Annals of Internal Medicine
COMMENTS AND RESPONSES
Report of Specific Cardiovascular Outcomes of the
ADVANTAGE Trial
TO THE EDITOR: In their letter to the editor, Braunstein and Polis
(1) address the omission of important cardiovascular safety results
from a 2003 report on the ADVANTAGE (Assessment of Differ-
ences between Vioxx and Naproxen To Ascertain Gastrointestinal
Tolerability and Effectiveness) trial (2). The 2003 article, authored
by Lisse and colleagues (including Polis), reported no statistically
significant differences for any cardiovascular end points in compari-
sons of Vioxx (rofecoxib, Merck & Co., Inc., Whitehouse Station,
New Jersey) and naproxen in a 12-week clinical trial. In fact, almost
3 years earlier, Merck gave the U.S. Food and Drug Administration
(FDA) a data table that showed a statistically significant relative risk
of 7.0 for cardiac events among patients receiving Vioxx compared
with those receiving naproxen in the ADVANTAGE trial. The
Merck data table is included in a 3-page letter (3) that was written in
2001 in response to the FDA’s request for additional information
regarding the ADVANTAGE trial; it shows 7 serious adverse cardiac
events in patients receiving Vioxx (6 events adjudicated as myocar-
dial infarction and 1 event adjudicated as a “sudden/unknown”
death) and 1 adjudicated myocardial infarction in a patient receiving
naproxen. In addition, another event that was “reported as hyperten-
sive heart disease and death by Merck” was reclassified by the FDA in
November 2001 as a “sudden/unknown” death, pushing the cardiac
events for Vioxx to 8 (4).
Merck’s letter and the FDA’s reclassification are emphatic evi-
dence of Merck’s knowledge that the ADVANTAGE trial showed
that Vioxx use caused an important and statistically significant excess
risk for cardiac events, namely myocardial infarction and “sudden/
unknown” death. Merck’s omission of this important risk informa-
tion from its 2003 article in Annals presents a serious public health
hazard because it misled the medical community. Braunstein and
Polis said that Merck’s predefined procedures kept hypertensive heart
disease and death from being externally adjudicated, which would
explain their exclusion from the Antiplatelet Trialists’ Collaboration
“cardiac event” group. Because Merck defined a list of external ad-
judication events that excluded such relevant events as hypertensive
heart disease, it was fortunate that the FDA adjudicated the case.
Merck still omitted this information when they published the trial.
Underlying this already egregious omission of safety data is
more deeply disturbing evidence of medicine that has run amok.
Internal Merck documents reveal that the ADVANTAGE trial did
not have a medical purpose. Instead, it was a “seeding trial,” meant
to seed the medical community with Merck’s new drug before it was
approved for the market. Thus, the deaths of trial patients receiving
Vioxx were deaths in an unnecessary trial. The real participants in
the trial were the physician “investigators” (nonparticipants served as
controls) who were chosen to participate by Merck sales representa-
tives (5). Merck intended to promote the drug to influential doctors
and their patients and then analyze the prescribing information of
these physician-investigators for marketing purposes. A Merck public
relations supervisor instructed employees to avoid revealing the true
purpose of the trial: “I eliminated the reference to seeding. It may be
a seeding study, but let’s not call it that in our internal documents” (6).
Downloaded From: http://annals.org/ by a Brown University User on 01/04/2018
Letters
Merck’s conduct in designing, conducting, analyzing, and pub-
lishing the ADVANTAGE trial is disturbing at best. The ethical
ramifications of drug-related deaths in a “seeding trial” deserve a
more thorough examination than is possible here. The practice of
selectively reporting drug safety data is evidence for the need for
complete and public disclosure—perhaps on the Internet— of clini-
cal trial data for new drugs. If anything, ADVANTAGE teaches us
that we cannot rely on drug companies to honestly report all of the
important data.
David S. Egilman, MD, MPH
Amos H. Presler, BA
Brown University
Attleboro, MA 02703
Potential Financial Conflicts of Interest: Expert testimony: Dr. Egilman
has served as an expert witness in Vioxx litigation. Mr. Presler is em-
ployed by Dr. Egilman.
References
1. Braunstein N, Polis A. Report of specific cardiovascular outcomes of the ADVAN-
TAGE trial [Letter]. Ann Intern Med. 2005;143:158-9. [PMID: 15968005]
2. Lisse JR, Perlman M, Johansson G, Shoemaker JR, Schechtman J, Skalky CS, et al.
Gastrointestinal tolerability and effectiveness of rofecoxib versus naproxen in the treat-
ment of osteoarthritis: a randomized, controlled trial. Ann Intern Med. 2003;139:539-
46. [PMID: 14530224]
3. Silverman RE. Merck Research Laboratories letter to FDA, Center for Drug Eval-
uation and Research: NDA 21-042/S-007: Vioxx (rofecoxib tablets). Response to FDA
request for information; 2001. Presented in evidence in Cona v. Merck, 2006.
