Individual and community level determinants of unmeet pregnancy complication education among pregnant ANC visitor women in Sub-Saharan Africa: a multilevel analysis of the recent demographic and health survey | BMC Pregnancy and Childbirth

The most recent available data from the Demographic and Health Surveys (DHS) across three countries—Ethiopia, Gabon, and Senegal—was collected in October 2015, as these were the only countries with the unmet pregnancy complication education (UPCE) variable. The DHS datasets are nationally representative surveys of households with at least one pregnant woman of reproductive age, specifically focusing on antenatal care (ANC) visits at health facilities.

To facilitate multilevel analysis, datasets were selected based on the availability of data regarding pregnant mothers’ unmet education about pregnancy complications during ANC visits. In this study, a multilevel mixed-effects logistic regression model was applied to analyze UPCE, enabling the examination of individual- and community-level factors influencing access to education during ANC. By adopting this approach, the study identified barriers contributing to unmet educational needs and their relationship with pregnancy outcomes, providing insights for targeted interventions to enhance maternal health education and outcomes.

A secondary analysis was conducted using a sample of pregnant women who attended ANC, drawing from the recent DHS data. Statistical analysis was performed using Stata 14 software, employing a multilevel logistic regression model to determine associated factors influencing prevalence and weight determinants. Variables with an adjusted odds ratio and a P-value of less than 0.05 were considered statistically significant.

Data collection was carried out by in-country research teams and statistical agencies, with technical assistance. The surveys followed a two-stage sampling design, where enumeration units or ‘clusters’ were first selected from larger regional units within each country, and households were then randomly sampled within these clusters [13].

To ensure data validity and credibility, DHS adheres to rigorous methodologies, including standardized questionnaires and trained data collectors. The national representativeness of the surveys enhances the generalizability of the findings, while established statistical techniques reinforce the reliability of the results. The study defined unmet education on pregnancy complications as its outcome variable, categorizing mothers who did not receive any information about pregnancy complications from healthcare providers during their pregnancies at a healthcare facility as lacking education on pregnancy complications during ANC visits.

Study setting

The Sub-Saharan region of Africa lies south of the Sahara Desert and is divided into four extensive and distinct areas: Eastern Africa, Central Africa, Western Africa, and Southern Africa. It spans approximately 9.4 million square miles and is home to around 1.3 billion people. This study is based on recent DHS survey data from three Sub-Saharan African countries—Ethiopia, Gabon, and Senegal.

Study design and period

This study utilized a multilevel, community-based, cross-sectional design with mixed effects to evaluate unmet education on pregnancy complications among pregnant women in Ethiopia, Gabon, and Senegal. The multilevel approach allowed for the assessment of both individual and contextual factors influencing unmet education, recognizing that outcomes may be shaped by variables at multiple levels, including individual characteristics and community-level determinants.

The community-based perspective ensured that the findings were representative of the broader population, strengthening their relevance for public health initiatives focused on improving maternal education and healthcare access. The cross-sectional nature of the study enabled data collection at a single point in time, facilitating the identification of unmet education prevalence and the exploration of associations between variables without implying causation.

Data were obtained from the Demographic and Health Surveys (DHS), comprehensive and nationally representative surveys conducted every five years, ensuring the robustness and validity of the analysis. The study employed a two-stage sampling design: in the first stage, enumeration units or clusters were selected from larger regional units within each country, and in the second stage, households were randomly chosen from within these clusters, ensuring a diverse and representative sample of pregnant women.

The analysis incorporated data collected from 2015 to 2022, spanning multiple years to enhance reliability and provide a comprehensive understanding of trends and patterns related to unmet education on pregnancy complications. Statistical analysis was performed using Stata 14 software, applying a multilevel mixed-effects approach to account for the hierarchical structure of the data. A P-value of less than 0.05 for the adjusted odds ratio was used to identify statistically significant associations.

By employing this rigorous study design, the research aims to offer valuable insights into the barriers pregnant women face in accessing education about pregnancy complications, ultimately informing public health strategies to improve maternal and fetal health outcomes in the region.

Population and eligibility criteria

The source population comprised women living in Sub-Saharan African countries, while the study population included all pregnant mothers who had attended antenatal visits within the designated enumeration areas analyzed in the study. Eligible participants were those who were currently pregnant and had received care during their antenatal visits, ensuring that the data accurately captured the experiences of women actively engaged in maternal healthcare.

Data source and sampling procedure

The data for this study was obtained from the Demographic and Health Surveys (DHS) conducted in three Sub-Saharan African countries. The DHS provides extensive datasets covering various health indicators, including mortality, morbidity, fertility, reproductive health, and education related to pregnancy complications and antenatal care (ANC). Each country’s survey consists of distinct datasets, which were combined to examine unmet pregnancy education and its determinants among pregnant women who had ANC follow-up. The DHS employs a stratified two-stage cluster sampling design, selecting enumeration areas in the first stage and generating a sample of households from each enumeration area in the second stage. For this study, the Individual Record (IR) dataset was utilized to extract both dependent and independent variables relevant to the analysis. The variable “unmet pregnancy complication education m43_1” from the IR dataset was recoded to create the outcome variable representing unmet pregnancy complication education.

A binary logistic regression model was applied to analyze the data and identify factors associated with unmet pregnancy complication education. Determinants were reported as adjusted odds ratios with a 95% significance level. The variable selection process for multivariable analysis was based on a p-value threshold of 0.25 in the univariate analysis, ensuring the inclusion of potentially relevant variables for a thorough assessment of their relationship with pregnancy complication education [14].

