model for introducing safer care detailed in the preceding section are highlighted in bold print.

CASE STUDY 1
Detecting and Preventing Adverse Drug Events

In 1988, researchers working within a 520-bed, tertiary teaching hospital’s departments of clinical epidemiology, pharmacy, and medical informatics (the improvement team) questioned whether the hospital’s existing nurse incidence reporting system adequately detected ADEs. They compared three different ADE detection systems in a parallel trial: (1) traditional nurse incidence reporting; (2) enhanced reporting; and (3) prospective expert case review, driven by a data-based clinical trigger system. Enhanced reporting allowed nurses to simply flag a patient through the computerized charting system, avoiding the time and effort of filling out an incident report. A representative from the improvement team reviewed the patient’s chart, determined whether an ADE had occurred, and completed the documentation. The clinical trigger system involved a series of treatment markers for ADEs, such as the use of antidote drugs (e.g., naloxone to counteract an opiate), abnormal values on specific laboratory tests (e.g., a twofold increase in blood creatinine), or other clinical indicators (e.g., reports of rash or itching in nursing notes). A positive clinical trigger led to prospective review by a clinical pharmacist within 24 hours, using explicit criteria. A clinical pharmacist from the improvement team also used explicit criteria to review all cases detected by traditional nurse incidence reporting to confirm whether an actual ADE had occurred. During the review, all ADEs were staged as mild, moderate, or severe, and their causes and patient outcomes were documented.

Over 18 months (May 1, 1989, through October 31, 1990), covering 36,653 hospitalizations, standard nurse incidence reporting, enhanced reporting, and prospective expert review driven by data-based clinical triggers found, respectively, 9, 92, and 731 confirmed ADEs (Classen et al., 1991). While enhanced reporting increased ADE detection rates by an order of magnitude, prospective expert review driven by data-based clinical triggers increased detection 80-fold.

Three members of the improvement team, expert in ADEs, reviewed more than 200 charts to identify ADE causes. Early analyses that classified ADEs by hospital location (e.g., emergency department versus operating room versus nursing unit) and by drug type (i.e., narcotics versus antibiotics) were not as useful as those that classified failures by process mechanism (epidemiological analyses and hypotheses for change generation). The team organized its findings as a cause-and-effect diagram (see Figure 5-2), then tallied actual ADEs to generate a Pareto chart of prioritized causes



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement