Decision Making Autonomy and Maternal Healthcare Utilization among Nigerian Women

1.1 Background of the study

Poor maternal health care utilization is of great concern in developing countries leading to high maternal and child mortality. Globally, maternal mortality is very high with about 830 women dying every day from pregnancy or childbirth related.1 In 2010, World Health Organization (WHO) reported that 358,000 maternal mortalities occurred globally, with more than 354,000 occurring in developing countries and about 200,000 deaths in sub-Saharan African countries.2 This has, however, dropped in 2015 to 303,000 women death globally.3,1

Studies have established that maternal mortality and morbidity are inversely proportional to health care services utilization. That is, when healthcare service utilization is low, this will translate into high maternal and child mortalities. Also, where reproductive healthcare utilization is high, maternal and child mortalities would be low.4 One of the most common indicators of a woman’s health and reproductive behavior is the state of her antenatal care utilization rate.5

The prevalence of non-utilization of healthcare services among several Nigerian women of reproductive age remains a major concern to all stakeholders in the actualization of Sustainable Development Goals (SDGs) by year 2030. It’s been reported that more than 33% of Nigerian pregnant women do not use antenatal care (ANC) service during pregnancy.4 For instance, the WHO’s World Health Statistics reported that in 2015 only 61% of Nigerian pregnant women had attended ANC at least once during their pregnancy period and only 51% met the WHO standard of a minimum of 4 visits.6

Women autonomy as collected in the Nigeria Demographic and Health Survey (NDHS) is defined as the extent to which women are independent on finances, level of participation in family decision making on matters pertaining to her health and that of the household, and freedom to visit outside of her matrimonial home without having to obtain permission.7 Recent studies have reported mixed conclusions on the relationship between women’s autonomy and reproductive healthcare utilization. In some Asian countries, studies found association between women autonomy and use of reproductive healthcare services.8,9 On the other hand,10 one Nepal study concluded that women’s participation in decision making was not associated with the use of prenatal and postnatal care, while another in another study, Deo et al11 found that autonomy is associated with ANC services in Eastern Nepal. However, some studies in Nigeria and sub-Saharan Africa (SSA) found that women autonomy is associated with maternal healthcare services.12,13,14

There are a number of other studies that have investigated the factors responsible for poor maternal healthcare services utilization and many of these studies used the classical regression analysis. For instance, a study conducted by Bamiwuye et al15 used composite score to obtain an overall index for autonomy by adding the dichotomous variables in all the situations for which data were available to have a minimum score of ‘0’ and maximum of ’4’ with higher scores meaning ‘more autonomy’. These were further dichotomized into ‘0’ (less autonomy) and ‘1’ (more autonomy). However, to our knowledge, studies adopting factor analysis are not available. Factor analysis (or principal component factor) is a more robust method that captures the totality of the constituents including their interactions by examining the correlation matrix to capture the underlying factors that explain substantial amount of variations.

1.2 Aims of the study

The aim of this study therefore was to determine the patterns of women decision making autonomy and establish how it relates with maternal healthcare utilization in Nigeria using factor analysis, factor score, and logistic regression with and without adjusting for confounding variables.

2.1 Study variables

The outcome variable (Dependent Variable) is the number of times a woman attended ANC during the last pregnancy. This was categorized as ‘0’ if she attended less than 4 times and ‘1’ if she attended at least 4 times. The principal independent variable (predictor variable) is the decision making status of the woman and was dichotomized as ‘0’ representing ‘low autonomy’ and ‘1’ representing ‘more autonomy’. Other confounding variables of interest examined include: age of respondent and age at 1^st marriage, where classified in group as (15-24, 25-34 and 35+), others were marital status (never in union, formerly in union and currently in union); number of children ever born (0-4, 5+) and number children living (0, 1-2, 3-4, 5+), the household wealth status classified as ‘low’ or ‘high’, education status (no education, primary education and secondary education and above); place of residence (whether rural or urban); and region (North East, North West, North Central, South East, South West and South South).

2.2 Statistical analysis

Four levels of analyses procedures were adopted:

• Level one, at the univariate level, percentage frequency distribution of the study sample was used to show the distribution of respondents by their characteristics and ANC prevalence.

• Level two, to construct autonomy status, factor analysis/score were used while considering the three dimensions as dichotomous where ‘1’ means participating and ‘0’ not participating. In view of this, we extracted the commonalities, the proportion of the variance explained, the loading of the variables on the factor, the factor score coefficients for each variable and the test of model appropriateness.18

• Level three, to establish the relationship of the independent variable (decision making autonomy index derived from factor analysis) with maternal health care utilization, we used chi square at 5% level of significance.

