Pharmacological Approaches to StutteringTreatment: Reply to Meline and Harn (2008)
Purpose: To address questions raised by T. Meline and W. E.
used in our review, however, Maguire et al.’s results met
Harn (2008) in their critique of our previous article (A. K.
neither of our two outcomes criteria (stuttering, disfluency,
Bothe, J. H. Davidow, R. E. Bramlett, D. M. Franic, & R. J.
or any such measure below 5%, or documented improvement
in any nonspeech variable) at either of the two times we
Method: Additional information is provided to address several
assessed (immediately posttreatment and 6 months post-
treatment). Maguire et al.’s results also did not meet our
Results and Conclusions: Our previous systematic reviewomitted 1 relevant article about the use of olanzapine in
secondary outcome criterion of a 50% reduction in stuttering
stuttering, but the minimal effectiveness and the known
(or any such variable), which we assessed in case the
serious side effects of this drug limit the implications of this
absolute 5% criterion could be viewed as too stringent.
omission. While we do not agree with many of Meline and
Meline and Harn questioned whether Maguire et al.’s
Harn’s critiques of our review, we do agree with them that
results might nevertheless be interpreted as positive. They
several larger points raise interesting questions about the
mentioned the quality of life literature in this context, writing
structure, analysis, and usefulness of literature reviews in
that “a standardized effect size of 0.50 (half a standard de-
stuttering and in other areas. Fundamentally, we reassert our
viation) is equivalent to increasing the success rate from
agreement with Meline and Harn that there is insufficient
25% to 75%” (p. 95). Equating standard deviations of
evidence to support the use of existing pharmacological
test scores and effect sizes is problematic, however; the
agents in the treatment of stuttering.
intended meaning within the quality of life literature was
Key Words: stuttering, olanzapine, systematic review
that 0.5 SD represents a change of 8% of the quality of lifescale (Sloan, Symonds, Vargas-Chanes, & Fridley, 2003).
Overall, in addressing our choice of outcomes criteria,Meline and Harn suggested that some improvements canbe important even if they do not represent a completecure, and of course we agree. The point of our review,
Meline and Harn (2008) raised several interesting however,wasnottoidentifyalltreatmentsthatmight
issues about pharmacological treatments for stut-
represent some improvement to someone, but to identify
tering, about our recent review of the literature
those treatments that met one set of reasonable, objective,
in this area (Bothe, Davidow, Bramlett, Franic, & Ingham,
and actually rather lenient outcomes criteria.
2006), and about reviews of the literature in general. We
It is also important to note that Maguire et al. (2004, p. 66)
are pleased to have this opportunity to respond to, and
described olanzapine as “well-tolerated.” Meline and Harn,
similarly, quoted both a well-known stuttering textbook(Guitar, 2006) and a new chapter from a book on stuttering
Olanzapine in the Treatment of Stuttering
(Ludlow, 2006) as suggesting that side effects are not prob-lematic for olanzapine, “indicating that further research
Among the most straightforward of Meline and Harn’s
would be warranted” (Ludlow, 2006, p. 248; Meline & Harn,
points is their first: that our review did not include or refer
2008, p. 93). The known side effects of olanzapine are
to Maguire et al.’s (2004) double-blind, placebo-controlled trial
serious and numerous, however, even if they have not
of the use of olanzapine for developmental stuttering. In the
been reported in the relatively small studies of its use for
terms used in our review, Maguire et al.’s study met three out
stuttering. It is critical for readers to be aware, for example,
of five methodological criteria (random assignment to groups,
that a “Dear Doctor” letter about olanzapine (marketed
pretreatment and posttreatment data, and reliability data, but
by Eli Lilly as Zyprexa) was released on March 1, 2004.
no speech rate or speech naturalness data and no beyond-
“Dear Doctor” letters are typically required by the U.S. Food
clinic data). It should have been included in our review as
and Drug Administration to be forwarded to physicians by
a relatively well-designed study and discussed in more detail
manufacturers if there have been problems with a product.
in our article. Our only response to Meline and Harn on this
In this case, it was noted that “hyperglycemia, in some cases
point is simply to agree that we missed it (and, indeed, to
extreme and associated with ketoacidosis or hyperosmolar
wonder with them how we could possibly have missed it).
