The study illustrated the combined effect of women's timing of ANC attendance, health worker's IPTp delivery and different delivery schedules of national IPTp guidelines on IPTp coverage. This is in line with findings of previous studies from Tanzania showing that low IPTp coverage levels can not be attributed solely to women's late enrolment to ANC [26, 30]. Instead, health worker's IPTp delivery practices and unclear IPTp guidelines led to lost effectiveness of the IPTp strategy. Thus, solutions need to be found at individual, facility and policy level if the government of Tanzania aims at reaching at least 80% of pregnant women with two IPTp doses .
Compared to the restricted IPTp guidelines, pregnant women initiated ANC attendance late, but still in time to receive a first IPTp dose. Pregnant women in Tanzania are recommended to attend ANC clinics for the first time at 16 weeks of gestation . Most women started ANC attendance in their second trimester around 20 weeks of gestation. This is consistent with the national average among facility users  and with findings from other studies [26, 28, 30, 31, 35]. However, contrary to health workers' perception and assumptions in the literature [29, 33, 49, 50], women's late ANC enrolment did not interfere with the IPTp schedule. This has also been stressed by other studies [26, 30, 31]. Only 15% of the women started ANC attendance after 24 weeks of gestation and were therefore according to the restrictive guidelines no longer eligible for a first dose of IPTp. Overall, women's timing of ANC visits matched well with the restrictive IPTp schedule that was practiced in health facilities. The majority of the participants attended the ANC clinic at least once between 20-24 weeks and between 28-32 weeks of gestation. This shows that women's ANC attendance follows health workers' instructions and was not the main cause for low IPTp levels. IPTp coverage could theoretically have been high. The high proportion of women attending during the specific periods is not surprising as women are given return dates by the health workers. Although women's knowledge about the timing of IPTp was not investigated, it can be presumed that women rely on health workers to correctly administer drugs [18, 49]. Findings of Marchant et al.  support this assumption: over 90% of the women who had not received a dose of IPTp said that health workers had not offered it to them. The participants' knowledge on malaria prevention and its effects was not very high. Almost a quarter of the women did not know why they were supposed to get malaria prevention at the ANC clinic. Pregnant women's knowledge concerning IPTp but also women's power and ability to actively demand IPTp and to protect themselves from erratic timing or missed delivery of IPTp  needs to be improved.
Health workers' IPTp delivery was characterized by its focus on the restrictive IPTp guidelines and by low delivery levels between 28-32 weeks of gestation. Timing of IPTp delivery was in accordance with the restrictive IPTp schedule as most IPTp doses were delivered within the specified periods of 20-24 weeks and 28-32 weeks of gestation. Only 11% of the total IPTp doses were administered outside these periods. Awareness of the IPTp strategy was high among the health workers, however, the concurrent existence of different IPTp guidelines led to some confusion about when and how many IPTp doses to administer.
Consistent with findings from several other studies [26, 27, 29, 31, 48], most women in the study got a first dose of IPTp, but many did not receive a complete course of two IPTp doses. HMIS data collected in the study area in 2008 provided a similar picture: less than half of the women who were given a first IPTp dose, also received a second one. Health workers' low performance provides one possible explanation [50, 51]: Observations of ANC consultations in the context of a study on quality of care showed that return visits were usually of very short duration and reduced to the most basic examinations such as abdominal palpation and the measurement of blood pressure and weight (Gross et al., submitted to BMC Pregnancy and Childbirth). In this context, IPTp as well as the delivery of other drugs and lab examinations might easily be skipped or forgotten. This pattern has also been reported in a study from Malawi . Given the shortages of SP often observed in Tanzania [26, 30, 52] health workers' rationing of SP, especially of the second IPTp dose, might represent another possible explanation. Information on SP stock-outs was not collected at the time of the study as it would only have served to explain the missed delivery of the most recent IPTp doses. Monitoring SP stocks at the health facilities through the collection of end of month drug stock data was not possible due to low quality of record keeping by the facility staff. However, a quality of care study conducted in October 2008 in the same area showed that all the selected health facility except one had SP available in the three preceding months (unpublished data, ACCESS Programme). As this study leaves open questions on why health workers delivered the second IPTp dose significantly less well than the first one, it calls for future studies on health workers IPTp delivery practices.
IPTp delivery practices at ANC clinics in the study area and other regions of Tanzania differ critically from the simplified WHO recommendations to distribute SP to all pregnant women twice after quickening and one month apart [26, 30]. The government's failure to implement the simplified IPTp schedule caused on the one hand confusion among health workers due to the concurrent existence of different and contradictory IPTp policies. On the other hand it represents a missed opportunity for high coverage levels of this important intervention. Calculations based on the simplified guidelines suggest that IPTp coverage could potentially be increased by up to 20 percentage points if IPTp delivery were no longer limited to the narrow gestational range of 20-24 and 28-32 weeks. Better outcomes are also to be expected as health workers' difficulties to assess gestational age is taken into account .
The Ministry of Health should therefore overcome inconsistent IPTp messages and advocate one clear IPTp recommendation. The study showed that implementing the simplified IPTp policy recommended by WHO has the potential to reach more pregnant women with the important intervention of IPTp. The fact that the simplified IPTp guidelines are already integrated in the Focused Antenatal Care guidelines highlights the need for training health workers on the new policy and disseminating the information to the periphery.
Not only coverage but also the number of administered doses could be easily increased through the implementation of the simplified guidelines. WHO currently recommends a three dose regimen for areas in which antenatal HIV prevalence exceeds 10% and where HIV-testing is not available . Trials from Kenya, Malawi and Zambia showed that receiving at least three courses of IPTp was associated with a better protective outcome among HIV-positive pregnant women [15, 54–56]. Tanzania so far continues with a two-dose regimen due to its relatively low HIV prevalence rates. However, if levels of parasite resistance continue to increase, alternative drugs for IPTp need to be urgently found . Any replacement drug to SP will most likely require a more complicated drug regimen. Thus, achieving high coverage levels will become even more challenging.
Collecting data through exit interviews at health facilities imposed two main limitations on the study: Firstly, information on women's ANC attendance behaviour was incomplete. Most women were at the beginning or at the middle of their pregnancy and data were usually available for less than four ANC visits. Moreover, although data collection took place over several months and at different ANC clinics it might not be representative for other places and periods of the year. Secondly, the conduct of exit interviews may have resulted in improved health worker performance. Including ANC card data on services received at earlier visits certainly lessened this type of bias. In reality, having a record of IPTp receipt is neither a guarantee that the drug was given nor taken - despite the fact that IPTp should be provided under direct observation. However, comparing ANC records with women's self-report revealed no major inconsistencies.
Finally, between the interviews with the health workers and the exit interviews with ANC attendees was a time gap of one year, since the exit interview survey was conducted in order to verify and validate certain results from the health worker interviews. Although health workers did not receive any training on IPTp delivery in the meantime, the time order of the studies did not allow to investigate health workers' reasons for the lower level of second dose IPTp delivery.