Trends in Maternal Health Care Utilization in West Africa: Diffusion or Modernization?

Abstract

Maternal health care remains a major concern in developing countries, but many sub-Saharan countries have achieved substantial reduction in maternal and neonatal mortality rates. This chapter seeks to account for the sources of these reductions, focusing on the role of socioeconomic changes and the health system and using Mali, Burkina Faso, Niger, and Senegal as case studies. The results of our decomposition analysis suggest that, overall, changes in the use of maternal health care reflect behavioral changes associated with improvements in the performance of health systems in these countries. Exceptions were observed for pregnancy and delivery assistance in Mali between 2001 and 2006, and in Burkina Faso between 1999 and 2003. Despite this overall progress, inequalities between socioeconomic groups persist, particularly in Mali and Senegal.

1. Introduction

Maternal and neonatal mortality remain a major concern in developing countries. Sub-Saharan Africa is the region in the world with the highest level of maternal mortality and the lowest rate in the use of maternal health services. Despite numerous policies and programs to improve access to maternal care and reduce maternal deaths, many populations are excluded from modern care, especially the most socioeconomically vulnerable groups [1, 2, 3]. Costs (both direct and indirect) are disincentives for women and their households to use health services during pregnancy and childbirth [4, 5].

Women who belong to a disadvantaged social group [6], living in poor households [7], or have difficulties paying for pregnancy [8] may be less likely to attend antenatal clinics or seek assistance from skilled health personnel in child delivery. As the costs of antenatal care and child delivery-related services are not affordable for all households, women living in poor households are more likely to receive less care during pregnancy and more likely to give birth at home. Thus, exorbitant costs (both direct and indirect) of care services for some poor households partly condition their decisions to use maternal health services. These costs represent factors that may discourage women from going to health facilities for pregnancy monitoring or for assistance at the time of child delivery [4, 5, 9]. In this way, Titaley and his colleagues [10] note that “Cost was one of the main reasons reported by participants in all villages for using traditional birth attendants’ services.” Referring to additional financial barriers induced by the increase in the cost of service provision in rural Zaire, a longitudinal study showed a decrease in the use of health services by almost 40% between 1987 and 1991 [11]. According to these authors, 18 to 32% of this decrease can be explained by the cost of using these services. Therefore, a reduction in these costs could lead to an increase in the use of health services. Ridde and colleagues [3] point to an increase in the use of facility-based delivery following cost suppression in some African countries.

However, the impact of cost reduction on the increase in the use of health services does not seem to be verified in all studies. This is the case, for example, in Kenya and Tanzania, where the quality of services seems to be more important than cost issues, supporting the complexity of the relationship among cost reduction, provider motivation, quality of services, and use of health service during pregnancy and childbirth [4, 12].

Furthermore, women’s education is widely reported in almost all studies on maternal health because of its undeniable effect on health care-seeking behavioral changes. Education facilitates, among other things, more egalitarian relationships within the couple, better communication with the spouse, and consequently greater decision-making power; better adaptability and negotiation skills; greater familiarity with health services; and better ability to communicate with health care providers and thus to demand adequate services [7, 13, 14, 15, 16]. Thus, the vast majority of educated women are able to turn away from traditional practices that are harmful to their health. Education also increases knowledge of motherhood-associated risks. An educated woman is more likely to be better informed about the need for antenatal visits or skilled attendance at birth and the benefits of care for her and her children. All studies indicate that relatively better educated women make better use of maternal health services than those with lower levels of education, who are more likely to seek care late in pregnancy and give birth at home [6, 7, 15].

The same is true for the education of spouses. The education level of husband is equally important in the decision to use maternal health care, although it appears to have less influence than that of the wife [7, 17]. When educated, husbands can understand the benefits of antenatal care and assisted delivery for their wives, allowing them to be more open to modern medicine [18]. Furthermore, financial access of pregnant and parturient women to care is facilitated by the fact that husband’s education is also associated with occupation and wealth of the household [19]. It is true that in some cases, the education of the partner has no significant effect on the use of antenatal care and assisted delivery compared to that of the woman [20]. But in most cases, this is because there is a high probability that an educated woman is in a relationship with an educated man [21].

The effect of education can go beyond individuals and affect the role of the community on health issues. For example, communities with more educated people may get organized to demand better public services and a higher profile for health on political agenda [7].

Clear disparities in access to obstetric care between rich and poor people and by level of education are combined with uneven distribution and inadequacy of human resources (health workers) and infrastructure, and poor quality of care provided by public health system [3, 22, 23].

