Participants
The present cross-sectional, population-based study is based on the National Family Health Survey (NFHS-4) data, conducted between 2015 and 2016. The NFHS-4 was conducted by the Ministry of Health and Family Welfare (MoHFW) of India, which designated the International Institute for Population Sciences (IIPS) Mumbai as the nodal agency to be responsible for the coordination and technical guidance for the survey, which was based on the DHS methodology. The surveys have been carried out since 1992 and provides state- and national-level estimates, while the latest round (NFHS-4) for the first time provided estimates at the district level, of basic demographic characteristics, reproductive health, maternal and child health, family planning, HIV/AIDS and gender-based violence, among others. Both men and women were interviewed, but for the scope of this study only women respondents have been included. Data on husband’s characteristics were only obtained from women respondents when their husbands were also eligible for interviewing. All women aged 15–49 who were usual members of the selected households or who spent the night before the survey in the selected households were eligible for interviewing. A total of 699,686 women aged 15–49 were interviewed to obtain a nationally representative sample.
Our study participants, however, consisted only of currently married, sexually active, never pregnant women aged 15–24. Never pregnant was defined as women who had never given birth, who are currently not pregnant, and who had never had an abortion, miscarriage or stillbirth. Women were considered sexually active if they had ever had sexual intercourse with their current partners. Women currently married but without a gaunaFootnote 1 performed were excluded from the sample. Based on these criteria, our study participants consisted of 16,475 young married women aged 15–24. Women who had missing data on partner and IPV characteristics were excluded from our logistic regression model analysis.
Sampling
To obtain a nationally representative sample of eligible households, a stratified, two-stage, cluster sampling strategy was used. The survey followed a uniform sampling design for each state; nonetheless, oversampling was required in certain areas to ensure national representativeness. Data on IPV and data on men were collected from a subsample of around 15% of households that were selected for the state-level module. If more than one eligible woman lived in a selected household, only one woman was randomly selected for the IPV module of the questionnaire. The module was only administered if complete privacy was obtained, and it was stopped if the respondent requested it at any time or if someone else was trying to listen to the respondent’s answers. In total, 83,397 women were selected for this module and 79,729 completed it (response rate of 95.6%). Pre-determined weights for IPV were assigned on the subsample for nationally representative estimates at the analytical stage. Further details about the sampling strategy, data collection and survey instruments can be found in the final report of the NFHS-4 and on the Demographic Health Survey (DHS) Program website [6, 25].
Variable characteristics
In the analysis, the outcome measure was preferred waiting time for the first childbirth. Data from the NFHS-4 dataset for this measure was solely accessible in categorized form according to the following categories: less than 12 months, 1 year, 2 years, 3 years, 4 years, 5 years and 6 years, based on the response to the question “How long would you like to wait from now before the birth of (a/another) child?” For our analysis, the outcome was dichotomised as: 1) a preferred waiting time of less than 2 years, and 2) a preferred waiting time of 2 years or more. The reasons for this were two-fold. First, the mean waiting time in India is 2 years [14]; and, second, because when calculating family planning indicators, such as the unmet need for family planning, women with a need for family planning are those who want to delay their next pregnancy for 2 years [26].
The independent factors included sociodemographic, partner and IPV characteristics. The sociodemographic factors used during the analysis were age group (15–19 years, 20–24 years), age at marriage (younger than 18, 18 or older), place of residence (urban, rural), religion (hindu, muslim, others), caste (scheduled tribe/caste [SC/ST], other backwards class [OBC], general), education level (no education, primary, secondary, higher), wealth index (poorest, poorer, middle, richer, richest), and media exposure to family planning (yes or no). The partner characteristics were education level of husband, age of husband, and spousal age difference, which was calculated by subtracting the age of the respondent from the age of her husband. Women were considered exposed to family planning on media if they had heard or seen a family planning message on radio, television, in a newspaper or magazine, or on a wall painting or hoarding in the past few months. IPV characteristics were exposure to physical, emotional, and/or sexual violence and/or controlling behaviour from husband and each was classified as either “yes” for ever exposed or “no” for never exposed.
Information about IPV from currently married women was obtained by asking them if their current husband ever did any of a series of violent behaviours. Exposure to physical IPV was present if their husband had ever pushed, shook, or threw something at them; slapped them; twisted their arm or pulled their hair; punched them with his fist or with something that could hurt them; kicked, dragged, or beat them up; tried to choke or burn them on purpose; or threatened or attacked them with a knife, gun, or any other weapon. Exposure to sexual IPV was present if their husband had ever physically forced them to have sexual intercourse with him even when they did not want to; physically forced them to perform any other sexual acts they did not want to; or forced them with threats or in any other way to perform sexual acts they did not want to. Exposure to emotional IPV was present if their husband had ever said or done something to humiliate them in front of others; threatened to hurt or harm them or someone close to them; or insulted them or made them feel bad about themselves. Exposure to controlling behaviour was present if their husband was jealous or angry if they talked to other men; frequently accused them of being unfaithful; do not permit them to meet their female friends; tried to limit their contact with their families; insisted on knowing where they are at all times; or did not trust them with any money.
Statistical analysis
The statistical analysis was done using IBM-SPSS Version 21.0. Contingency tables and a chi-square test of independence were used to determine whether there were any associations between the preferred waiting time for the first childbirth and the respondent’s sociodemographic characteristics (age group, place of residence, religion, caste, education level, wealth index, media exposure to family planning, and age at marriage) the partner’s education level, the age of the husband, and exposure to intimate partner violence (physical violence, emotional violence, sexual violence, and/or controlling behaviour). Independent t-tests were used to determine differences between preferred waiting time for the first childbirth and the means of our continuous variables (partner’s age and age difference between spouses). A p-value of less than 0.05 was consider statistically significant. The percentages reported in the descriptive tables were weighted to account for the NFHS-4 survey methodology.
A bivariate logistic regression analysis was performed to calculate the unadjusted odds ratios to determinate associations between sociodemographic, partner and IPV characteristics of the respondents and their preferred waiting time for their first childbirth (less than 2 years vs. 2 years or more). Then, a step-wise multivariable logistic regression analysis was performed to calculate the adjusted odds ratios (ORadjusted) between the same variables used for the bivariate logistic regression analysis. The variables selected for the multivariable logistic regression analysis were based on a review of the current literature and all measures were adjusted for simultaneously [23, 24]. Before performing the multivariable logistic regression analyses, a matrix of correlations of estimates was generated and we used a threshold of 0.8 to rule out multicollinearity among our independent variables; none of them were excluded from the model. 95% Confidence Interval (95%; CI) have been used to indicate statistical significance.
Ethical considerations
Informed consent was obtained from all the participants prior to the administration of the NFHS-4 survey. NFHS-4 data are freely available to the public upon request. We submitted our request stating our aims and objectives and permission was granted to download the women’s dataset.