Each year 30 million pregnancies occur in areas of stable malaria transmission in sub-Saharan Africa [1]. Malaria in pregnancy can adversely affect the health of the mother and that of the unborn child and, if untreated, can lead to severe anaemia, maternal death, pregnancy loss or intrauterine growth restriction and/or preterm delivery leading to low birth weight [2]. In sub-Saharan Africa, it is estimated that about 40 % of all pregnancies would experience Plasmodium falciparum placental infection without malaria in pregnancy (MiP) preventive interventions, which would result in an estimated 900,000 (Credibility Interval: 530,000–1,240,000) low birth weight deliveries every year [3]. However, MiP and its adverse effects are preventable. Through the use of the lives saved tool (LiST model) developed by the global Child Health Epidemiology Reference Group [4], it has been estimated that 94,000 neonatal deaths were prevented between 2009 and 2012 through MiP interventions, and a further 200,000 could have been averted if coverage with MiP interventions were to have met the 2010 international target of 80 % [5].

WHO recommends effective case management, prevention through insecticide-treated bed-nets (ITN) and intermittent preventive treatment (IPTp) with sulfadoxine–pyrimethamine for the control of malaria in pregnancy [6], which are proven to be efficacious and cost-effective interventions delivered through the antenatal care (ANC) platform. WHO recommendations, and Kenyan national guidelines state that IPTp should be administered at every ANC visit as early as possible during the 2nd trimester. However, there is wide discrepancy between ANC coverage and MiP intervention coverage pointing to substantial missed opportunities. The Kenya Malaria Indicators Survey carried out in 2010, found that 86 % of pregnant women in Kenya received antenatal care from a medical professional [7, 8]. However, coverage of at least one dose of IPTp with SP was 46 % and only 25 % of women received two or more doses (with at least one dose received through ANC). Although the proportion of women who receive two doses of IPT during pregnancy increased from 13 % in 2007–2009 to 21 % in 2010. Only half (49 %) of pregnant women reported that they slept under an ITN the previous night in 2008–09 [7]. A recent systematic review of 98 studies undertaken in sub-Saharan Africa points to obstacles at all levels of implementation: from healthcare providers, health facility, as well as at higher levels of the health system [9]. Implementation research is needed to identify barriers to the scale-up of MiP interventions that can be used by local decision makers. The aim of this study was to measure the effectiveness of ANC to deliver IPTp and ITNs, and to gain an understanding of where bottlenecks occur in the delivery process. The study does not intend to measure coverage of MiP interventions, which would require assessment of the whole pregnancy period (i.e. after birth) at the population level.

The study was approved by the ethics committees of the Kenya Medical Research Institute (KEMRI), the U.S. Centers for Disease Control and Prevention (CDC), the London School of Hygiene and Tropical Medicine, and the Liverpool School of Tropical Medicine. Health workers gave signed informed consent at the initial meeting for structured questionnaires and for structured observations. Pregnant women gave signed informed consent for observations and exit interviews immediately prior to beginning the ANC process. Permission from the District Health Management Teams was sought and heads of health facilities were informed before initiation of data collection.

Data collection took place between first of March and 28th of May 2010 in nine of the ten selected health facilities. The Nursing Officer in charge of one health facility refused permission to collect data. A total of 792 pregnant women were enrolled and observations were carried out for 753 participants while exit interviews were conducted with 613. For 39 participants enrolled, no observation was available as they did not receive antenatal care on the day of the survey and were therefore not included in the analysis: 23 women were sent home by the health facility staff due to lack of time either because they arrived late or because of long waiting time; six women came to the clinic but there were no ANC healthcare providers at the health facility on that day; four women were sent home as they came before their recommended ANC return visit date; two were sent to the labour ward; one was sent home as she didn’t have money to pay for the ANC laboratory tests and one was referred to another health facility with a laboratory to do the required ANC laboratory tests. Furthermore, two observations could not be carried out due to refusal from one healthcare provider and withdrawal from one participant after providing informed consent. Thirty-three participants completed part of the ANC process and were told to come back for the remaining processes on another day. These 33 women were included in the analysis. One health facility, the mission hospital, was excluded from the systems effectiveness analysis as the IPTp policy was not implemented in this facility.

