AMC Medication Error Program
The AMC consists of a 631-bed tertiary care teaching hospital and the Albany Medical College with its faculty physician practices. The hospital provides care for approximately 26,000 admissions per year. The hospital grants attending privileges to both private community physicians and faculty of the Albany Medical College. Most orders are written by AMC house staff using either standard "blank" physician order sheets, a number of "special" order sheets (eg, Antibiotic Order Sheet, Chemotherapy Order Sheet), and printed "physician-dependent protocols" and clinical pathways. AMC is planning for implementation but does not presently utilize physician order entry. The Department of Pharmacy operates with a central drug distribution process, using an HBOC Series 2000 pharmacy computer (HBOC, Orlando, Fla.) and a McKesson Automated Health RxRobot (McKesson, San Francisco, Calif.). Unit based cabinets are implemented throughout the institution for controlled substances and limited immediate floor stock medications.
Identification of Medication Prescribing Errors
All medication orders are reviewed by staff pharmacists and entered into the pharmacy computer system prior to dispensing. Patient-specific information routinely available to the reviewing pharmacist includes patient date of birth, attending service, admitting diagnosis, reported allergies to medications, body weight for pediatric patients, and a complete medication profile for the present hospitalization. Additional information regarding indication for use is provided for antimicrobials, which must be ordered on an "Antibiotic Order Sheet." A computerized reporting system enables the centralized staff pharmacists to access patient laboratory results from the computer terminals within the pharmacy. The pharmacy computer system has standard automatic dose range checking, allergy checking, and drug interaction checking capabilities. Pharmacists routinely use the described information sources as well as additional patient-specific information obtained as needed from the patient, medical record, nursing staff, physicians, and clinical pharmacists, to evaluate all medication orders for appropriateness.
Following the identification of medication orders potentially in error, the pharmacist contacts the prescriber or a cross-covering physician to obtain additional information and discuss the order. Potential prescribing problems are defined as medication orders for: the wrong patient, drug, dose, dosing frequency, route, or dosage form; inappropriate indication for use; inappropriate combinations of drugs; documented allergies to ordered medications; contraindicated therapy; which critical information is missing; and other miscellaneous problems. The medication order(s) in question are either confirmed as written, clarified, changed, or discontinued following the discussion between the pharmacist and physician. All identified problem orders that are jointly determined by the physician and pharmacist to require "correction" and subsequently changed are considered "confirmed problem orders." Each of these "confirmed" medication prescribing orders are further reviewed by a clinical pharmacist within 24 hours, and further information is obtained or action is taken to fully elucidate the problem and assure appropriate drug therapy is provided. Problem orders that are determined by this secondary review to be in error are defined as "confirmed medication prescribing errors."
For each confirmed prescribing error, the following data are collected:
patient medical record number
time of day order is written
pertinent patient-specific characteristics or other factors related to the error and relevant to the assessment of potential patient risk resulting from the error
The assessment of the potential significance of the adverse outcome of each error is based on available patient and pharmacologic information regarding risk for adverse events, potential of the error to be carried out, and, if carried out as ordered, potential of the error to result in adverse consequences (either an increased risk of adverse effects or an inadequate therapeutic response).The potential significance of errant orders is evaluated using our previously described rating scale.
This scale is based on the original scale developed by Folli and colleagues.
Confirmed medication prescribing errors are classified as either: (A) potentially fatal or severe, (B) potentially serious, or (C) potentially significant.
Consistency and agreement of assigning an error severity classification to specific errors has been previously evaluated.[8,9] Errors are then analyzed for "likely" contributing factors as described by Lesar, Briceland, and Stein.
Each error is "abstracted" and coded. The data are entered into a database (Excel, Microsoft, Redmond, Wash). A template for data collection and coding is shown in Appendix 1. This template facilitates the recording of errors both in "descriptive" (eg, drug class, overdose) and in (limited) "causation" (eg, failure to account for patient characteristic) categories. At AMC, the process is simplified by recording the error codes from the template in Appendix 1 onto a 2" x 3" pre-printed peel-off label that is placed on a copy of the errant order (Figure 1).
The copy of the order with attached abstract is filed for future reference and/or to be used as a teaching tool. A key factor to success learned over time is the use of a very limited number of individuals reviewing detected errors. An understanding of drug therapy, clinical and operational patient care processes, and quality improvement processes is necessary to maximize use of collected data to improve patient care.
A summary of all errors detected by the program are presented to the AMC Medication Use Quality Improvement Team along with summaries of errors detected by other programs (internal pharmacy, risk management occurrence report). The data and examples collected are used on an ongoing basis for staff education, monitoring of pharmacy error prevention activities, and use in analysis of need for system improvements to reduce error risks.
"Generalizability" of the Prescribing Error Program Process
The described process began in 1986, with 1987 as the first year of complete data. We believe a similar process is applicable to other hospitals. Relevant organizational process, information availability, communication, and pharmacist skill, among other factors, will define the nature of errors detected by a program. Given this context, it is expected that the data collected over time or within different organizations will differ. For example, the primarily centralized deployment of staff pharmacists at AMC, with obvious limitations to patient information and communication with other caregivers, defines limits on the types of prescribing errors detected, whereas such limits would be minimized for unit-based pharmacists working in a system using physician computer order entry. For these reasons, it is not expected that the identical "utility" of the specific described program is directly transferable to other healthcare organizations. The data collected by the described program provides information about certain types of errors, not all prescribing errors, and additional error-reporting efforts should address mistakes not readily monitored with the described process. Despite limitations, use of the process has been successfully implemented in a number of hospitals of varying size and has led to numerous process improvements within these organizations (L. Kleinsorge, personal communication, VHA Mountain States Performance Improvement, March 2000).
Medscape Pharmacists. 2000;1(2) © 2000 Medscape
Cite this: Medication Prescribing Error Reporting and Prevention Program: A 14-Year Experience - Medscape - Aug 14, 2000.