Women Empowerment and Antenatal Care Utilization in Bangladesh

17
:RPHQ (PSRZHUPHQW DQG $QWHQDWDO &DUH 8WLOL]DWLRQ LQ %DQJODGHVK %HOD\HW +RVVDLQ $KPHG $ +RTXH The Journal of Developing Areas, Volume 49, Number 2, Spring 2015, pp. 109-124 (Article) 3XEOLVKHG E\ 7HQQHVVHH 6WDWH 8QLYHUVLW\ &ROOHJH RI %XVLQHVV DOI: 10.1353/jda.2015.0045 For additional information about this article Access provided by Thompson Rivers University (24 Nov 2014 16:22 GMT) http://muse.jhu.edu/journals/jda/summary/v049/49.2.hossain.html

Transcript of Women Empowerment and Antenatal Care Utilization in Bangladesh

n p r nt nd nt n t l r t l z t nn B n l d hB l t H n, h d . H

The Journal of Developing Areas, Volume 49, Number 2, Spring 2015,pp. 109-124 (Article)

P bl h d b T nn t t n v r t ll f B nDOI: 10.1353/jda.2015.0045

For additional information about this article

Access provided by Thompson Rivers University (24 Nov 2014 16:22 GMT)

http://muse.jhu.edu/journals/jda/summary/v049/49.2.hossain.html

T h e J o u r n a l o f D e v e l o p i n g A r e a s

Volume 49 No. 2 Spring 2015

WOMEN EMPOWERMENT AND ANTENATAL

CARE UTILIZATION IN BANGLADESH1

Belayet Hossain

Thompson Rivers University, Canada

Ahmed A. Hoque

University of Victoria, Canada

ABSTRACT

This study attempts to examine the role of women empowerment in the utilization of antenatal care

in Bangladesh. Four dimensions are considered to measure the women empowerment: the highest

level of education, freedom of choice/movement, power in the household decision making process

and involvement in economic activities. Factor analysis technique is employed to construct the last

three dimensions. The probit and the zero-inflated negative binomial regression models are

specified and estimated using the 2011 Bangladesh Demographic and Health Survey data. Results

show that all four dimensions of women empowerment contribute positively and significantly to

the decision and intensity of utilization of antenatal care in Bangladesh. Findings of the study have

a number of policy implications on this issue for a developing country like Bangladesh.

JEL Classifications: O20; I10; C25

Keywords: Women Empowerment, Antenatal care, Probit Model, Zero-Inflated Negative Binomial

Model, Bangladesh

Corresponding Author’s Email Address: [email protected]

INTRODUCTION

The risk of infant and maternal mortalities is often high in most of the developing

countries. Adequate antenatal care is necessary to improve the health status of a country

as it contributes to the reduction of infant and maternal mortalities. Studies (Hong and

Ruiz-Beltran, 2007; Halim, Bohara and Ruan, 2011; Mcdonagh, 1996; and Habibov,

2011) show that antenatal care is negatively related to both infant and maternal

mortalities. Women receiving adequate antenatal care usually have better opportunity to

learn more about their health during pregnancy, child birth and parenting; and have their

health problems diagnosed and treated early. These cares result in higher chances of

normal delivery with better health for both newborn baby and the mother, which may

lead to lower infant and maternal mortalities.

There is a healthy body of literature on the determinants of antenatal care.

Simkhada et al. (2008) and Say and Raine (2007) have done a systematic review and

summarized most of the studies on this issue. The former examine 28 peer-reviewed

studies published since 1990, while the later examine 30 for the same period. These

studies identify and analyze main factors affecting the utilization of antenatal care in

developing countries. The major determinants of antenatal care utilization in developing

countries are divided into four categories: (1) women socio-demographic characteristics

110

(e.g., age, education and age at birth); (2) current pregnancy characteristics (e.g., birth

order, gender of the child, multiple births and wanting the last child or current

pregnancy); (3) affordability (e.g., socio-economic status such as spouse’s education,

household wealth/income and insurance coverage); and (4) accessibility to the services

(e.g., place of residence such as rural/urban and region, travel time to health facilities and

public/private facilities). Other factors such as ethnicity, religion and culture are also

found to be significant determinants of antenatal care utilization.

Women empowerment is believed to be an important factor in determining the

utilization of antenatal care. Duflo (2012) highlights three dimensions of women

empowerment: (i) education; (ii) participation in the decision making process; and (iii)

involvement in economic activities. She refers education as the first and most important

dimension of empowerment followed by other two.

Higher the level of education of a woman, better the chance that she would be

enlightened and be more conscious; and therefore can make better decisions for herself as

well as for her family. Educated women can easily comprehend the benefits of antenatal

care and thus expected to utilize antenatal care more effectively. They may also be better

informed about how the healthcare system functions and hence can get better access to it.

Furthermore, educated women can easily interact and communicate with the health

service providers to get required cares (Bloom et al., 2001).

In many developing countries, where a man is typically the head of a household

makes most of the household decisions and the wife experiences no equity in this process

(Holland and Hogg, 2001). For instance, men often control the financial matters no

matter who earns the cash and make decisions regarding major household purchases. This

translates into financial vulnerability and lack of autonomy of a woman and more

dependent on her spouse. In such cases, a woman who needs antenatal care has to rely on

her husband’s judgment and willingness to spend money on such cares. So, we

hypothesize that the higher the participation in the household decisions making process

by a woman, the higher is the likelihood that she would utilize the antenatal care in the

developing countries.

