Paper
Sunday, November 4, 2007
179
Variables that Predict Medication Errors
Debra M. Picone, PhD1, Marita Titler, PhD, RN, FAAN1, Joanne McCloskey Dochterman, PhD, RN, FAAN2, Leah L. Shever, RN3, Taikyoung Kim, MS4, Paul Abramowitz, PharmD5, and Mary F. Kanak, PhD6. (1) Department of Nursing Services and Patient Care, University of Iowa Hospitals and Clinics, Iowa City, IA, USA, (2) University of Iowa College of Nursing, Iowa City, IA, USA, (3) Nursing Interventions and Outcomes Effectiveness Grant, Univeristy of Iowa College of Nursing, Iowa City, IA, USA, (4) Nursing Interventions and Outcomes Effectiveness Grant, University of Iowa College of Nursing, Iowa City, IA, USA, (5) Department of Pharmaceutical Care, University of Iowa Hospitals and Clinics, Iowa City, IA, USA, (6) Mercy Medical Center, Cedar Rapids, IA, USA
Purpose: The purpose of this study was to describe the type and number of medication errors experienced by a group of hospitalized elderly patients and to determine what factors predict medication errors.
Conceptual Framework: A nursing effectiveness model (Titler, Dochterman, & Reed, 2004) was used to guide this study.
Sample: The sample consisted of 10,187 hospitalizations of patients aged 60 or greater admitted to a Midwest tertiary academic medical center between July 1, 1998 and June 31, 2002 who were at risk for falling.
Methods: Data were abstracted from 9 electronic data repositories. An explanatory model for predicting medication errors, considering patient characteristics, clinical patient conditions, treatments, and characteristics of the patient care unit was built using generalized estimated equations (GEE). The dependent variable, the occurrence of a medication error, was measured using the internal, voluntary incident reporting system.
Results: There were 871 medication errors reported and 92% may have been preventable. Most of the errors were due to transcription (36%) or omission (22%). Variables significant in the explanatory model included unique number of medications
(poly pharmacy), patient gender and race, RN staffing changes, medical and nursing interventions, and specific pharmacological agents.
Conclusion: The results of this study lend support to the literature that suggests that specific variables are associated with medication errors. It emphasizes the need for appropriate RN staffing. It also emphasizes the importance of a medication error incident reporting system to avail this type of research. These results suggest that automated systems including E-Mar, CPOE and decision support may reduce errors. Interdisciplinary approaches to reduce polypharamcy may also be important. Special attention to procedures for high risk medication administration is warranted. Limitations of the study are reliance on the incident reporting system, and limited ability to generalize findings as this study occurred at one academic medical center.