Lee-Han H, McGuire V, Boyd N F
Ludwig Institute for Cancer Research, Toronto Branch, Ontario, Canada.
J Clin Epidemiol. 1989;42(3):269-79. doi: 10.1016/0895-4356(89)90062-0.
Studying the association between diet and disease requires reliable and valid methods for the assessment of diet. The authors reviewed the literature concerned with the assessment of these aspects of the measurement of dietary intake. Studies were examined for the stated purpose and scope of the dietary instrument, for a description of the instrument itself, for any methods employed to train individuals in its use and for the methods used to assess its reliability and validity. Of the 59 studies reviewed, 54% described fully the dietary method used. Of the 39 studies that described the results using questionnaires, 51% gave specific information on questions asked and only 18% included the questionnaire itself. Reliability was assessed in 26 studies and 74% (19) used the test-retest reliability and 22% (6) used proxies to assess reliability. Validity was assessed in 46 studies and 83% (38) used indirect methods that compared the results of one dietary method (e.g. 24 hr recall) with another more extensive one (e.g. diet history). Thirty five percent (16) used biochemical and 15% (7) used other methods. This review suggests several directions that might be usefully followed in conducting and reporting further research in the development of methods to assess diet.
研究饮食与疾病之间的关联需要可靠且有效的饮食评估方法。作者回顾了有关饮食摄入量测量这些方面评估的文献。对各项研究进行了考察，内容包括饮食工具的既定目的和范围、工具本身的描述、用于培训个人使用该工具的任何方法以及用于评估其信度和效度的方法。在所回顾的59项研究中，54%充分描述了所使用的饮食方法。在39项使用问卷描述结果的研究中，51%提供了所提问题的具体信息，只有18%包含问卷本身。26项研究评估了信度，74%（19项）使用重测信度，22%（6项）使用替代指标评估信度。46项研究评估了效度，83%（38项）使用间接方法，即将一种饮食方法（如24小时回顾法）的结果与另一种更全面的方法（如饮食史法）的结果进行比较。35%（16项）使用生化方法，15%（7项）使用其他方法。这篇综述提出了几个方向，在开展和报告饮食评估方法开发的进一步研究时可能会有用。
