Influence of using different sources of carotenoid data in epidemiologic studies.