Poor coverage and quality for poor women: Inequalities in quality antenatal care in nine East African countries

Abstract

The use of quality antenatal care (ANC) improves maternal and newborn health outcomes. Ensuring equity in access to quality maternal health services is a priority agenda in low- and middle-income countries. This study aimed to assess inequalities in the use of quality ANC in nine East African countries using the most recent Demographic and Health Surveys. We used two outcome variables to examine ANC service adequacy: four or more ANC contacts and quality ANC. We defined quality ANC as having six of the recommended ANC components during follow-up: blood pressure measurement, urine sample test, blood sample test, provision of iron supplements, drug for intestinal parasite and tetanus toxoid injections. We used the concentration index (CCI) to examine inequalities within and across countries. We fitted a multilevel regression model to assess the predictors of inequalities in the contact and content of ANC. This study included 87 068 women; among those 54.4% (n = 47 387) had four or more ANC contacts, but only 21% (n = 15 759) reported receiving all six services. The coverage of four or more ANC and receipt of all six services was pro-rich within and across all countries. The highest inequality in four or more ANC contacts was in Ethiopia with a CCI of 0.209, while women in Burundi had the highest inequality in coverage of all six services (CCI: 0.318). Higher education levels and media exposure were predictors of service uptake, while women who had unintended pregnancies were less likely to make four or more ANC contacts and receive six services. Interventions to improve access to quality ANC require rethinking the service delivery mechanisms in all countries. Moreover, ensuring equity in access to quality ANC requires tailoring service delivery modalities to address the social determinants of service uptake.

Socioeconomic factors, inequalities, quality of care, antenatal care, East African countries

KEY MESSAGES

• This study systematically examined inequalities in the use of quality antenatal care (ANC) in Burundi, Ethiopia, Kenya, Malawi, Rwanda, Tanzania, Uganda, Zambia and Zimbabwe.

• Most women received low-quality ANC across all countries and socioeconomic levels, even when they managed to have adequate contact with providers.

• The coverage of four or more ANC contacts and receipt of all six services were pro-rich within and across all countries. Those women who received quality care had primary or higher education, reside in urban areas or had plans to have a child.

• Across all study countries, there was inconsistent delivery of ANC services that are proven to enhance positive pregnancy experience, which implies the need to prioritize and ensure the delivery of all ANC components to improve the maternal and health outcomes.

Introduction

Reducing the high rates of maternal and neonatal mortality has been a persistent challenge for health systems in low- and middle-income countries (LMICs) (Kruk et al., 2018). In 2015, an estimated 303 000 women died from causes related to pregnancy and childbirth worldwide, with more than two-thirds of these deaths occurring in sub-Saharan Africa (WHO, 2015). The number of deaths from causes related to pregnancy and childbirth in East Africa was 7050 in 2015 and ranged from 224 per 100 000 live births in Zambia to 789 per 100 000 live births in South Sudan (WHO, 2015). Ensuring access to quality health care is a key strategy to reduce the high rates of maternal mortality (Kruk et al., 2018). In many LMICs, encouraging progress has been made in improving access to health services (Victora et al., 2016). However, the quality of care delivered across different countries remains low and hinders progress in improving health outcomes (Arsenault et al., 2018).

Antenatal care (ANC) is a proven intervention for reducing complications during pregnancy, childbirth and the postpartum period (Campbell and Graham, 2006; Bhutta et al., 2014). ANC provides a unique opportunity to advise, educate and reach women with interventions that can be crucial for maternal and newborn health (WHO, 2016b). Moreover, receiving quality ANC is associated with higher retention in the continuum of care for maternal, newborn and child health (Kerber et al., 2007).

Three indicators are commonly used to measure the quality of ANC: number of ANC contacts, timing of the first ANC visit (initiation of ANC before 12 weeks) and provision of all recommended ANC services (Heredia-Pi et al., 2016; WHO, 2016b). The WHO recommendations for quality ANC include: the provision of caring and respectful evidence-based care that involves examinations (e.g. history taking and physical assessments) and diagnostic tests (e.g. blood sample test or urine sample test); appropriate preventive and curative treatments (tetanus vaccinations or iron supplementation); and provision of proper counselling and education (e.g. on healthy eating or danger signs of complications) (WHO, 2016b). However, ANC contact is commonly used as the principal monitoring and evaluation indicator to measure the provision of ANC (Carvajal-Aguirre et al., 2017; Benova et al., 2018). This has drawn the attention of healthcare managers and providers to focus on the number of contacts rather than the content and process of care. However, failure to ensure effective coverage (contact with complete delivery of all recommended services) leads to ineffective care resulting in a missed opportunity to achieve improved health outcomes (Carvajal-Aguirre et al., 2017).

The urgent need to tackle the causes of maternal and child morbidity and mortality is a target of the Sustainable Development Goals (SDGs). Specifically, SDG target 3.8 aims at ‘achieving universal health coverage, including financial risk protection, access to quality essential healthcare services and access to safe, effective, quality and affordable essential medicines and vaccines for all’ (UN, 2016). Exploring variations in the quality of services provided across socioeconomic levels and examining the gap between those who do or do not have access to quality services are the important step towards devising and focusing interventions to ensure equitable access to quality ANC.

