A computerized method for identifying incidents associated with adverse drug events in outpatients
Introduction
Adverse drug events (ADEs) are common, costly and responsible for significant morbidity and mortality in hospitalized patients. According to one estimate, the costs of problems with medications may be higher than the total cost of cardiovascular or diabetes care [1]. Both the Food and Drug Administration and the Joint Commission on Accreditation of Healthcare Organizations emphasize the need for reporting ADEs as important markers of the quality of medical care [2], [3]. In addition, the American Society for Health-Systems Pharmacists (ASHP) recommends that all health care systems develop ongoing ADE reporting programs [4].
A recent meta-analysis suggested that ADEs account for 106 000 deaths annually in the United States [5]. Even though this figure may represent an overestimate, the true figure is likely to be much higher than had been generally recognized [6]. Most studies of ADEs have been in hospitalized patients [7], [8], [9], [10], [11], and 2–5% of all hospital admissions each year are due to ADEs [12], [13]. In a study of inpatients, Bates et al. found that nearly 28% of ADEs that occurred were preventable [14].
However, identification of ADEs on a routine basis has been relatively ineffective. Spontaneous reporting is the most widely used technique but it identifies only 5% of events [15]. In inpatients, manual chart review is more effective but is too costly to be routinely practical [16]. Another alternative is computerized detection, which has been effective at several sites for identifying a variety of events [6], [16], [17], [18], [19]. We reasoned that, given the increase in availability of electronic medical records (EMR) and the development of natural language processing (NLP), it should be possible to detect ADEs in outpatients as well. Although electronic ADE detection has usually been limited to ‘rule-based’ searches, the growing availability of EMR and NLP creates an opportunity to use these technical advances in this important area of quality measurement. This paper describes the methodology used to identify potential ADEs from the information contained in an outpatient EMR.
Section snippets
Methods
Brigham and Women's Hospital is a 667-bed institution delivering primary, secondary and tertiary care in Boston, MA. Approximately 170 primary care physicians work at a diverse array of clinical sites including hospital-based practices, community-based practices and neighborhood health centers; most now use an electronic ambulatory record. Since 1993, all information pertaining to patients and their visits has been collected and stored in the Brigham Integrated Computer System (BICS). This
ICD-9 classification rules
The providers entered ICD-9 codes for each patient visit after the patient encounter. These ICD-9 codes were selected by the providers from an abbreviated list made available at the point of service. ICD-9 codes that have been previously found to be associated with the presence of possible adverse drug events were selected from this list to search the database (see Table 1) [20]. The ICD-9 list was compared using SQL to the codified visit records, and when they matched an incident was
Discussion
For several decades, the primary use of computers in hospitals in the United States was to facilitate reimbursement for care rendered and to automate results for high-volume, time-critical tests such as clinical laboratory procedures [22]. However, as increasing quantities of data became available electronically, investigators have begun to use this information to assess quality of care by hospitals, outpatient practices and providers [23], [24]. Although primarily utilized to evaluate
Conclusion
In this report, we describe a program that combines four search methods including data mining of the electronic medical record to detect ADEs in outpatient settings. Such approaches may make routine detection of quality problems in outpatients more practical and less costly. Further refinements to this methodology can improve the overall accuracy of detection.
Acknowledgements
This research was supported in part by a grant from Micromedex, Inc. In addition, Micromedex provided technical assistance in this project. We would like to thank Josh Lee, M.D., Jeff Rothschild, M.D., Alice Chang, M.D., Karen Steward, Lisa Zygel, Julie Backus and Jeanie Cornish for their assistance in identifying adverse events, and Julie Fiskio for her assistance in extracting computerized medical data.
References (31)
- et al.
The incident reporting system does not detect adverse drug events: a problem for quality improvement
Jt. Comm. J. Qual. Improv.
(1995) - et al.
Automated detection of hereditary syndromes using data mining
Comput. Biomed. Res.
(1997) - et al.
Drug related morbidity and mortality: a cost illness model
Arch. Intern. Med.
(1995) - et al.
National adverse drug reaction surveillance: 1985
J. Am. Med. Assoc.
(1985) - et al.
Food and drug administration monitoring of adverse drug machine
Am. J. Hosp. Pharm.
(1986) - American Society of Health-Systems Pharmacists, ASHP guidelines on adverse drug reaction monitoring and reporting, Am....
- et al.
Incidence of adverse drug reactions in hospitalized patients: a meta-analysis of prospective studies
J. Am. Med. Assoc.
(1998) Drugs and adverse drug reactions: how worried should we be?
J. Am. Med. Assoc.
(1998)- et al.
Adverse drug events in hospitalized patients: excess length of stay, extra costs, and attributable mortality
J. Am. Med. Assoc.
(1997) - et al.
The costs of adverse drug events in hospitalized patients
J. Am. Med. Assoc.
(1997)