close

Вход

Забыли?

вход по аккаунту

?

Outcome assessment in clinical trials evidence for the sensitivity of a health status measure.

код для вставкиСкачать
1344
OUTCOME ASSESSMENT IN CLINICAL TRIALS
Evidence for the Sensitivity of a Health Status Measure
ROBERT F. MEENAN, JENNIFER J. ANDERSON, LEWIS E. KAZIS, MARLENE J. EGGER,
MARY ALTZ-SMITH, CECIL 0. SAMUELSON, JR., ROBERT F. WILLKENS,
MARILYN A. SOLSKY, STANLEY P. HAYES, KENNETH L. BLOCKA, ARTHUR WEINSTEIN,
MARIA GUTTADAURIA, STANLEY B. KAPLAN, and JOHN KLIPPEL
Health status measures are conceptually relevant
to the assessment of clinical outcome in the rheumatic
diseases, but their ability to detect meaningful changes
in health has not been clearly demonstrated. This report
describes the performance of a self-administered health
status questionnaire in a randomized, double-blind, 21week comparison of placebo, oral gold, and injectable
gold in rheumatoid arthritis patients. Outcome was
assessed by standard clinical measures, including joint
count, grip strength, and laboratory tests, and by the
Arthritis Impact Measurement Scales, a reliable and
valid health status measure that assesses physical dis-
From the Multipurpose Arthritis Center, Boston University
School of Medicine, Boston, Massachusetts and the Cooperative
Systematic Studies of Rheumatic Diseases Program, University of
Utah School of Medicine, Salt Lake City.
Supported by contract AM-6-22 18, Multipurpose Arthritis
Center grant AM-20613, and training grant AM-07014 from the
National Institute of Arthritis, Diabetes, Digestive and Kidney
Diseases.
Robert F. Meenan, MD, MPH, Jennifer J. Anderson, PhD,
and Lewis E. Kazis, ScD: Boston University Multipurpose Arthritis
Center; Marlene J . Egger, PhD: Cooperative Systematic Studies of
Rheumatic Diseases Program; Mary Altz-Smith, MD: University of
Alabama in Birmingham; Cecil 0. Samuelson,.Jr., MD: University
of Utah; Robert F. Willkens, MD: Harborview Medical Center,
Seattle, Washington; Marilyn A. Solsky, MD: University of Michigan; Stanley P. Hayes, MD: University of Missouri-Columbia;
Kenneth L. Blocka, MD: University of California, Los Angeles;
Arthur Weinstein. MD: University of Connecticut; Maria Guttadauria, MD: State University of New York. Downstate Medical
Center; Stanley B. Kaplan, MD: University of Tennessee; John
Klippel, MD: National Institutes of Health.
Address reprint requests to Robert F. Meenan, MD, MPH,
Multipurpose Arthritis Center, Boston University School of Medicine, K-123, 80 East Concord Street, Boston, MA 02118.
Submitted for publication November 14, 1983; accepted in
revised form May 21, 1984.
Arthritis and Rheumatism, Vol. 27, No. 12 (December 1984)
ability, psychological status, and pain. Data from the
clinical and health status measures produced highly
similar conclusions: injectable and oral gold are more
effective than placebo for rheumatoid arthritis, and
injections are slightly more effective than oral gold. The
health status measure was thus quite sensitive to clinically meaningful drug-induced improvements. These findings provide justification for the further application of
health status measures to clinical trials of chronic disease.
The ultimate goal of medicine is to improve the
health of individuals. Unfortunately, it is not always
easy to determine whether a particular treatment substantially advances that goal. The clinical trial has
been developed as a method to determine, in a more
scientific fashion than simple clinical observation,
whether a therapy is significantly more effective than
some alternative therapy or a placebo (1). In the
theory and practice of the clinical trial, much attention
has been paid to such issues as sample size, pertinent
control groups, probabilities of Type I and Type I1
errors, and other matters of epidemiologic and statistical importance (2-4). All too often, however, the
broader, critical issues of how improved health
should be defined and measured are given very little
attention ( 5 ) .
