급성췌장염의 조기 예후평가 - 다단계판별분석법을 이용한 예후평가

Abstract

Objectives: Early prediction of the severity in patients with acute pancreatitis is very important because severe complications usually occur in the first 48-72 hours. Many studies have been carried out to identify clinical and objective criteria which can be used to predict the outcome of the disease. Multifactor-scoring systems adopted by Ranson et al. and Irmie et al. are generally accepted. But the systems have some disadvantages: (1) too many factors and values have to be assessed, (2) full assessment of the severity takes 48 hours or longer. We, therefore, undertook this study to assess a new method of the early identification of severity which was based on the admission data. Methods: We investigated 339 patients with acute pancreatitis admitted to Severance Hospital of Yonsei Medical Center between January, 1981 and December, 1990. The patients were divided into two groups: those who had no complications or mild complications were classified as mild group (Group I) and those who had severe, systemic complications including death were classified as severe group (Group II). The clinical courses and laboratory findings were analyzed. Results: 1) One hundred two patients experienced one or more complications. Forty patients were classified as severe group (11.8%) and twenty two patients among this group expired (mortality rate, 6.5%). 2) Complications of Group II had developed earlier than those of Group I (2.9 vs. 6.4 days, p<0.05). 3) By stepwise discrimination analysis, a discrimination equation was identified which was composed of calcium, aibumin, BUN, creatinine, uric acid, base deficit, and PaO2 on the admission. Z =4.55－039×Calcium －0.53×Albumin +0.53×BUN×10－1 －0.39×Creatinine+0.14×Uric acid+0.68×Base deficit×l0－1 －0.23×PaO2×10－1 4) The mean Z scores of Group I and Group II were 1.08±0.81 and 2.25±1.34, respectively. The cutting value (Zc) of Z score was identified by chi-square test and it was 1.0. When the cases with higher Z score than cutting value (Z) were considered as severe group, the sensitivity and specificity were 88.2% and 100.0%. The total accuracy of discrimination was 96.2% (p<0.001). The Z score revealed a good correlation with mortality raie (r =0.9825, p<0.01). Conclusion: This prognosis-discriminating equation (Z score) had reliable predictive ability and significant correlation with mortality. Because Z score is a summation of different influences of each factors to prognosis, this method may be more sensitive. And early and relatively simple prediction of severity ran be possible because the parameters of this system are usually obtained on admission. Use of this system should allow early intensive treatment and diagnosis of those who need it. We think that prospective studies should be followed to verify the predictive ability of the Z score of prognosis-discriminating equation.