Getting properly done studies is hard. In particular the statistics are even harder and easier to mess up (the retracted Lancet paper had also statistical problems).