Estimates of improvement have traditionally
focused on “objective” measures of disease activity,
such as physical examinations and laboratory tests.
Although such measures have gained acceptance
through longstanding use, they are often conceptually
and empirically limited in terms of their ability to
measure improvements in health. They fail to estimate
the impact of disease or therapy on the functional and
OUTCOME ASSESSMENT
psychological aspects of health, and their reliability
and validity may be surprisingly poor (6-8).
In the last few years, there has been a growing
interest in attempting t o measure health status more
broadly (9,lO). Instruments have been developed t o
estimate physical function and other aspects of health
status in community populations, general patient populations, and specific disease populations (1 1-13). This
latter trend has been particularly notable in rheumatology, where a number of health status instruments have
recently been developed, primarily in response t o the
perceived need t o measure health more comprehensively in patients who have chronic, pervasive diseases such as rheumatoid arthritis (14-17). T h e new
health status measures are conceptually relevant t o the
measurement of improved health outcomes, and extensive methodologic work has shown that these selfadministered questionnaires are highly reliable and
valid. Their reliability and validity, in fact, equal or
exceed that of many of the so-called objective measures that are traditionally utilized (18).
The potential applicability of such measures t o
clinical trials is becoming more widely recognized. In a
recent international conference on outcome measures
in rheumatology trials, participants rated health status
measures such as self-care, physical activity, pain, and
social role performance very highly in terms of their
potential utility for clinical drug studies (19). The
major question that needs t o be answered before such
health status measures can be more widely adopted for
clinical trials is whether they are sensitive enough to
measure the differences in outcome that might be
expected to occur, given the duration of the trial and
the relative effectiveness of the treatments (20).
In the present study, we address this question
by directly comparing the sensitivity of a specific
health status measure with that of standard clinical
outcome measures in a double-blind, multicenter clinical drug trial. Our results document the sensitivity of
the health status measure and lend support t o the
further application of such measures t o clinical trials of
chronic disease.
PATIENTS AND METHODS
Study design. This study was conducted in conjunction with a 2 I-week, double-blind, multicenter, parallel trial
comparing placebo, oral gold (auranofin), and injectable gold
(gold sodium thiomalate, GST) in patients with active rheumatoid arthritis (RA). Injectable gold is an established
treatment, having been shown in numerous trials to be
effective in controlling the inflammation of RA (21), while
oral gold is a relatively new drug that was undergoing premarketing evaluation at the time of this study.
1345
A comprehensive description of the study design is
provided elsewhere (22). In brief, the trial was conducted at
10 university rheumatology centers under the auspices of the
NIH-sponsored Cooperative Systematic Studies of Rheumatic Diseases project. A minimum of 20 patients within
each clinic were randomly assigned to receive placebo (20%
of subjects), oral gold (40%), or injectable gold (40%). All
patients were 18 years of age or older, had definite or classic
RA by American Rheumatism Association criteria (23), and
had been treated unsuccessfully with therapeutic doses of
salicylates and at least 1 other nonsteroidal antiinflammatory
drug. All patients had 6 or more swollen joints considered
capable of responding to drug therapy and 2 of the following:
1) 9 or more joints tender on pressure; 2) 45 minutes or more
of morning stiffness; and 3) Westergren erythrocyte sedimentation rate (ESR) of 28 mm/hour or greater.
All patients received weekly supplies of tablets and
were given weekly injections. Those in the placebo group
received weekly placebo injections and daily placebo tablets. Those in the oral gold group received 3.0 mg of
auranofin twice daily and weekly placebo injections. Those
in the injectable gold group received a placebo tablet twice
daily and weekly gold injections (50 mg weekly following 2
test doses). Subjects continued taking stable doses of aspirin
andlor another nonsteroidal antiinflammatory drug and,
when applicable, stable doses of prednisone for the duration
of the trial. Subjects and study physicians remained “blinded” as to treatment assignment until all participants at a
particular clinic had completed the trial.
