📄 medicine

Life expectancy nexus mortality in 14 selected SADC countries with trend analysis

This study analyzes 13 years of data from 14 SADC countries using GMM and Random Forest models to identify infant mortality, under-5 mortality, HIV prevalence, and GDP per capita as key determinants of life expectancy, projecting continued regional growth with significant disparities through 2027 while recommending targeted health interventions to address these gaps.

Original authors: Talent Kondo, Sithokozile Bafana, Mosty Kenama

Published 2026-07-03

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Talent Kondo, Sithokozile Bafana, Mosty Kenama

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: A Health Report Card for Southern Africa

Imagine the Southern African Development Community (SADC) as a large neighborhood with 14 different houses. Some houses are very well-maintained with shiny roofs and green lawns, while others are struggling with leaks and broken windows.

This study is like a 13-year inspection report (from 2010 to 2022) for this neighborhood. The researchers wanted to answer one main question: "What makes the people living in these houses live longer, and what makes them pass away sooner?"

They looked at how long people live (Life Expectancy) and compared it to things like how many babies die, how much money the country has, and how widespread diseases like HIV are. They used two types of "detective tools" to solve the mystery:

The Math Detective (GMM): A strict, rule-following method that looks at historical patterns.
The Pattern Spotter (Random Forest): A smart computer program that looks for hidden connections and predicts the future.

The Main Suspects: What Drives Life or Death?

The researchers found that life expectancy isn't just about having a lot of money. It's more like a recipe where some ingredients matter much more than others.

1. The "Baby Health" Factor (The Heavy Hitters)
The study found that Infant Mortality (babies dying) and Under-5 Mortality (young children dying) are the biggest villains.

The Analogy: Think of life expectancy as a chain. If the weakest link (the babies) breaks, the whole chain snaps. The study shows that if you save the babies, you instantly make the average lifespan of the whole country go up.
The Finding: These two factors were the strongest predictors. If infant deaths go down, life expectancy goes up significantly.

2. The "HIV Shadow"
HIV prevalence (how many people have the virus) is another major negative factor.

The Analogy: Imagine HIV as a heavy backpack that slows everyone down. The study found that as long as this backpack is heavy, it drags down the average life expectancy.
The Finding: Countries with high HIV rates saw their life expectancy drop. Managing this "backpack" is crucial for recovery.

3. The "Money" Factor (GDP)
Having more money (GDP per capita) helps, but it's not a magic wand.

The Analogy: Money is like the foundation of a house. You need a solid foundation to build a tall building, but just having a foundation doesn't guarantee the roof is fixed. The study suggests that money helps, but only if it's actually used to fix the "leaks" (healthcare) and isn't just sitting in a bank account.
The Finding: Money matters, but it's not the only thing that matters. You can have money, but if the health system is broken, people still don't live longer.

4. The "2021 Dip" (The Storm)
The report noticed a sudden drop in life expectancy around 2021.

The Analogy: Imagine the neighborhood was slowly getting better every year, like a garden growing. Then, in 2021, a massive storm (the COVID-19 pandemic) hit, knocking down some of the plants.
The Finding: The pandemic caused a temporary setback, but by 2022, the garden was starting to grow again.

The Neighborhood Divide: Who is Winning and Who is Losing?

Even though the whole neighborhood is getting better on average, there is a huge gap between the "rich" houses and the "struggling" houses.

The Star Performer (Mauritius): This country is like the mansion on the hill. It has the highest life expectancy (over 73 years) and is predicted to stay there. It's the model for how things should look.
The Struggling Houses (Lesotho and Eswatini): These countries are like houses with a leaky roof. They are at the bottom of the list (around 50–55 years). While they are trying to fix the leaks, they are still far behind the mansion.
The Middle Ground: Countries like Botswana and South Africa are like well-kept family homes. They are doing well but are sensitive to storms (like the pandemic).