In a developing country like Bangladesh, the opportunity of women involvement

in economic activities is limited. Microcredit organizations increase the scope of being

involved in economic activities for women as most of their members are women in

Bangladesh (Microcredit Regulatory Authority, 2013). When a woman is involved in

economic activities outside home, it improves the financial stability of the family and

uplifts her social status and self-confidence. It directly contributes to her power in any

household decision making process, including the decision to receive health care. It

makes sense to presume that such a woman would receive more antenatal care.

Several studies (Adamu and Salihu, 2002; Pallikadavath et al., 2004; Matsumura

and Gubhaju, 2001 and Mumtaz and Salway, 2005) suggest that the lack of women

autonomy is one of the reasons for under-utilization of the antenatal care. They show that

women autonomy is positively related to the antenatal care utilization. Ahmed et al.

(2010) find that economic status, education and empowerment (called as the 3Es) are

significantly associated with the utilization of maternal health services. They compare

two groups of women for each factor using data from 33 developing countries and

demonstrate inequality in maternal healthcare services utilization. However, the study

shows only the association between factors not the causality - it controls neither the other

111

the relevant variables nor the heterogeneity across countries. Most of these studies are

either regional in nature or fail to address the women empowerment broadly using

representative national level data The present study is an attempt to examine the role of

different dimensions of women empowerment on the antenatal care utilization in

Bangladesh to void the gap in the literature.

Over the last thirty years, Bangladesh has emerged as one of the success stories

worldwide in healthcare achievements despite of high population density, low per capital

income, widespread poverty and low per capita expenditure on healthcare, corruption and

political fragmentation. Since 1980, the maternal mortality has dropped by more than 75

percent, while the infant mortality has dropped by more than half since 19902. Maternal

and child under five mortality rates are estimated to be 2.4 and 48 per 1000 in 2010

respectively (WHO, 2013). As a result, life expectancy at birth climbs to 70 years in 2011

(WHO, 2013). Women empowerment is considered as one of the key factors behind the

success of healthcare in Bangladesh. Appropriate education policy targeting girls and

access to microcredits institutions which provide credit, healthcare services and education

have increased the women empowerment remarkably in Bangladesh (World Bank, 2013).

Given the prevailing situation in Bangladesh, it is of interest to examine the role of

women empowerment in the utilization of antenatal care.

The reminder of the paper is organized as follows. The next section presents a

discussion on data and methodology. Results and Discussion section presents the findings

and their interpretations of the study and concluding remarks and policy

recommendations are made in the Summary and Policy Implications section.

METHODOLOGY AND DATA

Data and Weights

The data used in this study are from the 2011 Bangladesh Demographic and Health

Survey (BDHS), a nationally representative sample survey. The 2011 BDHS contains

information of all ever-married women aged 12-49 residing in the selected households. A

sample of size 7,318 women, those who have given birth in last five years has been

selected. The BDHS provides useful statistics that can be used to determine changes in

key areas of development in Bangladesh including maternal and child health, women

involvement in economic activities, education and poverty reduction (see, NIPORT et al.,

2013). The survey is conducted through the collaboration between the National Institute

of Population Research and Training (NIPORT) of the Ministry of Health and Family

Welfare, ICF International/Macro International/ORC Macro, and Mitra and Associates.

The BDHS samples are collected using stratified multi-stage cluster sampling.

Hence the sample data is not an identical and independently distributed (iid) sample and

observations are selected using non-simple random sampling technique. Non-simple

random sampling, which is well known as complex survey sampling, consists of various

probabilities of selections at different levels. The weight assigned to each observation is

inversely proportional to the probability of selection. Sampling weights that come with

the survey data are used in estimation, instead of simple random sampling weight. The

weight series is rescaled following the BDHS manual and normalized such that the sum

of all normalized weights is equal to one.

112

Measuring Women Empowerment

The BDHS contains a number of variables that can be used to measure different

dimensions of women empowerment. Besides education, other dimensions of

empowerment are constructed using factor analysis technique. In the survey, respondents

were asked about who makes particular decisions, e.g., major household purchases,

respondent’s own healthcare and use of contraception. Four options were offered to each

question in the survey: a) respondent alone; b) respondent and spouse jointly; c) husband

alone; and d) someone else. Binary response variables are created by merging categories

a) and b) to be “1” to denote that the respondent participates in decision making; and c)

and d) to be “0” to denote that the respondent does not participate in the decision making

process. Nine variables included in the factor analysis are: choosing contraceptive

method, deciding the use of contraception, decision about respondent’s own health,

participation in major household purchases, visiting families and relatives, involvement

in microcredit organizations, decision about child healthcare, freedom of going to heath

center alone, and freedom of going to health center with a young child.

Using the principal component analysis for the above variables, three factors are

extracted that represent three dimensions of women empowerment. These dimensions are

named as freedom of choice and movement, decision making power in the household and

involvement in economic activities outside home. Table 1 presents the summary of factor

analysis, where three factors together account for more than 63 percent of the variations

in the sample data. Adding respondent’s level of education to these three factors, we

have four indicators to measure four dimensions of women empowerment.