Measures of response. Clinical assessments were performed at 0, 4, 8, 12, 16, and 21 weeks, with the same
physician assessing a given patient throughout the trial.
Physicians’ assessments included: 1) joint count for tenderness on pressure and/or pain on motion for 68 diarthrodial
joints, graded on a scale of 0 = none, 1 = positive response
on questioning, 2 = spontaneous response elicited, 3 =
withdrawal by patient on examination; 2) joint swelling for
66 joints graded on a scale of 0 = none, 1 = detectable
synovial thickening without loss of bony contours, 2 = loss
of distinctness of bony contours, 3 = bulging synovial
proliferation with cystic characteristics; 3) joint tenderness1
pain and joint swelling scores derived by summing the
severity scores for each joint; 4) grip strength using a
mercury column sphygmomanometer with a standard grip
bag; 5) physician assessment of disease activity on a scale of
I = asymptomatic, 2 = mild, 3 = moderate, 4 = severe, 5 =
very severe; and 6) functional class. Patients also provided
an overall assessment of arthritis activity graded as 1 = very
good, 2 = good, 3 = fair, 4 = poor, 5 = very poor.
The laboratory studies included a weekly urinalysis
and complete blood count. Multi-test chemical surveys were
performed and Westergren ESR and serum salicylate levels
recorded at entry and at 4-week intervals. Rheumatoid factor
and antinuclear antibodies (ANA) were measured at entry
and at the time of completion or withdrawal.
The Arthritis Impact Measurement Scales (AIMS)
questionnaire was used as the health status measure. AIMS
is a 67-item instrument which has been thoroughly evaluated
for reliability and validity (24). Its content and approach are
patterned closely after the health status instruments developed by Brook and colleagues at the Rand Corporation (12).
It can be easily completed in 10-15 minutes and is entirely
MEENAN ET AL
1346
Table 1. Initial scores for clinical measures illid health status scales, by treatment group
Clinical measures'
Tenderness
score
No. tender
joints
Swelling
score
No. swollen
joints
Mean grip
strength
(right)
Mean grip
strength
(left)
Functional
class
Physician
assessment
Patient
assessment
Health status scales#
Mobility
Physical
activity
Dexterity
Activities
of daily
living
Household
activity
Anxiety
Depression
Pain
Social
activity
Arthritis
impact
*n
=
n
$n
=
=
Gold
shots$
Placebo*
Oral goldt
43.9 2 3.7
46.8 f 3.6
45.5
f
4.4
0.877
28.7
2.2
31.2
f
1.9
30.0
f
2.2
0.712
27.5 f 2.1
32.2
f 2.1
29.9
f
3.0
0.408
1.3
23.9
5
1.5
21.8
5
1.7
0.218
20.0
5
4
F-test§
121.4 ? 8.9
123.9 5 8.0
115.7 ? 8.6
0.773
112.8 f 8.9
114.3 5 7.5
116.5 f 8.7
0.953
2.1
f
0.1
2.1
f
0.1
2.2
f
0.1
0.898
2.9
f
0.1
3.0
f
0.1
2.9
f
0.1
0.810
3.1 f 0.1
2.9
* 0.1
2.8
f
0.1
0.162
2.5
0.5
1.6
5
0.4
1.9 f 0.5
0.408
7.3 2 0.4
6.4 t- 0.6
6.2
5.7
5
0.3
6.9
4.2
5
f 0.5
f
0.4
0.6
0.069
0.017
1.1 2 0.3
1.1 5 0.2
1.2
f
0.3
0.931
1.4 f 0.2
4.7 f 0.3
3.8 f 0.3
7.5 f 0.3
1.3 f 0.2
3.9 f 0.3
3.0 f 0.3
7.2 0.2
1.4 f 0.2
3.3 f 0.3
2.6 f 0.3
6.3 5 0.3
0.941
0.012
0.045
0.013
4.9
f 0.3
4.6 t- 0.3
4.3
0.3
0.434
5.2
f
5.3 2 0.4
0.410
5
0.4
*
4.7
f
0.3
5
43 for clinical measures; n = 34 for health status scales.
