| Summary: | 博士 === 國立陽明大學 === 公共衛生研究所 === 99 === Healthcare-associated infections affect wellness of patients and hospital staffs. It also consumes a lot of medical care resources. Manual review of patient records and clinical laboratory data to detect healthcare-associated infections remains cumbersome and time-consuming tasks and is lack of time efficiency. The purpose of this study was to develop an automatic surveillance system for healthcare-associated urinary tract infections (UTI) and evaluate the effectiveness of this system.
The retrospective study was conducted at a 2,990-bed medical center. A total of 10,787 hospitalized patients with positive urine microbiological culture results from 2005 to 2007 were selected. The study data set contains demographic information, admission-discharge-transfer data, laboratory reports, and medication orders. This study was carried out in two stages. In the developing stage, a total of 4,220 patient electronic records were utilized as training data set to construct an automatic surveillance system. The rule-base method was used to develop detection rules for healthcare-associated UTI. In the validation stage, a total of 7,250 patient electronic records were used as testing data set to evaluate the performance of this system. The accuracy between the manual review by infection control practitioners (ICPs) and automatic surveillance system were compared to establish sensitivity, specificity, positive predictive values (PPV), and negative predictive values (NPV) of this system.
The results of this study were (1) an automatic surveillance system was established. It includes the architecture of data warehouse, the algorithm of detection in healthcare-associated UTI according to standard definitions proposed by the Centers for Disease Control and Prevention, and information technology environment. (2) In the
developing stage, 980 of 4,220 patients had healthcare-associated UTI were identified by infection control practitioners (infection rate is 23.2 %). The automatic surveillance system had a sensitivity of 99.1 %, specificity of 68.2 %, PPV of 48.6 %, and NPV of 99.6 %. In the validation stage, 1,821 of 7,250 patients had healthcare-associated UTI were identified by infection control practitioners in (infection rate is 25.1 %). This automatic surveillance system had a sensitivity of 99.0 %, specificity of 74.6 %, PPV of 56.7 %, and NPV of 99.5 %. The automatic surveillance system can effectively reduce 55.9 % workload of ICPs on chart review or survey of healthcare-associated UTI.
The results of this research demonstrate that this automatic surveillance system had an approximately 100 % of sensitivity and can effectively facilitate the survey of healthcare-associated UTI. The automatic surveillance system could be applied as a useful screen tool of healthcare-associated UTI. It is hoped that with the help of the system, the overtime work and manpower shortage problems of ICPs could be improved to some extent. Eventually, patients with healthcare-associated UTI could be identified and treated earlier and largely increase the quality of healthcare-associated infection detection for hospitalized patients.
|
|---|