The majority of studies in this area have addressed coverage indicators, but most do not reflect on the process and content of ANC (Zere and McIntyre, 2003; Boerma et al., 2008; Zere et al., 2010; Gebre et al., 2018; Mwase et al., 2018). Many of these studies reported the presence of socioeconomic inequalities in access to ANC and indicated that poorer women are less likely to use these health services in LMICs. Arsenault and colleagues reported that 89.7% of women in 91 LMICs had at least one ANC visit with a skilled provider. However, 33% of the women who attended ANC did not receive three important services of ANC including blood and urine sample tests and blood pressure measurement at any point during their ANC follow-ups. The richest women were four times more likely to get their blood pressure checked, and urine and blood tested than the poorest women during ANC contacts (Arsenault et al., 2018).

In the present study, we aimed to comprehensively assess inequalities in the use of quality ANC by examining the adequacy of both contact and content of ANC. First, we examined socioeconomic inequalities in ANC contact and content across all countries. Then we identified predictors of inequalities in four or more ANC contacts and receipts of six services separately.

Methods

Data

We used the Demographic and Health Surveys (DHS) of nine East African countries collected between 2013 and 2020. The DHS are nationally representative household surveys with large sample sizes and high response rates (the lowest was 98%). The DHS use multistage sampling procedure in all countries. Standardized questionnaires across time and countries are used to ensure data collected from the survey are comparable. Sampling design and methods have been described elsewhere (Rutstein and Rojas, 2006). We used the DHS from Burundi 2016, Ethiopia 2016, Kenya 2014, Malawi 2016, Rwanda 2014/15, Tanzania 2015, Uganda 2016, Zambia 2018 and Zimbabwe 2015. The study population includes all women of reproductive age (15 to 49 years) who had at least one live birth during the 5 years preceding the respective surveys.

Measures

We assessed two dimensions of ANC service adequacy: four or more ANC contacts and the services provided during contacts (processes and procedures of care provided). Based on the World Health Organization’s (WHO) recommendations (WHO, 2016b) on ANC for a positive pregnancy experience, we examined the provision of recommended components of care which are potential indicators of ANC quality.

Women who attended ANC were asked whether they received specific services during contacts. We identified 13 ANC services: weight and height measurement, blood pressure measurement, urine and blood sample tests, iron supplements, malaria prophylaxis, drugs for intestinal parasites, counselling on signs of complications, tetanus vaccination and counselling on where to go in case of complications. For this analysis, we included six of 13 ANC services that were reported and consistently available across all the DHS. These were blood pressure measurement, urine sample test, blood sample test, tetanus protection at birth, iron supplements and drug for intestinal parasites. We defined quality of ANC as a binary outcome where women who received all six services were labelled as having quality ANC.

Independent variables

We used the WHO Commission on Social Determinants of Health framework (Solar and Irwin, 2010) to assess the potential determinants of inequalities in four or more ANC contacts and all six services received separately. We considered indicators that define women’s socioeconomic position including household wealth index, educational status and occupational status. In the DHS datasets, the wealth index is a composite variable created using principal component analysis and measures the woman’s household living standards. It is constructed by collecting and analysing information on the ownership of selected household assets such as radio, television, refrigerator and vehicle; materials used for housing construction; and access to sanitation facilities and water. We ranked households into quintiles from the poorest (Q1) to richest (Q5) depending on their level of wealth.

Educational status of the women is also an important indicator for socioeconomic position. We used four categories of women’s education level (no education, primary, secondary or higher). We merged secondary and higher education levels because women with post-secondary education were few (only 4.3% of women had post-secondary education). For the purpose of this analysis, we grouped related occupations together and formed four categories of women’s occupation (not currently working, skilled and professional, agricultural and sales and services). Exposure to media was also included as one of the covariates in this study. Frequency of listening to radio and watching television (TV) were both categorized as not at all, less than once a week and once a week or more.

Individual level variables include women’s age and use of healthcare services. For use of health services, we included pattern of contraceptive method use at the time of the survey and timing of first ANC. We defined the early initiation of ANC as starting ANC before 12 weeks of gestation, while we considered women who were unable to attend ANC or initiated later than 12 weeks of gestation to have delayed ANC initiation. We used timing of the first ANC visit as a determinant because it affects the number of components and the content of care that women receive. A woman was also asked whether she wanted to be pregnant with her last child (wanted then or planned, wanted later or mistimed or unwanted), which we also included as a covariate.

Statistical analysis

We performed separate analyses for: (1) quantifying socioeconomic inequalities in coverage of (i) four or more ANC contacts and (ii) receipt of all six services and (2) a multilevel logistic regression analysis to determine factors driving inequality were assessed separately for four or more ANC and the six services women received during follow-ups. We adjusted for sampling design (stratification and clustering) and sampling weights in analyses.

Measuring inequalities

Socioeconomic inequalities in the coverage of (1) four or more ANC contacts and (2) distribution of the six services were estimated using the concentration curve and concentration index (CCI) (O'Donnell et al., 2007). The concentration curve is a plot of the cumulative percentage of the population, ranked by a wealth index from the poorest to the richest on the x-axis against a cumulative percentage of the health variable under study (e.g. four or more ANC contacts) on the y-axis (O'Donnell et al., 2007). If all women had an equal proportion of four or more ANC contacts irrespective of their socioeconomic status, then the curve would coincide with the 45° line, which indicates the presence of perfect equality in the use of four or more ANC contacts. If the concentration curve falls below the 45° line of equality, it indicates that the use of four or more ANC contacts is more concentrated among the rich. The opposite is true if the curve falls above the line of equality (O'Donnell et al., 2007).

