Weight loss and increased mortality: epidemiologists blinded by observations?