Journal of Psychiatric Practice Vol. 14, Suppl. 1 28 March 2008
The science of evidence-based practice and systematic
review has advanced substantially over the past decade.
Systematic reviews are now done much more frequent-
ly and are updated more regularly; they are also more
generally accepted. The evidence on which to base sys-
tematic reviews of clinical efficacy and effectiveness is
much broader than in previous decades. Unfortunately,
the translation of the science of systematic reviews into
clinical practice has lagged behind.
Systematic reviews, conducted by experts in informa-
tion science, clinical practice, and clinical epidemiology,
are planned and written to rigorous quality standards.
This results in reviews that are often very long and
sometimes employ a substantial amount of jargon. The
average practitioner is likely to have difficulty with long
web-based or even academic journal-length reviews.
Therefore, the translation of systematic reviews into
evidence-based messages is a key intermediate step in
producing changes in clinical practice.
Obviously, the process of developing and disseminat-
ing such evidence-based messages is not sufficient in
itself to produce a change in practice, but this does rep-
resent a necessary step in the process of practice
improvement. Additional steps include use of appropri-
ate reimbursement systems, administrative simplicity,
technical and other support for practice change, and the
availability of alternative treatment strategies.1–5
This article describes the process by which key mes-
sages were derived from a systematic review. This
process was part of a 3-year project, the goal of which
was the national dissemination of fair and balanced
information concerning use of antiepileptic drugs
(AEDs) in the treatment of bipolar disorder. While vari-
ations of key messages derived from evidence reviews
Extracting Key Messages from
Systematic reviews have become a common method of synthesizing literature to examine the comparative effec-
tiveness of medical interventions. Groups such as the Cochrane Collaboration and the U.S. Agency for Healthcare
Research and Quality (AHRQ) conduct dozens of such reviews each year. However, these documents are often
hundreds of pages long and the intended audiences, including medical providers and payers, may have difficul-
ty interpreting the often technical terminology used in these reports. In this article, we describe the derivation
of “key concepts” on the efficacy of antiepileptic drugs (AEDs) in the treatment of bipolar disorder from a pre-
viously published systematic review of the use of AEDs (AED Review) for multiple indications, including mood
disorders and chronic pain. With the aid of a multidisciplinary science panel, we derived the key concepts from
the source report and subsequent updates of that report. Because we found that the key concepts were still quite
technical, we subsequently derived four less technical “key messages” and revised these through multiple addi-
tional iterations. The concepts and messages were then tested with key informants and focus groups. At all stages
of the process, we found that it was critical to maintain fidelity to the initial systematic review. The structured
approach used in the derivation process described here proved to be very helpful in developing key messages
and concepts. (Journal of Psychiatric Practice 2007;14(suppl 1):28–34)
KEY WORDS: bipolar disorder, antiepileptic drugs, carbamazepine, gabapentin, lamotrigine, valproate, off-label use,
TIMOTHY S.CAREY, MD, MPH
CATHY L. MELVIN, PHD, MPH
LEAH M. RANNEY, PhD
CAREY, MELVIN, and RANNEY: The Cecil G. Sheps Center for
Health Services Research, The University of North Carolina at
Copyright ©2008 Lippincott Williams & Wilkins Inc.
Please send correspondence and reprint requests to: Cathy L. Melvin,
PhD, MPH, The Cecil G. Sheps Center for Health Services Research,
The University of North Carolina at Chapel Hill, 725 Martin Luther
King Jr. Boulevard, CB#7590, Chapel Hill, NC 27599-7590.
Funded by a grant administered by a consortium of state attorneys
will likely have some unique characteristics specific to
each project, we believe our experience is illustrative of
the challenges inherent in the process itself. If the infor-
mation contained in systematic reviews is to become
useful and result in practice improvement, such a trans-
lation process is essential.
