Malaria Infection and Human Behavioral Factors: A Stochastic Model Analysis for Direct Observation Data in the Solomon Islands MINATO NAKAZAWA, 1 * HIROSHI OHMAE, 2 AKIRA ISHII, 3 AND JUDSON LEAFASIA 4 1 Department of Human Ecology, School of International Health, The University of Tokyo, Tokyo 113, Japan 2 Bureau of International Cooperation, International Medical Centre of Japan, Tokyo 162, Japan 3 Department of Medical Zoology, Jichi Medical School, Tochigi 329-04, Japan 4 Solomon Islands Medical Training and Research Institute, Ministry of Health and Medical Services, Honiara, Solomon Islands ABSTRACT The effect of human behavior on malaria infection was in- vestigated in a holoendemic area in the Solomon Islands, using a newly developed mathematical model based on the results of blood examinations and 2 weeks of direct observations. The present study indicated that the place where individuals spent the evening had no significant relation to malaria infection, but that some kinds of clothes significantly reduced malaria infec- tion. However, as the result of prediction by the Susceptible-Exposed- Infective-Recovered (SEIR) model, effective control of malaria transmission would require as much as 95% coverage by protection even in the case that the protective measure was highly effective. The results strongly suggested protection measures besides bed net distribution in addition to the necessity of health education. Am. J. Hum. Biol. 10:781–789, 1998. © 1998 Wiley-Liss, Inc. Medical and entomological studies have been conducted for several years to control malaria in several villages of the Solomon Islands (Ohmae et al., 1991; Okazawa et al., 1991; Ishii, 1993; Wataya et al., 1993). Un- der the malaria control program, acceptance of the program by residents is an important factor for its effectiveness (Lengeler and Snow, 1996). For instance, the efficacy of distribution of bed nets largely depends on whether the nets would be properly used or not. In Malaysia, bed nets were distributed but were not properly used (e.g., people watched television until midnight outside of the net), so that malaria incidence had not been reduced (Leake and Jeffrey, 1994). In the Solomon Islands, because the main ma- laria vector is Anopheles farauti, which has its highest activity in the early evening, people have to stay under bed nets from just after sunset in order to achieve sufficient efficacy. Thus, an aim of this study is to ob- tain information on human behavior involv- ing mosquito bites by direct observation. Human activity surveys have been devel- oped in the discipline of ecological anthro- pology. In 1930s, diaries were sometimes used, but these were not adequate for quan- titative measurement. Direct observation is necessary for individual-based risk assess- ment. Suda (1994) has summarized direct observation methods as follows: 1) indi- vidual tracing, 2) observation at a fixed lo- cation, and 3) random spot-checking. Indi- vidual tracing has advantages in accuracy of observation but has limitations in terms of sample size. Observation at a fixed loca- Contract grant sponsor: Japanese Ministry of Education, Sci- ence and Culture. *Correspondence to: Minato Nakazawa, Department of Hu- man Ecology, School of International Health, the University of Tokyo, Hongo, Bunkyo-Ku, Tokyo, 113 Japan. Fax: +81-3-5684- 2739. E-mail: minato@humeco.m.u-tokyo.ac.jp Received 19 August 1997; Accepted 20 December 1997 AMERICAN JOURNAL OF HUMAN BIOLOGY 10:781–789 (1998) © 1998 Wiley-Liss, Inc.