According to International Conference on Population and Development, maternal health services which based on the concept of informed choice should include the following: education on safe motherhood, prenatal/antenatal care that is focused and effective, maternal nutritional programs, adequate delivery assistance that avoids excessive resource to Caesarian sections and provides for obstetric emergencies; referral services for pregnancy, childbirth and abortion complications; postnatal care and family planning [₁₃] . Among those maternal health services, mentioned above, this study focuses on antenatal care, assistance during delivery, and postnatal care. Also place of delivery as a maternal health service included in this study, because delivery at institutional level (government/private hospitals/hospitals running by NGOs/clinics/centers or at other type of health facility) receives better facilities and assistances than delivery at home.

The largest part of country's health infrastructure and health service system has been established by Government's management and control. Ministry of Health and Family Welfare (MOHFW) is responsible for comprehensive health policy formulation, planning and decision making in Bangladesh. There are two implementation wings under the MOHFW: (i) Directorate General of Health Services (DGHS) and (ii) Directorate General of Family planning (DGFP). The DGHS and DGFP are responsible for implementing all health programs and family planning programs respectively. In Bangladesh, the health service delivery system in public sector is divided in primary, secondary and tertiary levels. Health services delivery and care facilities providing by public health sector at various levels in Bangladesh are presented in table 1.

The Community Clinics are running for every six thousand population at village level in the country. Also, there are 96 maternal and child welfare centers (MCWCs) established at district level (also with some upazila level).To achieve the Millennium Development Goal 5 (Reduction of maternal mortality ratio to two thirds between 1990 and 2015), maternal health services have been given uppermost priority in the country's health system.

Figure 1

Table 1: Level and type of health facility in public sector health system, Bangladesh.

The services are provided through nationwide facility system as represented in table 1. Services are provided by the Family Welfare Assistants and Health Assistants at the village level. A Family Welfare Visitor (FWV) and a Sub-assistant Community Medical Officer (SACMO) or Medical Assistants (MA) is/are responsible for providing services at the union level. There are also some graduate medical officers who are involved at UHFWCs for providing maternal health services. At the Upazila level, the Maternal and Child Health (MCH) unit of the Upazila Health Complex (UHC) is providing child health services and maternal health services as well. Through an out patient consultation center and labor ward, district hospitals (DHs) at the district headquarters provide maternal health services.

Besides government sector health services, some private for-profit providers and private not for-profit providers play significant roles for providing health services including maternal health services in the country as well.

For analyzing married adolescents' use of maternal health services in Bangladesh, data extracted from Individual Recode Data file of Bangladesh Demographic and Health Survey (BDHS)–2004. The sample, used for this study, included those ever married adolescent women (EMAW) who had at least one birth in the five years preceding the survey interview. For those EMAW who had more than one birth, only utilization behavior of maternal health services associated with most recent pregnancy was considered. The sample of this study (for antenatal care, place of delivery and assistance during delivery) consists of 891 ever married adolescent women. For studying postnatal care, this study considers 776 ever married adolescent women (whose last delivery was at home).

In order to estimate the effects of individual's characteristics on maternal health services utilization, four dependent variables were considered in this study: antenatal care, place of delivery, assistance during delivery and postnatal care. All of these four variables (y) are dichotomous (see 3.3.1). Since dependent variables are dichotomous, logistic regression models were estimated for those four dependent variables. Logistic regression models provide an opportunity to estimate the probability of health service utilization depending on the independent variables included in the model. The logistic regression model for each of four dependent variables was considered as following:

Figure 2

Where, β= regression coefficient

α= constant

x _{= independent variable}

And exp(β_i)=odds ratio of one individual having characteristics i, (3)

Values of Probability of maternal health services utilization were estimated by using the equation (2).

Dependent variables

The study analyses four types of health service utilization: antenatal care, place of delivery, assistance at delivery and postnatal care, and those four health services are dichotomous and included as dependent variables in this study. In this study, dichotomous dependent variables indicate the non-use or use of respective maternal health service. Table 2 shows the descriptive statistics for four dependent variables.

Figure 3

Table 2: Descriptive statistics for dependent variables (maternal health services).