The CCI is two times the area between the line of equality and the concentration curve. The index takes a value between −1 and +1; an index of 0 indicates the presence of equality in the use of the health variable (e.g. four or more ANC contacts) (O'Donnell et al., 2007). If socioeconomic inequalities exist, there are two main forms: the first and the most common is when there is uneven concentration of favourable health variable among the rich; in this case, the concertation index takes on a positive value. The second is negative value CCI, which implies high concentration of health variable among the poor (commonly mortality and morbidity).

Multilevel analysis

The concentration curve and CCI can only quantify the level of inequalities related to wealth in the use of health services. However, wealth is not the only factor that determines the use of health services. In our study, we assessed individual and community level factors that determine coverage of four or more ANC contacts and receipts of the six services during contacts. We specified a 3-level model: level 1 variables included women and household factors, at level 2 we adjusted for clustering and at level 3 we adjusted for country (Figure 1). Results are presented with adjusted odds ratios (ORs) and statistical significance was declared when the P-value was <0.05. Analyses were conducted using Stata 14.2 and IBM Statistical Package for Social Sciences (SPSS, Chicago, IL, USA) version 25.0.

Figure 1

Flow chart describing included samples across nine East Africa countries

Results

We used information on 87 068 women surveyed from nine East African countries. The number of women ranged from 4988 in Zimbabwe to 14 429 in Kenya. Three in 10 women were between 15 and 24 years of age at the time of survey, 53.1% had primary level of education, 41.9% were engaged in agriculture and 74.8% lived in rural areas (Table 1).

Table 1

Factors associated with four or more ANC contacts

Study variables	Total		Women who received four or more ANC contacts, % (95% CI)	Unadjusted OR (95% CI)	P	Adjusted OR (95% CI)	P
	Number	%
Age-group (years)					<0.001		0.141
15–24	26 886	30.9	54.4 (53.4, 55.3)	1.00 (Referent category)		1.00 (Referent category)
25–29	22 373	25.7	55.6 (54.6, 56.7)	1.09 (1.02, 1.16)		0.98 (0.93, 1.05)
30–34	17 836	20.5	55.3 (54.2, 56.4)	1.08 (0.98, 1.19)		1.05 (0.94, 1.17)
35–49	19 974	22.9	52.4 (51.2, 53.5)	0.98 (0.85, 1.14)		1.07 (0.92, 1.24)
Wanted to be pregnant with the last live birth					<0.001		<0.001
Planned	48 936	61.5	56.5 (55.6, 57.4)	1.00 (Referent category)		1.00 (Referent category)
Mistimed	22 986	28.9	51.7 (50.7, 52.7)	0.73 (0.68, 0.78)		0.79 (0.75, 0.84)
Unwanted	7597	9.6	46.7 (45.0, 48.3)	0.65 (0.56, 0.76)		0.74 (0.64, 0.84)
First ANC visit <12 weeks of gestational age
No	59 980	68.9	42.7 (41.9, 43.5)	1.00 (Referent category)		1.00 (Referent category)
Yes	27 089	31.1	80.3 (79.6, 81.0)	7.53 (5.64, 10.05)		6.95 (5.50, 8.79)
Highest educational level					<0.001		<0.001
No education	16 693	19.2	40.4 (38.9, 41.8)	1.00 (Referent category)		1.00 (Referent category)
Primary	46 225	53.1	53.4 (52.7, 54.2)	1.31 (1.15, 1.49)		1.19 (1.05, 1.34)
Secondary or higher	24 150	27.7	66.0 (65.1, 67.0)	1.84 (1.38, 2.44)		1.37 (1.19, 1.57)
Pattern of use of contraceptive method at the time of survey					<0.001		<0.001
Never used	21 433	24.6	59.2 (58.4, 60.0)	1.00 (Referent category)		1.00 (Referent category)
Currently using	42 318	48.6	56.2 (54.9, 57.6)	1.54 (1.32, 1.79)		1.33 (1.19, 1.48)
Used since last birth	12 728	14.6	52.7 (51.3, 54.0)	1.34 (1.18, 1.54)		1.29 (1.18, 1.41)
Used before last birth	10 590	12.2	44.8 (43.5, 46.1)	1.21 (1.01, 1.45)		1.21 (1.04, 1.41)
Frequency of listening to radio					<0.001		0.001
Not at all	36 017	41.4	49.8 (48.7, 50.8)	1.00 (Referent category)		1.00 (Referent category)
Less than once a week	15 241	17.5	53.9 (52.7, 55.1)	1.17 (1.05, 1.30)		1.05 (0.96, 1.14)
At least once a week	35 793	41.1	59.4 (58.6, 60.1)	1.39 (1.27, 1.52)		1.14 (1.06, 1.22)
Frequency of watching television					0.002		0.003
Not at all	60 509	69.5	51.0 (50.2, 51.8)	1.00 (Referent category)		1.00 (Referent category)
Less than once a week	9631	11.1	54.0 (52.5, 55.4)	1.14 (1.03, 1.27)		0.96 (0.90, 1.02)
At least once a week	16 908	19.4	67.0 (65.7, 68.2)	1.61 (1.22, 2.12)		1.09 (1.00, 1.18)
Woman’s occupation					<0.001		<0.001
Not currently working	22 688	28.6	52.8 (51.4, 54.1)	1.00 (Referent category)		1.00 (Referent category)
Skilled and professionals	6151	7.8	65.2 (63.4, 66.9)	1.61 (1.32, 1.96)		1.26 (1.12, 1.41)
Sales and services	17 196	21.7	59.2 (57.9, 60.5)	1.20 (1.12, 1.29)		1.13 (1.04, 1.24)
Agriculture	33 262	41.9	50.4 (49.5, 51.3)	1.05 (0.94, 1.17)		1.13 (1.03, 1.25)
Household wealth index in quintiles					<0.001		<0.001
Poorest	19 265	22.1	48.6 (47.3, 49.9)	1.00 (Referent category)		1.00 (Referent category)
Poorer	18 035	20.7	51.1 (49.9, 52.3)	1.08 (0.99, 1.17)		1.06 (1.00, 1.11)
Middle	16 689	19.2	52.9 (51.7, 54.1)	1.26 (1.06, 1.51)		1.07 (1.01, 1.14)
Richer	16 697	19.2	55.6 (54.3, 56.9)	1.48 (1.14, 1.93)		1.09 (0.96, 1.24)
Richest	16 383	18.8	65.3 (64.0, 66.6)	2.08 (1.44, 2.99)		1.30 (1.10, 1.54)
Distance to health facility							0.030
Not a big problem	47 643	59.9	57.0 (56.2, 57.7)	1.00 (Referent category)		1.00 (Referent category)
Big problem	31 882	40.1	50.0 (48.8, 51.1)	0.88 (0.83, 0.93)		0.95 (0.91, 0.99)
Place of residence					<0.001		0.056
Urban	21 983	25.2	63.6 (62.4, 64.8)	1.00 (Referent category)		1.00 (Referent category)
Rural	65 086	74.8	51.3 (50.5, 52.1)	0.65 (0.48, 0.87)		0.91 (0.83, 1.00)

