INTRODUCTION:

Maternal mortality and morbidity are pathetic phenomenon for a family and for a society. The effect of a mother’s death results in vulnerable families and their infants. If maternal mortality is unexpectedly large particularly in impoverished communities like ours, then there arise so many crisis. That’s why, we need to reduce maternal death and ensure proper care for mothers. Through proper antenatal care (ANC), maternal mortality and morbidity can be reduced. That’s why, we have selected number of ANC visits as response variable in our study.

As the case of many other parts of the world, maternal mortality and child mortality are major problems in Bangladesh. Bangladesh is country with high maternal mortality and child mortality and because maternal and newborn health is inextricably linked, of those women who die, only one in four of their babies will survive their first week of life. These high mortality rates are underpinned by the fact that 9 out of every 10 deliveries take place at home, most with unskilled attendants or relative assisting. The low status of women, poor quality and low uptake of services all add to this problem. A study on safe motherhood programs in Bangladesh found that women’s low status in society, the poor quality of maternity care services, lack of trained providers, low uptake of services by women and infrastructure all contribute to the high rate of maternal deaths. This is compounded by strong cultural and traditional ties that deter women from delivering at health centers or with medically- trained attendants because their mothers have given birth “naturally” for generations. There is also little understanding about taking rest or additional nutritious food during pregnancy. Moreover, the low status of women within the family means one in every two women will have her health care decided by her husband. Often her mother-in-law will be a key decision maker.

Overall, women’s use of ANC is central to achieving the Millennium Development Goals to improve maternal health and to reduce child mortality. Unfortunately, many women in development countries do not receive such care. Report from neighboring countries show that a high utilization rate of ANC service results in lowing the risk of maternal mortality. A lots of research work have done using Bangladesh Demographic and Health Survey (BDHS) 2014 data to determine the crucial factors affecting antenatal care in Bangladesh. But most of them are analyzed using single level model. The 2014 BDSH data is based on two stage stratified sampling. Our study found that for such hierarchical structured data the multilevel effects (division level) have been found significant that resembles with (Chitalu Chiliba and Steve Koch, 2013) studies and have taken into consideration multilevel regression model. As, a result, this multilevel analysis enables the proper investigation of the effects of all independent variables measured at division level on the response variable the number of antenatal care visits of women during their pregnancy period in Bangladesh. The main reason for which there exists a significant multilevel effect for our data might be that there might be dependencies between individual observations due to sampling is not taken randomly but cluster sampling from geographical areas is used instead. In recent years, the modeling of count data is of a primary interest in many fields such as insurance, public health, epidemiology, psychology, and many other research areas.

Multilevel model is appropriate for finding contextual effect of antenatal care visits using BDHS 2014 data because in this data individual nested under different level of hierarchy. In this study the units at lower level (level-1) are individuals who are nested within units at higher level (division: level-2). We will specify two mixed effect models: Poisson regression model and negative binomial regression model with random effect which are also known as multilevel model with two level, for the count data. We can estimate these models by maximum likelihood method, generalized estimating equation approach, Bayesian approach etc. In this research we apply MLE method which are known as generalized linear mixed model approach. Using this method, we will compare the above two mixed effect models which are multilevel model with two levels.

This study, based on Bangladesh Demographic Health Survey (BDHS) 2014 data, attempts to identify important factors influencing the number of antenatal care visits of women during their pregnancy period in Bangladesh. Factors considered including socio-economic and demographic characteristics such as respondent’s age, place of residence, division, source of drinking water, respondent’s education, wealth index, respondent’s husband’s education, decision maker on respondent’s health care, access to mass media.

The purpose of this study is to examine the factors associated with the number of antenatal care visits of women to illustrate the options facing governments for setting strategies for ensure the antenatal care by increasing the number of antenatal care visits of women during their pregnancy period in Bangladesh.

