Bayesian Spatio-Temporal Modelling of Reported Terminated Pregnancy Across Nigerian States (2013-2024)

This study employs a Bayesian spatio-temporal framework to analyze 2013–2024 NDHS data, revealing a dual reproductive regime in Nigeria where southern states exhibit rising termination rates driven by reproductive agency and access, while northern states face increases due to unmet contraceptive needs and structural vulnerability, thereby advocating for state-specific rather than homogenous national reproductive health policies.

The Gist

Imagine Nigeria as a massive, bustling neighborhood with 36 different districts (states). For a long time, the people running the neighborhood (the government) have been trying to count how many families are choosing to end a pregnancy early. But because the rules in the neighborhood are very strict about this topic, most families keep it a secret. It's like trying to count the number of people sneaking out of a house at night when the lights are off and everyone is whispering. The official numbers are blurry, and the leaders don't really know where the problems are or why they are happening.

This paper is like a team of detectives with a high-tech, magical map (called a Bayesian Spatio-Temporal Model) that helps them see the invisible patterns in the dark.

Here is what they discovered, broken down simply:

1. The "Magic Map" vs. The Old Way

What we knew before: Previous studies were like taking a blurry photo of the whole neighborhood and saying, "Okay, on average, things are okay." They looked at the whole country as one big block and didn't notice that one street might be very different from the next. They also just looked at one snapshot in time, missing how things change.

What this study did: Instead of a blurry photo, they used a GPS tracker that updates every few years (2013, 2018, and 2024). Their "magic map" doesn't just look at one state in isolation; it understands that neighbors influence neighbors. If a problem is happening in one town, the map knows to check the town next door because they often share similar struggles. This helps fill in the gaps where data is missing, making the picture much clearer.

2. The Great Divide: Two Different Worlds

The detectives found that Nigeria isn't just one place; it's actually two different worlds with different stories, split roughly between the South and the North.

• The Southern "Coastal" Districts (like Lagos):
  Think of these areas as a busy, modern city. Here, the increase in terminated pregnancies isn't necessarily a sign of desperation. Instead, it's like people having more choices and more control. As women get more education and access to services, they are making active decisions about their family planning. It's a story of agency—people choosing when to start a family.

• The Northern "Inland" Districts (like Kano and Yobe):
  Think of these areas as facing a storm. Here, the story is different. The increase in terminations is driven by vulnerability and a lack of options. Many women here want to avoid pregnancy but don't have access to contraception (birth control). It's like trying to stop a leak in a boat without having any buckets or plugs. The decisions here are often made out of necessity and a lack of support, not just free choice.

3. The Rules of the Game Changed Over Time

The detectives also noticed that the "rules" for what causes these decisions have shifted over the last decade.

• In the past (2013): If you were rich, you were less likely to have an unintended pregnancy. Money was your shield.
• Today (2024): Money matters less. Now, education and the ability to read are the strongest shields. A woman who is educated is much better equipped to make informed choices about her body, regardless of how much money she has.

Why This Matters (The "So What?")

Imagine if the neighborhood leaders tried to fix a leaky roof by handing out the exact same tool to every single house. It wouldn't work because some roofs need a hammer, while others need a patch kit.

• For the Leaders (Policy): This study tells them to stop using a "one-size-fits-all" approach. They need a customized plan for each state. The South needs support for reproductive choice and services, while the North needs urgent help with access to contraception and education to fix the root causes of vulnerability.
• For the Future: This "magic map" method can be used for other health issues too, like tracking malaria or malnutrition, helping leaders spot trouble spots before they become disasters.

In a nutshell: This paper shines a light on a hidden problem, showing us that while the whole country is facing the same issue, the reasons and the solutions are completely different depending on where you live. It's time to stop guessing and start giving each neighborhood the specific help it needs.

Technical Summary

1. Problem Statement

The study addresses a critical data gap in Nigerian reproductive health policy: the disconnect between restrictive abortion legislation and the estimated annual incidence of 1.8 million terminations. Due to legal constraints, abortion data is characterized by systematic under-reporting and spatial instability, making it difficult for policymakers to identify specific geographic locations and drivers of terminated pregnancies. Previous research has largely relied on cross-sectional analyses and national-level aggregates, failing to account for spatial dependencies between states or the temporal evolution of reproductive decisions. This lack of granular, subnational data hinders the development of targeted health interventions.

2. Methodology

To overcome the limitations of existing data, the authors employed a Bayesian spatio-temporal framework to analyze data from the 2013, 2018, and 2024 Nigerian Demographic and Health Surveys (NDHS).

• Statistical Approach: The study utilizes a Bayesian hierarchical model to detect latent socio-demographic causes of terminated pregnancy. This approach allows for "borrowing strength" from neighboring states, thereby improving estimation accuracy in areas with sparse data or high spatial correlation.
• Variables Analyzed: The model incorporates key socio-demographic predictors, including:
  • Household wealth
  • Education levels
  • Literacy rates
  • Contraceptive prevalence
• Temporal Scope: The analysis tracks the evolution of these factors across three distinct survey waves (2013, 2018, 2024) to capture temporal trends and space-time interactions.

3. Key Contributions

This study makes three primary contributions to the field of reproductive health research and policy:

• Methodological Innovation: It introduces a robust Bayesian spatio-temporal modeling approach for analyzing sensitive maternal health indicators in low-resource settings. This method effectively handles spatial dependence and provides more reliable small-area estimates than traditional cross-sectional methods.
• Granular Risk Mapping: The study moves beyond national averages to identify specific "hotspots" and consistent high-risk conditions at the state level, decoding risk patterns that were previously obscured by aggregate data.
• Temporal Dynamics: It is one of the few studies to explicitly model how the determinants of abortion change over time, revealing shifts in the primary drivers of reproductive choices across different regions.

4. Key Results

The analysis reveals a highly polarized and spatio-temporally intense landscape of terminated pregnancy in Nigeria, characterized by a "dual reproductive regime":

• Regional Divergence:
  • Southern and Coastal States (e.g., Lagos, Bayelsa): These regions show sustained increases in prevalence. The drivers here are linked to a high fertility transition, increased reproductive agency, better access to services, and higher reporting rates.
  • Northern States (e.g., Yobe, Kano): Increases in termination rates in these areas are primarily driven by unmet contraceptive needs and structural vulnerability rather than reproductive choice.
• Shifting Determinants: The predictive power of socio-economic factors has evolved over the decade. While household wealth was a protective factor in earlier surveys (2013/2018), by 2024, education and literacy had emerged as the strongest predictors of terminated pregnancy patterns.
• Statistical Evidence: The model provides probabilistic evidence supporting the need for state-specific interventions, highlighting that a "one-size-fits-all" national policy is insufficient.

5. Significance and Implications

The findings have profound implications for research, practice, and policy in Nigeria:

• Policy Shift: The study argues for a transition from homogenous national approaches to state-specific reproductive health policies. It highlights the necessity of tailoring interventions to the specific context of each region (e.g., addressing structural vulnerability in the North vs. supporting reproductive agency and service access in the South).
• Resource Allocation: By identifying emerging hotspots and high-risk states, the study enables health planners to allocate limited resources more efficiently and implement focused interventions.
• Future Research: The methodology serves as a blueprint for analyzing other sensitive maternal and public health indicators using small-area estimation techniques, potentially improving data quality for other hidden health phenomena in Nigeria and similar contexts.

In conclusion, this paper provides the first comprehensive, spatially explicit, and temporally dynamic evidence base for understanding terminated pregnancy in Nigeria, urging a paradigm shift toward localized, evidence-based reproductive health governance.