Introduction

Adverse drug reactions and drug interactions present a growing concern in the health care setting. A number of studies have found that the incidence of drug interactions ranges from 3%₁ to 30%₂. Drug-drug interactions can lead to a variety of adverse events, and it has been suggested that preventable adverse events are the eighth leading cause of death in the United States₃.

Adverse drug related events can take a number of forms, with drug-drug interactions one of the most common. A large number of drugs are available to the physician; the Food and Drug Administration has approved over 19,000 drugs, and an average of 340 new drugs are added to the formulary every year₄. Many outpatients are on complex regimens of medication, and up to 20% report significant adverse events as a result of their medications₅. A large retrospective study of 30,000 medical records in New York in 1984 found adverse events in 4% of hospitalizations₆. Drug related complications caused 20% of these adverse events₇. Adverse drug related events are largely preventable₈, and it is critical to address barriers to their recognition and prevention.

The task of recognizing drug interactions is somewhat more difficult in the emergency department, where every patient is new and unknown to the staff. The incidence of potential drug interactions in this patient population is high. A number of studies have estimated the incidence of potential drug-drug interactions in 20-30% of patients, and the incidence of clinically relevant interactions at 4-10%₉,₁₀,₁₁,₁₂,₁₃,₁₄. However, most studies have used a variety of screening criteria, and evaluate between 300-400 patients. The incidence of drug-drug interactions or adverse events in an unselected emergency department population is unclear.

We are designing a computerized triage program which will allow patients to review their existing list of medications or choose their current medications from a list of 7000 medication names. In the background, it will check for interactions and then present a list of potential interactions to the emergency physician. This system should provide a critical resource to help improve the recognition of drug-drug interactions by emergency physicians.

In order to focus our efforts on the most common medications and the most common drug-drug interactions to screen for, we attempted to evaluate the frequency and type of drug-drug interactions in our ED patients by screening records over a two week time period. This method resulted in a sample of 1355 patients, an order of magnitude greater than that in many of the studies cited above. All medications were cross-compared using a computerized database of known drug interactions.

Methods

Design

Setting

Data Analysis

Results

The most common medications reported at triage are listed in Table 1. Of note, 35 patients (2.5%) listed a medication that was illegible on the triage sheet and not clarified elsewhere on the chart. The most common class was oral contraceptives. The second most common was albuterol inhalers, due to the prevalence of asthma in the general population of approximately 7% ₁₅. Table 2 shows the frequency of polypharmacy in this population. Up to 60% of patients were taking at least one medication, with 6.3% taking more than 7 medications. These frequencies are similar to those reported by other groups. ₁₄, ₁₆

Figure 1

Table 1: The 20 most common medications reported.

Figure 2

Table 2: Frequency of polypharmacy. The number and proportion of patients reporting multiple medications are listed.

Table 3 lists drug allergies reported at triage and Table 4 the frequency of drug allergies, with 24% of patients reporting at least one medication allergy. Surprisingly, four patients reported they were currently taking drugs to which they claimed to be allergic. Two of these were aspirin, one was sulfamethoxazole, and one was penicillin. There were no documented instances of an adverse event related to a patient taking a medication to which they reported an allergy.

Figure 3

Table 3: Allergies reported at triage. The number of patients with each medication allergy are listed.

Figure 4

Table 4: Frequency of reported drug allergies. The number and proportion of patients listing multiple allergies are listed.

Table 5 lists the most common medications given or prescribed in the ED and Table 6 the frequency of ED prescriptions. Most patients (72%) were given medication, and 2% received at least five new medications. In combination with the frequency of polypharmacy on presentation, this suggests a high risk of drug-drug interactions.

Figure 5

Table 5: The most common ED prescriptions. The number of patients given or prescribed each medication in the emergency department is listed.

Figure 6

Table 6: Frequency of additional prescriptions from the emergency department. The number and proportion of patients receiving prescriptions in the ED are listed.

Table 7 lists the most common of the 1435 potential interactions of patients' current medications generated by the Medical Letter Drug Interactions Program. Some of these drug combinations are logical, such as a bronchodilator and a corticosteroid, or two antihypertensives, with potential interactions that are usually tolerated.

Figure 7

Table 7: Most common potential interactions of current medications at triage.

Figure 8

Table 8 lists the most common of the 450 potential interactions of medications prescribed in the ED. A full cross tabulation of frequencies is in Table 9, with current medications in the left lower half, and ED prescriptions in the right upper half.

Figure 9

Table 8: Most common potential interactions of ED prescriptions

Figure 10

Table 9: Cross tabulation of frequencies of potential interactions. Current medications are listed along the lower left, and ED prescriptions are listed along the upper right. The number of patients with each interaction is listed in each box. The graph is color coded as follows: Blue, 1-15 patients; Green, 16-30 patients; Red, >30 patients.

Table 10 summarizes the 14 notes in the 1355 ED charts suggesting a drug related adverse event. The most notable are increased bleeding from multiple anticoagulants; bradycardia from multiple rate control agents; and Dilantin toxicity from a Bactrim/Dilantin interaction. The number of adverse drug related events documented by emergency department caregivers was 14 out of 1355 visits, or approximately 1% of all ED visits. Five of these adverse drug related events (0.3% of all visits) were due to drug-drug interactions.

