It's worth pointing out that the study cited in the article does use an analysis that looks for causality.
The study says "To formally test mediation [...] calculate indirect effects similar to the Sobel method." I think this refers to a paper "Direct and indirect effects in linear structural equation models" by Michael Sobel. I don't understand enough to say what the limitations are of this method -- but it does sound like rather more advanced causal analysis than you typically see in a medical paper.
I'm guessing that the limitation of the Sobel method is "garbage model in, garbage causal inferences out". It'd be interesting to hear from someone who knows the field, who can say whether the model used in this paper was rich enough to test your ideas (about excess sleep versus insufficient sleep).
The study says "To formally test mediation [...] calculate indirect effects similar to the Sobel method." I think this refers to a paper "Direct and indirect effects in linear structural equation models" by Michael Sobel. I don't understand enough to say what the limitations are of this method -- but it does sound like rather more advanced causal analysis than you typically see in a medical paper.
I'm guessing that the limitation of the Sobel method is "garbage model in, garbage causal inferences out". It'd be interesting to hear from someone who knows the field, who can say whether the model used in this paper was rich enough to test your ideas (about excess sleep versus insufficient sleep).