Identifying flaws in the GWAS datasets of a published Mendelian randomization study: complementary re-evaluation and suggestion for analytical refinements.