• Level four, logistic regression analysis was used to establish the likelihood effects of principal variable (unadjusted) and (adjusted for confounding factors) with ANC visits. Stata14SE was used for the computation.19

2.3 Ethical approval

Being a secondary data survey, the ethical permission to use the data for this study was obtained from Opinion Research Corporation (ORC) Macro International, Incorporated, Calverton, USA with its approval for survey already approved by Ethics Committee of ORC Macro Inc. and by the National Ethics Committee of Federal Ministry of Health, Nigeria.7

3.1. Factor analysis of autonomy status

Using principal component factor of factor analysis, one factor was extracted accounting for 81% of the total variance. The Bartlett test of sphericity for decision making autonomy among Nigerian women was highly significant (p<0.000) with chi-square of 44793 indicating homogeneity of variance by the decision making patterns.

The following equation was used to construct the autonomy score (AS):

The resulting values which lie between 0 and 1.1 were rescaled and multiplied by 100 to give values between zero and 100. This was finally dichotomized into 50/50,5,15 resulting into autonomy index used for the analysis classified as ‘0’ low autonomy and as ‘1’ more autonomy as presented below in Table 1a and 1b.

This puts the prevalence rate of low decision making autonomy among Nigerian women to 60% (table 1b).

3.2. Distribution of participant’s characteristics

As show in table 2, the mean age of respondents was 28.86 years with standard deviation of 9.68 years. More than 90% of women with age at first marriage between 15 and 24 years participated in the survey. The prevalence of ANC visits meeting the WHO standard (4 times and above during pregnancy) among Nigerian women was slightly above 52%. Over 11,000 of the women in the survey do not have any living child and 22% have more than 4 children. The table also reveals that about three quarters of the women participants are currently in union with about 5% formerly in union, but now no longer in union. An equal number of women who have never had a child also had five or more children.

Table 3 shows that ANC visits at the WHO standard varied with a number of the background variables of participants. Other variables that were significantly associated with ANC include: age in group, number of children ever born, household wealth index, number of children living, level of educational attainment, etc. The proportion of women whose age at first marriage was 35+ years (86%) that met WHO ANC standard was more than those for any other age group. ANC utilization was significantly lower among the rural women (38.9%) than among the urban women (77.6%). ANC utilization significantly increased among women with higher education from 2699 with no formal education to 5377 for those who have completed secondary education and above.

3.3 Logistic regression analysis

4.1 Discussion

We applied factor analysis to construct the Autonomy Index (AI) for the respondents using the principal component factor option. We used the scores (expressed in 10 scales) for the three dimensions of women decision making autonomy (deciding on their own health care; making major household purchases, and visits to family or relatives) as collected in 2013 NDHS. These variables were further converted into dichotomous variables of ‘0’ & ‘1’, where ‘1’ means having the item and ‘0’ not having it.

The kmo of 0.736 measures the adequacy of sample-size of women decision making autonomy indicating that a higher correlation existed between different decision making autonomy statuses to warrant using factor analysis.20 The finding that only about 40% Nigerian women has autonomy in deciding on issues that pertain to her health, large purchases for family and visit outside the home either singly or jointly with her husband is relatively low. Furthermore, the principal exposure variable, women autonomy was found to be significantly associated with ANC utilization in Nigeria. This agrees with similar studies in Nigeria12, 13 and in some Asian countries,8, 9 using different analytical approaches, but the result was at variance with other findings in Nepal.10 The significance of women autonomy as a risk factor for ANC utilization even after adjusting for other confounding factors makes it very relevant especially in resource limited settings like Nigeria.

4.2 Limitations

However, the interpretations of the results from this study are subject to a number of limitations. First, the study is cross sectional and as such causal effects could not be determined. Second, the study did not examine the independent predictor effects of other significant independent variables. Third, in view of the multi ethnic diversities of Nigeria, cultural impediments to ANC were not examined. Fourth, the constituents of women decision status as collected by NDHS were in 3 dimensions. These could have been more to further justify the use of factor analysis. Overall, the strength of the results from the study draws upon the nationally representativeness of the large sample size of the NDHS. Future studies should identify more research gap in the areas of identifying cultural impediments in the use of ANC in Nigeria. For example, interactions of both maternal and paternal variables can be investigated and how much both household and paternal variables can explain the relationships between maternal characteristics and ANC.