Data source
We used data from the 2016 UDHS which was a nationally representative population based cross-sectional survey funded by the United States Agency for International Development (USAID) implemented between 20th June 2016 to 16th December 2016 by the Uganda Bureau of Statistics. UDHS is a periodical survey that is carried out every five years as part of the MEASURE DHS global survey and collects Information on demographic, health and nutrition indicators. The UDHS used stratified two-stage cluster sampling design that resulted in the random selection of a representative sample of 20,880 households 11. A detailed explanation of the sampling process is available in the UDHS 2016 report 11. Women who were aged 15–49 years and were permanent residents of the selected households or visitors but had stayed in the household the night before the survey were eligible for inclusion in the survey.
Our secondary analysis included women who had given birth in the last 5 years preceding the survey and had attended at least one ANC contact. Out of the total 18,506 women in the survey, 10,263 women had given birth in the last 5 years preceding the survey, and a final weighted number of 9,957 women had attended at least one ANC contact. Where a woman had more than one birth in the last 5 years, ANC information for the most recent birth was considered. UDHS 2016 had four different questionnaires that included; household, men’s, women’s and the biomarker questionnaires. Our secondary analysis used data collected by the women’s and household questionnaires that focused on women’s background characteristics, reproductive health, domestic violence and nutrition. Women were asked a number of questions about antenatal health care that included; type of ANC provider, the number of ANC visits, the timing of the first ANC visit and the components included in the ANC provided.
Study Setting
Uganda currently has a population of about 44 million people 15. Uganda’s health system has six levels ranging from the highest level of national referral hospitals to the lowest level at the community level 16. Agriculture is the main source of income for the 84% of Ugandans living in rural areas 16. The Ugandan government abolished user fees in 2001 in all public health facilities. However, the health system still faces inadequate staffing, low pay, shortage of medicines and poor infrastructure that have negatively affected health service provision and utilisation 17. The private sector owns almost half of Uganda’s health facilities 13. However, less than 20% of these private health facilities are at the level of health centre III or above, which is the minimum level at which maternal health services are to be provided 13.
Variables
Outcome variable
Based on WHO’s definition of ‘effective ANC’ and previous studies 9, 18, 19 regarding quality and access of ANC, a final criterion was developed for determining optimal access to ANC. The outcome variable ‘optimal access to ANC’ was measured as a composite index and was dichotomized. The outcome optimal access to ANC was coded as 1 and suboptimal access as 0. Three indicators were used to define access to ANC and these included; timing of initial visit, number of visits and provider of the ANC. Optimal access to ANC was considered as ANC package that was initiated in the first trimester, included a minimum of four ANC contacts and was provided by a skilled health provider 13, 20. Timing of initiation of contacts describes how many months of pregnancy a woman was during her first ANC contact. This was initially numerical ranging from 0-10 months and was categorized as: ‘first three months’ (coded as 1) and ‘more than three months’ (coded as 0). Total number of ANC contacts is the number of times a woman received ANC and was initially a numerical variable categorized as ‘four or more visits’ (coded as 1) and ‘less than 4 visits’ (1-3 visits and coded as 0). We considered a minimum of four visits since it was the recommendation of WHO at the time of the survey. Type of ANC provider referred to the service provider who provided ANC to the woman in her last pregnancy. It was categorized as the ‘skilled provider’ (doctor, clinical officer, nurse and midwife coded as 1) and ‘non-skilled provider’ (coded as 0). Where multiple providers provided care to one woman, the highest skilled provider was recorded. The total score from the three indicators for each woman was determined, with the maximum score being ‘3’ if the woman initiated ANC in the first trimester, had a minimum of four ANC visits and was seen by a skilled provider. Score of 3 was considered as optimal ANC access and any score below 3 was considered as suboptimal ANC access.
Explanatory variables
Measures of women’s empowerment
Four indices were created to measure empowerment of women: decision making, economic empowerment, sexual empowerment and exposure to media indices. Women’s empowerment indices were measured as composite scores 1, 19, 21.
