"approved uses" may lead physicians into the mistaken notion that they are somehow prohibited from medically sound prescribing merely because a manufacturer and the FDA, for whatever reason, have not concluded a transaction between themselves to include a use in the labeling.

A third-party provider may refuse to furnish or pay for a drug based on the absence of some recommendation in the manufacturer's FDA-approved literature. If that is the only reason for refusal, it is a deplorable administrative mistake. The FDA has no legal authority to impose such action.

If a malpractice suit should arise from real or alleged injury by a drug, the plaintiff's lawyer would probably attempt to strengthen his case if he could point to lack of recommenda tion in the manufacturer's literature for the use involved. Injustice might result if the defense failed to point out that the FDA does not regulate the practice of medicine. The labeling might be given some consideration in how well it reflects proper practice, but it should not be allowed to establish what is proper. Other medical literature or expert testimony can quite validly support correct use of a drug.

The fact that, based on adequate clinical trials, the FDA often approves additions to labeling recommendations for uses that have been employed for years gives mute testimony that the uses were proper all along. Some examples among many are worth repeating: propranolol for angina pectoris and hypertension, metronidazole for amebiasis, amantadine for parkinsonism, diazepam for status epilepticus, imipramine for childhood enuresis, colestyramine resin for hyperlipidemia, lidocaine for arrhythmias. Thus, physicians, not the FDA, still determine how drugs are used in the practice of medicine.

For anyone who might continue to consider package inserts as dogma, the preceding list involves some remarkable incongruities. Long before propranolol was labeled for angina pectoris, many cardiologists considered it a form of malpractice to perform a coronary bypass operation unless a patient had had a therapeutic trial with the drug.'

For years after injectable diazepam was recognized as the drug of choice for status epilepticus, its labeling bore warnings against its use in patients with epilepsy. After that absurdity was corrected regarding a largely pediatric disorder, the labeling long continued to advise that "The safety and efficacy of [the drug] in children under age 12 have not been established." (!

For about a decade after imipramine was used successfully for childhood enuresis, the labeling contained warnings against giving the drug to children. Before labeling for amantadine disclosed that the drug could be useful in treatment of early A, influenza (as contrasted with mere prophylaxis), it actually denied, contrary to fact, that such evidence existed. Almost amusing, for years after colestyramine resin was used successfully for hyperlipidemia, manufacturers listed this property of the drug as an "adverse reaction" or "side effect.""

Some other valid uses of marketed drugs may never reach the status of being an addition to the existing labeling. A manufacturer may never see any financial incentive for pursuing the approval to advertise a drug for an uncommon need. Another negative motive could be even more persuasive. As JAMA, Aug 24/31, 1984-Vol 252, No. 8

an example, certain urinary tract infections may be cured by a single dose of an appropriate drug. If a manufacturer's approved labeling recommended, say, a ten-day regimen, that manufacturer might not choose to supplicate the government for the privilege of reducing sales.

Fortunately, the myth about the authoritarian status of the package insert is disappearing. An honorable and welcomed statement by the FDA' has confirmed what I have said for two decades about "approved uses" of drugs: there is no such thing. The FDA statement even endorsed the same alternate and correct phrasing that I coined": An "unapproved use" should not connote a disapproved use, but merely an "unlabeled use." Uses in the labeling are merely that: "labeled uses."

The House of Delegates of the American Medical Association, at its 1982 Interim Meeting, adopted a report that quoted in full the FDA statement on this subject. The report called for the publisher of the Physician's Desk Reference (PDR) (Medical Economics Company, Oradell, NJ) to include this statement in future editions. Accordingly, in the 1983 and 1984 editions of the PDR, a summary of the FDA statement regarding the use of approved drugs for purposes not in the labeling appears in the FOREWORD.

Yet, habit is tenacious. Every edition of the AMA Drug Evaluations (American Medical Association, Chicago), beginning in 1971, has contained a discussion noting that the FDA has no authority to approve (or disapprove) how physician may use a marketed drug in his practice. Ironically, however, the fifth edition' organizes its discussion of at least one drug in terms of "approved" and "unapproved uses." (That will be avoided in future editions-John C. Ballin, PhD, oral communication, 1984.)

