New antiepileptic drug safety information is not transmitted
systematically and accepted by U.S. neurologists☆
Sarah G. Bella, Martha Matsumotoa, Susan J. Shawb, Jason Brandtc, Gregory L. Kraussa,⁎
aJohns Hopkins University, Department of Neurology, 600 N. Wolfe St., Meyer 2-147, Baltimore, MD 21287, USA
bUniversity of Southern California, Keck School of Medicine, Department of Neurology, Rancho Los Amigo National Rehabilitation Center, 7601 E. Imperial Highway, HB 145, Downey,
CA 90242, USA
cJohns Hopkins University, Division of Medical Psychology, 600 N. Wolfe St., Meyer 218, Baltimore, MD 21287, USA
a b s t r a c ta r t i c l ei n f o
Received 1 May 2013
Revised 3 June 2013
Accepted 8 June 2013
Available online 7 August 2013
Methods of education
Error in medicine
Professional conduct and ethics
We surveyed U.S. neurologists in order to evaluate their knowledge of, and sources for, recent FDA safety warn-
ings regarding antiepileptic drugs (AEDs) and whether they incorporate this information into their practices.
Surveyrespondents (N = 505)werepredominantlyboard-certifiedAmericanAcademyofNeurologymembers.
Approximately 20% of respondent neurologists were not aware of warnings about four drug safety risks:
suicidality with newer AEDs, increased birth defect risks from in utero divalproex exposure, impaired cognitive
descent starting carbamazepine. Most respondents were aware of a recommendation for haplotype screening,
yet did not routinely perform the safety screening, and 18 reported patients that had hypersensitivity reactions
to carbamazepine. Respondents learned about drug safety risks from varied sources; only notifications from
specialty organizations were associated with accurate knowledge of drug safety warnings. Most surveyed neu-
rologists would prefer implementing “a formal warning process via specialty organizations” with e-mails of
updated product insert warnings.
© 2013 The Authors. Published by Elsevier Inc. All rights reserved.
Neurologists need up-to-date information on the safety of widely
used medications to care for their patients effectively. The safety
warnings of the FDA, however, are not systematically transmitted to
ty risks. The FDA reports these risks in “Drug Safety Communications”
and “MedWatch Alert” e-mails, but these are only directly transmitted
ly receive e-mail warnings of safetyrisks from theFDA, which maythen
be passed along to members. The FDA also requires that drug manufac-
turers update product inserts and highlight particularly severe or un-
usual drug safety risks in “black box” warnings and “Dear Health Care
Provider” letters [1,2]. Many neurologists learn of drug safety risks
from published articles and continuing medical education (CME). It re-
rologists effectively and whether physicians implement recommended
drug safety screening into their practices. Previous studies, for example,
show that physicians are often aware of updated drug safety informa-
tion but may not carry out recommended practices [3–5].
Four new safety risks for antiepileptic drugs (AEDs) have been re-
cently identified: increased suicidal thoughts or behavior with newer
AEDs , high risks for birth defects  and cognitive impairment 
sitivity reactions in patients of Asian descent starting carbamazepine
. The empirical basis and causative relationships underlying some
. However,webelieve itisimportant forneurologists tobe aware of
major drug safety warnings in order to incorporate appropriate safety
screening procedures and patient counseling into their practices. We
cent AED safety warnings, their sources of drug safety information, and
whether they incorporate this safety information into their practices.
This information may help identify optimal methods for imparting im-
portant safety updates to neurologists.
2.1. Standard protocol approvals and patient consents
This survey study meets the criteria for research exempt from IRB
monitoring under Exemption 45 CFR 46.101(b)(2).
Epilepsy & Behavior 29 (2013) 36–40
☆ This is an open-access article distributed under the terms of the Creative Commons
⁎ Corresponding author at: 600 N. Wolfe St., Meyer 2-147, Baltimore, MD 21287, USA.
Fax: +1 410 955 0751.
E-mail address: firstname.lastname@example.org (G.L. Krauss).
