Background
The last few decades have marked substantial successes in increased coverage of essential maternal and perinatal health services in low- and middle-income countries (LMICs). In particular, coverage of antenatal care (ANC) has risen dramatically, as measured by whether women had four or more ANC visits, which alongside skilled birth attendant coverage, is one of the most widely used summary measures of maternal health programme performance [
1,
2]. There are concerns that the ANC 4 + visit indicator has focused on advances in mere contact, rather than the process and content of ANC, obscuring large gaps between coverage of services and the quality of care received [
1,
3]. This coverage-quality gap has been blamed for the persistent burden of maternal and perinatal mortality and morbidity [
4,
5].
Several studies have combined indicators of ANC contact with capture of the care components received in order to measure ‘effective coverage’ for pregnancy care [
1]. These studies mostly rely on household surveys, such as the Demographic and Health Surveys, which use women’s self-reports of care components received during the most recent pregnancy that ended in a live birth. This includes some routine elements that should be done at each ANC visit (e.g., blood pressure (BP) measurement), but women are asked only if they received the component at least once. The studies found that ANC 4 + visits and coverage across selected components of care correlated relatively well, as fewer visits meant fewer opportunities to offer/obtain care components [
1,
3]. But while some LMICs had high coverage and high content of ANC, many did not; perhaps more troubling, some had high coverage but poor content [
1]. In India, for example, the National Family Health Survey 2015–16 found that 51.2% of women had at least four ANC visits [
6]; further analysis, however, revealed that only 23.5% of all women received adequate ANC which was defined as care delivered by skilled health personnel, registration of pregnancy and first ANC visit within the first trimester, 4 + ANC visits and with appropriate content [
7].
New measures are needed to understand the care pregnant women receive [
1,
3,
8,
9]. In 2016, the World Health Organization (WHO) released new ANC guidelines, recommending an increase from four to eight or more ANC visits, emphasising person-centred care and well-being, and recognising the complexities of providing and monitoring quality ANC in diverse health systems [
10]. The WHO conceptual framework for quality ANC highlights the multiple dimensions of quality, including content and women’s experience of care, and various inputs needed to deliver routine ANC, including equipment and competent healthcare providers [
9]. Measures reflecting services received at least once during pregnancy, such as those assessed on household surveys, are limited in assessing whether women were adequately followed throughout their pregnancies [
8]. Among screening components, such as for syphilis or anaemia, there are often no indicators for whether women screening positive received adequate treatment, resulting in data gaps for capturing maternal and foetal assessment and appropriate response [
9]. Further, few studies consider women's experience of care. Examining these different dimensions and inputs is critical to creating a holistic picture of quality of ANC.
Rethinking ANC quality assessment is particularly helpful in settings with high coverage like Telangana, India where nearly all pregnant women access ANC [
11]. In Telangana’s ANC programme, pregnant women are expected to receive frequent ANC visits, including two visits in the first trimester to a sub-centre and one to a primary health centre (PHC) to register the pregnancy, provide an obstetric history and receive preventive and screening interventions (such as haemoglobin and syphilis testing). If no risk factors are identified, then pregnant women should have monthly primary care level facility visits and two visits to a higher-level facility with a gynaecologist in the second and third trimester. Women identified with high-risk pregnancies have monthly primary care level visits alongside multiple visits with a gynaecologist at a higher-level facility. The National Family Health Survey 2015–16 showed that in Telangana, 75.0% of women with a live birth in the previous five years had 4 + ANC visits, and among those who received ANC, reporting of selected components, such as BP measured, was nearly universal [
12]. Yet these coverage measures do not reveal whether pregnant women received the care components correctly, at the right time and frequency, and with an appropriate response.
This paper takes a multi-dimensional approach to examine quality of ANC in Telangana, India based on four different sources of data.
Discussion
We analysed four data sources from Telangana that examined different aspects of quality of ANC, finding some important deficiencies in the quality of care despite high levels of utilisation. Analysis of the NFHS-5 for Telangana showed very high statewide coverage of assessed components, though counselling on pregnancy complications was the least performed component of care. Likewise, HMIS data showed high coverage of ANC visits but significant gaps in screening for syphilis and gestational diabetes. The facility survey in selected districts showed moderately equipped facilities. Some key services such as urine protein testing, which should be monitored regularly throughout pregnancy, and syphilis testing, which should be performed at least once during pregnancy, were unavailable in most of the facilities surveyed. While many ANC services or equipment items were commonly available individually, no facility in our sample offered all 10 components of routine ANC services. In the ANC observations, most women received adequate physical examinations, though some quality issues were noted in performance of BP measurement. However, symptom checking and client education were poorly done.
We found that clinical content of care, in particular maternal and foetal assessments, had high coverage and examinations – where we could evaluate these – were generally performed correctly. Despite high coverage of some important screening assessments (e.g., haemoglobin testing), there were also notable gaps in coverage and service availability (e.g., syphilis testing). This finding was obscured in the NFHS-5 results because women were only asked if they ever had their blood tested, not which specific tests performed. Some assessments that should be performed at every ANC visit, such as BP measurement, had nearly universal coverage in both the ANC observations and in the NFHS-5, which asked only if BP had been measured at least once during the pregnancy. However, other assessments, such as urine testing, had high coverage in the NFHS-5 when asked if urine testing had been performed at least once during the pregnancy, but coverage for urine protein testing, or referral for a test, showed considerable gaps in the ANC observations. This echoes findings from a survey of pregnant women in Kenya which found substantial disparities between receipt of key services at any point in pregnancy and receipt of those services at the recommended frequency [
20].