64 for clinical measures; n = 50 for health status scales.
54 for clinical measures; n = 37 for health status scales.
§ P values for analysis of variance F-test for dserences between groups in initial scores.
ll Mean ? SEM; see text for description of scoring.
# Mean f SEM; scores are all on a 0 (well) to 10 (ill) range.
self-administered. The AIMS instrument was completed by
patients at the time of entry and once again at the time of
study completion or withdrawal.
The AIMS instrument consists of 9 scales which
assess the following aspects of health status: Mobilityability to move about the community; Physical Activitylower extremity function; Dexterity-hand function; Household Activities-routine household tasks; Activities of Daily
Living-basic self-care tasks; Social Activity-interaction
with friends and family; Anxiety; Depression; and Pain.
Previous work has demonstrated that these 9 scales can be
summarized into 3 broad components of arthritis health
status: Physical Disability (including Mobility, Physical Activity, Dexterity, Household Activities, and Activities of
Daily Living), Psychological Status (including Depression
and Anxiety), and Pain (25). In addition, the AIMS instrument contains a 10-cm visual analog scale that assesses
overall arthritis impact. All AIMS scale and component
scores are standardized to a 0-10 range where higher numbers represent poorer health status.
Statistical analysis. Data were analyzed using differences between the initial and final values (change scores) for
all variables. A linear covariance adjustment was applied to
change scores to adjust for any differences between the
treatment groups in the initial value of the measures. Group
comparison analyses were performed on the adjusted change
scores (26). Analysis of variance F-tests of ranked data were
used to explore for any differences between the change
1347
OUTCOME ASSESSMENT
Table 2. Change scores in clinical measures, by treatment group
Treatment group
Clinical measure
P*
Placebo
(n = 43)
Oral gold
(n = 64)
-2.65t
-15.03
-19.70
0.001
0.003
0.001
0.439
0.49
-4.09
-8.48
-7.51
-11.35
-11.40
0.001
0.022
0.001
0.059
0.001
0.004
0.617
0.534
-2.21
-4.00
-7.39
0.068
0.032
0.006
0.336
Tenderness score
No. tender
joints
Swelling score
No. swollen
joints
Mean grip strength
(right)
Mean grip strength
(left)
Functional class
Physician assessment
Patient assessment
Gold shots
Any
Placebo Placebo Shots vs.
(n = 54) difference vs. oral vs. shots
oral
17.51
17.53
22.22
0.825
17.22
-0.07
-0.12
-0.33
20.13
-0.17
-0.47
-0.44
21.82
-0.19
-0.60
-0.50
0.518
0.399
0.015
0.727
* A preliminary test for differences between the groups in initial scores was negative for each variable.
P values are shown for analysis of variance F-tests on ranked change scores. If this result was
significant ( P < 0.05), pairwise group comparisons were performed using the Mann-Whitney test.
t Minus sign indicates lower mean score at end of trial; no sign indicates higher score.
scores of the 3 groups. If the results of the overall tests
revealed any significant difference (P < 0.05) among the 3
groups, then pairwise comparisons were performed using
Mann-Whitney tests or, equivalently, t-tests on ranked data
(27). Since multiple comparisons were involved at this step,
the exact P values are reported.
