Adherence
The apparent non-linear relationship between coffee consumption and mortality (benefit at moderate doses, null at high doses) in the general population is largely driven by confounding from smoking, as the relationship becomes linear and inverse when restricted to never smokers.
If you do not smoke, drinking coffee (even heavily, >5 cups) is associated with a lower risk of death compared to non-drinkers. The 'danger' of heavy coffee drinking seen in general population studies is likely due to the high correlation between heavy coffee drinking and smoking. For non-smokers, more coffee (up to 5+ cups) may offer greater mortality benefits than moderate drinking.
Given that this association became linear and inverse after restricting to never smokers, it is likely that the non-linear association observed in the total population was due to the residual confounding by smoking.
Why this rating
Strong observational evidence from large cohorts with extensive adjustment and stratification.
Source
Association of Coffee Consumption With Total and Cause-Specific Mortality in 3 Large Prospective Cohorts
Ming Ding et al. · Circulation · 2015
This is one finding among thousands. Every one is graded and traced to its source, so you can see what the evidence actually supports. Browse the research →