A twin study is just one form of (natural) experiment. It's common in medicine/epidemiology to perform double-blind trials with placebos, etc. Good medical research utilizes experiments rather than assuming causation from correlation like this study does.

I am a big fan of twin studies, though (my background is in behavior genetics). Take the example of smoking. The most solid demonstration that smoking was bad for your health (aside from maybe the experiments on other mammals) was comparison of twins where one twin smoked and the other didn't.