The project described here was funded by the
Neurontin Executive Committee, a consortium of state
attorneys general (see Melvin, et al. in this supplement
for a more detailed description of this project, p. 96).The
Neurontin Executive Committee contracted with the
Cecil G. Sheps Center for Health Services Research at
the University of North Carolina at Chapel Hill to
develop a national dissemination program concerning
the utilization of AEDs for bipolar disorder based on a
systematic drug class review conducted as part of the
Drug Effectiveness Review Project (DERP), which
involved a series of drug class reviews coordinated by
the Oregon Health Sciences University.7,8The DERP is
an established systematic review project conducting
drug class reviews for a broad consortium of state
Medicaid agencies. The original systematic review con-
cerning the AEDs conducted by DERP was posted in
December 2005 and the final report was completed in
May 2006. The scope of the AED Review was quite
broad and addressed use of AEDs for a variety of indi-
cations, including bipolar mood disorder, neuropathic
pain, and fibromyalgia. The original AED Review,
including evidence tables, was over 700 printed pages.
Thus, the need to distill this voluminous evidence into
messages that would be accessible to clinicians and pol-
icymakers was evident.
We first assembled a multi-disciplinary “science
panel” that included experts in systematic review, psy-
chiatry, and pharmacology. We were fortunately able to
include the primary author of the original systematic
review on the panel. All panelists disclosed any poten-
tial conflicts of interest and no relevant competing inter-
ests with pharmaceutical companies were identified.
The panel was assisted by staff members. The initial
task of the science panel was to read the report in detail
and discuss the findings with the primary author. The
panel also reviewed multiple background papers as well
as pivotal trials of AEDs in bipolar disorder. Because the
focus of the dissemination effort was the role of
gabapentin in the treatment of bipolar disorder, the
panel focused on primary publications concerning use of
this agent in bipolar disorder.
Meetings, conference calls, and web-based sharing of
documents greatly facilitated this process. Although the
focus of the dissemination process was largely utiliza-
tion of gabapentin, the panel members quickly recog-
nized that simple cautionary messages regarding use of
this medication would likely be perceived as essentially
negative without giving physicians a more positive mes-
sage as to therapeutic options and efficacious agents for
the treatment of this serious and often life-long illness.
The panel members were also reluctant to proceed
immediately from the 700 page report to actionable
messages; instead, they adopted an intermediate strate-
gy of moving from the report to “key concepts.”
Another activity undertaken early in the project was
the pharmaco-epidemiologic analysis of national data-
bases in order to gather information concerning pre-
scribing practices for AEDs in psychiatric disease.
Based on this analysis, the staff easily concluded that
the audience for the messages being developed would be
practicing psychiatrists, since they prescribe the vast
majority of AEDs for bipolar disorder.
The panel also found it necessary to expand some-
what on the data presented in the evidence report, since
they wished to examine the strength of the evidence
regarding the efficacy (or lack of efficacy) of the phar-
macologic agents under consideration in more detail. In
consultation with the staff of the UNC evidence-based
practice center, the science panel used the Grading of
Recommendations, Assessment, Development and
Evalution (GRADE) system to evaluate this evidence.9
The GRADE system is a method for assessing the
strength of evidence. It provides a global qualitative
assessment, which is often operationalized as an assess-
ment of the likelihood that additional work could lead to
a different conclusion. Thus, the GRADE system
includes high ratings (“further research is very unlikely
to change the estimate of effect”), moderate ratings
(“further research is likely to have important impact on
the confidence in the estimate of effect and may change
the estimate”), and low ratings (“further research is
very likely to have an important impact on the estimate
effect and is likely to change the estimate”). Some vari-
ations of the GRADE system also include a “very low”
rating. GRADE ratings were done in parallel with gen-
eration of the initial draft key concepts from the evi-
dence report. Over the course of 7 months, key concepts
were drafted by staff, presented to the science panel,
revised, redrafted, and presented again during multiple
conference calls and face-to-face meetings.In total,eight
successive versions of the key concepts were elaborated.
