close

Вход

Забыли?

вход по аккаунту

?

Low-dose prednisone in rheumatoid arthritis patientsPlacebo treatment.

код для вставкиСкачать
50 1
LETTERS
On blaming the placebo
To the Editor:
In a retrospective analysis of 5 trials of diseasemodifying antirheumatic drugs (DMARDs) versus placebo in
adults with rheumatoid arthritis (RA) (l), Paulus et a1 addressed the problem of “excessive” benefit attributable to
placebo obscuring the benefit of DMARD therapy. Interestingly, in another multicenter, multinational study reported in
the same issue of Arthritis and Rheumatism involving children with rheumatoid arthritis treated with auranofin or
placebo, the authors also attribute failure to demonstrate a
difference between DMARD and placebo to excess or unexpected activity of the placebo (2).
Paulus and colleagues set out to find a group of
outcome variables and a degree of improvement in those
variables that occurs in 5 5 % of the 130 placebo-treated
participants in 4 of their published studies. Using 6 outcome
measures selected “arbitrarily” from those 4 studies, they
developed a post hoc composite index that provides the
desired 5 5 % response for the placebo treatment group.
They then apply this discriminant function to data from a
different CSSRD study, the initial analysis of which showed
placebo to be as effective as either parenteral gold or
sulfasalazine in the treatment of RA in adults (3). Using an
intent-to-treat orientation, which we believe to be most
appropriate, this new index again shows there is no difference between gold sodium thiomalate (26% improved) and
placebo (24% improved) (see their Table 6, last line). Despite
this elaborate reworking of their data with multiple, arbitrary
decisions to maximize the post hoc difference between
placebo and DMARDs, they still are unwilling to conclude
that it is likely that parenteral gold does not benefit the
patient any more than does a placebo.
It is striking that in the same issue of Arthritis and
Rheumatism, 2 well-done, randomized, double-blind studies
find that drugs commonly prescribed by rheumatologists to
treat both adults and children with RA are ineffective when
compared with placebo. Paulus and colleagues suggest that
this “problem” can be resolved by redefining the outcome
criteria and then reanalyzing the data, yet, their analysis
using the new criteria also finds the gold to be no better than
placebo.
We remain unconvinced that failure to demonstrate a
significant difference between a DMARD and a placebo
effect in the several trials carried out by the Cooperative
Systematic Studies of the Rheumatic Diseases Group is due
to overactivity of the placebo. Perhaps the old saying “If it
ain’t broke don’t fix it” applies in this situation.
Wallace V. Epstein, MD
Curtis J. Henke, PhD
University of California San Francisco
Sun Francisco, C A
I Paulus HE. Egger MJ, Ward JR, Williams HJ, and the Cooperative Systematic Studies of Rheumatic Diseases Group: Analysis
of improvement in individual rheumatoid arthritis patients
treated with disease-modifying antirheumatic drugs, based on the
findings in patients treated with placebo. Arthritis Rheum 33:477484, 1990
2. Giannini EH, Brewer EJ Jr, Kuzmina N, Shaikov A, Wallin B,
for the USA Pediatric Rheumatology Collaborative Study Group
and the USSR Cooperative Children’s Study Group: Auranofin in
the treatment of juvenile rheumatoid arthritis: results of the
USA-USSR double-blind, placebo-controlled trial. Arthritis
Rheum 33:46&476, 1990
3. Williams HJ, Ward JR, Dahl SL, Clegg DO, Willkens RF,
Oglesby T, Weisman MH, Schlegel S , Michaels RM, Luggen
ME, Polisson RP, Singer JZ, Kantor SM, Shiroky JB, Small RE,
Gomez MJ, Reading JC, Egger MJ: A controlled trial comparing
sulfasalazine, gold sodium thiomalate, and placebo in rheumatoid
arthritis. Arthritis Rheum 31:702-713, 1988
Low-dose prednisone in rheumatoid arthritis patients:
placebo treatment?
To the Editor:
I read with interest the article by Paulus and coworkers regarding a proposed index for determining improvement
of individual rheumatoid arthritis (RA) patients treated with
disease-modifying drugs (DMARDs) (Paulus HE, Egger MJ,
Ward JR, Williams HJ, and the Cooperative Systematic
Studies of Rheumatic Diseases group: Analysis of improvement in individual rheumatoid arthritis patients treated with
disease-modifying antirheumatic drugs, based on the findings in patients treated with placebo. Arthritis Rheum 33:
477-484, 1990). The authors reasoned that the responses of
the placebo-treated patients in controlled clinical trials of
DMARDs represent the background progression of their
disease. From this, they reanalyzed the data from controlled
trials performed by their group to develop a criterion for
improvement in individual patients that could be met by no
more than 5% of placebo-treated patients.
Although their analysis yielded interesting data, I
have problems with the composition of the placebo treatment groups. Specifically, I object to the inclusion of patients treated with corticosteroids. Using the numbers
quoted in the original publications, patients taking placebo
were treated with corticosteroids in 18 of 52 in the Dpenicillamine trial, 8 of 46 in the gold trial, and 59 of 94 in the
methotrexate trial. Together, 85 of 192, or 44%, of the
patients taking placebo were also taking corticosteroids.
Overall, a similar number of patients in the active drug
treatment groups received corticosteroids. The study criteria
were uniform, in that the dosage was not allowed to exceed
the equivalent of 10 mg/day of prednisone, had to be stable
for at least 1 month prior to the start of the trial, and had to
be maintained constant throughout the trial. In 1 trial, the
average dosage of prednisone was 6.9 mg/day.
The efficacy of such dosages of corticosteroids in the
treatment of RA is not known, although there is evidence to
support a disease-modifying effect at higher doses (for a
recent review, see Weiss MM: Corticosteroids in rheumatoid arthritis. Semin Arthritis Rheum 19:9-21, 1989). However, the fact that the corticosteroids were not withdrawn
prior to entry into the study indicates that the investigators
believed the drug was exerting some effect. It follows that in
corticosteroid-treated patients, the trials analyzed compared
the effect of DMARDs with the effect of placebo superimposed on the effect of a stable dosage of corticosteroids.
