Shutterstock
You may have seen renewed recent media claims that alcohol is linked to cancer, based on a new commentary in Addiction. While overuse of alcohol can certainly cause something like fatty liver disease and then cirrhosis, so can too many cheeseburgers. Lots of things can be harmful when misused, that is why the American Council on Science and Health talks about dose-response when environmentalists only want to talk about hazard. (1)
It has been commonly established in longitudinal studies that alcohol in moderation has some positive effects, like with cardiovascular disease risk, but then some meta-analyses say it causes cancer. Which is it? If you are in media, it doesn't matter; you write about both as the Big Story. Many of us are reductionists; we want to eliminate bad luck and true randomness as much as possible (2), so we try to find cause and effect for everything. Which is why we love to read articles claiming X is a Miracle Product or later that X will Cause Cancer - and we do, over and over, like last week with alcohol causing cancer. Again.
In March, Stat News and others also warned us again that alcohol may be not be good for you after all. Yet the "after all" is limited to media accounts, both the National Toxicology Program of the US Department of Health and Human Services and the American Heart Association refused to jump on that 'drink away' bandwagon.
What do they know that we don't? Nothing. Despite toxicology being in the name of the HHS group, their warnings are based on epidemiology. And that's the issue. When you are using statistics as science, it can go either way. The reason there is disagreement on drinking and alcohol is because of how analyses are framed. If ex-drinkers and lifelong teetotalers are considered "healthy abstainers" that skews the results, and alcohol is bad. If another paper says they controlled for such abstainer bias and drinkers still live longer, then alcohol is good. Analyses dealing with all mortality can show drinking alcohol is bad, but if it's limited to cardiovascular issues, alcohol is good.
As has been noted many times, using epidemiology, everything can cause or prevent cancer if someone wants it to - that is why environmental groups hate toxicology studies but latch on to every epidemiology meta-analysis that can affirm their beliefs. (3)
For example, in America of 2016 it is clear that epidemiologists are going to have to create a special category for the "Netflix" lifestyle, where people sit on the couch and snack for an entire weekend. That does not mean Netflix causes disease. (4)
Except it can mean just that. In the right hands, namely someone who went into science or health to 'Make A Difference' in their sociological pet cause, a scholar can find a way to show their cause is killing people.
So here it is, the way to show Netflix causes cancer in 10 easy steps:
1. Get a random sample of 1,000 people and ask them to recall their viewing habits
2. Find out how many watch Netflix
3. Find out which disease they share in common
4. Do some data-dredging and use the seductive certainty of significance to suggest that your p-value is relevant
5. Pay to publish it in an open access journal, or find a niche relevant journal desperate for impact factor
6. Pay AAAS Eurekalert to carry the press release
7. Answer emails from harried science journalists writing three articles that day. Send snappy quotes about implications for longevity. Slate, Vox, etc. then write blog posts about the newspaper articles and then Inquisitr, Mashable, etc. rewrite those blog posts
8. Someone writes a New York Times bestselling book based on it
9. Academics needing R01 grants rush to produce papers validating the claims in the book
10. Academics like Marion Nestle write about all of the new papers and declare anyone disagreeing is a shill for Big Streaming.
There you have it. Your first epidemiology paper. And this will get read, because a lot of people have Netflix.
One caution: This is a template that is time-honored, but you don't want to get too specific. Using this method, not only can I show that Netflix causes cancer, I can show watching "Daredevil" does, but the more specific you get, the less suspension of disbelief the public will have.
And suspension of disbelief is important in modern-day epidemiology. It is why the International Agency for Research on Cancer (IARC) backed off on claims that coffee causes cancer - when they were fully prepared to declare it did. Their credibility had already been clobbered with epidemiological suggestions that glyphosate caused cancer despite cancer plummeting as pesticide use went up, and that eating sausage is as hazardous as plutonium.
NOTES:
(1) We can detect parts per quadrillion now, so we can detect a trace chemical in 11,000,000 gallons of water. Environmentalists are basically homeopaths when they claim that is doing anything.
(2) Lots of people get lung cancer that have never smoked, so we want to believe it is caused by second-hand and even third-hand smoking. No one can die from lung cancer today without it being caused by something.
(3) 
(4) Whereas Netflix and chill is good for health. Unless it is bad for health, because you are binge-watching with multiple anonymous partners. See what I mean about epidemiology going either way?