TABLE 1. FACTOR ANALYSIS RESULTS FOR WOMEN EMPOWERMENT

VARIABLES

Factor/Dimension Eigen

value

Explained

Variance

(in proportion)

Cumulative

proportion

of variance

Factor 1:

Freedom of choice/ movement

Factor 2:

Power in household decisions

Factor 3:

Involve in economic activity

2.89

1.79

1.03

0.3026

0.2123

0.1198

0.3026

0.5149

0.6347

Models and Variables

Two models have been used to examine the influence of women empowerment along

with other variables that determine the decision of receiving and frequency of antenatal

care. A discrete choice model, with the dependent variable being a binary outcome of

whether a woman, who has given a birth in last five years has received any antenatal care

from health professionals or not, can be estimated by logit or probit regression model.

This study uses a probit model, as given below, to estimate the probability of receiving

antenatal care:

113

( | ) ∫ ( )

( )

where is the probability of receiving antenatal care by a woman i. if antenatal

care is received from health professionals and , otherwise.

To examine the impact of women empowerment on the frequency of antenatal

care visits, the Zero-Inflated Negative Binomial (ZINB) regression model has been

specified. In ZINB model the count response variable is with

probability for zero count and ( ) with probability

( ) for non-zero cunt so that

( ) ( )( )

( ) ( ) (

)

( ) (

)( )

where is the mean of the underlying negative binomial distribution and is the over-

dispersed parameter. The parameter is modeled as = exp( ) where is the

( ) vector of unknown parameters associated with the known covariates

( ) where is the number of covariates.

The ZINB model appears to be appropriate when dependent variable is a count

variable, number of nonnegative counts of an event has extra-Poisson variation (over-

dispersion)3 and there exists excessive zero values of dependent variable (Yau, Wang and

Lee, 2003). The distribution of the frequency of antenatal care visits (the dependent

variable) as presented in Figure 1 shows a high zero value frequency, which supports the

use of ZINB model4.

In our model, there are four categories of covariates. First category includes four

dimensions of women empowerment (highest level of education, power in household

decision making process, freedom of choice/movement and involvement in economic

activities). The specification of all variables included in empowerment category is

presented in Table 2.

In second category of covariates, individual characteristics of the respondent other

than education, such as age of the respondent, if pregnancy wanted, experience from

previous child birth and age at first birth are included. Third category includes household

characteristics, such as household’s economic condition, spouse’s level of education and

occupation, and if the household has a television or/and radio to access information.

Geographical location (e.g., town, city or country side) of residence of the respondent is

included in the final category. Table 3 presents the list of all explanatory variables and

their specifications under category 2, 3 and 4.

114

FIGURE 1: DISTRIBUTION OF THE ANTENATAL CARE VISITS

Source: Bangladesh Demographic and Health Survey, 2011.

The estimated coefficients of probit and ZINB models provide the signs of the

partial effects of each of the explanatory variables on the response probability

(Wooldridge 2002, p. 465). For instance, the sign of an estimated coefficient in probit

model determines whether the variable has a positive or negative effect on the probability

of receiving antenatal care of by a woman. But to get the magnitude of the impact, the

marginal effect has to be calculated. Using STATA software commands we estimate

marginal effects of all covariates separately for both models.

RESULTS AND DISCUSSIONS

Descriptive Statistics

Table 2 presents descriptive statistics of explanatory variables related to women

empowerment, while descriptive statistics of other explanatory variables are presented in

Table 3. Sixty seven percent of the total respondents (7,318) receive antenatal care from

the health professionals with a mean visit of 3.59. The WHO guideline (WHO, 2003)

recommends that every expecting mother should have at least four antenatal care visits

per pregnancy period. The survey data (NIPORT, 2013) shows that about 40% of the

pregnant women who received antenatal care have four or more visits with a mean of

6.06. Other 60% have visits between 1 and 3 with a mean of 1.96.

The utilization of antenatal care increases significantly with the increase in

respondent’s level of education. Among the respondents, 30% have primary, 43% have

secondary and 8.5% have post-secondary education. Fifty eight percent of the

respondents having primary education received such cares with a mean visit of 2.95. On

the other hand, 96% of the respondents having post-secondary education received

antenatal care with mean of 5.5 visits. Similarly, empowered women, when measured by

other indices also show higher utilization of antenatal care (Table 3).

Of the total respondents, 62%, 21% and 3% belong to 20-29, 30-39 and 40-49

age groups respectively. The percentage of respondents receiving antenatal care as well

01

02

03

04

0

Pe

rcen

t

0 5 10 15 20antcarefr

Number of Antenatal Care Visits

Mean: 2.39

Variance: 7.22

115

TABLE 2. SPECIFICATION AND DESCRIPTIVE STATISTICS OF VARIABLES

MEASURING WOMEN EMPOWERMENT

Variables

Description of variable

No. of

Obs.

