Zanzibar is comprised of two main islands approximately 40 km from mainland Tanzania: Unguja (2012 population = 896,721; 1,666 km²) and Pemba (2012 population = 406,848; 998 km²).15,16 This study was conducted at a purposive sample of 6 of 38 health facilities in Zanzibar where routine deliveries occur (3 facilities in Unguja [Mnazi Mmoja Hospital, Mwembeladu Hospital, and Kivunge Health Center] and 3 facilities in Pemba [Chake Chake Hospital, Wete Hospital, and Micheweni Health Center]) from August of 2011 to September of 2012. These facilities were engaged in an ongoing program supported by the US Agency for International Development (USAID) designed to improve maternal and newborn health services. Although these six hospitals covered approximately 82% of hospital deliveries on both islands, 50% of women in Zanzibar deliver at home.17 The peak rainfall period occurs from March to June, and the peak malaria transmission period is from April to July.18

Residents of Unguja or Pemba delivering at a study facility were considered eligible if their antenatal care (ANC) card indicated they had received no doses of IPTp SP. Women were enrolled 24 hours/day every day of the week by the clinic staff. Consecutive sampling was used to enroll all eligible women during the 12-month study period. However, because of the added burden on the clinic staff, no listing was maintained of women who were eligible but not enrolled. The 12-month enrollment ensured that both low- and high-transmission seasons were represented.

Women who delivered at the six facilities were screened by examining their ANC card, which contains a designated space for recording doses of IPTp SP received. No additional questions were asked, because eligible clients were typically in active labor. Eligible women who agreed to participate were read an informed consent statement by trained maternity nurses or nurse midwives, and their verbal consent was obtained and recorded. No additional assent options were offered to minors. A brief form, including age of mother, gravidity, hemoglobin or hematocrit and week of gestation at which the measurement was taken, whether it was a singleton or multiple birth, outcome of the birth, and birth weight of the baby, was filled out by the health worker with information extracted from each consenting mother's ANC card. The form was placed into the client's file and transferred to the delivery room with the client.

After management of the birth, a trained maternity nurse or nurse midwife at each facility obtained a placental dried blood spot (DBS) specimen from the maternal side of the placenta. A single deep incision was made, and five drops of blood were drawn up with a pipette and dotted onto Whatman 903 Protein Saver Card filter paper labeled only with the woman's study identification (ID) number.19,20 DBS specimens were labeled, packed with desiccant, and transferred to the University of California, San Francisco for polymerase chain reaction (PCR) to detect infection with Plasmodium (all species). Although exact timing was not monitored, collection of each specimen generally occurred within 30 minutes of delivery. After collection of the specimen, the placenta was disposed of per the facility's routine procedures. Multiple births were noted on the client's study form. In the case of twins with one placenta, one ID was issued. In the case of twins with two placentas, each placenta was assigned a unique ID number and tested. Corresponding placental ID numbers were noted on the client form.

DNA from each DBS was extracted using the QIAamp DNA Microkit following the manufacturer's instructions (Qiagen, Germantown, MD). Plasmodium DNA was amplified by 18S ribosomal DNA PCR, with the species-specific nested round confirming falciparum species.21 The presence of PCR product was evaluated using agarose gel electrophoresis stained with ethidium bromide. Testing of control samples extracted from DBSs with known parasite densities confirmed a limit of detection of < 10 parasite/μL.

The required sample size was calculated to be 1,825 assuming a parasitemia prevalence of 5% with a 95% confidence interval with precision of ±1% point. The sample size was calculated assuming independence of observations, because intracluster correlations could not be assessed, but the expected prevalence was liberally estimated to help account for unknown variables. Initial results indicated prevalence well under the liberal prevalence estimate used in the sample size assumptions, indicating that fewer samples were needed to attain the necessary power for the study, but data collection continued to allow for coverage throughout both the rainy and dry seasons.

Client forms were entered into an Access database and date stamped, and hard copies were filed chronologically along with the consent forms. Data on PCR status were provided in an Excel spreadsheet, and the ID numbers were matched between the two files. Cleaned data were exported into SPSS, version 20 and Stata, version 11 for analysis. Although we attempted to enroll all eligible women, given the total number of deliveries and the expected proportion of women who did not receive IPTp (25%),14 it is likely that some of the eligible women were not enrolled. Given that the participants in this study were a non-probability sample of all eligible women, the 95% confidence interval given for the proportion of women with placental parasitemia who did not receive IPTp is considered a precision interval rather than a confidence interval.

Ethical approval was provided by the Zanzibar Research Council, Johns Hopkins School of Public Health Institutional Review Board, and the US Centers for Disease Control and Prevention, Atlanta.

In total, 1,411 pregnant women were enrolled at the time of delivery. Data and specimens from 74 enrollees were not analyzed because of mislabeling, duplicate IDs, missing client cards, or loss. After exclusion of these 74 women, our analysis included 1,339 women from Pemba (N = 694) and Unguja (N = 645). The average age of enrolled women was 27 years (range = 15–48 years), and 433 (32%) women were primigravidae (Table 1). The proportion of primigravidae varied by facility (from 14% of the enrolled women at Mwembeladu to 42% of the enrolled women at Mnazi Mmoja). The enrollees included both singleton (N = 1,326) and twin (N = 13) pregnancies. Birth outcomes experienced by these women included 1,308 (97%) live births, 22 (2%) stillbirths, and 73 (5%) LBW deliveries (Table 1). The majority of enrolments (67%) occurred in the first 4 months of the study (Figure 1), likely because of SP stockouts preceding and during this period. Overall, 6% of all deliveries at the six facilities were enrolled (ranging from 4% to 15%). Although only 23% of all deliveries at Zanzibar health facilities occurred on Pemba in 2011, deliveries from Pemba comprised 52% of the total enrolment in this study.