Systems effectiveness of delivery of IPTp-SP

Out of the 748 observations in the eight health facilities, 546 participants were eligible for IPTp according to national policy (Table 4). Of the 202 not eligible, 117 (16 %) were HIV positive and assumed to be taking daily cotrimoxazole, 14 had a gestation <16 weeks and/or hadn’t felt the baby move, and gestational age was unknown for 86 participants (note these are not mutually exclusive categories). Of the eligible women, only 45.8 % were given IPTp by DOT per policy. This was similar for level 4 (39.8 %) and levels 2 and 3 combined (53.3 %) (Fig. 3 a). Two intermediate steps were identified as ineffective in the delivery of IPTp in level 4 facilities, namely: being given any SP (74.0 %) and being given IPTp by DOT (67.3 %). For health facilities level 2 and 3, the only ineffective step was being given IPTp by DOT (70.1 %) as 80.3 % of eligible women were given SP during their ANC visit.

	Level 4			Levels 3 and 2
	n	Intermediate	Cumulative	n	Intermediate	Cumulative
		% (95 % CI)	% (95 % CI)		% (95 % CI)	% (95 % CI)
With DOT
IPTp eligiblea	304			242
SP in stock	304	100.0	100.0	242	100.0	100.0
Receive SP during visit	208	68.6 [60.2, 75.9]	68.6 [60.2, 75.9]	198	80.3 [61.7, 91.1]	80.3 [61.7, 91.1]
Receive three doses SP during visit	206	99.0 [96.1, 99.7]	67.8 [60.6, 74.3]	187	94.1 [77.5, 98.7]	75.5 [55.4, 88.5]
Took SP by DOT	121	64.6 [22.2, 92.1]	39.8 [12.5, 75.4]	129	70.1 [22.1, 95.1]	53.3 [21.0, 83.1]
With or without DOTb
IPTp eligiblea	208			220
SP in stock	208	100.0	100.0	220	100.0	100.0
Being given SP during visit	140	67.5 [60.3, 74.0]	67.5 [60.3, 74.0]	181	79.7 [60.8, 90.9]	79.7 [60.8, 90.9]
Being given three doses SP during visit	139	99.2 [96.2, 99.9]	67.0 [60.6, 72.8]	170	93.4 [74.5, 98.6]	74.5 [53.6, 88.1]
Has SP on exit	52	88.0 [63.3, 96.9]	40.3 [18.2, 67.2]	50	93.4 [45.9, 99.6]	54.0 [17.6, 86.5]
Knows to take three tablets SP	48	92.1 [78.1, 97.5]	37.2 [17.7, 61.9]	49	96.4 [59.6, 99.8]	52.5 [16.3, 86.2]
SP by DOT	80	58.3 [19.3, 89.1]	39.1 [15.9, 68.6]	118	71.0 [22.9, 95.3]	52.9 [20.2, 83.2]
Knows to take three tablets SP or took SP as DOT	128	92.1 [87.4, 95.1]	61.7 [55.7, 67.4]	167	97.1 [70.6, 99.8]	72.3 [46.9, 88.6]

^aIPTp eligibility according to Kenya national guidelines was women not taking cotrimoxazole (or being HIV positive as a proxy for cotrimoxazole use), having felt the baby move (i.e. past quickening) or being 16 weeks gestation or over

^bAnalysis limited to participants completing exit interviews where information on availability of SP at exit was collected

ANC antenatal care; CI confidence interval, DOT directly observed therapy, IPTp intermittent preventive treatment SP sulfadoxine–pyrimethamine

When either having SP at exit and having the knowledge of how to take the three tablets at home or taking IPTp by DOT were considered as the final step in the delivery process, only the 576 participants with exit interviews were included. Of these, 428 were eligible for IPTp. The cumulative effectiveness was higher at 61.7 % for level 4 and 72.3 % for levels 2 and 3 combined respectively (Fig. 3b) and the only ineffective step in the delivery process was being given any SP at level 4 facilities (Table 4).

Systems effectiveness for delivery of ITNs

The overall cumulative effectiveness for the delivery of ITN during a first ANC visit was 65.3 % which is better than for the delivery of IPTp per policy (Table 5). Hospitals (level 4 health facilities) performed similarly to dispensaries and health centres combined (levels 2 and 3 respectively) where 63.1 and 67.4 % of first ANC attendees respectively were offered and took an ITN (Fig. 4). The ineffective step in the ITN delivery process for level 4 facilities was a woman being offered an ITN. Despite the cumulative effectiveness for level 2 and 3 facilities not reaching 80 %, all intermediate steps were considered effective (completed by >80 % of women). Stock-outs of ITNs was recorded in four of the eight health facilities included in this analysis restricted to first ANC visits, varying from 1 to 4 days of the survey.

