Levels of and changes in socioeconomic inequality in delivery care service: A decomposition analysis using Bangladesh Demographic Health Surveys.

BACKGROUND: Socioeconomic inequality in maternity care is well-evident in many developing countries including Bangladesh, but there is a paucity of research to examine the determinants of inequality and the changes in the factors of inequality over time. This study examines the factors accounting for the levels of and changes in wealth-related inequality in three outcomes of delivery care service: health facility delivery, skilled birth attendance, and C-section delivery in Bangladesh.
METHODS: This study uses from the Bangladesh Demographic and Health Survey of 2011 and 2014. We apply logistic regression models to examine the association between household wealth status and delivery care measures, controlling for a wide range of sociodemographic variables. The Erreygers normalised concentration index is used to measure the level of inequalities and decomposition method is applied to disentangle the determinants contributing to the levels of and changes in the observed inequalities.
RESULTS: We find a substantial inequality in delivery care service utilisation favouring woman from wealthier households. The extent of inequality increased in health facility delivery and C-section delivery in 2014 while increase in skilled birth attendance was not statistically significant. Wealth and education were the main factors explaining both the extent of and the increase in the degree of inequality between 2011 and 2014. Four or more antenatal care (ANC4+) visits accounted for about 8% to 14% of the observed inequality, but the contribution of ANC4+ visits declined in 2014.
CONCLUSION: This study reveals no progress in equity gain in the use of delivery care services in this decade compared to a declining trend in inequity in the last decade in Bangladesh. Policies need to focus on improving the provision of delivery care services among women from poorer socioeconomic groups. In addition, policy initiatives for promoting the completion of quality education are important to address the stalemate of equity gain in delivery care services in Bangladesh.

Introduction

By 2030, the Sustainable Development Goals (SDGs) aims to reduce global maternal mortality to less than 70 deaths per 100,000 live births and the incidence of neonatal and infant deaths to as low as 12 and 25 deaths per every 1,000 live births, respectively [1]. Achieving such progress requires an unreserved commitment by governments worldwide, through the provision of high-quality maternal care that prioritises improved access to adequate antenatal care (ANC) and facility-based birth deliveries attended by qualified health professionals. Access to high-quality maternal delivery care potentially minimises the risk of having a stillbirth or child loss due to intrapartum-related complications by about 20% [2]. Moreover, skilled birth attendants have the capacity to conduct safe deliveries including ability to detect, address and refer any complications such as haemorrhage or sepsis which are known to kill mothers during and after childbirth [3]. Despite the well-known and apparent benefits of high-quality delivery care for both mother and child, there is persistent and large socioeconomic inequality in access to and use of delivery care services in many low-middle income countries [4, 5].

Bangladesh has remarkably improved maternal and neonatal health in the last decade [6]. However, there is well-documented socioeconomic inequality in various indicators of maternal healthcare services [7-10]. Socioeconomic inequality in delivery care service utilisation favouring better-off women remains a topical issue among research and public policy circles in Bangladesh [8]. Several studies reported that socioeconomic inequality in delivery care services has decreased in Bangladesh until 2011. For example, socioeconomic inequality in skilled birth attendant and institutional delivery declined over 1993/94-2011 [11, 12]. In addition, pro-rich inequality in Caesarean section (C-section) delivery lessened by about 25% between 2004 and 2014 [7]. Despite the observed declines of inequality in the uptake of delivery care services, the overall magnitude of such inequality remains high compared to that of ANC services [13]. There is also higher regional variation in socioeconomic inequalities in delivery care services compared to ANC services [8]. Furthermore, a recent study has projected that existing socioeconomic inequality in delivery care services is most likely to persist until 2030 [14].

Improving maternal healthcare coverage at the national level and closing the gap among different countries was a key target in the Millennium Development Goals (MDGs) while reducing socioeconomic inequality within country is considered fundamental to the SDGs [15]. The idea is that policy strategies targeted at narrowing the socioeconomic divide could help bridge the gap in utilisation of delivery care services. A clear understanding of the underlying drivers of the observed inequality in health facility delivery, skilled birth assistance, and C-section delivery is not only an important step in the design of effective policies to improve the health and wellbeing of Bangladeshi women and their children, but also critical towards the monitoring and evaluation of progress towards goals 3.1 and 3.2 of the SDGs. Quantifying and explaining socioeconomic inequalities are essential for health planners to design policies that target specific sub-groups of the population where health resources are the most needed.

There are limited studies to explain the factors accounting for socioeconomic inequalities in the utilisation of delivery care services in Bangladesh [7, 16]. However, these studies did not consider several methodological issues associated with the measurement and decomposition of socioeconomic inequality in binary healthcare outcomes. For example, changes in the average level of C-section delivery was ignored when comparing socioeconomic inequality between 2004 and 2014 in the study by Khan et al., 2018 [7]. In addition, the decomposition method was used to explain the underlying factors that contribute to inequality, but statistical inference of the estimates from the decomposition analysis was not provided. In addition, there is no evidence to know how the role of contributing factors of inequality changes between two periods. Therefore, this study aims to address limitations of earlier studies by measuring and explaining wealth-related inequalities in health facility delivery, skilled birthattendance, and C-section delivery between 2011 and 2014. We also unravel the factors associated with the changes in inequalities. The findings of this study have important implications in terms of enlightening health planners, government, and other public health stakeholders interested in contributing towards the design and implementation of policies to alleviate socioeconomic inequalities in delivery care services in Bangladesh and other developing countries.

Materials and methods

Data and sample

The study is based on data from the two most recent rounds of the Bangladesh Demographic and Health Survey (BDHS), implemented in 2011 and 2014. The BDHS is a nationally representative cross-sectional survey conducted in every three years as a part of the global DHS programme since 1991/92. The BDHS employed a complex multistage sample design to collect data on maternal health and healthcare utilisation from women of reproductive age (15–49 years). The response rate of these surveys was about 98% [17]. We restrict our analysis to women who have had at least one live birth in the last three years preceding the surveys and only consider data related to the most recent delivery if there were multiple live births within the timeframe. After discarding the observations (around 2%) with missing information on selected variables, our final sample consists of 4638 and 4481 women in 2011 and 2014, respectively.

Measures of delivery care service

We analyse three dichotomous indicators of delivery care services utilisation: health facility delivery, skilled birth attendance, and C-section delivery. Health facility delivery is coded as 1 if a mother gave birth in a health facility (e.g. public hospital, district hospital, maternal and child welfare centre, upazila health complex, upazila health & family welfare centre, other public sector facility, community clinic, private hospital/clinic, NGO static clinic, other NGO facility etc.) and 0 if the birth occurred at home. Skilled birth attendance takes the value of 1 if the birth was assisted by a medically trained professional (e.g. qualified doctor, nurse, midwife, family welfare visitor and community skilled birth attendant) and 0 otherwise. Lastly, C-section delivery is a binary variable which is equal to 1 if the birth was a C-section delivery and 0 otherwise. Definitions and measurements of outcome variables used in this study were similar in both surveys.