These versions intentionally began at a very general
EXTRACTING KEY MESSAGES FROM SYSTEMATIC REVIEWS
Journal of Psychiatric Practice Vol. 14, Suppl. 1March 200829
level and became successively more specific. However, at
some point, the panel found that the specificity of many
of the key concepts was too great and not sufficiently
supported by evidence in the report. When this
occurred, these key concepts were revised to be some-
what more general. The members of the science panel
emphasized the importance of maintaining fidelity to
the evidence throughout the process as they examined
the strength of the findings concerning possible benefits
that can be ascribed to the use of AEDs in the treatment
of bipolar disorder. A final critical issue for the science
panel was making a clear distinction between presenta-
tion of evidence in the form of key concepts and mes-
sages and development of a practice guideline.
Essentially all the members of the panel had previously
participated in guideline development activities and
recognized that evidence is the raw material upon
which guidelines should be based.Thus, the members of
the panel felt strongly that the key messages being
developed should describe the evidence and should not
be prescriptive statements concerning what providers
“should” or “should not” do in practice.
After multiple iterations of the key concepts had been
developed, the panel then developed more simply word-
ed messages derived from the somewhat more technical
key concepts. These messages went through six major
revisions in a process that once again required that the
messages be frequently checked against the core litera-
ture of the original evidence report and sometimes
against constituent trials and observational studies.
Ensuring that the key concepts and messages main-
tained fidelity with the systematic review of best evi-
dence was the top priority throughout the project.
Evidence reports and systematic reviews have a finite
life span, since scientific evidence is in constant evolu-
tion. Thus, in the late fall of 2006, the AED Report was
updated with additional trials,10,11necessitating further
review by the staff and the members of the science
panel. These additional findings were then compared
against the findings presented in the key concepts and
messages.While no major changes were made in the key
concepts and messages as a result of this update, sever-
al findings were reinforced.
Finally, the messages were tested using interviews
with key informants and focus groups of practicing psy-
chiatrists (for a description of this process, see Kish-
Doto et al., p. 3512).The valuable feedback obtained from
this process did not change the content but did alter the
presentation and wording of the messages. Wording in
the draft messages that was perceived as being more
indicative of guideline development rather than mes-
sage dissemination was changed; the final messages
were also more focused.
As discussed above, members of the science panel came
from different disciplines, and the panel’s initial work
reflected an excellent collegial relationship among the
panel members characterized by respect for the special-
ized expertise of each of the other disciplines represent-
ed. Some of the work also required that members of the
panel acquire specialized knowledge (for example, not
all of the members were familiar with the GRADE rat-
ing system for strength of evidence at the outset of the
The final key concepts are listed in Table 1 and can
also be accessed on our website (www.prescribingforbet-
teroutcomes.org). While the key concepts that were
developed were of course much briefer than the com-
plete evidence report, they were still relatively wordy
and the reviewers thought the language was somewhat
equivocal. For example, the wording of the first key con-
cept in Table 1, which maintains fidelity to the underly-
ing report, is quite dense and subject to more than one
interpretation. As noted above, multiple iterations were
required to arrive at this level of specificity. However,
with regard to our primary purpose of describing the
role of gabapentin in the treatment of bipolar disorder,
the key concept related to that issue (see Concept 8 in
Table 1) was more succinct.
Panel members thought that these key concept for-
mulations were necessary in order to be clear regarding
the acceptability of the evidence. For example, multiple
case series have tried to describe the efficacy of
gabapentin in bipolar disorder, but the science panel
unanimously believed that such case series evidence
was not acceptable in demonstrating efficacy of a treat-
ment in a condition which has a waxing and waning
course, such as bipolar disorder.