coma or death” (U.S. Food and Drug Administration, 2004,
More interesting than this study’s absence, however,
para. 2) had been reported as associated with use of Zyprexa
are its implications. Maguire et al. (2004) concluded that
(olanzapine). More recently, a New York Times article (“Lilly
olanzapine was “statistically superior to placebo” (p. 65)
Settles,” 2007) reported that the company has now settled
for three different dependent variables and described it as
lawsuits involving 18,000 individuals who claimed they had
“useful” (p. 66) in the treatment of stuttering. In the terms
“developed diabetes or other disorders after taking Zyprexa”
American Journal of Speech-Language Pathology • Vol. 17 • 98–101 • February 2008 • A American Speech-Language-Hearing Association
( p. C1). This means that, along with earlier settlements, at
1970 because we wanted to address stuttering treatments
least 28,500 individuals have received compensation from
developed since the introduction of prolonged speech; we
the manufacturer for claims that they were injured by
searched certain journals and used the key words we selected
because they seemed a relevant and comprehensive subset
Such issues are obviously serious and raise questions in
(and we note that Meline and Harn suggested only the one
our minds about describing this agent as “well tolerated.”
article that we missed, an omission that was a human error,
While there may yet be situations in which olanzapine could
not a flaw in our selection of databases or a problem that
be used, its known side effects must be considered in any
would have been solved by more databases); and we chose,
calls for “further research.” We might suggest that any
given the structure of our project and after consultation with
further research of its use in stuttering must, at a minimum,
the editors during the review process, to make the larger
include repeated fasting and postprandial blood glucose
list of all the articles we had consulted available to interested
testing; the blood testing reported by Maguire et al. was
readers upon request rather than taking the journal space
insufficient for an agent known to have hyperglycemic side
to publish them. Meline and Harn are correct in noting that
effects (Cavalot et al., 2006; Kendall, 2005; Liebl, 2003).
we did not describe in detail the process through which we
In addition, because this agent is an antipsychotic drug,
removed inappropriate articles (e.g., those that were about
further information is needed regarding its cognitive impact
pharmacological agents that caused stuttering-like speech)
in nonschizophrenic patients before it could reasonably
from the initial pool retrieved by our searches; that task, like
be recommended as a treatment for stuttering.
the others in our project, was completed by multiple readers.
Meline and Harn also commented on the fact that our
review included only published articles. They are correct in
noting that some definitions of systematic reviews emphasizethe importance of including unpublished material, but there
is by no means universal agreement about this point. One
The bulk of Meline and Harn’s letter proceeded to critique
important reason to seek unpublished literature is because of an
the Bothe, Davidow, Bramlett, Franic, and Ingham (2006)
accepted general tendency for positive results to be published
article in several larger senses, or to review our review. Their
and for negative results not to be. Drawing the conclusion
efforts in this area raise several interesting issues.
from published literature that a treatment is effective may be
Previous summaries. Meline and Harn noted, to begin
problematic, therefore, if there is important information to the
with, that the review’s conclusions were more negative than
contrary in unpublished sources. This problem did not occur for
the conclusions reached in some previous reviews of related
the Bothe et al. review, given our conclusion of the relative
literature. As they also noted, we did address this question
ineffectiveness of the pharmacological agents.
in our original article, which also discussed part of the an-
Several other lines of reasoning also suggest an emphasis
swer: It has become relatively well established that more
on published literature; it is available to all readers, for ex-
stringently conducted clinical research often leads to more
ample, and it has already passed the test of peer review. The
negative results than those obtained in smaller or less con-
decision to concentrate on published literature also draws
trolled studies. Because our review incorporated trial quality
an objective line that avoids the complex problems inherent
assessment, and based its results on those articles that met
in deciding which unpublished literature to include and which
a trial quality criterion, its conclusions could be predicted to
to exclude, because as reasonable as the inclusion of some
be more negative than the conclusions reached in less selec-
unpublished information might be, the decision to include all
tive reviews. We also must point out that our conclusions
unpublished information about a given topic will usually
did not differ from all previous conclusions. Meline and
be unreasonable. Biddle, Watson, Hooper, Lohr, and Sutton
Harn quoted Ingham’s (1984, p. 413) suggestion, for ex-
(2002), as one example, in a relevant recent review, initially
ample, that some pharmacological agents “are undoubtedly
concentrated on published peer-reviewed information only,
reliable I for reducing state anxiety.” That particular quote
but then made the decision to include published but not peer-
was less relevant to the issues at hand, however, than the
reviewed test manuals as another source of information for
same chapter’s succinct and quite negative summary a few
some specific reasons. While other lines can be defensible
pages later: “For the present, it is safe to conclude there is
and could certainly serve legitimate purposes, in summary, it
little evidence that stuttering can be usefully treated by drug
is not an error to draw a line between published peer-reviewed
therapy” (Ingham, 1984, p. 424). In fact, if we have learned
research and other types of information.