Determinants of delivery care services

This study follows previous literature from Bangladesh and other developing countries to select the determinants of delivery care service utilisation [13, 18–22]. These include women’s age at the time of survey, age at marriage, religion, number of children ever born (parity), history of pregnancy complications, at least four visits for antenatal care (ANC4+), exposure to mass media, involvement in microcredit and education. Education of husband is also included as the predictor of delivery care services [22]. Region and place of residence (rural/urban) are the geographic variables. Since BDHS does not collect information on household expenditure or income, the measure of socioeconomic status (SES) in this study is an asset-based wealth index [23].

The wealth index was constructed using Principal Components Analysis (PCA), which comes pre-calculated in the BDHS [24, 25]. The wealth index includes a range of assets or belongings owned by the households. Examples of the assets are radio/TV, refrigerators, farmland, farming animals, house construction materials, water, sanitation infrastructure etc [24]. PCA is a multivariate statistical method that is widely used as a data reduction technique [25]. This method creates uncorrelated components with each component consisting of a linear weighted combination of the original asset variables [25, 26]. The resulting components are arranged in such a way that the first principal component explains the largest variability in the data [25]. The asset score (continuous variable) is used to rank households from the lowest to the highest to compute our measures of SES-related inequality. Households are categorised into five wealth quintiles from the poorest (quintile 1) to the richest (quintile 5) [26, 27]. A detailed description of the methods underlying the construction of the asset-based index are described elsewhere [25, 28].

Statistical analysis

This study uses logistic regression analysis to examine the association between the outcomes of delivery care service with demographic and socioeconomic variables. We measure and explain socioeconomic inequality in the utilisation of delivery care services using the concentration index (CI) and the decomposition method [29, 30]. We estimate the CI as follows: where y is indicator of delivery care use for individual i, r is the fractional ranking of individuals according to wealth index and μ is the mean of y. The value of this index falls between −1 and 1 [30]. A negative CI indicates higher utilisation among the poor (pro-poor) while a positive value suggests greater utilisation among the rich (pro-rich). The higher the absolute value of the CI is, the greater the extent of inequality.

The range of the CI becomes smaller when the variable of interest is a binary indicator. This is because of the lower and the upper bounds of the CI depending on the mean of the outcome variable [31]. As a result, the change in socioeconomic inequality measured by the CI could be affected considerably if the mean of the variable of interest changes over time [32]. Therefore, we use the Erreygers Index (EI) to address this problem [32]. The EI is basically a normalised version of the CI as below:

The interpretation of EI is like the standard CI. A positive EI shows the distribution of delivery care services favouring women from wealthier households i.e pro-rich inequality and vice versa. We then employ the decomposition technique to partition wealth-related inequalities in delivery care services. We assume that y, utilisation of delivery care services is modelled by an additively separable linear function of X (a vector of covariates) as shown below:

Using Eq (3), the EI can be decomposed into the weighted sum of the socioeconomic inequality in the determinants for delivery care use [33]. Weight refers to the sensitivity of utilisation with respect to each covariate, which is defined by . As our outcome variable is binary-thus providing more justification for the use of EI, the decomposition of the EI follows:

In this expression, β is the partial effects of healthcare determinants, CI is the concentration indices of X and GCI is the generalized CI of the error term. Eq 5 suggests that a variable contributes to inequality in the use of delivery care services when two conditions are satisfied: (1) it must be correlated with delivery use and (2) it must be unequally distributed across socioeconomic status as measured by the EI [23]. The higher is the partial effect of a variable and the more unequally the variable is distributed with respect to SES, the higher the contribution of that variable. We run linear probability models to estimate the coefficients of determinants of delivery care utilization in the decomposition analysis as a non-linear model could induce approximation error [34].

We are also interested in understanding how inequality has changed between 2011 and 2014 and the role of contributing factors in this change. For this purpose, we apply the Oaxaca-type decomposition following [33] to the EI. This procedure is shown in the following equation:

In Eq (5), Δ stands for differences across time. This method allows us to decompose the evolution of SES-related inequality in delivery care use into two components. The first component ( represents the changes in socioeconomic inequality in the determinants while the second one measures the changes in the sensitivity of utilisation with respect to each covariate over time [29]. We follow the method of van Doorslaer et.al. [34] to obtain the standard errors of the CIs and their contributions by applying the bootstrapped method with 1,000 replications. This allows us to make the statistical inferences of the point estimates. We account for the multistage survey design in descriptive, regression, and decomposition analyses by using the sample weights, clusters, and strata provided in the BDHS data sets. All analyses are performed using Stata/MP version 15.1(Stata Corp., College Station, TX, USA).

Ethics statement

The BDHSs 2011 and 2014 were implemented under the authority of the National Institute of Population Research and Training (NIPORT), the Ministry of Health and Family Welfare, Bangladesh. Mitra and Associates, a Bangladeshi research firm located in Dhaka conducted the surveys with technical assistance from the ICF International of Calverton, Maryland, USA. Institutional Review Board of the InnerCity Fund (ICF) Macro, Maryland, USA, and the National Research Ethics Committee of Bangladesh Medical Research Council (BMRC), Dhaka, Bangladesh approved the protocol of these surveys. Verbal consents were taken from the participants before conducting in the interviews. We obtained the de-identified data for this study from the DHS online [35].

Results

Summary statistics

Table 1 reports that the percentage of women having skilled birth attendance increased from 31.9% in 2011 to 42.9% in 2014. The coverage of facility delivery increased from about 29% in 2011 to about 39% in 2014. There was an increase in C-section delivery by about 7 percentage point over the same period (17.2% in 2011 to 25.2% in 2014). About 27.1% of women were married by the age of 12–14 in 2014 compared to about 33.4% in 2011. The number of women with more than three children declined from 34.6% in 2011 to 30.0% in 2014. The coverage of at least four ANC visits increased from 25.5% in 2011 to 31.2% in 2014. The completion rate of secondary and higher education increased among both women and their husbands in 2014.

Table 1

Weighted distribution of respondents by selected background characteristics.

