Medication error in the care of HIV/AIDS patients: electronic surveillance, confirmation, and adverse events.

BACKGROUND: Medication error occurring during the care of HIV-infected patients may lead to treatment failure, drug toxicity, or even death.
OBJECTIVE: The objective of this study was to ascertain and confirm 5 categories of medication error in the care of HIV-infected patients. RESEARCH
DESIGN: This study was a retrospective study to describe the occurrence of preventable medication error and to determine if adverse events were associated with confirmed errors. A roster of medications for each category of potential errors was created. Computerized pharmacy records were scanned for all dispensing of these medications. Potential errors were confirmed by medical records abstraction. For the incorrect dosing, coadministration of contraindicated medications, and antiretroviral monotherapy error categories, random samples were chart reviewed for confirmation. For the remaining 2 error categories, all potential errors were chart reviewed. The positive predictive value (PPV) of potential errors, the incidence of confirmed error among all new prescription orders filled and the patient characteristics predicting likelihood of error confirmation were estimated for each error category.
SUBJECTS: The study sample involved 5473 HIV-infected patients of the Kaiser Permanente Northern California (KPNC) health plan.
RESULTS: Among the 5 error categories, PPVs ranged from a high of 80% for coadministration of contraindicated medications to <1% for antiretroviral monotherapy. Incidence of confirmed errors was 9.80 errors per 1000 new prescriptions dispensed for incorrect dosing, 9.51 errors per 1000 for contraindicated medications, and <1.00 for all other categories. Adverse events associated with confirmed errors were observed only in the contraindicated medications error category. The likelihood of a contraindicated medications error was significantly increased among patients >or=50 years of age and decreased among black patients.
CONCLUSIONS: Use of electronic pharmacy records to ascertain true medication errors appears most reliable when conducting surveillance for contraindicated medications errors and less reliable for other error categories. Lack of confirmation is likely the result of patients' lack of adherence to drug regimens or providers' intentional deviation from accepted prescribing guidelines. Only confirmed contraindicated medications errors appear to be linked to adverse events.