简易呼吸衰竭评估系统的初步探讨

Preliminary study of simplified useful respiratory failure evaluation system

目的建立适用于评估呼吸重症监护病房(RICU)患者病情、预后及疗效的评分系统简易呼吸衰竭评估系统(SURFES),并分析运用SURFES评估呼吸衰竭患者病情的效果,以期更好地适用于基层医院。方法采集模型组102例患者与病情、预后及疗效有关的数据,采用因子分析方法,筛选有关的指标;采用Logistic回归方程确定各指标的等级分值,形成初步评分系统;根据SURFES评分系统重新评估模型患者的分值,建立死亡概率方程;根据实际病死率对初步形成的权重系统进行校正,结合专家意见形成最终的权重系统;最后对验证组101例患者同时采用SURFES以及急性生理和慢性健康状况评分系统Ⅱ(APACHEⅡ)进行评分,对比分析两系统的优劣,评估SURFES的价值。结果最终建立SURFES指标包括:二氧化碳分压(PaCO2)、脉搏、pH、神志评分、收缩压、呼吸频率;采用SURFES系统对模型组102例评分,与APACHEⅡ的结果比较,两者准确度均为86.30%,卡帕(Ksppa)值=0.570,P=0.000;SURFES对验证组评估得到预测的灵敏度为83.33%、特异度88.68%、准确度86.10%、阳性预告值(+PV)86.96%、阴性预告值(-PV)85.46%、阳性似然比(+LR)7.36、阴性似然比(-LR)0.19;两系统受试者工作特征曲线(ROC)面积比较,Z=0.1218,P>0.50;SURFES对验证组进行预后评估得到的预测结果与实际结果比较差异无统计学意义(χ2=7.207,P=0.514),与APACHEⅡ系统相比(χ2=10.243,P=0.175)其一致性检验示Ksppa=0.681,P=0.000;SURFES分值与实际病死率的相关性,r=0.683,P=0.000;随着分值的增加,病死率随之增高,但是当分值超过9分时,病死率急剧增高(约70%),这一点与模型组的ROC曲线的截断点(8.5分)一致。结论SURFES指标较少,且较其他系统更适用评估RICU患者病情的变化,可作为RICU专用的评分系统。

Objective To establish a scoring system-simplified useful respiratory failure evaluation system （SURFES） - to predict the illness severity of patients in respiratory intensive care unit（RICU） and to evaluate the SURFES so as to make this scoring system applicable to the grassroots hospitals. Methods Data consisted of 102 respiratory failure patients （estimation group） from RICU were collected in order to screen for prognostic factors affecting the outcomes of respiratory failure patients. By factor analysis, scores were allocated to the corresponding interval respectively by means of Logistic regression to establish a primary scoring system. Based on this primary scoring system, the probabilistic equation was established to reappraise patients from this estimation group. A definitive evaluation scheme is resulted by combining actual mortality and by a panel of pulmonary experts＇ evaluations. The patients from validation group （101 cases） were scored and calculated by their predicted mortality with SURFES and acute physiology and chronic health evaluation Ⅱ （APACHE Ⅱ ）. Comparing the performance of the SURFES and APACHEⅡ by means of statistical analysis,the validity of SURFES can be delineated. Results PaCO2, pulse, pH, conscious state,systolic blood pressure, respiratory rate were associated with prognosis of respiratory failure. To the estimation group, the accuracies of SURFES and APACHEⅡ were both 86.30 % and the Ksppa value was 0. 570, P = 0. 000. The sensitivity, specificity, accuracy, positive predictive value （ ＋ PV ）, negative predictive value （ - PV）, positive likelihood ratio（＋ LR） and negative likelihood ratio（-LR） of SURFES for the validation group were 83.33%, 88.68%, 86.10%, 86.96%, 85.46%, 7.36 and 0.19, respectively. In the areas under the receiver operating characteristics curves （AUC）, no significant differences between the two systems （ Z = 0. 1218, P 〉 0.50） can be detected. There are no significant differences between the predictive results and the actual results obtained by SURFES for the validation group （X^2= 7. 207, P = 0. 514）. In comparison to APACHE Ⅱ （X^2= 10. 243, P = 0. 175）, the Ksppa value of SURFES was 0. 681, P = 0. 000. The scores correlated positively with the mortality rate. The mortality rate increasd rapidly when the score exceeded 9 points （about 70%）, which was uniform with the cutoff of the operating characteristics curves （ROC） in the estimation group （8.5 points）. Conclusion SURFES has less variables and is much superior to the APACHE Ⅱ scoring system in predicting the illness severity of patients in RICU and can be employed as a valuable scoring system in RICU.