We used a data-set series of four nationally representative Pakistan Demographic Health Surveys (1990–91, 2006–07, 2012–13, and 2017–18). These PDHS surveys are conducted in order to gain information on mother and child health, fertility, family planning, reproductive health, nutritional status, and immunisation status. The representative samples and response rates of each PDHS of ever-married women were 6,611 (95%) in 1990–91, 10,023 (95%) in 2006–07, 13,558 (93%) in 2012–13 and 15,509 (96%) in 2017–18. Our study focused on the personal, socio-cultural, community-level, and supply-side factors dealing with the use of contraceptives. To obtain the sample for this study, women who had given birth in the previous five years and participated in the family planning module were selected. The sample from each data set is: 4,092 women in 1990–91, 5,742 women in 2006–07, 7,461 women in 2012–13, and 8,219 in 2017–18 (27–30).
Instrumentation and data classification
Current use of contraceptive methods was defined as the dependent variable, including traditional methods (periodic abstinence [rhythm], withdrawal, and abstinence) and modern methods (pill, intrauterine devices, injections, diaphragm, condom, female or male sterilisation, implants, female condom, foam/jelly, and lactational amenorrhea).
Women’s socio-demographic, demand and supply-side factors were considered as independent variables. Women’s socio-demographic factors were: age, type of residence (urban vs. rural), region, ethnicity, education (no education, primary, secondary, higher), women’s occupation/employment status (not working, unskilled employment [sales, household domestic, unskilled manual], skilled employment [sales, agriculture self-employed, agriculture employees, skilled manual, clerical], and professional [professional/technical/managerial services]). The wealth index of women was calculated through quintiles and categorised into three categories (poor, middle, rich) for this analysis.
Among the demand-side factors, questions regarding media exposure, desire for children, number of living sons, number of living daughters, history of intimate partner violence, decisional autonomy, permission to attend medical or health facilities, as well as unwillingness to go alone and concerns about going to female health providers were included. Exposure to any source of information (TV, radio, newspaper) was computed and recoded as overall media exposure. Desire for more children was categorised into two categories: Either “No” (wanted no more children, sterilised [respondent or partner], declared infecund) or “Yes” (wanted within or after the next two years, unsure about timing, undecided). Intimate partner violence was part of PDHS 2012–13 in which emotional, physical, and sexual violence were included. Women were asked if they had ever faced any humiliating attitude from their husband, physical violence such as being beaten, having their arms twisted, hair pulled or threatened with a harmful weapon such as a knife or gun, or sexual violence from their husbands, which includes forced sex. All forms of violence were coded as binary categories and, thereafter, combined as overall violence by intimate partners. In women’s decisional autonomy, their independent or joint control of income, purchases, healthcare decisions, and visits to relatives were included. Each variable was first coded into two categories: “Yes” if the respondent contributed to any type of decision either individually or jointly, and “No” if she did not participate in any decision-making. Then all types were combined into overall autonomy as women who have any type of autonomy and those who do not have autonomy. The responses to questions regarding permission to attend medical or health facilities were categorised into two categories (“Big problem” and “Not a big problem”). Similarly, going alone to get medical treatment was categorised into the same two categories. The number of living sons has also been taken as a variable to determine whether use of contraceptives is contingent upon son preference. It has been coded as “No living son”, “1–3”, “4–6”, and “7–10”.
Among supply-side factors, the facilitation and provision of governmental and non-governmental family-planning services was measured through questions about distance to health facilities, transport availability, visits by lady health workers, unmet needs and the availability of contraceptives through different sources. Distance to health facilities and transport availability were categorised as either “Big problem” or “Not a big problem”. The sources of family planning were first categorised into public (government hospitals, family-planning clinics, lady health workers), private (private hospitals, pharmacies, and clinics), and others (such as shops, friends or relatives, and traditional practitioners). Unmet needs included those for the spacing and limiting of births.
Data was analysed using SPSS version 24. Absolute numbers and weighted percentages were obtained through descriptive analysis. The relationship between demographic characteristics, demand and supply-side factors, along with current use and non-use of contraceptives was assessed through the Chi square test (X2) on categorical variables. A p-value of < 0.05 was considered statistically significant. Associations between demand and supply-side indicators of non-use of contraceptives were measured using binary logistic regression models to present Odds Ratios (OR). Multivariate analysis was conducted by assessing Adjusted Odds Ratios (AOR) through controlling the demographic variables (age, education, income, wealth, residence). In PDHS 1990–91 and 2017–18, data on ethnicity is missing, so it was not included in the analysis. Furthermore, several variables relating to demand-side factors (media exposure, intimate partner violence, decisional autonomy, permission to attend medical or health facilities, not wanting to go alone for medical help) and supply-side factors (distance to health facility, transport availability, visit by a family-planning worker during the previous 12 months) were not included in the questionnaire in 1990–91 or 2006–07, and therefore were not included in the analysis for these years.