臺灣博碩士論文加值系統

為了解三種病人住院日數預估法（醫師預估法、 DRGs 預估法、
ICD-9-CM 預估法）在預估日數與病人實際住院日數方面之差異，本研究
立意取樣台大醫院 1995.4.1 至 1995.7.31 日新入院於內、外科五個病
房之 MDC6、7 病人共 707 人，經過樣本處理後實得 564 人， 藉由無母
數分析比較三種預估法預估之準確性。研究結果發現， 三種預估法中以
DRGs 法準確性最高，ICD 法居中，醫師法準確性最低； 醫師法在外科組
別之預估準確性較低，ICD 法在腫瘤診斷碼組別的預估準確性較低，而
DRGs 法在各組別均具預估準確性。 一般說來， 內科病人通常病因較不
明確，預估外科病人之住院日數應較預估內科病人來得容易， 但本研究
發現醫師預估日數在多數外科病人組別顯著較樣本實際住院日數短，其原
因有待進一步探討。本研究進一步探討了病人入出院診斷更動狀況，發現
入出院診斷完全相同的比例為 72.16 ％；入出院診斷不同，但有關者為
22.34 ％； 入出院診斷完全無關者佔 5.50 ％，顯示依 DRGs 法， 採用
入院臨時診斷所編訂之 DRGs 碼為預估病人住院日數之作業方式應具可行
性。由於不確定性（ uncertainty ）因素繁多，病人住院日數預估須作
更多之研究， 尤其若採 DRGs/PPS 支付制度，需要在預估之方法學上作
進一步之探討。

To understand the difference among three methods (MD
method, DRG method and ICD-9-CM method) in predicting
patients'' length of sta y, 707 cases were chosen
purposely. These patients'' admission ICD -9-CM codes were
classfied as MDC6 or MDC7 and they also admitted in five
units of a major medical center during Apr. 1- Jul. 31,
1995. After dealing with inappropriate data, 564 cases were
analy zed with Wilcoxon rank sum test. The author found that
DRG method is the most precise one among three methods. The
MD method has lower precision in surgical groups and the
ICD-9-CM method has lower precision in tumorous groups.
Because of the uncer- tainty about the causes in medical
cases, it''s usually harder to predict patient length of stay
in medical than surgical groups. The reason why MD method
has lower precision in surgical groups in this study needs
further study. The research also found that in all
admission diagnoses, 72.16% are equal to discharge
diagnoses, 22.34% are related to discharge diagnoses,
and 5.5% are unrelated to discharge diagnoses. This result
manifests the DRG method may be used practically. Due to
numerous uncertain factors, prediction of patients'' length of
stay needs to be more studied. If the DRGs/PPS payment
system is implemented, the design of prediction method should
be further discussed.