NONPARAMETRIC ANALYSIS OF TRAFFIC ACCIDENT IN IRAN USING GENERALIZED ADDITIVE MODELS

Background: Increasing traffic casualties is a major challenge for human societies that affects health promotion and development. In this study, generalized additive mixed model was used to investigate the nonlinear pattern of traffic injury and examine the relationship between the rate of crash injury and its associated factors. Methods: In this longitudinal study, generalized additive mixed models for count data were used to analyze six-year traffic crash data of injured persons referred to 21 hospitals in Hamadan province of Iran. The rate of hospital admissions due to crash from car occupants, motorcyclists, and pedestrians was analyzed separately. Data was analyzed using "gamm" package of R software. Results: About 75.5% of the subjects were males. The mean age of the subjects was about 30 and did not show a significant effect on the traffic injury. The gender (p-value < 0.001) and rural residential when compared with non-residential area (p-value < 0.001) showed a significant linear effect on the traffic injury. There was also a nonlinear significant relationship between the time variable and the rate of the traffic injuries. The rate of traffic injuries is significantly different over time (p-value < 0.001). Conclusion: The generalized additive modeling results for the data of this study suggest that such relationship curves may not be monotonic. The traffic injury pattern when the generalized additive models are used suggests that the major crash prevention policy should be established to reduce and control the traffic injuries especially among motorcyclist.