	2011		2014
Variables	N	Proportion	N	Proportion
Facility delivery	1350	29.1%	1730	38.6%
Skilled birth attendance	1480	31.9%	1922	42.9%
C-section delivery	798	17.2%	1084	24.2%
Current age (Ref: 15–19)	914	19.7%	941	21.0%
20–24	1739	37.5%	1506	33.6%
25–34	1725	37.2%	1770	39.5%
35+	260	5.6%	269	6.0%
Age at marriage: Year 18+	1085	23.4%	1268	28.3%
Year: 15–17	2004	43.2%	1999	44.6%
Year: 12–14	1549	33.4%	1214	27.1%
Parity (Ref: 1 child)	1674	36.1%	1788	39.9%
2 children	1359	29.3%	1349	30.1%
3 or more children	1605	34.6%	1344	30.0%
Religion (Ref: Islam)	408	8.8%	372	8.3%
Pregnancy complication (Ref: No)	775	16.7%	636	14.2%
ANC4+ visits (Ref: No)	1183	25.5%	1398	31.2%
Mass media exposure (Ref. No)	1665	35.9%	1716	38.3%
Irregular	682	14.7%	484	10.8%
Regular	2291	49.4%	2285	51.0%
Microcredit involvement (Ref: No)	1498	32.3%	1304	29.1%
Women education (Ref: No)	821	17.7%	636	14.2%
Primary	1396	30.1%	1250	27.9%
Secondary	2078	44.8%	2137	47.7%
Higher	343	7.4%	457	10.2%
Husband education (Ref: No)	1285	27.7%	1071	23.9%
Primary	1382	29.8%	1344	30.0%
Secondary	1382	29.8%	1420	31.7%
Higher	589	12.7%	645	14.4%
Wealth quintile (Ref. Poorest)	1062	22.9%	972	21.7%
Poorer	918	19.8%	851	19.0%
Middle	914	19.7%	856	19.1%
Richer	904	19.5%	923	20.6%
Richest	835	18.0%	883	19.7%
Urban resident (Ref: Rural)	3576	77.1%	3311	73.9%
Region (Ref: Barisal)	260	5.6%	260	5.8%
Chittagong	1076	23.2%	977	21.8%
Dhaka	1415	30.5%	1582	35.3%
Khulna	441	9.5%	358	8.0%
Rajshahi	612	13.2%	448	10.0%
Rangpur	487	10.5%	435	9.7%
Sylhet	343	7.4%	417	9.3%

Predictors of delivery care services

Regression results in Table 2 show that women having more than three children were significantly (OR: 0.40 with p<0.01) less likely to use all three services, compared with women with only one child. At least four ANC visits was positively associated with the likelihood of the uptake of delivery care services. For example, the likelihood of facility delivery for those with ANC4+ was about 2.4 (p ≤ 0.01) times higher in 2011 and 1.9 (p ≤ 0.01) times higher in 2014 than those who had less than four ANC visits. Educational attainment of both women and their husbands was significantly associated with the use of delivery care services. For example, women married to a husband with higher educational achievement were about 2.6 times more likely to have a C-section delivery, compared with women whose husbands had no primary education attainment in 2014. Women from poorer households were significantly less likely to use delivery care services compared to their counterparts from the richer household. The wealth gradient in the uptake of delivery care services became steeper in 2014 as shown by the higher OR of the richer quintiles. For example, the OR of giving birth at a health facility in the richest quintile increased from 3.19 in 2011 to 4.59 in 2014.

Table 2

Multivariate logistic regression results for the factors associated with the use of delivery care services.

	Facility delivery						Skilled birth attendance						C-section delivery
	2011			2014			2011			2014			2011			2014
	AOR	SE	P	AOR	SE	P	AOR	SE	P	AOR	SE	P	AOR	SE	P	AOR	SE	P
Age (years) at survey date (Ref: 15–19)
20–24	1.04	(0.16)	0.81	1.16	(0.16)	0.27	0.99	(0.15)	0.95	1.03	(0.14)	0.81	1.51	(0.26)	0.02	1.09	(0.17)	0.60
25–34	1.48	(0.28)	0.04	1.66	(0.30)	0.01	1.42	(0.25)	0.05	1.49	(0.28)	0.04	2.30	(0.49)	0.00	1.71	(0.34)	0.01
35+	2.00	(0.57)	0.02	2.41	(0.61)	0.00	1.81	(0.50)	0.03	1.92	(0.51)	0.01	3.29	(1.07)	0.00	2.80	(0.86)	0.00
Age (years) at marriage (Ref: 18+)
15–17	0.89	(0.12)	0.37	0.77	(0.09)	0.02	0.95	(0.12)	0.70	0.79	(0.08)	0.02	1.02	(0.14)	0.89	0.66	(0.10)	0.01
12–14	0.69	(0.11)	0.03	0.70	(0.11)	0.02	0.69	(0.11)	0.02	0.69	(0.10)	0.01	0.78	(0.15)	0.18	0.73	(0.14)	0.10
Parity (Ref: 1 child)
2 children	0.59	(0.07)	0.00	0.57	(0.08)	0.00	0.58	(0.07)	0.00	0.60	(0.09)	0.00	0.59	(0.09)	0.00	0.55	(0.07)	0.00
3 or more children	0.40	(0.07)	0.00	0.41	(0.06)	0.00	0.42	(0.07)	0.00	0.41	(0.07)	0.00	0.36	(0.08)	0.00	0.39	(0.08)	0.00
Religion (Ref: Islam)	1.66	(0.27)	0.00	1.12	(0.27)	0.62	1.71	(0.26)	0.00	1.01	(0.22)	0.97	1.36	(0.26)	0.11	1.03	(0.18)	0.85
Pregnancy complication (Ref: No)	1.36	(0.15)	0.01	1.15	(0.13)	0.21	1.30	(0.14)	0.02	1.13	(0.13)	0.30	1.37	(0.20)	0.03	1.17	(0.14)	0.19
ANC4+ visits (Ref: No)	2.39	(0.23)	0.00	1.88	(0.17)	0.00	2.35	(0.23)	0.00	1.89	(0.17)	0.00	1.98	(0.23)	0.00	1.80	(0.20)	0.00
Mass media exposure (Ref. No)
Irregular	1.21	(0.18)	0.21	1.20	(0.19)	0.25	1.19	(0.17)	0.23	1.08	(0.16)	0.60	1.54	(0.29)	0.02	0.86	(0.17)	0.46
Regular	1.40	(0.18)	0.01	1.22	(0.18)	0.17	1.46	(0.17)	0.00	1.15	(0.16)	0.30	1.53	(0.24)	0.01	1.11	(0.17)	0.51
Microcredit involvement (Ref: No)	1.09	(0.11)	0.36	1.35	(0.22)	0.07	1.21	(0.11)	0.04	1.33	(0.21)	0.07	0.98	(0.12)	0.88	1.40	(0.20)	0.02
Women education (Ref: No)
Primary	1.15	(0.21)	0.43	1.54	(0.27)	0.01	1.10	(0.18)	0.56	1.57	(0.25)	0.00	1.14	(0.27)	0.58	1.40	(0.29)	0.11
Secondary	1.44	(0.26)	0.04	1.82	(0.32)	0.00	1.44	(0.24)	0.03	1.96	(0.32)	0.00	1.42	(0.34)	0.14	2.19	(0.50)	0.00
Higher	2.46	(0.61)	0.00	2.39	(0.57)	0.00	3.20	(0.78)	0.00	2.69	(0.61)	0.00	2.43	(0.70)	0.00	2.26	(0.60)	0.00
Husband education (Ref: No)
Primary	1.00	(0.14)	0.99	1.13	(0.13)	0.30	1.09	(0.14)	0.53	1.14	(0.13)	0.22	1.33	(0.26)	0.13	1.25	(0.28)	0.30
Secondary	1.32	(0.19)	0.05	1.34	(0.19)	0.04	1.41	(0.19)	0.01	1.45	(0.20)	0.01	1.77	(0.33)	0.00	1.65	(0.39)	0.03
Higher	1.85	(0.33)	0.00	2.10	(0.39)	0.00	1.80	(0.32)	0.00	2.60	(0.48)	0.00	2.54	(0.56)	0.00	2.58	(0.71)	0.00
Wealth quintile (Ref. Poorest)
Poorer	1.30	(0.22)	0.13	1.42	(0.19)	0.01	1.16	(0.19)	0.36	1.49	(0.20)	0.00	2.47	(0.66)	0.00	1.17	(0.30)	0.54
Middle	1.51	(0.25)	0.01	1.73	(0.32)	0.00	1.50	(0.23)	0.01	1.64	(0.30)	0.01	2.80	(0.70)	0.00	1.67	(0.52)	0.10
Richer	2.25	(0.38)	0.00	2.43	(0.38)	0.00	2.12	(0.34)	0.00	2.38	(0.37)	0.00	3.72	(0.98)	0.00	2.40	(0.54)	0.00
Richest	3.18	(0.64)	0.00	4.52	(0.80)	0.00	2.92	(0.56)	0.00	4.00	(0.65)	0.00	5.68	(1.60)	0.00	4.08	(1.04)	0.00
Place of residence (Ref: Rural)	0.68	(0.08)	0.00	0.71	(0.09)	0.01	0.62	(0.07)	0.00	0.77	(0.09)	0.03	1.18	(0.16)	0.22	0.84	(0.11)	0.19
Region (Ref: Barisal)
Chittagong	1.20	(0.22)	0.33	1.03	(0.20)	0.90	1.05	(0.18)	0.78	1.09	(0.23)	0.68	1.03	(0.21)	0.89	0.84	(0.16)	0.34
Dhaka	1.34	(0.25)	0.11	1.26	(0.22)	0.20	0.96	(0.16)	0.80	1.02	(0.20)	0.92	1.66	(0.35)	0.02	1.56	(0.26)	0.01
Khulna	2.72	(0.50)	0.00	2.61	(0.54)	0.00	2.04	(0.35)	0.00	2.13	(0.48)	0.00	2.03	(0.43)	0.00	2.03	(0.36)	0.00
Rajshahi	1.97	(0.38)	0.00	1.69	(0.32)	0.01	1.36	(0.25)	0.10	1.31	(0.27)	0.18	1.77	(0.38)	0.01	1.40	(0.26)	0.07
Rangpur	1.49	(0.30)	0.05	1.21	(0.26)	0.37	0.99	(0.18)	0.94	0.95	(0.21)	0.82	0.99	(0.23)	0.96	0.87	(0.20)	0.53
Sylhet	1.16	(0.24)	0.47	0.84	(0.17)	0.40	0.97	(0.17)	0.86	0.79	(0.17)	0.29	1.11	(0.24)	0.64	0.72	(0.15)	0.11

