Historical reconstruction and prospective scenarios of road traffic mortality in Colombia (1986-2025)

Abstract

This study shows that the historical evolution of road traffic mortality in Colombia does not follow continuous growth but a non-monotonic dynamic comprising three distinct temporal regimes. Road traffic mortality is a central public-health and transport-sustainability challenge, yet longitudinal evidence characterizing its long-term behavior and projecting its trajectory remains scarce. We analyzed the annual series of road traffic fatalities and the rate per 100,000 inhabitants between 1986 and 2025 (n = 40), compiled from official records under explicit traceability criteria. Arima models with a pulse intervention term for the 2020 pandemic shock were fitted, selected by AICc and validated out of sample, and contrasted with an unobserved-components structural model and a segmented regression. The series displays three phases —growth (1986-1997), decline (1997-2006) and recovery (2006-2025): the rate peaked in 1995 (22.0 per 100,000) and, although 2025 records the all-time high in counts, per-capita risk (16.4) remains below that peak. The Arima (0,1,2) model with intervention showed residuals free of autocorrelation and heteroscedasticity, with an out-of-sample mean absolute percentage error of 11.4 %. Inference to 2030 and 2035 sets a range between stabilization around 8,850 annual fatalities and a continued rise toward 10,000. The projected trajectory more than doubles the target of the National Road Safety Plan 2022-2031, which is unattainable without unprecedented structural change. The findings encourage strengthening the safe system approach, continuous monitoring of the series and multisectoral governance involving the health and education sectors.