Climate Change and Deaths from Heat and Cold
It is thought that as global temperatures rise, there will be an increase in heat waves and, consequently, an increase in temperature-related deaths. However, it is far from clear whether the number of heat-related deaths will outpace the reduction in cold-related deaths. Of course, as we've come to expect, cherry-picking climate alarmists and climate skeptics can manipulate the data to predict whatever they would like, so it's not easy to find unbiased information on the topic.
Instead of relying on models, we can analyze mortality data from the CDC, which has searchable records going back to 1999. (ICD-10 categories X30 and X31 provide data on heat- and cold-related deaths.) Conveniently, the website also age-adjusts the data. Plotting this along with the U.S. surface air temperature anomaly allows us to determine if there is a correlation between temperature and heat- and cold-related mortality from 1999-2010. (See chart.)
Is there a correlation between temperature and temperature-related deaths? Yes and no. Using the correlation (CORREL) function on Excel, one can see that there is a decent correlation (0.616) between temperature and heat-related deaths, but there is no correlation (-0.026) between temperature and cold-related deaths. (Note: "1" and "-1" are considered perfect correlations, albeit in opposite directions.)
So, what does this mean? Strictly based on the data given, we should expect a warmer planet to increase the number of heat-related deaths, but to have little to no impact on reducing cold-related deaths.
But of course, this data set is rather limited (only 12 years' worth of data from the U.S.), so much more information would be needed to make any firm conclusions. Even the IPCC's AR4, the current "gold standard" on climate change information, is unsure of the impact of climate change on temperature-related mortality.
We know that's a sort of "wet noodle" conclusion, but in science, there is nothing wrong with that: At least it has the virtue of being evidence-based.