Decision making included women’s ability to be involved in making decisions regarding; their own health; large household purchases; visits to their family and control over their earnings 5. Re-coding to the responses was done to have two categories (1 = woman involved in decision making alone or with partner, 0 = woman not involved in decision making). To create an index, we added up all the scores for each woman with the total score ranging from 0 to 4 and we finally categorized the score into four groups. The highest score was four which meant that the woman was involved in the decision making for the four used indicators. Medium decision making ability meant that women were involved in 2 or 3 indicators, low decision making meant that the woman was involved in only one indicator and no decision making meant that the woman was not involved in any decision making 21, 22. Decision making had about 1,846 missing responses since these questions were asked during the domestic violence survey sessions and yet not all women in the UDHS were included in the domestic violence section of the survey. These missing observations were assumed to be zero 19 and hence we risked overestimating low decision making. To ensure that this doesn’t affect our findings, we did a sensitivity analysis were we only considered women who had decision making responses and excluded those who were missing decision making responses. This showed no significant difference with the original analysis and more details are included in the sensitivity analysis section of the results. For background characteristics, we provided frequencies of decision making considering only women who had responded to decision making questions.
Economic empowerment included women’s owning of a house, land and the type of earning from her work 22, 23 . Each of the three economic empowerment indicators was re-coded 1 if the women owned a house or land (either alone or jointly with a partner) or received cash payment for their work and 0 if not owning a house, land or cash payment for work. To create an index, we added up all the scores for each woman with the total score ranged from 0 to 3 and we finally categorized the score into four groups. The highest score was 3 which meant that the woman was owning a house, land and earned cash for her work. Score 2 was considered medium economic empowerment, score 1 considered low economic empowerment and 0 no economic empowerment.
Sexual empowerment was considered as women’s ability to refuse sex and ask a partner to use condoms 23, 24. Responses were coded (1 if the woman could refuse sex or ask for a condom and 0 if the woman could not) and sexually empowered women were those who were able to refuse sex or ask their partners to use condoms. To create an index, we added up all the scores for each woman with the total score ranged from 0 to 2 and we finally categorized the score into three groups. Highest score was 2 and these were considered to have high sexual empowerment, score 1 low sexual empowerment and 0 as no sexual empowerment.
Exposure to media was considered as the women’s ability to have the opportunity to read newspaper or a magazine, listen to the radio and watch TV. Responses were re-coded (1 if the woman was exposed to newspaper, radio or TV and 0 if the woman was not). To create an index, we added up all the scores for each woman with the total score ranged from 0 to 3 and we finally categorized the score into four groups 21. A value of 0 meant no access to any of the three, 1 exposure to only one medium (low) , 2 as exposure to only two (media) and 3 as exposure to all the three media channels (high) 21, 22.
Other explanatory variables or potential confounders
We included determinants of ANC access basing on available literature and data 1, 25, 26. Ten variables were considered and of these, two were community level factors that included; place of residence categorized into rural and urban, and region of residence categorized into Central, East, West and North. Two household level factors included; household size which was categorized into less than six and six and above and wealth index that was categorized into quintiles that ranged from the poorest to the richest quintile. Wealth index was calculated by DHS from information on household asset ownership using Principal Component Analysis. Six individual/maternal level factors were included in the analysis and these were; age (categorized as 15-24, 25-34, 35-49), parity (categorized as 0-4 and above 4), working status (categorized as working and not working), marital status (categorized as married and unmarried), health insurance (categorized as having insurance and not having) and level of education (categorized as no education, primary, secondary and tertiary levels).
Statistical analysis
We used the SPSS analytic software version 25.0 Complex Samples package for this analysis. Weighted data was used to account for the unequal probability sampling in different strata. Frequency distributions were used to describe the background characteristics of the women. We used cross-tabulations to examine associations between ANC access and women’s empowerment indicators and socio-demographic factors. Pearson’s chi-squared tests were used to investigate the significant differences between ANC access and the explanatory variables with the level of statistical significance set at p-value < 0.25.
Bivariable logistic regression was also conducted and we present crude odds ratio (COR), 95% confidence interval (CI) and p-values. Independent variables found significant at p-value less than 0.25 were included in the multivariable model. Independent variables that were non-significant at bivariable analysis level but had been shown to be associated with access to ANC in previous studies were also included in the multivariable logistic regression models. Two models were constructed in the multivariable analysis; one with only women empowerment variables and the final model that included the women empowerment and other variables. Adjusted odds ratios (AOR), 95% Confidence Intervals (CI) and p-values were calculated with statistical significance level set at p-value < 0.05. We conducted sensitivity analysis where we separated the three indicators of optimal ANC access and we analyzed the association between women empowerment and each of the indicators (timing, frequency and provider) separately. Further sensitivity analysis was done with eight or more ANC contacts as the outcome and lastly, analysis where we only considered women with decision making responses and excluded those who had these responses missing. All variables in the model were assessed for collinearity, which was considered present if the variables had a variance inflation factor (VIF) greater than 10. However, none of the factors had a VIF above 3.