4. Villalba ML. Medical officer review, NDA 21-042 and NDA 21-052 (rofecoxib
tablets and rofecoxib oral solution). Re: Complete response to approvable letter for
21-042/S 007 and 21-052/S 004. U.S. Food and Drug Administration; 2001.
5. Smith B. Communication to J. Webb re: Vioxx seeding study selection. 8 January
1999. Bates no. MRK-AFB0001598. Presented in evidence in Humeston v. Merck,
2005.
6. Weiner JD. E-mail to K. Lindemann, C. Fanelle, and R. Higbee. Subject: ADVAN-
TAGE ideas. 19 March 1999. Bates nos. MRK-ADI0024344–MRKADI0024346.
Presented in evidence in Humeston v. Merck, 2005.
Socioeconomic Status and Mortality
TO THE EDITOR: Alter and colleagues (1) address an important
question regarding mediating factors that potentially account for the
contribution of socioeconomic status to health care disparities. How-
ever, because of the potential social and political implications of
these results, careful consideration should be given to several key
issues that limit the authors’ interpretations. First, it has been previ-
ously suggested that socioeconomic status is a multidimensional con-
struct. Although operational definitions are numerous, most incor-
porate aspects of educational attainment, occupation, and social
class. Use of self-reported income as a single measure to represent
this construct therefore has the potential to markedly reduce strength
of the intended “signal” and underestimate its association with the
outcome of interest (2). Second, the authors applied exclusion crite-
ria that potentially attenuate an association between socioeconomic
status and mortality and may introduce bias. The authors observe,
for example, that patients with lower income had a significantly
higher prevalence of cardiac risk factors and were less likely to receive
specialty care. By eliminating from analysis those patients who died
© 2006 American College of Physicians 781
 Letters
within 24 hours of admission or those with “very severe illness,” the
authors have essentially removed the patients with the greatest “ex-
posure” to the potential health effects of socioeconomic status. We
find the work by Alter and colleagues potentially informative, but
their failure to attend to the multidimensional nature of socioeco-
nomic status leads us to conclude that they may have underrepre-
sented the importance of this construct’s effects on health.
Mehdi H. Shishehbor, DO, MPH
David Litaker, MD, PhD
Cleveland Clinic
Cleveland, OH 44195
Potential Financial Conflicts of Interest: None disclosed.
References
1. Alter DA, Chong A, Austin PC, Mustard C, Iron K, Williams JI, et al. Socioeco-
nomic status and mortality after acute myocardial infarction. Ann Intern Med. 2006;
144:82-93. [PMID: 16418407]
2. Kaplan GA, Keil JE. Socioeconomic factors and cardiovascular disease: a review of
the literature. Circulation. 1993;88:1973-98. [PMID: 8403348]
IN RESPONSE: We appreciate the comments of Drs. Shishehbor and
Litaker. We agree that socioeconomic status is a multidimensional
construct of which income serves as only 1 of many social measures.
Although such limitations were acknowledged in our paper, our
study did adjust for individual education, employment status, eth-
nicity, and social support. Adjustment for such variables partially
accounted for some of the heterogeneous features of socioeconomic
status. Furthermore, the inclusion of more elaborative social mea-
sures, although intriguing, would not have mitigated the importance
of exploring the causal pathway factors that mediate income–mortal-
ity associations—associations that have been consistently observed in
the literature and require explanation (1).
We acknowledge that the exclusion of very high-risk patients
(that is, those receiving mechanical ventilation or those who died
before enrollment) may have introduced bias and attenuated the
association between socioeconomic status and mortality. Unfortu-
nately, the exclusion was unavoidable because income was ascer-
tained by using self-administered surveys. Enrollment into the
SESAMI (Socio-Economic and Acute Myocardial Infarction) study
required patient consent, which also probably contributed to selec-
tion bias (2). Consequently, the magnitude of association between
income and mortality after acute myocardial infarction might have
been less than otherwise expected had we been able to examine a
more representative “real-world” population.