anything from the drug treatments of stuttering, especially
Outcome criteria. Meline and Harn discussed the pos-
investigations into the efficacy of anxiolytic drugs, it must
sibility that our use of a 5% criterion “biased” the con-
be that stuttering is neither inextricably nor functionally
clusions drawn because it is an “imprecise measure of
related to anxiety. There is simply no evidence that an-
treatment success or failure” (p. 94). While we certainly
xiolytics have helped to eliminate or reduce stuttering.
agree wholeheartedly with the latter, we do not believe
Search strategies, inclusion criteria, and references.
that selecting a binary outcome criterion, or using speech
Meline and Harn also raised some specific questions about
outcomes as one of two types of outcome criteria in a review
the databases and search strategies we had used, and they
of stuttering treatment, creates “bias” in a review. Quite
questioned the information we provided about the articles
the opposite: Such a method provides an objective (i.e.,
selected for review. The answer to most of their questions is
specifiable and repeatable) method through which a set of
simply that one must draw a line somewhere: We selected
articles can be divided into those that meet a relevant
outcome criterion and those that do not. There are other
sets of criteria (see Bothe, Davidow, Bramlett, & Ingham,
ways to create reviews of the literature, but such methods
as meta-analysis (combining results mathematically) or cal-
Meline and Harn, in a parallel fashion, sought to deter-
culation of effect sizes (as Meline and Harn discussed)
mine whether our review had been done “well enough.”
impose multiple restrictions on the articles that can be in-
Their efforts to define “well enough” led them to Oxman and
cluded (because of the need for individual data or variability
Guyatt’s (1988, p. 698, Table I) “guidelines for assessing
data) in a manner that we believed to be too restrictive for our
research reviews” and to Oxman et al.’s (1991, p. 92, Table 1)
purposes in reviewing stuttering treatment research.
and Oxman and Guyatt’s (1991, p. 1272, Table 1) related
It is also important to note that Meline and Harn (2008,
“criteria for assessing the scientific quality of research
p. 94) summarized our review as having required “the
overviews.” Their abstract stated that they had adopted
frequency of ‘stuttering’ to be reduced to 5% or less.”
“A. D. Oxman and G. H. Guyatt’s (1998) guidelines I and
This summary is misleading. We used a 5% criterion for
A. D. Oxman and G. H. Guyatt’s (1991) criteria” for their
any measure of disfluency, nonfluency, stuttering, or any
review of our article. The structure of their letter was not
reported variable; we investigated the alternative possibility
based on Oxman and Guyatt’s lists, however; instead, Meline
that any such measure had been reduced by half if the
and Harn seem to have structured their letter around a Web
absolute 5% criterion were not met; we identified all studies
site (Davies & Crombie, undated, cited by Meline & Harn
where pretreatment values had been below 5% to assess
at their p. 94) and around a previous Meline (2006) article.
them separately; and we also looked for any documented
Their intent, as described in their abstract, to use a “7-point
pretreatment to posttreatment change in any social, emotional,
scale from extensive flaws on the high end to minimal flaws
or cognitive variable. We certainly agree with Kazdin’s (2001,
on the low end” (Meline & Harn, 2008, p. 93) also does
p. 461) warning, quoted by Meline and Harn in this con-
not seem to appear explicitly in their letter, nor does such a
text, that “we cannot merely decide a criterion as a cutoff and
scale appear anywhere in the Oxman and Guyatt (1988,
assume those who have made the cutoff have improved.”
1991) or Oxman et al. (1991) articles.
Kazdin’s (2001) point is not that no criterion or cutoff should
The structure of their letter thus leads to some interesting
ever be used, however, only that one must also consider
observations. First, it is clearly possible to review a set of
the impact of treatment on quality of life or patient preferences
studies, or to review a review, using more than one set of
as well. In stuttering, one of the many possible variables
criteria, and identifying or establishing those criteria is not
that must be assessed in comparing treatment options is
necessarily straightforward. It is also possible, even given
whether they tend to result in improved speech; our review
the same criteria, to reach different conclusions. As an
did this. It is also reasonable to address other variables,
example, we must respectfully disagree with Meline and
and our review did that as well (as has some of our other
Harn’s (2008, p. 94) conclusion that our review “lacked
work; see Bramlett, Bothe, & Franic, 2006; Franic & Bothe,
transparency” in a way that created “major flaws.” We
specified the publication years, publication language, search
Sample size, external validity, and confounding variables.
