High burden of malaria infection in pregnant women in a rural district of Zambia: a cross-sectional study

This was a prospective cohort study of ANC attendees involving recruitment at ANC booking with follow-up to delivery. The study site was the catchment area for two health centres in Nchelenge District which is located on the shores of lake Mweru and has a population of 173,680 [32]. Recruitment of participants was done at the two health centres between November 2013 and April 2014, the high malaria transmission period.

The main study was designed to estimate the prevalence of malaria, curable sexually transmitted and reproductive tract infections and their co-infection, to explore determinants of poor birth outcomes in pregnancy, and to assess in vivo efficacy of SP over 28 days following administration.

The sample size of 1086 was based on the assumption that the incidence of adverse birth outcomes (low birth weight, stillbirth, preterm delivery or intra-uterine growth retardation) among women who have a malaria infection or a curable sexually-transmitted and reproductive tract infections infection during pregnancy would be at least 10 %. Thus, the sample size had 80 % power with 95 % CIs to detect risk factors that have an odds ratio >2 (n = 984); 10 % (n = 102) more pregnant women were recruited to account for expected losses to follow-up as observed in another MIP study in Zambia [33]. Women were enrolled upon providing informed written consent if they had not been exposed to anti-malarial and/or antibiotic therapy within the previous 4 weeks, were willing to have a member of the study team record their HIV test results following routine HIV screening, and had a gestational age of <32 weeks.

Data and sample collection were conducted alongside the provision of routine ANC activities. Routine HIV testing was done using finger-prick blood with two rapid test kits (Determine^® HIV-1/2 [Abbott] and Uni-Gold™ Recombigen^® HIV-1/2 [Trinity Biotech]). Women were generally tested together with their spouses/partners. Individuals found to be seropositive using Determine^® HIV-1/2 were then tested with Uni-Gold™; HIV status was concluded based on the result of the second test.

Trained field workers administered a questionnaire to participating woman in a private room. The questionnaire was used to collect information on socio-demographics, malaria prevention interventions, HIV status and obstetric history. Finger-prick blood was collected and used to prepare a thick smear from each participant for diagnosis of malaria by microscopy. Approximately 160 μl of peripheral blood was placed on four circles of Whatman^® filter paper and labelled with a unique identifying number for each participant and was subsequently used for malaria diagnosis by PCR. IPTp-SP was administered by direct observation to participants whose gestational age was >16 weeks. Women who were receiving cotrimoxazole prophylaxis were not given IPTp-SP due to toxicity concerns. Strips of filter paper containing blood spots were then air-dried and individually stored in envelopes pending deoxyribonucleic acid (DNA) extraction for malaria diagnosis by PCR. Thick blood films were stained using 10 % Giemsa solution and read by two independent microscopists. Parasite density was determined by assuming 8000 white blood cells (WBCs) per µl and counting the number of parasites per 200 WBCs. In cases where fewer than nine parasites were counted against 200 WBCs, parasites were counted against 500 WBCs. Two-hundred high-power fields were read before declaring a slide negative. If results between two microscopists were discordant, a third observer read the slide to determine the diagnosis and parasite count.

DNA extraction was carried out using the Chelex method as described elsewhere [34]. P. falciparum was detected using the same nested PCR method described by Snounou et al. [35], with modifications to the PCR parameters [36]. Briefly, all PCR reactions were carried out in total volumes of 25 µl using Thermo Scientific^® Dream Taq PCR Master Mix (2X) and 0.5 µM of each primer and 2 µl of template; 2 µl of the primary amplicon was used as a template in the secondary reaction. The primary reaction PCR programme was as follows: initial denaturation at 94 °C for 3 min, followed by 30 cycles of denaturation at 94 °C for 1 min, annealing at 60 °C for 2 min and extension at 72 °C for 2 min. The final extension after the 30 cycles was at 72 °C for 2 min. The secondary PCR reaction consisted of 30 cycles of reaction as above and the annealing temperature was at 55 °C. Negative controls were included in all extraction batches and, separately, a positive and a negative control were added to all PCR runs. The secondary amplicon was analysed by electrophoresis on 2 % ethidium bromide stained agarose gel and visualised under ultra-violet light.

Data were double-entered in EpiData version 3.1 software [37], cleaned, processed and analysed using Stata software version 13 [38]. Frequencies of all variables were generated to check for missing data. Variables were then recoded. Data on sources of income, level of education, and fixed and durable assets (type of roof and floor, type of fuel used for cooking, ownership of assets such as television, fridges and radios) were used to create an index of household wealth using principal components analysis [39]. Geometric means and medians were used to describe non-normally distributed continuous variables such as parasite density and age. Characteristics of study participants were described using percentages and the Chi-squared test was used to assess differences in proportions between women recruited at the two sites and between HIV-infected and HIV-uninfected women. Standard statistical tests (i.e. Mann–Whitney U test and t tests) were used to assess significance (P < 0.05) of differences in the distribution of continuous variables. Parasite densities were log-transformed before conducting t-tests or analyses of variance. The prevalence of malaria infection, determined by microscopy or by PCR, and their 95 % CI, were estimated. The pre-specified primary outcome was malaria measured by PCR, therefore, results obtained from PCR diagnosis were used to assess potential risk factors for malaria infection. Univariate analyses of potential predictors of malaria infection were conducted using logistic regression, and crude odds ratios were estimated. Potential predictors that showed significance at P < 0.1 in the univariate analysis were entered into a multivariable model and assessed using a likelihood ratio test. Factors that were found to be independently associated with malaria infection at P < 0.05 were entered in a final model and adjusted odds ratios obtained. Independent variables in the final model were checked for interaction.

