Can outbreak signal of the Bureau of Epidemiology really detect aberration?: A case study of typhoid fever, 2009 - 2013

Abstract

Early outbreak detection is one of important functions of epidemiological surveillance which can initiate timely and efficient control measures. Several statistical methods that support this surveillance function are currently available. This study aimed to assess attributes of outbreak signal system of the Bureau of Epidemiology (BOE), adapted from a method developed by the US Centers for Disease Control and Prevention, and typhoid fever was chosen as a case study. Weekly number of typhoid fever reported in national notifiable disease reporting system was analyzed to identify signals and aberrations occurred during 2009-2013. In this study, 44 abnormal weeks, which were grouped into 10 abnormal events, were identified. At the signal threshold level of 1 standard deviation (SD), sensitivity of detecting abnormal weeks and abnormal events were 43% and 40% respectively whereas those of the signal level of 2 SD were 25% and 30% respectively. For both signal threshold levels, positive predictive values were 100% and median delayed signal times were one week after abnormal events. These findings suggested that sensitivity of the BOE signal system should be increased, either by reducing threshold level or using other alternative outbreak detection methods.