strategy, and topics covered by our project; we published
Finally, Meline and Harn mentioned several problems with
references to all the articles that met our methodological
many clinical studies that they believed the Bothe et al. re-
inclusion criteria, which were the articles our review dis-
views should have emphasized. The problems discussed in
cussed; we offered to make lists of all the articles we had
this section of Meline and Harn’s letter are indeed problems
initially reviewed available to interested readers (and we have
with many studies. They were not addressed in detail in our
in fact since done so); and we described in detail the methods
original articles in part for reasons of space and in part because
and the results from the multiple reviewers who assessed
we sought to use a relatively short list of extremely basic
each article using a written data-extraction instrument. That
methodological criteria to assess all studies in an objective way.
it would have been possible to select different publicationyears, different publication languages, or different searchstrategies is not a problem of the transparency of our review;
it is simply a statement that other authors or researchers couldhave made other equally defensible decisions.
Criteria for the Criteria for the Criteria
One problem with assessing a review of a review, how-
The most intriguing feature of Meline and Harn’s letter
ever, is that these issues are in danger of becoming the old
may be that it parallels so closely several of the complex
joke represented visually by a person looking at a picture
issues we also faced in developing and reporting our sys-
of herself looking at a picture of herself looking at a picture
tematic review articles. The essential goal of our project,
of herself. The present authors enjoy these issues as aca-
which was a systematic review with trial quality assessment,
demic exercises, and we suspect Meline and Harn do as well.
was to identify those stuttering treatment research articles
In the particular case of pharmacological treatments for
that had been done “well enough” that we believed they
stuttering, we agree with Meline and Harn’s basic conclusion
could reasonably serve as a body of literature to be sum-
that “the existing evidence for the use of pharmacological
marized and potentially acted upon. Because there is no
agents with persons who stutter is insufficient to recommend
single definition of “well enough,” we turned to multiple
them in practice” (p. 95). We cannot agree with some of
previous sources and assembled five common methodolog-
the other implications in their final paragraph, such as that
ical requirements, and we also compared the results from
being “acceptable” to clients can be equated with “no un-
those criteria with the results obtained using two existing
desirable side effects.” Undesirable side effects are common
100 American Journal of Speech-Language Pathology • Vol. 17 • 98–101 • February 2008
and do not contraindicate the use of a drug, as long as the
Guitar, B. (2006). Stuttering: An integrated approach to its nature
benefits are viewed as outweighing the risks.
and treatment (3rd ed.). Philadelphia: Lippincott, Williams &
Such minor differences aside, we appreciate Meline and
Harn’s careful reading of our articles. We also appreciate
Ingham, R. J. (1984). Stuttering and behavior therapy: Cur-
rent status and experimental foundations. San Diego, CA:
their help in bringing the problems of olanzapine to light
for speech-language pathologists, and we appreciate the
Ingham, R. J., & Cordes, A. K. (1997). Self-measurement and
opportunity to address the larger issues raised by their ques-
evaluating stuttering treatment efficacy. In R. F. Curlee & G. S.
tions. We also suspect that Meline and Harn would agree
Siegel (Eds.), Nature and treatment of stuttering: New direc-
with us that all of us who have the luxury of expending
tions (Rev. ed., pp. 413–437). Boston: Allyn & Bacon.
energy on such exercises as reviewing reviews of reviews
Kazdin, A. E. (2001). Almost clinically significant ( p < .10):
of reviews also have the responsibility to ensure that our
Current measures may only approach clinical significance.
efforts lead, somehow, to improved assessment, treatment,
Clinical Psychology: Science and Practice, 8, 455–462.
and management methods for the people living with the
Kendall, D. M. (2005). Postprandial blood glucose in the man-
disorders we study. We hope that this exchange of letters
agement of Type 2 Diabetes: The emerging role of incretinmimetics. Medscape Diabetes & Endocrinology, 7(2), Retrieved
might represent some small progress in that direction.
August 30, 2007, from www.medscape.com/viewarticle/515693.
Liebl, A. (2003). Postprandial blood glucose as the driving
force in pathogenesis of type 2 diabetes. Medizinische Klinik,
Lilly settles with 18,000 over Zyprexa. (2007, January 7).
Ludlow, C. L. (2006). Neuropharmacology of stuttering: Concepts
and current findings. In N. B. Ratner & J. Tetnowski (Eds.), Cur-
rent issues in stuttering research and practice ( pp. 239–254).
Mahwah, NJ: Erlbaum.
Maguire, G. A., Riley, G. D., Franklin, D. L., Maguire, M. E.,
Nguyen, C. T., & Brojeni, P. H. (2004). Olanzapine in the
treatment of developmental stuttering: A double-blind, placebo-controlled trial. Annals of Clinical Psychiatry, 16, 63–67.
