I would like to see the original inputs that went into the calculation myself. I've searched quite a bit without being able to locate them. As far as I can tell, it's calculated by dividing the (very low) number of acute fatalities from civilian nuclear accidents by the (very high) number of terawatt hours generated by civilian nuclear power plants. If I'm correct that would kind-of be cheating because they're not counting "mundane" fatal accidents (forklift crushing, arc flash, falls from high places, etc.) that also happen on nuclear sites. The spirit of the comparison that fossil combustion is far more dangerous to human health than nuclear power is correct, but the exact number given for nuclear may be a bit too low.
Additionally, I'd like to see fatality numbers for solar overall, not just rooftop solar. In the US at least more solar capacity is installed at ground-mount plants than on rooftops. The rooftop number itself could probably use an update; increasing module efficiency, if nothing else, is going to reduce the number of rooftop trips required for a given capacity, hence the number of accidents per terawatt hour.
Yes, though both nuclear and solar require significantly less primary materials extraction than fossil generation. The highest impact from coal is actually air pollution affecting the general population rather than on-the-job deaths of miners.
Additionally, I'd like to see fatality numbers for solar overall, not just rooftop solar. In the US at least more solar capacity is installed at ground-mount plants than on rooftops. The rooftop number itself could probably use an update; increasing module efficiency, if nothing else, is going to reduce the number of rooftop trips required for a given capacity, hence the number of accidents per terawatt hour.