All of those “death count” arguments people make (“communism killed more people than Nazism,” “communism is bad because it killed this many people,” “actually capitalism is worse because it’s killed this many more people”) are all awful and anyone who makes them seriously would flunk out of a high school-level research design course.

1. You aren’t accounting for population, allowing you to attribute the magnitude of deaths in a country resulting purely from its size to the social system it adopts (Preventable deaths in the US are almost certainly higher than in North Korea right now, purely because we have 13 times the population they do)
2. You aren’t accounting for duration of a social system, allowing you to attribute the magnitude of deaths resulting purely from longer timespans to social systems (Nazi Germany existed for 12 years, the USSR existed for 69 years; of course the USSR would have a larger death toll)
3. You’re selectively cherrypicking which forms of preventable death can be attributed to social systems in order to achieve your desired results (Was Germany destined to return to war following it’s post-WWI instability, or did they only return to war due to Nazism’s revanchist militancy? To what extent were various Soviet and Chinese famines the result of poor management or intentional murder, and to what extent were they the result of uncontrollable climatic and agricultural conditions? Depends: how bad do you want these systems to look for the point you’re making?)
4. You’re only accounting for deaths, and not deaths prevented or lifespans expanded under various social systems (if American capitalism is to be blamed for deaths due to lack of access to medicine, is it not also to be attributed the life years gained by capitalist improvements in medicine? If Maoist communism is to be blamed for the atrocities of the Great Leap Forward, is it not also to be attributed the fact that, on net, life expectancy grew faster under Mao than in most other developing countries of the time period (see the “Life expectancy at birth” tool here)?)

If you were serious about doing something like this, you’d need to find objective measures of annual mortality rates and then do something like averaging them across the years of a systems duration and then somehow controlling them in a consistent fashion for starting conditions, age distributions, environmental conditions, and more so as to try and only leave the deaths that could be reasonably attributable to social, economic, and political decisions, but even then you’re still missing important stuff like cross-national spillover effects.