Nonetheless, the extent to which such limitations altered our
results remains speculative. For example, available evidence suggests
that wealth– health gradients are more likely to narrow, not widen,
among elderly patients than among younger subgroups—subgroups
that disproportionately make up higher-risk “real-world” populations
(1, 3). Of importance, the objective of our study was not to measure
the true magnitude of association between income and mortality
after acute myocardial infarction but to quantify the extent to which
income–mortality associations were explained by traditional athero-
genic or vascular factors, noncardiac comorbid conditions, and
health service use. On the basis of our results and those of others (4),
there is no reason to believe that age and cardiovascular risk factors
782 16 May 2006 Annals of Internal Medicine Volume 144 • Number 10
Downloaded From: http://annals.org/ by a Brown University User on 01/04/2018
would not have exerted similar explanatory effects on income–mor-
tality associations if higher-risk populations had been examined.
To what extent, if any, can disparities between socioeconomic
status and mortality rates be modified through intensive secondary
prevention strategies? Are socially disadvantaged patients predestined
to die after acute myocardial infarction (regardless of intensive sec-
ondary prevention initiatives) because of their baseline cardiovascular
risk profiles at the time of presentation? These remain the pertinent
questions for future study. Social– epidemiologic and health service
research must now explore the impact of secondary preventive inter-
ventions to determine whether outcomes can be improved effectively
and efficiently among high-risk populations in the real world.
David A. Alter, MD, PhD, for the SESAMI Study Group
Institute for Clinical Evaluative Science
Toronto, Ontario M4N 3M5, Canada
Potential Financial Conflicts of Interest: None disclosed.
References
1. Kaplan GA, Keil JE. Socioeconomic factors and cardiovascular disease: a review of
the literature. Circulation. 1993;88:1973-98. [PMID: 8403348]
2. Tu JV, Willison DJ, Silver FL, Fang J, Richards JA, Laupacis A, et al. Impractica-
bility of informed consent in the Registry of the Canadian Stroke Network. N Engl J
Med. 2004;350:1414-21. [PMID: 15070791]
3. House JS, Lepkowski JM, Kinney AM, Mero RP, Kessler RC, Herzog AR. The
social stratification of aging and health. J Health Soc Behav. 1994;35:213-34. [PMID:
7983335]
4. Lynch JW, Kaplan GA, Cohen RD, Tuomilehto J, Salonen JT. Do cardiovascular
risk factors explain the relation between socioeconomic status, risk of all-cause mortal-
ity, cardiovascular mortality, and acute myocardial infarction? Am J Epidemiol. 1996;
144:934-42. [PMID: 8916504]
CLINICAL OBSERVATIONS
Acute Myocardial Infarction Associated with the Serotonin
Syndrome
Background: The serotonin syndrome is an adverse drug reaction
manifesting as mental status changes, autonomic hyperactivity, and
neuromuscular abnormalities caused by excess stimulation of central
nervous system and peripheral serotonin receptors. Diagnosis of the
syndrome is based on characteristic clinical findings and history of
exposure to serotonergic agents (1). Although peripheral serotonin
activity is important for maintaining vascular tone, cardiac ischemia
is not a typical complication of the serotonin syndrome or of selec-
tive serotonin reuptake inhibitor therapy. On the contrary, use of
these agents has been postulated to decrease the risk for acute myo-
cardial infarction (MI), possibly by inhibiting platelet activation (2).
Objective: To describe a case of the serotonin syndrome causing
an acute MI.
Case Report: A 31-year-old woman presented to the emergency
department because of a change in mental status. She had a history
of depression and bipolar disorder and had changed therapy from
quetiapine to duloxetine 3 weeks before presentation. Her other
medications were paroxetine, bupropion, and clonazepam. During
the week before hospitalization, her mother described the patient as
www.annals.org
 Letters
“twitchy.” On the day of admission, the patient had vomited and Ivan E. Liang, MD
was then found somnolent and confused. She had no history of Caritas St. Elizabeth’s Medical Center
intentional overdose, and no empty pill bottles were found at the Boston, MA 02135
scene. Findings on physical examination were consistent with the
serotonin syndrome (1). Her vital signs were significant for a heart
Potential Financial Conflicts of Interest: None disclosed.
rate of 120 beats/min and blood pressure of 160/104 mm Hg. She
was awake and confused with intermittent lucidity; she did not fol-
low commands. She had nystagmus, a fine tremor, lower-extremity References
hyperreflexia, and inducible ankle clonus. Initial electrocardiographic 1. Boyer EW, Shannon M. The serotonin syndrome. N Engl J Med. 2005;352:1112-
findings were unremarkable. Results of laboratory studies were sig- 20. [PMID: 15784664]
nificant for a serum creatine kinase level of 1638 U/L with normal 2. Schlienger RG, Fischer LM, Jick H, Meier CR. Current use of selective serotonin
fractionation and a serum troponin I level of 3.83 ␮g/L. reuptake inhibitors and risk of acute myocardial infarction. Drug Saf. 2004;27:1157-
The patient was treated with aspirin, carvedilol, and lisinopril 65. [PMID: 15554748]
for cardiomyopathy. Over the next 3 days, her serum troponin I level 3. Reznik I, Rosen Y, Rosen B. An acute ischaemic event associated with the use of
decreased, and repeated electrocardiography showed inverted T venlafaxine: a case report and proposed pathophysiological mechanisms. J Psychophar-
macol. 1999;13:193-5. [PMID: 10475726]