Although the equivocal nature of the key concepts was
judged to be inappropriate for a broad dissemination
campaign, the key concepts remain valuable as backup
documentation and references for the multiple itera-
tions of the key messages that were developed based on
The science panel had the most difficulty describing
the adverse events reported in association with the
AEDs. Such issues are unfortunately ubiquitous in drug
class reviews. Randomized trials of efficacy are often too
small to allow calculation of rates of adverse events
(also called “harms”) due to the medication. Pharmaco-
EXTRACTING KEY MESSAGES FROM SYSTEMATIC REVIEWS
Journal of Psychiatric Practice Vol. 14, Suppl. 130March 2008
EXTRACTING KEY MESSAGES FROM SYSTEMATIC REVIEWS
Journal of Psychiatric Practice Vol. 14, Suppl. 1March 200831
Table 1. Key Concepts about antiepileptic drugs (AEDs) derived from the Drug Effectiveness
Review Project (DERP) Report*,7
1. Current evidence supports the conclusion that three AEDs (carbamazepine, valproic acid/valproate, and lamotrigine)
are efficacious in maintaining remission for adult outpatients with primary diagnoses of bipolar I disorder with recent
mania or mixed episodes.
a. The overall magnitude of benefit obtained with AEDs in bipolar I disorder with recent mania or mixed episodes was
an absolute improvement of the probability of attaining remission ranging from 7% to 28%; the relative rate of
attaining remission was between 1.17 and 2.87 compared with placebo.The strength of evidence for this indication
is low (Grading of Recommendations,Assessment, Development and Evaluation [GRADE] criteria).
b. Carbamazepine is the only AED that has been shown in fair-quality published trials to be significantly better than
placebo in reducing mania scores in acute therapy of adult outpatients. Evidence is stronger for lamotrigine in
prevention of depressive than manic episodes.
c. There was no acceptable evidence to support choice of one agent over another based on speed of onset in attaining
2. Current evidence provides only modest support for the efficacy of the same three AEDs in achieving and maintaining
remission in adult outpatients with bipolar I disorder with a recent depressive episode or bipolar II disorder.The
strength of evidence in either of these subtypes of bipolar disorder is less than the strength of evidence regarding
efficacy in bipolar I disorder with recent mania or mixed episodes.
a. The overall magnitude of benefit obtained with AEDs relative to placebo in bipolar I disorder with a recent
depressive episode was an absolute improvement in attaining remission of 11%, with a relative rate of attaining
remission of 1.44 compared with placebo.The GRADE strength of evidence for use of AEDs for this indication is low.
b. The overall magnitude of benefit in bipolar II disorder is an absolute difference of 15%, with a relative rate of
attaining remission of 1.58 compared with placebo.The GRADE strength of evidence for use of AEDs for this
indication is low.
3. Efficacy of these agents in maintaining remission is generally based on the percentage of patients who do not
experience symptomatic recurrence or prematurely discontinue study treatment because of symptoms.There is great
variability among the methods used to measure symptoms and functional status in these populations.The evidence
indicating improvement in manic or depressive symptoms is much less clear due to difficulties comparing different
symptom scores and functional status measures in these populations. Most available evidence does, however,
demonstrate improvement in symptom scores compared with placebo.The level of absolute improvement in reducing
recurrence ranged from 1% to 23%, with a risk reduction ranging from 0.63 to 0.96. The GRADE strength of evidence
for use of AEDs for this indication is low.
4. Carbamazepine, valproic acid, and lamotrigine appear to have similar magnitudes of benefit in inducing remission,
although the risk of recurrence is substantial for all agents.While the three drugs have similar magnitudes of benefit
based on indirect comparisons, few studies directly compare these medications with each other; therefore, these
conclusions should be considered tentative.The GRADE strength of evidence supporting this conclusion is low.
