A data-driven approach to setting trigger temperatures for heat health emergencies.

OBJECTIVES: Unprecedentedly hot weather during the summer of 2009 resulted in considerable excess mortality in Greater Vancouver, Canada. Local municipalities and public health authorities requested a rapid, evidence-based recommendation for the temperature above which emergency action plans should be triggered to reduce potentially-avoidable mortality during future events.
METHODS: Candidate trigger temperatures were identified by examining the coincidence of extreme mortality days with extreme temperature days, using temperatures observed at two regional airports. Days when the two coincided between 2005 and 2009 were defined as historical heat health emergencies. Forecast and observed temperatures were combined in multiple early warning scenarios to retrospectively test the capacity to predict those heat health emergency dates, and results were expressed in terms of true positive (emergency predicted when one occurred) and false positive (emergency predicted when one did not occur) triggers.
RESULTS: Extreme mortality was observed when the 2-day average of maximum temperatures was > or =31 degrees C at the coastal airport and > or =36 degrees C at the inland airport. When observed and forecast temperatures were combined in different early warning scenarios, all historical heat health emergencies were correctly identified in four of twelve cases, with a minimum of two false positive triggers.
CONCLUSIONS: A heat health emergency should be triggered for Greater Vancouver when the average of the current day's 14:00 observed temperature and the next day's forecast high is > or =29 degrees C on the coast and/or > or =34 degrees C inland. This condition provided 19 hours of lead time for preparation and was clearly understood by emergency responders and other users.