In this context of inequalities in access to health care, several African countries have adopted subsidy and/or free health care policies in recent years. Among these, we can mention Burkina Faso, Mali, Niger, and Senegal, which have implemented measures of total or partial exemption from payment in 2006 (Burkina Faso) and 2005, respectively. Indeed, Cesarean sections are officially free in Mali, Niger, and Senegal, and deliveries are subsidized in Burkina Faso [3, 22, 23]. Even though these policies vary from one country to another, the objective is the same everywhere, namely, to make health care more accessible, particularly to the most vulnerable. These actions to improve health also include free treatment of malaria, one of the major causes of mortality in these countries, among pregnant women and children under five, particularly through the free provision of antimalarial drugs and impregnated mosquito nets.

Since the implementation of these different free and/or subsidized interventions, several studies have looked at their effects on the use of health services. This is the case for Ridde and his colleagues [3] who show an increase in the use of delivery in a health center induced by the removal of costs in some African countries (Ghana, Kenya, Madagascar, Senegal, South Africa, Tanzania, and Uganda). In the context of Burkina Faso, the subsidy and free care at the time of delivery has allowed for some empowerment of women in a context where the decision to seek care was subject exclusively to the husband’s discretion [24, 25]. Subsidies and free health care seemed to contribute to an increase in the use of health services. According to Samb [25], the majority of women report that they no longer need their husbands to pay for childbirth-related costs, although most still seek their husband’s “approval” before visiting health centers, out of respect for them.

As mentioned above, the role of health service provision and health system, more generally, in improving the level of service use during pregnancy and childbirth is more or less documented. However, the effect of changes on the population socioeconomic structure on the use of healthcare services at national level remains largely unexplored. It is true that in contexts such as those of poor countries, the costs of antenatal care and assisted childbirth remain out of reach for most individuals and their households. From this point of view, subsidy and/or exemption policies seem important for greater access to care, especially for the poorest, even if difficulties are observed in their implementation¹. But certainly, the improvement in the supply of services alone cannot explain the increase in the national level of use of maternal care.

There are still very few studies that seek to identify the contribution of changes in average living standards and educational attainment on national maternal health service utilization patterns. Canning et al. [27] find that the reduction in total female fertility induced by the increase of an additional year of schooling is smaller than that due to the average improvement in schooling in many developing countries. This implies that the change, observed at an aggregate level, produces a larger effect than that at the individual level. This is a valid argument for exploring the contribution of changes in the socioeconomic structure of the female population to recent trends in the national level of the use of health care. Examining the contribution of changes in the structure of the female population, and thus of improvements in their level of socioeconomic and human development, is all the more relevant as the living conditions of populations change and those of women with them.

Therefore, in the overall effort to improve health, socioeconomic development also matters. Beyond the improvement of supply through health policies and programs, highlighted by some authors [28], it is important to identify the effect of socioeconomic changes (change in the socioeconomic structure of the female population). Understanding the precise dimensions of this development would help guide actions to improve maternal health. This raises questions about the sources of progress in improving maternal health. Does the increase in the use of maternal health care stem from effective health programs and policies (more generally from the health system) or from global changes in the level of socio-economic and human development?

The objective of this chapter is to identify the respective roles of changes in the socioeconomic composition and the health system in general in the upward trend in the use of maternal care at the national level. The aim is to examine whether positive changes in the use of maternal health care are the result of the performance of health policies and programs (contextual health resources) or of an overall improvement in the ability of women of childbearing age to make better use of available services in favor of effective use of maternal health care. In addition, we examine whether inequalities in health care utilization between socioeconomic groups have increased or decreased by examining their respective contributions to the total change observed at the national level by comparing the situation in Mali with that in three neighboring countries (Burkina Faso, Niger, and Senegal).

The rest of the chapter is structured as follows: Section 2 present the data, the methodology, and the variables, while Section 3 reports and discusses the results. Section 4 concludes.

2. Data, methodology, and variables

2.1 Data sources

The dataset used for this chapter comes from the Demographic and Health Survey (DHS). The DHS are nationally representative surveys, conducted in different countries including Sub-Saharan Africa. We focus on three rounds for each country, namely, Mali (2001, 2006, 2012), Burkina Faso (1999, 2003, 2010), Niger (1998, 2006, 2012), and Senegal (1997, 2005, 2011).