waves in leads I, aVL, II, aVF, and V3 to V6. She had septal, anterior,
4. Hack JB. Oral sumatriptan-induced myocardial infarction. J Toxicol Clin Toxicol.
and lateral hypokinesis; ejection fraction was 0.30 on echocardiogra-
2004;42:309-11. [PMID: 15362600]
phy. Her mental status gradually improved. She was discharged
5. Golino P, Piscione F, Willerson JT, Cappelli-Bigazzi M, Focaccio A, Villari B, et al.
home after 4 days but returned 1 week later with pericarditis-like
Divergent effects of serotonin on coronary-artery dimensions and blood flow in pa-
chest pain. During this second admission, a cardiac catheterization
tients with coronary atherosclerosis and control patients. N Engl J Med. 1991;324:
showed patent coronary arteries and an ejection fraction of 0.74.
641-8. [PMID: 1994246]
Discussion: Our patient had a cardiac biomarker leak with focal
wall motion abnormalities that resolved after 1 week, consistent with
myocardial stunning after acute MI. To our knowledge, the only case
report of an acute MI associated with selective serotonin reuptake CORRECTIONS
inhibitor therapy was a 69-year-old patient with diabetes and coro-
nary artery disease who experienced an acute MI 5 days after starting
venlafaxine therapy. This patient did not have clinical findings of the
serotonin syndrome and, unlike our patient, had evidence of coro- Correction: Nutrition and Blood Pressure: Is Protein One
nary atherosclerosis on cardiac catheterization (3). Link? Toward a Strategy of Hypertension Prevention
Excess serotonergic activity may be associated with ischemia, as A recent editorial regarding nutrition for prevention of hypertension
evidenced by multiple reports of sumatriptan-induced acute MI (4). (1) contained an incorrect reference citation. The authors cited a
The serotonin syndrome may result in ischemia through constriction study by Elliott and colleagues (2) that was still in press but has since
of coronary arteries because serotonin constricts most vascular beds. been published by a different journal.
Paradoxically, serotonin dilates normal coronary arteries while con-
stricting diseased ones (5). Endothelial 5-HT1 receptors release vaso-
dilatory endothelium-derived relaxing factor; consequently, vessels References
with atherosclerosis may not have this protective effect, which would 1. Cutler JA, Obarzanek E. Nutrition and blood pressure: is protein one link? Toward
result in unopposed 5-HT2 receptor–mediated vasoconstriction (5). a strategy of hypertension prevention [Editorial]. Ann Intern Med. 2005;143:74-5.
[PMID: 15998756]
However, our patient must have had ischemia caused by vasospasm
2. Elliott P, Stamler J, Dyer AR, Appel L, Dennis B, Kesteloot H, et al. Association
in the presence of normal coronary vasculature.
between protein intake and blood pressure: the INTERMAP Study. Arch Intern Med.
Conclusion: As the number of approved serotonergic agents in-
2006;166:79-87. [PMID: 16401814]
creases and as more patients are given combination serotonergic ther-
apy, the incidence of the serotonin syndrome will probably increase.
Treatment is generally supportive; benzodiazepines (and possibly
cyproheptadine) can be used to ameliorate serotonergic end-organ
effects. Also, a conscious effort must be made not to prescribe new
Correction: Efficacy and Safety of Inhaled Insulin Therapy
In a recent letter regarding efficacy and safety of inhaled insulin
serotonergic agents for these patients because they may cause symp-
therapy (1), Dr. Zinman’s institutional affiliation should have read as
toms to worsen. Clinicians should consider the presence of myocar-
follows: Mount Sinai Hospital and University of Toronto; Toronto,
dial ischemia in patients with serotonergic findings, particularly
Ontario, Canada.
those known to have coronary artery disease or its risk factors.

Michael Ganetsky, MD Reference

Steven B. Bird, MD 1. Rosenstock J, Zinman B, Murphy LJ, Clement SC, Moore P, Bowering CK, et al.
University of Massachusetts Medical School Efficacy and safety of inhaled insulin therapy [Letter]. Ann Intern Med. 2006;144:
Worcester, MA 01655 533-4.

www.annals.org 16 May 2006 Annals of Internal Medicine Volume 144 • Number 10 783

Downloaded From: http://annals.org/ by a Brown University User on 01/04/2018