- People have become more aware of the risk factors, and behave accordingly. Awareness would be connected to prevalence but have a lagged effect.
- Deadly diseases cause evolutionary responses. Perhaps a disproportionate number of people with predisposition to these illnesses passed away before having offspring.
- Reporting bias? While an illness looms large in the minds of the medical community, doctors are more likely to either wrongly attribute to the illness (false positive), or do false negatives less often.
- Highly speculative: combined effects are non-linear. I don't know what they use to do these studies, but typically you hear something along the lines of "for every x, there's m*x effect", which makes things sounds nice and linear. Maybe better prevention and better treatment does better than either on its own summed up, and so you won't be able to find the "reason" by splitting into each feature.