With respect to Mali, it is important to note that the last survey (2012) occurred in the context of high political instability coupled with terrorism conflict in the northern and the central regions. So, it did not fully cover the national territory, missing the northern area (Timbuktu, Gao, and Kidal regions) and three circles of the Mopti region located in the center of the country (Douentza, Ténenkou, and Youwarou). Therefore, the 2012 data may not be comparable to the previous data, which were carried out on all regions of the country and according to a sampling procedure stratified by the region and place of residence. To circumvent this problem, we chose to work only on the southern zone, consisting of the first five regions (Kayes, Koulikoro, Sikasso, Ségou, and Mopti located in the south and center of the country, respectively) and the district of Bamako (the capital) for the examination of trends and comparison with other countries.

2.2 Methodology

In terms of methodology, the questions and hypotheses formulated in this work are addressed by means of the decomposition method developed by Eloundou-Enyegue and Giroux (2010). The decomposition method is based on the principle that the prevalence of maternal health service utilization at the national level ($Y_t$) is an average of the prevalence in the different socioeconomic categories ($y_{\mathit{it}}$) weighted by the proportion of women in each category at the same time t (_wit). The basic decomposition formula is obtained by expressing the national performance as a weighted average of group performance.

$$Y_t=\sum w_{\mathit{it}}\ast y_{\mathit{it}}$$E1

Y is the national average for the substantive variable (health service utilization during pregnancy or delivery at the national level). $y_{\mathit{it}}$ is the prevalence of maternal health service utilization for group I in year t. $w_{\mathit{it}}$ is the proportion of women belonging to group i at time t.

Based on this formula, the change in the prevalence of maternal health service utilization at the national level can be broken down as follows:

$$∆Y=\sum \overline{y_l}\ast ∆w_i+\sum \overline{w_l}\ast ∆y_i$$E2

The decomposition thus expresses the total variation due to the change in “composition” (first term in the above equation) and due to “performance/behavior” within different socioeconomic classes.

For the advanced decomposition, the performance effect is further decomposed, noting that the performance of a given group (i) can be expressed as a function of another variable(s). In the case of a linear relationship, for example,

$$y_{\mathit{it}}={\alpha }_t+{\beta }_t{x}_{\mathit{it}}+{\mu }_{\mathit{it}}$$E3

where the constant $\alpha$ represents baseline performance when $x=0$, β is the increase in prevalence of service use associated with a unit increase in variable $X$ (in our case, household socioeconomic status or female education level), and $\mu$ is the error that is interpreted as the residual effect of factors other than $X$ not considered in the analysis.

In this case, the change in the value of $y_i$ between two periods is obtained as follows:

$$∆y_i=∆\alpha +\overline{\beta }∆x_i+\overline{x_l}∆\beta +∆{\mu }_i$$E4

Since the definitions of the categories of $x$ do not change between dates $t_1$ and $t_2$, the second term in this equation is equal to zero, and $x_i=x_i$.

Thus, the equation is reduced to:

$$∆y_i=∆\alpha +x_i∆\beta +∆{\mu }_i$$E5

Replacing this Eq. (5) in the basic decomposition Eq. (2) gives the general equation for the advanced decomposition below:

$$∆Y={\sum }_j\overline{Y_J}\ast ∆w_j+{\sum }_j\overline{w_J}\ast ∆\alpha +{\sum }_j\overline{w_J}\ast x_j∆\beta +{\sum }_j\overline{w_J}\ast ∆{\mu }_j$$E6

The first term in this advanced decomposition equation represents the compositional effect, and the last three components refer respectively to the change attributable to improvement in basic health conditions, the effect of differentiating the prevalence of health service use by the classification variable used, and the residual effect of other variables not considered.

2.3 Variables

2.3.1 Prevalence of prenatal consultations and assisted deliveries

The main interest variable is the maternal health care service utilization. The variable is constructed using the women’s questionnaire. Women aged 15 to 49 who had their last birth in the 5 years preceding the surveys were asked whether they had received antenatal care from a health worker at least once and whether they were assisted by a trained provider (doctors, nurses, midwives, and matrons) at the time of childbirth. The maternal health service utilization is defined as a dummy variable if a woman who had a child in the last 5 years responded “yes” and 0 otherwise. These individuals’ responses are used to calculate the national prevalence of antenatal visits and assisted deliveries for all countries and years.

2.3.2 Socioeconomic resources of the context: Decomposition of the socioeconomic and human structure of the female population of reproductive age

National variables of living standard indicators and education are calculated. More concretely, the national proportions of women of childbearing age in each standard of living (low, intermediate, and high) and education category (none, primary, and secondary or higher) are calculated to see how these proportions change over time. Any temporal increase in the national proportions of women living in the most affluent households with high levels of education is broadly interpreted as an improvement in national socioeconomic and human development.