The bold emphasis indicates variables that are significantly associated with the outcome variable (P-value < 0.05).

Coverage of ANC contact and content

Across all countries, 94.8% (95% CI: 94.2–95.3%) of women received one or more ANC visits, but only 31.1% (95% CI: 30.5–31.7%) of women had a first visit before 12 weeks of gestation. Coverage of four or more ANC contacts across all countries was 54.4% (95% CI: 53.7–55.1%).

Coverage of four or more ANC contacts and the six services varied across countries (Figure 2). Women in Zimbabwe had the highest four or more ANC contacts 75.7% (95% CI: 73.8–77.6%), while Ethiopian women had the lowest 31.8% (95% CI: 29.2–34.5%). The gap between contact and content of care was large across all countries. More than half of women (54.4% 95% CI: 53.7–55.1%) had four or more ANC contacts, but only 21% (95% CI: 20.2–21.7%) of women who had attended ANC received all six services. The largest gap was in Zimbabwe, where 75.7% (95% CI: 73.8–77.6%) of women had four or more ANC contacts, but only 1.7% (95% CI: 1.3–2.2%) received all six services. Women in Zambia were more likely to receive all six services 41% (95% CI: 39–43.1%) compared to other countries in our study.

Figure 2

Coverage of four or more ANC contacts and the six services across nine East Africa countries

The majority 91.7% (95% CI: 91.3–92.2%) of women reported having a blood sample test, ranging from 72.5% in Ethiopia to 96.7% in Rwanda. The urine sample test had the overall lowest coverage among the six services across all countries 52.8% (95% CI: 51.9–53.7%), ranging from 27% in Burundi to 88.8% in Kenya. We found extremely low coverage of drug for intestinal parasite in Zimbabwe 3.6% (3.0–4.3%) and Ethiopia 7.6% (95% CI: 6.5–9.0%). In Malawi, Rwanda, Zimbabwe, Uganda, Tanzania and Zambia, coverage of iron supplement was >80% (Figure 3).

Figure 3

Coverage of the six ANC services across nine East Africa countries

As women’s education status and wealth increased, coverage of each ANC services also increased (Figure 4). Women from urban areas were more likely to have better coverage of each ANC service compared to women from rural areas: coverage of all six services among rural women was 17.2% (95% CI: 16.5–17.9%), but coverage was 32.4% (95% CI: 30.7–34.2) among women from urban areas.

Figure 4

Coverage of ANC services by place of residence and educational status across nine East Africa countries

Socioeconomic inequalities

The positive values of concentration indices in Figure 5 show that attendance of four or more ANC contacts and receipt of all six services was disproportionately concentrated among wealthier women across all countries (Figure 5). The inequality in four or more ANC contacts was statistically significant in the majority of the countries, but there were no differences in Burundi, Rwanda and Zambia. The highest CCI in having four or more ANC contacts was registered in Ethiopia (0.209) and lowest in Zambia (Figure 6). Countries with higher four or more ANC contacts had lower socioeconomic inequalities in four or more ANC contacts, e.g. in Zimbabwe where the coverage was highest (75.7%), the inequality was lower (CCI: 0.029). We observed similar patterns—higher coverage and lower inequalities—in Uganda and Zambia.

