In Japan, few new cases of tuberculosis are registered during the winter, and the incidence tends to increase during summer, but few detailed statistical analyses of seasonality have been reported. In the present study, we evaluated data regarding newly registered cases of bacillus-positive pulmonary tuberculosis in patients older than 70 years and living in the Kanto district; these cases were identified through clinical examinations during 1992–1998. We obtained our information from the tuberculosis surveillance data of the Ministry of Health, Labour and Welfare in Japan. We evaluated seasonality and the relationship between disease onset and various meteorological factors for 2 groups within our study population, which were categorized according to the interval between the estimated onset of disease and its registration, i.e., <2 or <3 months (4117 and 5281 cases, respectively). Cyclical components with a 12-month period were most prevalent in the power spectrum analysis of the time series fluctuation in morbidity. Monthly morbidity tended to be high from June to October and was reduced during December and January, and the morbidities during summer and autumn were significantly higher than during winter. A positive correlation was observed between the morbidity, air temperature, and relative humidity when the estimated onset was 1 month before registration. In addition, morbidity and relative humidity seemed to be correlated when the onset was estimated to have occurred 2 months before registration, but the correlation coefficients were under 0.5, and the relationship between morbidity and air temperature or relative humidity was not always clear. For our elderly patients whose disease was registered <2 or <3 months after onset, morbidity began to increase in April or May, because the onset of tuberculosis is before about 1 or 1.4 months in the registered as midpoint value was estimated, respectively.
View full abstract