Antenatal care indicates whether care was sought from medically trained provider (doctor, nurse/midwife, family welfare visitor, MA/SACMO) which coded as 1; if care was sought from other than medically trained provider or no antenatal care was sought then it was coded as 0. More than 54% of total sample of EMAW who gave birth five years preceding the survey interview received antenatal care from medically trained provider for their last birth. Urban EMAW were more likely to have received antenatal care from medically trained provider for their last birth than rural EMAW (69.2% and 48.3% respectively).

Place of delivery specifies whether the delivery was at home — respondent's or others' home — (coded as 0) or other than home (coded as 1). The other than home category includes all types of deliveries which were done at government/private hospitals/hospitals running by NGOs/clinics/centers or at other type of health facility.

Assistance during delivery indicates assistance received at delivery from medically trained provider. Assistance received from medical doctor, nurse/midwife, family welfare visitor or MA/SACMO is considered as assistance received from medically trained provider and coded as 1. Assistance received during delivery from traditional birth attendant (TBA), relative/friend, untrained TBA, nonqualified doctor, health assistant, and family welfare assistant or from other person except medically trained provider is considered as assistance from medically non-trained provider and coded as 0.

Postnatal care indicates after delivery (at home only, because it is assumed that deliveries at health facility will receive postnatal checkup as a part of routine institutional delivery care) health checkup for mother from health professional (doctor, nurse/midwife, family welfare visitor, MA/SACMO, health assistant or family welfare assistant) which coded as 1 and health checkup for mother from TBA, non-qualified doctor, other person except health professional or no any health checkup is considered as no postnatal care (coded as 0).

Independent variables

This section displays the results of logistic regression analyses predicting the utilization of maternal health services using various independent variables mentioned in section 3.3.2. Three models for three dependent variables — antenatal care, place of delivery and assistance at delivery - were fitted for all ever married adolescent women who had had at least one birth for national (both urban and rural), urban and rural areas respectively. One model for postnatal care was fitted only for national and rural area, because postnatal care users in urban areas were very few.

Use of antenatal care

The results of logistic regression model fitted for antenatal care are presented in table 3. Among the predisposing characteristics of women only education variable has positive and significant effects on antenatal care in national level. Results show that adolescent women with primary education and above primary education are more likely to use antenatal care than adolescent women with no education. With respect to enabling characteristics of women, household wealth index and place of residence (urban-rural) have positive and statistically significant impact on use of antenatal care in national level. Also some disparities exist in antenatal care utilization between regions. The results derived from rural sample are similar to the results derived from national sample regarding significance of independent variables (but the magnitudes of their impact is not similar); but the results derived from urban sample are not similar with the result of national sample.

Figure 4

Table 3: Logistic regression estimates for use of antenatal care by ever married adolescent women (EMAW), Bangladesh.

For urban sample, no predisposing characteristic of women has significant impact on antenatal care utilization. Only household wealth index has statistically significant impact on antenatal care in urban areas of Bangladesh.

The expected odds, of the adolescent women belong to the household of the highest quintile of the household wealth index, for using antenatal care increase by 11.87%, 23.13% and 1.73% compared to the adolescent women belong to the household of the lowest quintile of the household wealth index for national, urban and rural areas respectively. Also, regional variations regarding antenatal care utilization exist in this study population.

Place of delivery

The results of logistic regression estimate for place of delivery are presented at table 4. Education level and type of family from predisposing characteristics of ever married adolescent women are significantly associated with place of delivery. Ever married adolescent women with primary and above primary level education are more likely, the respective odds are 1.87 times and 3.43 times more than odds of EMAW with no education, to give birth at health institution (other than home). The odds of EMAW from extended type family are 76% less to use health institution for delivery than EMAW from nuclear type of family. Couple (EMAW and her husband) from nuclear type of family may feel fear to perform the delivery (of the EMAW) at home, because there are no other family members to whom they could depend for getting helps during delivery. Because of the above reasons, Bangladeshi EMAW from nuclear type of family are more likely to use health institution (other than home) as place of delivery compared to EMAW from extended type of family.

Figure 5

Table 4: Logistic regression estimates for place of delivery of ever married adolescent women (EMAW), Bangladesh.

Assistance at delivery

Table 5 shows the results of logistic regression estimates for assistance during delivery. Education level and type of family from predisposing characteristics of women have statistically significant impact on assistance at delivery in national level of Bangladesh. The EMAW with primary and above primary education are more likely, estimated odds are 1.54 and 2.93 respectively, than EMAW with no education to take assistance at delivery from medically trained provider in national level of Bangladesh.

