The trouble with observational cohorts is that it is very difficult to adjust effectively for things that cloud the results (confounders).

If these confounders and biases are not correctly identified and adequately adjusted for, then it doesn't matter how big the cohort is, it just means the magnitude of erroneous results are just on a bigger scale.

The Nurses Health study as a classic example of how to kill people using observational trials. This was a very famous cohort of nearly 50,000 women, which set out to determine factors that influenced women't healt. It was well put together with meticulous follow up over 10 years. One of the findings was that the use of HRT post menopause was protective against cardiovascular events. As a result, hundreds of thousands of women around the world were prescribed HRT to protect their hearts.

It was dead wrong!!!

When the randomised controlled trial was completed, it was discovered that HRT INCREASES the likelihood of cardiac events. (NB for all you women who are on HRT and reading this, my comments are heavily simplified so don't panic and take yourself off HRT - there are good reasons to be on it depending on your circumstances).

In this latest study with coffee, it appears they have also failed to adjust for some significant confounders. As Brian has already mentioned, there are some significant disparities in survival for non-white, lower socio-economic status, insurance status, which cancer centre one attends (US only) as well as the well known ones such as tobacco use, alcohol use and gender. This study appears not to have adjusted for all of these biases. HPV infection does not present in all people equally, so this is likely to be a significant bias as well.

There is also a very different profile for men and women coffee drinkers. High use is most frequent in women who are younger than their non-drinking counterparts whilst the reverse is true for the men. I find that fascinating ...

Having said all that, coffee makes Alex happy and in the absence of data to suggest harm, I am happy to take this information with a pinch of salt and add it to the list of things that MIGHT improve Alex's well being. If Alex were NOT a coffee drinker, I certainly would not push him to start based on these data.