Steps for ITN delivery	Level 4			Levels 3 and 2
	n	Intermediate	Cumulative	n	Intermediate	Cumulative
		% (95 % CI)	% (95 % CI)		% (95 % CI)	% (95 % CI)
Attend ANC	142	100.0		135	100.0
ITN in stock	127	90.0 [76.9, 96.1]	90.0 [76.9, 96.1]	115	82.6 [47.6, 96.1]	82.6 [47.6, 96.1]
Given ITN during consultation by healthcare provider	91	70.1 [54.3, 82.2]	63.1 [55.9, 69.7]	94	83.0 [61.2, 93.8]	68.6 [38.0, 88.6]
Women took ITN	91	100	63.1 [55.9, 69.7]	93	98.4 [81.8, 99.9]	67.4 [35.6, 88.6]

ANC antenatal care, CI confidence interval, ITN, insecticide-treated net

This study found that the systems effectiveness of ANC to deliver IPTp was unacceptably low and only slightly better for ITNs. The majority of health facilities failed to deliver these interventions to 80 % of eligible women. The findings illustrate clear missed opportunities for protecting women and their unborn babies from the adverse effects of MiP and the need to maximize the public health impact of these cost-effective interventions.

Despite presenting to ANC and being eligible for IPTp, over 54 % of pregnant women did not receive a dose of IPTp by DOT and 33 % did not receive any SP (either by DOT or to take home) despite SP being in stock. The system effectiveness was slightly better for health centres and dispensaries compared to hospitals with 80 % of eligible women receiving some SP for IPTp, although only 53 % received the correct dose by DOT overall, compared to 62 and 40 % in hospitals for any IPTp and IPTp by DOT, respectively. Although stock-out of SP has been reported as a major issue in previous studies, in this study no stock-outs were observed on the days of visit despite the fact that facilities reported SP stock-outs in the preceding 12 months [9]. These cumulative system effectiveness estimates cannot be compared to overall IPTp coverage (i.e. 25 % for Kenya in 2010 [8] or 17 % for Nyanza Province [7]) as the presented estimates are derived from a cross-section of pregnant women not capturing the IPTp coverage for the whole pregnancy period and represent IPTp eligible women coming to ANC whereas estimates from demographic health surveys and/or malaria indicator surveys are based on household surveys involving women who gave birth in the previous years disregarding if they have been to ANC or if they had a contraindication, such as taking cotrimoxazole and fail to capture pregnancy not ending in a live births (estimate close to 30 % of pregnancies for Kenya [1]). The system effectiveness for delivery of ITNs was better than for IPTp, but still underlines missed opportunities with close to 35 % of first ANC attendees leaving a health facility without an ITN. Even though all the intermediate steps for the delivery of ITNs (namely presenting for a first ANC visit; ITNs being in stock at ANC on the day; being offered an ITN and taking the ITN), were effective for health centres and dispensaries, only 67 % of women coming for a first ANC visit left these facilities with an ITN. In hospitals, only 70 % of women were offered an ITN when they were in stock and the cumulative effectiveness was 63 %. Furthermore, missed opportunities were highlighted by 37 women (5 %) who presented to the clinic but didn’t receive ANC services for a variety of reasons ranging from late arrival at ANC, absent ANC healthcare providers, presenting before the recommended ANC return visit date, and need for ANC laboratory tests that were not available at the facility. Thirty-three (4 %) women didn’t complete their ANC visit, mainly due to delays at the laboratory, and were asked to come back the next working day. Of these only 3 % received IPTp and 12 % received an ITN. It is unknown what proportion of women returned to complete their visit as scheduled by the healthcare provider, but given the known barriers for attending ANC (such as commitment to farming, childcare, employment and transport cost) this is likely to be a small proportion [9].

A household cross-sectional study conducted concurrently alongside the health facility survey [12] found low IPTp coverage, with 59 % of recently pregnant women who attended ANC at least once between 4 and 9 months gestation reporting to have received one dose of SP. The household survey showed that among women who visited ANC at least twice during pregnancy 27 % received two doses of IPTp and only 14 % by DOT. Due to the cross-sectional nature of the health facility survey, the authors were unable to assess the proportion of pregnant women receiving at least two doses of IPTp during pregnancy, which requires data for the whole pregnancy period. Findings from the household survey suggest that the major bottlenecks occur at the health facility level, as reflected by the discrepancy between the proportions of women who attended ANC at least twice in an eligible gestation (78 %) and those receiving at least two doses of IPTp (27 %). Three quarter of participants who received either dose of IPTp said they received it by DOT, which is higher than the proportion observed in the health facility survey (58 % for hospitals and 65 % for health centres and dispensaries combined).

