The Impact of Demographic Factors on Poverty and Reproductive Health in Bangladesh: A Multivariate Analysis

Abstract

Poverty and reproductive health are important issues in Bangladesh as well as in the developing countries around the world. The number of reproductive health study is increasing day by day. On the aspect of socio-economic, demographic and reproductive health related characteristics, the present study investigates the impact of demographic factors on poverty and reproductive health among ever married women in Bangladesh based on Bangladesh Demographic and Health Survey (BDHS) 2014 data. In this study univariate analysis, bivariate analysis and multivariate analysis such as logistic regression analysis (multinomial logistic regression analysis and ordinal logistic regression analysis), multiple classification analysis (MCA), factor analysis (FA) and structural equation modeling (SEM) have been used to identify and examine the impact of demographic factors on poverty, reproductive health and health of ever-married women. To fulfill the objectives of this study, three main determinants have been considered: Standard of living status of ever-married women, reproductive health status of evermarried women at the time of last pregnancy and health status of ever-married women. The results of univariate analysis revealed that the current age of 36.4% ever-married women are in the age group 20-29 years, 30.0% in 30-39 years, and remaining 22.3% ever-married women age in 40 years or above. Early age at marriage for women is a serious concern in Bangladesh. The percentages distribution of the respondents according to their age at marriage 71.3% respondents have married within 18 years of age and only 29% women have married after age of 18 years. This is one of the major issue in women’s health and poor pregnancy-related outcome in Bangladesh. Literacy rate, especially among the women in Bangladesh is very low (66.3%). A countable part of total respondents 23.5% are illiterate and second largest number of respondents 29.3% completed their primary level and secondary education level are 37.6% and only 9.6% of the ever-married women have higher education.

It has been found that a substantial proportion of ever-married women included in the study have started childbearing before age at 20 years, thus it might happen that they have a high number of children born. More than 48% ever-married women have one or two children ever born, 41.4% ever-married women have three or more children, and only 10% ever-married women have no children ever born. And, also about 26% evermarried women have only one living children, 32.5% have two living children, and 40.9% ever married women have three or more living children in Bangladesh. Bivariate analysis has been used in this study to determine the association or relationship between the variables and also measures the strength of relationship. The chi-square test shows that current age, marital status, age at marriage, children ever born, number of living children, place of residence, religion, and family size are significantly (p < 0.01) associated with standard of living status of ever-married women. Again, variables current age, age at marriage, children ever born, number of living children, and place of residence are significantly (p < 0.05) associated with reproductive health status of evermarried women and remaining variables marital status, religion and family size are insignificantly associated with reproductive health status of ever-married women in Bangladesh. In the case of health of ever-married women, chi-square test shows that women’s current age, marital status, children ever born, number of living children, place of residence and family size are significantly (p < 0.01) associated with health status of ever-married women of reproductive age group in Bangladesh. Multinomial logistic regression model shows that current age of ever-married women has significant (p < 0.05) influence on their health status in Bangladesh. The reproductive women aged before 20 years, women belonging to the reproductive age group 20-29 years and age group 30-39 years are 2.70 times and 1.33 times more likely and 0.83 times less likely to be underweight whereas the likelihoods of being overweight or obese are 0.23 times and 0.55 times less likely and 1.02 times more likely among ever-married women aged before 20 years, women belonging to the reproductive age group 20-29 years, and 30-39 years than those women aged 40 years or above with normal health status respectively. Women’s marital status seems to have highly significant (p < 0.001) influence on health status of ever-married women in Bangladesh. Married women are less likely to be underweight (OR = 0.54, 95% CI: 0.43 - 0.69) while more likely to be overweight or obese (OR = 1.74, 95% CI: 1.33 – 2.27) than those women who are divorced or separated having normal health status. Another demographic factor such as children ever born have significant (p < 0.01) influence on ever-married women health status. Among the women having one to two children ever born are less likely to be under weight (OR = 0.67, 95% CI: 0.50 – 0.89) and more likely to be overweight or obese (OR = 1.42, 