(%)

Women with

any ant. care

Women with

1-3 ant. care

visits

Women with

4 ant. care

visits

No. of

women

(%)

Aver.

visits

No. of

women

(%)

Aver.

visits

No. of

women

(%)

Aver.

visit

Number of women received

antenatal care

4873

(67%)

4873

(67%)

3.59 2940

(60%)

1.96 1933

(40%)

6.06

Dimension 1: Education

Level of Education: Respondent’s highest level of education (default: no education)

Primary

If the highest level education is primary then 1, else 0

2191 (29.9%)

1265 (58%)

2.95 898

(41%) 1.87

367 (17%)

5.6

Secondary If the highest level education

is secondary then 1, else 0

3171

(43.3%)

2475

(78%) 3.64

1450

(46%) 2.01

1025

(32%)

5.94 Post-

Secondary

If the highest level education is

post-secondary then 1, else 0 624

(8.5%)

601

(96%) 5.5

179

(29%) 2.27

422

(68%)

6.86

Other dimensions of women empowerment: Contraception

Decision

If respondent has say on the

use of contraception then

1,else 0

4489 (61.3%)

3082 (69%)

3.67 1833

(41%) 2

1249 (28%)

6.12

Contraception

Method

If respondent can 4choose a

method then 1, else 0

4875

(66.6%)

3339

(68%) 3.66

1991

(.41%) 2

1348

(27%)

6.11

Own Health If respondent has say about seeking own healthcare then 1,

else 0 otherwise

4460

(61%)

3065

(69%) 3.71

1795

(40%) 2

1270

(29%)

6.11

Major HH Purchase

If respondent has say on major household purchase then 1,

else 0

4062

(55.5%)

2795

(69%) 3.78

1595

(39%) 2

1200

(30%

6.15

Visiting Families

If respondent has freedom over visiting family members

then 1, else 0

4282

(58.5%)

2958

(69%) 3.77

1707

(40%) 2

1251

(29%)

6.18

Children

Health

If respondent has say about

seeking children’s healthcare

then 1, else 0

5098 (70%)

3499 (68%)

3.74 2030

(40%) 2

1469 (29%)

6.15

Going Alone to

Health Centre

If respondent has freedom to

go to health centre alone then

1, else 0

4637 (63%)

3161 (68%)

3.68 1849

(40%) 1.98

1312 (28%)

6.1

Going to

Health Centre with a Young

Child

If respondent has freedom to

go to health centre with a young child then 1, else 0

2458 (33.6%)

1664 (68%)

3.53 1022

(42%) 1.93

642 (26%)

6.1

Association with Micro-

Credit Inst.

If respondent associates with microcredit institutions then 1,

else 0

2555

(34.9%)

1632

(64%) 3.43

1008

(39%) 1.97

624

(24%)

5.78

Source: Bangladesh Demographic and Health Survey, 2011.

116

TABLE 3. SPECIFICATION AND SUMMARY STATISTICS OF OTHER

VARIABLES IN THE STUDY

Variables

Description of variable

No. of

women

(%)

Women with any ant. care

Women with 1-3 ant. care

visits

Women with

4 ant. care visits

No. of

women (%)

Aver.

visits

No. of

women (%)

Aver.

visits

No. of

women (%)

Aver.

visits

Respondent’s other characteristics

Age group

Age1:

If respondent belongs to age group 20-29 then 1; else 0

(Default group: 12-19)

4523

(62%)

3095

(68%) 5.58

1963

(41%) 1.98

1232

(27%) 6.0

Age2 If respondent belongs to age group 30-39 then 1; else 0

1544 (21%)

936 (61%)

3.94 544

(35%) 1.96

392 (26%)

6.45

Age3 If respondent belongs to age

group 40-49 then 1; else 0

218

(3%)

95

(44%) 2.91

72

(33%) 1.81

23

(11%) 6.35

Pregnancy

wanted

If pregnancy wanted then 1;

else 0

5188

(71%)

3585

(69%) 3.65

2121

(41%) 1.98

1464

(28%) 6.07

No. of births in last 5 years

A count variable Mean: 1.2; Sd=0.43; Max 4 and Min 1

Age at first

birth

Respondent’s age at first birth:

continuous variable Mean: 18.084; Sd=3.29; Max 40 years and Min. 12 years

Household Characteristics

Poor

If a respondent belongs to

wealth index 2 then 1; else 0 (default: wealth index 1,

poorest)

1408

(19%)

924

(66%)

3.09

625

(44%)

1.88

299

(21%)

5.61

Middle If respondent belongs to wealth index group 3 then 1;

else 0

1473

(20%)

1122

(76%) 3.45

697

(47%) 2

425

(29%) 5.83

Rich

If respondent belongs to

wealth index group 4 then 1;

else 0

1513

(21%)

1382

(91%) 4.85

542

(36%) 2.17

840

(56%) 6.58

Spouse education

A continuous variable indicating years of schooling

5.45; Sd=4.68; Max18 and Min 0

Spouse

occupation

If spouse occupation is agri.

then 1; else 0

5391

(73.7%)

3860

(72%) 3.79

2200

(41%) 2

1660

(31%) 6.16

Access to

information

If household has TV and/or

Radio then 1; else 0

2978

(41%)

2385

(81%) 4.1

1223

(41%) 2.02

1162

(40%) 6.28

Religion If respondent’s religion is Islam then 1; else 0

6594 (90%)

4335 (66%)

3.54 2653

(40%) 1.96

1682 (26%)

6.03

Geographical location of residence

Place of residence (Default: country side/rural)

Town If respondent resides in small

city or town then 1; else 0

1573

(21.5%)

1233

(78%) 4.03

643

(41%) 2.07

590

(37%) 6.16

City If respondent resides in city then 1; else 0

751 (10.3%)

649 (86%)

4.9 240

(32%) 2.07

409 (54%)

6.56

Source: Bangladesh Demographic and Health Survey, 2011.

117

as the mean number of visits, decreases as the level of age group increases (Table 3). On

the other hand, the percent and mean number of visits gradually increases with the

increase in the wealth level. For example, 66% of the respondents belong to wealth index

2 receives antenatal care with a mean visits of 3.09, whereas for the wealth index 4, it is

91% with a mean visits of 4.85. More respondents residing in towns or cities receive

antenatal care with a higher mean number of visits compared to respondents residing in

country sides or rural areas (Table 3).

