Quality of antenatal care in Zambia: a national assessment

Every day, approximately 800 women die from preventable causes related to pregnancy and childbirth [1]. In 2008 over 300,000 maternal deaths occurred worldwide and almost all of these in low- and middle-income countries [2, 3]. It is also estimated that every year 3 million newborn babies die within the first month of life [4] and between 2.1 to 3.8 million babies are stillborn [5‐7]. Antenatal care (ANC) is one of the recommended interventions to help reduce these alarming maternal and newborn mortalities [8‐10]. Most low-income countries have adopted a focused ANC strategy promoted by WHO which includes four visits and structured elements within these [11‐14].

Although the percentage of women attending ANC (for at least one visit) generally tends to be satisfactory even in low-income countries, maternal and neonatal mortality remain high. In Zambia, for example, 94% of women attend ANC at least once, while maternal mortality is estimated at 591 maternal deaths per 100,000 live births, and neonatal mortality at 34 neonatal deaths per 1,000 live births [15]. This weak relationship between ANC use and maternal and newborn survival has motivated a recent call to focus on content and quality of care provided rather than mere ANC attendance as we aim at achieving MDG 4 and MDG 5 [16]. Although researchers frequently highlight the importance of quality of maternal care in improving maternal and newborn health [17‐19], the quality of ANC remains insufficiently studied.

Donabedian proposed a framework for assessing quality of care which distinguishes between the attributes of the health care setting (structure), the actual care delivered (process) and the end result of the interaction between an individual and the health care system (outcome) [20‐22]. Maxwell added to this initial concept by arguing that quality of care cannot be measured in a single dimension and suggested six dimensions of quality [23]. These quality of care dimensions have been adopted by various organizations [24, 25].

Tools to capture several dimensions of quality of ANC received by mothers have been developed for high-income settings [26, 27], but are not easily transferable to low-income settings. Studies in low-income countries usually rely on population-based surveys such as the Demographic and Health Survey (DHS) [28], which provide some information on quality of ANC received by mothers, or on community surveys in areas with interventions aiming to improve the quality of maternal health care [29, 30] where the quality of care is not necessarily representative for the country or region as a whole.

Little conceptual work has been done on measurement of ANC quality at health facilities [31], although availability of ANC interventions is a prerequisite for women receiving good quality care. In particular, it is necessary that drugs and equipment are available at the facility, that health workers are present and have the necessary knowledge and skills, and that they actually provide the recommended interventions. In settings where health facilities often lack drugs or skilled personnel, evaluating level of service provision may be more diagnostic in terms of identifying where the problem lies, than collecting information from the population on the care received. So far, hardly any published studies have assessed quality of ANC provided at health facilities. One study in Tanzania used health facility information to compare quality of ANC between public and private providers, but was limited to few facilities in an urban area and thus the results are not representative for the country [31].

Possible reasons for this neglect or inadequate measurement of quality of care, besides the conceptual difficulties in capturing the multiple dimensions of quality, include the weakness of routine health information systems, and the subsequent dependence on household surveys, which lack appropriate information on technical aspects of quality of care, as pointed out by Graham and Varghese [17]. Only few low-income countries have detailed information on the services provided by their health facilities and can thus evaluate quality of care at the point of provision.

In Zambia, both a national Health Facility Census from 2005, and a national household survey (Zambia DHS 2007) are available, presenting a unique opportunity to illustrate how the barriers to the measurement of quality of ANC can be overcome methodologically. We draw upon these datasets to comprehensively assess and describe the quality of ANC in Zambia at that point in time, which can serve as a baseline to show any improvements against. The specific objectives were to: (i) assess various quality of care dimensions for ANC facilities in Zambia; (ii) ascertain how many of Zambia’s ANC facilities met developed criteria for optimum and adequate level of service provision, and (iii) assess and compare level of service provision at health facilities with the quality of ANC received by expectant mothers in Zambia.

Drawing upon two national datasets with detailed provider and user information, the Zambia 2005 HFC and the Zambia 2007 DHS, we demonstrated that although ANC attendance in Zambia was high, insufficient provision of important antenatal screening tests and interventions were limiting the level of ANC delivered, which was reflected in the high proportion of women not receiving good quality ANC. To our knowledge, this is the first study to present data on ANC quality at health facilities on a national scale and the first to develop a tool for classifying the level of ANC service provision in health facilities in low-income settings that includes several quality dimensions.

Rural health centers, the main ANC providers, usually opened at least once a week for ANC, employed at least one skilled health worker and provided at least three ANC functions, however, more than half did not provide any ANC screening test. In contrary, almost all the hospitals had at least three skilled health workers and the majority of them performed all five essential ANC functions as well as three or more screening tests. Hospitals, however, form the minority of ANC providers in Zambia and are used by less than 20% of expectant mothers.

Various important ANC interventions that could contribute to improving maternal and newborn health remain largely unavailable in Zambia. Only 23% of ANC facilities were able to perform a urine protein test, reflected by a similarly low number of mothers reporting to have had their urine checked. Coupled with the fact that 17% of mothers did not have their blood pressure measured, this will make it difficult to detect and manage hypertensive complications of pregnancy before potentially fatal eclamptic fits develop. This defeats the objective of focused ANC, which hinges on early detection and treatment of complications [14].

