The value and hazards of standardization in clinical epidemiologic research