1525-5050/$ – see front matter © 2013 The Authors. Published by Elsevier Inc. All rights reserved.
Contents lists available at ScienceDirect
Epilepsy & Behavior
journal homepage: www.elsevier.com/locate/yebeh
A master list of active members of the American Academy of
Neurology (AAN) was stratified by state and systematically sampled.
Surveys were transmitted to 4627 U.S. neurologists by three e-mail
requests. The survey was conducted using “SurveyGizmo,” an Internet-
consulting outside sources.
2.3. Survey questions
Survey questions concerned the following four recently identified
safety risks: (1) A high risk for hypersensitivity reactions (e.g., Stevens–
Johnson syndrome and toxic epidermal necrolysis) with carbamazepine
use has been associated with the HLA-B*1502 haplotype marker, which
is more common in patients of Asian descent. The FDA recommends
haplotype screening for patients of Asian descent before initiating
carbamazepine therapy . (2) The FDA identified a doubling of risk
cidal ideation risk increasing from 2.4 to 4.3 per 1000 patients. Though
controversial among neurologists, the FDA recommends warning
and monitoring patients who are prescribed these AEDs. (3) Recent
studies have shown an increase in the previously reported risks for
malformations in the offspring of women treated with divalproex,
with a risk of approximately 10.7% at the time of the survey . (4) A
major longitudinal study recently identified increased risk for cognitive
limitations in children after exposure in utero to divalproex, with a de-
three alternative AEDs during pregnancy . Only preliminary findings
are reported in the drug product insert , and this was used as a test
of knowledge of an evolving safety risk. We also included a control
question; neurologists were asked whether they knew that lacosamide
did not have “black box” safety warnings. Neurologists were also asked
to rank five methods of distributing drug safety information from most
effective to least effective. The complete list of survey questions and
neurologist responses can be found in the supplementary addendum.
2.4. Statistical analysis
Respondents were scored on their knowledge of the exact level of
risk cited by the FDA and on their knowledge of the existence of a sig-
nificantly increased risk for the four new drug safety questions and
the lacosamide control question. Knowledge of safety risks in neurol-
ogists who did and did not use each source was compared using a
t-test. Additionally, chi-squared tests or ANOVA was used to test for
dependence between demographics, practice characteristics, answers
to individual questions, and use of each source of information.
3.1. Characteristics of survey respondents
Six hundred five (13.1%) of the 4627 survey recipients completed
and returned the survey within six weeks. One hundred respondents
reported that they did not care for patients with epilepsy; responses
from the remaining 505 were analyzed. Responses were received
from all 50 states and the District of Columbia. The backgrounds of
respondents were similar to national AAN membership: one-half
(49.5%) were in solo or group practice, 66.5% were male, mean age
was 48.9 years (range: 29–77), and survey respondents were in prac-
tice an average of 22.4 years (range: 3–52) . The majority (54.0%)
of survey respondents treat at least 100 patients with epilepsy each
year. Compared to all AAN members, a slightly larger proportion of
respondents had academic or government-based practices (27.1%
AAN members; 39.8% survey respondents) and had board certifica-
tion (76%; 96.4%)  (Table 1).
3.2. Sources for drug safety information
Neurologists reported receiving drug safety information from
multiple sources. The most common sources were the following: no-
tifications from specialty organizations (67.1%), published literature
(67.1%), colleague and/or peer (53.1%), and CME or other educational
programs (52.9%) (Fig. 1). Respondent neurologists in academic and
private clinical practices used many of the same sources of drug safety
information. Private practice neurologists, however, were more likely
to acquire their knowledge from pharmaceutical representatives
(54.8% private, 23.4% academic), while academic neurologists were
(59.2% private, 75.1% academic).
3.3. Drug safety survey responses (Table 2)
3.3.1. FDA recommendation to perform haplotype screening prior to
initiating carbamazepine therapy in patients with Asian heritage
Seventy percent of respondents reported that they were aware
of this 2007 FDA recommendation. While 147 neurologists (29.1%)
reported initiating carbamazepine treatment in patients of Asian de-
scent since the warning, only 33 of them (22.5%) said they performed
haplotype screening. Eighteen neurologists reported that their pa-
tients of Asian descent had developed carbamazepine-related hyper-
sensitivity reactions during this time period.
