You don't trust health data from tobacco companies either, but you can still figure out ways to analyze the health of tobacco.

This certainly is not my wheelhouse, but I assume someone has made some attempts to analyze this?