Background
Approximately 35 million pregnant women are at risk of malaria infection each year in sub-Saharan Africa [
1]. Adverse consequences of malaria infection during pregnancy include maternal anaemia, intra-uterine growth retardation [
2], preterm delivery [
3], stillbirth [
4,
5] and low birth weight [
6]. Low birth weight is associated with a marked increase in neonatal death [
7‐
10].
Meta-analysis of malaria in pregnancy (MiP) studies conducted in Eastern and Southern Africa between 1990 and 2011 showed that 32.0 % (95 % confidence intervals [CI], 25.9–38.0; N = 47,433) of pregnant women attending antennal care (ANC) facilities had peripheral parasitaemia. When the time period was restricted to studies conducted between 2000 and 2011, parasitaemia was 29.5 % (95 % CI, 22.4–36.5; n = 18,375) [
11]. These estimates were calculated using a standard method for correcting errors of magnitude based on the known specificity and sensitivity of individual diagnostic methods [
12].
Various diagnostic methods can be used in malaria detection. In this study microscopy and polymerase chain reaction (PCR) methods were employed. One advantage of standard microscopy is that the method requires a relatively short time for diagnosis when used in areas of high transmission and parasites are present in high concentrations [approximately 1000 parasites per microlitre (µl) of blood]. However, if parasite densities are very low, examination of each slide is labour-intensive [
13]. Moreover, the sensitivity and specificity are greatly influenced by the skills and workload of microscopists [
13]. In contrast, the use of PCR for the diagnosis of malaria is highly sensitive and consistent in the detection of parasites; PCR has the ability to measure infections where parasite counts are as low as 5 per µl of blood [
14,
15]. However, some problems arise with false-negative results when DNA isolation is inappropriate [
16,
17].
In Zambia, malaria is endemic in all 10 provinces and
Plasmodium falciparum is responsible for approximately 95 % of all cases [
18]. Since 2006, the National Malaria Control Programme has conducted a Malaria Indicator Survey (MIS) every 2 years to measure the prevalence of malaria by microscopy in children under 5 years of age in selected sites. The prevalence of malaria parasitaemia decreased in Zambia from 2006 to 2010 in some regions, while little change was observed in others [
19]. Of concern is that parasitaemia declined between 2006 and 2008 in Eastern, Northern, and Luapula Provinces, but was higher in the 2010 survey. There has been a 48.3 % decline in the national overall malaria prevalence in children under 5 years of age between 2006 and 2012 (22.1 versus 14.9 %), although there was only a 2.5 % reduction in Luapula Province (32.9 versus 32.1 %) [
20].
Zambia revised its national malaria drug treatment policy between 2000 and 2005 to adopt artemisinin combination therapy (ACT) as the first-line treatment for uncomplicated malaria [
21]. Quinine is the first-line treatment for uncomplicated malaria in pregnancy in the first trimester, whereas ACT is for use in second and third trimesters. For complicated malaria, parenteral quinine is recommended in all trimesters; sulfadoxine-pyrimethamine (SP) is used by policy for intermittent preventive treatment of malaria in pregnancy (IPTp) and also serves as a drug of choice for people who cannot tolerate ACT and during periods of ACT stock-outs in health facilities [
22]. The Zambian malaria policy states that IPTp-SP doses are to be administered during pregnancy at scheduled ANC visits, spaced 1 month apart after 16 weeks of gestation. The 2012 MIS reported that 84.6 % of rural and 93.0 % of urban women who had given birth in the five preceding years reported taking at least one dose of IPTp-SP [
22].
Infection with HIV has been shown to impair the capacity of pregnant women to control peripheral and placental infection [
23]. A review of 11 studies showed that HIV-infected women experience more peripheral and placental malaria, higher parasite densities, and more febrile illnesses, severe anaemia, and adverse birth outcomes than HIV-uninfected women [
24].
The prevalence of HIV among ANC attendees in Zambia is estimated to be 16.8 % in urban and 8.3 % in rural areas [
25]. In 2010, national HIV testing among ANC attendees was estimated to be 94.0 % [
26], and virtually all pregnant women are now tested for HIV. Zambia is listed among the 22 priority African countries with the highest numbers of pregnant women living with HIV who are in need of antiretroviral therapy for the prevention of mother-to-child transmission of HIV [
27]. According to the 2013 ‘
Towards Universal Access report on HIV and AIDS’, Zambia has made notable progress with coverage of maternal antiretroviral treatment (prophylaxis and therapy). The coverage of antiretroviral drugs for prevention of mother-to-child transmission, excluding single dose nevarapine, was estimated to be >95 % (95 % CI, 87—>95 %) [
27].