Figure 5

Socioeconomic inequalities in the coverage of ANC 4+ contacts and the six services within and across nine East African countries

Figure 6

Concentration curves for four or more ANC contacts across nine East Africa countries. Note: Ethiopia had the highest inequality in four or more ANC contacts, while Zambia showed no inequality in four or more ANC contacts. The concentration curve for four or more ANC contacts across all countries show that the coverage is pro‐rich.

The inequality in the distribution of six services was highest in Burundi favouring wealthier women (CCI: 0.318) (Figure 7). There were no socioeconomic differences in receipts of all six services in Malawi and Zimbabwe. Within and across all countries, we found low coverage and high inequality in the coverage of the six services of ANC (Figure 5). Among the six services of ANC analysed in our study, urine sample test had the highest inequalities (CCI: 0.121).

Figure 7

Concentration curves for recipients of all six services across nine East Africa countries. Note: Burundi had the highest inequality in receipt of all six services, while Malawi showed no inequality. The concentration curve for receipt of all six services across all countries shows that the coverage is pro‐rich.

The concentration curve for receipt of all six services across all countries shows that the coverage is pro-rich. Inequalities in the coverage of all six services among women living in rural areas were statistically significant, and the degree of inequality was (CCI: 0.042), but there was no statistically significant difference among women living in urban areas. We observed similar patterns across the specific services including drug for intestinal parasite, urine sample test, blood sample test and blood pressure measurement services (Figure 8).

Figure 8

Socioeconomic inequalities in the coverage of the six services by area of residence across nine East African countries.

Table 1 shows the multivariable models for the associations between the use of four or more ANC contacts and women’s sociodemographic factors. Timing of first ANC visit, educational status, experiences of contraceptive use, exposure to media (radio or TV), occupation of women and wealth of women’s household were positively associated with the use of four or more ANC contacts. The odds of four or more ANC contacts attendance increased with increasing level of education: women with secondary or higher education had higher odds (AOR 1.37, 95% CI: 1.19, 1.57) compared to women who had no education. With increasing wealth, status of women’s household, the odds of four or more ANC contacts also increased as the richest fifth of women had the highest odds of attendance (AOR 1.30, 95% CI: 1.10, 1.54), followed by those in third quintile (Q3) (1.07, 95% CI: 1.01, 1.14) and Q2 (AOR 1.06, 95% CI: 1.00, 1.11) than women from the poorest household.

We found two factors that lowered the odds of four or more ANC contacts: unintended pregnancies and distance to health facility. Women with unintended pregnancy had lower odds of four or more ANC contacts: women who had mistimed pregnancy had 21% (AOR 0.79, 95% CI: 0.75, 0.84) lower odds of attendance, while those with unwanted pregnancy had 26% (AOR 0.74, 95% CI: 0.64, 0.84) lower odds of attending four or more ANC contacts compared to women who wanted to be pregnant with the last live birth.

Women who reported distance to health facility as a big problem had 5% lower odds of attending four or more ANC contact (AOR 0.95, 95% CI: 0.91, 0.99) compared to women who did not consider distance to health facility as a big problem.

In the multivariable models constructed to assess associations between receiving all six services and women’s sociodemographic factors, the following factors were positively associated: women’s age, timing and number of ANC contact, media exposure (radio or TV) and primary or above levels of education and occupation of women (Table 2).

