تجزیه و تحلیل شاخص های کیفیت مراقبت های بهداشتی با استفاده از داده کاوی و سیستم های پشتیبانی تصمیم گیری

This study presents an analysis of healthcare quality indicators using data mining for developing quality improvement strategies. Specifically, important factors influencing the inpatient mortality were identified using a decision tree method for data mining based on 8405 patients who were discharged from the study hospital during the period of December 1, 2000 and January 31, 2001. Important factors for the inpatient mortality were length of stay, disease classes, discharge departments, and age groups. The optimum range of target group in inpatient healthcare quality indicators were identified from the gains chart. In addition, a decision support system (DSS) was developed to analyze and monitor trends of quality indicators using Visual Basic 6.0. Guidelines and tutorial for quality improvement activities were also included in the system. In the future, other quality indicators should be analyzed to effectively support a hospital-wide continuous quality improvement (CQI) activity and the DSS should be well integrated with the hospital order communication system (OCS) to support concurrent review.

An increasing concern with improving the quality of care in various components of the health care system has led to the adoption of quality improvement approaches originally developed for industry. These include ‘total quality management’ (TQM) (Deming, 1986), an approach which employs process control measures to ensure attainment of defined quality standards, and ‘continuous quality improvement’ (CQI) (Juran, 1988), a strategy to engage all personnel in an organization in continuously improving quality of products and services. CQI was originally based on the quality assurance (QA) paradigm, which emphasizes monitoring of incidents, mortality and morbidity audits, and hospital infection audits. However, manufacturing industry experiences have shown that QA programs, which focus on end-product evaluation/audit, have little effect on improving quality or decreasing costs (Sakofsky, 1996). In manufacturing industries, process quality improvement strategies have been proven to be far effective than product oriented quality control programs. At the beginning of the nineties, the emphasis was shifted from the QA paradigm to that of process oriented TQM/CQI, concurrently with the realization of the advantages of the latter throughout the industry. Process improvement strategies operationalize the plan-do-check-act (PDCA) process quality management cycle (Deming, 1986). The outcome targets from the continuous and final QA criteria to be used at quality evaluation-and-improvement checkpoints. The key inputs to the PDCA process are patient assessment/outcome data that are compared to the expected outcome targets and best practice guidelines or protocols. Each set of evaluation results can be used as part of the decision support information for revising the care plan and improving the intervention strategies. In Korea, QA activity has been launched in 1981 as a part of the Hospital Standardization Project organized by the Korean Hospital Association. Since the Korean Society of Healthcare QA was established in 1994, more comprehensive quality management, evaluation, and research have been implemented. The Hospital Service Evaluation System began in 1995 for an evaluation of CQI activities at the tertiary hospitals initially, but was later expanded to the hospitals with less than 200 beds. Recently, for a more systematic and practical evaluation of hospital quality, researches on reconceptualization of CQI, development of QI standard and QI indicators, establishment of a QI department, and development of a QI manual are actively in progress. But because of inadequate utilization of QI evaluation results and feedback, heavy workloads, and lack of motivation of this endeavor, CQI has not been successfully implemented in most hospitals in Korea. Moreover, the majority of QI activities heavily relied on manual processes such as chart audit. However, manual QI activity without its connection to underlying clinical information produced by hospital information system had been criticized as contributing nothing to quality improvement (Luttman, 1993). Therefore, there is a need for a decision support system (DSS) that provides patient assessment/outcome information and a clinical pathway to support the PDCA process. For process quality improvement to be successfully implemented, information on patient care process and the factors influencing quality or treatment outcome must be available at real time for comparison against the desired progress/outcome criteria and development of quality improvement strategies by integrating with the hospital information system. In this study, the factors influencing quality were identified using data mining, and a DSS for process-oriented CQI based on these factors is another key information for the PDCA process. Data mining is a knowledge discovery method from a large-scale information bank such as a data warehouse. Data mining was used in this study in order to identify pattern or rules about various quality problems or indicators from a large-scale data warehouse. While there were several studies on data mining such as identifying significant factors influencing prenatal care (Prather et al., 1997) and automatic detection of hereditary syndromes (Evans, Lemon, Deters, Fusaro, & Lynch, 1997), these systems did not explicitly deal with management issues on CQI activities.

This study presents an analysis of healthcare quality indicators using data mining for developing quality improvement strategies. Important factors influencing the inpatient mortality were identified using a decision tree method for data mining based on 8405 patients who were discharged from the study hospital during the period of December 1, 2000 and January 31, 2001. Important factors for the inpatient mortality were LOS, disease classes, discharge departments, and age groups. The optimum range of target group in inpatient healthcare quality indicators were identified from the gains chart. The cumulative statistics in the gains chart show us how well we do at finding inpatient mortality cases by taking the best segments of the sample. We demonstrated how the decision tree could be used in developing CQI strategies. While statistical methods such as logistic regression can also be used for identifying important factors influencing quality, it does not provide information about the segment characteristics of such factors that may be useful for CQI activity. The CHAID algorithm provided cumulative statistics that show how well we do at finding the inpatient mortality cases by taking the best segments of the sample. The gains chart also provided valuable information about which segments to target and which to avoid. In addition, a DSS was developed to analyze and monitor trends of quality indicators using Visual Basic 6.0. Guidelines and tutorial for quality improvement activities were included in the system. This system has the potential to prevent adverse medical events, improve the quality of care and produce significant savings on healthcare costs. Such system can also provide nurses with valuable process quality and decision support for them to function as effective nurses as well as CQI staff. There were several literatures related to this study. In the study of the factors influencing unscheduled readmission, which is another quality indicator, Oh (1996) found that when age is older, the readmission rate was 1.03 times higher, and when the disease was in the lower risk group, the rate was 0.36 times higher. Furthermore, it was reported that the higher the insufficient discharge schedule points are, the unscheduled readmission rate could be up to 10 times higher. Solberg et al. (1997) used CQI program to improve quality of clinical prevention services for chronically ill patients, especially those with diabetes and reported a reduction in unscheduled readmission. Chu (2001) reported that computerized clinical pathway and DSS could improve the clinical process. In order to apply the DSS for healthcare quality enhancement, the following are recommended: First, in order for the DSS to be successful in improving healthcare quality, there must be strong top management support, active participation and effort of the clinical department to obtain system structure and resources, and monitoring and continuous development according to the amount of task process must occur. Second, for a more effective DSS, hospitals must build a hospital-wide information infrastructure to obtain quality information from various sources and then convert them into useful information to be applied in the decision making process. Third, the DSS should be actively applied to aberrance monitoring, goal achievement monitoring, status progress monitoring, cohort pursuit monitoring, and test monitoring as a continuous clinical work monitoring tool. There are several limitations of this study. First, the data collection period was only one month, from December 2000 to January 2001, and so it is insufficient to support the decision for the entire quality improvement of healthcare. Second, healthcare quality improvement must be accomplished through prospective method rather than retrospective method, but the data in this study retrospectively used the discharge summary database. In the future, other quality indicators should be analyzed to effectively support a hospital-wide CQI activity based on comprehensive database. In addition, the DSS should be well integrated with the hospital order communication system (OCS) to support concurrent review.