This may be a good place to start: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5739436/

Conclusion: Our results indicate excessive mortality among the homeless as well as the weak and rather typical influence of atmospheric conditions on mortality rates in this subpopulation, except for a greater risk of cold related deaths than in the general population... Deaths caused by hypothermia were thirteen-fold more frequently recorded among the homeless than for the general population. A relative risk of death for a homeless person even in moderate cold stress conditions is higher (RR = 1.84) than in thermoneutral conditions.

Averaged? 1.9 per day, 700 yearly. https://nationalhomeless.org/wp-content/uploads/2014/02/Wint...

Throughout the US from 1999 to 2011, an average of about 1,301 per year (but not exclusive to homeless). https://www.cdc.gov/mmwr/preview/mmwrhtml/mm6151a6.htm https://www.cdc.gov/nchs/data/nhsr/nhsr076.pdf

Also there was an average of 658 deaths per year due to heat or humidity (again, not exclusively homeless). https://www.accuweather.com/en/weather-news/prolonged-exposu...

These deaths are mostly preventable by just giving people a room to sit in that's warmer/colder than the outside air, for a very small number of nights/days. An outreach program during these periods of extreme temperatures could have a big impact.

Some for sure, I've heard plenty of news about incidences. Sub-freezing isn't too bad though, especially with layers and a good sleeping bag. The worst happens if you get wet or are exposed to wind.

In the past, I notice cold snaps affect southern homeless more. Those up north know how to prepare for -20, some down south might not survive an overnight freeze.