Table 2

Factors associated with receiving all six services of ANC

Study variables	Total		Women who received all six services of ANC, % (95% CI)	Unadjusted OR (95% CI)	P	Adjusted OR (95% CI)	P
	Number	%
Age -group (years)					0.001		<0.001
15–24	25 812	31.3	19.5 (18.6, 20.4)	1.00 (Reference category)		1.00 (Reference category)
25–29	21 238	25.7	21.8 (20.7, 22.9)	1.30 (1.16, 1.45)		1.25 (1.12, 1.39)
30–34	16 901	20.5	22.1 (21.0, 23.2)	1.35 (1.09, 1.68)		1.31 (1.08, 1.61)
35–49	18 551	22.5	21.1 (20.0, 22.2)	1.23 (0.97, 1.55)		1.28 (1.02, 1.59)
Wanted pregnancy at the time					<0.001		0.009
Wanted at the time	46 256	61.4	21.5 (20.6, 22.4)	1.00 (Reference category)		1.00 (Reference category)
Mistimed	22 024	29.3	21.2 (20.3, 22.1)	0.85 (0.74, 0.97)		0.92 (0.83, 1.01)
Unwanted	7001	9.3	16.9 (15.6, 18.3)	0.83 (0.74, 0.93)		0.87 (0.79, 0.95)
Timing of ANC					<0.001		<0.001
Initiated late	55 413	67.2	19.4 (18.5, 20.2)	1.00 (Reference category)		1.00 (Reference category)
Initiated early	27 089	32.8	24.1 (23.1, 25.0)	1.55 (1.36, 1.77)		1.29 (1.21, 1.38)
Number of ANC contacts					<0.001		<0.001
Less than four ANC contacts	35 115	42.6	17.0 (16.2, 17.8)	1.00 (Reference category)		1.00 (Reference category)
Four or more ANC contacts	47 387	57.4	23.9 (23.1, 24.9)	1.56 (1.34, 1.83)		1.37 (1.20, 1.57)
Frequency of listening to radio					<0.001		0.004
Not at all	32 780	39.7	18.0 (17.0, 19.0)	1.00 (Reference category)		1.00 (Reference category)
Less than once a week	14 748	17.9	21.7 (20.5, 22.9)	1.27 (1.09, 1.47)		1.15 (1.00, 1.31)
Once a week or more	34 956	42.4	23.7 (22.9, 24.6)	1.35 (1.21, 1.50)		1.14 (1.05, 1.23)
Frequency of watching television					<0.001		<0.001
Not at all	56 559	68.6	17.3 (16.6, 18.1)	1.00 (Reference category)		1.00 (Reference category)
Less than once a week	9291	11.3	24.6 (23.2, 26.0)	1.38 (1.24, 1.54)		1.20 (1.07, 1.34)
Once a week or more	16 634	20.2	32.4 (30.8, 34.1)	1.52 (1.12, 2.05)		1.07 (0.89, 1.29)
Highest educational level					<0.001		0.007
No education	14 027	17	11.9 (11.1, 12.8)	1.00 (Reference category)		1.00 (Reference category)
Primary	44 692	54.2	20.7 (19.9 21.4)	1.27 (1.17, 1.39)		1.21 (1.12, 1.31)
Secondary or higher	23 783	28.8	27.3 (26.0, 28.7)	1.57 (1.22, 2.02)		1.28 (1.12, 1.47)
Women’s occupation					<0.001		<0.001
Not currently working	20 512	27.3	22.3 (21.1, 23.6)	1.00 (Reference category)		1.00 (Reference category)
Skilled and professionals	6000	8	27.6 (25.9, 29.3)	1.52 (1.32, 1.74)		1.31 (1.16, 1.48)
Sales and services	16 468	21.9	27.0 (25.7, 28.5)	1.20 (1.03, 1.38)		1.11 (0.97, 1.26)
Agriculture	32 103	42.8	15.8 (15.1, 16.5)	0.85 (0.74, 0.98)		0.92 (0.85, 0.99)
Household wealth					<0.001		0.048
Poorest	17 704	21.5	15.8 (14.9, 16.8)	1.00 (Reference category)		1.00 (Reference category)
Poorer	16 954	20.6	17.1 (16.2, 18.1)	1.08 (0.99, 1.17)		1.03 (0.97, 1.09)
Middle	15 827	19.2	19.8 (18.6, 21.1)	1.26 (1.06, 1.51)		1.13 (0.99, 1.28)
Richer	15 971	19.4	23.4 (22.0, 24.9)	1.48 (1.14, 1.93)		1.16 (1.00, 1.35)
Richest	16 046	19.4	29.6 (28.0, 31.3)	2.08 (1.44, 2.99)		1.26 (1.01, 1.59)
Distance to health facility					<0.001		0.004
Not a big problem	46 113	61.3	23.4 (22.4, 24.4)	1.00 (Reference category)		1.00 (Reference category)
Big problem	29 167	38.7	17.2 (16.4, 18.0)	0.82 (0.76, 0.88)		0.92 (0.87, 0.97)
Place of residence					<0.001		0.002
Urban	21 585	26.2	32.4 (30.7, 34.2)	1.00 (Reference category)		1.00 (Reference category)
Rural	60 917	73.8	17.2 (16.5, 17.9)	0.51 (0.39, 0.67)		0.74 (0.61, 0.90)

The bold emphasis indicates variables that are significantly associated with the outcome variable (P-value < 0.05).

With increasing levels of women’s education, the odds of receiving the six services also increased, as the odds of women who had secondary or more education levels were higher (AOR 1.28, 95% CI: 1.12, 1.47) compared to women who had no education. Women who had their first ANC contact before 12 weeks of gestation had higher (AOR 1.29, 95% CI: 1.21, 1.38) odds of receiving all six services compared to women who had no or late ANC contact. Women with four or more ANC contacts had higher odds of receiving all six services (AOR 1.37, 95% CI: 1.20, 1.57) compared to women who had three or lower ANC contacts. The odds of receiving all six care components increased with increasing access to radio: women who listen to radio once a week or more had higher (AOR 1.14, 95% CI: 1.05, 1.23) odds of receiving all six services than women who did not listen to radio at all.

Women from rural areas had 26% lower odds of receiving all six services (AOR 0.74, 95% CI: 0.61, 0.90) compared to women from urban areas. Compared to women who wanted to be pregnant with their last live birth, women who had unwanted pregnancy had 13% (AOR 0.87, 95% CI: 0.79, 0.95) lower odds to receive all six services. Women who reported distance to health facility as a big problem had 8% (AOR 0.92, 95% CI: 0.87, 0.97) lower odds of receiving all six services compared to women who did not consider distance to health facility as a big problem.

Discussion

We assessed inequalities in access to quality ANC services women received in nine East African countries. We systematically examined inequalities in having the recommended number ANC contacts (four or more) and receipt of quality care, which we defined as having six of the recommended components of ANC. The majority (95%) of women had at least one ANC contact (ranging from 62.9% in Ethiopia to 99.3% in Burundi), but only 31.1% had their first visit within the first 12 weeks of gestation. More than half of women across all countries had four or more ANC contacts; this ranged from 31.8% in Ethiopia to 75.7% in Zimbabwe. Many women did not receive all six components, although they had an adequate number of ANC visits. About 21% of women across all countries received all six components of care, ranging from 1.7% in Zimbabwe to 41% in Zambia. This implies that across all the countries and socioeconomic levels, the majority of women received low-quality care even when they managed to be in contact with health care providers.

