"Some deaths involved more than one type of drug; these deaths were included in the rates for each subcategory. Therefore, categories of deaths presented are no

"Some deaths involved more than one type of drug; these deaths were included in the rates for each subcategory. Therefore, categories of deaths presented are not mutually exclusive." So the count includes people who died after taking both cannabis and a more dangerous drug. These crude death rates aren't a good indication of safety. For instance, antidotes like Naloxone are presumably also present in a large fraction of overdose autopsies, but that doesn't mean Naloxone itself is unsafe. Determining causation from statistical data is hard. I think USAFacts might be doing a disservice overall, by presenting raw data from which wrong conclusions can be easily drawn.