3.3.2. Suicidal ideation with newer AEDs
Of the respondents,80.6% were awareoftheFDAwarningregarding
suicidal thoughts; 70.1% reported counseling patients on this risk in the
three years since the safety notification. Only 60.2% of respondents cor-
rectly identified the risk as 4.3 per 1000 (Fig. 2). Many neurologists
commented that the FDA's findings and recommendations are contro-
versialandthatknowingtheexactrisk of “suicidality”(itselfa nebulous
concept) is not essential to good clinical care.
3.3.3. Increased risks for birth defects with divalproex exposure
Only 33.5% of respondents were aware of the recently identified
increased risk of birth defects in the offspring of mothers treated
with divalproex. Many noted that it may be sufficient for them to
be aware of an increased risk for birth defects but not of this specific
risk. Nearly all (93.3%) respondents reported counseling women who
were planning pregnancies on birth defect risks with divalproex
Demographics of neurologist survey respondents compared to active AAN members:
Survey respondents' demographics were similar to active AAN members, but a higher
proportion of respondents were board-certified and were in academic practice.
Survey neurologists Active AAN neurologists
Mean age (range) (years)
Work settings (%)
aThe AAN membership practice demographics describe this as “University-Based
Group” and “Government Hospital or Clinic.”
bThis represents U.S. neurologists in the AAN membership practice demographics
who fell into the “Neurology Group” or the “Multispecialty Group.”
cThis represents U.S. neurologists in the AAN membership practice demographics
who fell into “Staff-Model HMO” or “Other Public or Private Hospital or Clinic.”
S.G. Bell et al. / Epilepsy & Behavior 29 (2013) 36–40
3.3.4. Cognitive impairment in children with divalproex exposure in utero
The risk of lowered IQ in children with in utero divalproex expo-
sure was recognized by approximately one-half of the respondent
neurologists (48.9%), while 16.9% were not aware of any risk for
decreased IQs (Table 2). The remaining one-third of neurologists
(30.7%) thought that “cognitive and development risks in offspring
exposed to divalproex in utero has not been established.” The current
divalproex product insert does not include this specific risk but notes
that “there have been reports of developmental delay, autism and/or
autism spectrum disorder in the offspring of women exposed to
valproate during pregnancy.”
3.3.5. No “black box” warnings for lacosamide
Seventy-three percent of respondents correctly answered the con-
trol question and recognized that there was no “black box” warning
for lacosamide (Table 2).
3.4. U.S. neurologists' overall knowledge of drug safety warnings
Of the respondents, 73.9% knew of the increased risks associated
with AEDs, as judged by their recognition of at least 4 of 5 safety issues
queried. However, those who recognized specific levels of drug risks,
such as rate of birth malformation with divalproex, were fewer: only
29.3% recognized specific drug risks for at least 4 of the 5 items que-
ried, and only 6.7% knew all five.
Only one source of information, notifications from specialty organi-
zations, was associated with increased specific knowledge and general
recognition of drug safety risks (specific knowledge: t(333) = 2.53,
p = 0.012; general recognition: t(321) = 3.33, p = 0.001).
Respondents' practice type, region of practice, years in practice,
and age were not associated with their knowledge of drug safety
issues. The number of patients with epilepsy treated each year
was associated with a modest increase in knowledge of drug safety is-
sues (specific knowledge: rs= 0.14, p = 0.002; general recognition:
rs= 0.17, p b 0.001).
3.5. Neurologists' preferences for methods to receive updated drug safety
Most neurologists prefer to receive safety information from struc-
tured, easily obtained sources. A large number endorsed the creation
of “a formal warning process via specialty organizations” (n = 190)
or e-mails with “updated product insert warnings to specialists in
that field” (n = 176). Another 101 neurologists thought that phar-
maceutical companies should be required to “notify physicians of all
major safety updates on product inserts, not just the very serious
‘black box’ warnings.” Few respondents (n = 24) prefer to “continue
thecurrentsystemunchanged,”andevenfewer(n = 14)preferrelying
on safety information from FDA websites.
