If the 20-somethings today are substantially healthier than the 20-somethings 10 years ago, then quality statistics _should_ capture the effect.