Research published by The BMJ suggests inaccurate ... it may have something to do with the accuracy of coding data, since many "weekend effect" studies use hospital administrative data.