95% CI: 1.09 – 1.85) than those women who have three or more children ever born with normal health status. From these findings, it is observed that the relative chance to the health problem of ever-married women increases with the increasing of their children ever born in Bangladesh. Women’s place of residence has highly significant (p < 0.001) influence on their health status. With regards to the place of residence, urban evermarried women are less likely to be underweight (OR = 0.73, 95% CI: 0.64 – 0.85) while more likely to be overweight or obese (OR = 2.22, 95% CI: 1.98 – 2.49) relative to rural women having normal health status in Bangladesh. For the ordinal response, it is also necessary to use ordinal logistic regression model to get precise and efficient estimates of the regression coefficients. The proportional odds model has been used in this study to estimate the odds ratios. Women current age below 20 years are significantly 0.38 times (p < 0.001) less likely to being high standard of living compared to those women aged 40 years or above and women belonging to the age group 20-29 years are significantly 0.56 times (p < 0.001) less likely to deserve high standard of living than those women aged 40 years or above but women age group belonging to 30-39 years are significantly 0.92 times (p < 0.01) less likely to being high standard of living than those women aged 40 years or above. This study also shows that the married women are significantly 1.47 times (p < 0.001) more likely to expect higher standard of living compared to those women who are divorced or separated. Ordinal logistic regression model also shows that the early married women (age below or at age of 18 years) are significantly 0.73 times (p < 0.001) less likely to being high status of living than those women who married after age of 18 years in Bangladesh. Mothers who have one to two children ever born are significantly 1.79 times (p < 0.001) more likely to being high standard of living compared to women with three or more children ever born. The women who have one and two living children are insignificantly 1.27 times (p = 0.055) and 1.13 times (p = 0.229) more likely to deserve high standard of living compared to the women with three or more living children respectively. Women who live in urban areas are significantly 0.48 times (p < 0.001) more likely to deserve higher standard of living than those women who live in rural areas in Bangladesh. The present study also reveals that women with one to four and five to six family members are significantly 0.59 times (p < 0.001) and 0.71 times (p < 0.001) less likely to being high standard of living compared to those women with seven or more family members respectively. On the other hands, it is observed that women’s current age has an important effect on reproductive health of ever-married women in Bangladesh. The results of ordinal logistic regression model show that women current age below 20 years are insignificantly 0.98 times (p = 0.963) less likely to being high status of reproductive health compared to those women aged 40 years or above and women belonging to the age group 20-29 years are insignificantly 1.15 times (p = 0.757) more likely to deserve high status of reproductive health than those women aged 40 years or above. Moreover, ever-married women belonging to the age group 30-39 years are insignificantly 1.92 times (p < 0.159) more likely to being higher reproductive health compared to those women aged 40 years or above. This study also reveals that the married women are significantly 2.95 times (p < 0.05) more likely to expect higher reproductive health compared to those women who are divorced or separated. The early married women (age below or at age of 18 years) are significantly 0.77 times (p < 0.05) less likely to being high status of reproductive health compared to those women who married after age of 18 years. Mothers who have one to two children ever born are insignificantly 1.25 times (p < 0.498) more likely to being high status of reproductive health compared to the women with three or more children ever born. The women who have one and two living children are significantly (p < 0.05) 2.18 times and 1.83 times more likely respectively to deserve higher reproductive health compared to the women with three or more living children in Bangladesh. Another demographic factor such as place of residence has highly significant (p < 0.001) influence on reproductive health status of ever-married women. Women who are live in urban areas significantly 1.92 times (p < 0.001) more likely to deserve higher reproductive health status compared to those women who are live in rural areas in Bangladesh. From this finding, it can be concluded that urban women are more expected to being high quality of reproductive health compared to rural counterparts in Bangladesh. The results of multiple classification analysis (MCA) show that the effect of respondents’ age at marriage ( = 0.05 unadjusted and = 0.14 adjusted), children ever born ( = 0.11 unadjusted and = 0.13 adjusted), living children ( = 0.09 unadjusted and = 0.05 adjusted), place of residence ( = 0.44 unadjusted and = 0.43 adjusted) and family size ( = 0.06 unadjusted and = 