There is strong experimental evidence that physiologic stress from high temperatures is greater if humidity is higher. However, "heat indices" developed to allow for this have not consistently predicted mortality better than has (dry-bulb) temperature. To clarify, we explored models for daily mortality as a function of both temperature and humidity in a large international dataset. Methods. In each of 395 cities in 17 countries we fitted a time series regression model to summer mortality with distributed lag non-linear model (DLNM) for temperature (to lag 3) and supplemented this with a range of terms for relative humidity and its interaction with temperature. City-specific associations were summarised using meta-analytic techniques. Results. There was significant variation in the association between cities and countries, with the overall association of humidity with cumulative lagO-3 mortality close to null. At very high temperatures risk did on average rise slightly with high humidity. For example, on average at 99th centile summer temperature (Vs MMT) increase=13.6% at median humidity and 0.8% higher (95%CI 0.1-1.5) at 99th centile humidity. In contrast, below the 90th centile temperature the 99th centile humidity increment was null 0.0%(-0.1,0.2). The humidity-related excess risk was overwhelmingly at lag 0, with high humidity at lag 1 usually associated with LOWER mortality, possibly reflecting an increased risk with upward change in humidity between lags 1 and 0. Results using dewpoint were very similar. Discussion. In general, allowing for humidity in addition to temperature added only marginally to ability to predict daily summer mortality in a city in models considering lags 0-2, with risk increments for higher humidity restricted to periods of very high temperature. However, in specific countries or cities atypical for this data set, or for different lag intervals, patterns may be different. On behalf of the MCC Collaborative Research Network.
展开▼