The Prediction (2023–2027):
The computer model (the Pattern Spotter) looked at the trends and said: "If we keep doing what we are doing, the garden will keep growing."

Mauritius will stay on top.
Lesotho and the DRC will still be at the bottom, though they will improve slightly.
Botswana and Malawi are expected to make big jumps if they keep their current momentum.

The Solution: How to Fix the Neighborhood

The researchers didn't just point out the problems; they offered a blueprint for repair.

Focus on the Babies: Don't just throw money at general problems. The study says the most effective way to fix the "chain" is to focus entirely on Maternal and Child Health. If you save the babies, the whole country's life expectancy rises faster than anything else.
Tackle the HIV Backpack: We need to move beyond just giving medicine. We need to make sure people are tested, treated, and that the virus is suppressed. It's about making the "backpack" lighter for everyone.
Neighbors Helping Neighbors: Diseases don't respect borders. If one house has a leak, the water can seep into the next house. The study suggests that these 14 countries need to work together as a team, sharing health protocols so that progress in one country isn't ruined by a crisis in the next.
Spend Money Wisely: It's not enough to just increase the health budget. The money needs to go to the front lines (doctors, vaccines, clinics), not get stuck in paperwork.

The Bottom Line

This paper tells us that the Southern African neighborhood is healing. Despite the storms of HIV and COVID-19, the trend is moving upward. However, to make sure everyone in the neighborhood lives a long, healthy life, we need to stop looking at the big picture and start fixing the specific cracks: save the babies, manage the HIV, and make sure the neighbors help each other.

Note: The study uses advanced math and computer models to prove these points, but the core message is simple: Save the children, fight the virus, and work together.

Technical Summary: Life Expectancy Nexus Mortality in 14 Selected SADC Countries with Trend Analysis

Problem Statement
The Southern African Development Community (SADC) region faces a complex demographic and epidemiological landscape characterized by the interplay of infectious diseases (notably HIV/AIDS), non-communicable diseases, and varying levels of socioeconomic development. While global studies provide broad estimates, there is a significant gap in granular, region-specific longitudinal analyses that decompose the "nexus" between life expectancy (LE) and specific mortality indicators. Previous literature often lacks the capacity to account for the unique mortality shocks in the region, such as the HIV epidemic and the recent COVID-19 pandemic, or to address data quality challenges and cross-country heterogeneity. This study aims to bridge these gaps by analyzing the historical impacts of health and economic indicators on life expectancy across 14 SADC countries from 2010 to 2022 and forecasting regional trends up to 2027.

Methodology
The study employs a comprehensive panel dataset covering 14 nations (Angola, Botswana, Comoros, DRC, Eswatini, Lesotho, Madagascar, Malawi, Mauritius, Mozambique, Namibia, South Africa, Zambia, and Zimbabwe) over a 13-year period (2010–2022). Data sources include the World Bank's World Development Indicators and UN World Population Prospects.

The analytical framework utilizes a multi-stage approach:

Diagnostic Testing: Cross-sectional dependence (Pesaran, 2004) was tested to identify common unobserved shocks. The Hausman test was used to determine the appropriate model specification between Fixed Effects (FE) and Random Effects (RE).
Econometric Modeling:
- Pooled OLS: Used as a baseline to assess overall explanatory power.
- Pooled Mean Group (PMG): Applied to a subset of 12 countries to estimate long-run relationships and error correction mechanisms.
- Dynamic Panel GMM: A Generalized Method of Moments (Arellano-Bond) estimator was deployed to address potential endogeneity in mortality variables (e.g., HIV prevalence) and the persistence of life expectancy (lagged dependent variable). This model accounts for dynamic panel bias and unobserved heterogeneity.
- Non-linear Analysis: The GMM model included squared terms for Government Health Expenditure (GHE) and GDP to test for diminishing or accelerating marginal returns.
Machine Learning: A Random Forest (RF) model was implemented alongside the econometric models to enhance predictive accuracy and identify feature importance hierarchies.
Forecasting: Linear trend analysis and the RF model were used to project life expectancy from 2023 to 2027.

