Impact of Simulation-Based Learning on Medication Error Rates in Critically Ill Patients
Intensive Care Medicine
Authors
Daniel G. Ford, PharmD
Critical Care Clinical Pharmacist, NorthBay Medical Center
At the time this study was conducted, Dr. Ford was the Critical Care Pharmacy Specialty Resident, University of
Pittsburgh Medical Center
NorthBay Medical Center, Department of Pharmacy 1200 B. Gale Wilson Blvd, Fairfield, CA 94533
Amy L. Seybert, PharmD
Associate Professor, Pharmacy and Therapeutics, University of Pittsburgh School of Pharmacy
Pharmaceutical Care Coordinator for Critical Care, University of Pittsburgh Medical Center
Director, Critical Care and Cardiology Pharmacy Specialty Residencies, University of Pittsburgh Medical Center
Associate Director of Pharmacy Programs, Peter M. Winter Institute for Simulation, Education and Research (WISER), University of Pittsburgh
200 Lothrop Street Suite 302, Pittsburgh PA 15213
Pamela L. Smithburger, PharmD
Medical ICU, Clinical Specialist, University of Pittsburgh Medical Center
Assistant Professor, Pharmacy and Therapeutics, University of Pittsburgh School of Pharmacy
200 Lothrop Street Suite 302, Pittsburgh PA 15213
Lawrence R. Kobulinsky
Simulation Specialist, Peter M. Winter Institute for Simulation, Education and Research (WISER), University of Pittsburgh
230 McKee Place Suite 300, 3rd Floor Pittsburgh, PA 15213
Joseph T. Samosky, Ph.D.
Assistant Professor, Anesthesiology, University of Pittsburgh School of Medicine
Director of Research and Development for Healthcare Simulation, Peter M. Winter Institute for Simulation, Education and Research (WISER), University of Pittsburgh
230 McKee Place Suite 300, 3rd Floor Pittsburgh, PA 15213
Sandra L. Kane-Gill, PharmD, MSc, FCCM
Associate Professor, Pharmacy and Therapeutics, University of Pittsburgh School of Pharmacy
Critical Care Specialist, Center for Pharmacoinformatics and Outcomes Research
Critical Care Patient Safety Officer, University of Pittsburgh Medical Center, Department of Pharmacy
Pharmacy; 918 Salk Hall 3501 Terrace Street, Pittsburgh, PA 15261
Address correspondence and reprint requests to Dr. Seybert: University of Pittsburgh Medical Center, 302 Scaife Hall, 200 Lothrop Street, Pittsburgh, PA 15213. Phone: (412) 647-6170, Fax: (412) 647-5847,
Discussion of Error Types
Certain types of administration errors did not clearly fit into the standardized categories, thus required further explanation. Prescribing errors included circumstances where an inappropriate amount of drug was administered to the patient; however the nurse had accurately followed the orders from the prescriber. Errors classified as “mislabeled” included either the wrong medication displayed on the infusion pump or doses administered at one time but documented as being given at another time. The latter type of mislabeling error was not classified as “wrong time” as the medication was still given at an appropriate time considering the properties of the drug and physician’s order, but were considered to be mislabeled on the medication administration record, thus constituting an error in the administration process.
Table 1 displays the types of administration errors; the most common types of errors were omitted doses, incorrect preparation of the medication and use of the wrong administration technique. The most common medications associated with administration errors were (incidence in parentheses): insulin (11), famotidine (11), subcutaneous heparin (9), levothyroxine (9), fentanyl (7), pantoprazole (7), aspirin (5), propofol (4) and furosemide (4).
Although this study was not designed to detect or analyze changes in individual types of administration errors, we did observe changes in frequencies of the different types. There was a reduction in errors due to incorrect administration techniques during the initial post-intervention observation in both ICUs. However, the final post-intervention observation for the MICU showed a subsequent increase in wrong administration techniques. The increase was not due to a single study participant. This could be explained by the fact that nurses in the MICU did not actively practice administration techniques as the CCU nurses did during the simulation. Fewer preparation errors were seen in the CCU after the SBL in-service, but not in the lecture-trained group. We saw an increase in mislabeling errors in the MICU during the initial post-intervention observation, mostly due to incorrect documentation of the time the dose was given. Since this particular error was not seen in the baseline period, it was not addressed in the in-services. The reason for the increase in labeling/documentation errors is not clear, however it may be explained by changes in workflow or staffing. Although omission and prescribing errors are typically beyond the control of the nurses, they were addressed in the in-services by simply encouraging communication with physicians and pharmacists. There was a decrease in both omission and prescribing errors during initial post-intervention observation, with several instances of nurses being observed calling pharmacy for missing doses and asking physicians to clarify orders. Unfortunately, this decrease was not noted in final post-intervention observation as omission errors increased in the MICU. Examples of these final post-intervention observation omission errors included not administering medications because of an ongoing procedure and not applying prescribed ointments because the nurse was unable to turn the patient without assistance.