Notes: AOR = adjected odds ratio, SE = standard error, and robust standard errors in parentheses.

Inequality in delivery care

There was an overall increase in the use of delivery care services across all wealth quintiles from 2011 to 2014, but the uptake of these services was lower among women from poorer households compared to women in wealthier households (Fig 1). It is noticeable that the absolute increase in the utilisation was higher among the women from wealthier households between 2011 and 2014. We also show this gradient in Fig 2 which plots the predicted probabilities from logistic regression models which include the interaction between year and wealth quintiles. The increase in predicted probabilities for facility delivery and C-section delivery were the highest and significant among women from the richest wealth quintile since there was no overlap in the 95% confidence intervals between 2011 and 2014. On the other hand, there was no notable change in the predicted probabilities between 2011 and 2014 for women from other wealth quintiles for these two outcomes.

Fig 1

Proportion of delivery care service utilisation by wealth quintiles.

Fig 2

Predicted probabilities of delivery care service utilisation across wealth quintile.

Table 3 shows that the estimates of inequality were positive and statistically significant in both years. This result suggests that the distribution of delivery care services utilization was concentrated among women from wealthier households. There was an increase the value of the EI for all three outcomes. For example, the EI for facility delivery increased from 0.41 in 2011 to 0.47 in 2014 while that for C-section delivery increased from 0.31 in 2011 to 0.38 in 2014. However, the increase in the EI for skilled birth attendance was not statistically significant.

Table 3

Inequality in delivery care services in Bangladesh, 2011 and 2014.

	Facility delivery				Skilled birth attendance				C-section delivery
	2011		2014		2011		2014		2011		2014
Inequality index	Estimate	SE	Estimate	SE	Estimate	SE	Estimate	SE	Estimate	SE	Estimate	SE
EI	0.41***	(0.02)	0.47***	(0.02)	0.44***	(0.02)	0.47***	(0.02)	0.31***	(0.02)	0.38***	(0.02)
Observations	4638		4481		4638		4481		4638		4481

Notes: Standard errors in parentheses and significance level * p<0.10, ** p<0.05, *** p<0.01. EI = Erreygers corrected concentration index, SE = standard error.

Decomposition of inequality in delivery care

The decomposition results in Table 4 suggest that household wealth explains about 36.3% of the socioeconomic inequality in facility delivery in 2011 and about 48.1% in 2014. For skilled birth attendance, its contribution increased from around 34.7% in 2011 to 45.0% in 2014. For C-section delivery, there was also an increase in the relative contribution of wealth status from about 38.4% in 2011 to 44.2% in 2014. The relative contribution of women’s education reduced to 8.3% in 2014 from 10.8% in 2011 for facility delivery. On the other hand, husband’s education explained about 10.4% of the inequality in skilled birth attendance in 2011 which increased to about 15.5% in 2014. ANC4+ visits had a positive contribution to the wealth-related inequality in delivery care, but its relative importance declined in 2014. For instance, its relative contribution to inequality in facility delivery was 14.0% in 2011, but it declined to about 8.8% in 2014.

Table 4

Factor contributions to wealth-related inequalities in the use of delivery care services (decomposition of the EI).