MATERIALS AND METHODS:

Data and Variable:

In this study, one of our main purposes is to determine the potential factors influencing the number of antenatal care (ANC) visits during pregnancy and we use 2014 BDHS data set which is based on two stage stratified sampling. Various literatures shows that the appropriate approach to analyzing this data is multilevel or hierarchical model approach.

In this study, the dependent variable is “Number of Antenatal Care Visits of Women in Bangladesh” which contains minimum 0 to maximum 10 and 9 (nine) predictor variables have been included for this analysis which are respondent’s age, place of residence, division, source of drinking water, respondent’s education, wealth index, respondents husband’s education, decision maker on respondent’s health care and access to mass media. From these independent variables, we consider ‘Division’ as the level-2 variation or the random effect and the rest are fixed effect. Several types of tests, goodness of fit, description of AIC, BIC etc. are used for selecting the best model among a set of models in the data analysis.

The Model and Estimation Procedures:

Multilevel analysis is a suitable approach to take into account the social contexts as well as the individual respondents or subjects (Snijders, 2011). Normally these situations can be seen in the data collected by multi-stage stratified clustered sampling. The simplest and the most common multilevel model consider only two level of analysis and this study deals only with this.

A multilevel model or a mixed model can be represented as,

_Where,

_{Y is known vector of observation, with mean}

_{E (Yi) =}_q_i_{= log µi = xi}_¢_β

_xi_{is
fixed effect covariates.}

_β_{is an unknown regression coefficients of fixed effects.}

_Zi_{is
the unknown random effects.}

_e_{is an unknown random errors, with mean E (}_e_{) = 0 and variance, Var(}_e_{) = R}

_Let,_q_i_=
xi_¢_β_{= β0 + β1x1+β2x2 + .+ βkx
k}

In this process, we consider a generalized linear model with link log,

i.e. we get,

In this section it is given the description of two multilevel models with two levels:

1. Poisson Regression Model with Random Effect

2. Negative Binomial Regression Model with Random Effect.

Poisson Regression Model with Random Effect:

In Poisson model with random effects we consider,

So, in our analysis the Poisson regression model is,

Here,

Negative Binomial Regression Model with Random Effect:

In negative binomial regression with random effects the parameter is modeled

log=

Where β is the (p+1)×1 vector of unknown parameters associated with the known covariate vector where p is the number of covariates not including the intercept, and Z_i be the random effects which follows a multivariate normal distribution with mean zero and variance-covariance matrix . i.e. Z_i~ MVN (0, ψ)

_{In
our analysis the model is,}

₌

Modeling “The Number of Antenatal Care Visits of Women” in Bangladesh:

Multilevel (with two level) regression is given below:

glmer (factor (no. of antenatal care visit) ~ factor (respondent’s age) + factor (place of residence) + factor (source of drinking water) + factor (respondent’s education) + factor (wealth index) + factor (respondents husband’s education) + factor (decision maker on respondent’s health care) + factor (access to mass media) + (1| factor (division)).

Fitting different regression models with random effect for the number of antenatal care visits:

Fitting the Poisson regression model with random effect

Using the number of antenatal care visits as dependent variable our proposed model is

Log(µ_i) = β₀+ β_1jX_1j+ β_2jX_2j+ β_3jX_{3 j}+ β_4jX_4j+ β_5jX_5j+ β_6jX_6j+ β_7jX_7j +β_8jX_8j+ Z_ij

Where, the variables used in the model are defined as

X₁= Respondent’s age (3 categories)

X₂ = Place of residence (2 categories)

X₃= Source of drinking water (3 categories)

X₄= Respondent’s education (4 categories)

X₅ = Wealth index (3 categories)

X₆ = Respondent’s husband’s education (4 categories)

X₇ = Decision maker on respondent’s health care (4 categories)

X₈ = Access to mass media (2 categories)

All of these are fixed effects and Z_ijis the random effect or level-2 variation or cluster variation, where, Z_ij = Division with seven categories.

RESULTS:

At first, we find whether there exists a significant association between the continuous variable, number of ANC visits and a categorical variable. The results of ANOVA are given in the Table 1.

