Determinants of early marriage among married women in nine high fertility sub-Saharan African countries: a multilevel analysis of recent demographic and health surveys

This study utilized pooled data from the latest Demographic and Health Surveys (DHS) conducted between January 2010 and December 2018 of nine countries in SSA. Niger, Democratic Republic of Congo, Mali, Chad, Angola, Burundi, Nigeria, Gambia, and Burkina Faso were included in this study. These countries were selected because they are the top ten countries with high fertility rates in SSA with fertility rates above 5.0, a higher value than the rate of 4.44 in SSA and 2.47 worldwide [37]. One country (Somalia) with no DHS data was excluded from the analysis. The data for these countries were obtained from the official database of the DHS program, www.​measuredhs.​com after authorization was granted via online request by explaining the purpose of our study. We used the women record (IR file) data set and extracted the dependent and independent variables. The DHS is a nationally representative household survey that uses face-to-face interviews on a wide range of population, health, nutrition tracking, and effect assessment measures. Study participants were selected using a two-stage stratified sampling technique. Enumeration Areas (EAs) were randomly selected in the first stage, while households were selected in the second stage [38]. A total weighted sample of 121,077 reproductive-age women was included in the study (Table 1).

The outcome variable for this study was early marriage, defined as young girls married before their 18th birthday [36, 39, 40]. It was dichotomized and coded as “yes” =1 if the age at first cohabitation among the women occurred before their 18th birthday and “no” =0 if the first marriage was at 18 years and above.

Individual and community level independent variables were included in this study.

Individual level variables; Educational status of respondents, husband education, occupation of respondents, husband occupation,wealth status, media exposure, number of living children.

Community level variables; Community level variables included residences and some were derived from the individual level data of all community members in the primary sampling unit (PSU), which includes the community level poverty, community education, community employment and community level media exposure.

For data analysis Stata version 16 software was used. To ensure the representativeness of the DHS sample and obtain reliable estimations and standard errors, data were weighted (v005/1000000) before analyzing it.

The study fitted four models: the null model with no explanatory variables, model I with individual factors, model II with community factors, and model III with both individual and community factors. As the models were nested, the Intra class Correlation Coefficient (ICC), Median Odds Ratio (MOR) and, deviance (−2LLR) values were used for model comparison and fitness, respectively. Model III was the best-fitting model due to its low deviance. In the multivariable analysis, variables with a p-value less than 0.2 in bivariable analysis were used. Finally, in the multivariable analysis, adjusted odds ratios with 95% confidence intervals and p-values less than 0.05 were utilized to identify factors of early marriage.

Out of the total respondents, 53.85% women were not attended formal education, 67.62% had no work, and 60.62% had media exposure towards early marriage. Among the participants, 44.54% had seven and above family size. With regard to their economic status, 40.42% women were from the poor wealth quintiles and 39.51% were from the rich wealth quintiles (Table 2).

Of the respondents, 68.20% were rural dwellers. More than half (51.56%) of the respondents were from communities with low proportion of poverty level. Half (50.41%) of women had media exposure. Above 50% (50.64%) of participants were from communities having high proportion of community level education (Table 3).

Overall, the prevalence of early marriage in top nine highly fertile sub-Saharan African countries was 55.11% (54.8, 55.4). The prevalence of early marriage ranged from 28.11% in Burundi to 80.77% in Niger (Fig. 1).

Regarding individual level factors, the study found that women with a secondary or higher education were 90% less likely to be married below the age of 18 years than those who had no formal education (AOR = 0.1; 95% CI: 0.09, 0.11) and those women with primary education were 61% less likely to be married below the age of 18 years compared to women who have never had formal education (AOR = 0.39; 95% CI: 0.38, 0.41). Women who were working were 27% less likely to be married below the age of 18 years compared to those who had no working (AOR = 0.73; 95% CI: 0.71, 0 .75). The odds of being married below at 18 years in the rich level were 13% less likely compared to women who live in poverty (AOR = 0.87; 95% CI: 0.85, 0.91). The likelihood of women’s early marriage was high among women who had ≥7 families size (AOR = 1.28; 95% CI: 1.23, 1.33) compared to 1–3 families size.

About the community level factors, married women classified as high Community level poverty were more likely to have early marriage (AOR = 1.09; 95% CI: 1.01, 1.17) than low Community level poverty. In addition the odds of being married below the age of 18 in rural area were 1.16 more likely than living in an urban area (AOR = 1.16; 95% CI: 1.12, 1.21) (Table 4).