The U.S. neurologists in the survey received drug safety warnings
from multiple sources; however, safety information was not system-
atically delivered, and approximately one-fifth of survey respondents
(17–27% across the five safety issues) were unaware of recently
reported safety risks with AEDs. Neurologists who treated large num-
bers of patients with epilepsy had an increased knowledge of drug
safety risks compared to other neurologists, though geographic distri-
bution, years in practice, board certification, and age were not associ-
atedwithincreasedknowledge.Theonly methodofsafety notification
tifications from specialty organizations.
Many neurologists said that they disagreed with the significance of
several of the FDA warnings. For example, many felt that the FDA's ev-
idence for increased risk of suicidality with AEDs was flawed, did not
Fig. 1. In general, what is/are your main source(s) for drug safety information? Each bar represents the percentage of respondent neurologists who answered that they did use the
source. Many neurologists reported having several drug safety information sources.
Neurologists' general and specific knowledge of five safety risks for AEDs: approximately
1/5 of neurologists did not recognize general safety risks for the drugs; knowledge of
specific levels of risks for AEDs varied widely (33 to 74%).
N (%) correct
exact level of riskb
N (%) correct
Hypersensitivity reactions in Asian
Suicidal ideation with newer AEDs
Congenital malformations with
Safety risks for lacosamide
IQ changes with divalproex
410 (81.2%)373 (73.9%)
Total knowledge scores
(for the previous 5 questions)
N (%) correct
exact level of riskb
N (%) correct
0–1 questions correct
2–3 questions correct
4–5 questions correct
aDid the neurologist identify whether a safety risk existed or not?
bDid the neurologist select the exact level of the increased risk?
S.G. Bell et al. / Epilepsy & Behavior 29 (2013) 36–40
include key risk factors for suicide such as depression, and should not
necessarily be included in counseling. In a previous 2008 survey
by Shneker et al. , many neurologists thought that the FDA
warning on suicidality risks with AEDs was unclear, and they were
not counseling patients on this risk. Some respondents also felt
that having a general awareness of drug safety risks was sufficient
in making clinical decisions. Many, however, did not recognize the
general levels of risk which would be necessary to assess even contro-
versial safety warnings. For example, knowing that the FDA's reported
risk for suicidality with new AEDs is relatively low (4.3 per 1000) is
helpful in deciding whether to counsel patients on the risk.
The FDA's recommended pharmacogenomic screening to avoid
carbamazepine-associated hypersensitivity reactions did not appear to
be effective—many survey neurologists had recently treated Asian pa-
tients with carbamazepine and yet, despite recognizing an association
between the HLA-B*1502 haplotype and carbamazepine hypersensitiv-
ogists reported Asian patients developing hypersensitivity reactions.
This is consistent with previous studies showing that the majority of
physicians eventually become aware of FDA drug safety advisories but
cedures [4,5,15]. Physicians, for example, often continue prescribing
contraindicated concurrent medications despite drug safety warnings
. Pharmacogenomic safety screening may have failed for multiple
reasons, including the following: the FDA did not provide technical in-
formation on how to obtain and interpret haplotype testing, specialty
organizations have not endorsed or encouraged testing, and logistical
difficulties in obtaining testing while delaying treatment may be a bar-
rier for neurologists. In addition, 30% of respondents were not aware of
a requirement to perform haplotype screening.
The difficulty of incorporating new information into neurologists'
practices is not exclusivetoAED safety risks.For example, anAAN prac-
tice parameter recommending early epilepsy surgery screening for pa-
tients with medically resistant seizures did not reduce the 17-year
average period of patients' pharmacotherapy before they were referred
for surgery evaluation . It is possible that, similar to AED suicidality
warnings, neurologists disagreed with the practice recommendation.
Our survey showed, though, that neurologists receive medical up-
dates nonsystematically from varied sources, and it is likely that
many neurologists were not aware of the practice recommendation.