Perhaps the year 1984 will see one reversal of George Orwell's prediction." I have often read well-meaning statements that something was the "drug of choice" or "well established" or "fully recognized" for treatment of a disease-combined with the caveat that such use was "not approved." Such reasoning is Orwellian doublethink: the process of considering two opposite concepts at the same time and believing both. The doublethink under discussion resulted in part own naiveté 22 years ago, when I contributed some unfortunate language to a federal statute. That, however, is another matter: I apologized to the world as best I could in a previous publication."

from my

JOHN D. ARCHER, MD American Medical Association Chicago

1. Archer J: Instrument or impediment? The regulatory monograph in medical communications. JAMA 1972;220:1474-1477.

2. Archer JD: A guide into chaos: Resist it. JAMA 1974;227:1397-1398. 3. Lasagna L, Wardell WM: The FDA, politics, and the public. JAMA

1975;232:141-142.

4. Treatment of urinary tract infections. Med Lets Drugs Ther 1981;23:69-72. 5. Use of approved drugs for unlabeled indications. FDA Drug Bull

1982;12:4-5.

6. Archer J: Eternal vigilance—the price of liberty. JAMA 1972;222:1553

1555.

7. AMA Division of Drugs: AMA Drug Evaluations, ed 5. Chicago, American Medical Association, 1983, pp 1738-1739.

8. Orwell G: 1984. New York, Harcourt Brace Jovanovich Inc, 1946.

9. Archer JD: The confession of an erstwhile bureaucrat. JAMA 1978;239:1608.

Editorial 1055

Sumatriptan succinate, a 5-HT,, receptor agonist, constricts human cranial arteries. Two parallel-group trials for treatment of acute migraines were conducted in the United States. Adult patients were randomized and given either 6 mg of sumatriptan succinate subcutaneously (n = 734) or placebo (n=370). At 1 hour, sumatriptan was significantly more effective than placebo in reducing moderate or severe headache pain to mild or no pain (70% vs 22% ), in completely relieving headaches (49% vs 9%), and in improving clinical disability (76% vs 34%). Sumatriptan also reduced nausea and photophobia significantly better than placebo. Patients with residual migraines received another injection; those who had originally received sumatriptan received either a second active injection (n=187) or placebo (n=178), while those who had received placebo received a second placebo injection (n=335). Statistical evidence for benefit of second sumatriptan injection is absent. Adverse events associated with sumatriptan were tingling, dizziness, warm-hot sensations, and injection-site reactions. Sumatriptan is effective and well tolerated in patients with acute migraine.

EIGHT million Americans have migraines.' Sufferers often experience two or more incapacitating attacks every month, causing considerable disruption to both work and leisure time." BBlockers, calcium channel antagonists, and antidepressant drugs are sometimes used for prophylaxis. Current therapies for acute migraine include ergot derivatives, analgesics, and antiemetics. The effectiveness of existing rescue treatments for migraines is inconsistent, and the side effects of the treatments may be intolerable.

The pathogenesis of migraine is not well understood. Dilation of cranial blood vessels is thought to play an important role." Serotonin (5-HT) is a potent vasoconstrictor and is effective in treating migraines, but its unpleasant side effects prevent its routine use."

The brain has three classes of 5-HT receptors: 5-HT,, 5-HT,, and 5-HT,. The 5-HT, class is further subdivided; stimulation of the 5-HTD receptors causes

From the Shealy institute for Comprehensive Health Care, Springfield, Mo (Dr Cady); Department of Neurology, Health Science Center, University of Arizona, Tucson (Dr Wendt); the Migraine Treatment Clinic, Omaha, Neb (Dr Kirchner); Headache and Internal Medicine Research Center, Menninger, Topeka. Kan (Dr Sargent): Department of Neurology, University of California at San Diego Medical Center (Dr Rothrock); Neurology Section, Center for Clinical Research, Austin, Tex (Dr Skaggs); on behalf of the US Sumatriptan Research Group

Dr Cady owns stock in Glaxo Pharmaceuticals Inc. Reprint requests to Shealy Institute for Comprehensive Health Care, 1328 E Evergreen St, Springfield, MO 65803 (Dr Cady).

JAMA, June 5, 1991-Vol 265, No. 21

Otherwise healthy adults were eligible for enrollment in these studies between May and November 1989. Migraines were diagnosed using a 1-year history of classic migraine (with aura) or common migraine (without aura) and the criteria established by the International Headache Society." Written, informed consent was obtained from all patients. The protocol and consent form were approved by an institutional review board for each clinic.