Results of the Probit Model

The probit model is estimated to determine how different indicators of women

empowerment influence the decision of receiving antenatal care. The results of the probit

model are presented in column 2 of Table 4. The Wald Statistic suggests that the data fit

the model very well. The signs of the coefficients of all explanatory variables meet our

prior expectations. Given that the probit regression model provides only the signs of the

coefficients, marginal effects of all covariates are estimated separately to observe the

magnitudes of the impacts. Marginal effects of the probit model are shown in column 3

of Table 4. Of the 21 explanatory variables, 15 are significant at 1% level, three are at 5%

level and three are at 10% level (column 2, Table 4). All factors representing four

dimensions of women empowerment used in the study are found positive and statistically

significant. Note that three categorical variables are used to see the effect of education on

the utilization of antenatal care. The marginal effects of primary, secondary and post-

secondary level of education are 0.086, 0.173 and 0.310 respectively (column 3, Table 4).

The marginal effect of 0.173 for the secondary education level, for instance, indicates that

respondents having secondary level of education have a 17% higher likelihood of

receiving antenatal care by health professionals compared to those who have no

education or other levels of education. Similarly, the likelihood of receiving antenatal

care is 31% higher for those who have post-secondary level of education compared to

others. It also indicates that a respondent with the post-secondary level of education has a

14% higher chance of receiving such cares compared to one who has secondary level of

education.

The estimated marginal effects of the variables representing the participation in

the household decision making process, freedom of choice/movement and involvement in

economic activities are 0.025, 0.025 and 0.022 respectively (column 3, Table 4). These

results imply that respondents, who participate in household decision making process,

have about three percent higher chance of receiving antenatal care compared to those,

who do not participate in such activities. Similarly those, who have more freedom of

choice in using contraception or going out alone have three percent and those who are

involve in economic activities have two percent higher chances of receiving antenatal

care. Note that the marginal effects of these three dimensions of women empowerment

are not very large individually. But if the combined effect of all four dimensions of

women empowerment is calculated, it is significantly large. For instance, an empowered

woman, defined by having secondary level of education, participates in household

decision making process, has freedom to choose her contraception use and method, and is

involved in economic activities, has a 24% percent higher likelihood to receive antenatal

care compared to one who has no education, and is not involved in those activities. Thus,

118

all four dimensions of women empowerment under study play a vital role in receiving

antenatal care in Bangladesh. Among the variables representing respondent’s other

characteristics, age and numbers of births in last five years are negatively related to

receiving antenatal care. Three dummy variables are specified to represent respondents’

age groups. The marginal effects of all age groups are negative and they increase

successively in absolute value (column 3, Table 4), meaning a respondent belongs to a

higher age group is less likely to receive antenatal care.

It is believed that the older women are more experienced and utilize their

experience to stay well during their pregnancy periods instead of going out to receive

antenatal care from the health professionals. Similarly, if a respondent has given multiple

births in the last five years, she is less likely to receive antenatal care. On the other hand,

the marginal effect of the variable indicating age at first birth is positive. When a woman

becomes pregnant at older age, she is expecting more complications and thus she will be

more inclined to receive antenatal care.

Among household characteristics, the marginal effects of spouse’s level of

education, wealth and access to information (TV/Radio) are positive, as expected. The

spouse’s level of education is measured in number of years of schooling. So, the

estimated marginal effect of this variable 0.007 indicates that the likelihood of receiving

antenatal care rises by 0.7% for the woman for every additional year of schooling of her

spouse.

Economic status, represented by three wealth indices, is an important predictor

of receiving antenatal care. The marginal effects of all three wealth indices increase

successively, implying that higher the level of wealth the more likely to receive the

antenatal care. Access to information is represented by owning TV and/or Radio. The

marginal effect of this variable is significant at 10% level. This may be because of the

multicollinearity issue as the BDHS constructs the wealth index taking TV/Radio as one

of the financial assets into account. The marginal effects of agriculture being spouse

occupation and household being Muslim are negative. Women belong to Muslim families

are expected to be more conservative and less likely to go outside to receive antenatal

care from health professionals.

In Bangladesh, the accessibility to health care depends on the geographical

location as well, as more healthcare facilities are located in towns and cities compared to

the country sides. Thus, location of a household influences the probability of using

antenatal care. Two dummy variables for town and city are specified to represent three

locations. The marginal effects of these two dummies are positive and statistically

significant suggesting that a respondent who lives in a city has the largest probability of

receiving antenatal care than one who lives in a town or a country side.

