Objectives: In order to understand the pattern of utilization of migra
ine prophylactic drugs by US physicians, we reviewed the scientific ri
gor of published trials of anti-migraine medications, assessed their c
ost, and tested the correlation, if any, between utilization, scientif
ic rigor and cost. Materials and methods: scientific rigor of publishe
d reports. We identified all placebo-controlled, randomized, double-bl
ind trials of migraine prophylactic agents through Medline search, maj
or Headache textbooks and proceedings of major scientific meetings whe
re headache-related topics are discussed. We excluded trials that did
not include placebo treatment during the active phase of the study. Th
e trials were reviewed and rated for scientific rigor using a 5-point
scale (scientific score [ss]; 1=low, 5=good), blinded to the physician
s' utilization data and cost of the drugs. Studies that did not show b
enefit of the active drug over placebo were scored -1 to -5, thus allo
wing for the reverse logic of negative studies. US physicians' utiliza
tion. Neurologists and primary care physicians (PCP) completed phone-m
ail-phone questionnaires which inquired about first and second choices
of migraine prophylaxis. These choices were averaged to obtain a weig
hted average percent usage of each drug. Cost. The average wholesale p
rice (AWP) of each drug was obtained from data published by Adelman an
d Von Seggern, and from the Amerisource (7/9/96) catalog. Statistical
analysis: Spearman's correlation coefficient was used to assess the re
lationship between the average ss, physician use, and cost of each dru
g. Results: Propranolol (ss=1.44), amitriptyline (ss=2.33) and verapam
il (ss=1.00) were the three preferred migraine prophylactic drugs by b
oth neurologists and PCPs. Approximately 10% of neurologists said that
divalproex (ss=3.75) would be their first or second choice. The selec
tive serotonin reuptake blockers were favored by 13.21% of PCPs. All o
ther prophylactic drugs were felt to be first or second line of treatm
ent by less than 10% of either neurologists or PCPs. Except for one st
udy (ss=1) that showed that propranolol reduced the migraine frequency
by 76% over placebo, trials of the three most preferred medications f
ailed to demonstrate that the active drug is >50% better than placebo,
i.e. the difference in headache frequency when on placebo vs active d
rug is >50%. Of the drugs available in the United States, flurbiprofen
and metoprolol achieved the best ss (5.00 and 4.33, respectively) but
their efficacy over placebo (23% and 14-33%, respectively) and cost (
$67.2 and $65.6) were unfavorable. Neurologists and PCPs chose migrain
e prophylaxis on the basis of scientific merit (r=0.644, p=0.018; r=0.
576, p=0.05, respectively) but not cost (r=-0.254, p=0.45; r=-0.255, p
=0.455). Conclusion: The three most commonly chosen migraine prophylac
tic agents have not been shown irrefutably to prevent migraine. Furthe
rmore, their benefit, if any, does not exceed 50% over placebo. The we
ll-conducted recent trials that demonstrated the efficacy of divalproe
x in migraine prevention are steps in the right direction of finding t
he ''ideal migraine preventative agent''. Until that drug is discovere
d, it is difficult to argue that one migraine prophylactic medication
is superior to another and accordingly should be used as a first line
of treatment.