2.3.3 Change in socioeconomic situation at the national level: Standardized living by country

Given that the surveys were conducted at different dates, the living standard indicators available in the databases cannot be used for comparison purposes. Indeed, the households in the different surveys are not classified according to the same standard of living score scale, which varies from 1 year to another. We combined the data from three DHS round to estimate how the distribution of women by living standard categories changes over time within each country and the implication that this structural change may have on the national level of health care utilization.

We then use the characteristics of the housing and the facilities owned by households in a Principal Component Analysis (PCA) procedure to construct our country-specific standard of living indicators. These indicators are calculated, taking into account the specificities between urban and rural areas, in order to assess the living environment of a household compared to households living in a similar environment. The choice of an indicator calculated separately by environment is justified by the vast socioeconomic and infrastructure disparities that are found between urban and rural areas in most sub-Saharan countries [29, 30, 31]. The main factors derived from the PCA are used to construct categories of households, prioritized on a standard of living scale. For each of the four countries, these factors are divided into three quantiles: The first third of households are considered as having a low standard of living, the second as the intermediate group, and the last third represents households with a high standard of living.

The distribution of women of childbearing age by standard of living categories, shown in Table 1, shows an increase in the proportion of women living in households with better living conditions, except for the period 1998–2006 for Niger.

Country	DHS round	Indicator of living conditions			Total
		Low	Middle	High
Burkina Faso	1999	53.7	23.4	22.8	100.0
	2003	43.3	25.0	31.8	100.0
	2010	33.4	31.2	35.3	100.0
	Total	40.5	27.7	31.9	100.0
Mali	2001	18.0	59.0	23.0	100.0
	2006	14.7	56.8	28.5	100.0
	2012	7.1	58.7	34.2	100.0
	Total	13.7	58.1	28.2	100.0
Niger	1998	73.7	12.9	13.4	100.0
	2006	72.7	14.0	13.3	100.0
	2012	60.3	20.6	19.1	100.0
	Total	68.0	16.3	15.6	100.0
Senegal	1997	19.8	21.1	59.2	100.0
	2005	11.2	19.8	69.1	100.0
	2011	8.4	19.5	72.1	100.0
	Total	11.9	20.0	68.1	100.0

Table 1.

Distribution of women aged 15–49 by standardized standard of living, by country and year.

2.3.4 Trends in educational attainment at the national level

Another indicator used in the decomposition procedure is the change in the distribution of women by educational attainment at the national level. The level of education, which we use to perform the aggregate calculation, refers to the highest level of education attained by women at the time of the surveys. This variable, available in the databases, has been categorized in three modalities: “No level,” “Primary,” and “Secondary or higher.”

The data (Table 2) show progress in the average education levels of women of reproductive age. Progress is slow overall, although Senegal is ahead of the others. Progress is slowest in Niger, followed by Burkina Faso, and Mali.

Country	DHS round	Education level			Total
		No education	Primary	Secondary and more
Burkina Faso	1999	85.8	8.4	5.8	100.0
	2003	80.3	11.0	8.7	100.0
	2010	74.0	13.6	12.4	100.0
	Total	78.3	11.8	9.9	100.0
Mali	2001	80.0	11.3	8.7	100.0
	2006	78.2	11.4	10.3	100.0
	2012	75.8	9.3	14.9	100.0
	Total	78.2	10.8	11.0	100.0
Niger	1998	84.8	9.9	5.3	100.0
	2006	83.5	10.4	6.1	100.0
	2012	80.1	11.4	8.5	100.0
	Total	82.5	10.7	6.8	100.0
Senegal	1997	66.6	20.9	12.5	100.0
	2005	59.6	25.2	15.2	100.0
	2011	57.9	21.8	20.4	100.0
	Total	60.4	22.9	16.7	100.0

Table 2.

Distribution of women by standardized standard of living, education level by country, and year.

3. Results

3.1 Trends in maternal health services utilization

The results indicate that antenatal visits are more frequent in Burkina Faso and Senegal ahead of Mali and Niger. Mali has the lowest current rate of antenatal care utilization. Over the past twenty years, the prevalence of prenatal consultations has increased in all four countries but at different rates. Maternal care has greatly increased in Burkina Faso and Niger, particularly during the periods 2003–2010 in Burkina and 2006–2012 in Niger, when it almost doubled. Senegal and Mali also registered an increase in maternal care use, but it was slower compared to Burkina Faso and Niger. In terms of absolute levels, Senegal has been ahead of these other countries for a long time and already had a higher prenatal consultation rate in 1997 than Niger and Mali in 2012 and Burkina Faso in 2003. In the case of Mali, it was during the period 2001–2006 that a sharp increase² was observed.

