Without a clear mechanism of action, this sure requires monitoring (like any drug), but the conclusions are terribly oversold. Correlation does not imply causation, no matter the sample size