Results of the Zero-Inflated Negative Binomial (ZINB) Regression Model

Upon estimating the probit model, which predicts the probability of receiving antenatal

care, we examine the intensity of the use of antenatal care in Bangladesh based on the

same set of explanatory variables. Intensity of care is measured by the frequency of

antenatal care. The ZINB model is estimated to examine the effects of women

empowerment on the frequency of using antenatal care. The estimated coefficients and

marginal effects of covariates in the ZINB are presented in columns 4 and 6 of Table 4

119

respectively. The data fits the negative binomial regression model even better than that of

the probit model as shown by the Wald Statistic. The Vuong test shows that ZINB model

is more appropriate than standard Negative Binomial regression model. All explanatory

variables of the ZINB model meet a priori expectation. Out of 21 variables, 11 are

statistically significant at less than 1% level, two are at 5% level, three are at 10% level

and five are not significant at an acceptable level. The marginal effects of all women

empowering variables are positive and have large magnitudes, reflecting the importance

of women empowerment on the intensity of using antenatal care. The marginal effects of

three dummy variables representing respondent’s level of education which are 0.42, 0.89

and 1.95 respectively (column 6, Table 4), suggest that the impact of the level of

education on the frequency of visiting health professionals for antenatal care. For

instance, the marginal effect of 1.95 for post-secondary level of education implies that a

respondent having post-secondary level of education is expected to have on average two

more visits compared to one who does not have the post-secondary level of education.

We also observe that that the respondent’s level of education is the largest contributing

factor in explaining the intensity of receiving antenatal care in Bangladesh.

Positive marginal effects of 0.21 and 0.20 suggest that women who participate in

the household decision making and are involved in economic activities respectively are

expected to visit health professionals more often. However, the coefficient with the

variable representing the freedom of choice and movement is found not significant at an

acceptable level. This may be because of multicollinearity problem. Thus, most of the

women empowering factors appear to play an important role in determining the

frequency of antenatal care visit. Further, an empowered woman having post-secondary

education, who participates in household decision making process and is involved in

economic activities is expected to have at least two more (2.3) antenatal care visits

compared to the woman having no empowering factors. Note that the coefficients of all

of the variables representing women empowerment in the inflated model are negative;

indicating that an empowered woman is less likely to have zero antenatal care visits.

Among other characteristics of a respondent, age and number of births in the last

five years negatively influence the frequency of visiting health professionals. The

marginal effects of the age groups suggest that as age of a respondent increases the

number of visits to a health professional declines. This finding is consistent with the

previously found lower probability of receiving antenatal care by an older woman.

Similar conclusion can be drawn for women, who have given multiple births in the last

five years. It is also found that when a respondent’s age at first birth is higher and

pregnancy is wanted, the expected number of visits rises. However, the marginal effect of

wanted pregnancy is not significant at an acceptable level. The economic condition of a

respondent’s household measured by the wealth index is the most important factor

determining the frequency of antenatal care visits. All three indices of wealth have

positive impacts on the intensity of visits and increase when the economic condition of

the household improves. Spouse’s level of education also influences such intensity

positively. The marginal effect of TV/Radio is positive which implies that a respondent

having more information is expected to make more antenatal care visits. On the other

hand, agriculture as being spouse occupation and the respondent being a Muslim

negatively affect the intensity. These findings are plausible. Because, for

120

TABLE 4. RESULTS OF THE PROBIT AND ZERO-INFLATED

NEGATIVE BINOMIAL REGRESSION MODELS

Covariates Probit Model Zero-Inflated Negative Binomial Model4

Coefficient Marginal

effect

Coefficient of

ZINB

Model

Coefficient of

Inflated model Marginal effect

of ZINB model

Women Empowerment

Education (Default no education)

Primary 0.282 (0.053)*** 0.086

(0.017)*** 0.117

(0.060)** -0.476

(0.126)***

0.425

(0.115)***

Secondary 0.581 (0.060)*** 0.173

(0.017)*** 0.231

(0.059)*** -1.109

(0.157***

0.887

(0.122)***

Post-secondary 1.142 (0.131)*** 0.310

(0.027)*** 0.415

(0.074)*** -5.211

(9.523)

1.946

(0.443)***

Power in HH decision 0.080

(0.038)** 0.025

(0.012)** 0.091

(0.029)*** -0.022

(0.100)

0.211

(0.062)***

Freedom of

choice/movement

0.082

(0.040)** 0.025

(0.013)** 0.003

(0.030)

-0.208

(0.101)**

0.078

(0.065)

Involvement in economic

activities

0.071

(0.040)* 0.022

(0.013)* 0.076

(0.032)*** -0.084

(0.102)

0.199

(0.068)***

Individual Characteristics

Age2:20-29 group -0.193

(0.056)*** -0.059

(0.017)*** -0.043

(0.039)

0.405

(0.160)***

-0.236

(0.090)***

Age3: 30-39 group -0.339

(0.070)*** -0.104

(0.021)*** -0.095

(0.052)* 0.678

(0.188)***

-0.444

(0.119)***

Age4: 40-49 group -0.552

(0.130)*** -0.165

(0.037)*** -0.382

(0.119)*** 0.817

(0.330)***

-1.155

(0.272)***

Pregnancy wanted 0.073

(0.042)* 0.022

(0.013)* -0.010

(0.033)

-0.146

(0.108)

0.028

(0.071)

Birth in last 5 years -0.160

(0.044)*** -0.049

(0.013)*** -0.060

(0.036)* 0.289

(0.105)***

-0.232

(0.086)***

Age at first birth 0.023 (0.007)*** 0.007

(0.002)*** 0.014

(0.005)***

-0.027

(0.017)

0.041

(0.010)***

HH characteristics

Wealth index2 0.140 (0.053)*** 0.043

(0.016)*** -0.018

(0.047)

-0.343

(0.138)***

0.080

(0.096)

Wealth index3 0.350 (0.063)*** 0.107

(0.019)*** 0.063

(0.050)

