健保新版住院診斷關聯群變異係數習性之實證研究

Study on Behavior of Coefficient of Variation for Taiwan's Newly Version DRGs

DRGs 變異係數對其相對權數計算或是DRGs 實施前後的費用異常管理， 具有很大的分析意義。鑒於傳統上僅做DRGs 醫療總費用的剖析，對於各項子 費用上的變異情形甚少探討。本研究利用試題反應理論（現代測驗理論）的 Rasch （ 1960 ）分析，進行（ 1 ）醫療費用變異係數特質能力表現上的單一向 度檢測、（2） Pearson 相關分析及複迴歸分析單一向度內的費用子項目，以選 出影響DRGs 總醫療費用變異係數的關鍵項目、（3） 統計分析各主要診斷分類 （MDC） 的費用變異係數排名及其在各子費用上統計顯著（p<.05 ）變異的異 常反應、（4 ）探討子費用變異異常（p<.05 ）項次最多的DRG排名及其異常分 析。 研究發現：手術及麻醉費未呈相同的費用變異特質向度；診察費、病房 費、藥事費、檢查費、特材費、藥費等六項子費用對醫療總費用變異係數影 響最大，迴歸分析解釋力達0.72 ；各MDC 中以傳染疾病在變異係數值最大 （p>.05） ； MDC在各費用指標上困難度走勢呈U 字型態的分佈，常規費用變 異小而技術費用變異大； DRG40001 呈現最多個（6 個）異常費用變異反應。 現代測驗理論較傳統測驗理論提供更多的訊息，後續研究者可以再對健 保第三版DRGs 的相對權數計算、標竿學習醫院間DRGs 費用異常之挑選、病 例組合（CMI） 對層級醫院間影響及醫院開DRGs 住院日數等進行比較與分 析，並提供健保局或是醫院管理階層管理決策上的參考。

Coefficient of variation (CV) on DRGs renders a paramount important tool, especially on its total medical fees, for hospital management in calculating DRG relative weights and the containment of medical fees. Due to the difficulty in analyzing medical sub-item fees with classic test theory (CTT), we adopted Rasch (1960) analysis of modern test theory (MTT) to proceed following depositions: (1) examining a single construct of CV latent trait on medical fee-items; (2) selecting key influencing factors on total medical fee of DRGs through Pearson's Correlation and multi-regression analysis on those medical fee-items under a unidimensionality; (3) ranking CV of each MDC and statistically significantly abnormal responsiveness on sub-item fees; (4) investigating and ranking specified DRGs encountering most unexpected sub-items. Results found in this study were (1) fees of operation and anesthesia excluded from a single construct associated with other fees; (2) fees of diagnoses, wards, pharmacist services, examination, special materials and drugs associated with key factors on CV of total medical fees (R squared =0.72); (3) infectious disease ranking first on CV in MDC, difficulty distribution on sub-item fees shaped a U-type; (4) DRG40001 had much most frequency of unexpected item fees encountered. Offering much more information than CTT, Rasch analysis could be used on further researches such as DRG relative weights calculating on Taiwan's newly version, selections of abnormal fees for insurance company, impacts of CMI on various levels of hospitals and CV of MDC across sub-item fees in an effort to help hospital managers make proper decisions.