The study protocol was approved by the University of Zambia Biomedical Research Ethics Committee (reference number 004-02-13) and the London School of Hygiene & Tropical Medicine Observational/Interventions Research Ethics Committee (reference number 6292). Participation was voluntary and participants could withdraw at any time. To ensure confidentiality interviews were conducted privately and each participant was assigned a unique identifying number that was used to anonymize biological samples, questionnaires, and result record forms.

Women and/or their partners who tested HIV positive for the first time underwent assessment for eligibility to receive antiretroviral therapy and cotrimoxazole prophylaxis as per routine procedures. All women, except those currently receiving cotrimoxazole prophylaxis, were provided IPTp-SP according to national guidelines. In the rare event that a woman showed up for ANC booking with symptomatic malaria, routine laboratory testing was done using a rapid diagnostic test kit (SD Bioline Malaria Ag P.f/Pan brand; Standard Diagnostics, Inc.) and treatment was administered based on national policy guidelines. Strengthening the reporting of observational studies in epidemiology (STROBE) guidelines (Additional file 1) were followed in reporting this study [40].

Of the 1237 women who came for ANC booking, 1095 (89 %) were eligible for enrolment. Only 0.8 % (n = 9) of eligible women refused to take part in the study. A total of 142 women were ineligible for the following reasons: six were not pregnant, 116 came after 32 weeks gestation; 13 had taken anti-malarial treatment in the previous 4 weeks; five had taken anti-malarial treatment in the 4 weeks prior to recruitment and were more than 32 weeks pregnant, and two had taken antibacterial agents in the previous 4 weeks. One woman withdrew consent among the 1086 participants who had been successfully recruited.

The median age of participants was 25 years (interquartile range 20–30) and the median gestational age was 22 weeks (interquartile range 19–25.5). Nearly one-half of the participants, 49.2 % (n = 534) reported ownership of a bed net (treated or untreated combined). Of 531 bed net owners, 78.3 % (n = 416) reported having slept under a bed net the previous night. Overall, only 38.5 % (n = 416) of participants slept under a bed net the previous night. Table 1 presents characteristics of women according to the recruitment sites. No association was found between site of recruitment and socio-demographic characteristics of the participants. Socio-demographic details of participants are presented in full within Table 1.

The prevalence of HIV infection was 13.2 % (95 % CI, 11.3–15.3) and the highest burden was observed in multigravidae (Table 2). The prevalence of malaria infection measured by microscopy was 31.8 % (95 % CI, 29.1–34.6) and by PCR was 57.8 % (95 % CI, 54.9–60.8). Parasite density ranged from 64 to 24,760 parasites per µl of blood with a geometric mean of 1082 (95 % CI, 962–1217) asexual parasites per µl of blood. Parasite density was highest among primigravidae with geometric means being significantly different across gravidae (P < 0.001). Of the 343 malaria positive samples detected by microscopy, seven were negative for P. falciparum by PCR.

Of the 1074 women whose results were available for malaria as detected by PCR, 8.4 % (n = 90, 95 % CI, 6.7–10.2 %) were co-infected with HIV. Table 3 presents characteristic of HIV-infected and HIV-uninfected women. Infection with HIV was associated with age, gravidity and bed net usage on the night before enrolment. Among women who tested positive for HIV, combined with those with known HIV-positive status, 42.7 % were receiving antiretroviral therapy at the time of recruitment. The prevalence of malaria among HIV-infected and HIV-uninfected women detected by PCR was 65.2 % (90/138) and 56.7 % (531/936), respectively (P = 0.059). In the case of microscopy, 31.3 % (294/939) of HIV-uninfected had parasitaemia compared to 35.0 % (49/140) among HIV-infected women (P = 0.382). HIV-infected women had significantly higher parasite densities than HIV-uninfected women (P < 0.001). Parasite density among HIV-infected women receiving antiretroviral therapy and those who were not at recruitment was not significantly different: 1310 (95 % CI, 693–2476, n = 15) and 2131 (95 % CI, 1408–3224, n = 34), respectively (P = 0.189).

Figure 1 shows the prevalence of malaria infection among study participants each calendar month during the 6-month recruitment period from November 2013 to April 2014. In the univariate analyses of recruitment site, age, marital status, gravidity, bed net ownership, bed net usage, wealth quintiles, and HIV status were associated with malaria infection at P < 0.1. In the multivariate analysis of HIV status, site of recruitment and wealth quintile were independently associated with malaria infection (Table 4). Wealth quintile was not associated with malaria infection in a model with HIV infection and site of recruitment. In the final model, the risk of malaria infection was higher among women recruited at Nchelenge health centre (adjusted odds ratio = 1.81; 95 % CI, 1.38–2.37, P < 0.001) and

HIV-infected women across both centres (adjusted odds ratio = 1.46; 95 % CI 1.00–2.13, P = 0.045). There was no interaction between HIV status and site of recruitment (P = 0.751).