Characteristics of Adverse Drug Events Originating During the Hospital Stay, 2011 #164
Audrey J. Weiss, Ph.D. and Anne Elixhauser, Ph.D.
An adverse drug event (ADE) involves patient injury resulting from medication use. These events pose a serious patient safety concern. Some ADEs are known side effects that may arise even when a medication is taken correctly. Other ADEs, however, are the result of medication errors that may occur when a drug is prescribed or administered improperly. In 2007, the Institute of Medicine estimated that approximately 1.5 million preventable ADEs occur each year in the United States. 1
Many preventable ADEs occur during patient hospitalizations. Indeed, conservative estimates are that hospitalized patients experience 380,000 to 450,000 preventable ADEs each year.2 Overall, ADEs are the most common nonsurgical adverse events that occur in hospitals.3 ADEs are associated with an increased length of stay in the hospital (by 1.7 days), increased hospital cost (an additional $2,000), and increased risk of death (1.9 times higher than those not experiencing an ADE).4
This Statistical Brief presents hospital inpatient data on the four most common specifically identified ADEs from 32 States participating in the Healthcare Cost and Utilization Project (HCUP) that included a designation of whether ADE-related diagnoses were present on admission (POA) or originated during the hospital stay. In particular, patient and hospital characteristics of the most frequent specific causes of ADEs originating in the hospital are presented.
Differences in rates of 20 percent or greater are described in the text. All numbers noted in the text and included in the tables are actual values, not estimates, because the data include a census of discharges rather than a sample of discharges. In other words, we count the actual number of hospital stays with ADEs originating during the stay in the 32 States. Because we analyze numbers for the actual population rather than a sample, there is no need to estimate how well the sample represents an underlying population. As a result, there is no sampling error associated with the calculated values presented, and significance testing is not necessary.5