Thus, the data as published may indicate the background
502
LETTERS
progression of disease in a heterogeneous population,
roughly half of which was treated with corticosteroids. I
would be very interested to see the results of an analysis
excluding corticosteroid-treated patients.
Maarten Boers, MD, PhD
University Hospital
Maastricht, The Netherlands
Composite index methodology
To the Editor:
The paper by Paulus and coworkers was welcome for
its attempt to provide a composite index to be used in trials
of disease-modifying antirheumatic drugs (Paulus HE, Egger
MJ, Ward JR, Williams HJ, and the Cooperative Systematic
Studies of the Rheumatic Diseases Group: Analysis of
improvement in individual rheumatoid arthritis patients
treated with disease-modifying antirheumatic drugs, based
on the findings in patients treated with placebo. Arthritis
Rheum 33:477484, 1990). We would be interested in the
authors’ response to the following questions.
First, they selected their criterion for improvement
(320% change in 2 4 variables) by inspecting combinations
of different levels of percentage of change and different
numbers of variables. They chose the combination which
gave the greatest difference between placebo-treated and
active drug-treated patients in 3 trials. The percentage of
placebo-treated patients who improved was much greater in
the trial of oral gold. The baseline characteristics of the
placebo-treated patients in the oral gold trial differed significantly from those in the other 2 trials, even with a Bonferroni correction. The placebo-treated patients in the trial of oral
gold had a shorter duration of disease, a lower joint swelling
score, etc. It appears that the ability of the selected criterion for
improvement to discriminate the response to placebo versus
active drug may depend on the baseline severity of disease.
Combining disparate placebo-treated patient groups therefore
diminishes the validity of the criterion.
Since these trials were conducted at different times,
a second question arises: Are these completely different
patients in all the trials? With multiple participating centers
and the probable use of study identification numbers, are the
authors confident that the same patients were not used in
more than one trial?
Margaret G. E. Peterson, PhD
Pamela G. Williams, MD, MPH
Cornell Arthritis and Musculoskeletal Disease Center
New York, N Y
Reply
To the Editor:
We appreciate the interest of Drs. Felson and
Meenan, Epstein and Henke, Boers, and Peterson and
Williams and wish to respond to their thoughtful comments.
Clinicians are required to treat patients one at a time
and find that, frequently, the responses of individual patients
differ markedly from the reported, statistically adjusted,
mean group responses to the same treatments. Indeed, even
for the same individual, some outcome measures may indicate improvement while others do not. We would like to be
able to identify the favorable responders before starting the
treatment; unable to do that, we would like to provide the
patient with a reasonable estimate of the probability that he
or she will improve on undergoing the proposed treatment.
Since “improvement” is a binary, yes or no, decision, some standard must be established by which to decide
whether a given patient, among a group of patients exhibiting
a wide spectrum of responses, has improved; using the same
group of subjects, the number declared “improved” will
vary depending on the standard that is selected. We sought
a measure of global, whole-body improvement sufficiently
conservative that only 5% of placebo-treated patients would
respond. This definition assumes that a few placebo-treated
patients (continuing background nonsteroidal antiinflammatory drugs [NSAIDs] and/or corticosteroids) will experience
major spontaneous improvement that is similar to that desired from treatment with an effective disease-modifying
antirheumatic drug (DMARD), but that the clinical course of
most patients will be less than desirable. Thus, our goal was
to develop a quantitative standard for the qualitatively
important, but ill-defined, term “clinically significant improvement,” and to apply it to individual patients. The
proposed outcome was also to be defined in terms of
categories, so that it could be easily interpretable clinically.
Felson and Meenan are correct that logistic regression and discriminant analysis have optimal properties in
their class with respect to discriminating between 2 groups,
and can be used retrospectively to determine the minimum
number of outcome measures needed (1). Recursive partitioning is another such option ( 2 ) . However, analytically
sophisticated discrimination between “active drug” groups
and placebo groups from randomized, double-blind clinical
trials can be a seductive way to lose track of the fact that
there is no gold standard for improvement in rheumatoid
arthritis. Problems associated with the use of a proxy for
improvement, such as membership in an “active drug”
group as opposed to a “placebo” group have been discussed
by others (3). Therefore, our choice was to avoid statistical
modeling methods which required a gold standard and
simply to graph what the data tell us in several clinical trials
from the same cooperative group, utilizing clinically interpretable combinations of familiar outcomes and the requirement that very few placebo-treated patients be defined as
responders.
Unfortunately, if one attempts to classify individuals
as improved or not improved by using the single “best”
treatment group discriminator, an unacceptably large number of individuals will be misclassified by chance, because of
the substantial methodologic variability (and large standard
deviation) inherent in all of our clinical measurements of
rheumatoid arthritis (RA). In group analyses, this measurement variability is overcome by increasing the number of
subjects. When making decisions about individual subjects,
it can be overcome by increasing the number of (somewhat
related) clinical measurements analyzed; requirement of
simultaneous improvement in several relatively imprecise
clinically based response variables would decrease the proportion of individuals falsely judged to be improved. Hence,
Документ
Категория
Без категории
Просмотров
0
Размер файла
263 Кб
Теги
prednisone, treatment, dose, low, arthritis, rheumatoid, patientsplacebo
1/--страниц
Пожаловаться на содержимое документа