically significant [20]. In Mozambique, the odds of clinical malaria in infants born to mothers who were randomized to IPTp-SP were higher, but not statistically significantly so, than in infants born to mothers randomized to placebo (OR 1.28; 95% CI 0.90–1.83) [16].

Risk of bias in individual studies was assessed using the Newcastle–Ottawa scale for observational studies (Table 6) and the Cochrane Collaboration tool for RCTs (Table 7). One study where IPTp was randomized was assessed using the Newcastle–Ottawa Scale for assessing risk of bias in observational studies because the study only assessed an association between MiP (and not IPTp regimens) and the risk of clinical malaria or parasitaemia in the infant [40]. All cohort studies had a representative exposed cohort, selected the non-exposed comparison group adequately and measured malaria exposure during pregnancy, and demonstrated that the outcome was not present at the beginning of follow-up. Most of the studies adjusted for IPTp and ITN use but only two studies [20, 35] adjusted for malaria exposure. In three of the studies, the number of participants lost to follow-up or the reasons for losses to follow-up were not reported [16, 36, 41]. Three studies had > 20% losses to follow-up [17, 36, 40]. The overall risk of bias in cohort studies was rated as high in nine studies [17, 18, 21, 34, 36, 37, 39‐41] and medium in two studies [20, 35]. In all the RCTs, allocation concealment was adequate, and no trial was stopped before completion. Overall, the risk of bias in all three RCTs was low.