System-level factors associated with fatal-crash involvement and 30-day mortality among child pedestrians in Türkiye: a national retrospective study

Child pedestrian records may be common in settings that differ from those associated with severe crashes or the child’s own death. This study examined demographic, road, vehicle, environmental, and police-coded driver factors associated with two complementary outcomes among child pedestrians in Türkiye: fatal-crash involvement and individual 30-day mortality. National police records for 31,499 pedestrians aged 0–14 years involved in injury or fatal crashes during 2021–2023 were analyzed. The primary estimand was the adjusted record-level association with involvement in a crash in which at least one person died within 30 days; the secondary estimand was the adjusted record-level association with death of the recorded child within 30 days. Category-specific rates and prespecified multivariable logistic regressions were estimated, with sensitivity analyses using alternative specifications. Overall, 507 records were associated with fatal crashes, and 420 represented children who died within 30 days. Fatal-crash involvement was associated with heavy vehicles (OR = 8.25; 95% CI: 5.45–12.48), police-coded speed-related driver fault (OR = 3.56; 95% CI: 2.91–4.35), major/higher-order roads (OR = 3.60; 95% CI: 2.62–4.94), buses/minibuses (OR = 3.09; 95% CI: 2.26–4.21), locations outside built-up areas (OR = 2.25; 95% CI: 1.44–3.52), and nighttime conditions (OR = 1.90; 95% CI: 1.52–2.37). The child-mortality model showed a broadly similar pattern, including heavy vehicles (OR = 7.14), major/higher-order roads (OR = 3.52), and police-coded speed-related driver fault (OR = 3.32). The primary model characterizes the severity context of crashes involving child pedestrians, whereas the secondary model directly addresses the child’s own 30-day mortality. The broadly concordant pattern across these related outcomes identifies road and vehicle contexts associated with greater severity and may inform Safe System prevention, while causal interpretation—particularly of police-coded fault variables—should be avoided.