%0 Journal Article
%T Determinants of mortality type in a high altitude Andean context using a multivariable logit regression model in Puno, Peru‎
%A Percy Huata-Panca
%A Jahir Manuel Huata Apaza
%A Angel Javier Quispe Carita
%A Godofredo Quispe Mamani
%A Fred Torres-Cruz
%J Journal of Advanced Pharmacy Education and Research
%@ 2249-3379
%D 2025
%V 15
%N 3
%R 10.51847/1vvhNPv5Vy
%P 198-204
%X Understanding the determinants of mortality type is central to evidence-based public health, particularly in high-altitude Andean regions where demographic, clinical, and health-system factors intersect. This study aimed to identify factors associated with the type of death (natural versus violent) in the Puno region, Peru, in 2024, using a parsimonious and statistically robust logit modeling framework. A cross-sectional analytical design was applied to the entire population of registered deaths (n = 6,455), of which 3,701 cases with complete cause-of-death information were retained for multivariable analysis. Logistic regression models were estimated by maximum likelihood, with backward elimination based on likelihood-ratio tests to ensure parsimony and inferential stability. The final multivariable model included nine predictors, including clinical causes of death, demographic characteristics, health insurance coverage, place of registration, and necropsy status. The model performance was high, with a Nagelkerke R² of 0.658 and a classification accuracy of 98.7%, indicating excellent probabilistic discrimination. Odds-ratio estimates showed that primary and tertiary causes of death, male sex, increasing age, health insurance type, province, and registration in health facilities significantly increased the probability of natural death. In contrast, secondary causes and necropsy were inversely associated. A reduced model that included only the first three causes of death preserved strong predictive capacity (Nagelkerke R² = 0.272; accuracy = 97.9%), highlighting the dominant explanatory role of multi-cause mortality coding. These findings suggest that high-quality civil registration data, combined with rigorously specified logit models, can accurately characterize mortality patterns in high-altitude regions. This study provides actionable epidemiological evidence to support mortality surveillance, health-system planning, and risk stratification in the Peruvian Andes and comparable low-resource, high-altitude contexts.
%U https://japer.in/article/determinants-of-mortality-type-in-a-high-altitude-andean-context-using-a-multivariable-logit-regress-0fivhmnetcyrlko