Meline, T. (2006). Selecting studies for systematic review: In-
clusion and exclusion criteria. Contemporary Issues in Com-munication Science and Disorders, 33, 21–27.
Biddle, A., Watson, L., Hooper, C., Lohr, K. N., & Sutton,
Meline, T., & Harn, W. E. (2008). Comments on Bothe, Davidow,
S. F. (2002). Criteria for determining disability in speech-
Bramlett, Franic, and Ingham (2006). American Journal of
language disorders (Evidence Report/Technology Assessment
Speech-Language Pathology, 17, 93–97.
No. 52; Prepared by the University of North Carolina Evidence-
Oxman, A. D., & Guyatt, G. H. (1988). Guidelines for reading
Based Practice Center under Contract No. 290-97-0011;
literature reviews. Canadian Medical Association Journal, 138,
AHRQ Publication No. 02-E010). Rockville, MD: Agency
for Healthcare Research and Quality.
Oxman, A. D., & Guyatt, G. H. (1991). Validation of an index of
Bothe, A. K., Davidow, J. H., Bramlett, R. E., Franic, D. M.,
the quality of review articles. Journal of Clinical Epidemiology,
& Ingham, R. J. (2006). Stuttering treatment research
1970–2005: II. Systematic review incorporating trial quality
Oxman, A. D., Guyatt, G. H., Singer, J., Goldsmith, C. H.,
assessment of pharmacological approaches. American Journal
Hutchison, B. G., Milner, R. A., & Streiner, D. L. (1991).
of Speech-Language Pathology, 15, 342–352.
Agreement among reviewers of review articles. Journal of
Bothe, A. K., Davidow, J. H., Bramlett, R. E., & Ingham,
R. J. (2006). Stuttering treatment research 1970–2005: I. Sys-
Sloan, J., Symonds, T., Vargas-Chanes, D., & Fridley, B.
tematic review incorporating trial quality assessment of behav-
(2003). Practical guidelines for assessing the clinical signifi-
ioral, cognitive, and related approaches. American Journal of
cance of health-related quality of life changes in clinical
Speech-Language Pathology, 15, 321–341.
trials. Drug Information Journal, 37, 23–31.
Bramlett, R. E., Bothe, A. K., & Franic, D. F. (2006). Using
U.S. Food and Drug Administration. (2004, March 1). Safety
preference based measures to assess quality of life in stuttering.
data on Zyprexa ® (olanzapine)—hyperglycemia and diabetes
Journal of Speech, Language, and Hearing Research, 49,
[Dear Doctor letter]. Retrieved January 8, 2007, from the U.S.
Food and Drug Administration MedWatch Web site: www.fda.
Cavalot, F., Petrelli, A., Traversa, M., Bonomo, K., Fiora, E.,
Conti, M., et al. (2006). Postprandial blood glucose is astronger predictor of cardiovascular events than fasting blood
glucose in type 2 diabetes mellitus, particularly in women: Lessons
from the San Luigi Gonzaga Diabetes Study. The Journal of
Clinical Endocrinology and Metabolism, 91, 813–819.
Franic, D. M., & Bothe, A. K. (2008). Psychometric evaluation
Contact author: Anne K. Bothe, Department of Communication
of condition-specific instruments used to assess health-related
Sciences and Special Education, 516 Aderhold Hall,
quality of life, attitudes, and related constructs in stuttering.
The University of Georgia, Athens, GA 30602.
American Journal of Speech-Language Pathology, 17, 60–80.
Z O/A v U Donnerstag, 5. Januar 2006 Gesundheit Spezial Aufklärung Abnehmen Wer Fett statt Muskeln abbauen will, sollte auf Blitzdiäten verzichten Bei 56 Prozent der 16- bis 35-jährigenFrauen, die schon Geschlechtsverkehrhatten, war das «erste Mal» spontan zu-bis zu 25 Jahren liegt dieser Anteil sogarnoch etwas höher. Rund jede fünfte Frauschlechtsverkehr nicht vor Aids und an
MODELO COMUNITARISTA VISTO CON LOS OJOS DE LA SOCIEDAD CIVIL Y EL ESTADO “Pero padre si Uds. los líderes se desinflan, nosotros, los de la base, peor”. Son las palabras de Martín un líder sin muchos títulos de un barrio sub-normal de una ciudad: la mía. A que lideres se refiere a los líderes del sector social y solidario, los investigadores sociales, maestros y estudiantes