Table 10: Adverse drug-related events documented in ED charts

Allergic / Idiosyncratic

Side Effect

Drug Interaction

Discussion

Drug related adverse events have been identified as a major source of morbidity and mortality in the United States, and drug-drug interactions are a significant source of these events. In this study, we documented the frequency of various reported medications and allergies in an unselected emergency department population. In addition, we calculated the frequency with which potential drug-drug interactions would be highlighted by a computer-based system that might screen patients at triage.

The most common medications taken by our emergency department patients were oral contraceptives, albuterol, and aspirin. The frequency of polypharmacy is high in our ED population, with six percent of patients taking more than 7 medications. These frequencies are similar to those reported by other groups ₁₄, ₁₆. These patients are likely to be sicker, to have more medical problems, and to be at higher risk of drug-related interactions than the average patient₁₀, ₁₃, ₁₄.

The most frequent classes of medications implicated in potential drug-drug interactions were NSAIDs, beta-blockers, steroids, ACE inhibitors, and anticoagulants. This compares similarly to a previous report which also implicated antibiotics, diuretics, hypoglycemics, and calcium-channel blockers₁₂. In contrast, when only those interactions that were thought to be clinically relevant by emergency physicians are analyzed, antihypertensives (most commonly beta blockers) and carbamazepine are the most commonly cited.₁₀ Our study and others recognize NSAIDs as the most common source of potential drug-drug interactions.

Although many medications have been implicated in numerous studies as a source of potential interactions, it has been difficult to evaluate the frequency of clinically relevant adverse drug related events. Gandhi et al recently reported that in a series of ambulatory care patients, up to 25% suffered an adverse drug event, of which 13% were serious₅. Our data suggest that only 1% of presentations to the emergency department can be primarily attributed to an adverse drug related event; however, these data can be reconciled by hypothesizing that only a small percent of patients with an adverse drug related event present to the emergency department rather than to their doctor's office. In the study by Gandhi et al, the most commonly implicated drugs causing an adverse event were selective serotonin reuptake inhibitors (SSRIs), antihypertensives, and NSAIDs. We similarly found that NSAIDs and antihypertensives were commonly implicated, but found a lower rate with SSRIs. This may again reflect the differences in populations. When controlling for frequency of use, Gandhi et al found that corticosteroids were the most common adverse event associated medication, while we found it to be one of the most common medications associated with potential interactions.

The frequency of clinically significant adverse drug related events in this study was approximately 1%, as compared with the 10% documented by Hohl et al in 2001₁₂. One explanation for this difference is that Hohl et al prescreened for ED visits by patients 65 years of age and older, a group that has been shown to demonstrate a much higher risk of adverse drug related events. In addition, their criteria was quite broad and included patient noncompliance. Our study only documented adverse drug related events that were recognized as such by the treating clinician.

The frequency of potential drug-drug interactions, however, was one hundred fold higher than that of documented adverse drug related events. Drug-drug interactions are known to comprise a small subset (3%-6%) of all adverse drug reactions₇, ₁₇, ₁₈. Many of these drug interactions are known and expected, and include medications that are frequently used in concert (such as corticosteroids with albuterol). However, many are of potentially high relevance, and we noted four cases here in which a foreseeable drug-drug interaction led to a clinically significant adverse event requiring a presentation to the emergency department.

Potential drug-drug interactions are present not only at triage, but are often compounded by medications given in the ED. Only 28% of out patients received no medications. The most common drugs given and prescribed in the ED, NSAIDs and benzodiazepines, are among the most common sources of potential interactions.

Emergency physicians have been shown to demonstrate a poor sensitivity and specificity for drug-drug interactions compared to a computer system₄, and a number of authors have suggested that computer-aided order entry and prescription writing can reduce the number of medication errors₂, ₃, ₁₈,₁₉,₂₀,₂₁,₂₂,₂₃,₂₄,₂₅,₂₆.

In our hospital, use of computerized order entry for inpatients, in which all medications are entered and cross-checked for interactions, has been shown to decrease medication errors and adverse drug related events₁₈ and to generate cost savings as well.₂₄ Our data suggest there may be a great deal of added value in translating a similar system to the emergency department. Our department has implemented a computer based order entry system,₂₇ but this can only identify drug-drug interactions in those medications prescribed in the ED. In order to detect and screen for all medication interactions that may contribute to patient care, it may be appropriate that the triage nurse not only document all medications but also enter them along with allergies into the ED order entry system.

While a number of potential interactions have been identified, not all are clinically relevant. It is important for expert systems to alert physicians only to the most important interactions, as they otherwise run the risk of showing too many alerts for the patients on multiple medications. We have documented the most common potential drug-drug interactions in an unselected emergency department population, which should help focus information systems development on highlighting the most common and most under-recognized drug-drug interactions.

Limitations

Conclusions

The incidence of potential drug-drug interactions is high in an unselected emergency department population. The risk is higher among current medications recorded at triage than it is for new medications prescribed in the ED. Actual interactions are relatively few. Physicians should be vigilant for potential drug-drug interactions, especially among the most high risk patients taking multiple medications, but existing software that automatically alerts physicians to all potential interactions may generate excessive noise to signal.

Correspondence to

Thomas O. Stair, MD, Department of Emergency Medicine, Brigham & Women's Hospital, 75 Francis Street, Boston, MA 02155. tstair@partners.org