Besides early detection and treatment of pre-eclampsia, screening for syphilis with treatment of positive pregnant women and screening for HIV infection in order to implement PMTCT are effective interventions that are recommended for low-resource settings [11]. Despite the high HIV prevalence of 16% among women of reproductive age in Zambia [15], only 30% of ANC facilities offered VCT and PMTCT to pregnant women and only half of mothers reported to have been offered this service. It is evident that the new model integrating VCT and PMTCT into ANC had not been fully implemented by the time of the 2005 Zambia HFC. Syphilis prevalence among women aged 15–49 in Zambia was 4% at the time of the DHS survey [15], and it is estimated that 69% of pregnant infected women will have an adverse pregnancy outcome [39]. Improving the coverage of antenatal syphilis screening could reduce the syphilis-attributable incidence of stillbirth and perinatal death by 50% [39]. However, implementation was still a big challenge in Zambia in 2005 as our data showed that only about half of the country’s ANC facilities offered syphilis screening then.

Using information both from ANC providers and ANC users enabled us to get a complementary picture of ANC quality in Zambia. Unfortunately, we did not have information on facility case load in the HFC. Knowing how many women sought ANC in each facility could contribute towards understanding the reasons why most women in Zambia did not receive good quality ANC. For instance, one could establish whether women in an area used certain inadequate facilities at all, and in case women mostly used facilities offering an optimum level of service but still did not receive such care, this would indicate that facilities were not always providing available ANC functions or screening tests. This information would help to differentiate between the problems of availability of services and correct provision of available services, which require different approaches to their solution.

In interpreting our findings, it is worth noting certain limitations of this study. The Zambian HFC, on which our assessment of levels of service provision is based, only provides a snapshot of the health facilities at the time of visit in 2005. The situation now is likely to be different and possibly improved. With assistance from the World Bank, the Japan International Cooperation Agency and the UK Department for International Development, the Zambian Ministry of Health has bought necessary equipment in recent years which will have improved structural quality. A number of new health facilities have been constructed and a direct entry mid-wifery programme has made it possible to produce, recruit, and place midwives at a rapid rate. A results-based financing project targeting maternal and child health is also contributing to improve the situation [40‐44].

Additionally, important information such as opening hours, training of health workers, and client satisfaction were not collected in the HFC and could thus not be included in our quality classification. Proxy indicators such as ANC days per week and the availability of skilled health workers were included in assessing level of provision instead. The assumption that provision of certain interventions implies availability of the necessary equipment and drugs also does not necessarily hold; providers may improvise and provide an intervention in a suboptimal way. There was furthermore no information on whether interventions were given in the correct way or information on provider-client interaction, as the HFC - given its national scope, covering around 1400 facilities - did not allow observation of service provision. For these reasons, it is likely that some facilities have been wrongly classified in terms of level of ANC provision. We unfortunately also did not have any complementary qualitative information, e.g. from key informant interviews which would have helped to interpret the quantitative findings. This is lamentable because perceived quality of care by mothers as well as other qualitative factors have been shown to be critical determinants of ANC uptake [45]. Finally, mothers may not have correctly remembered the details of the ANC they received, causing measurement error in the user information; this was reduced by restriction to the most recent birth in the DHS.

In summary, very few facilities in Zambia provide an optimum level of ANC services, as judged by the criteria we developed using technical process of care attributes, and this is reflected in the less than optimal quality of ANC received by the majority of expectant mothers.

Although 94% of mothers had at least one ANC visit with a skilled health worker and 60% had at least four visits – the indicators tracked by “Countdown to 2015” [16] -, only 29% of mothers in Zambia received good quality ANC and only 8% received good quality ANC and attended in the first trimester (Figure 4). “Effective coverage”, defined as “the proportion of the population who needed a service that received it with sufficient quality to be effective” [46, 47], is thus much lower than suggested by current service coverage indicators. This “quality gap” [48, 49] is an indication that there are still many “missed opportunities” at ANC for delivering effective interventions to improve maternal and newborn health [13, 50].

There have been numerous calls to “improve the content and quality of services offered” at ANC [13, 17, 19, 49, 50] and to develop indicators that capture quality of care [17‐19, 46, 51]. We have shown that it is possible to develop an indicator of “effective ANC coverage”, using information on interventions received from the DHS. Furthermore, we have demonstrated that evaluating level of ANC provision in health facilities can be enlightening and point towards deficiencies that should be addressed. It has been shown that to improve the quality of ANC received by mothers, efforts should focus on both improving the level of services provided at ANC facilities and their accessibility [52]. Especially in settings where services are highly deficient, collecting data upstream at health facilities is advisable. Health Facility Censuses could be adapted for continuous monitoring and could then provide actionable data at the local level at shorter intervals and probably lower cost than population surveys.

This study was granted ethical approval by the London School of Hygiene & Tropical Medicine ethics committee (application number 5172).