Electronic medical record (EMR) technology would be ideal for
personalizing safety warnings at point of care; for example, pre-
scribers ordering carbamazepine for Asian patients could be warned
that the patient needs haplotype screening. EpiCare (EPIC), a widely
used electronic medical record tool, automatically warns prescribers
of potentially dangerous drug interactions, but it does not link patient
factors and drug safety warnings, e.g., when women of childbearing
age are prescribed divalproex, it does not warn prescribers of high
We used an Internet survey to permit adaptive questions; however,
the low yield of 13.1% could select for specific responses—for example,
physicians more knowledgeable about safety risks (or perhaps less
knowledgeable) might be more likely to respond. In a national survey
on colorectal cancer, a much higher proportion of physicians preferred
mail (90%) than Internet (6%) practice surveys, suggesting that neurol-
ogists might also be best surveyed via mailings . Nonetheless, over
500 U.S. neurologists from all U.S. states and practice types responded,
with relatively uniform patterns of responses. Moreover, the sample
was typical for U.S. board-certified neurologists and AAN members:
all practice types were represented, with the majority of neurologists
in private practice; neurologists' years in practice ranged from 1 to
across respondents; practice type, location, and years in practice had
little effect on knowledge of drug safety issues. Another limitation
of the questionnaire was that multiple-choice questions were used,
individual responses could not be explored, and only warnings for
antiepileptic drugs were probed. Overall, survey respondent back-
grounds were typical of board-certified U.S. neurologists, with the ma-
jority in private practices and a slightly higher proportion of academic
neurologists responding than in the overall AAN membership.
Our findings raise several important issues on how drug safety
screening could be better communicated and implemented. Many
neurologists received new safety information from the FDA, pub-
lished literature, and product insert updates; however, the sources
of safety information varied considerably among neurologists, and
neurologists noted that there was no systematic delivery of safety in-
formation. Despite using multiple sources, neurologists reported that
drug safety information was most effectively transmitted via specialty
organizations. These survey results suggest that it may be helpful for
the FDA to review controversial safety warnings and to implement
Fig. 2. Neurologists' estimate of risk for suicidality with newer AEDs. Eighty percent of respondent neurologists recognized an increased risk associated with newer AEDs, and 60%
knew of the specific risk of 4.3 per 1000. Nineteen percent did not recognize an increase compared to a background risk of 2.4 per 1000.
S.G. Bell et al. / Epilepsy & Behavior 29 (2013) 36–40
drug safety screening programs with advisory panels and specialty Download full-text
organizations. Neurologists would prefer that FDA safety warnings be
communicated more systematically via e-mail summaries specific to
their specialties. For pharmacogenomic screening, an FDA risk evalua-
tion and mitigation screening (REMS) program could be implemented
which requires safety screening before prescribed drugs are released
by pharmacies, e.g., the vigabatrin vision safety SHARE program .
A risk mitigation program may, however, be difficult to implement for
a small at-risk U.S. population and may be more appropriate for more
frequent, serious drug risks.
Previous studies have identified the absence of a structured, sys-
tematic approach to disseminating important drug safety information
to U.S. health-care providers [2,3]. They have also shown that FDA
drug safety communications are most effective when they are specific
and repeated and when they provide alternative care options to
providers . Our survey showed that only notifications from specialty
organizations improved neurologists' knowledge of drug safety risks.
Neurologists in the survey strongly suggested that the current system
of disseminating drug safety warnings be changed. Most frequently,
they expressed a preference for a formal drug safety warning process
updates for drugs in their specialty field. Whether a revised drug safety
notification system could be devised to improve knowledge among the
approximately 20% of neurologists who do not recognize important
safety risks despite reporting exposure to multiple sources for drug in-
formation remains to be determined.
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://
 U.S. Food and Drug Administration. Drug safety communications. Available at:
http://www.fda.gov/drugs/drugsafety/ucm199082.htm. [Accessed July 19, 2012].