5. The rates of achieving and maintaining remission during treatment with the three AEDs mentioned above are similar
to those obtained with lithium treatment for bipolar I disorder. For adult outpatients with acute mania,
carbamazepine and valproate were similar, relative to lithium, in terms of response rates.
a. The incidence of recurrence in the studies examined ranged from 16% to 70% with placebo and from 6% to 65% with
medication treatment.The broad range of these estimates is due to the variable definitions of recurrence used and
variable durations of follow-up. Recurrence is a significant problem for these patients even with treatment with AEDs.
epidemiologic studies may be limited by lack of speci-
ficity regarding the nature of the adverse event and dif-
ficulty assigning a rating of causality. Data on adverse
events occurring in association with medication are
often limited to case reports, which make comparison of
rates of adverse events among drugs very difficult, since
there is no denominator available with which to calcu-
late a rate. Pharmaco-epidemiology studies may allow
calculation of a rate of adverse events from a drug, but
are often limited by lack of detail regarding patient
characteristics and comorbidities. A particular issue
with bipolar disorder is the relatively high rate of sui-
cide attempts and completion. The panel discussed, at
some length, this devastating consequence of the dis-
ease as well as the important distinctions between sui-
cide as a consequence of the underlying disease and
suicide as a consequence of medication. Rather than a
consequence of medication, the panel felt that rates of
EXTRACTING KEY MESSAGES FROM SYSTEMATIC REVIEWS
Journal of Psychiatric Practice Vol. 14, Suppl. 132March 2008
Table 1. continued
6. The overall risk of adverse events resulting in medication discontinuation is similar among the three AEDs listed
above, and the overall risk of adverse events for AEDs is similar to that for lithium across all clinical subtypes of
bipolar disorder. However, the types of adverse events encountered differ among the three AEDs and lithium. Serious
adverse events, although uncommon, may occur with each agent; meaningful comparisons of the rates of these serious
events among agents could not be performed. Overall evidence regarding comparative adverse event rates is based on
a small number of studies.
a. Between 5% and 24% of patients discontinue an AED due to adverse events. The rates of serious adverse events
are on the order of 0% to 10.3%.The GRADE strength of evidence for this finding is low.
b. Rates of skin reactions are higher for carbamazepine and lamotrigine; the most common predictor of rash is history
of previous rash with an AED.
7. The risk of suicide or suicide attempt is present in bipolar disorder.
a. Compared with placebo, patients treated with AEDs for psychiatric conditions may have a small increase in suicidal
b. Evidence regarding protection against the risk of suicide attempt is limited and does not support a difference
between valproate and carbamazepine. Both of these agents have been associated with a lower protective effect
against suicide attempt than lithium in observational studies; however, the evidence of less protective effect is
stronger for valproate.There are insufficient data to compare the risk of suicide attempt among other AEDs.
8. There is some evidence (although not extensive) showing that gabapentin is no more, and perhaps less, efficacious
than placebo in the treatment of bipolar I disorder with recent mania and rapid cycling bipolar disorder. No acceptable
evidence was found to support the use of gabapentin in achieving remission or preventing relapse in bipolar disorder.
Evidence regarding efficacy of topiramate for any of the conditions discussed above (bipolar I disorder with recent
mania, hypomania, or mixed episodes and bipolar II disorder) is sparse. Current evidence regarding use of topiramate
for acute mania shows no evidence of efficacy.
9. The available evidence regarding potential differential efficacy among AEDs in the treatment of patients with rapid
cycling and other patient subgroups is extremely limited.Available evidence does not allow prediction of which patient
subpopulations will respond to any given AED.Available evidence does not allow prediction of response to a
subsequent AED based on response to an initial trial of therapy.
10. Little evidence is available regarding differential efficacy of the three AEDs in subpopulations defined by gender, age,
ethnicity, or comorbidity.
*These key concepts are derived from the Drug Class Review on antiepileptic drugs in bipolar mood disorder. The population of inter-
est is adults with new onset or chronic bipolar I and II disorder. While most of the studies in the review were conducted in outpatients,
some of the studies involved patients with acute disease who were hospitalized for initiation of therapy.
suicidality were more likely a result of inadequate doses
of medication, insufficient duration of treatment, or
individual pharmacokinetics. However, the lack of large
randomized trials made any conclusions regarding
these issues very tentative.