In the univariate analysis, variables with a p-value of < 0.25 were considered candidates for inclusion in the multivariable analysis. In the subsequent multivariable logistic regression, variables with p-values < 0.05 were deemed statistically significant. The final analysis incorporated a total weighted sample of 20,467 pregnant women, ensuring the results accurately represent the population and reflect the complexities of antenatal education (Table 1).

Study variables

The primary outcome of this study was the prevalence of unmet pregnancy complication education among pregnant women who had attended antenatal care (ANC) visits and were of reproductive age. This unmet pregnancy complication education was assessed by recoding the variable (m43_1) from the Women’s Status dataset.

Independent variables were drawn from both individual and community levels due to the hierarchical nature of the DHS data. At the individual level, variables included women’s age (categorized as 15–24, 25–34, and 35–49), education level (no formal education, primary, secondary and above), Marital status was classified as living in union and living non-union, and occupation (no occupation or employed). Additionally, partner’s occupation was assessed (no occupation or employed), along with the number of ANC visits (less than 4 or more than 4), previous obstetric history (yes or no), and women’s wealth index (poor, intermediate, and rich). Media exposure was also considered, classified as having no exposure or having exposure.

At the community level, independent variables included women’s residence (urban or rural), community ANC utilization level (low or high), community media exposure (low or high), community illiteracy (low or high), and community poverty (low or high). The specific countries involved in the analysis were also included as community-level variables. These comprehensive variables were essential for analyzing the factors associated with unmet pregnancy complication education among pregnant women in the selected Sub-Saharan African countries.

Data processing and statistical analysis

The data were sourced from the latest DHS datasets and subsequently cleaned, re-coded, and analyzed using STATA version 14 statistical software. Prior to performing any statistical analysis, sampling weights, primary sampling units, and strata were applied to ensure the survey’s representativeness and account for its sampling design when calculating standard errors, thereby producing reliable statistical estimates. The weighting variable (m43_1) served as a normalized relative weight tailored to the specific survey.

For pooled data, the individual standard weight for unmet pregnancy complication education was de-normalized by adjusting it according to each country’s sampling fraction, calculated as follows: (unmet pregnancy complication education adjusted weight = m43_1 × (total unmet pregnancy complication education pregnant women aged 15–49 years in the country at the time of the survey)/(number of unmet pregnancy complication education pregnant women)). Given the hierarchical structure of DHS data, standard logistic regression assumptions—such as independence of observations and homogeneity of variance—were not met. Women grouped within clusters may share similar characteristics, influencing these assumptions. To account for between-cluster variability in assessing the association with unmet pregnancy complication education, a multilevel mixed-effects logistic regression model was implemented [13, 15].

The multilevel mixed-effects logistic regression analysis incorporated four models: the null model (including only the outcome variable), Model I (accounting for individual-level variables), Model II (considering community-level variables), and Model III (integrating both individual- and community-level variables). The null model was utilized to examine the variability in unmet pregnancy complication education prevalence across clusters. Model I explored the relationship between individual-level variables and the outcome, while Model II assessed the influence of community-level variables. In the final model, Model III, both individual and community-level variables were simultaneously included with the outcome variable. Model fitness was evaluated, and the model with the lowest deviance and highest log-likelihood ratio was identified as the best-fit model [16, 17].

Random effects (Measures of variation)

To assess the variation in the prevalence rates of unmet pregnancy complication education among pregnant women across clusters, several random effects measures were computed, including the Likelihood Ratio test (LR), Intra-class Correlation Coefficient (ICC), and Median Odds Ratio. Taking clusters as a random variable, the ICC quantifies the degree of heterogeneity in unmet pregnancy complication education prevalence rates between clusters. It reflects the proportion of the total observed variation that is attributable to differences between clusters.

The ICC is calculated using the formula; \(\:\text{I}\text{C}\text{C}=\frac{VC}{VC+3.29}\times\:100\%\). The Median Odds Ratio is the median value of the odds ratio which quantifies the variation or heterogeneity in pregnant women unmeet pregnancy complication education prevalence rates between clusters in terms of odds ratio scale and is defined as the median value of the odds ratio between the cluster at high likelihood of pregnant women unmeet pregnancy complication education on antenatal visit prevalence rates and cluster at lower risk when randomly picking out individuals from two clusters; \(\:\text{M}\text{O}\text{R}={e}^{0.95\surd\:\text{V}\text{C}}\) Moreover, the PCV demonstrates the variation in the pregnant women unmeet pregnancy complication education prevalence rates explained by determinants and computed as; \(\:\:\text{P}\text{C}\text{V}=\frac{Vnull-Vc}{Vnull}\times\:100\%\); where Vnull = variance of the null model and VC = cluster level variance.

The fixed effects were used to estimate the association between the likelihood of pregnant women unmeet pregnancy complication education prevalence rates and individual and community level independent variables. It was assessed and the strength was presented using adjusted odds ratio and 95% confidence intervals with a p-value of < 0.05. Because of the nested nature of the model, Deviance = −2(log likelihood ratio) was used to compare models, and the model with the lowest deviance and the highest log likelihood ratio was selected as the best-fit model [13]. The models’ variables were examined for multicollinearity using variance inflation factors (VIF). The results showed that all VIF values were within the acceptable range of one to ten, indicating no significant multicollinearity among the variables. The study centered on unmet pregnancy complication education and its influencing factors among women of reproductive age who had attended antenatal care (ANC) visits at healthcare facilities during their pregnancy. Unmet pregnancy complication education was assessed by recoding the variable (m43_1) from the Individual Record (IR) dataset. This approach ensured that the analysis accurately reflected the educational needs of pregnant women regarding potential complications during pregnancy [15].