0.09 adjusted) have the strongest influence for explaining the variability on standard of living status of ever-married women among the selected variables. This study also shows that women marital status ( = 0.06 unadjusted and = 0.07 adjusted), age at marriage ( = 0.10 unadjusted and = 0.06 adjusted), children ever born ( = 0.17 unadjusted and = 0.05 adjusted), number of living children ( = 0.17 unadjusted and = 0.18 adjusted), place of residence ( = 0.19 unadjusted and = 0.17 adjusted) and family size ( = 0.06 unadjusted and = 0.01 adjusted) have strong influence for explaining the variability on reproductive health status of ever-married women among the selected demographic variables. In the case of health of ever-married women, MCA shows that the effect of women’s current age ( = 0.19 unadjusted and = 0.23 adjusted), marital status ( = 0.06 unadjusted and = 0.09 adjusted), children ever born ( = 0.01 unadjusted and = 0.11 adjusted), number of living children ( = 0.11 unadjusted and = 0.18 adjusted) and place of residence ( = 0.18 unadjusted and = 0.17 adjusted) have strong influence for explaining the variation on health status of ever-married women among the included variables. Factor analysis has been applied to investigate and determine responsible demographic, poverty and reproductive health factors to explain the respective set of variables in an effort to find a new sets of variables, fewer in number than the original variables. From the factor analysis it is observed that there are nine eigen values greater than unity and almost 69.017 percent of the total variance is attributable to the nine factors. Five variables; household electricity (0.649), household television (0.755), household refrigerator (0.397), household cooking fuel (0.782) and household floor materials (0.773) are loaded on Factor 1 and an inspection of these items clearly shows that the majority of these items reflect the socio-economic characteristics. Factor 2 contains seven variables; number of visits antenatal care (0.527), place of delivery (0.804), health care after delivery (0.892), delivery by caesar (0.845), postnatal care provider (0.884), place of first postnatal care (0.496) and fertility planning status (0.754) with larger loading that clearly reflect the reproductive health. Factor 3 contains five variables; current age of respondents (0.775), total children ever born (0.919), number of living children (0.919), place of residence (0.663) and family size (0.538) with larger loading that clearly reflect the demography. Factors 5, 7, and 9 contain the variables; household drinking water (0.678), respondents mass media exposure (0.630), household radio (0.816), household toilet facility (0.496) and household bicycle (0.539) with larger loading that appear to reflect the related motives of socio-economic and an inspection of these items clearly shows that these items reflect socio-economic conditions. Factor 4 contains two variables such as current use of contraceptive (0.710) and contraceptive method types (0.784) with larger loading that clearly reflect the related motive of reproductive health. Factor 6 and 8 contain the variables current marital status (0.681), age at marriage (0.716) and religion (0.837) with higher loading that appear to reflect the changing pattern in composition of population and an inspection of these items clearly shows that these items reflect the demography. The present factor structure appears to be represented by three dimensions (Socioeconomic, Reproductive health, and Demography). So, it was decided to rerun factor analysis, stipulating the extraction of only three factors and rotated component matrix presents only three rotated factors. Factor 1 contains ten variables i.e., household electricity (0.697), household radio (0.819), household television (0.787), household refrigerator (0.658), household bicycle (0.536), household cooking fuel (0.624), household drinking water (0.675), household toilet facility (0.557), household floor materials (0.740) and respondents mass media exposure (0.554) with larger loadings that clearly reflect the socio-economic conditions and was thus Factor 1, labeled as Poverty Factor. Factor 2 contains nine variables i.e., number of visit antenatal care (0.503), place of delivery (0.806), health checkup after delivery (0.872), current use of contraceptive (0.706), delivery by caesar (0.843), postnatal care providers (0.862), place of first postnatal care (0.561), contraceptive method types (0.781) and fertility planning status (0.686) with larger loadings that reflect the reproductive health conditions and was thus Factor 2, labeled as Reproductive health Factor. Factor 3 contains eight variables i.e., current age of respondents (0.788), marital status (0.619), age at marriage (0.711), children ever born (0.888), number of living children (0.901), place of residence (0.491), Religion (0.814) and family size (0.535) with larger loadings that reflect the demographic conditions and was thus Factor 3 labeled as demographic factor. Finally, this three-factor model represents the combination of the nine original factors and appears to reflect adequately the underlying factor structure of the 27 variables. In