As mislabeling and omission errors exhibited an increase in the post-intervention observations but were a small portion of the in-services, a post-hoc analysis was conducted of the data with these categories removed. This statistical analysis was consistent with the results of the complete data set showing a significant decrease in administration errors after SBL, no detectable difference initially after the lecture and a significant increase in administration errors in the final observation of the lecture-trained group. The robustness of our findings is demonstrated through this additional analysis.
Table 1. Types of Errors*
Baseline / Initial post-intervention observation / Final post-intervention observation / Baseline / Initial post-intervention observation / Final post-intervention observation
Drug Prepared
Incorrectly / 9 (18.8%) / 3 (50%) / 0 / 1 (3%) / 11 (28.2%) / 2 (3.7%)
Expired Product / 2 (4.2%) / 0 / 0 / 1 (3%) / 0 / 2(3.7%)
Improper Dose or
Quantity / 5 (10.4%) / 0 / 1 (16.7%) / 7 (21.2%) / 4 (10.3%) / 1 (1.9%)
Mislabeling / 2 (4.2%) / 1 (16.7%) / 0 / 1 (3%) / 16 (41%) / 1 (1.9%)
Omission / 15 (31.3%) / 1 (16.7%) / 0 / 11 (33.3%) / 4 (10.3%) / 17 (31.5%)
Prescribing / 2 (4.2%) / 0 / 0 / 4 (12.1%) / 0 / 0
Unauthorized or
Wrong Drug / 0 / 0 / 1 (16.7%) / 1 (3%) / 0 / 0
Wrong Administration
Technique / 12 (25%) / 0 / 3 (50%) / 7 (21.2%) / 4(10.3%) / 25 (46.3%)
Wrong dosage form / 0 / 0 / 0 / 0 / 0 / 1 (1.9%)
Wrong Route / 1 (2.1%) / 0 / 0 / 0 / 0 / 0
Wrong Time / 0 / 1 (16.7%) / 1 (16.7%) / 0 / 0 / 5 (9.3%)
Total / 48 / 6 / 6 / 33 / 39 / 54
CCU = unit receiving simulation based learning
MICU = unit receiving traditional didactic lectures
*Deteriorated Product, Extra Dose and Wrong Patient categories were omitted as they were not observed during the study.
Table 2. Subjective Assessments in Percent Response (number)
CCU
(Simulation-Based Learning0 / This in-service was beneficial / 0 / 0 / 0 / 56% (5) / 44% (4)
I enjoyed participating in this in-service / 0 / 0 / 11% (1) / 44% (4) / 44% (4)
I prefer this teaching method / 0 / 0 / 0 / 44% (4) / 56% (5)
I learned something from this in-service / 0 / 0 / 0 / 44% (4) / 56% (5)
I would like more in-services like this / 0 / 0 / 11% (1) / 56% (5) / 33% (3)
MICU
(Didactic Lecture) / This in-service was beneficial / 0 / 0 / 8% (1) / 58% (7) / 33% (4)
I enjoyed participating in this in-service / 0 / 0 / 8% (1) / 58% (7) / 33% (4)
I prefer this teaching method / 0 / 0 / 17% (2) / 42% (5) / 42% (5)
I learned something from this in-service / 0 / 0 / 8% (1) / 42% (5) / 50% (6)
I would like more in-services like this / 0 / 0 / 8% (1) / 58% (7) / 33% (4)
Behavioral learning vs traditional learning
Simulation based learning offers the opportunity for behavioral change in a safe, yet realistic environment when compared to traditional, didactic based learning. This study shows that the use of high-fidelity training with patient simulation can have a positive impact on patient care by decreasing the medication administration error rate in critically ill patients. By altering the behavior and practice of the nurses administering medications with simulation based education, the potential to impact patient care is a novel area of research.
5