• Jun Xiao, Rong Zhong, and Gui-shan Ye.
• Intensive Care Unit, Affiliated Hospital of Guilin Medical College, Guilin541001, Guangxi, China.
• Zhongguo Wei Zhong Bing Ji Jiu Yi Xue. 2006 Dec 1;18(12):743-7.

ObjectiveTo compare and evaluate the prognostic value of three prognostic systems namely acute physiology and chronic health evaluation II(APACHEII), simplified acute physiology score II(SAPSII), and Logistic organ dysfunction system (LODS),to predict mortality rate in intensive care unit (ICU).MethodsAPACHEII, SAPSII and LODS were used to assess probabilities of hospital mortality rate and to compare with actual outcome. The overall goodness-of-fit of three prognostic systems were assessed by using the areas under the receiver operating characteristic (ROC) curves and classification tables. Calibration curves and Lemeshow-Hosmer chi square statistics were used to test accuracy of prediction.ResultsHospital mortality rate was basically equivalent to that predicted by APACHEII but higher than those predicted by SAPSII and LODS. The areas under the ROC curves were 0.881 for APACHEII, 0.904 for SAPSII and 0.875 for LODS, and 95% confidence interval (CI) were 0.858-0.904,0.884-0.924 and 0.851-0.898, respectively.The overall correct classification rates, applying the decision criterion of 50%, were 81.60% for APACHEII, 82.09% for SAPSII and 79.26% for LODS. Calibration of APACHEII(chi (2)=9.69) and SAPSII (chi (2)=13.50) were superior to that of LODS (chi (2)=87.22).ConclusionThe prognostic value of discrimination for APACHEII, SAPSII and LODS is good and basically equivalent, but calibration of acuity of APACHEII and SAPSII is much better than that of LODS.

15 keywords...

hide…