Performance of different data sources in identifying adverse drug events in hospitalized patients.

PURPOSE

The incidence of adverse drug events (ADE) is an important parameter in determining the quality of medical care. We identified the probability that a specific data source would identify ADEs in patients on the oncology ward, that could be assigned to one substance.

METHODS

We captured all medical adverse events (AE) from five different data sources. Each AE was determined to be drug-related according to the WHO criteria and classified according to the severity, category, and causality of the ADE.

RESULTS

The study recorded 129 patients with 252 hospitalizations over a 5-month period. A total of 3,341 medical events were captured and resulted in 1,121 ADEs. In 122 patients, at least one ADE (95%) was observed. Only 39 hospitalizations were believed not to have an ADE (15%). No ADE was captured by all data sources. The patient record captured 550, the nursing record 569, the laboratory tests 387, the questionnaire 63, and the event monitoring during grand rounds 141 ADEs. Only the nursing record and the laboratory tests had a significantly different probability of observing indicative ADEs.

CONCLUSION

For all AEs reported in the data sources, physicians and nurses were the best source for ADEs. Data sources differed in identifying indicative ADEs and were influenced by specific patient parameters.