In all four countries, less than half of pregnant women consult a health care professional more than three times. In Niger and Burkina Faso, only one-third of pregnant women visit a health care professional at least four times during their pregnancy. We observe that the prevalence of maternal care utilization doubled in Burkina Faso during the period 2003–2010 (after a deceleration between 1999 and 2003), in Niger during 2006–2012, and in Senegal already during the period 1997–2005. In Mali, however, the increase was less marked.

Overall, assisted delivery is less common than antenatal care, and Niger stands out for having a much lower level than the others. Its prevalence in 2012 is still below that of Mali in 2001 and Senegal in 1997 and is equivalent to that of Burkina Faso in 1999. In terms of trends, Burkina Faso stands out for a more significant increase, especially during the 2003–2010 period. The 1998–2006 period was marked by stagnation in Niger (Figure 1).

Figure 1.

Trends in maternal health service utilization by study countries.

3.2 Basic decomposition

3.2.1 Compositional effect of improving women’s socioeconomic status at the national level on increasing trends in care utilization

Using a basic decomposition, we account for the observed changes in national-level utilization of care in terms of changes in the educational composition of women versus changes in group-specific rates of care utilization (behavior or performance effect).

The results of the prenatal consultation highlight the preponderance of the effect of behaviors in all countries and for all periods except Mali between 2006 and 2012. In Mali, during this period, the improvement in the socioeconomic situation of households where women of childbearing age live had a very significant effect on the increase in the national level of care utilization. In other words, the increase in antenatal care at the national level during this period is mainly the result of the improvement in women’s collective capacity to use antenatal care services. Except in this case, progress in antenatal care attendance is mainly attributable to behavioral change across socioeconomic groups.

The results indicate that the behavioral effect predominates, apart from the case of Mali reported above, but even if it is in a relatively smaller proportion, a composition effect of the female population is also observed. Thus, in Senegal for the period 1997–2005, nearly one-third of the increase in the prenatal follow-up rate was due to this compositional effect and a little more than one-quarter in Burkina Faso for the period 1999–2003. This effect is therefore not negligible (Figure 2).

Figure 2.

Simple decomposition of the effect of standardized living in prenatal care utilization patterns.

For assisted delivery, the behavioral effect always dominates in all countries (Figure 3). However, this effect is smaller than for antenatal care, as the compositional effect, which is related to improvements in women’s socioeconomic conditions, is larger for assisted childbirth. In Burkina Faso, nearly half of the increase in the assisted childbirth rate between 1999 and 2003 was due to the compositional effect. Similarly, the share of the increase related to women’s collective capacity to use services exceeded one-third in Niger between 2006 and 2012 and in Mali in all periods and exceeded one-fifth in Senegal during the period 1997–2005. This result highlights a distinct role of improvement in socioeconomic status according to the maternal health component: It has a greater effect in increasing the overall level of use of assisted childbirth than of prenatal care.

As with the prenatal consultation, in Burkina Faso, the compositional effect was much smaller during the 1999–2003 period, during which the socioeconomic situation improved only slightly. In Niger, since the socioeconomic situation of women did not improve at all between 1998 and 2006, the compositional effect was negative, thus constituting an obstacle to increasing the national use of assisted childbirth. Moreover, this could explain, in part, the stagnation of the assisted childbirth rate during this period in Niger (Figure 3).

Figure 3.

Simple decomposition of the sources of change in health care utilization: Effect of changes in standard of living (composition effect) versus changes in group-specific behavior (behavior effect).

3.2.2 Compositional effect of overall improvement in women’s human capital on changes in care use

The results of the decomposition associated with women’s education level are similar in all the countries studied regardless of the period except in Niger between 1998 and 2006 for assisted childbirth. This case aside, the trends in both antenatal care and assisted childbirth, mostly result from changes in the behavior of mothers regardless of education level. This preponderance of the behavioral or performance effect is slightly stronger for the prenatal consultation component than for assisted delivery. Still, the educational component is also important. In Mali, the improvement in women’s education made a substantial contribution to the increase in the national prevalence of antenatal visits during the 2006–2012 period. During this period, slightly more than a quarter of the increase in the national level of use of assisted childbirth was due to the overall increase in women’s human capital. The same is true in Burkina Faso and to a lesser extent in Senegal with respect to assisted childbirth.

