Data source
This study utilized data from the TDHS-2018, a nationally representative, household-based survey conducted every five years. The TDHS-2018 conducted interviews with 7,345 women aged 15–49 using a standard DHS questionnaire. We analyzed data from currently married women (n = 5,156) in this study. To ensure the representativeness of the sample, all analyses were weighted. As a result, the tables show a total of 4,820 currently married women.
Dependent variables
The study included eight dependent variables, as described below. Table 1 also shows the classification of these variables and the number of women examined for each variable.
1) Total number of births: It describes the total number of children ever born to women.
2) Ideal number of children: This refers to the number of children women desire to have, regardless of their current family size. In cases where women provided non-numerical responses regarding their ideal number of children (n = 30), the mean ideal number of children (3.0) was assigned.6,24
3) Having more children than the desired number: This variable indicates the difference between the woman's actual number of living children and her ideal number of children. It was calculated by subtracting the ideal number of children from the number of living children. If the difference was greater than zero, the woman was regarded as having more children than her desired number. Given that younger women are less likely to achieve their fertility goals, only married women aged 35 and above were included in the analysis.6,24
4) Unwanted birth: This variable was used to determine whether the most recent child born within the previous five years was wanted or unwanted (wanted later or not wanted at all).
5. Demand for family planning satisfied with modern methods: This variable refers to the proportion of women using modern contraceptive methods compared to the total number of women with a demand for family planning (the sum of women with unmet needs for family planning and women using any method).25
6) Contraceptive category: Contraceptive methods were divided into female-only methods (female sterilization, pills, injectables, intrauterine devices and implants) and couple methods (condoms, withdrawal and periodic abstinence). The main distinction between the two categories is that couple methods necessitate a certain level of awareness, support, involvement, or cooperation from the husband.26
7) Induced abortion: It refers to whether the woman has ever had a pregnancy terminated by induced abortion.
8) Miscarriage: It refers to whether the woman has ever had a pregnancy terminated by miscarriage.
Table 1. The classification of the dependent variables and the number of women analyzed in the study
Dependent variables
|
Classification
|
Study population (weighted)
|
1. Total number of births
|
0-3; ≥ 4
|
Currently married women (n=4820)
|
2. Ideal number of children
|
0-3; ≥ 4
|
Currently married women (n=4820)
|
3. Having more children than the desired number
|
Yes; No
|
Currently married women aged ≥35 (n=2693)
|
4. Unwanted birth
|
Wanted at that time; Wanted later; Not wanted at all
|
Women who gave birth to a child in the last five years (n= 2002)
|
5. Demand for family planning satisfied with modern methods
|
Yes; No
|
The total number of women with a demand for family planning (n= 3924)
|
6. Contraceptive category
|
Couple methods; Female-only methods
|
Married women currently using any contraceptive method (n= 3358)*
|
7. Induced abortion (at least one)
|
Yes; No
|
Currently married women (n=4820)
|
8. Miscarriage (at least one)
|
Yes; No
|
Currently married women (n=4820)
|
* Women using the folkloric method (n= 8) were not included.
Explanatory variable
The study's main independent variable was women's social independence, which was measured using the SWPER Global Index. The SWPER index has three domains: 1) attitudes toward violence, 2) decision-making, and 3) social independence. We excluded the attitudes toward violence domain because a relatively low proportion of women (approximately 10%) endorsed statements about violence in the TDHS-2018. We also excluded the decision-making domain because the TDHS-2018 only asked questions about one of its three items. As a result, we included the social independence domain, as the TDHS-2018 encompassed all items within it. The original social independence domain comprises six items designed to identify the prerequisites for women to achieve their goals and the components that contribute to agency and autonomy. We found that two items, ‘internet usage’ and ‘do you have money that you can spend by yourself?" were highly correlated with the items of the original social independence domain. Subsequently, we decided to use the modified form, adding these two items (Table 2). In the TDHS-2008, 7.1% of currently married women had never given birth. Their age at first birth was imputed through hot-deck imputation.12 Principal component analysis was employed to estimate the factor loadings of the items, and individual standardized scores were calculated accordingly (Supplementary file 1). Social independence scores were divided into quintiles, with the lowest 20% of scores representing the least empowered women and the highest 20% representing the most empowered women.
Table 2. Items and code/units of the original and modified social independence domains
|
The original social independence domain of the SWPER Global Index
|
The modified social independence domain
|
Items
|
Code/unit
|
Code/unit
|
1. Frequency of reading a newspaper or magazine
|
Not at all = 0; less than once a week = 1; at least once a week and more = 2
|
Not at all = 0; less than once a week = 1; at least once a week and more = 2
|
2. Woman’s education in completed years of schooling
|
Years
|
Years
|
3. Age of woman at first birth
|
Years
|
Years
|
4. Age at first cohabitation
|
Years
|
Years
|
5. Age difference: woman’s minus husband’s age
|
Years
|
Years
|
6. Education difference: woman’s minus husband’s years of schooling
|
Years
|
Years
|
7. Internet usage
|
-
|
No = 0; irregularly = 1; regularly = 2
|
8. Whether the woman has money to spend by herself
|
-
|
No= 0, Yes= 1
|
Covariates
These variables included women's age, place of residence (rural/urban), region (west, south, central, north, east), and household wealth categories (poorest, poorer, middle, wealthier, wealthiest). To maintain focus on the study's primary objective, we excluded the associations between these variables and reproductive health outcomes from the tables.
Data analysis
In the preliminary analyses, hot-deck imputation was initially applied for nulliparous women using R software, version 4.2.1. Principal component analysis was utilized to estimate the factor loadings of the items and individual standardized scores using Minitab 17 software. Chi-squared tests were used to compare the proportion of reproductive outcomes by social independence quintiles. Prevalence ratios and corresponding 95% confidence intervals (PRs, 95% CI) were estimated to explore the relationship between dependent variables and social independence quintiles using robust Poisson regression analysis with R software. PRs were adjusted by the woman’s age, place of residence, region, and household wealth quantiles. The statistical significance of the results was accepted when p < 0.05 for all comparisons. All estimates were weighted using the weight variable provided by the TDHS-2018.
Ethical approval: The Hacettepe University Ethics Commission granted ethical approval for the collection of TDHS-2018 data to the Hacettepe University Institute of Population Studies, which conducted the survey. The dataset analyzed in this study was anonymized and made publicly available. We obtained permission to use the data from the official website of the Hacettepe University Institute of Population Studies (https://hips.hacettepe.edu.tr/en/menu/demographic_and_health_survey_serie-101).