These aren't controlled experiments so how could they possibly rule out selection bias affecting the coffee drinking cohort? Maybe healthier people can handle

These aren't controlled experiments so how could they possibly rule out selection bias affecting the coffee drinking cohort? Maybe healthier people can handle high coffee consumption better for example, and maybe they are more likely to be in sociable and professional settings, both of which are associated with better health.