Table 1: Mean number of ANC visits by the selected Socioeconomic and Demographic variables.

Variables (n=Sample Size)	Mean ± SD	Percentage	95% CI	p-value
Respondent’s age				0.000
Under 19 (915)	2.61 ± 2.313	20.8	(2.46, 2.76)
19-29 (2629)	2.83 ± 2.484	59.8	(2.74, 2.93)
Above 29 (852)	2.48± 2.349	19.4	(2.32, 2.64)
Place of residence				0.000
Urban (1407)	3.56± 2.566	32	(3.43, 3.70)
Rural (2989)	2.32 ± 2.252	68	(2.24, 2.40)
Division				0.000
Barisal (522)	2.39±2.237	11.9	(2.19, 2.58)
Chittagong (847)	2.57±2.429	19.3	(2.41, 2.74)
Dhaka (779)	3.01 ±2.502	17.7	(2.83, 3.18)
Khulna (506)	3.31±2.406	11.5	(3.10, 3.52)
Rajshahi (533)	2.70 ±2.473	12.1	(2.49, 2.91)
Rangpur (545)	3.21 ±2.434	12.4	(3.00, 3.41)
Sylhet (664)	2.00±2.208	15.1	(1.83, 2.17)
Source of drinking water				0.000
Other (806)	3.09 ± 2.572	18.3	(2.91, 3.27)
Tap water (93)	3.57 ± 2.688	2.1	(3.02, 4.12)
Tube-well water (3497)	2.61± 2.373	79.5	(2.53, 2.69)
Respondent’s education				0.000
Illiterate (587)	1.45 ± 1.815	13.4	(1.31, 1.60)
Primary (517)	2.10 ± 2.269	11.8	(1.91, 2.30)
Secondary (323)	3.62 ± 2.303	7.3	(3.37, 3.87)
Above secondary (2969)	2.98 ± 2.463	67.5	(2.89, 3.07)
Wealth index				0.000
Poor (1757)	1.83 ± 2.115	40.0	(1.73, 1.93)
Middle (840)	2.46 ± 2.158	19.1	(2.31, 2.60)
Rich (1799)	3.71 ± 2.465	40.9	(3.60, 3.82)
Respondent’s husband’s education				0.000
Illiterate (1004)	1.79 ± 2.043	22.8	(1.67, 1.92)
Primary (1331)	2.28± 2.314	30.3	(2.16, 2.41)
Secondary (1393)	3.08 ± 2.335	31.7	(2.96, 3.20)
Above secondary (668)	4.23 ± 2.501	15.2	(4.04, 4.42)
Decision maker on respondent’s health care				0.000
Respondent alone (520)	2.97 ± 2.511	11.8	(2.75, 3.18)
Respondent and husband (2123)	2.83 ± 2.439	48.3	(2.73, 2.94)
Husband alone (1417)	2.44 ± 2.380	32.2	(2.32, 2.57)
Someone else (336)	2.80 ± 2.319	7.6	(2.55, 3.05)
Access to mass media				0.000
No (1699)	1.78 ± 2.063	38.0	(1.68, 1.88)

Estimate of Fixed Parameter: (For Poisson Regression Model)

