Background
As recognized by WHO, health is influenced climate change. Several studies have already provided the association between ambient temperature and mortality, hospital admissions and affluence to urgency services, in elderly population, especially due to cardiovascular and circulatory diseases. However, few studies have explored the association between elderly mortality in winter and extreme cold weather, considering as covariables different meteorological indices.
The present study aimed to assess which meteorological index is the best descriptor for winter mortality in elderly population living in Lisbon district.
Methods
Mortality data was provided by Statistics Portugal (INE), meteorological data from The Portuguese Institute for Sea and Atmosphere (IPMA) and influenza-like-illness rates from Portuguese general practitioners (GP) sentinel network (Rede Médicos-Sentinela).
Distributed lag linear and non-linear models (DLNM) were applied to study the effect of cold on mortality by all causes of death, and, particularly, by circulatory and respiratory diseases, in the Lisbon district, in the winter season (from November to March) between 2002 and 2012. Based on different combinations of the meteorological variables (that is, mean temperature, mean temperature and wind speed, mean temperature and humidity, and windchill temperature), several models were fitted and their performance compared. All models were adjusted for trend and seasonality, and for the confounding effect of flu activity. As a reference for relative risk (RR) calculations, the 50th percentile of each temperature series was used.
Results
The best fit was found from a linear relation between temperature (either mean or windchill) and both mortality causes under study (all causes, and circulatory and respiratory diseases). The results showed that the effect of cold appears with delay and persisted for about 23 to 30 days. The maximum effect occurs with the lowest temperature registered (Mean Temperature=-0,4ºC and windchill Temperature=-3,96ºC) and with a delay of 5 days.
The highest cumulative relative risk for all causes of death was found using the windchill temperature [RR=1,8 (CI95%: 1,7; 2,0)]. For mortality by circulatory and respiratory diseases, the highest cumulative relative risk was also found using the windchill temperature [RR=2,0 (CI95%: 1,8; 2,3)].
Conclusions
Cold weather seems to be a strong predictor of mortality in Lisbon district, with the strongest association found out between cold temperature and both circulatory and respiratory mortality. Windchill temperature seems to be a better predictor of mortality than mean temperature.