Refining seasonal mortality estimates through age adjustment: Evidence from Serbia, 2015–2023

Background: Seasonal fluctuations in mortality are a persistent demographic and public health phenomenon. The ideal mortality (IDE) framework estimates seasonal excess mortality by comparing observed outcomes with a counterfactual based on the lowest-mortality seasonal window. Objective: This study evaluates the validity of the IDE framework when applied to age- and sex-specific mortality in Serbia and proposes an age-structure adjustment (IDEadj) to address anomalies in age-specific estimates. Methods: Using mortality data for Belgrade and Vojvodina for the period 2015–2023, we construct the IDE baseline based on the three months with the lowest total mortality in each year. Age-specific mortality rates and life expectancy at birth (e₀) are compared across observed, IDE, and IDEadj scenarios. Results: The IDE framework yields higher life expectancy than observed mortality but produces systematic age-specific inconsistencies. IDE mortality rates occasionally exceed observed values at younger ages and fall to implausibly low levels at older ages, reflecting a mismatch between the age distribution of deaths in the lowest-mortality window and the annual pattern. The IDEadj approach corrects these distortions by aligning age-specific mortality with the observed annual age structure, while preserving the overall magnitude of seasonal gains in life expectancy. Conclusions: Age-structure adjustment improves the internal consistency and interpretability of seasonal mortality estimates without altering their aggregate magnitude. The IDEadj framework refines the original IDE approach and provides a demographically coherent basis for assessing seasonal mortality effects. Contribution: By identifying and correcting age-specific artefacts inherent in the IDE framework, this study provides a demographically coherent extension that enhances the analysis of seasonal mortality patterns.