Table 2: Estimated Parameters of Poisson Regression Model

Independent variable	Categories	Estimated parameter	Standard Error	Odds ratio	Z value	Pr(>\|z\|)
Intercept		0.386068	0.074114		5.209	1.90e-07 ***
Respondent’s age	Under 19	…….	………	………	………	………..
	19-29	0.031875	0.023920	1.0324	1.333	0.182682
	Above 29	-0.007978	0.031329	0.9921	-0.255	0.798994
Place of residence	Urban	…..	…….	……	……..	…………
Place of residence	Rural	-0.159515	0.020995	0.8526	-7.598	3.02e-14 ***
Source of Drinking water	Other	……...	……………	………….	……………	……………
	Tap water	0.043635	0.059242	1.0446	0.737	0.461387
	Tube-well water	-0.020969	0.023454	0.9792	-0.894	0.371298
Respondent’s education	Illiterate	……….	…………	…………	…………	……….
	Primary	0.224886	0.047401	1.2522	4.744	2.09e-06 ***
	Secondary	0.416301	0.049798	1.5163	8.360	< 2e-16 ***
	Above secondary	0.367202	0.039642	1.4437	9.263	< 2e-16 ***
Wealth index	Poor	………	…………	……….	…………	………..
	Middle	0.107417	0.030329	1.1134	3.542	0.000398 ***
	Rich	0.328293	0.029206	1.3886	11.241	< 2e-16 ***
Respondent’s husband’s education	Illiterate	…………	…………….	………….	……………	………
	Primary	0.058598	0.031109	1.0603	1.884	0.059618.
	Secondary	0.151740	0.031906	1.1639	4.756	1.98e-06 ***
	Above secondary	0.323437	0.035691	1.3819	9.062	< 2e-16 ***
Decision maker about respondent’s health	Respondent alone	……….	…………	…………..	…………	……….
	Respondent & husband	-0.040414	0.028858	0.9604	-1.400	0.161378
	Husband alone	-0.110774	0.031111	0.8951	-3.561	0.000370 ***
	Someone else	-0.088591	0.041867	0.9152	-2.116	0.034343 *
Access to mass media	No	………..	………………..	……………	…………….	…………
Access to mass media	Yes	0.215419	0.026212	1.2404	8.218	< 2e-16 ***

Signif. Codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Estimate of Fixed Parameter: (For Negative Binomial Regression Model):

Table 3: Estimated Parameters of Negative Binomial Regression Model

Independent variable	Categories		Estimated parameter	Standard error	Odds ratio	Z value	Pr(>\|z\|)
Intercept			0.376866	0.100835		3.737	0.000186 ***
Respondent’s age		Under 19	……	…….	…….	……	……
		19-29	0.020538	0.037190	1.0208	0.552	0.580789
		Above 29	-0.024357	0.048742	0.9759	-0.500	0.617270
Place of residence		Urban	…….	…….	…….	……	……..
Place of residence		Rural	-0.160040	0.033908	0.8521	-4.720	2.36e-06 ***
Source of Drinking water		Other	……	……	……	……	……
		Tap water	0.080005	0.099708	1.0833	0.802	0.422327
		Tube-well water	-0.002849	0.037800	0.9972	-0.075	0.939918
Respondent’s education		Illiterate	…….	…….	……	……	……
		Primary	0.224581	0.066486	1.2518	3.378	0.000730***
		Secondary	0.435050	0.074896	1.5450	5.809	6.29e-09 ***
		Above secondary	0.369339	0.054836	1.4468	6.735	1.64e-11 ***
Wealth index		Poor	……	……	……	…….	…….
		Middle	0.113738	0.044861	1.1205	2.535	0.011233 *
		Rich	0.338726	0.044822	1.4032	7.557	4.12e-14 ***
Respondent’s husband’s education		Illiterate	…….	…….	…….	…….	…….
		Primary	0.062354	0.044770	1.0643	1.393	0.163697
		Secondary	0.150528	0.047367	1.1624	3.178	0.001483 **
		Above secondary	0.324062	0.055663	1.3827	5.822	5.82e-09 ***
Decision maker about respondent’s health		Respondent alone	…….	…….	……	…….	……..
		Respondent & Husband	-0.047363	0.046297	0.9537	-1.023	0.306299
		Husband alone	-0.131258	0.049247	0.8770	-2.665	0.007692 **
		Someone else	-0.102311	0.066397	0.9027	-1.541	0.123340
Access to mass media		No	…….	…….	…….	…….	…….
Access to mass media		Yes	0.224251	0.038442	1.2514	5.834	5.43e-09 ***

Signif. Codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

The Table 3 presents estimated parameters of the Negative Binomial regression model for number of antenatal care visits. The value of AIC of the above fitted model is 17942.9, BIC is 18070.7 residual deviance 17902.9 on 4376 degrees of freedom following chi-square with 1 degree of freedom. The dispersion parameter is found to be 17902.9/4376= 4.09115. From the Table 3 it is observed that the parameter estimates have been increased and the standard errors have also increased. It is also observed that Negative binomial regression model reduced over-dispersion problem.