Four key messages were eventually derived from the
10 key concepts (Table 2). Pre-testing with members of
the practice community was especially valuable in refin-
ing the first and fourth key messages (see Kish-Doto et
al. p. 3512and Melvin et. al. p. 4413for a full description
of the audience research).While practitioners often stat-
ed that they “didn’t use gabapentin” as a primary drug
in the treatment of bipolar disorder, they often followed
this up by saying that they sometimes used it as adjunc-
tive rather than primary treatment. Therefore, the sci-
ence panel added an explicit statement reflecting the
lack of evidence for the efficacy of gabapentin either in
primary or adjunctive use.
The fourth key message regarding adverse events was
also modified. While the research literature showed no
significant differences among the AEDs and lithium
with regard to withdrawal from therapy or hospitaliza-
tion for adverse events, clinicians consistently focused
on the greater clinical impact of overdoses of lithium
compared with overdoses of other agents such as the
AEDs. The panel acknowledged this clinical insight and
revised the fourth key message to address this issue.
Our experience in developing key messages based on a
systematic review of the use of AEDs in bipolar disorder
illustrates several of the challenges in disseminating
the best evidence available for treatment of serious
chronic conditions. Systematic reviews are certainly
necessary for practice change since they summarize
available evidence. As such, they serve as the necessary
but not sufficient “science” base for the dissemination of
best available treatments in practice. Moreover, even
the relatively brief “key concepts” we derived from the
systematic review were not the same as the key mes-
sages that we ultimately developed for dissemination.
The process described in this article required the active
participation of individuals with expertise in the meth-
ods and content of both psychiatry and systematic
reviews as well as input from our intended audience of
practicing psychiatrists and from experts in health com-
munication and dissemination.
Modern systematic reviews, such as those commis-
sioned by the Agency for Healthcare Research and
Quality (www.ahrq.gov), address many key questions
and often examine large classes of medications. We
found that the science panel required at least several
months to digest, interpret, and boil down one such
extensive review into key concepts and messages.
EXTRACTING KEY MESSAGES FROM SYSTEMATIC REVIEWS
Journal of Psychiatric Practice Vol. 14, Suppl. 1March 200833
Table 2. Key Messages about antiepileptic drugs (AEDs) derived from AED Key Concepts (Table 1)*
Key Concept #Key Messages
1. There remains no scientifically acceptable clinical trial evidence which supports use of either
gabapentin or topiramate in bipolar mood disorder, either as monotherapy or as an adjunct
to other therapies.8
2. Research supports the use of three AEDs—1) carbamazepine, 2) valproic acid/valproate,
and 3) lamotrigine—in achieving and maintaining remission for adult outpatients with
primary diagnoses of bipolar I disorder. Evidence of efficacy is less clear for these treatments
for bipolar II disorder. 1, 2, 3, 4
3. Carbamazepine, valproic acid/valproate, and lamotrigine work as well as lithium in achieving and
maintaining remission in bipolar I disorder, but the strength of the evidence supporting this conclusion
is low, and additional research is needed to clarify the relative roles of these agents in bipolar disorder.5
4. The types of adverse events vary among AEDs and lithium. There is insufficient evidence to determine
if the overall risk of adverse events differs among AEDs. Unlike the AEDs, lithium poses a significant
risk when taken in an overdose.6, 7
*Key Concepts 9 and 10 refer to specific subpopulations that are not included in the Key Messages.
The utilization of a multidisciplinary “science panel” Download full-text
throughout the process provided insights into the
importance of obtaining the perspectives of multiple dis-
ciplines, including the end-user community. Similarly,
early pre-tests of the messages were extremely useful
and led to significant revisions. Even after revision, not
all potential users agreed with the messages, but the
panel recognized that transmission of evidence-based
messages to individuals who might not necessarily
agree with all of them is part of the process of practice
Maintaining fidelity to the original evidence report
while deriving messages couched in practical straight-
forward terms represents a challenge to all translation-
al researchers. Recently, the National Institutes of
Health (NIH) launched a national Clinical Transla-
tional Science Award (CTSA) program. Translational
research is of two types: “Type I” speeds the testing of
novel therapies in humans—i.e., it involves translation
from bench to bedside. “Type II” research, in contrast,
seeks to speed the movement of efficacious therapies
from clinical trial settings into practice. This second
type of translational research requires assembling data
through systematic reviews, since few clinical policy
changes are based on a single clinical trial. The results
of these trials and systematic reviews must be dissem-
inated to clinicians and the public in clear terms that
also discuss the limitations of the evidence base.