 Duestzina SB, Higashi AS, Dorsey ER, Conti R, Huskamp HA, Zhu S, et al. Impact of
FDA drug risk communications health care utilization and health behaviors. Med
 Cluston Jr RJ, Li Z, Heaton PC, Weiss SR, Zuckerman IH, Moomaw CJ, et al. Impact of
regulatory labeling for troglitazone and rosiglitazone on hepatic enzyme monitor-
Pharmacoepidemiol Drug Saf 2005;14:1–9.
 Graham DJ, Drinkard CR, Shatin D, Tsong Y, Burgess MJ. Liver enzyme monitoring
in patients treated with troglitazone. JAMA 2001;286:831–3.
 Wagner AK, Chan KA, Dashevsky I, Raebel MA, Andrade SE, Lafata JE, et al. FDA
drug prescribing warnings: is the black box half empty or half full?
Pharmacoepidemiol Drug Saf 2006;15:369–86.
 U.S. Food and Drug Administration. Information for healthcare professionals: sui-
cidal behavior and ideation and antiepileptic drugs. Available at: http://www.fda.
ucm100192.htm. [Accessed May 14, 2012].
 Wyszynski DF, Nambisan M, Surve T, Alsdorf RM, Smith CR, Holmes LB. Increased
rate of major malformations in offspring exposed to valproate during pregnancy.
Neurology 2005;64:961–5 [serial online].
 Meador KJ, Baker GA, Browning N, Clayton-Smith J, Combs-Cantrell DT, Cohen M,
et al. Cognitive function at 3 years of age after fetal exposure to antiepileptic
drugs. N Engl J Med 2009;360:1597–605.
 Alfirevic A, Jorgensen AL, Williamson PR, Chadwick DW, Park BK, Pirmohamed
M. HLA-B locus in Caucasian patients with carbamazepine hypersensitivity.
 Hesdorffer DC, Kanner AM. The FDA alert on suicidality and antiepileptic drugs:
fire or false alarm? Epilepsia 2009;50(5):978–86.
 FDA Alert. Information for healthcare professionals: dangerous or even fatal
skin reactions — carbamazepine (marketed as Carbatrol, Equetro, Tegretol, and
generics). Available at: http://www.fda.gov/Drugs/DrugSafety/PostmarketDrugSafety
InformationforPatientsandProviders/ucm124718.htm. [Accessed May 29, 2012].
 Impact of FDA drug risk communications on health care utilization and health
behaviors: a systematic review. North Chicago (IL): Abbott Pharmaceuticals PR
Ltd.; 2009 [Package insert. NDC 0074-6212-13].
 Neurologists. AAN member demographics. Available at: https://www.aan.com/
globals/axon/assets/6723.pdf; 2009. [Accessed April 10, 2012].
 Shneker BF, Cios JS, Elliott JO. Suicidality, depression screening, and antiepileptic
drugs: reaction to the FDA alert. Neurology 2009;72:987–91.
 Busch SH, Frank RG, Leslie DL, Martin A, Rosenheck RA, Martin EG, et al. Antide-
pressants and suicide risk: how did specific information in FDA safety warnings
affect treatment patterns? Psychiatr Serv 2010;61:11–6.
 Smalley W, Shatin D, Wysowski DK, Gurwitz J, Andrade SE, Goodman M, et al.
Contraindicated use of cisapride: impact of Food and Drug Administration regula-
tory action. JAMA 2000;284:3036–9.
 Haneef Z, Stern J, Dewar S, Engel J. Referral pattern for epilepsy surgery after
evidence-based recommendations. Neurology 2010;75:699–704.
primarycare physicians' colorectal cancer screening recommendations andpractices.
Prev Med 2003;36:352–62.
 Pellock JM, Faught E, Sergott RC, Shields WD, Burkhart GA, Krauss GL, et al. Regis-
try initiated to characterize vision loss associated with vigabatrin therapy. Epilep-
sy Behav 2011;22:710–7.
state of Ohio medicaidprogram.
S.G. Bell et al. / Epilepsy & Behavior 29 (2013) 36–40