Statistical Assessment of Maternal Health Outcomes in Nigeria

Background to the Study

WHO defines maternal health as health during pregnancy, childbirth and postpartum. Despite Millennium Development Goals and SDG 3.1, burden remains unequal. Sub-Saharan Africa 66% of global maternal deaths 2020. Nigeria MMR 512 per 100,000 live births NDHS 2018 masks subnational disparities: north-west states Sokoto and Zamfara exceed 1,500 per 100,000 while southern states lower. Most common direct causes: obstetric haemorrhage particularly postpartum haemorrhage, hypertensive disorders including eclampsia and pre-eclampsia, sepsis, obstructed labour, unsafe abortion. Indirect causes malaria, anaemia, HIV contribute substantially.

Maternal near-miss defined by WHO as woman who nearly died but survived severe acute complication during pregnancy, childbirth or within 42 days of termination. Near-miss concept increasingly adopted in biostatistics because larger sample enables more powerful inference about risk factors and causal pathways than death alone.

Biostatistical toolkit includes descriptive characterization of maternal age, parity, gestational age, complications; contingency tables and chi-square for associations; logistic regression quantifying independent contributions controlling for confounders; survival analysis Kaplan-Meier and Cox proportional hazards addressing time dimension; ANOVA comparing mean outcomes across groups; PCA reducing dimensionality of inter-correlated risk indicators facilitating cleaner specification. Application to Nigerian data provides empirical foundation for evidence-based intervention design, identifying amenable risk factors, priority geographies and quantitative impact of scaling ANC in resource-constrained public health system.

University of Ibadan and UCH Ibadan represent premier maternal health research site with long obstetric data collection tradition and referral catchment spanning Oyo, Ogun, Osun. Combined with NDHS national data, analysis offers richness at institutional and population levels.

WHO – Maternal Mortality Fact Sheet 2023

NDHS 2018 – Nigeria Demographic and Health Survey

Statement of the Problem

Despite policy attention and investment, maternal mortality decline insufficient to achieve SDG 3.1. NDHS 2018 MMR 512 per 100,000 only marginally improved from 576 in 2013, far below 7.5% annual reduction required. Contributing factor is inadequate use of statistical evidence in programme design and resource allocation. While broad determinants qualitatively understood, detailed quantitative analyses simultaneously controlling confounders, accounting for time dimension, incorporating near-miss concept are scarce in Nigerian literature. Limited evidence on relative quantitative contributions of risk factors among women attending tertiary referral facilities in south-western Nigeria where clinical picture differs due to referral bias. This study addresses gap by comprehensive methodologically rigorous assessment using population-level NDHS and facility-level UCH records.

Aim and Objectives

Aim: To conduct comprehensive statistical assessment of maternal health outcomes in Nigeria, focusing on identifying, quantifying and modelling determinants of adverse maternal outcomes.

1. Describe sociodemographic and obstetric characteristics of study population using appropriate descriptive measures.

2. Assess association between key risk factors and adverse maternal outcome using chi-square tests and correlation analysis.

3. Identify independent predictors of adverse maternal outcome through binary logistic regression analysis.

4. Assess time-to-complication using Kaplan-Meier survival curves and Cox proportional hazards regression.

5. Compare mean birth weights across parity groups using one-way ANOVA and post-hoc testing.

6. Reduce dimensionality among inter-correlated obstetric risk indicators using Principal Component Analysis.

7. Formulate evidence-based recommendations for improving maternal health outcomes in Nigeria.

Research Questions

What are sociodemographic and obstetric characteristics of maternal study population?

What is association between antenatal care attendance, delivery mode, parity and adverse maternal outcomes?

Which risk factors are independent predictors of adverse maternal outcome after multivariate adjustment?

How does time-to-complication differ between women with and without antenatal care?

Is there statistically significant difference in mean birth weight across parity groups?

What latent risk constructs underlie inter-correlated obstetric risk indicators?

Significance of the Study

Clinically and programmatically, quantification of adjusted odds ratios and hazard ratios provides actionable evidence for obstetricians, midwives and public health officers to identify high-risk patients and allocate scarce emergency obstetric care resources. Finding that antenatal care is most modifiable predictor directly supports universal ANC coverage programmes. Methodologically, demonstrates integrated suite of biostatistical tools (logistic regression, survival analysis, ANOVA, PCA) to single maternal dataset, providing template adaptable across sub-Saharan Africa. For policymakers, translates statistical findings into concrete investment recommendations, bridging data production and evidence-based decision-making weakness in Nigerian health information system. For students and academics, provides self-contained worked example of biostatistical analysis serving as teaching resource for advanced undergraduate and postgraduate statistics, epidemiology and public health. For further learning, see our tutorials on maternal health data analysis and Cox proportional hazards modelling.

Scope of the Study

Focuses on adult women aged 15-49 who delivered at or referred to University College Hospital, Ibadan January 2015-December 2023, plus nationally representative sample from 2018 NDHS dataset. Outcome variable is adverse maternal outcome defined as maternal near-miss or death during index delivery admission. Neonatal outcomes reported as secondary findings not primary focus. Does not include abortions or ectopic pregnancies.