	CI of factors		Facility delivery				Skilled birth attendance				C-section delivery
	2011	2014	2011		2014		2011		2014		2011		2014
Variables	CI	CI	Contr.	(%)	Contr.	(%)	Contr.	(%)	Contr.	(%)	Contr.	(%)	Contr.	(%)
Age (Ref: 15–19)
20–24	0.000	0.041**	0.000	0.000	0.002	0.391	-0.000	-0.001	0.000	0.087	0.000	0.003	0.001	0.340
25–34	0.033**	0.004	0.003	0.774	0.001	0.127	0.003	0.683	0.000	0.086	0.005*	1.722	0.000	0.131
35+	-0.097*	-0.005	-0.002	-0.501	-0.000	-0.040	-0.002	-0.421	-0.000	-0.032	-0.003*	-0.929	-0.000	-0.052
Age at marriage (Ref: 18+)
15–17	-0.014	-0.038**	0.001	0.129	0.003	0.693	0.000	0.071	0.003	0.645	0.000	0.012	0.004*	1.163
12–14	-0.150***	-0.138***	0.011*	2.708	0.010*	2.083	0.012**	2.733	0.010*	2.182	0.004	1.407	0.008*	1.983
Parity (Ref: 1 child)
2 children	0.060***	0.040*	-0.006**	-1.531	-0.005*	-1.064	-0.007**	-1.523	-0.005*	-0.989	-0.005**	-1.558	-0.004*	-1.174
3 or more children	-0.171***	-0.177***	0.033***	7.939	0.034***	7.260	0.032***	7.408	0.035***	7.415	0.027***	9.001	0.028***	7.392
Religion (Ref: Islam)	0.060*	-0.049	0.002	0.453	-0.000	-0.072	0.002	0.457	0.000	0.004	0.001	0.287	-0.000	-0.050
Pregnancy complication (Ref: No)	-0.006	0.029	-0.000	-0.042	0.000	0.081	-0.000	-0.035	0.000	0.067	-0.000	-0.043	0.000	0.074
ANC4+ visits (Ref: No)	0.329***	0.261***	0.058***	13.986	0.041***	8.750	0.058***	13.196	0.042***	8.904	0.035***	11.574	0.033***	8.650
Mass media exposure (Ref. No)
Irregular	-0.165***	-0.073*	-0.002	-0.432	-0.001	-0.200	-0.002	-0.438	-0.000	-0.104	-0.002	-0.776	0.001	0.192
Regular	0.329***	0.336***	0.031**	7.532	0.026	5.592	0.039***	8.969	0.021	4.435	0.022*	7.129	0.009	2.465
Microcredit involvement (Ref: No)	-0.148***	-0.144***	-0.003	-0.653	-0.009*	-1.877	-0.006*	-1.300	-0.009*	-1.881	0.000	0.014	-0.007*	-1.847
Women education (Ref: No)
Primary	-0.184***	-0.256***	-0.001	-0.170	-0.015*	-3.223	0.000	0.080	-0.017*	-3.695	0.001	0.418	-0.005	-1.240
Secondary	0.189***	0.141***	0.014*	3.413	0.024**	5.062	0.016*	3.682	0.030***	6.328	0.006	1.875	0.021**	5.441
Higher	0.635***	0.553***	0.031***	7.531	0.030**	6.505	0.038***	8.760	0.035***	7.366	0.033***	10.828	0.026**	6.903
Husband education (Ref: No)
Primary	-0.117***	-0.171***	0.001	0.208	-0.004	-0.752	-0.001	-0.159	-0.004	-0.942	-0.002	-0.609	-0.004	-1.028
Secondary	0.233***	0.197***	0.012*	2.933	0.015*	3.116	0.016**	3.559	0.021**	4.408	0.014**	4.458	0.016**	4.125
Higher	0.545***	0.523***	0.034***	8.24	0.044***	9.413	0.031***	7.037	0.056***	12.026	0.039***	12.836	0.050***	13.326
Wealth quintile (Ref. Poorest)
Poorer	-0.344***	-0.377***	-0.004	-0.987	-0.011	-2.445	-0.001	-0.255	-0.017*	-3.566	-0.008*	-2.573	0.002	0.440
Middle	0.051***	0.003	0.001	0.334	0.000	0.044	0.002	0.382	0.000	0.038	0.002	0.493	0.000	0.016
Richer	0.443***	0.401***	0.039***	9.415	0.052***	11.166	0.040***	9.179	0.055***	11.779	0.026***	8.472	0.030**	7.835
Richest	0.819***	0.803***	0.114***	27.565	0.184***	39.340	0.111***	25.382	0.172***	36.742	0.098***	32.036	0.136***	35.912
Place of residence (Ref: Rural)	-0.142***	-0.158***	0.029***	6.957	0.029**	6.265	0.036***	8.309	0.022*	4.744	-0.008	-2.532	0.014	3.806
Region (Ref: Barisal)
Chittagong	0.050***	0.099***	0.001	0.290	0.000	0.080	0.000	0.090	0.002	0.339	0.000	0.088	-0.002	-0.605
Dhaka	0.085***	0.112***	0.005*	1.142	0.006	1.314	-0.000	-0.045	0.000	0.095	0.006**	1.892	0.010*	2.651
Khulna	0.087***	-0.015	0.006***	1.338	-0.001	-0.188	0.004**	0.983	-0.001	-0.158	0.003**	0.935	-0.001	-0.145
Rajshahi	-0.094***	-0.138***	-0.005**	-1.194	-0.005**	-1.079	-0.002	-0.538	-0.003	-0.567	-0.003*	-0.927	-0.003	-0.673
Rangpur	-0.233***	-0.210***	-0.005*	-1.172	-0.002	-0.494	0.000	0.085	0.001	0.229	0.001	0.263	0.002	0.553
Sylhet	-0.023	-0.171***	-0.000	-0.038	0.001	0.245	0.000	0.005	0.002	0.444	-0.000	-0.031	0.001	0.357
Explained inequality			0.391	95.167	0.449	96.093	0.419	96.335	0.451	96.429	0.292	95.765	0.366	96.941
Residual			0.016	4.833	0.018	3.907	0.016	3.665	0.017	3.571	0.013	4.235	0.012	3.059

Note: Significance level *** p<0.01, ** p<0.05, and * p<0.1. CI = concentration index, Contr = contribution (absolute).

Decomposing changes in inequality in delivery care

Table 5 presents the results from the analysis of the decomposition of changes in inequality. Most of the change in inequality was associated with wealth status. The aggregate contribution (a sum of total changes for wealth variable) was 0.075 for facility delivery, 0.059 for skilled birth attendance, and 0.050 for C-Section. Notably, the large positive contributions were due to the sensitivity effects of the richest quintile. Their positive contributions were triggered by the steeper pro-rich gradient of delivery care service use in 2014. This result is clear from the finding that the coefficients of the richest quintile increased in 2014 for all the three outcomes. For example, the OLS coefficient estimate for facility delivery shows an increase from 0.188 to 0.272. On the other hand, the inequality effects of wealth variable show little contributions to the changes in the CIs of delivery care utilisation. In fact, the CI for the richest wealth quintile changed from 0.819 to 0.803 only. Husband’s education was the second largest contributor to changes in inequality in facility delivery. The proportion of husbands who accomplished higher education increased from 12.7% to 14.4% between 2011 and 2014, and it was more concentrated among wealthier women in 2014.

Table 5

Changes in the contributing factors of inequalities in delivery care service use (decomposition of the EI).

