JMIR Publications

ABSTRACT

Background:

Life expectancy at birth is a key indicator of population health and human development. In Uganda, life expectancy has steadily improved over the past decades, driven by demographic changes, reductions in child mortality, and advancements in public health. Despite this progress, the extent to which specific demographic factors—such as infant mortality, number of births, population doubling time, and mean age at childbearing—influence life expectancy remains inadequately documented in empirical literature. Understanding these relationships is essential for informing policy interventions aimed at improving population health outcomes

Objective:

This study examines the short-run and long-run determinants of life expectancy in Uganda, focusing on infant mortality rate, number of births, mean age at childbearing, and population doubling time.

Methods:

The study employs annual time series data from 1955 to 2024 obtained from international and national statistical repositories. An Autoregressive Distributed Lag (ARDL) model is used to estimate both long-run and short-run relationships between life expectancy and the selected demographic variables. Unit root tests (ADF) are conducted to assess stationarity, followed by ARDL bounds testing for cointegration. Model estimation includes both long-run coefficients and short-run dynamic adjustments through the error correction mechanism. Statistical significance is determined at conventional confidence intervals, and robustness checks are conducted.

Results:

The ARDL model demonstrates strong explanatory power with R² = 0.7227 and adjusted R² = 0.6963. The error correction term is negative and highly significant (ECT = –0.7083, P = .000), confirming a stable long-run relationship between life expectancy and the selected demographic variables. In the long run, infant mortality is the only significant predictor of life expectancy, with a coefficient of –0.3500 (p = .000), indicating that a one-unit reduction in IMR increases life expectancy by approximately 0.35 years. In contrast, the number of births (β = –4.46×10⁻⁶, p = .566), population doubling time (β = –0.0042, P = .886), and mean age at childbearing (β = –1.3516, P = .286) have negative but statistically insignificant effects, suggesting that improvements in LEB are primarily driven by reductions in early-life mortality rather than raw fertility or population growth metrics. Short-run dynamics reveal that only the lagged change in life expectancy is statistically significant (β = –0.3460, P = .000), indicating rapid adjustment to deviations from long-run equilibrium. The immediate effects of IMR, NB, PDT, and MAC are not significant in the short run, highlighting that year-to-year changes in life expectancy are dominated by prior values and short-term fluctuations rather than contemporaneous demographic shocks.

Conclusions:

Infant mortality remains the strongest determinant of life expectancy in Uganda, underscoring the importance of sustained investments in maternal and child health. Other demographic factors, while directionally aligned with theory, do not display statistically significant effects, suggesting that improvements in survival—rather than demographic quantity—drive life expectancy gains. Policies aimed at reducing infant mortality through improved neonatal care, immunization, nutrition, and access to primary healthcare are likely to yield the greatest improvements in population health outcomes. Clinical Trial: Not applicable (observational time-series analysis)