Quantitative forecasting trend of dengue fever and dengue haemorrhagic fever cases in Peninsular Malaysia

• Main Author: Cheng, Kueh Yee
• Format: Thesis
• Language: English
• Published: 2006
• Subjects: RA643-645 Disease (Communicable and noninfectious) and public health
• Online Access: http://eprints.usm.my/47743/1/DR.%20KUEH%20YEE%20CHENG-24%20pages.pdf; http://eprints.usm.my/47743/

Introduction: Forecasting dengue fever (DF} cases (including dengue haemorrhagic fever, DHF) is important in improving the effectiveness of control measures by early identification of period with higher DF cases, areas and populations at risk. Objective: This study was carried out to describe epidemic, seasonal and trend-cycle of DF patterns, to determine the best time series forecasting model, to test the predictability of rainfall and temperature and finally, to forecast the monthly OF cases in the next two years (2005-2006) in Peninsular Malaysia which includes 11 states and Federal Territory of Kuala Lumpur. Methods: A quantitative forecasting study was conducted using reported monthly DF cases (including DHF), rainfall and temperature during eight years, from 1997 to 2004. The OF data were obtained from the Vector Borne Diseases Section, Ministry of Health Malaysia and rainfall and temperature data was obtained from Malaysian Meteorological Service, Malaysia. The patterns of DF were described using decomposition time series method. The best time series forecasting model was identified among six time series models. The predictability of rainfall and temperature was tested using linear regression model and ARIMA mixed model. Finally, the best model was used to forecast the DF cases in the next two years (2005-2006).Results: During eight years, all states in Peninsular Malaysia faced epidemic outbreak, especially in 1998 and 2002. Seasonal variation was found in all states of Peninsular Malaysia except Selangor, Malacca, Perak and Kedah. Increasing trend of DF cases was observed in seven out of twelve study areas. Decomposition m~thod was identified as the best forecasting model for all states except Perak and Terengganu where Winter's and ARIMA with seasonal components model were the best respectively. However, rainfall and temperature were not found to be good predictors for DF. With a cut point (i.e. mean+2SD) for epidemic, several states such as Malacca, Kedah, Penang, Perlis, and Terengganu were forecasted to have at least one month of epidemic during 2005 to 2006. Conclusion: The study successfully identified the best forecasting model, and calculate forecasts for the next two years for each state in Peninsular Malaysia, which could help the health management in carrying out the activities of DF prevention and control by making the appropriate choice of strategy.