	Facility delivery			Skilled birth attendance			C-section delivery
	ΔSensitivity	ΔCI	Total	ΔSensitivity	ΔCI	Total	ΔSensitivity	ΔCI	Total
Current age (Ref: 15–19)
20–24	0.0000	0.0018	0.0018	0.0000	0.0004	0.0004	0.0000	0.0013	0.0013
25–34	0.0016	-0.0042	-0.0026	0.0007	-0.0033	-0.0026	-0.0008	-0.0040	-0.0048
35+	-0.0014	0.0032	0.0018	-0.0007	0.0024	0.0017	-0.0005	0.0031	0.0026
Age at marriage: Year 18+
Year: 15–17	0.0006	0.0021	0.0027	0.0008	0.0020	0.0028	0.0015	0.0029	0.0044
Year: 12–14	-0.0005	-0.0009	-0.0014	-0.0006	-0.0011	-0.0017	0.0040	-0.0008	0.0032
Parity (Ref: 1 child)
2 children	-0.0010	0.0024	0.0014	0.0000	0.0020	0.0020	-0.0016	0.0020	0.0004
3 or more children	-0.0003	0.0013	0.0010	0.0009	0.0014	0.0023	-0.0007	0.0011	0.0004
Religion (Ref: Islam)	-0.0015	-0.0008	-0.0023	-0.0021	0.0000	-0.0021	-0.0006	-0.0004	-0.0010
Pregnancy complication (Ref: No)	0.0001	0.0005	0.0006	0.0001	0.0004	0.0005	0.0001	0.0003	0.0004
ANC4+ visits (Ref: No)	-0.0061	-0.0105	-0.0166	-0.0052	-0.0105	-0.0157	0.0057	-0.0083	-0.0026
Mass media exposure (Ref. No)
Irregular	-0.0003	0.0012	0.0009	0.0008	0.0006	0.0014	0.0040	-0.0009	0.0031
Regular	-0.0058	0.0005	-0.0053	-0.0191	0.0004	-0.0187	-0.0126	0.0002	-0.0124
Microcredit involvement (Ref: No)	-0.0062	0.0002	-0.0060	-0.0034	0.0003	-0.0031	-0.0072	0.0002	-0.0070
Women education (Ref: No)
Primary	-0.0101	-0.0043	-0.0144	-0.0129	-0.0048	-0.0177	-0.0047	-0.0013	-0.0060
Secondary	0.0177	-0.0082	0.0095	0.0239	-0.0104	0.0135	0.0221	-0.0072	0.0149
Higher	0.0040	-0.0045	-0.0005	0.0016	-0.0051	-0.0035	-0.0030	-0.0039	-0.0069
Husband education (Ref: No)
Primary	-0.0033	-0.0011	-0.0044	-0.0023	-0.0014	-0.0037	-0.0008	-0.0012	-0.0020
Secondary	0.0051	-0.0027	0.0024	0.0088	-0.0037	0.0051	0.0048	-0.0028	0.0020
Higher	0.0117	-0.0018	0.0099	0.0280	-0.0024	0.0256	0.0134	-0.0022	0.0112
Wealth quintile (Ref. Poorest)
Poorer	-0.0064	-0.0010	-0.0074	-0.0142	-0.0014	-0.0156	0.0094	0.0001	0.0095
Middle	0.0018	-0.0030	-0.0012	0.0015	-0.0030	-0.0015	-0.0004	-0.0010	-0.0014
Richer	0.0189	-0.0057	0.0132	0.0211	-0.0061	0.0150	0.0071	-0.0033	0.0038
Richest	0.0736	-0.0037	0.0699	0.0650	-0.0036	0.0614	0.0409	-0.0028	0.0381
Place of residence (Ref: Rural)	-0.0026	0.0030	0.0004	-0.0165	0.0023	-0.0142	0.0206	0.0015	0.0221
Region (Ref: Barisal)
Chittagong	-0.0011	0.0002	-0.0009	0.0004	0.0008	0.0012	-0.0015	-0.0011	-0.0026
Dhaka	-0.0002	0.0015	0.0013	0.0004	0.0001	0.0005	0.0018	0.0025	0.0043
Khulna	-0.0005	-0.0060	-0.0065	-0.0002	-0.0049	-0.0051	0.0002	-0.0036	-0.0034
Rajshahi	0.0016	-0.0015	0.0001	0.0006	-0.0008	-0.0002	0.0011	-0.0008	0.0003
Rangpur	0.0025	0.0002	0.0027	0.0010	-0.0001	0.0009	0.0015	-0.0002	0.0013
Sylhet	0.0003	0.0010	0.0013	0.0003	0.0018	0.0021	0.0003	0.0012	0.0015

Discussion

The study measures and examines the extent of wealth-related inequalities in the utilisation of delivery care services in Bangladesh between 2011 and 2014. This study also explains the contributing factors that characterise the dynamics and the changes in the observed inequality. The findings reveal a substantial pro-rich inequality in the utilisation of three key components of delivery care services. Most importantly, the magnitude of absolute inequalities increased between 2011 and 2014 in health facility delivery and C-section delivery. Findings from the decomposition analysis indicate that household’s wealth and education of both women and their husbands were the most important factors to explain the extent of and change in socioeconomic inequalities in delivery care in Bangladesh over the study period.

Our findings reveal a significant pro-wealth inequality in three outcomes of delivery care services, which is in line with earlier studies on socioeconomic inequalities in the use of maternal healthcare in Bangladesh [8, 9, 13, 36–38]. In a multi-country study, Bangladesh was ranked as the fourth most inequitable country in skilled birth attendance among 54 developing countries [39]. The extent of socioeconomic inequality in health facility delivery in Bangladesh is also one of the highest among the countries in South and East Asia [40]. We find that wealth-related absolute inequality in health facility delivery and C-section delivery increased between 2011 and 2014. This finding contradicts previous studies which measured socioeconomic inequality in maternal healthcare services in Bangladesh in the last two decades. For instance, a considerable reduction in inequality in delivery care service outcomes was documented between 1991 and 2011 [9]. A declining trend was also shown in other studies [13, 38] in the last decade. However, these studies also acknowledged that the equity gain in delivery care services over time was not substantial compared to ANC services.

Our results show that inequality measured by the EI increased. This is because of an increase in the probability of delivery care services use between 2011 and 2014 among all the wealth quintiles (Figs 1 and 2). However, the absolute size of this improvement was larger among the richer women compared to their poorer counterparts. As a result, the percentage point changes were greater among the richer women. Thus, it is expected that the EI estimates would increase, since it accounted for absolute changes in the outcome variable. On the other hand, the relative size of the improvement was larger among the poorer, which means that the percentage changes are greater among the poorer.