History and results of physical examination, 12-lead electrocardiogram, and routine clinical laboratory tests were recorded at patient screening. Patients with hepatic or renal impairment, histo

ry of ischemic heart disease, Raynaud's disease or syndrome, uncontrolled hypertension, those who had previously been treated with sumatriptan, and those who were pregnant, were using inadequate contraception, or were lactating were excluded.

STUDY DESIGN-PROTOCOL

Patients presenting to the clinic with acute migraines gave information on headaches, concomitant symptoms, and clinical disability. Headaches were verbally rated by patients using a scale from 0 through 3, where 0 indicated no pain; 1, mild pain; 2, moderate pain; and 3, severe pain. Patients had to have moderate (grade 2) or severe (grade 3) headaches in order to be treated. Use of opioids or ergotamine within 24 hours or simple analgesics within 6 hours of study-drug administration disqualified the patient. Long-term prophylactic medications for migraine were not reasons for disqualification.

Qualified patients (n=1104) were randomly assigned to 6 mg plus 6 mg of sumatriptan succinate, 6 mg of sumatriptan succinate plus placebo, or placebo plus placebo, according to a randomization schedule generated prior to the trial and administered based on the chronological order that patients presented for treatment (Fig 1). Each dose was administered as a 0.5-mL subcutaneous injection over the deltoid muscle of the left or right arm. Absence of pain (grade 0) 1 hour after the first injection disqualified the patient from receiving the second injection. The second dose was given to evaluate whether remedication would provide additional efficacy if the first dose was not effective or if partial relief was achieved. Rescue therapy was administered at the discretion of the investigator if migraine persisted 1 hour after the second dose. Patients could use their usual rescue medications, such as aspirin, acetaminophen, meperidine hydrochloride, and promethazine hydrochloride, but excluding ergotamines.

Efficacy Measurements

Severity of headaches was rated by patients at 10, 20, 30, 40, 50, 60, 90, and

120 minutes after each dose. Pain relief was prospectively defined as reduction of moderate or severe headache pain (grade 2 or 3) to mild or no headache pain (grade 1 or 0). Mean pain scores and summed pain intensity differences scores are also reported." Patients who received rescue medication were defined as treatment failures.

Clinical disability and presence or absence of nausea, vomiting, and photophobia were assessed on the same schedule as headaches. Clinical disability was rated by patients using the following scale: 0 indicated the ability to work and function normally; 1, working ability mildly impaired; 2, working ability severely impaired; and 3, bed rest required. After discharge and for 48 hours after receiving treatment, patients kept a diary of headaches and of use of rescue medications.

Safety Assessments

Physical examinations and routine clinical laboratory tests were performed before treatment and prior to discharge. Vital signs (heart rate and blood pressure) were measured every 30 minutes after each dose until discharge from the clinic. Adverse events were recorded throughout the in-clinic treatment period and in the diary period.

Statistical Analysis

Nonparametric analyses of pain, clinical disability, nausea, vomiting, and photophobia were used. The last efficacy score prior to rescue medication was carried forward to subsequent time points. The P values were computed for each time point using Mantel-Haenszel and extended Mantel-Haenszel tests. All tests were two-sided, with P<.05 prospectively defined as statistically significant. There was no adjustment of P values to account for multiplicity of testing, reported P values were usually far smaller than those defining statistical significance.

2). At 1 hour, 515 (70%) of 734 patients who had received a single dose of sumatriptan reported mild pain (grade 1) or no pain (grade 0) compared with 81 (22%) of 370 patients who had received placebo (P<.001). Of these, 356 (49%) of 734 patients who had received sumatriptan were completely pain free (grade 0) compared with 35 (9%) of 370 patients who had received placebo (P<.001). Of the 356 patients who were pain free at 1 hour, 351 (99%) were still pain free at 2 hours.

Mean pain scores for patients receiv

ing sumatriptan at 10 minutes were significantly lower than those for patients receiving placebo and remained lower throughout the entire observation period (at 1 hour they were 0.91 vs 2.09; P<.001). The summed pain intensity differences scores at 1 hour were significantly higher in patients receiving sumatriptan compared with placebo (P<.001).

