Abstract
OBJECTIVE: This study aimed to characterize the epidemiological trends and mechanisms of pediatric unintentional injuries in Kunshan, China (2018-2023) and to develop time series models to predict future trends. METHODS: A retrospective analysis was conducted on 77,379 pediatric unintentional injury cases, stratified by age, gender, and injury categories. Subgroup analyses targeted children under 5 years. To adjust for pandemic-related disruptions, separate comparisons between 2018 and 2023 were performed. Time series analysis employed an ARIMA model, with model selection based on information criteria and residual diagnostics, and a non-COVID-19 dataset (2018, 2019, and 2023) for forecasting future trends. RESULTS: Males constituted 62.76% of cases, with a mean age of 5.37 ± 3.55 years. The primary age groups were 3-6 years and 6-12 years, which accounted for 60% of the total population. Falls (21.36%) and transport injuries (4.00%) predominated, with limbs being the most injured body part (59.08%). Contusions/abrasions (41.54%) and sprains/strains (31.21%) were common. Subgroup analysis was performed in children under 5 years old, with 22,110 being males (57.5%) and 16,291 being females (42.5%). Among this group, falls and burns were identified as the most frequent incidents. Unintentional injury cases decreased significantly during COVID-19 (2020-2022). The refined ARIMA(1,1,2) (0,1,1)[12] model, excluding pandemic effects, achieved a mean absolute percentage error of 6.46% while revealing seasonal patterns and predicting a slight downward trend for 2024-2026. CONCLUSION: Pediatric unintentional injuries in Kunshan exhibited gender and age-specific patterns, with COVID-19 altering injury profiles. The ARIMA model can capture the seasonal patterns of unintentional injuries to a certain extent, facilitating public health planning and intervention strategies.