Publication: Time series modelling and forecasting commuting accident and association with meteorological parameters in Johor Bahru and Kuala Lumpur, Malaysia (2015-2019)
No Thumbnail Available
Date
2024-12
Authors
Hanafi, Nur Sujaihah
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Introduction: In tropical countries such as Malaysia, rainfall and temperature serve as primary meteorological parameters that potentially increase the risk of road traffic accidents (RTAs). The working population in major cities is particularly vulnerable, especially during commuting hours, resulting in a surge of commuting accidents. However, research evidence utilizing objective measures, such as time series data, to model the association between meteorological factors and commuting accidents or even the RTAs in general, particularly within the Malaysian context are substantially limited. Therefore, this study aims to model the association between rainfall and temperature with commuting accidents and forecast their future occurrence.
Methods: This study employed retrospective secondary data on RTAs, rainfall, and temperature from Kuala Lumpur and Johor Bahru from 1 January 2015 to 31 December 2019. In the first phase, the outcome variable was the number of RTAs among the working population. A quasi-Poisson regression model was utilized to investigate the influence of different temporal factors on the outcome variable. Its association with temperature and rainfall was then examined using a combination of Hurdle Negative Binomial (HNB) regression and Distributed lag non-linear models (DLNM). The second phase focused on the number of commuting accidents as an outcome variable, defined as RTAs involving the working population during commuting hours (7:00 am to 8:00 pm). A combined approach using time-stratified
case-crossover design and DLNM was implemented to investigate the association between temperature and rainfall with commuting accidents and subsequently, an ARIMA model was employed to forecast their future incidence.
Results: In phase one, 11,877 and 8833 workers were involved in 11,531 and 8632 RTAs over five years in Kuala Lumpur and Johor Bahru, respectively. The majority of workers were male (> 75%) and aged 15-44 years (> 70%). The time of day, divided into two time zones, was significantly associated with RTAs among workers. The incidence risk ratio (IRR) was highest during T3 (12:00 pm - 5:59 pm) compared to T1 (12:00 am – 5:59 am) and during PEAK 2 (4:30 pm-7:30 pm) compared to OTHER (12:00 am – 7:29 am, 7:31 pm – 11:59 pm) in both cities. In Kuala Lumpur, workers had higher RTA risk during weekdays compared to weekend (IRR of 1.18, 95% CI: 1.10, 1.27). Weeks comprising of three or more school holiday showed a reduced RTA risk in both cities (Kuala Lumpur: IRR 0.78, 95% CI: 0.69-0.88; Johor Bahru: IRR 0.83, 95% CI: 0.72-0.95). Lower temperatures significantly affected RTA occurrence, peaking at 22°C in Kuala Lumpur (OR 1.46, 95% CI 1.10-1.94) and 26°C in Johor Bahru (OR 1.36, 95% CI 1.04-1.78). The strongest lag effect of extremely low temperature (23°C) was at lag 0, and for extremely high temperature, it was delayed up to 16 hours in Kuala Lumpur (36°C) and 24 hours for Johor Bahru (34°C). Rainfall intensity increased RTA odds, with the most pronounced effects at lag 9-hour and 12-hour in Kuala Lumpur and Johor Bahru, respectively. In phase two, 8941 and 6548 commuting accidents were recorded in Kuala Lumpur and Johor Bahru from 1 January 2015 to 31 December 2019. Kuala Lumpur showed a significant inverse linear association between daily maximum temperature and commuting accidents, with the highest odds at 28°C (OR 1.76, 95% CI 1.14-2.72), while Johor Bahru had a non-significant positive association, peaking at 35°C (OR 1.21, 95% CI 0.63-2.32). No significant association between rainfall and commuting accidents was observed in both cities. The best forecasting model for Kuala Lumpur was a regression with ARIMA (1,1,1) error, with a weekly daily minimum temperature utilized as regressor. In Johor Bahru, ARIMA (1,1,1) was identified as the best fitted model. Both models indicated no substantial changes in the projected trajectory of weekly commuting accidents during 2020.
Conclusion: The study findings elucidate the association between temperature, rainfall, and their lag effect with RTAs among the working population, and subsequently on commuting accidents, underscoring the significance of incorporating meteorological parameters in public health policy related to occupational safety and health, urban planning, and road safety. These insights can also inform targeted intervention and enhance the preparedness of urban workers for extreme weather events resulting from climate change.