An estimated 200,000 pregnancies in Zambia are at risk of malaria each year [
28], however the MIS has not reported any estimates of malaria infection among pregnant women to date. Thus, this study was conducted to determine the prevalence of peripheral parasitaemia and risk factors for malaria infection from a cross-section of first ANC attendees in Luapula Province, an area known to have intense malaria transmission [
29]. According to the Zambia Demographic and Health Survey 2013–2014, 94.6 % of women between 15 and 49 years of age in Luapula Province reported having received ANC from a skilled health care provider during pregnancy for their most recent birth in the preceding 5 years [
30]. The coverage of IPTp-SP reported in the MIS at provincial level was as follows: 89.7, 76.6, and 57.6 % of women took at least one dose, two or more doses, and three or more doses of IPTp-SP, respectively [
20]. The prevalence of HIV among ANC attendees at the two recruitment sites for the study, Kashikishi and Nchelenge health centres, in 2014 was 13.0 and 13.3 %, respectively [
31]. This estimate combined new cases and already known HIV positive cases.
Discussion
This study provides a robust estimate of the prevalence of peripheral malaria infection among pregnant women of Nchelenge District. A high burden of
P. falciparum malaria diagnosed by PCR was observed in this population. The prevalence of malaria detected by microscopy was much lower than that detected by PCR, suggesting that a considerable proportion of this study population had sub-microscopic infections (smear negative but PCR positive). Sub-microscopic infections have been associated with placental malaria and decreased maternal haemoglobin [
41]. Although no association was found between sub-microscopic infections and adverse maternal and foetal outcomes, the importance of these infections in pregnancy cannot be ignored in malaria policy formulation, especially when factors such as their role in maternal morbidity and malaria transmission are considered [
41]. Moreover, pregnant women are a significant reservoir of gametocyte transmission and will require particular attention in elimination efforts [
42].
Seven samples that were positive by microscopy in this study were found negative by PCR. Because PCR diagnostic method is more precise than standard microscopy [
43], we would expect that a malaria sample found positive by microscopy would also be positive by PCR. The observation above could be attributed to a number of factors including malaria infection due to other
Plasmodium species and insufficient DNA extraction. A negative PCR test result may be an outcome if the amount of blood collected on filter paper was less than 25 µl, coupled with low parasitaemia.
The prevalence of malaria infection by microscopy was very similar to the prevalence of parasitaemia observed in children less than five-years-old in 2012 (32.1 %, 114/356) in the MIS in Luapula Province [
20]. Another study conducted in the same area among children less than 10 years of age reported a similar parasite prevalence, 30.2 % (236/782) by microscopy [
29].
Analysis of monthly-reported district-level malaria cases among pregnant women showed that MiP in Zambia decreased between 2010 and 2013, although persistent hot spots were reported in the southeast and northeast of the country [
44].
The prevalence of malaria among 375 pregnant women in the study conducted in Nchelenge District between February and April 2013 by Siame et al. was much lower than what was observed in the current study, 15.0 % by microscopy and 22.0 % by PCR [
45]. The differences in the prevalence of malaria in the two studies could be attributed to a number of factors. Firstly, in the current study women attending the Nchelenge and Kashikishi health centres were recruited, whereas in the study by Siame et al. a third health centre, Kabuta, was included in addition to these two. Secondly, in the study by Siame et al. pregnant women who sought ANC services were included, regardless of visit number; in this study, only first ANC attendees who had not taken anti-malarial treatment in the previous month before recruitment were enrolled. Some of the participants who were included in the study by Siame et al. had received IPTp-SP during their earlier ANC visit(s) as well as anti-malarial treatment before enrolment, and it is unclear what time period may have elapsed between treatment and subsequent screening. The prevalence of malaria is likely lower in a group of women exposed to any IPTp-SP because even one dose of IPTp-SP has been shown to protect against maternal parasitaemia [
46].
There was a difference in malaria infection and parasite densities among HIV-infected and HIV-uninfected women, a finding similar to other studies [
23,
24,
47‐
51]. The proportion of malaria infection detected by microscopy and by PCR in this study was consistently higher in HIV-infected than HIV-uninfected women. HIV-infected women had a 46 % increased risk of malaria infection detected by PCR at first ANC visit (
P = 0.045). However, the association between HIV and malaria infection was not statistically significant.
Previous studies have reported stronger association between HIV and peripheral malaria detected by microscopy in pregnant women. A hospital-based study in Zimbabwe in 2000–2001 reported a high odds of malaria parasitaemia in HIV-infected women compared to HIV-uninfected women (OR 3.96; 95 % CI, 2.42–6.46) [
50]; a study in Rwanda conducted between 1992 and 1993 reported that the risk of parasitaemia was moderately higher in HIV-infected compared to HIV-uninfected women (adjusted risk ratio 1.40; 95 % CI, 1.1–1.6,
P = 0.016) [
47]; a similar finding was reported from Kenya in a study conducted between 1996 and 1999 among HIV-infected women in the third trimester (risk ratio 1.70; 95 % CI, 1.52–1.90) and at delivery (risk ratio 1.56; 95 % CI, 1.34–1.81) [
51]. The prevalence of malaria parasitaemia among HIV-infected women at first ANC visit was higher than among HIV-uninfected women in a study in Malawi conducted between 1987 and 1990 (56.6 versus 43.6 %,
P < 0.001) [
49].