Communication between the systematic review com-
munity and the practice community should occur early
in the process. This second type of translation should
not be viewed as information that is being “handed
over” from systematic reviewers. Rather, the systemat-
ic review team must remain involved with the dissemi-
nation of the information by conducting appropriate
updates and especially by maintaining fidelity between
the messages and the underlying evidence. Although
this fidelity can never be 100% precise at the first pass,
ongoing communication among all parties will assure
the best possible fidelity. Simply instructing clinicians
to go to a website and absorb very bulky and technical
reports will likely only result in the perpetuation of
practices that are not evidence-based or potentially
1.Oxman AD, Thomson MA, David DA, et al. No magic bullets:
A systematic review of 102 trials of interventions to improve
professional practice. CMAJ 1995;153:1423–31.
Pathman DE, Konrad TR, Freed GL, et al. The awareness to
adherence model of the steps to clinical guideline compliance.
Med Care 1996;34:873–89.
Grimshaw JM, Shirran L,Thomas R, et al. Changing provider
behavior: An overview of systematic reviews of interventions.
Med Care 2001;39:112–45.
Grimshaw JM,Thomas RE, MacLennan G, et al. Effectiveness
and efficiency of guideline dissemination and implementation
strategies. Health Technol Assess 2004;8:1–72, iii–iv.
Grimshaw JM, Eccles M, Thomas R, et al. Toward evidence-
based quality improvement. Evidence (and its limitations) of
the effectiveness of guideline dissemination and implementa-
tion strategies 1966-1998. J Gen Intern Med 2006;21:S14–20.
Melvin CL, Carey T, Oldham JM. Introduction: Antiepileptic
medications for bipolar disorder. J Psychiatr Pract 2008;14
(suppl 1): 7–8.
Goodman F, Glassman P, Maglione M, et al. Drug class review
on antiepileptic drugs in bipolar mood disorder, neuropathic
pain, and fibromyalgia. Portland: Oregon Evidence-based
Practice Center; 2006 (available online at either www.pre-
scribingforbetteroutcomes.org or http://www.ohsu.edu/drugef-
Helfand M. Using evidence reports: Progress and challenges
in evidence-based decision-making. Health Aff 2005;24:123–7.
Guyatt G, Vist G, Falck-Ytter Y, et al. An emerging consensus
on grading recommendations? ACP J Club 2006;144:A8–9.
10. Kushner SF, Kahn A, Lane R, et al. Topiramate monotherapy
in the management of acute mania: Results of four double-
blind, placebo controlled trials. Bipolar Disord 2006;8:15–27.
11. Vieta E, Manuel-Goikolea J, Martinez-Aran A, et al. A double-
blind, randomized controlled prophylaxis study of adjunctive
gabapentin for bipolar disorder.
12. Kish-Doto J, Evans WD, Squire C, et al. Patterns of prescrib-
ing antiepileptic drugs for bipolar mood disorder. J Psychiatr
Pract 2008;14 (suppl 1):35–43.
13. Melvin CL, Ranney L, Carey T, et al. Disseminating findings
from a drug class review: Using best practices to inform pre-
scription of antiepileptic drugs for bipolar disorder. J
Psychiatr Pract 2008;14(suppl 1):44–56.
J Clin Psychiatry
EXTRACTING KEY MESSAGES FROM SYSTEMATIC REVIEWS
Journal of Psychiatric Practice Vol. 14, Suppl. 134 March 2008