Analytical Sciences

Detection Limit Used for Early Warning in Public Health Surveillance

Tsuyoshi KOBARI,*¹ Kazuo IWAKI,*² Tomomi NAGASHIMA,*² Fumiyoshi ISHII,*³ Yuzuru HAYASHI,*⁴ and Takehiko YAJIMA*⁵

*1 Kosumo Chouzai Yakkyoku Co., Ltd., 2-129-3 Miyahara, Kita, Saitama 331-0812, Japan
*2 School of Pharmaceutical Sciences, Ohu University, 31-1 Misumido, Tomita-machi, Koriyama, Fukushima 963-8611, Japan
*3 Department of Pharmaceutical Sciences, Meiji Pharmaceutical University, 2-522-1 Noshio, Kiyose, Tokyo 204-8588, Japan
*4 National Institute of Health Sciences, 1-18-1 Kami-Yoga, Setagaya, Tokyo 158-8501, Japan
*5 Faculty of Pharmaceutical Sciences, Toho University, 2-2-1 Miyama, Funabashi, Chiba 274-8510, Japan

A theory of detection limit, developed in analytical chemistry, is applied to public health surveillance to detect an outbreak of national emergencies such as natural disaster and bioterrorism. In this investigation, the influenza epidemic around the Tokyo area from 2003 to 2006 is taken as a model of normal and large-scale epidemics. The detection limit of the normal epidemic is used as a threshold with a specified level of significance to identify a sign of the abnormal epidemic among the daily variation in anti-influenza drug sales at community pharmacies. While auto-correlation of data is often an obstacle to an unbiased estimator of standard deviation involved in the detection limit, the analytical theory (FUMI) can successfully treat the auto-correlation of the drug sales in the same way as the auto-correlation appearing as 1/f noise in many analytical instruments.

J-STAGE: View this article in J-STAGE