This study employed the 2022 GDHS, a component of the international Demographic and Health Survey (DHS) program that gathers health and demographic information on women, men, and children worldwide [24]. The DHS has been conducted in over 90 LMICs since its establishment, with more than 350 surveys [25]. The 2022 DHS in Ghana is the eighth standard DHS conducted since the first survey in 1988 [24]. The data collection process utilised structured questionnaires, employing a cross-sectional design and a multistage sample procedure [26]. The study comprised 2231 married and cohabiting women aged 40 to 49 years in Ghana. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist [27].

High completed fertility was the outcome variable, defined in the DHS as the total number of children ever born by women. We recoded the variable into low and high total fertility by assigning a value of 1 to individuals with five or more children (high) and a value of 0 to individuals who had four or fewer children (low) based on the 2022 GDHS report [24].

Following a detailed literature review on predictors of high completed fertility [28, 29], we included nineteen explanatory variables based on their availability in the GDHS. The variables consisted of the current working status(working, not working), educational level(No education, primary, secondary/higher), marital status(married, cohabiting), listening to the radio(yes, no), watching television (yes, no), reading newspapers or magazines(yes, no), internet use(yes, no), ethnicity(Akan, Ga/Dangme/Ewe, Mole dagbani, Others), religion(Christian, Islam Traditional/no religion), partner educational level(No education, primary, secondary/higher), the ideal number of children(0–3, 4–5, 6 +), decision-making on healthcare(respondent alone, Joint with partner, others), contraceptive use(yes, no), wealth index(poorest, poorer, middle, richer, richest), sex of household head(female, male), type of place of residence(rural, urban), and region(Western, Central, Greater Accra, Volta, Eastern, Ashanti, Western North, Ahafo, Bono, Bono East, Oti, Northern, Savannah, Northeast, Upper East, Upper West). Further, the variables were segregated into individual and contextual levels. Aside from the latter four variables, grouped as contextual (household and community level variables), the remaining were individual-level variables. Please see the attached Supplementary file that shows the coding scheme of the variables.

The statistical analyses were conducted using Stata software version 17.0 (Stata Corporation, College Station, TX, USA). First, a spatial map was used to show the proportion of women with high total fertility. Next, we determined the distribution of the explanatory variables across the outcome variable and used a Pearson chi-square test to show their associations. Finally, a mixed-effect multilevel binary logistic regression analysis was conducted using four models to identify the high completed fertility predictors. Model I, which did not include any explanatory variables, revealed the changes in high completed fertility ascribed to the clustering at the primary sampling units (PSU). In Model II, the individual level variables were included, while in Model III, the contextual level variables were included. Model IV included all the explanatory variables. The mixed-effect regression analysis yielded results that included both fixed effects and random effects. The fixed-effect analysis revealed the correlation between the explanatory predictors and high completed fertility. The results were reported as an adjusted odds ratio (aOR) and their corresponding 95% confidence intervals (CI). The random effect results, however, indicate the variations in high completed fertility. All four models used the intra-cluster correlation coefficient (ICC) values to determine the variation. All the analyses were weighted, and the svyset command in Stata, which contains the sampling weights, one or more stages of clustered sampling, and stratification, was used to deal with the complex nature of the DHS dataset.

Since the GDHS dataset is publicly available, we did not require a separate ethical clearance process for this study. However, we ensured proper access by obtaining permission to use the data for publication from MEASURE DHS (Monitoring and Evaluation to Assess and Use Results—Demographic and Health Surveys). For more information on ethical considerations and privacy principles related to using DHS data, please refer to the MEASURE DHS website: https://dhsprogram.com/Methodology/Protecting-the-Privacy-of-DHS-Survey-Respondents.cfm.