-0.835

(0.190)***

0.423

(0.107)***

Wealth index4 0.728 (0.085)*** 0.213

(0.022)*** 0.260

(0.056)*** -1.776

(0.346)***

1.140

(0.149)***

Spouse education 0.022 (0.006)*** 0.007

(0.002)*** 0.019

(0.004)*** -0.029

(0.016)*

0.052

(0.010)***

Agri. being spouse

occupation

-0.10

(0.043)** -0.031

(0.013)** -0.070

(0.040)* 0.123

(0.111)

-0.198

(0.084)**

TV/Radio 0.079

(0.049)* 0.024

(0.015)* 0.067

(0.035)** -0.075

(0.140)

0.176

(0.077)**

Muslim -0.161

(0.064)*** -0.050

(0.019)*** -0.105

(0.042)*** 0.272

(0.172)

-0.328

(0.104)***

Geographic location

Town 0.180 (0.054)*** 0.055

(0.017)*** 0.144

(0.035)*** -0.302

(0.152)**

0.425

(0.085)***

City 0.218 (0.090)*** 0.067

(0.027)*** 0.247

(0.041)*** -0.203

(0.265)

0.626

(0.125)***

Wald Statistics 996.32 870.74

Vuong Test

14.9

(0.000)

Sample 7307 7318

Notes: Robust standard errors are reported in the parenthesis ***, ** and * denote that coefficient/marginal effects are significant at 1%, 5% and 10% level of

significant respectively

121

instance, if a respondent belongs to a Muslim family she is expected to visit less for

antenatal care as Muslim women in general are more conservative. The marginal effects

of two dummy variables (town and city) are positive and their impact increases

successively, suggesting that respondents living in towns are likely to make more visits

compared to those who live in country sides. Similarly, respondents living in cities are

likely to have more visits compared to their counterparts who live in towns.

SUMMARY AND POLICY IMPLICATIONS

Bangladesh has achieved a remarkable success in health, evidenced by the fall in

maternal and infant mortalities and the rise in life expectancy. Antenatal care plays a key

role in reducing maternal and infant mortalities in Bangladesh. This study examines the

role of women empowerment in the utilization of antenatal care in Bangladesh using the

2011 Bangladesh Demographic and Health Survey (BDHS) – a nationally representative

survey sample. Following Duflo (2012) and using rotating factor analysis technique, four

dimensions of women empowerment have been constructed to examine their effects on

antenatal care utilization. They are (i) women’s highest level of education; (ii) decision

making power in the household; (iii) freedom of choice and movement; and (iv)

involvement in economic activities. Hypotheses have been tested to determine if these

dimensions of women empowerment contribute to the utilization of antenatal care in

Bangladesh.

The probit and the zero-inflated negative binomial regression models are

specified to see if women empowerment along with other variables plays any significant

role in the utilization of antenatal care. The probit model is used to determine factors

affecting the decision of receiving antenatal care, whereas the zero-inflated negative

binomial regression model is employed to determine the intensity of receiving such cares.

The intensity of care is measured by the frequency of receiving antenatal care.

The data fit both models very well. All explanatory variables in both models

meet our prior expectations. The results show that each dimension of the women

empowerment contributes positively to both the decision and frequency of using

antenatal care in Bangladesh. Of four dimensions, women education is found as the most

dominating factor in the utilization of antenatal care. This finding suggests that with an

increase in the level of empowerment through higher level of education, the likelihood of

antenatal care utilization increases substantially. The other three dimensions of

empowerment also contribute to the utilization of antenatal care positively.

Other important factors affecting the utilization of antenatal care are economic

condition (measured by wealth), spouse’s level of education, age of the respondent,

religion, location of the household and access to TV/Radio. With the increase in the level

of wealth which enhance the affordability, utilization of antenatal care increases.

Spouse’s level of education helps to realize the need for antenatal care. Accordingly, the

utilization of antenatal care rises with the rise in the level of spouse’s level of education.

Location of the household determines the accessibility of the healthcare facilities. More

healthcare facilities are available in cities and towns compared to country sides, therefore

respondents living in cities and towns are utilizing more antenatal care than those who are

living in the country sides. Older women are utilizing relatively less antenatal care

compared to younger women. It is believed that the older women are more experienced

122

and they feel that they need less antenatal care provided by health professionals. Owning

TV/Radio provides information and exposures, which may enlighten the respondents to

utilize more antenatal care.

The study has number of policy implications for all developing countries like

Bangladesh. First, women empowerment is one of the key factors in the utilization of

antenatal care, which subsequently contributes to the improvement of health status. All

efforts should be made to increase the women empowerment, particularly the girls’

education. Increase in women involvement in economic activities increases women

empowerment. For a developing country like Bangladesh, scopes and opportunities for

women employment are limited. The experience from Bangladesh shows that microcredit

institutions, which not only provide credits to members but also provide access to

information, education and health services, play an important role in increasing women

involvement in the economic activities.

Secondly, poverty emerges as one of the foremost obstacles in the utilization of

antenatal care. Respondents belong to poorer households cannot afford enough antenatal

care. Poverty eradication policy should be pursued effectively. In the short-run, special

programs targeting the poorest segment of the community should be undertaken to

provide adequate antenatal care. Thirdly, the findings of the study show that there exists a

disparity between rural and urban areas in antenatal care services utilization. To address

this disparity, the resource allocation should be taken into account based on the regional

need and degree of inequality. Finally, the study shows that TV/Radio can influence the

utilization of antenatal care, particularly the frequency of antenatal visits. This suggests

that more educated and informative programs and/or advertisement should be

broadcasted through TV/Radio aiming the target groups.