Figure 6

Table 5: Logistic regression estimates for assistance at delivery of ever married adolescent women (EMAW), Bangladesh.

Those EMAW who are member of extended type family are less likely to receive assistance at delivery from medically trained provider than EMAW from nuclear type family. The estimated odds of EMAW who are member of extended type family receiving assistance at delivery from medically trained provider decreased by 79% compared with EMAW from nuclear type of family. Use of health services by EMAW from extended type of family may depends on decision of older members (except husband of respected EMAW) of the family which may, some times, hinders to take assistance at delivery from medically trained provider.

Use of postnatal care

Logistic regression estimates for postnatal care use regarding national sample and rural sample are presented in table 6. Only age at the time of last birth from predisposing characteristics of women has statistically significant impact on postnatal care utilization. Education is also positively associated to raise the postnatal care utilization but not statistically significant.

Figure 7

Table 6: Logistic regression estimates for postnatal care use by ever married adolescent mothers, Bangladesh.

Predicted probabilities for using antenatal care are calculated by using those independent variables which were statistically significant in the logistic regression model. Education level, wealth index and place of residence were shown to have significant impact on antenatal care utilization from medically trained provider in the logistic regression model. Therefore, a separate logistic regression model was fitted with these independent variables to predict the probabilities for receiving antenatal care from medically trained provider:

logit(probability for using antenatal care)= 0.275+β₁X₁+β₂X₂+β₃X₃

where, X₁=level of education= (x₁₀=0 _{x11=1 x12=2) ^/ , β1=(β10=0 β11=0.311 β12=0.851)}

X₂=household wealth index= (x₂₀=0 x₂₁=1 _{x22 =2 x23=3 x24=4) ^/ ,}

β₂= (β_{2 0}=0 β₂₁= 0.645 _{β22=0.943 β23=1.399 β24=2.496).}

As shown in figure 1, probability for receiving antenatal care increases as household wealth index increases. There are differences in predicted probabilities for receiving antenatal care depending on education level and place of residence. The predicted probabilities are clustered together towards the upper end of household wealth index.

Figure 8

Figure 1: Predicted probabilities for antenatal care of ever married adolescent women (EMAW), Bangladesh, 2004.

When the household wealth index is high, differences in educational level and place of residence do not seem to affect the antenatal care utilization, but differences in educational level and place of residence seem to affect the antenatal care utilization when the household wealth index is low. At the lowest household wealth index, group of ever married adolescent women with no education and from rural areas have the lowest probability for receiving antenatal care compared to women with primary and above primary and from rural areas; also same results hold for women from urban areas. Probability for receiving antenatal care of a woman in rural areas with no education and belongs to lowest household wealth index is [1/{1+e ^{-[0.275+0(0)+0(0)+0(0)]} =0.5683] only 0.56, those (probabilities) of woman in rural areas with primary and above primary education and belongs to lowest quintile of household wealth index are [1/1+e ^{-[0.275+0.311(1)+0(0)]} =0.6424] 0.64 and [1/1+e ^{-[0.275+0.851(2)+0(0)]} =0.8784] 0.87 respectively. The probability for receiving antenatal care of a woman in urban areas with no education and belongs to lowest quintile of household wealth index is [1/1+e ^{-[0.275+0(0)+0(0)+0.363(1)]} =0.6543] 0.65, and those (probabilities) of woman belongs to lowest quintile of household wealth index in urban areas with primary and above primary education are [1/1+e ^{-[0.275+0.311(1)+0.363(1)]} =0.7209] 0.72, and [1/1+e ^{-[0.275+0.0.851(2)+0.363(1)]} = 0.9121] 0.91 respectively. However, even in scarce of resources, though women from urban areas are more likely to use antenatal care than women from rural areas, but education has an impact to raise the antenatal care utilization both in urban and rural areas.