Key Results

Determinants of Life Expectancy: Correlation and regression analyses identified Infant Mortality (IM), Under-5 Mortality (U5M), and HIV Prevalence (HIVP) as the most significant negative predictors of life expectancy, with correlation coefficients oscillating between -0.72 and -0.90. GDP per capita emerged as a critical positive predictor.
Model Performance:
- The Pooled OLS model demonstrated high explanatory power ( $R^2 = 0.959$ ), though it suffered from potential bias regarding unobserved heterogeneity.
- The Fixed Effects model yielded an $R^2$ of 0.855, confirming the importance of controlling for country-specific unobserved factors.
- The Random Forest model achieved the highest predictive accuracy with an $R^2$ of 0.94, outperforming the linear OLS model ( $R^2 = 0.924$ ).
Dynamic and Non-linear Findings:
- Persistence: The lagged life expectancy variable showed strong persistence, indicating that past health outcomes heavily influence current trends.
- HIV Impact: HIV prevalence was a statistically significant negative driver in both long-run (PMG) and dynamic (GMM) models. A one-unit increase in HIV prevalence was associated with a long-run decrease of approximately 1.9 years in life expectancy.
- Non-linear Effects: The GMM results revealed a non-linear relationship for health spending and economic growth. Specifically, $GDP^2$ showed a positive, accelerating impact on life expectancy at higher economic levels, while $GHE^2$ indicated diminishing marginal returns on health expenditure.
Trends and Projections:
- The region generally exhibited an upward trend in life expectancy from 2010 to 2022, interrupted by a notable dip around 2021 attributed to the COVID-19 pandemic.
- 2027 Projections: Mauritius is forecasted to maintain the highest life expectancy (approx. 74.44 years). Conversely, Lesotho and the DRC are projected to remain at the lower end of the spectrum (approx. 61.41 and 62.92 years, respectively), highlighting persistent regional disparities.

Key Contributions
The paper contributes to empirical literature by:

Methodological Integration: Blending classical dynamic econometric panels (GMM) with non-linear machine learning architectures (Random Forest) to provide a robust analysis of the LE-mortality nexus.
Granular Regional Analysis: Providing country-specific forward projections for 14 SADC nations, offering more detailed insights than global aggregate studies.
Policy-Relevant Evidence: Demonstrating that while economic growth is foundational, the most immediate levers for increasing life expectancy in the SADC region are the reduction of infant/under-5 mortality and the management of HIV prevalence.
Diffusion Analysis: Utilizing heatmap analysis to visualize the slow convergence of health progress across the region, driven by the diffusion of medical technology and antiretroviral therapy.

Significance and Claims
The authors claim that the study provides evidence-based insights to inform health policy, resource allocation, and intervention strategies. The high predictive accuracy of the models ( $R^2$ up to 0.94) suggests that the identified indicators (IM, U5M, HIVP, GDP) are the correct focus for policy interventions.

The study emphasizes that regional governments must shift from a singular focus on economic output to targeted interventions in maternal/child health and HIV control. It argues that for lower-income nations, overcoming the "HIV-mortality complex" is the primary objective, while middle-income nations must ensure economic growth is pro-poor and channeled into accessible primary healthcare. The paper concludes that coordinated cross-border health protocols are essential, given the spillover effects of health outcomes between neighboring countries, to sustain long-term mortality improvements and achieve regional health equity by 2027.

Limitations
The authors acknowledge limitations including potential bias from forward/backward filling of missing data, the assumption of linear trend continuation in forecasting (which may not account for future shocks like new pandemics or political instability), and the exclusion of specific variables such as vaccine access quality and political stability indices. Additionally, high correlation between GDP and access to electricity (ATE) may affect individual coefficient reliability in OLS, though not the overall predictive power.