Introduction

Adverse drug interactions pose a significant threat to hospitalized patients. Most patients receiving potentially interacting drugs do not experience adverse drug events, however serious adverse events are known to occur.₁,₂,₃ Evidence from epidemiologic studies suggest that drug interactions contribute to a small but significant number of adverse events in hospitalized patients.₄,₅,₆,₇

Athough the incidence of adverse drug events attributable to drug interactions in most studies is low, the majority of patients in these studies were not critically ill. In addition, the number of drugs that these patients received may have been less than the number of drugs generally received by intensive care unit patients. Critically ill patients usually receive multiple drugs and complex drug combinations. The probability of an adverse drug event increases as the number of drugs received increases and this association has been attributed to drug interactions.₈ In addition, the pharmacologic effects of many drugs are altered in critically ill patients.₉ For these reasons, critically ill patients may be more susceptible to adverse drug events related to drug interactions. However, the incidence of drug interactions and associated adverse drug events in critically ill patients is not known. Knowledge of the epidemiology of drug interactions may help critical care practitioners reduce the risk for adverse drug events in this patient population. The purpose of this study was to estimate the incidence of major drug interactions (MDI) and associated adverse drug events (ADE) in critically ill surgical intensive care unit patients.

Materials And Methods

The University of Texas M.D. Anderson Cancer Center (MDACC) is a comprehensive cancer center that is devoted exclusively to cancer patient care, research, education, and prevention. The Center has 518 beds of which twenty-two are dedicated to surgical intensive care. The surgical intensive care unit (SICU) serves the intensive care needs of a wide array of adult and pediatric patients undergoing general, neurological, thoracic, head and neck, and plastic and reconstructive surgery.

All patients admitted to the SICU for a period of at least four days from July 1, 1993 through June 30, 1994 were identified from the MDACC Patient Care Information System (PCIS). The PCIS is a health information database containing comprehensive information for all MDACC inpatients including demographic, diagnostic, admission, discharge, and transfer data. Medication use records were extracted from the MDACC Division of Pharmacy database and matched with each patient record in a database program (MicroSoft Access, Redmond, Washington). The study was limited to drugs that were ordered upon and during SICU admission. Queries were designed (RCL Systems, Houston, TX) to identify potential MDIs. We defined a drug interaction as those interactions which are well-documented in the literature and which have the potential of being harmful to the patient.₁₀ Pharmaceutical incompatibilities and drug-enteral feeding interactions were excluded. The specific major drug interactions and the associated adverse events that were searched for are depicted in the appendix. Upon identication of a MDI, the patient medical record was carefully reviewed by an experienced clinical pharmacist for evidence of adverse drug events.

Results

Two-hundred and seventy five patients were admitted to the SICU for 4 or more days during the study period. Some of these patients were admitted more than once for a total of 290 SICU stays. Of these patients, 20 patients were found to have been exposed to 24 major drug interactions. The population at risk for ADEs was therefore comprised of 20 patients. This corresponds to a one-year incidence of 7.3% in our critically ill SICU population. The median length of SICU stay for the patients exposed to MDIs was 18 (range 4-151) days and these patients received a mean of 29.1 (SD 11.9, median of 29, range 12-55) different drugs per SICU stay. The table depicts the types of MDIs identified. No adverse drug events attributable to the MDIs could be identified from the careful chart review.

Figure 1

Table 1: Major drug interactions identified.

The presence of an interaction was evident in two patients receiving phenytoin for seizure prophylaxis who were subsequently placed on rifampin. The total phenytoin plasma concentrations were noted to be low, following the addition of rifampin, suggesting that rifampin had increased phenytoin clearance. Although phenytoin concentrations (< 2.5 g/mL, 4.9 g/mL) were low in each case, no overt clinical seizure activity was documented.

Discussion

Major drug interactions with the potential for adverse events were common in our population. Despite an overall one-year incidence of major drug interactions of 7.3%, no associated adverse drug events were identifed after careful review of the medical records. The observed incidence of drug interactions in our critically ill patient population was higher than the incidence reported in other hospitalized patients.₄,₅,₆,₇

Puckett and Visconti prospectively screened the drug therapy of 2,422 patients admitted to large community hospital over a two-month period time and found 113 patients who were receiving potentially interacting drug combinations.₄ Seven patients developed clinical evidence of interaction representing 6.2% of the 113 patients at risk and 0.3% of the population screened. The Boston Collaborative Drug Surveillance Program examined 83,200 drug exposures in 9,900 hospitalized patients and identified 3,600 adverse drug reactions.₅ A total of 234 (6.5%) of the adverse drug reactions were attributed to drug interactions. The Harvard Medical Practice Study screened 30,195 medical records from 51 hospitals in New York state and identifed 1,133 adverse events caused by medical management that resulted in measurable disability₆. The leading cause of adverse events in this study was drug complications which was projected to account for 19.4% of all adverse events and 14.1% of serious disabilities in the population. More than half (58%) of the adverse events were judged to be due to one or more management errors. Drug treatment errors accounted for 8.9% of the management errors and failure to recognize possible antagonistic or complementary drug interactions, in turn, accounted for 8% of the total number of drug treatment errors. The ADE Prevention Study Group studied all admissions to eleven medical and surgical units in two tertiary care hospitals over the course of 6 months₇. During the study period, 334 errors were detected that led to 264 preventable and potential adverse drug events. Drug-drug interactions were involved in 10 (3%) of these errors. The differences in incidence of MDI in our patients suggest that critically ill patients are more likely to have a drug interactions. With the exception of the Boston Collaborative Drug Surveillance Program, which defined a drug interaction in terms of a pharmacodynamic phenomenon, drug interactions in all other studies were not well-defined.