Estimate of Random parameter (level-2 variation):

Groups Name Variance Standard Deviation

Division (Intercept) 0.02115 0.1454

The estimate of the random intercept for level-2 (division) is 0.02, which means that the average variation of the number of antenatal care visits of women between divisions is 0.02.

DISCUSSION:

Based on AIC, BIC and dispersion parameter it is found that between two mixed effect models, Negative Binomial regression model with random effect is better model for modeling the number of antenatal care visits of women in Bangladesh which is over-dispersed count data. So, now we interpret estimated parameters of Negative Binomial regression model with random effect.

The women in the selected model visit for antenatal care at least one time in their pregnancy period.

The intercept was found to be 0.376866 in the Table 3 which is statistically significant at 0.1% level of significance. The intercept term 0.376866 represents that log of the average number of antenatal care visits of women during pregnancy who is under 19 and illiterate, comes from poor family in urban area of Bangladesh, whose husband is uneducated, who drink other source of water, take the decision on their health care alone and have no access to mass media (Radio, Television and Newspaper) in the group of women who visit for antenatal care. In this case the expected number of antenatal care visits of above characterized women is exp (0.376866) =1.45771. That is the above characterized women who visit for antenatal care, take antenatal care 1.45771 times on average during their pregnancy period in Bangladesh.

The estimated parameters corresponding to respondent’s age are statistically insignificant.

The estimated parameter of place of residence of rural category is -0.160040 which is statistically significant at 0.01% level of significance. The women who live in rural are visit for antenatal care on average 0.8521 times less than the women who live in urban area in the group of women who visit for antenatal care at least one time. The estimated parameters corresponding to source of drinking water are statistically insignificant.

From the Table 3 it is found that primary, secondary and above secondary educational levels of respondent all are highly significant. From the odds ratios it is clear that, the ratio of women who receive antenatal care and who never receive antenatal care is 1.2518, 1.5450 and 1.4468 times more among the primary, secondary and above secondary educated women respectively than the illiterate women. It also clear that the primary educated respondents visit for antenatal care on average about 25.2%, the secondary educated respondents visit for antenatal care on average about 54.5% and the above secondary educated respondents visit for antenatal care on average about 44.7% more than the illiterate respondents in the group of antenatal care visited respondents.

The odds ratios for middle and rich levels of wealth index are 1.1205 and 1.4032 respectively, which are strongly associated with the number of visits for antenatal care of women during pregnancy period at 5% and 0.01% level of significance respectively. From this result it is clear that, the women of middle class family and rich family visit for antenatal care on average respectively 1.1205 and 1.4032 times more as compared to the women of poor family.

The parameters corresponding to respondent’s husband’s education indicate that primary level of respondent’s husband’s education is statistically insignificant but secondary and above secondary both levels of respondent’s husband’s education are statistically significant at 1% level of significance and 0.01% level of significance respectively. The odds ratios secondary and above secondary levels of respondent’s husband’s education are 1.1624, and 1.3827 respectively. From the odds ratios it is clear that the average number of antenatal care visit of respondents whose husband is secondary and above secondary educated are respectively 1.1624, and 1.3827 times more than the respondents whose husband is illiterate in the group of respondents who at least one time take antenatal care.

The decision maker on respondent’s health care is statistically significant. Only husband alone level of decision maker on respondent’s health care is statistically significant at 1% level of significance. It is clear that the respondents whose health care decision is taken by husband alone they visit for antenatal care on average 0.8770 times less than the respondents who take the decision on their health care alone among the antenatal care taken respondents.

