"We're interested in cause; that's why we often are doing observational research. Authors never face that head on, and don't really come to grips with it," says Andrew J. Vickers, PhD.
In this video, Andrew J. Vickers, PhD, discusses the rationale behind the publication, “Guidelines for reporting observational research in urology: The importance of clear reference to causality,” for which he served as the lead author. Vickers is the statistical editor for European Urology and.an attending research methodologist at Memorial Sloan Kettering Cancer Center in New York, New York.
I'm the statistical editor of European Urology, and I have colleagues on European Urology who are statisticians, Rodney Dunn, and Melissa Assel. Melissa is also a very senior statistical editor at Journal of Urology. So, everybody reads the urology literature; I keep up with urology literature. Then we get all the things that we have to peer review. The way that it works is pretty much every paper with substantive statistics that goes through European Urology or Journal of Urology is going to be seen by one of us. So, we read a very large volume of papers. It struck us that there was a repeated common problem in the papers that we were seeing. They were observational studies, and they hemmed and hawed about the question of causality.
Causality is generally what we want to know. So, for example, we were just discussing this morning, there's some research on Viagra and biochemical recurrence after radical prostatectomy. The only point in doing that research is to think about cause. Does Viagra cause an increase in recurrence rates? Because if it does, we should maybe advise patients not to take Viagra. And yet the papers on it never say, "We're worried that Viagra causes an increase in recurrence rate." In fact, most of these papers never say the word cause or causal or causative. There is no reference to cause. They start by saying something general about prostate cancer and then something general about [how] Viagra is taken by prostate cancer patients, then the hypothesis is something like is there an association between Viagra and biochemical recurrence. Then of course, all the stats methods will talk about control of confounding, which is bizarre because you control confounders to establish cause. That's why you do it, but they don't say that's why they're doing it; they're saying it's to adjust.
Then an odd thing happens in the discussion section. Causal language is then used without being explicit about it. So, words like "impact" or "caution is advised" or "patient should avoid" is used somewhere in the discussion. Then there is the generic boiler plate, Get Out of Jail Free card, "correlation does not prove causation". And then you go back to the very end, the final conclusion is something such as "there may be an association between Viagra and biochemical recurrence," which I guess that may be right. That's why we did the study in the first place. And if you said there is an association, then what do you do with that? We're interested in cause; that's why we often are doing observational research. Authors never face that head on, and don't really come to grips with it. So, that was the problem, and that's why we decided to write these guidelines.
This transcription has been edited for clarity.
Clinicians referring a patient to MSK can do so by visiting msk.org/refer, emailing email@example.com, or by calling 833-315-2722