老年呼吸衰竭机械通气患者预后评估系统的建立与评价

A system for evaluating treatment efficacy and outcome prediction for senile patients with respiratory failure undergoing mechanical ventilation

目的设计老年呼吸衰竭机械通气患者的预后评估系统并对其进行分析验证。方法回顾性分析呼吸衰竭患者238例,随机分为模型组(138例)和验证组(100例)。对模型组数据先后采用单因素和多因素分析筛选变量,对筛选出的变量划分等级分值,建立补充评分系统,采用Logistic多元回归方程组合APACHEⅡ系统和补充评分系统,形成老年呼吸衰竭机械通气专用评分系统(简称SRFMV系统)。采用SRFMV系统评价验证患者,并与采用APACHEⅡ系统者比较,以评价SRFMV系统的真实性与可靠性。结果筛选氧合指数、呼气末正压(PEEP)值、潮气量、痰液量和性状以及肺部听诊情况作为老年呼吸衰竭机械通气患者的补充评分标准。运用Logistic回归方程建立死亡概率方程。SRFMV系统对验证组的预测灵敏度和特异度(分别为0.878、0.821)均优于A-PACHEⅡ系统(分别为0.818、0.771);SRFMV系统的ROC曲线下面积、Lemesshow-Hosmer拟和优度检验的χ2值(分别为0.911±0.034、13.77)也较APACHEⅡ系统有所提高(分别为0.860±0.037、11.808)。结论SRFMV系统具有较高的灵敏度和特异度,预测死亡率和患者实际死亡率具有较好的一致性,可用于老年呼吸衰竭机械通气患者的预后评估。

Objective On the basis of the Acute Physiology and Chronic Health Evaluation Ⅱ （APACHE Ⅱ）, to establish and appraise a new prognostic evaluation system for elder patients with respiratory failure undergoing mechanical ventilation in order to predict hospital mortality in the respiratory intensive care unit. Methods Two hundred and thirty-eight senile patients with respiratory failure having had mechanical ventilation during their hospitalization were retrospectively analyzed. Patients were randomly divided into 2 groups： model group （n=138） and validation group （n=100）. Data of model group were analyzed by monofactorial and multifactorial regression analysis to screen the risk factors. Risk factors were given numerical values according to different grades, and complementary scoring system was then established. APACHE Ⅱ system was combined with complementary scoring system with Logistic multiple regression to form SRFMV system, a special evaluation system for the senile patients with respiratory failure undergoing mechanical ventilation. One hundred patients from validation group were evaluated by both SRFMV system and APACHE Ⅱ system, and the results were then compared to assess the validity and reliability of SRFMV system. Oxygenation index, positive end expiratory pressure （PEEP）, tidal volume, phlegm quantity and character, and pulmonary auscultation were selected to serve as standard for complementary scoring system. Mortality equation was set up with logistic multiple regression analysis. Results The predication sensitivity and specificity evaluated by SRFMV system in validation group （0.878 and 0.821, respectively） outstripped that evaluated by APACHE Ⅱsystem （0.818 and 0.771, respectively）; the area under ROC curve in SRFMV system and the X2 value in Lemesshow-Hosmer statistic （0.911 and 13.77, respectively） also outstripped that in APACHE Ⅱsystem （0.860 and 11.808, respectively）. Conclusions SRFMV system is of better sensitivity and specificity, in which the prediction of mortality is closely consistent with the reality. SRFMV system can be applied to obtain valid predictions of relevant outcomes in senile patients with respiratory failure underwent mechanical ventilation.