My own take. The point is that the link may be the other way around. The population of people who tend to drink coffee might age slower. Perhaps due to some third variable like wealth or race. Correlation is not causation.