In the 1980s, Brown and colleagues determined that giving 1 mg of epinephrine to a 70-kg patient was no more than homeopathic and should produce little if any effect on blood flow to the heart and brain during cardiopulmonary resuscitation [4-6]. Their dose-response experiments confirmed that the original dose of 0.1 mg/kg used by Redding and Pearson achieved the best hemodynamic effects and should used in human resuscitation.

Persuaded by these animal data, many groups and clinicians became early adopters of high-dose epinephrine, and editorials, case series, and retrospective reports lent support to the higher dose. Editorialists lamented the "tyranny of the AHA," which declined to issue a special recommendation for the use of high-dose epinephrine. Conscientious researchers, despite their intuitive sense that the higher dose was best, embarked on prospective clinical trials. However, the results presented as prepublication data at the 1991-1992 AHA Guidelines Conferences were disappointing. These trials involved more than 2400 adult patients in nine cities [7-10]. Although increased return of pulses was associated with higher epinephrine doses, no statistically significant improvements were found in survival to hospital discharge. Subset analyses failed to confirm a benefit for several patient groups that many thought would respond to the higher dose of epinephrine. However, no trials detected substantial harm to patients who received the higher doses of epinephrine.

By 1992, the AHA had adopted graded resuscitation recommendations, reflecting an increasing reliance on evidence-based guidelines [11]. An intravenous epinephrine dose of 1 mg continues to be the first and most important pharmacologic intervention in adult cardiac arrest. Higher doses of epinephrine (5 mg or about 0.1 mg/kg) recommended on the basis of the available evidence remain a class IIb recommendation for adults, to be used only after the 1-mg dose has failed. (A class IIb recommendation means that the intervention is acceptable and possibly helpful, but the evidence provides neither support nor discouragement.)

How have clinicians responded to such waffling recommendations? The adult guideline accurately reflects the evidence-no established benefit, no confirmed harm-but offers little clinical direction. Informal surveys at advanced cardiac life support courses across the United States reveal that most emergency specialists and resuscitation protocols follow an escalating approach of 1 mg-3 mg-5 mg given 3 minutes apart. However, recent data from the study by Behringer and coworkers [12] reported in this issue should send out warning signals about the routine use of higher, escalating doses of epinephrine. Use of large amounts of epinephrine for patients who do not respond to the first dose may harm the few patients who do survive. Using the cerebral performance category, Behringer and coworkers related functional neurologic outcomes to the total epinephrine dose administered. An average epinephrine dose of 4 mg (range, 0 to 50 mg) per resuscitated patient was used. Fifty-eight percent of patients had an unfavorable neurologic outcome, defined as a cerebral performance category score of 3 or less. The total amount of epinephrine used was significantly higher in resuscitated patients with unfavorable neurologic outcomes than in those with favorable outcomes (4 mg compared with 1 mg; P < 0.001). Clearly, the duration of resuscitation is a confounding variable because clinicians routinely administer epinephrine every 3 to 5 minutes during resuscitation attempts. However, even when duration of resuscitation and other potential confounders were controlled for, the negative effect of total epinephrine dose remained significant.

As a retrospective cohort study, this study merits an intermediate level of evidence ranking of 4 [13]. The researchers thoughtfully designed the study to address an important topic, carried it out with strict adherence to the internationally accepted Utstein Style guidelines [14], and analyzed the data in a sophisticated manner. Researchers have not reported a direct toxic effect of higher amounts of epinephrine in previous clinical trials of high-dose epinephrine, but none of those trials closely examined the possibility of adverse outcomes. Furthermore, several nonclinical trials reported multiple negative effects of the total cumulative epinephrine dose [15, 16].

As Behringer and coworkers caution, further prospective studies are needed to determine whether high-dose epinephrine may be a neurotoxin. However, this unfavorable effect will be seen only in patients whose heartbeats are not restored. Many researchers have observed that the only way to recover the brain is to first restore the heartbeat. This attitude has produced the low threshold among clinicians for using large amounts of epinephrine in the hope of restoring a heartbeat.

Researchers have already performed many clinical trials of high-dose epinephrine. Since the 1992 AHA Guidelines, randomized clinical trials on high-dose epinephrine from France [17], Australia [18], and Sweden [19] have been published. Each of these studies found the same negative outcomes. Eventually, the persistently negative results will convince researchers that additional trials are unnecessary. At that point, the opportunity to test the hypothesis generated by Behringer and coworkers will end. Perhaps the answer will come from a reanalysis of the existing prospective studies of high-dose epinephrine. Analysts can search for evidence that the few patients who survived did worse over time if they were randomly allocated to receive higher doses of epinephrine.

The AHA will publish the next set of resuscitation guidelines in the year 2000. The AHA and the International Liaison Committee on Resuscitation are committed to an evidence-based approach to all guideline additions or revisions. For the Guidelines 2000 discussions and debates, new data are needed on the use of high-dose epinephrine. Perhaps the casual attitude of emergency care providers toward administration of large amounts of epinephrine during cardiac arrest should end.

Ms. Hazinski: Vanderbilt University, 2100 Pierce Avenue, Nashville, TN 37212.