The use of antiretroviral therapy was not stated in all the studies above and, considering the years in which these studies were conducted, it can be assumed that coverage of antiretroviral therapy among pregnant women was very low if not non-existent. In the current study, however, the coverage of antiretroviral therapy at first ANC visit was 42.7 % among all HIV-positive women, newly-tested and known HIV-positive cases. Some of the women who were screened in the current study were receiving co-trimoxazole prophylaxis although coverage data of cotrimoxazole at recruitment was not collected. However, data from delivery showed that 37 % of HIV-infected study participants had received cotrimoxazole prophylaxis during pregnancy. Thus, the relatively a weaker association between HIV-infected status and malaria parasitaemia observed in the current study compared to the previously reported studies could be due to the fact that some of the HIV-infected women were on ARV and/or cotrimoxazole prophylaxis.
There was no difference in the geometric mean of parasite density in HIV-infected women on antiretroviral therapy and women who were not on treatment. This may be attributable to newer HIV infections. Women who were not already on antiretroviral therapy may not have progressed to a state of highly compromised immunity since some of them tested positive for the first time at their ANC booking. Furthermore, eligibility to receive antiretroviral drugs in this community was linked to CD4 count. Therefore, if women known to be HIV-positive were not on therapy, it was likely that their CD4 count was above 500 cells per cubic millimetre of blood.
The fact that bed net usage was much lower in HIV-infected women than in HIV-uninfected women, despite there being no significant difference in bed net ownership in the two groups, is surprising because one might expect the opposite. It is unclear if low bed net usage confers an additional risk of malaria infection in HIV-infected women in this study group as would be expected considering that no association was found between bed net use and malaria infection in this study. The significant difference in use of bed nets suggests that there could be differences in disease prevention behaviours between the two groups.
In the univariate analysis, the risk of infection was strongly associated with gravidity and age, but this was not observed in the multivariate analysis. Studies have shown that in malaria-endemic areas, the prevalence of malaria, both clinical and asymptomatic, is highest in young women and primigravidae and paucigravidae (primigravidae and secundigravidae combined) [
2,
52,
53]. This is due to acquisition of semi-immunity that is gravidity-dependent such that malaria infection tends to be less prevalent and less severe among multigravidae [
2,
6,
52]. However, HIV infection has been shown to impair the ability of multigravidae to manage malaria infection [
23,
51]. The fact that HIV prevalence was highest among multigravidae in this study population is a plausible explanation for the fact that there was no association between parity and malaria infection in multivariate analysis.
The low coverage of IRS and bed net use in an area of intense transmission may partially explain the high prevalence of malaria infection. Despite the progress that has been made in scaling-up malaria control interventions, a high burden of malaria still remains in Nchelenge District [
19,
54]. Several reasons may underlie this observation including population movements from high endemic areas (internally and across borders), increasing parasite resistance to insecticides [
29,
54], as well as homes having been built close to water sources that serve as breeding grounds for
Anopheles mosquitoes.
The risk of malaria infection was higher among women at Nchelenge health centre compared to Kashikishi health centre. Since there was no difference in socio-demographic characteristics and HIV infection at the two sites, this difference in the risk of malaria infection may be attributable to the intensity of malaria transmission within the catchment areas of Kashikishi and Nchelenge health centres. Apart from providing health services to the people living close to the Lake Mweru, Nchelenge health centre also provides health services to people in villages that are close to a rubber plantation, the Kenani River, and other in-land water bodies in the Robert Mutepuka village area which provide more breeding sites for mosquitoes.
Malaria transmission intensity is known to decrease during the winter months, April to August, in Nchelenge District [
18]. Thus, the estimated prevalence of MiP from this study conducted during the hot rainy season from November to April is probably higher than the prevalence at other times of the year. Zambia has different malaria transmission zones ranging from very low, to low-moderate and moderate-high [
55]. Thus, the prevalence of malaria estimated in this study can only be cautiously extended to areas with similar transmission patterns.
The prevalence of P. falciparum malaria detected by PCR in pregnant women of Nchelenge District is high. HIV infection increased the risk of malaria infection, although our sample size was not large enough to conclude this risk was statistically significant. Robust vector control, the provision of IPTp-SP, and community education on the importance of using the available interventions to prevent and to treat MiP are vital to reducing the malaria burden in pregnancy. The high prevalence of malaria infection that we observed in this population suggests that past prevention efforts have had limited impact in pregnant women.