Our finding of a large gap between contact and content of ANC that women received is consistent with similar studies from Nigeria (Fagbamigbe and Idemudia, 2015) and Nepal (Joshi et al., 2014). In such studies, the gap between contact and content of services was partly due to the challenges related to shortages of medical equipment and supplies. Surveys of health facilities in sub-Saharan African countries (Kruk et al., 2016; Leslie et al., 2017) have found shortages of basic supplies and equipment to collect urine and blood specimens and to measure blood pressure. Lack of adequate healthcare providers with the right set of skills, commitment, care and respect for women provides another explanation for the gap in ANC services in those studies (WHO, 2016a; Sumankuuro et al., 2018).

The service gaps in the present study were mainly due to the lower coverage of services that require the availability of medications and functional laboratories. The relevant services were urine sample tests, drugs for intestinal parasites and tetanus injections. We also found that these services were the underlying reasons for inequalities in the content of care received during contacts; in particular, the urine sample test had coverage of 42% among the poorest quintile of women but was 70.7% among the richest quintile. In 2016, WHO introduced a new ANC model that recommends a minimum of eight ANC contacts (WHO, 2016b). However, increasing the number of contacts alone cannot improve health outcomes without improving the quality of care delivered to women.

The limited availability of health resources, coupled with an inefficient use of existing resources, also compromises the quality of health services in many countries (WHO, 2000; 2010). Inadequate financing affects the health system's capacity to provide basic health equipment and supplies to health facilities that are serving communities (WHO, 2006). Due to a lack of adequate laboratory and pharmaceutical supplies, health facilities are often unable to provide good quality care (Leslie et al., 2017). The poor quality of care can also be explained by health systems performance issues that include long queues and waiting times as well as inadequate and sometimes disrespectful treatment women receive from providers (Kruk et al., 2018).

In agreement with previous studies (Amo-Adjei et al., 2018; Arsenault et al., 2018; Benova et al., 2018), our analysis revealed a pro-rich distribution of both contacts and content of ANC received by women across all countries. Inequalities in ANC contacts and receipt of six components of care in Kenya, Uganda, Tanzania and Zambia, favoured wealthier women. In Ethiopia, Malawi and Zimbabwe, four or more ANC contacts were higher among richer women. However, in Rwanda, we noticed relatively low, but evenly distributed coverage in terms of the four or more ANC contacts and all six services received.

There is typically a strong association between socioeconomic status and use of health services (Victora et al., 2010; Bobo et al., 2017; Arsenault et al., 2018). In our study, the socioeconomic well-being of women was associated with an increased number of ANC contacts. While maternal health services are provided free of charge in many developing countries (Arsenault et al., 2018), women may incur both direct (laboratory tests and medications) and indirect costs (e.g. transportation costs), which often limits their uptake of these services. Moreover, women may also have other competing priorities that may affect their use of ANC services. Most poor women travel longer distances and may wait for long periods to receive ANC services and consequently incurring the opportunity costs related to work and family responsibilities. A fee exemption for maternal health services is a vital step forward in reducing financial barrier to accessing care but demand-side barriers of indirect costs such as cost of transportation and missing productive work hours remain critical (Sumankuuro et al., 2018).

We observed that older women were more likely to attend four or more ANC contacts than younger women. A recent systematic review that examined determinants of ANC use (Okedo-Alex et al., 2019) across sub-Saharan Africa has had a similar finding. On the other hand, we found that after adjusting for age and other confounding variables as the number of pregnancies and births increased, women were less likely to attend ANC, which is consistent findings of other similar studies (Joshi et al., 2014; Afulani et al., 2019). The decision to seek ANC for second or later pregnancies can be affected by experiences of previous pregnancy and childbirth. If the previous pregnancy and childbirth were safe and uncomplicated, women might think later pregnancies carry lower risk and, therefore, be less likely to attend ANC follow-ups (Afulani et al., 2019).

We found that planning for pregnancy affects a woman’s decision to seek ANC. Women who had a mistimed or unwanted pregnancy were less likely to attend four or more ANC contacts or receive adequate content of care. Failure to recognize the pregnancy early on is one of the reasons for lack or delayed use of ANC, which could be worsened by unavailability of early pregnancy testing services (Gross et al., 2012; Amo‐Adjei and Anamaale Tuoyire, 2016).

Urine tests are done to screen for possible pregnancy and examine protein and glucose levels to screen for hypertensive disease of pregnancy and diabetes. In this study, coverage of urine sample test was lowest among the six components of ANC we analysed and this can partly be attributed to unavailability of the test at the facilities. It is also possible that the health care providers could have failed to request for it (Arsenault et al., 2018).

Previous studies have highlighted the fact that there are clear connections between ANC, institutional delivery and postnatal care (Engmann et al., 2016; Iqbal et al., 2017; Mohan et al., 2017). More specifically, an earlier study in sub-Saharan Africa has demonstrated that receiving higher quality ANC improves the uptake of skilled birth attendance (Chukwuma et al., 2017). The associations between use of contraceptives and ANC, as revealed in this study, further highlight the need to emphasize continuity of maternal health services.

