Menu
Menu
Menu
Menu
Background: Globally, there has been a decline in female child marriage (FCM) from 1 in 4 girls married a decade ago to approximately 1 in 5 currently. However, this decline is not homogenous because some regions are still experiencing a high prevalence of FCM. As such, the United Nations reiterated the need for concentrated efforts towards ending FCM to avoid more than 120 million girls getting married before their eighteenth birthday by 2030. Following this, we examined the prevalence and factors associated with FCM in Nigeria using multi-level analysis. Methods: We used cross-sectional data from the women’s file of the Nigeria Demographic and Health Survey (NDHS) conducted in 2018. A sample of 4143 young women aged 20–24 was included in the study. Our analysis involved descriptive, chi-square (χ2) and multi-level analyses. Results were presented in percentages, frequencies, and adjusted odds ratios (aOR) with their respective confidence intervals (CIs). Results: The prevalence of FCM in 2018 was 65.30%. Young Muslim women aged 20–24 [aOR = 1.40; 95% CI (4.73–7.52)], those with parity between one and two [aOR = 5.96, 95% CI 4.73–7.52], those residing in North East [aOR = 1.55; 95% CI (1.19–2.10)] and North West [aOR = 1.59; 95% CI (1.18–2.16)] had a higher odd of practicing FCM respondents with secondary education and above [aOR = 0.36; 95% CI (0.29–0.46)], those within the richer wealth index [aOR = 0.35; 95% CI (0.23–0.54)] and young women living in communities with high literacy level [aOR = 0.74; 95% CI (0.59–0.92)] were less likely to get married before age 18 years. Conclusion: Our findings indicate that FCM is high in Nigeria. Formal education, being rich and living in communities with high literacy levels were some protective factors that can be strengthened to ensure that FCM is reduced or eliminated in Nigeria. On the other hand, residing in North-East or North-West and having children between one and two were some prevailing factors that exacerbated the odds of experiencing FCM in Nigeria. Therefore, attention should be channelled towards mitigating these prevailing negative factors.
The 2018 Nigeria Demographic Health Survey (DHS) was used for this study. DHS is a nationwide survey executed every five years. The surveys focused on key maternal and child health measures such as marital age, FGM, unintended pregnancy, skilled birth attendance, contraceptive use, intimate partner violence and immunization among under-fives. A stratified dual-stage sampling approach was employed during the survey to collect information on the respondents. Furthermore, a cluster sampling process (i.e., enumeration areas [EAs]) was involved in the survey, followed by systematic household sampling within the chosen EAs. The sample framework generally excludes nomadic and institutional groups, such as inmates and hotel occupants [28]. The women’s files with responses by women aged 15–49 were accessed [29]; however, this study participant was limited to young women between the age of 20–24 as inclusion criteria [3, 30]. The eligible sample size for this study was 4,543 women aged 20 to 24. The study included only respondents who had complete information on the variables of interest. The outcome variable for this study was “Age at first marriage”. To derive this variable, respondents were asked the age at which they first got married. Respondents were categorized as married before age 18 years if the response falls between age 1 to 17 and categorized as married 18 years and above if respondent’s response falls between 18 years and above [1, 31, 32]. Eleven explanatory variables were considered in this study and were grouped into individual-level variables and household/community level variables. These variables were determined by priori from previously published studies and the availability of the variables of interest in the datasets before the selection of all the explanatory variables [33, 34]. The individual-level factors were educational level, working status, religious, ethnicity, ever circumcised, parity (children ever born), and media exposure. Educational level was coded as ‘no education,’ ‘primary education’, and ‘secondary/higher education,’ while ‘working’ and ‘not working’ were the categories for working status. Religious were recoded as “Islam”, “Christianity” and “Traditionalist/others”, ethnicity was recoded into three major ethnic groups in Nigeria, namely, “Yoruba”, “Igbo”, “Hausa” while the remaining tribes were coded as others. Ever circumcised was coded as “Yes” for young women who were circumcised and “No” for those that were not circumcised, while parity was coded as “No child” for young women with any child, “1–3” for young women with children between 1 and 3, “4–6” for young women between 4–6. ‘Frequency of reading newspaper/magazine, listening to the radio and watching television were recategorized as media exposure. Those who didn’t engage in any of these were coded ‘No’ while those exposed to at least one of these were coded ‘Yes’. The household/community level variables were the place of residence, region, wealth quintile, and sex of the household head. These selected household/community level variables were based on their categorization in the DHS [35]. Place of residence was coded as ‘urban’ and rural, the Sex of the household head was coded as ‘male’ and ‘female’ while the wealth quintile was computed using the standard DHS data on household ownership by selecting properties such as bicycles, house building materials, television, type of access to water and sanitation facilities. To make households on a continuous relative wealth scale, a composite variable, wealth status, was generated from these assets through Principal Component Analysis (PCA), and we divided the households into five quintiles of wealth: poorest, poorer, middle, wealthier, and wealthiest [36]. The analyses began with a descriptive analysis table to display the prevalence of age at first marriage, and a chi-square test of independence (χ2) was used to show the association between age at marriage and the explanatory variables (see Table Table11). Distribution of age at first marriage by explanatory variables Weighted NDHS, 2018 A multilevel logistic regression model (MLRM) was used to examine the association between the individual and household/community factors and age at marriage in Nigeria using the recent DHS dataset. The Stata command “melogit” was used in fitting these models. A 2-level model for binary responses was specified, reporting age at marriage below 18 or not for young women aged 20–24. At the first level, women were modelled from households (Individual level) and at the second level, households were modelled from PSUs (household/community levels). Four models were constructed in this study. The first model was the empty model/null model (Model 0), which is the model that shows the variance in the outcome variable, which is the age at marriage, attributed to the clustering of primary sampling units (PSUs), however, this model has no explanatory variable included. The second model contained only the individual-level factors (Model I), while the third model contained the household/community-level factors (Model II). The final model was the complete model (Model III) that simultaneously controlled for the individual and household/community factors. The MLRM consists of fixed and random effects [37, 38]. The fixed effects (measures of association) showed results of the association between the selected explanatory variables and the outcome variable (age at marriage) and were reported as adjusted odds ratios (aOR) with their 95% confidence intervals (CIs), while the random effects (measures of variations) were assessed using Intra-Cluster Correlation (ICC) [39]. This means that variations in factors influencing FCM in Nigeria were drawn from within PSUs, which enabled us to draw an appropriate conclusion [40]. The LR test was used to check for model adequacy. Both Akaike’s Information Criterion (AIC) and Bayesian Information Criteria (BIC) were used to measure how well the different models fitted the data. The sample weight (v005/1,000,000) was applied to correct for over-and under-sampling, while the ‘svy’ command was used to account for the complex survey design and generalizability of the findings. The analyses were carried out with Stata version 16.0 (Stata Corporation, College Station, TX, USA). Ethical approval was granted by the Institutional Review Board of ICF International, and the DHS Program approved the use of the dataset for this study, which the dataset is freely available at https://dhsprogram.com/data/available-datasets.cfm upon request. Individual informed consent was sought from all participants during data collection. All methods were performed according to the relevant guidelines and regulations in line with the World Medical Association Declaration of Helsinki Ethical principles [41].
N/A