Migraine Relief: Second Dose

Of the 734 patients who initially received sumatriptan, 178 received a sec

Sumatriptan for Acute Migraine-Cady et al

ond dose of placebo, and 187 received a second dose of 6 mg of sumatriptan succinate. There was no statistical evidence for differences in pain variables or associated symptoms between one and two doses of sumatriptan (P>.15 for comparison of 6 mg plus 6 mg of sumatriptan succinate vs 6 mg of sumatriptan succinate plus placebo). A comparison at 2 hours of sumatriptan-treated patients (one or two active doses) with

JAMA, June 5, 1991-Vol 265, No. 21

placebo-treated patients, prior to any rescue medication, revealed that 81% of those receiving sumatriptan had migraine pain relief compared with 34% of those receiving placebo (P<.001). Associated Migraine Symptoms: First Dose

Most patients reported nausea and photophobia at presentation for treatment (Table 1). Within 20 minutes of

Data are combined for two identical trials. Number of patients receiving rescue medication was cumulative throughout the 48-hour period.

†Sumatriptan succinate was significantly better than placebo (P<.001).

sumatriptan administration, fewer patients reported nausea or light intolerance compared with patients receiving placebo. At 1 hour, 43% of sumatriptantreated patients had photophobia compared with 76% of placebo-treated patients (P<.001). Twenty-seven percent of patients receiving sumatriptan had nausea at 1 hour compared with 51% of patients receiving placebo (P<.001) (Fig 3). Beginning 20 minutes after treatment, more sumatriptan-treated patients reported normal or slightly impaired working ability than did placebotreated patients (P<.001).

Rescue Medication

During the clinic period, 20% of sumatriptan-treated patients received rescue medication compared with 59% of placebo-treated patients (P<.001) (Table 2). However, within 24 hours, 61% of sumatriptan-treated patients and 88% of placebo-treated patients took rescue medication (P<.001). Throughout the diary period, more placebo-treated patients took rescue medications (Table 2).

Duration of Complete Migraine
Pain Relief

Of the 734 sumatriptan-treated patients, 69% (511/734) were pain free at discharge compared with 22% (80/370) of placebo-treated patients (P<.001). Of the sumatriptan-treated patients, 250 (34%) of 734 remained completely pain free for 24 hours compared with 40 (11%) of 370 of placebo-treated patients. Note that these results are in spite of the greater use of rescue medication by placebo- than sumatriptan-treated patients.

Safety Results

Vital sign measurements and clinical laboratory test results were similar before and after treatment for all groups. Seven patients (six sumatriptan-treated and one placebo-treated) withdrew

from the study before receiving a second dose because of adverse events.

One hundred ninety-seven placebotreated patients (53%) experienced 462 adverse events (1.25 adverse events per patient) compared with 622 sumatriptan-treated patients (85%) who experienced 2275 adverse events (3.1 adverse events per patient). Significantly fewer placebo-treated patients had adverse events that were considered by the investigator to be drug related (156 [79%] of 197 placebo-treated patients compared with 600 [96%] of 622 sumatriptan-treated patients (P<.001]). Adverse events were described as severe in 34 (17%) of 197 placebo-treated patients and in 122 (20%) of 622 sumatriptan-treated patients.

Table 3 shows the specific adverse events occurring at an incidence of 1% or more in the sumatriptan-treated patients. Most adverse events began within minutes of the injection and lasted less than 1 hour. COMMENT

Sumatriptan rapidly reduces the severity and duration of acute migraine. Pain scores analyzed by four methods (mean pain score, percentage of patients achieving a 0 or 1 pain score, percentage of patients who were pain free, and summed pain intensity differences scores) consistently showed sumatriptan to be more effective than placebo in relieving moderate or severe migraines. Sumatriptan also produced significant improvement in clinical disability scores, nausea, and photophobia. A second injection at 1 hour was not associated with significantly more relief of migraine or associated symptoms. A secondary measure of sumatriptan's efficacy was the comparison of rescue medication use between treatment groups. Overall, 88% of placebo-treated patients received rescue medication within 24 hours compared with 61% of sumatriptan-treated patients. This indicates both the need to evaluate the duration of action of a single dose and to define a multiple-dose regimen. However, rescue medications were taken at the patient's discretion. It was not uncommon for patients to self-medicate with rescue medication in anticipation of the return of headache or associated symptoms, in spite of being pain free.

Significantly more adverse events occurred following treatment with sumatriptan than placebo. Concern whether

this unblinds the trial must be considered. Since patients were treated only once, which limited their exposure to study medication, and since adverse events and efficacy parameters were reported by the patient, it is unlikely that 2834

JAMA. June 5, 1991-Vol 265, No. 21