Methods of analysis

The methods of analysis used were multinomial logistic regression and linear regression. Data analysis was carried out using STATA v.14, where descriptive statistics were used to provide sample characteristics. The first outcome variable, timing of the first ANC visit was analysed using bivariate and multinomial logistic regression since it was a three outcome variable coded as none, late and early. We fit one model to predict the linkage between pregnancy wantedness and timing for the first ANC visit in the presence of other explanatory variables. The second outcome variable, frequency of antenatal care visits is analysed using bivariate and multivariate logistic regression to determine the relationship with pregnancy wantedness in the presence of other explanatory variables.