Previous studies reported that the absolute gap in the utilisation of delivery care services among different socioeconomic groups was widening over time despite an increase in utilisation rate among poorer women [10, 37]. For example. socioeconomic inequality in ANC4+ visits increased between 2011 and 2014 [10, 37]. A recent study projected that existing socioeconomic inequality in delivery care services is most likely to persist until 2030 [14]. This study further cautions that reaching the goal of 80% utilisation of maternal health care services by this time would not be possible despite substantial coverage of maternal health care interventions. In this regard, our findings add to the above concerns about no progress equity gain in delivery care services. However, further studies are required to affirm this conclusion for policymakers.

Our decomposition analysis reveals that wealth of the household and education were the most important factors that explain pro-rich inequality in delivery care services in Bangladesh, which is consistent with the current literature from different countries in South Asia, Middle East and sub-Saharan Africa [16, 41–45]. This result is attributable to the strong and positive association of these variables with delivery care service utilisation as well as pro-rich inequality in these variables. Women from richer households are more willing to pay for these services in the private sector, while poorer women may not be able to even bear the transportation cost to go to a public health facility [42]. The decomposition results also show that household’s wealth status and husband’s education contributed to the increase of socioeconomic inequality in facility delivery during the study period. These findings can be explained by the fact that the association of facility delivery with the women from the richest wealth quintile in 2014 became more string while there was almost no reduction in wealth inequality. Women from richer households were typically married with men with higher education, which contributed to the observed increase in inequality.

ANC visits also played a significant role in explaining the pro-rich inequality in delivery care utilisation, which is consistent with the findings from other studies [16]. However, the role of ANC4+ in explaining the socioeconomic inequality declined in 2014. Because the extent of association between visits to ANC four or more times and delivery care services declined during the period. Living in rural areas explained a significant contribution of pro-rich inequality in health facility delivery and skilled birth attendance as our regression results suggest that women in rural areas had lower utilisation of both services. This could be the result of the higher concentration of poorer women in rural areas and lower use of these services. Poor access to health facilities in rural areas makes the poorer women commute longer to get necessary care in the majority of the developing countries [42].

The interpretation and implications of the findings of this study are subject to a few limitations. For example, the role of supply-side factors in inequality of delivery care services have not been included in the models. This is due to the lack of information about accessibility and the quality of delivery care provision in the BDHS [20]. It could be the case that the utilisation of delivery care services was negatively influenced by greater distance to a health facility as reported in some other studies [46]. Recall bias could also induce measurement error in the outcome variables and our findings could be affected by this problem. However, childbirth is a very momentous life event for women and limiting the analysis to the latest birth would have potentially mitigated this limitation [8]. Another limitation is that decomposition exercise does not provide any causal interpretations with the findings [47], and it is rather an accounting practice to understand the amount of contribution [48].

Our study has important implications for policy. Improving average utilisation is easier than socioeconomic targeting. The government of Bangladesh has taken several policy measures to improve access to maternal healthcare services for reducing maternal and child mortality. These policies have led to improved utilisation of ANC and institutional delivery care services over the last two decades [20]. However, there has been no improvement to reduce socioeconomic inequality in delivery care services in recent years. Therefore, policies should focus on improving the accessibility of maternal health services, especially among the socioeconomically disadvantaged women. Our study has shown that the education of both women and their husbands plays a critical role in explaining inequality in delivery care service utilisation. In this regard, we emphasise that policies for promoting the completion of quality education are important in addressing this growing inequality. Given that income inequality is growing in Bangladesh in the face of rapid economic development in recent times [49], undertaking a redistributive policy reform is imperative to ameliorate inequality maternity care.

Conclusion

This study adds to the literature by presenting robust empirical evidence on socioeconomic inequalities in the utilisation of delivery care services in Bangladesh. Our findings show that absolute inequalities in health facility delivery and C-section delivery increased 2011 and 2014. Compared to the progress in the reduction in socioeconomic inequality delivery care services measured by relative inequality indicator in the last decade, this study finds an increasing inequality measured by absolute inequality indicator in this decade in Bangladesh. Therefore, we emphasise to measure socioeconomic inequality using a robust indicator to present comprehensive evidence to policy makers. Our findings from this paper reinforce that policies need to focus on improving the provision of delivery care services among women from poorer socioeconomic groups. In addition, policy initiatives for promoting the completion of quality education are important to address the stalemate equity gain in the utilization of maternal healthcare services in Bangladesh.

26 May 2020

PONE-D-19-33387

Levels of and changes in socioeconomic inequality in delivery care service in Bangladesh

PLOS ONE

Dear Dr. Chirwa,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Jul 10 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

David Hotchkiss

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. In your Methods section, please provide additional information on how the wealth index was calculated."

3. Our internal editors have looked over your manuscript and determined that it is within the scope of our Health Inequities and Disparities Research Call for Papers. This collection of papers is headed by a team of Guest Editors for PLOS ONE: Clare Bambra, Hans Bosma, Diana Burgess, Joseph Telfair, Barbara Turner, and Jennie Popay. The Collection will encompass a diverse range of research articles on health inequities and disparities.  Additional information can be found on our announcement page: https://collections.plos.org/s/health-inequities

If you would like your manuscript to be considered for this collection, please let us know in your cover letter and we will ensure that your paper is treated as if you were responding to this call. If you would prefer to remove your manuscript from collection consideration, please specify this in the cover letter.

Additional Editor Comments (if provided):

Given that the purpose of the study is to investigate changes in inequality over time, I agree with the second reviewer's comment that the period from 2011 to 2014 is too short and his suggestion to include in your analysis other national surveys carried out prior to 2011.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: I Don't Know

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: No

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Major points:

1. The article’s message hinges on the choice of index (particularly between EI and WI)- while EI shows an increase over the study period the other two indices show a decline. The authors defend EI with their argument but it will be a good idea to mention a critique to their argument, as suggested in the paper below:

Kjellssona & Gerdthama, 2011. Correcting the Concentration Index for Binary Variables. Retrieved from https://project.nek.lu.se/publications/workpap/papers/WP11_4.pdf on April 2, 2020

I have pasted excerpts of the relevant text from pages 19-20:

“If the localization of the threshold within the distribution of the latent variable (compare Figure 5) is due to either arbitrariness or cultural differences, we argue that level independence is a desirable property; given that the level of prevalence is due to an arbitrary threshold, it is sensible if the measured degree of inequalities is invariant to the level of prevalence. ….

By contrast, the threshold for a diagnosed medical condition, e.g. diabetes, is less subjective and there is less variation between contexts. The prevalence of diabetes can therefore be considered to be both accurate and interesting information, and thus an analysis of relative inequalities may be appropriate. In such a case, level independence is not necessarily desirable and, given the normative acknowledgment of the mirror relativity of W, the choice of index should depend on the preferred value judgment.

Accordingly, we can sum up the discussion in a two by two matrix (see Figure 6). If the threshold of the latent variable is arbitrary or subjective, i.e. if there is a risk of reporting heterogeneity, one should consider using E. In turn, if the threshold is objective, the choice ought to depend on the imposed value judgment.”