Potential interactions involving pharmacodynamic mechanisms (interactions resulting when drugs with additive or antagonistic effects are combined) were more common in our patients than those involving pharmacokinetic (interactions involving the distribution, metabolism, or excretion) mechanisms (table). The most common potential MDI was the combination of neuromuscular blockade and aminoglycoside antibiotics. Although aminoglycoside antibiotics have been reported to potentiate the neuromuscular blockade induced by neuromuscular blocking agents, we did not find any evidence of potentiation or prolongation of blockade.₁₁,₁₂ However, potentiated neuromuscular blockade due to an interaction may have been difficult to document since in most instances, the dosage of neuromuscular blocker was titrated to effect and monitoring of blockade with peripheral nerve stimulation was not consistently documented.

Two of the patients identifed were receiving phenytoin and rifampin concurrently and had phenytoin serum concentrations that were low. This observation is consistent with the influence of rifampin which is known to induce the hepatic metabolism of many drugs including phenytoin.₁₃ Although no overt clinical seizure activity was documented, seizure activity may be difficult to recognize in some critically ill patients.₁₄

There are several limitations that may have influenced our ability to identify ADEs. The retrospective design of the study, which relied on chart review to identify adverse events, may have prevented the identification of adverse drug events even if they actually occurred. The ability to identify an adverse drug event from the medical record assumes that those caring for the patient recognize that a particular adverse event is related to the drug therapy and it assumes that the event, once recognized, is documented.

The clinical outcome of most drug interactions is highly situational and depends on several factors including the sequence of administration, duration of therapy, dose of each drug, even the influence of other drugs.₁₀ These factors not only influence whether an interaction will actually lead to a clinically measurable outcome but also the time that an interaction will manifest. Retrospective detection of the result of an interaction is therefore limited because the observation period cannot be controlled.

The institution of a satellite pharmacy and full time clinical pharmacy services in the SICU precluded a valid prospective design since most pharmacists help identify drug-related problems and avert adverse drug events, including ADEs related to drug interactions in hospitalized patients.₁₅,₁₆ The time period from July 1, 1993 through June 30, 1994 was therefore selected because this period preceded the implementation of a satellite pharmacy and full time clinical pharmacy services devoted to exclusively to our SICU patients. However, even before the institution of SICU-dedicated pharmacy services, drug orders were reviewed prior to dispensing by the MDACC pharmacy staff and prescribers were routinely notified of potentially serious drug interactions (J. Lajeunesse, Pharm.D., Personal communication). This notification may have led to closer patient monitoring or alterations in drug therapy (eg. dosage adjustments, additional laboratory tests, etc.) thereby averting ADEs.

Finally, our decision to only include patients who were in the SICU for 4 or more days may have limited our sample size and therefore our ability to detect ADEs. Many patients admitted to our SICU require a few days of special monitoring that cannot be provided in our non-intensive care surgical wards. Although these patients require intensive care, they are not necessarily critically ill. We therefore limited the study population to patients who were in the SICU for at least four days in an effort to improve the generalize our findings to more critically ill patients. However, this was probably a minor limitation since nearly 5,000 SICU patient-days were represented.

Conclusions

Major drug interactions with the potential for adverse events were common in our population of surgical intensive care patients. No associated adverse drug events were identified but the retrospective design of the study probably limited our ability to detect them.

Drug interactions pose a difficult dilemma for critical care practitioners. On one hand, there is ample evidence that drug interactions may lead to ADEs and therefore unfavorable outcomes. It is equally clear, however, that most patients receiving drugs with the potential to interact do not manifest adverse consequences. Therefore, clinicians must be alert for those particular situations in which the patient is truly at risk, without overreacting every time a patient is receives a potentially interacting combination of drugs. Unfortunately, it is not possible for clinicians to remember all of the details of every drug interaction. Nevertheless, knowledge of the interactive properties of drugs coupled with an appreciation of the incidence of significant drug interactions should enable one to predict many adverse drug interactions before they occur. Clinicians should be alert for drug interactions and the potential for associated adverse drug events. Prospective studies are needed to determine the true incidence of ADEs that result from major drug interactions.

Correspondence to

David E. Oeser, Pharm.D. Clinical Pharmacist and Medication Safety Coordinator Presbyterian/St. Luke's Medical Center 1719 East 19h Avenue Denver, Colorado 80218 David.oeser@healthonecares.com

Appendices

APPENDIX 1: Categories of major drug interactions

Figure 2

Figure 3

APPENDIX 2: Categories of major drug interactions

Figure 4

Figure 5