Protective effect of smoking: misleading use of statistics