The three indicators of delivery care belong to an objective category. The two indices differ mainly on the property of level independence and the choice of index is based on the “preferred value judgement”. The latter should be clarified, in explicit terms, in the text.

2. Please clarify the age-distribution in the data-sets. The 2011 Bangladesh DHS surveyed women from ages 13 years onward while 2014 Bangladesh DHS surveyed women from ages 15 years onward. The correction will most likely change the estimates for 2011.

3. The discussion section lacks clarity and needs to be strengthened. A major point of the study was to use EI instead of CI and WI but the authors have not touched on this point at all in the discussion. It would have been a good idea to expand on how the difference in calculations gives more insight into the changes in inequality over time. For example, the major recipients of delivery care in 2014 must have been in the richest 50% population (denominator of EI) thereby showing an increased EI but not in the richest 10% population (denominator of WI) thereby showing a decreased WI.

The study shows an increase in EI over time but this is due to changes in sensitivity to certain predictors and not due to changes in CI of the said predictors (which in fact show a decline). The authors fail to clearly demarcate this and explain or theorize on why this is so.

Additional points below.

Some minor points:

Lines 129-132: The description of health facility delivery: While there are no ‘health posts’ in the variable description of m15_1 in Bangladesh DHS 2011 or 2014, categories like upazila health & family welfare center; other public sector; community clinic; and other ngo sector have not been included in the description. Does it mean that these go into the “0” category?

Line 133: It will be a good idea to detail the categories included as ‘a medically trained professional’.

Line 247: Do the predicted probabilities refer to those obtained following the multivariate logistic regression? Please clarify.

Lines 249-251: That is a sweeping statement and does not apply to skilled care for the 2nd and 4th wealth quintile.

Lines 268-276: I will like to understand why the relative importance of 4+ ANC visits decreased. The same goes for the relative importance of household wealth, women’s education, and husband’s education. What does it signify? Were there some policy changes or some other ground realities that could be behind these findings?

Table 4 also raises the same questions as above. While we observe a reduction in inequality for independent factors (in 2nd & 3rd columns in Tables 4) like 4+ ANC visits, women’s and husband’s education (except for primary education), and in household wealth (except for the poorer), the relative contribution of these has changed in the models predicting the three delivery care indicators. Why? The explanation in “Lines 268-276” suggests substitution but again, the question is why?

Table 5: The above is further corroborated by the findings in Table 5. Here again, the findings suggest that the change in CI is mainly due to higher sensitivity to the parameters. On the other hand, the change in CI contributes negatively meaning that the direction of change for household wealth, women’s and husband’s education is the same as 4+ ANC visits. The discussion section should explain what this signifies.

Lines 324-325 & Lines 334-337: The authors refer to the same article for contradictory views, namely “reduction in inequality of delivery care services” and “increasing socioeconomic inequality”.

Lines 337-339: How did the authors reach this conclusion?

Lines 348-350: Please elaborate on this sentence.

Lines 353-354: 4+ ANC visits actually caused a reduction in CI from 2011 to 2014. When the authors say that it has contributed to inequality, I guess they refer to the positive CI in the 2nd & 3rd columns in Tables 4 (these are independent CI for each predictor for 2011 and 2014). This causes unnecessary confusion because so far, the narrative was about decomposition of inequality explained by a particular predictor within the multivariate model and its change over time.

Lines 387-388: How did the authors come to this conclusion? How and when was need defined?

Reviewer #2: This manuscript was previously submitted to other journal and unfortunately the authors did not follow the comments and suggestion that were made and they decided to submit the manuscript to PLOS ONE. The authors should consider reviewing the comments/suggestions from the reviewers before sending the manuscript to other journal.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Deepali Godha

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: Review of the Manuscript-PONE-D-19-33387.docx

Click here for additional data file.

8 Aug 2020

we have attached the response document

Submitted filename: Review_Response_R1.docx

Click here for additional data file.

4 Sep 2020

PONE-D-19-33387R1

Levels of and changes in socioeconomic inequality in delivery care service: A decomposition analysis using Bangladesh Demographic Health Surveys

PLOS ONE

Dear Dr. Chirwa,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Oct 19 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

David Hotchkiss

Academic Editor

PLOS ONE

Additional Editor Comments (if provided):

The manuscript has been re-reviewed by one of the reviewers that commented on the first version. Note that one of her suggestions -- incorporating the 2018 DHS data into the analysis -- has since been withdrawn, as it was determined that the 2018 dataset is not yet publicly available. However, there are still a few improvements that need to be made to the manuscript before we issue a decision to accept it for publication. Please review the reviewer's suggestions to modify the methodology section to reflect that CI is no longer presented in the study -- as well as the other comments and suggestions. I look forward to receiving a revised version of the manuscript, along with point by point responses.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors have done a good job in explaining some of their results. However, there is still quite a bit of confusion in the corrected version owing to the removal of CI results, taking a stand on EI as opposed to WI and justifying its use, and use of some incorrect words. Also, after reading the other reviewer’s and the editor’s comments, I will like to point out that the 2018 Bangladesh DHS is available. It seems odd that the authors have not used it.

- If authors have removed the CI, then why do they still describe it in Methods section (Line 173 onwards)? Given that the authors have removed all initial results on CI, it does not make sense to then provide a decomposition of CI. If it is decomposition of EI, then table headings need to be corrected.

- It seems that the authors do not think that WI is useful (Lines 189-190 & 198-199). If the authors’ preference is EI (Lines 199-200), the need of explaining and estimating WI seems superfluous. Accordingly, a lot of unnecessary technical detail in the methods section can be removed and an explanation about EI and why it has been chosen should be enough.

- The authors refer to WI as relative inequality in Lines 363-364 while they have earlier explained WI as “WI is neither an absolute nor a relative measure of inequality” (Lines 199-200).

- Should the word be “decomposed” in Lines 185, 188, and 197? These refer to the corrections made to CI for estimation of WI and EI.

- The sentence in Lines 453-455 needs to be rechecked. This is because the declining inequality trend in last decade was probably measured using CI which even now shows a similar trend (though results have been removed). Also, the authors should be careful in using the words equity and inequity as opposed to equality and inequality.

- I will suggest the use of ‘skilled birth attendance’ instead of ‘skilled delivery’.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

19 Oct 2020

attached

Submitted filename: R2_Response to Reviewer.docx

Click here for additional data file.

2 Nov 2020

Levels of and changes in socioeconomic inequality in delivery care service : A decomposition analysis using Bangladesh Demographic Health Surveys

PONE-D-19-33387R2

Dear Dr. Chirwa,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

David Hotchkiss

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

11 Nov 2020

PONE-D-19-33387R2

Levels of and changes in socioeconomic inequality in delivery care service: A decomposition analysis using Bangladesh Demographic Health Surveys

Dear Dr. Chirwa:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. David Hotchkiss

Academic Editor

PLOS ONE