比较识别医院感染的3个指标
本帖最后由 樵夫 于 2012-10-11 21:50 编辑Presented in part as an Abstract at the 5th Decennial International Conference on Healthcare-Associated Infections, March 21, 2010, Atlanta, GA.
Alan M. Stamm, MD, Christopher J. Bettacchi, MD
Department of Medicine, University of Alabama at Birmingham, Birmingham, AL
published online 22 June 2012.
Background
The best approach to measurement of health care-associated infection rates is controversial.
Methods
We compared 3 metrics to identify catheter-associated bloodstream infection (CA-BSI), catheter-associated urinary tract infection (CA-UTI), and ventilator-associated pneumonia (VAP) in 8 intensive care units during 2009. We evaluated traditional surveillance using National Healthcare Safety Network methodology, data mining with MedMined Data Mining Surveillance (CareFusion Corporation, San Diego, CA), and administrative coding with ICD-9-CM.
Results
A total of 65 CA-BSI, 28 CA-UTI, and 48 VAP was identified. Traditional surveillance detected 58 CA-BSI and no false positives; data mining identified 51 cases but 51 false positives; administrative coding documented 6 cases and 6 false positives. Traditional surveillance detected 27 CA-UTI and no false positives; data mining identified 17 cases but 19 false positives; administrative coding documented 3 cases and 1 false-positive. Traditional surveillance detected 41 VAP and no false positives; data mining identified 26 cases but also 79 false positives; administrative coding found 17 cases and 13 false positives. Overall sensitivities were as follows: traditional surveillance, 0.84; data mining, 0.67; administrative coding, 0.18. Positive predictive values were as follows: traditional surveillance, 1.0; data mining, 0.39; administrative coding, 0.57.
Conclusion
Traditional surveillance proved superior in terms of sensitivity, positive predictive value, and rate estimation. 作者评估了基于NHSN的传统监测法、使用Medmined Data Mining的数据挖掘法和ICD-9编码，结论了传统的方法有较好的敏感性、阳性预测值和率的估计 谢谢 又学习了收获很大{:1_1:}
页:
[1]