Education empowers women and reduces gender inequality (Solar and Irwin, 2010). In the present study, improvements in four or more ANC contacts and the six services received were consistent with improvements in women’s education. Education helps to create improved awareness and knowledge about maternal health services, which positively influences women’s desire/willingness to use health services (Say and Raine, 2007). Previous studies have also indicated that educated women are more likely to use ANC and other related maternal health services (Afulani, 2015; Heredia-Pi et al., 2016; Agho et al., 2018).

Access to media can be a powerful tool to reach all women of different socioeconomic status. We found that access to mass media (radio and TV) favourably influences the contact and content of ANC received. Disseminating organized and deliberate messages about the importance of attending maternal health services is crucial to reach not only women, but also their partners and opinion leaders in the community (Zamawe et al., 2016). Disseminating messages that target women’s partner and community leaders can help to create a supportive environment that can favourably influence women to use health services (Zamawe et al., 2016).

Place of residence is recognized as one of the key factors underpinning inequalities in the use of health services (Say and Raine, 2007; Solar and Irwin, 2010). Women from urban areas were more likely to receive the six services compared to women residing in rural areas. This finding is consistent with those of earlier studies (Afulani, 2015; Benova et al., 2018). Various explanations have been suggested for the low-quality ANC coverage in rural areas including less developed health infrastructure and fewer skilled providers (Kruk et al., 2016; Leslie et al., 2017). Similarly, some studies have suggested that women from rural areas who attend urban health facilities are more likely to receive poor quality ANC; which might be because they are probably less educated, so providers think they won't complain about quality (Afulani, 2015; Afulani et al., 2019). Others have pointed out that discriminatory treatment from providers can undermine the quality of ANC women receive (Bowser and Hill, 2010; Bohren et al., 2015). Often, women receive differential treatment as a result of their ethnic background, level of wealth and educational attainment (Bowser and Hill, 2010).

The strengths of our study include the use of nationally representative population-based survey collected from nine east African countries. Second, most studies in this area have focused on wealth related inequalities, while other critical social determinants of health such as education, place of residence, occupation and other sociodemographic factors have not been considered. This study provides a more comprehensive evidence by first assessing inequalities in the use of ANC relative to wealth using concentration curve and index and then looking at other important determinants of inequalities in a multilevel analysis.

Our study has some limitations. First, information on women’s last live birth was collected retrospectively, and the birth may have been up to five years before the survey. Thus, recall and self-reporting bias are limitations, as women may not accurately remember whether they received a service and/or the number of ANC contacts they have made. Second, measuring the quality of care is complex as conceptually quality is a multidimensional construct (Donabedian, 1988). In the current analysis, we were limited to analysing the content of care as a proxy to assess the quality of ANC without considering the process of delivering care due to the lack of variables to reflect on the later. Furthermore, we were unable to include all recommended 13 components of ANC in our analysis since seven of the services were not provided/available consistently in the survey datasets for the included countries. For example, data for nutritional counselling services were only available in Ethiopia 2016 DHS. Weight measurements were only reported in Kenya, Malawi and Uganda, while height measurements were reported in Kenya and Malawi. Moreover, cost of health services was not included in the datasets while it could have proved to be important determinant of inequitable access to quality ANC as we have shown for distance from health facility.

Conclusions

The findings of this study highlight a large gap between the contact and content of ANC. We found that the coverage of quality ANC was low across all countries included in our analysis. Women who received all the six services were largely educated, live in urban areas and had planned to have a child. In effect, this study has shown that there are multiple factors at play that could be responsible for inequalities in access to adequate ANC services. As such, interventions to address current inequalities in access to quality maternal care in sub-Saharan Africa should consider the underlying causes of such inequalities.

Our findings have some policy implications. First, we note that effective coverage is an important way to bridge the quality gap. Effective coverage requires that performance is measured by not only the number of people the health systems is able to reach but also by incorporating indicators to monitor content of care received by the people. The actual delivery of standard components of a health services package to the population should be the ultimate target. Second, while both demand and supply-side issues require attention, health systems need to ensure consistent delivery of all ANC components to enhance quality. Therefore, in addition to promoting more ANC contacts, ensuring consistent delivery of ANC components should be a priority in all countries. Third, efforts at strengthening family planning services are required as women who had unplanned pregnancies received lower-quality ANC. Fourth, our findings also imply that ministries of health in individual countries should at least study the DHS data with the aim of looking for evidence to improve equitable uptake of quality ANC. This helps identify quality gaps that arise from deficiencies in supplies and equipment versus those that may reflect the need for targeted training of staff. Finally, yet importantly, quality improvement efforts should begin in areas with poor quality ANC services and directly consider the needs and experiences of poor and vulnerable populations.

Acknowledgements

Authors are grateful to Measure DHS, ORC Macro, Calverton, MD, USA, for permission to access and Demographic and Health survey data for this analysis.

Availability of data and material

This study was based on an analysis of existing dataset in the DHS repository that are freely available online with all identifier information removed (http://www.dhsprogram.com).

Conflict of interest statement. We declare no confilicts of interest.

Ethical approval. This study used secondary data, which were not subject to ethical approval because it did not involve data concerned with human participants. The DHS surveys are fully available upon request without restriction.