Nonlinear and threshold of the association between meteorological factors and bacillary dysentery in Beijing, China.

Previous studies examining the weather-bacillary dysentery association were of a large time scale (monthly or weekly) and examined the linear relationship without checking the linearity assumption. We examined this association in Beijing at a daily scale based on the exposure-response curves using generalized additive models. Our analyses suggested that there were thresholds for effects of temperature and relative humidity, with an approximately linear effect for temperature >12·5 °C [excess risk (ER) for 1 °C increase: 1·06%, 95% confidence interval (CI) 0·63-1·49 on lag day 3] and for relative humidity >40% (ER for 1% increase: 0·18%, 95% CI 0·12-0·24 at lag day 4); and there were linear effects of rainfall (ER for 1-mm increase: 0·22%, 95% CI 0·12-0·32), negative effects for wind speed (ER: -2·91%, 95% CI -4·28 to -1·52 at lag day 3) and sunshine duration (ER: -0·25% 95% CI -0·43 to -0·07 at lag day 4). This study suggests that there are thresholds for the effects of temperature and relative humidity on bacillary dysentery, and these findings should be considered in its prevention and control programmes.