In the logistic regression model for assessing the influence of individual's characteristics on place of delivery, education level and type of family have shown statistically significant effects. So, for predicting probabilities for place of delivery at health facility (other than home) another logistic regression model was fitted with education level and type of family as independent variables. The form of the logistic regression model as the following:

logit(probability for place of delivery at other than home)= -1.217+β₁X₁+β₂X₂

where, X₁= level of education= (x₁₀=0 _{x11=1 x12=2) ^/ , β1=(β10=0 β11=0.605 β12=1.674)}

X₂= type of family= (x₂₀=0 x₂₁=1) ^/ , β₂= (β_{2 0}=0 β₂₁= -1.115).

Figure 9

Figure 2: Predicted probabilities for place of delivery of ever married adolescent women (EMAW), Bangladesh, 2004.

The predicted probabilities for delivering at health facility (other than home) show the impacts of type of family and education level on place of delivery (figure 2). EMAW from nuclear type of family were more likely to use health facility (other than home) for their last child delivery than EMAW from extended type of family. EMAW from extended type of families, maybe, have less decision making power regarding health services utilization. Probabilities for using health facilities (other than home) for recent (last) child delivery increases as education level of EMAW increases at both types of family. Probability for delivering at health facility (other than home) of EMAW from nuclear type of family with no education is 0.23 [1/1+e ^{-[-1.217+0(0)+0(0)]} =0.2285], and those (probabilities) of EMAW from nuclear type of family with primary and above primary education are 0.35 [1/1+e ^{-[-1.217+0.605(1)+0(0)]} =0.3516] and 0.89 [1/1+e ^{-[-1.217+1.674(2)+0(0)]} =0.8938 ] respectively.

The probability for delivering at health facility (other than home) of EMAW from extended type of family with no education is 0.09 [1/1+e ^{-[-1.217+0(0)-1.115(1)]} =0.0.08850], and those (probabilities) of EMAW from extended type of family with primary and above primary education are 0.15 [1/1+e ^{-[-1.217+0.605(1)-1.115(1)]} =0.150972] and 0.73 [1/1+e ^{-[-1.217+1.674(2)-1.115(1)]} =0.734193] respectively. Therefore, education has an impact to raise the use of health facilities (other than home) as place of delivery even for controlling the type of family.

Education level of EMAW and type of family, where they lived, has shown statistically significant effects on receiving assistance at delivery in the logistic regression model. Therefore, a different logistic regression model was fitted for predicting the probabilities for receiving assistance at delivery from medically trained provider with education level and type of family as independent variables. The shape of the logistic regression model as under:

logit(probability for receiving assistance at delivery

from medically trained provider)= -0.920+β₁X₁+β₂X₂

where, X₁= level of education= (x₁₀=0 _{x11=1 x12=2) ^/ , β1=(β10=0 β11=0.419 β12=1.577)}

X₂= type of family= (x₂₀=0 x₂₁=1) ^/ , β₂= (β_{2 0}=0 β₂₁= -1.127).

As displayed in figure 3, education shows remarkable impact on receiving assistance at delivery from medically trained provider. The probability for receiving assistance at delivery from medically trained provider of EMAW from nuclear type of family with no education is 0.28 [1/1+e ^{-[-0.920+0(0)+0(0)]} =0.2849578], and those (probabilities) for receiving assistance at delivery from medically trained provider of EMAW from nuclear type of family with primary and above primary education are 0.38 [1/1+e ^{-[-0.920+0.419(1)+0(0)]} =0.377305690] and 0.90 [1/1+e ^{-[-0.920+1.577(2)+0(0)]} =0.903261444] respectively. Probabilities, for receiving assistance at delivery from medically trained provider, of EMAW from extended type of family with no education, primary and above primary education are 0.11, 0.16 and 0.76 respectively. Therefore, education has an influence to EMAW for receiving assistance from medically trained provider during delivery even controlling for type of family. Also, type of family shows modest impact on receiving assistance at delivery from medically trained provider. EMAW from nuclear type of family are more likely to receive assistance at delivery from medically trained provider than EMAW from extended type of family. In the extended type of family, decision regarding health service seeking, may be made by other adult member(s) of the family and which tends to hold EMAW to receive assistance during delivery from medically trained provider.

Figure 10

Figure 3: Predicted probabilities for assistance at delivery of ever married adolescent women (EMAW), Bangladesh, 2004.

Almost of this study has been conducted in my stay at Center for Northeast Asian studies of Jilin University, China. I am grateful to Center for Northeast Asian studies of Jilin University for providing me computer facility and good research environment. There was no financial support for this study.