The situation in Niger highlights how gains in women’s education between 1998 and 2006 helped maintain the level of assisted childbirth constant during this period. This very large share, despite the stagnation of the national assisted childbirth rate, suggests that the overall proportion of assisted childbirth in Niger between 1998 and 2006 would have declined if there had not been this slight increase in women’s human capital.

In conclusion, progress in school enrolment did not raise the national rate at which women used delivery health services. Their effect remains below that relating to behavioral changes linked to the relative performance of the health system (Figures 4 and 5).

Figure 4.

Simple decomposition of the effect of education in prenatal care utilization patterns.

Figure 5.

Simple decomposition of the effect of education on health care utilization patterns at delivery.

3.3 Advanced decomposition

The advanced decomposition shows that the behavioral effect mostly reflects a baseline change for all women (baseline performance) rather than differential progress between groups (Table 3). The upward trend in health care utilization is the result of overall changes in behavior resulting mainly from improvements in the health system (service provision, awareness, policies and programs, etc.), with the exception of assisted childbirth during the periods 1998–2003 and 2001–2006 in Burkina Faso and Mali, respectively. Indeed, in these cases, differential changes in the propensity to use health care between socioeconomic categories are observed.

In Burkina Faso, the increase in the prevalence of childbirth, which occurred between 1998 and 2003 and during the 2001–2006 period in Mali, is due to a higher propensity to use health care among women belonging to less disadvantaged socioeconomic groups: intermediary or wealthy (Table 3).

Pays	Variables	DHS round	Antenatal follow-up
			Total change (b-a)	Composition (%)	Performance effect
					Base (%)	Difference (%)	Residual (%)
Burkina F.	Living standard	1999–2003	10.6	26.0	79.0	17.1	−22.1
		2003–2010	21.5	6.5	135.7	−51.1	−3.62
	Education	1999–2003	10.6	13.9	131.5	−45.4	−0.1
		2003–2010	21.5	4.7	179.8	−175.1	−0.13
Mali	Living standard	2001–2006	13.9	16.3	91.0	12.7	−20.1
		2006–2012	3.2	78.0	195.1	−122.7	−43.4
	Education	2001–2006	13.9	4.4	162.8	−68.8	1.6
		2006–2012	3.2	27.4	186.5	−207.1	−1.7
Niger	Living standard	1998–2006	6.1	4.9	130.6	−26.5	−9.0
		2006–2012	36.4	9.1	121.1	−33.5	−8.1
	Education	1998–2006	6.1	8.8	90.5	0.8	−0.2
		2006–2012	36.4	3.0	159.3	−118.7	0.1
Sénégal	Living standard	1997–2005	7.3	28.7	159.5	−62.8	−25.4
		2005–2011	3.0	13.1	246.5	−171.4	2.9
	Education	1997–2005	7.3	11.6	194.8	−103.7	−2.7
		2005–2011	3.0	3.8	257.6	−323.8	16.4
Assisted childbirth
Burkina F.	Living standard	1999–2003	8.5	42.5	26.7	69.7	−39.0
		2003–2010	29.2	8.7	130.1	−29.7	−5.01
	Education	1999–2003	8.5	25.6	51.5	23.2	−0.4
		2003–2010	29.2	8.1	172.1	−130.2	−0.15
Mali	Living standard	2001–2006	7.6	39.3	10.6	100.8	−50.6
		2006–2012	10.3	35.2	146.4	−40.9	−17.9
	Education	2001–2006	7.6	10.2	127.9	−41.4	3.3
		2006–2012	10.3	13.3	142.8	−81.8	−1.2
Niger	Living standard	1998–2006	−0.2	−34.1	110.1	−23.3	47.3
		2006–2012	13.8	35.3	115.7	−18.9	−27.6
	Education	1998–2006	−0.2	215.4	−1898.7	1790.2	−6.9
		2006–2012	13.8	12.2	134.1	−52.6	−0.9
Sénégal	Living standard	1997–2005	15.0	28.3	97.3	−2.2	−23.3
		2005–2011	10.2	14.3	178.3	−60.8	0.1
	Education	1997–2005	15.0	16.0	121.0	−34.3	−2.8
		2005–2011	10.2	9.4	207.3	−184.1	4.2

Table 3.

Results of the advanced decomposition for the contributions of baseline gains versus differential progress in maternal health care utilization.

