資訊模組提供健檢訊息之應用研究

A Module for Releasing Check-up Messages from Health Examination Reports

健康檢查報告除顯示檢查項目的正常與異常，醫師憑專業推斷各種病情的可能，對於實證醫學取代經驗醫學的統計分析，卻仍在努力中。本研究試圖利用試題反應理論的Rasch(1960)分析，以2005年參與自 費全身健檢者的21檢查項目，建置網路資訊模組檢測：(1)健檢結果在性別項目功能上是否有差異？(2)受檢者潛在變項的健康指數是否可予估計出來？(3)檢查報告能否依個人健康指數高低對各檢查項目難易度而推判出正常或異常的合理性？(4)由檢查項目推衍出罹患疾病（如骨質疏鬆症）的可能機率。結果顯示，多點計分的檢查報告之測量信度較2點（正常與異常）為高，性別在各檢查項目上存在有試題差異功能，因此應將男女檢查結果分別分析。單向度的健康指數可做為受檢者間「健康度」的相互比較，也是與檢查項目間之合理異常或不合理正常反應機率的分析基礎。釋例於檢查報告中揭示罹患骨質疏鬆症的可能機率，以實證資訊模組可提供健檢複檢更多重要的訊息。

How to release much more useful messages in exception of those normal or abnormal categories from health examination reports is a vital task to be implemented via information hi-tech. We developed a 21-item inventory of health examination, which can be used to investigate effects of examinees and their differences beyond expectation on health status. Rasch analysis was applied in this study to analyze dimensionality, differential item functioning, distribution of health ability and item difficulty, and group differences. A KIDMAP was illustrated to show the results of health examination on internet and the possibilities of related diseases such as osteoporosis. The results show that the inventory is unidimensional and is measuring 'health status of examinees.' The reliability of polytomous response inventory is higher than that of dichotomous one. Of the 21 items, 10 exhibited differential item functioning on gender. If these 21 items are to be treated as criteria, most subjects' health statuses are quite well. These items are too easy to well differentiate individual differences between healthy subjects. However, they are appropriate for less healthy subjects. The information module developed on healthcare will be recommended to further researchers.