Data source and sampling procedure

The data for this study were obtained from the latest DHS data of 16 sub-Saharan African countries from 2015 to 2020 (Angola, Benin, Burundi, Cameron, Ethiopia, Guiney, mail, Malawi, Mauritania, Nigeria, Uganda, Sierra Leone, South Africa, Tanzania, Zambia, and Zimbabwe). The data sets were downloaded in STATA format from the DHS website (http://www.dhsprogram.com). Countries were selected based on the availability of recent standard DHS data. The DHS data is nationally representative data that uses four main standard model questionnaires (the Household Questionnaire, the Woman’s Questionnaire, and the Man’s Questionnaire) to collect data that are comparable across countries. The questionnaire covers basic health indicators such as marriage, fertility, mortality, family planning, reproductive health, child health, nutrition, and HIV/AIDS. It had men, women, kids, and household dataset records. For this study, we used a kids’ recode file.

It uses a two-stage stratified sampling technique samples were stratified by geographic region and by urban/rural areas within each region. In the first stage primary sampling unit, clusters were selected from the enumerations area. The second stage was a complete listing and selecting of a total of 25–30 households from each cluster by equal probability systematic sampling, then the data was collected from each selected household. For our study, we use a total of 35,087 children from selected SSA countries to determine the magnitude and associated factor of incomplete immunization among children aged 12–23 months.

Study design, period, and area

The study was a community-based cross-sectional survey which was conducted in 16 selected sub-Saharan African countries from 2015–2020(Fig. 1).

Fig. 1Study population, inclusion, and exclusion criteria

The Source population was all alive children aged 12–23 months in sub-Saharan African countries and the study population was all alive children aged 12–23 months in sub-Saharan African countries in selected enumeration areas (Fig. 2). A total of 35,087 children weighted samples were pooled from sixteen sub-Saharan African countries (Table 1) to determine incomplete immunization coverage and associated factors among children aged 12–23 months in sub-Saharan African countries. A child who didn’t possess a vaccination card and a mother who didn’t know the vaccination status of her child was excluded from the study.

Fig. 2

Flow chart for selecting of sample from kids record dataset

Table 1 Sample size from each selected SSA countryVariable of the studyOutcome variable

The dependent variable was incomplete immunization coverage among children aged 12–23 months. According to WHO guidelines, children are fully immunized when they receive one dose of Bacillus Chalmette Guerin (BCG), three doses of DPT, three doses of polio vaccines, and one dose of measles-containing vaccination by the age of 9–12 months. We recorded each vaccine as “yes” and “no” for those who received and did not receive respectively. Then we added them up, recode them as vaccine status, and categorized them as “full” for those who received all, “partial “for those who missed at least one dose, and “no” for those children who had never taken a vaccine by the age of 12–23 month.

Independent variable

Socio-demographic characteristics (age, marital status, educational level and employment status of the mother, wealth of the family), and obstetric-related factors (ANC, place of delivery, parity, and use of family planning) were individual-level independent variables. At the community level place of residence and media exposure were considered. We generated a variable media exposure by summing up TV, radio, and newspaper we recorded each of them as “yes” and “no” for those who had been exposed and who hadn’t been exposed respectively. Then we added up and categorized them as “yes” if they were exposed to at least one of the three and “no” for those who had no exposure to at least one media.

Data management and analysisData analysis

The data were kept, cleaned, recorded, and appended by STATA version 14.2 and exported to R version 4.3.0 for analysis and descriptive statistics (percent, proportion, graph, and frequency table). The sample was weighted by sampling weight v005 to make valid inferences. As the data set had a hierarchical nature, we used two-level mixed-effect multinomial logistic regression models. We fitted a total of 4 models the first is a null model without any explanatory variable, the second is for individual level explanatory variable third community level explanatory variable and finally, both individual and community level explanatory variables were fitted. We made a model comparison by using the log-likelihood ratio (LR) as the model was nested with a different number of parameters. The final model was the best-fitted model with a low likelihood ratio (LR). We calculated the intra-class correlation coefficient (ICC) to verify the significance of performing mixed effect analysis instead of simple multinomial logistic regression. Proportional change in variance was also calculated to know the variability explained by random effect in the final model. Variables with p-value < 0.2 in bi-variable analysis in the final model were included for multivariable analysis. In multivariable analysis, a p-value less than 0.05 was considered a statistically significant associated factor for partial immunization.

Ethical consideration

This study was based on the existing survey data collected by the Demographic and Health Surveys (MEASURE DHS) project (www.measuredhs.com). All study participants gave written informed consent before participation and all information was collected confidentially. We requested the DHS program to use the data. The raw survey data and written consent of MEASURE DHS were obtained with authentication letter number 184,828 on May 23, 2023.