3.4 Analyses of trends in inequalities between social groups in the use of maternal care

We examine trends in inequality across socioeconomic categories through changes in their relative contributions (Table 4). Results vary across countries. In Mali, inequalities between the different socioeconomic categories increased during the 2006–2012 period. This is undoubtedly why composition is the dominant source of the increase in the use of prenatal care during this period (see Figure 2). This probably explains the timid increase in the use of care (from 70% in 2006 to 74% in 2012), compared, for example, to Niger during the same period. The prevalence of health care utilization in Niger in 2006 was lower than that observed in Mali in 2001 (46% in Niger and 57% in Mali), but by 2012, it had become much higher in Niger (84%) than in Mali (74%) (see Figure 1). This slower increase in Mali, where the current level of prenatal monitoring is the lowest of the four countries considered, is also the result of this significant increase in inequalities in contributions between socioeconomic groups. Niger is the country where the poorest group of women has contributed the most to the change in the level of prenatal care since 1998. The more rapid increase in the prenatal care rate observed in Niger, particularly between 2006 and 2012, also stems from this strong contribution of the poorest socioeconomic groups, even though a downward trend is observed between 2006 and 2012. It is especially in Burkina Faso that inequalities between socioeconomic categories, in terms of their contribution to changes in the use of prenatal care, have decreased. This is also the case in Senegal, but to a lesser extent, because even though the gaps between socioeconomic categories are gradually narrowing, the contributions of the poorest women’s group are still low.

Variables	Burkina Faso		Mali		Niger		Senegal
	1998–2003	2003–2010	2001–2006	2006–2012	1998–2006	2006–2012	1997–2005	2005–2011
Living standard
Low	−20.7	19.5	8.3	−100.7	96.4	58.0	−73.4	−67.5
Middle	32.4	52.8	45.9	63.0	11.1	24.5	6.6	22.4
High	88.3	27.7	45.8	137.6	−7.5	17.5	166.8	145.1
Education level
No education	44.5	64.7	79.2	28.8	69.7	83.8	−9.4	40.4
Primary	27.2	17.4	9.1	−66.4	12.5	8.8	68.0	−123.1
Secondary and more	28.3	17.9	11.7	137.6	17.9	7.4	41.3	182.7
Total	100.0	100.0	100.0	100.0	100.0	100.0	100.0	100.0

Table 4.

Relative contributions of women in different socioeconomic and educational categories to total change in use of antenatal care by country.

With regard to the education effect, there is an increase in the contributions of the least educated group of women to the increase in the prenatal follow-up rate in Burkina Faso and Niger, respectively, during the periods 2003–2010 and 2006–2012. These periods were also marked by a much greater increase in prenatal follow-up in these two countries, no doubt due to the observed decrease in inequalities between the least educated and educated women. On the other hand, in Mali and Senegal, there has been an increase in inequalities between the most educated and less educated women, but their contribution to the overall increase in the prenatal follow-up rate remains significant.

3.4.1 Trends in the unequal contributions of different socioeconomic groups to increased use of assisted childbirth

The results show a decrease in inequalities between socioeconomic groups in Burkina Faso and Niger in terms of their contribution to the overall change in the use of assisted childbirth. In both countries, there was an increase in the contribution to changes in the use of assisted childbirth of the most disadvantaged group of women (those in the lowest socioeconomic category or with no education at all). This therefore implies a reduction in inequalities in this respect between socioeconomic groups. This increase in the contribution of the most disadvantaged women, from a socioeconomic or educational point of view, is the most apparent in Niger, a country where the national level of use of assisted childbirth did not increase between 1998 and 2006. This stagnation, therefore, would be due to the significant disadvantage of the group of uneducated women (who are the most numerous). It can also be assumed that, during this period, health services were not easily accessible to the poorest. During the following period, between 2006 and 2012, the reduction in the disadvantages of the least advantaged women resulted in a significant increase in the use of assisted childbirth. It was also during the period 2003–2010 that the increase in the rate of assisted childbirth was the greatest in Burkina Faso.

In Senegal, we also observe a decrease in inequality, but contrary to what is observed in Burkina Faso and Niger, it is the middle class that benefits. It is the gap between the most affluent women and those in the middle class that is narrowing the most, while the contribution of the poorest women remains low. It could therefore be said that the disadvantage of the poorest women in Senegal continues. In terms of education, the opposite situation is observed in this country, that is, an increase in inequalities to the detriment of women at the primary level.

