This kind of "study" is probably the weakest kind of research there is, yet studies like this get published all the time. It is what I call a "correlational study." All it does is gather tons of data, and then apply statistical tests to see if any of the pieces of data can be correlated to other pieces of data.
At the p < 0.05 significance level, 1/20 of these correlations are false (of course, you don't know which one). And the remaining 19 correlations are worthless without further data. (A correlation is only meaningful if a reason for the correlation can be determined.) Correlation is not causation, yet all kinds of conclusions are made whenever a correlation is found.
Often, these "studies" are conducted by people who are very familiar with statistics (or have access to a statistician), and not at all familiar with research design or hypothesis formulation. For many MDs, it satisfies their research requirement (often required as part of their training), without actually making them do research. A correlation is the basis of a hypothesis and signals the need for more study; it is not an end result.
Okay, sorry for going off like that, but this is one of my pet peeves. I'm getting off the soapbox now...
That’s awesome and I agree.