Pharmacist Participation in Medical Rounds Reduces Medication Errors

Kimberly K. Scarsi, Michael A. Fotis, Gary A. Noskin


Am J Health Syst Pharm. 2002;59(21) 

In This Article

Analysis and Resolution

From May 1 through May 31, 2000, one clinical pharmacist actively participated in daily rounds on 1 of 19 medical services in a 600-bed academic medical center. The members of the general medicine team include a medical student, an internal medicine intern, a resident, and an attending physician. A pharmacist had not previously participated in rounds with this team. The pharmacist performed daily activities during rounds, including investigating allergy information, monitoring trends in laboratory test values, and reviewing medication orders for appropriateness of dose and medication selection. Closely reviewed were drug indications; patient age, weight, and organ function; the medication administration record; and the provision of patient education. The pharmacist recorded all medication errors as they were discovered during the intervention period and filed appropriate medication-error reports using MedMARx 3.0 (United States Pharmacopeial Convention, Inc., Rockville, MD).

In the control group, another pharmacist participated in rounds each morning with the admitting team from the previous evening and had direct contact with each patient admitted to the service on the first inpatient day. During admitting rounds, medication orders and laboratory test values were reviewed and patients receiving high-risk medications were identified. Patients were monitored for the rest of their stay only if they were prescribed a designated drug per pharmacy policy. All other medication orders received routine profile review as part of the medication-ordering process. Routine profile review consisted of screening for drug-drug interactions and evaluating the appropriateness of each dose on the basis of patient-specific values. Although the existing process allowed the pharmacist to see more patients per month, it did not provide the continuity of care nor the patient-medical team interaction that daily rounds provided. Pharmacist time spent on clinical activities was similar in each group (approximately two hours per day).

The study group included all patients admitted to the participating general medicine service during May 2000 (n = 35). These patients were matched after hospital discharge to concurrently hospitalized patients by random selection using an internally designed computer-generated report prepared by the pharmacy department's information systems manager, who was blinded to the study purpose. The report identified 35 patients for inclusion in the control group, to ensure that both groups were evenly matched for age, sex, length of stay, number of medication orders written, and nursing unit. All control group patients received the same care as the intervention group. The groups were compared by age, length of stay, and number of inpatient medication orders. Every patient initially selected for the control group was included in the study.

Three pharmacists with previous experience using chart review to assess for medication errors and adverse drug reactions were selected to review all patients' medical charts. All of the pharmacists were senior members of the clinical staff with at least 10 years experience in this area. One author provided orientation and training for the reviewers and also served as a reviewer, evaluating the consistency of the assessments completed by the other two reviewers. After patient selection was verified, the blinded reviewers conducted a retrospective evaluation of patient medical records. A structured chart review began with examining the interfaces of care. Each admission, transfer, and discharge note was examined for medications currently taken, medication allergies, and discharge prescriptions. The medication orders and the medication administration record were searched for matching entries. Any unexplained variance was considered an error. The remaining progress notes were similarly searched. Medication selection, dosing, and monitoring were compared with recommendations published in the clinical staff manual Optimizing Medication Use at NMH.[6] Again, unexplained variances were interpreted as errors. An error was noted if a consultant note included an indication for additional or different medications, if no order was written for the specified medication, or if follow-up consultant notes confirmed the need for treatment. The same process was used for each chart reviewed. Reviewers classified medication errors as prescribing, administration, pharmacy, or discharge errors by using definitions derived from the 1999 IOM report (appendix).[1] Reviewers also documented the duration of the error before it was corrected.

Data were compiled and statistically evaluated in a spreadsheet using Microsoft Excel 97 (Microsoft Corporation, Redmond, WA). Statistical evaluation was performed with median and mean variables, confidence intervals (CIs), and odds ratios with chi-square tests to determine statistical significance. A p value of less than 0.05 was considered significant.

The intervention and control groups were evenly matched for age, sex, average length of stay, and number of medication orders written ( Table 1 ). To further verify the internal validity between the intervention and control groups, the concurrent data reported by the clinical pharmacist participating in rounds with the medical service was compared with the retrospective data collected by the blinded reviewers. Overall, concurrent data collection identified 47 medication errors in the intervention group, compared with 46 errors documented by the reviewers in the intervention group.

When a pharmacist participated in daily medical rounds, medication errors were reduced by 51% (p < 0.05). The number of patients without a medication error during their hospitalization increased (22.9% in the control group versus 40.0% in the intervention group) (p < 0.05). The duration of time that an error continued after it occurred was also significantly less in the intervention group. An error persisted less than 1 day (mean, 0.73; 95% CI, 0.48-0.98) and with less than one dose of medication (median, 0) in the intervention group, compared with 2.4 days (95% CI, 2.3-2.9) and two doses of medication in the control group ( Table 2 ). Classification of errors was similar between the groups and consistent with the distribution of errors documented in the 1999 IOM report.[1] Prescribing errors occurred most frequently, followed by administration errors and pharmacy errors ( Table 3 ).

The reduction in the number of medication errors that occur at the time of prescribing is likely due to the pharmacist gaining a more thorough clinical picture of the patient related to medication prescribing after participating in rounds. After reviewing a medication order, the pharmacist could evaluate medication use with a more extensive knowledge of the patient's medical history, drug contraindications, organ function, and the reason a drug was prescribed.

This pharmacist could also serve in an interdisciplinary educational role for physicians, nurses, and other health care providers. Barriers to communication are greatly reduced when the pharmacist participates in daily rounds. It is easier for informal education to occur, such as answering questions about medication dose, adverse reactions, and drug interactions. Through this interaction, medication errors can be prevented.

The pharmacist had the opportunity on daily rounds to review the medication administration records for errors and educate the nursing staff on the appropriate administration of medications. Administration errors were primarily corrected by altering inappropriate administration time of food-dependent medications, preventing drug-drug interactions of coadministered medications, and encouraging timely administration of time-dependent medications, such as antimicrobials.

The reduction in pharmacy errors was primarily the result of clinical interventions on the basis of the application of the principles of therapeutic drug monitoring. For example, through interacting with the medical team and patients, the pharmacist could assess and document laboratory test values, whereas the control patients often did not receive routine follow-up of these values.

Medication errors occurring at the time of discharge were common in a pilot study at our hospital. The clinical pharmacist was able to reduce these by thoroughly evaluating the discharge summary written before patient discharge, ensuring that discharge medication orders were appropriate and complete. Routine pharmacy follow-up for patients in the control group did not allow for this, as discharge prescriptions are filled at community pharmacies.