this study structural equation modeling (path analysis) has been employed to examine and determine the direct and indirect effects of demographic factors on poverty, reproductive health and health of ever-married women in Bangladesh. For standard of living of ever-married women, it has been found that there are 20 paths out of 21 hypothesized paths statistically significant (p < 0.001). The SEM results reveal that that respondent’s marital status, place of residence, religion, children ever born and number of living children have direct significant (p < 0.001) negative effects on standard of living and respondent’s current age and family size have direct significant (p < 0.01) positive effects on standard of living of ever-married women in Bangladesh. Total effects of the respondent’s current age, marital status, age at marriage, place of residence, religion and family size on standard of living of ever-married women are transmitted through its implied effect about 67.9%, 77.0%, 45.8%, 83.0%, 75.7% and 77.5% respectively and indirect effects of those exogenous variables through endogenous variable children ever born on standard of living are 24.8%, 16.5%, 41.7%, 13.3%, 17.6% and 16.5% respectively. However indirect effects of the respondent’s current age, marital status, age at marriage, place of residence, religion and family size on standard of living through endogenous variable number of living children are 7.4%, 6.5%, 12.5%, 3.7%, 6.8% and 6.0% respectively. Total effect of children ever born and number of living children on standard of living of ever-married women are about -0.275 and -0.084 of which both are about 100.0% transmitted through its direct effect on standard of living of ever-married women. SEM results for reproductive health show that there are 18 paths out of 21 hypothesized paths statistically significant on reproductive health of ever-married women and out of eight variables, five variables are found to have significant (p < 0.05) direct effect on reproductive health of ever-married women. Among those respondent’s marital status, place of residence and children ever born have direct significant (p < 0.05) negative effects and respondent’s current age and age at marriage have direct significant (p < 0.01) positive effects on reproductive health of ever-married women in Bangladesh. Total effects of the respondent’s current age, marital status, age at marriage, place of residence, religion and family size on reproductive health of ever-married women are transmitted through its implied effect about 67.0%, 80.1%, 77.0%, 81.3%, 43.3% and 62.6% respectively and indirect effects of those exogenous variables through endogenous variable children ever born on reproductive health are 28.7%, 16.7%, 19.7%, 16.5%, 46.7% and 31.6% respectively. However indirect effects of the respondent’s current age, marital status, age at marriage, place of residence, religion and family size on reproductive health of ever-married women through endogenous variable number of living children are 4.3%, 3.2%, 3.3%, 2.3%, 10.0% and 5.8% respectively. Total effect of children ever born and number of living children on reproductive health are about -0.314 and -0.048 of which both are about 100.0% transmitted through its direct effect on reproductive health of ever-married women. For health of ever-married women, it is found that there are 19 paths out of 21 hypothesized paths statistically significant on health of ever-married women and out of eight variables six variables are found to have significant (p < 0.05) direct effect on health of ever-married women. Among those respondent’s marital status, place of residence and children ever born have direct significant (p < 0.01) negative effects and respondent’s current age, age at marriage and number of living children have direct significant (p < 0.01) positive effects on health of ever-married women in Bangladesh. Total effects of the respondent’s current age, marital status, age at marriage, place of residence, religion and family size on health of ever-married women are transmitted through its implied effect about 65.7%, 78.3%, 53.8%, 85.1%, 48.4% and 38.9% respectively and indirect effects of those exogenous variables through endogenous variable children ever born on health of ever-married women are 25.7%, 15.2%, 34.6%, 11.4%, 35.5% and 43.5% respectively. However indirect effects of the respondent’s current age, marital status, age at marriage, place of residence, religion and family size on health of ever-married women through endogenous variable number of living children are 8.6%, 6.5%, 11.5%, 3.5%, 16.1% and 17.6% respectively. Total effect of children ever born and number of living children on health of ever-married women are about -0.247 and 0.086 of which both are about 100.0% transmitted through its direct effect on health of ever-married women. Therefore the variables children ever born, number of living children and family size are needed to be controlled for better standards of living as well as reproductive health status of ever-married women, age at marriage and urbanization are needed to be enhanced in Bangladesh for the development of socio-economic and reproductive health status of ever-married women.