OBJECTIVES: Relatively little is known about rates of outpatient adverse drug events (ADEs), and most health systems do not routinely identify them. We developed a computerized ADE measurement process and used it to detect ADEs from electronic health records and then categorized them according to type, preventability, and severity. METHODS: The rules used represent combinations of variables including coded medication names, laboratory results, diagnoses, and specific items such as symptoms from free text clinician notes, all obtained from electronic health records. Rules targeted various diagnostic and laboratory abnormalities potentially caused by a broad range of outpatient medications commonly used in primary care. The rules were run on 4 months of data on primary care patients seen in the outpatient setting in 2 large health systems; possible incidents were identified by chart review and validated as ADEs by clinician reviewers, then rated by severity and preventability. RESULTS: The rates of ADEs were 75 ADEs/1000 person-years and 198/1000 person-years at the 2 sites, respectively. The overall rate was 138 ADEs/1000 person-years across the 2 sites. Eleven percent of ADEs were preventable, with a rate of 15 preventable ADEs/1000 person-years across sites. Approximately one-fourth of ADEs were serious or life threatening at both sites. The highest yield rules for identifying preventable ADEs included rules based on drug classes and symptoms, and drug-laboratory rules. CONCLUSIONS: Adverse drug events occurred frequently in routine outpatient care, and many were serious and preventable. Computerized monitoring represents an efficacious approach for identifying ambulatory ADEs, although it needs additional refinement. In addition, site-specific variations need further exploration.
OBJECTIVES: Relatively little is known about rates of outpatient adverse drug events (ADEs), and most health systems do not routinely identify them. We developed a computerized ADE measurement process and used it to detect ADEs from electronic health records and then categorized them according to type, preventability, and severity. METHODS: The rules used represent combinations of variables including coded medication names, laboratory results, diagnoses, and specific items such as symptoms from free text clinician notes, all obtained from electronic health records. Rules targeted various diagnostic and laboratory abnormalities potentially caused by a broad range of outpatient medications commonly used in primary care. The rules were run on 4 months of data on primary care patients seen in the outpatient setting in 2 large health systems; possible incidents were identified by chart review and validated as ADEs by clinician reviewers, then rated by severity and preventability. RESULTS: The rates of ADEs were 75 ADEs/1000 person-years and 198/1000 person-years at the 2 sites, respectively. The overall rate was 138 ADEs/1000 person-years across the 2 sites. Eleven percent of ADEs were preventable, with a rate of 15 preventable ADEs/1000 person-years across sites. Approximately one-fourth of ADEs were serious or life threatening at both sites. The highest yield rules for identifying preventable ADEs included rules based on drug classes and symptoms, and drug-laboratory rules. CONCLUSIONS: Adverse drug events occurred frequently in routine outpatient care, and many were serious and preventable. Computerized monitoring represents an efficacious approach for identifying ambulatory ADEs, although it needs additional refinement. In addition, site-specific variations need further exploration.
Authors: Richard W Grant; Jeffrey M Ashburner; Michael C Jernigan; Jaime Chang; Leila H Borowsky; Yuchiao Chang; Steven J Atlas Journal: J Gen Intern Med Date: 2015-01-06 Impact factor: 5.128
Authors: Sue Jordan; Timothy Banner; Marie Gabe-Walters; Jane M Mikhail; Gerwyn Panes; Jeff Round; Sherrill Snelgrove; Mel Storey; David Hughes Journal: PLoS One Date: 2019-09-11 Impact factor: 3.240