We found that care often lacked the communication between the healthcare provider and pregnant woman that is important to high-quality, person-centred care [
21]. The NFHS-5 and ANC observation data showed poor provision of information, with little counselling on potential signs of pregnancy complications. ANC observations showed poor psychosocial and emotional support. Few women in our ANC observations were asked about any current physiological symptoms or their mental health – important components of women’s experience of care [
9]. Poor counselling in ANC has been documented in other LMIC settings, with calls for better measuring and improving the quality of information provision in ANC [
22,
23]. The focus on guideline-driven care, particularly with increasing technical content of clinical care [
10,
24] and emphasis on examinations, can negatively impact interpersonal aspects of quality [
21]. Further, busy clinics or those with restricted hours or staff for ANC can often afford little time for meaningful provider-patient interaction [
24].
Our findings demonstrate how data sources build upon or contradict one another to provide a fuller picture of the quality of ANC in Telangana, contributing to a growing body of literature on measurement of ANC quality [
8,
9,
21,
25]. As others have found, components of ANC provision vary widely in quality and taking multiple data sources together can reveal quality gaps. For example, a study in rural Tanzania found that while pregnant women were highly satisfied by their care in exit interviews, data from observations and facility audits found ANC consultations frequently missed important care components, often due to stock-outs of medications and screening tests [
26]. Another study of hospitals in Nepal found poor provision of recommended components during ANC observations; qualitative data from providers and pregnant women echoed these findings, attributing the observed poor performance to insufficient human resources, infrastructure and supplies [
27]. Others have noted opportunities for integrating household survey and facility survey data to estimate composite measures of effective coverage of ANC interventions [
28].
Indicators frequently drive the focus of improvement efforts [
29]. Existing ANC quality measures mainly encompass indicators of content of care and of health system inputs with only a few measures of women's experience of care [
8,
9]. Our ANC observation tool attempted to address this by examining whether women were asked about current pregnancy symptoms, given an explanation about the physical examination or given an opportunity to ask questions, drawing on components of respectful, person-centred ANC [
20]. Given the historic relative emphasis on clinical assessments over counselling in ANC guidelines [
10,
21], it is unsurprising that we found limited provision of information to pregnant women. Incorporating better measures of women’s experience of care will require greater consensus on what matters to women and what can be effectively measured, including through ANC observations, exit interviews with pregnant women, and household surveys [
8,
9,
21].
Improving measurement of the quality of ANC includes opportunities to better assess responsiveness of care. For example, assessment of clinical practice could include whether women were told what their blood and urine samples were for and were given the results of screening tests. This could be assessed by observing ANC consultations or through exit interviews or surveys with pregnant women, although further validation work is needed on whether women can self-report this information. High-quality ANC should be responsive to individual women’s needs; where complications are identified, additional indicators on whether women received an appropriate response or treatment are needed.
Our analysis offers multiple strengths in bringing together four different data sources, but we encountered several limitations. Firstly, our data sources cover different time points, reducing some comparability of findings, particularly from the NFHS-5 five-year recall period. The facility survey and ANC observations were conducted in a small number of facilities in the selected five districts. While the facilities were randomly selected, the inclusion criteria for the sampling frame reflected logistical constraints and may not be representative of all PHCs and sub-centres in the districts.
Results from HMIS were hampered by questions about data quality and whether the available denominator – women registering their pregnancy at a public facility – was the most appropriate one. The counts of women extracted from the annual HMIS report (April 2019-March 2020) reflect imperfect numerators and denominators in a setting where pregnant women access care at many different facilities, including within different districts and within the public and private sector. So for example, while a pregnant woman might register at one public sector facility, and be recorded as receiving haemoglobin tests there, she might also receive multiple haemoglobin tests at different public facilities, leading to an overcounting of haemoglobin testing coverage as we observed. Despite this, the HMIS results yield a useful, though imperfect, picture of variability in service coverage.
The ANC observations offered invaluable insight into quality of care during a single ANC visit, but both data collection and analysis were challenging. We found it difficult to find the right balance between designing a data collection tool which covered all possible components of ANC and designing something which was feasible for fieldworkers to complete during the ANC observation. Pre-testing revealed the data collector would observe ANC consultations and later finish filling in the tool, as it was too challenging to observe and complete the long checklist. This introduced potential for misclassification or recall bias. Healthcare providers may also have improved the quality of care while under observation, though we note that substantial quality gaps remained. Additionally, analysing the observation data required integrating the results from the checklist with additional information about the woman’s stage of pregnancy and previous care received – elements that the tool was not fully designed to address. Each ANC observation was assessed individually by a clinically trained researcher, integrating information available in the woman’s handheld ANC card and whether or not it was the woman’s first ANC visit at that facility (or any facility). This limited the replicability of the analysis, and the amount of time needed to assess each ANC observation meant that this method would be challenging to do at large scale.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.