Big weather, small health (data) -- statistical models for understanding climate effects on mortality

Quantifying the effects of climate change on human health requires weather data as well as public health data. Arguably however, there are more (publicly available) weather data than health data, for various reasons but mostly due to the variability in how different countries manage their health data. Having too many data sources on weather, say for a specific country, presents the problem of having to either choose the most optimal one or somehow combine them. On the other hand, having too little health data to link to weather data means increased uncertainty in the estimated relationships. In this talk, we present the use of penalized splines in hierarchical models in a) a framework to integrate temperature data sources and b) quantifying the effects of temperature and other weather variables on human mortality data from Cyprus and the UK.