A related study evaluating the health systems effectiveness of ANC to deliver MiP interventions in Mali [15, 18] found that delivery of IPTp through ANC was ineffective, with 0 and 25 % of eligible women receiving IPTp by DOT at the district and community levels, respectively. The cumulative systems effectiveness of being given IPTp with or without DOT was also better in Mali, at 56 and 67 % for the district and community levels respectively. Despite the different study/country contexts, such as the higher HIV prevalence in Kenya and differing national guidelines on IPTp, similar cumulative effectiveness were observed in Kenya, with 62 % for hospitals and 72 % for health centres and dispensaries. As with the Kenya findings, the ineffective steps in Mali were being given any SP and receiving IPTp by DOT, with higher-level facilities having the least effective delivery.

Similar to the Mali study, there was a significant clustering of effective and ineffective steps and outcomes by health facility as reflected by the high design effects. Steps such as receiving IPTp by DOT or being given SP to take home varied across facilities with some facilities always and others never giving IPTp by DOT. This clustering calls attention to the need for targeted interventions to improve the delivery of MiP interventions at the health facility level. These could be implemented through enhanced supervisory visits to selected facilities.

This study provides insights on the health facility level barriers to effective delivery of MiP interventions, however, several limitations should be noted. First, the approach used in the study presented here differed from the health facility survey carried out in Mali in that survey staff were placed at different stations in ANC to capture all consenting women on the day of the survey. In Mali, each enrolled participant was followed by one study staff from the beginning of the ANC process until they exited the clinic. Although in Kenya all women could be observed on the day, in busy health facilities not all participants could be interviewed at exit as the turnover was usually faster than the time it took to complete the interview.

Secondly, the authors were not able to exclude women who received SP within the last 4 weeks which could under-estimate the cumulative systems effectiveness by including IPTp ineligible participants in the analysis. Assuming about a third of pregnant women make the minimum recommended number of four visits (34 % according to the concurrent household survey [12]), this could have been the case for some of the revisits. Also participants with unknown gestational age (4 %) could not be included in the IPTp effectiveness analysis as their eligibility could not be established. However, including these participants assuming that they were beyond the first trimester (median gestation at first ANC visit was 5 months in household survey, well into the second trimester [12]) only slightly affect the cumulative effectiveness of receiving IPTp by DOT (it was 42 % for hospitals and 52 % for lower level facilities compared to 40 and 55 % overall, respectively, and the cumulative effectiveness for with or without DOT were 55 and 74 % compared to 62 and 72 %, respectively).

ITN stock information was not available for nine survey days from two health facilities for which stock-out was assumed, however if ITNs were in stock on those days the cumulative effectiveness would not have been affected but the intermediate step of being offered an ITN in health centres and dispensaries becomes ineffective at 69 %. There could have been errors in data collection as data from observations may reflect the Hawthorne effect [19], data from exit interviews from social desirability bias [20], and data extracted from ANC registers may have been incomplete or erroneously abstracted. However, given the low observed cumulative effectiveness for each of the endpoints it seems unlikely that these factors would have an impact on the outcomes and, if anything, would over-estimate ANC delivery effectiveness. Lastly, the selected mission hospitals could not be included as one facility did not implement the policy for IPTp, and ITNs and SP for IPTp were out of stock for the whole study period and the other mission hospital refused to take part in the study. There is limited information on quality of care or system effectiveness of ANC for delivery of MiP interventions in these settings. It will be important for future studies to be inclusive of the mission run health facilities and private sector, given that care in that sector is growing and it provides health services to a significant proportion of the population in sub-Saharan Africa [21].

This study identified that the delivery of MiP interventions through the ANC platform in this setting—namely the delivery of IPTp and ITNs in hospitals, and of IPTp by DOT at all health facility levels—was ineffective. The substantial variation observed between health facilities suggests that future intervention studies to increase uptake of IPTp and ITNs, as well as other ANC services, should focus on identifying facility level problems and evaluating customized solutions. Special attention should be given to the private sector, for which very little information on ANC and quality of care in general is available. Overall, the ANC systems effectiveness for delivering MiP interventions was suboptimal and highlights missed opportunities to protect pregnant women and their babies from the adverse consequences of malaria. These are cost-effective interventions with the potential of saving many lives. Qualitative data collected alongside the quantitative data in this survey will provide further insights on the reasons for these missed opportunities and potential tailored interventions.