ENDNOTES

1 An earlier version of the paper was presented at the AABSS (Australian Academy of Business

and Social Sciences) conference held in Kuala Lumpur, Malaysia, August 25-26, 2014

2 The Guardian (2013) “Global Development”, http://www.theguardian.com/global-

development/2013/nov/21/bangladesh-healthcare-poverty-lancet-study. November 21, 2013.

Retrieved on February 20, 2014

3 The Negative Binomial model can accommodate overdispersion but not underdsipersion with

respect to the Poission model.

4 The ZINB assumes there are two distinct data generation processes. The result of a Bernoulli trial

is used to determine which of the two processes is used. For observation i, with probability the

only possible response of the first process is zero counts, and with probability of ( ) the

response of the second process is governed by a negative binomial with mean, . The zero counts

are generated from, both the first and second processes, where a probability is estimated for

whether zero counts are from the first or the second process. The overall probability of zero counts

is the combined probability of zeros from the two processes.

123

REFERENCES

Adamu, Y. M., and H. M. Salihu (2002)."Barriers to the use of antenatal and

obstetric care services in rural Kano, Nigeria." Journal of Obstetrics & Gynecology

22(6): 600-603.

Ahmed, Saifuddin, et al. (2010). "Economic status, education and

empowerment: implications for maternal health service utilization in developing

countries." PloS one 5(6): e11190.

Bloom, Shelah S., David Wypij, and Monica Das Gupta (2001). "Dimensions of

women’s autonomy and the influence on maternal health care utilization in a north Indian

city." Demography 38(1): 67-78.

Duflo, Esther (2012). “Women Empowerment and Economic Development”,

Journal of Economic Literature, 50(4):1051-1079.

The Guardian (2013) “Global Development”,

http://www.theguardian.com/global-development/2013/nov/21/bangladesh-healthcare-

poverty-lancet-study, Retrieved on February 20, 2014

Habibov, Nazim N. (2011). "On the socio-economic determinants of antenatal

care utilization in Azerbaijan: Evidence and policy implications for reforms." Health

Economics, Policy and Law 6(2) :175-203.

Halim, Nafisa, Alok K. Bohara, and Xiaomin Ruan (2011). "Healthy mothers,

healthy children: does maternal demand for antenatal care matter for child health in

Nepal?" Health policy and planning 26(3):242-256.

Holland, Karen, and Christine Hogg (2010). Cultural awareness in nursing and

health care: an introductory text. London: Hodder Arnold.

Hong, Rathavuth, and Martin Ruiz-Beltran (2007). "Impact of prenatal care on

infant survival in Bangladesh." Maternal and child health journal 11(2):199-206.

Matsumura, Masaki, and Bina Gubhaju (2001). "Women's Status, Household

Structure and the Utilization of Maternal Health Services in Nepal: Even primary-leve1

education can significantly increase the chances of a woman using maternal health care

from a modem health facility." Asia-Pacific Population Journal 16(1): 23-44.

McDonagh, Marilyn (1996)."Is antenatal care effective in reducing maternal

morbidity and mortality?." Health policy and planning 11(1):1-15.

Microcredit Regulatory Authority (2013). “Microcredit in Bangladesh”,

http://www.mra.gov.bd/index.php?option=com_content&view=category& layout=blog

&id=29&Itemid=80, retrieved on February 7, 2014.

Mumtaz, Zubia, and Sarah Salway (2005). "‘I never go anywhere’: extricating

the links between women's mobility and uptake of reproductive health services in

Pakistan." Social science & medicine 60(8):1751-1765.

National Institute of Population Research and Training (NIPORT) et. al. (2013),

2011 Bangladesh Demographic and Health Survey (BDS) 2011, Dhaka, Bangladesh and

Calverton, Maryland (USA): National Institute of Population Research and Training,

Mitra Associates and ICF International.

Pallikadavath, Saseendran et. al. (2004). "Antenatal care: provision and

inequality in rural north India." Social Science & Medicine 59(6):1147-1158.

124

Say, Lale, and Rosalind Raine (2007). "A systematic review of inequalities in

the use of maternal health care in developing countries: examining the scale of the

problem and the importance of context." Bulletin of the World Health Organization

85(10):812-819.

Simkhada, Bibha, et al. (2008). "Factors affecting the utilization of antenatal

care in developing countries: systematic review of the literature." Journal of advanced

nursing 61(3): 244-260.

World Bank (International Development Association) (2013). “Helping

Bangladeshi girls go further”, http://www.worldbank.org/ida/profile-gender.html,

retrieved on February 20, 2014

World Health Organization (WHO) (2003). Antenatal Care in Developing

Countries: Promises, Achievements and Missed Opportunities – An Analysis of Trends,

Levels and Differentials 1990-2001.

World Health Organization (WHO) (2013). Bangladesh: Country Profile,

http://www.who.int/gho/countries/bgd.pdf?ua=1, retrieved on February 20, 2014

Wooldridge, Jeffrey M. (2010). Econometric analysis of cross section and panel

data. MIT press.

Yau, Kelvin KW, Kui Wang, and Andy H. Lee. (2003). "Zero‐Inflated Negative

Binomial Mixed Regression Modeling of Over‐Dispersed Count Data with Extra Zeros."

Biometrical Journal 45(4):437-452.