Mali, unlike in the other three countries, shows an increase in inequalities in the use of assisted childbirth, which is expressed by the widening gap between the poorest women and others. However, despite this decrease in the contribution of uneducated women, it is mainly due to them that the slight overall increase in the national level of use of assisted childbirth in Mali between 2006 and 2012 was observed (Table 5).

Variables	Burkina Faso		Mali		Niger		Sénégal
	1998–2003	2003–2010	2001–2006	2006–2012	1998–2006	2006–2012	1997–2005	2005–2011
Living standard
Low	−9.3	29.0	5.6	−12.8	−19.6	48.7	−6.0	−7.3
Middle	26.0	47.8	35.4	56.3	20.8	19.0	4.3	28.8
High	83.2	23.2	59.0	56.5	98.8	32.3	101.7	78.5
Education level
No education	37.7	74.6	63.3	70.1	−380.3	73.2	37.4	68.8
Primary	30.9	14.3	21.5	−14.8	67.4	15.1	39.2	−20.7
Secondary and more	31.4	11.1	15.2	44.7	412.9	11.7	23.4	52.0
Total	100.0	100.0	100.0	100.0	100.0	100.0	100.0	100.0

Table 5.

Relative contributions of women in different socioeconomic and educational categories to total change in the use of assisted childbirth by country.

4. Discussion and conclusion

This chapter shed light on the roles of socioeconomic changes on living conditions of the female in maternal care service utilization in Mali, Burkina Faso, Niger, and Senegal. We also examine the trends in inequalities between socioeconomic groups in maternal health care use at the national level.

The results highlight both a behavioral and a compositional effect, but the former clearly predominates regardless of the country, except in Mali and Niger, respectively, for the periods 2006–2012 for prenatal follow-up and 1998–2006 for assisted childbirth. In Mali, during the 2006–2012 period, the increase in the use of antenatal care was marked by a predominant effect of improving women’s socioeconomic conditions. On the other hand, in Niger between 1998 and 2006, a negative effect of socioeconomic composition was observed: the stagnation in the level of use of assisted childbirth stems from the fact that the socioeconomic situation did not improve during this period. During this same period in this country, however, a slight improvement in the human capital of women of childbearing age helped keep the level of assisted childbirth constant. Except in these cases, the socioeconomic composition of the female population is not the main factor. Rather, we observe an overall trend toward behavioral change.

This predominant change in behavior is mainly due to basic performance (except for assisted childbirth in Mali between 2001 and 2006 and in Burkina Faso between 1999 and 2003). This increase in the use of maternal health services may be partly a reflection of different policy actions to improve the health status of mothers. The establishment of community health centers in the countries has brought health services closer to users. In addition, there are policies to waive the cost of Cesarean sections, to treat cases of severe malaria among pregnant women, and to make insecticide-treated nets available despite multiple dysfunctions and problems observed in their development and implementation [22, 23].

The gains permitted by service provision or policies and programs are often accompanied by inequalities across socioeconomic categories. While the contributions of the poorest people to changes in levels of health care utilization are rising in Burkina Faso and Niger, the same cannot be said for Mali and, to a lesser extent, Senegal. Contrary to our hypothesis 2, the relative disadvantage of the poorest women (especially the group of women living in poor households, slightly less so for the group of uneducated women), compared to those from the most advantaged socioeconomic strata, has increased in Mali and remained constant in Senegal despite the overall performance in health care utilization. This could explain the importance of compositional effects of Mali’s change on the national rate of antenatal monitoring between 2006 and 2012. Health services are still relatively inaccessible to the poorest socioeconomic strata, and new interventions often produce new inequalities [31].

With respect to inequalities, they tend to grow along economic, rather than educational, lines, both for prenatal monitoring and assisted childbirth. Such a result highlights the greater role of poverty as an obstacle to seeking maternal care. This suggests that the need for care to reduce the risks associated with pregnancy and childbirth is increasingly understood among the population. Maternal care tends to be widely perceived as a need, even among the least educated women, but it is financial inaccessibility combined with insufficient supply believed to be the real barriers to seeking care, rather than cultural barriers.

Finally, progress remains slow in Mali and in the other countries included in this analysis. These results warrant attention to inequalities that go beyond the health system alone. The local context (including availability of services) matters even when laudable policies and programs are adopted nationally. Another issue is women’s overall empowerment, that is, their real opportunity to use services if they so desire. This also translates into a lower contribution of the socioeconomic resources of the context (compositional effect) to the progress made in the use of care in these countries.