In total, 1,349 placental specimens were processed and analyzed for presence of malaria infection by PCR (Table 2). Nine (0.7%; precision estimate = 0.2–1.1%) were positive for P. falciparum, with no other Plasmodium species detected.

Placental infection was detected during 6 of 12 months of enrolment, with no seasonal variation. Unguja and Pemba accounted for six (66%) and three (33%) of the positive placental specimens, respectively (Figure 2). Detection of cases in the low-transmission season (September to December) was likely the result of increased study enrolment eligibility during these months as a result of SP stockouts (Figure 1). Seven (78%) positives originated from two of six enrolment facilities: Mwembeladu Hospital and Micheweni Health Center on Unguja and Pemba, respectively. In Pemba, placental infection (N = 3) was only detected at Micheweni, but at least one placental infection was detected at each of three Unguja facilities.

Birth outcomes among nine placental infections included one stillborn infant of LBW (< 2.5 kg). None of the remaining women with placental parasitemia delivered an LBW infant. Three women with placental parasitemia had hemoglobin tests; two women were anemic (hemoglobin < 11 g/dL) (Table 3). The proportion of anemic women at time of delivery was similar in women with and without placental malaria. Three positive specimens came from primigravidae women, proportionate to the percent of primigravidae sampled in the study.

This study is limited to the cross-sectional prevalence of placental parasitemia among a consecutive sample of women delivering in health facilities in Zanzibar, but more than one-half of the women in Zanzibar deliver at home. The representativeness of our study sample may be questioned, because it is not unreasonable to presume that women delivering at home are less likely to receive IPTp and thus, more likely to be at risk for malaria infection. However, the high level of concordance among the results of our study, MEEDS, and findings from ANC screening provides reassurance that these results are valid. Receipt or non-receipt of IPTp SP was determined solely by reviewing the client ANC card at delivery, which means that the quality of information for enrolment eligibility was limited by the recording practices of ANC providers. In most cases where women had not received IPTp SP, it was indicated rather than just an absence of a marking on the card. Because SP is not widely available outside of ANC (in 2011, the Zanzibar Food and Drug Board banned the sale of SP and artemisinin monotherapies from private drug shops), the proportion of women who could have received SP outside of ANC (and thus, been misclassified as not receiving SP) is negligible. National survey data confirm that the proportion of women reporting receipt of any antimalarial in pregnancy (85%) is virtually identical to the proportion reporting having received SP from the ANC (84%).14 Because of SP stockouts, most women were enrolled in low-transmission months (September to December), potentially resulting in lower prevalence of placental malaria. Although we attempted consecutive enrollment of all eligible women, it is likely that it did not occur. Because no record was maintained of women who were not enrolled, to minimize the study-associated workload among busy labor/delivery ward staff, the exact number of eligible women remains uncertain. However, failure to enroll eligible women was likely related to the workload at the time that the woman presented, which is unlikely to be related to the outcome of interest. This limitation did not likely compromise the generalizability of our results to all pregnant women who did not receive IPTp delivering at these facilities. Because the facilities were selected purposively from the facilities supported by a USAID-funded program designed to improve maternal and newborn health services, the generalizability of the findings to all pregnant women who deliver at health facilities in Zanzibar cannot be assured. The use of histopathology, rather than PCR alone, would have allowed us a better estimation of the prevalence of malaria throughout a longer proportion of the pregnancy; however, it was beyond the scope of this study. Finally, because this study did not enroll women who did receive IPTp SP, no information was collected on the current protective efficacy of IPTp SP in Zanzibar.

As malaria transmission in Zanzibar reaches the pre-elimination phase, MIP policy and strategies must be suitably adjusted. This study found a negligible level of malaria among women who had not received IPTp SP. Given the extremely low levels of malaria in both the general population and pregnant women, discontinuing IPTp SP is one reasonable option for Zanzibar to consider. A shift away from IPTp SP as part of Zanzibar's MIP control strategy should only be considered if increased commitment is made to ensure uninterrupted, high-quality access to malaria diagnosis and treatment during pregnancy combined with enhanced measures to drive current transmission levels down closer to zero. We further recommend enhanced surveillance of MIP through reinforcement of the existing surveillance system to ensure that cases among pregnant women are reported from both ANC and outpatient settings in a timely manner. Efforts should continue to ensure high ownership and use of ITNs, particularly among women of reproductive age. Findings from ongoing studies in higher-transmission areas on the cost effectiveness of integrating screen and treat programs into ANC services may not be directly relevant in Zanzibar, where prevalence of malaria is much lower. The Zanzibar Ministry of Health should consider an internal review of costs and findings from surveillance to inform on whether the cost of screening every pregnant woman in ANC is a sound public health investment.