Operational Definition of Terms

Maternal Mortality: Death of woman while pregnant or within 42 days of termination, irrespective of duration or site, from cause related to or aggravated by pregnancy or management (WHO 2020).

Maternal Near-Miss: Woman who nearly died but survived severe acute complication during pregnancy, childbirth or within 42 days of termination (WHO 2011).

Adverse Maternal Outcome: Composite endpoint comprising maternal near-miss events and maternal deaths during index delivery admission.

Antenatal Care (ANC): Organised healthcare services by skilled professional before delivery encompassing examination, lab tests, health education, complication screening. Adequate ANC defined as four or more visits per WHO 2016 model.

Parity: Number of previous deliveries whether live births or stillbirths. Grand multiparity five or more previous deliveries.

Postpartum Haemorrhage (PPH): Blood loss 500 ml or more within 24 hours vaginal delivery or 1,000 ml or more for caesarean section.

Hazard Ratio (HR): Relative risk of event occurring in one group vs another at any given time, derived from Cox proportional hazards regression.

Odds Ratio (OR): Ratio of odds of outcome in exposed vs unexposed group, derived from logistic regression.

WHO – Maternal Health and Near-Miss Approach

CDC – Maternal Mortality Prevention

Short Conclusion

Analysis of 1,200 records from NDHS and UCH Ibadan identifies maternal age >35 OR 2.84, absence of ANC OR 4.21, referral delivery OR 3.17, grand multiparity OR 2.51 and PPH OR 6.83 as strongest independent predictors of adverse maternal outcome. Cox model confirms same as hazard contributors, with Kaplan-Meier showing significant divergence in complication-free survival by ANC status (log-rank p<0.001). PCA clusters obstetric risks into two components explaining 61.3% variance, and ANOVA confirms parity influences birth weight (F=12.47, p<0.001). Findings underscore that while referral bias elevates risk in tertiary centre, modifiable factor ANC remains central. Evidence supports targeted ANC scale-up to four or more visits, skilled birth attendance improvement, emergency obstetric care strengthening with PPH bundles, referral system optimization, and grand multipara counselling. Future research should incorporate prospective design, wealth, distance and partner education confounders. Explore our resources on biostatistics for maternal health and survival analysis interpretation.

Frequently Asked Questions

Q: What defines adverse maternal outcome in this study?

A: Composite of maternal near-miss (woman who nearly died but survived severe complication within 42 days per WHO criteria) and maternal death during index delivery admission at UCH Ibadan or reported in NDHS.

Q: What data sources were used?

A: Secondary data from 2018 Nigeria Demographic and Health Survey nationally representative sample plus 1,200 maternal case records from University College Hospital, Ibadan spanning 2015-2023, enabling population and facility-level analysis.

Q: Which statistical methods were applied?

A: Descriptive statistics, chi-square and correlation for associations, binary logistic regression for probability of adverse outcome, Kaplan-Meier estimator and Cox proportional hazards for time-to-complication, one-way ANOVA with post-hoc for birth weight across parity, and Principal Component Analysis for dimensionality reduction.

Q: What were strongest predictors of adverse outcome?

A: Absence of antenatal care OR 4.21 (95% CI 2.93-6.05), postpartum haemorrhage OR 6.83 (4.51-10.34), maternal age >35 OR 2.84 (1.97-4.10), referral delivery OR 3.17 (2.12-4.74), grand multiparity OR 2.51 (1.74-3.62), all significant after multivariate adjustment.

Q: How did survival analysis add value?

A: Kaplan-Meier curves showed significant divergence in complication-free survival between ANC and no-ANC groups (log-rank p<0.001). Cox proportional hazards quantified hazard ratios over time, confirming same predictors increase instantaneous risk of complication, addressing time dimension ignored by logistic only.

Q: What did PCA reveal?

A: Principal Component Analysis among inter-correlated obstetric risk indicators reduced to two principal components explaining 61.3% total variance, revealing latent constructs such as haemorrhagic risk and hypertensive/age-related risk, facilitating cleaner model specification.

Q: Why ANOVA for birth weight across parity?

A: ANOVA tested mean birth weight differences across parity groups (nullipara, multipara, grand multipara). F=12.47 p<0.001 indicated significant difference, with post-hoc showing grand multipara often lower birth weight due to maternal depletion and higher risk.

Q: What are limitations?

A: Tertiary referral centre skewed to high-risk cases (referral bias), secondary hospital records may have missing values and miscoding with listwise deletion risking selection bias, NDHS cross-sectional limits causal inference, Cox model proportionality assumption may not hold perfectly, unobserved confounders like wealth, partner education, distance may modify associations.

Q: What policy recommendations follow?

A: Targeted ANC scale-up to WHO-recommended 4+ visits with community outreach, skilled birth attendance improvement, emergency obstetric care strengthening with PPH prevention bundles, referral system optimization to reduce delays, and counselling for women >35 and grand multipara on risks.

Q: Can this framework be replicated?

A: Yes. Combination of NDHS population data plus facility records, with logistic, survival, ANOVA and PCA, provides integrated methodological template adaptable for other tertiary hospitals in Nigeria and sub-Saharan Africa, serving as teaching resource for biostatistics students.