Access to mass media is statistically significant at 0.01% level of significance. This indicates access to mass media has impact on the number of antenatal care visit. From the analysis it is found that, the odds ratio of access to mass media is 1.2514 which means the average number of antenatal care visits of women who have access to mass media about 25.1% more than women who have no access to mass media in the group of antenatal care visited women.

The negative binomial regression model with random effect is presented in the following fitted equation:

Log (mean number of visits) = 0.376866 + 0.020538 X₁₂- 0.024357 X₁₃- 0.160040 X₂₂+ 0.080005 X₃₂- 0.002849 X₃₃+ 0.224581 X₄₂+ 0.435050 X₄₃+ 0.369339 X₄₄+ 0.113738 X₅₂+ 0.338726 X₅₃+ 0.062354 X₆₂+ 0.150528 X₆₃+ 0.324062 X₆₄- 0.047363 X₇₂- 0.131258 X₇₃- 0.102311 X₇₄+ 0.224251 X₈₂+ Z_ij

Where, X_1j’s represent the levels of respondent’s age, X_2j’s represent the level of place of residence,X_3j’s represent the level of source of drinking water, X_4j’s represent the level of respondent’s education, X_5j’s represent the level of wealth index categories of the respondent’s, X_6j’s represent the level of respondent’s husband’s education X_7j’s represent the level of decision maker on respondent’s health care and X_8j’s represent the level of access to mass media of the respondent’s and Z_ijis the level-2 variation or cluster variation of the model, where j for individual (Level-1), and i for division (Level-2).

Result interpretation of random parameter which is division level or level-2 variation of selected Negative Binomial Regression Model with random effect:

The variance of the random intercept term, which shows the extent to which outcomes between divisions differ, after controlling for the covariates. The estimate of the random intercept for level-2 (division) is 0.02, which means that the average variation of the number of antenatal care visits of women between division is 0.02.

Although both individual- and division-level characteristics have an influence on the inadequate and non-use of ANC, division-level factors have a stronger influence in the rural areas.

CONCLUSION:

In this study it is found that the multilevel effects (division level) have been found significant and have to take into consideration in mixed effect model which leads multilevel analysis. The study provides evidence that, while both individual and division-level factors are instrumental in determining the attendance and utilization of ANC. Based on findings of this study, we can say that the women who have secondary educational qualification, come from rich family, live in urban area of Bangladesh, whose husband’s educational qualification is above secondary, take the decision on their health care alone and have access to mass media (Radio, Television and Newspaper) visit more times for antenatal care among the women who visit for antenatal care, whereas in the class of women who have no educational qualification, come from poor family, live in rural area of Bangladesh whose husband is illiterate and have no access to mass media.

So i wish, this study can help policymakers and program managers to track the progress of mother’s health and refocus efforts to meet the goal of reducing maternal and child mortality and morbidity to a great extent

CONFLICT OF INTEREST:

There is no conflict of interest for this study.

REFERENCES:

1. Ana Maria Osorio, Luis Miguel Tovar, Katharina Rathmann. 2014. Individual and local level factors and antenatal care use in Colombia: a multilevel analysis. Saude Publica ol.30 no.5 Rio de Janeiro. Retrieved from http://dx.doi.org/ 10.1590 /0102-311X00073513.

2. Bangladesh Demographic and Health Survey (BDHS). 2014. NIPORT, Dhaka, Bangladesh; Mitra and Associates, Dhaka/ Bangladesh.

3. Bibha Simkha, E.R. 2008. Factor affecting the utilization of antenatal care in developing countries: Systematic review of the literature. Journal of advanced Nursing 61(3), 244-260.

4. Chitalu Chiliba and Steve Koch. 2013. Utilization of focused antenatal care in Zambia: Examining individual- and community-level factors using a multilevel analysis. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. Health Policy and Planning 2015; 30:78–87

5. Gardner, W., Mulvey, E.P. and Shaw, E.C. 1995. Regression analysis for counts and rates: Poisson, Over-dispersed Poisson and Negative Binomial models. Psychological Bulletin, Vol-118, 392-404.

