压疮危险因素分析与判别分析模型的建立

Logistic regression analysis of risk factors of pressure sore and building of discriminant analysis model

目的通过分析压疮的危险因素,建立压疮判别分析的模型。方法收集2005年1月—2006年12月我院预报压疮及压疮患者相关资料1 292例,运用多元Logistic回归法进行分析,找出与压疮密切相关的危险因素,建立压疮判别分析模型,以此模型对2008年1—6月入院的呼吸内科、神经内科和肿瘤科住院患者303例进行预测,并与其压疮实际发生情况对比,以检测模型的敏感度、特异度、假阳性率、假阴性率和符合率。结果与压疮发生密切相关的有潮湿情况、活动情况、移动情况、白蛋白、精神意识、年龄及血红蛋白等7大因素;由此建立的压疮判别分析模型敏感度为76.09%,特异度为84.82%,假阳性率为15.18%,假阴性率为23.91%,符合率为83.50%。结论根据压疮密切相关因素而建立的压疮判别分析模型预测压疮符合率高,能简便、有效、可靠地对压疮的发生进行预测,为临床准确防范压疮提供了依据,有利于提高护理工作的有效性。

Objective To analyze the risk factors of pressure sore using logistic regression, and build the discriminant analysis model to predict pressure sore incidence. Methods The relevant information of pressure sore of 1 292 patients in our hospital from Janu 2005 to December 2006 were collected, and the multiple logistic regression was used to analyze and identify the risk factors closely related to pressure sore. So the discriminant analysis model was built to predict pressure sore incidence of 303 patients in departments of respiratory, medicine, neurology and oncology from Janua to June 2008. The sensitivity, specificity, false-positive rate, false-negative rate and coincidence rate of the model were detected by comparing with the accurate occurrence data. Results Wet condition, physical activity, movement, albumin, mental awareness, age and hemoglobin were closely related with pressure sore. Of the discriminant analysis model, the sensitivity was 76. 09%, specificity was 84. 82% , false-positive rate was 15.18% , false-negative rate was 23.91 % , and coincidence rate was 83.50%, respectively. Conclusion The discriminant analysis model built based on the closely related risk factors of pressure sore can predict pressure sore incidence simply, efficiently and reliably. Moreover, it also can provide basis for clinical prevention of pressure sore and improve the efficiency of nursing work.