6. Gelman, A. 2012. Multilevel (hierarchical) modeling: what it can and cannot do. Technometrices.

7. Jerald F. Lawless. 1987. Negative Binomial and Mixed Poisson Regression. The Canadian Journal of Statistics. Vol. 15, No. 3. pp. 209-225.

8. Lambert, D. 1992. Zero-inflated Poisson regression, with an application to defects in manufacturing. Technometrics, Vol.-34, 1-14.

9. Littell, R., Henry, P., and Ammerman, C. 1998. Statistical analysis of repeated measures data using sas procedures. Journal of animal science, 76(4): 1216-1231.

10. MacDonald, R. B. 2014, March 24. Overdispersion and Poisson regression, 268-280.

11. Mullahy, J. 1998. Specification and Testing of Some Modified Count Data Models. Journal of Econometrics, Vol.-33, 341-365.

12. Nelder, J. and Wedderburn, R. 1972. Generalized Linear Models. Journal of the Royal Statistical Society, Vol.-135(3), 370-384.

13. NIPORT. 2014. Bangladesh Demographic and Health Survey 2014. Rockville, Maryland, U.S.A. National Institute of Population Research and Training, Dhaka, Bangladesh, Mitra and Associates, and ICF International.

14. Peiming Wang, Martin L. Puterman, Iain Cockburn and Nhu Le. 1996.Mixed Poisson Regression Models with Covariate Dependent Rates. International Biometric Society.: Biometrics, Vol. 52, pp. 381-400.

15. Peter F. Thall. 1998. Mixed Poisson Likelihood Regression Models for Longitudinal Interval Count Data. International Biometric Society, Vol. 44, pp. 197-209.

16. Priyanka Dixit, Laxmi Kant Dwivedi and Faujdar Ram. 2013. Estimating the impact of antenatal care visits on institutional delivery in India: A propensity score matching analysis. Health http://dx.doi.org/10.4236/health.2013.55114. Vol.5, No.5, 862-878

17. Rabbi, A. M. F. and Karmaker, S.C. 2015. The Socio-economic Determinants of Antenatal Health Care Utilization in Bangladesh: Evidence from Multivariate Techniques. Dhaka Univ. J. Sci., 63(1):19-23

18. Rahman, M, Islam, R. and Rahman, M. 2010. Antenatal Care Seeking Behaviour among Slum Mothers: A Study of Rajshahi City Corporation, Bangladesh. Sultan Qaboos University Medical Journal, 10(1): 50-56.

19. Robert D. Gibbons and Donald Hedeker. April, 2000. Applications of Mixed-Effects Models in Biostatistics. The Indian Journal of Statistics, Series B (1960-2002), Vol. 62, No. 1, pp. 70-103.

20. Shahjahan, M., Chowdhury, H.A., Akhter, J., Afroz, A., Rahman, M. M. and Hafez, M.A. 2012. Factors associated with use of antenatal care services in a rural area of Bangladesh. South East Asia Journal of Public Health, 2(2):61-66.

21. Simon J. Bond and Vernon T. Farewell. July, 2009. Likelihood Estimation for a Longitudinal Negative Binomial Regression Model with Missing Outcomes. Journal of the Royal Statistical Society. Series C (Applied Statistics), Vol. 58, No. 3pp. 369-382

22. Snijders, T. A. 2011. Multilevel analysis. Springer.

23. Venkatram Ramaswamy, Eugene W. Anderson and Wayne S. DeSarbo. 1994. A Disaggregate Negative Binomial Regression Procedure for Count Data Analysis. : INFORMS. : Management Science, Vol. 40, pp. 405-417.

24. WHO/UNICEF, Antenatal care in developing countries: promises, achievements and missed opportunities: an analysis of trends, levels and differentials. 1990-2001. World Health Organization. 2003. Geneva.

Received on 14.01.2020 Modified on 12.03.2020

Asian Journal of Management. 2020; 11(1):107-114.

DOI: 10.5958/2321-5763.2020.00017.7