Two methods were used to compare the clinical and
health status change scores more directly. Correlations between the 2 sets of variables were calculated for the subset of
121 subjects with complete data. On the basis of a principal
iterations factor analysis performed on the initial scores,
clinical measure changes were grouped into 3 factors: joint
tenderness, which combined tenderness score and count; joint
swelling, which combined swelling score and count; and grip
strength, which represented an average of left and right grip
strength changes. Pearson correlations were estimated after
standardizing both sets of measures within treatment groups to
means of 0 and standard deviations of 1, to adjust for d8eYent
degrees of improvement across the 3 groups. In the second
method of comparison, discriminant analyses were run using
the standardized change scores of clinical factors and AIMS
dimensions as the discriminating variables, and drug treatments as the groups. This approach addressed the question of
how accurately the 2 sets of measurements could distinguish
between different drug groups, by correctly assigning subjects
to their actual treatment.
Table 3. Change scores in laboratory measures, by treatment group
~~
~
P*
Treatment group
Laboratory
measure
Hemoglobin
(< 12 gm/dl)
at first visit)
Platelets
0 4 5 0 x I03/mm3
i t first visit)
Oral
gold
Placebo
0.33i
= 12)
(n
=
Placebo
vs. oral
Placebo
vs.
shots
Shots vs.
oral
12.5
0.023
0.500
0.029
0.011
0.048
0.889
0.023
0.034
0.001
0.002
0.001
0.056
(n = 10)
12)
-80
(n = 16)
(n = 9)
35)
(n
-24
=
Any
difference
0.06
= 21)
(n
(n
Gold
shots
(n
-179
ESRS
(>28 mm/hour
at first visit)
-3
- 22
=
35)
- 24
(n = 37)
* A preliminary test for differences between the groups in initial scores was negative For each variable.
P values are shown for analysis of variance F-test on ranked change scores. If this result was
significant ( P < 0.05), pairwise group comparisons were performed using the Mann-Whitney test.
t Minus sign indicates lower mean score at end of trial; no sign indicates higher score.
$ Erythrocyte sedimentation rate.
MEENAN ET AL
1348
RESULTS
comparable in terms of their initial mean scores on the
standard clinical measures (Table 1). There were differences between treatment groups, however, in terms
of initial health status scores. The mean health status
change scores were adjusted by using the initial health
status score as a covariate so that the groups would be
comparable.
A change score for each of the outcome variables was estimated for each patient by subtracting the
initial value from the final value, and the mean change
score for each treatment group was then calculated.
Laboratory change scores were calculated only for the
subset whose initial values were abnormal. The
change scores for the clinical and laboratory variables
are shown in Tables 2 and 3, respectively. For all
variables except grip strength and hemoglobin, change
scores with negative values represent improvement.
An overall test for differences between the
change scores of the 3 treatment groups was significant for the number and score of tender joints, the
overall swelling score, and the physician assessment.
In pairwise comparisons on all 4 of these variables,
injectable gold was found to be significantly better
than placebo. Oral gold was better than placebo in 3 of
the 4 cases, and there was no detectable difference
between the 2 active treatments on any clinical variable.
The change scores for the laboratory variables
demonstrated differences between groups for hemoglobin, platelet count, and ESR. In all 3 cases, inject-
Two hundred eight patients fulfilled the entry
criteria. One hundred ninety-three were considered
acceptable study patients since 15 were withdrawn for
non-drug-related reasons such as lack of cooperation,
incorrect diagnosis, moved from clinic area, or intercurrent illness. Four patients receiving placebo, 5
receiving oral gold, and 6 receiving injectable gold
were excluded for the above reasons. One hundred
sixty-one completed at least 20 weeks of treatment and
formed the basis of the efficacy analysis. Dropout
rates were higher in the injection group (34%) than in
the placebo (14%) or oral gold (17%) groups due to a
higher incidence of adverse drug reactions associated
with GST.
Complete clinical data were available on all 161
subjects, including 43 subjects receiving placebo, 64
receiving oral gold, and 54 receiving injectable gold.
Entry and exit AIMS data were also available on 121
subjects of this group (75%), including 34 (79%) of the
placebo subjects, 50 (78%) of the oral gold group, and
37 (69%) of the injectable gold group. The lower
capture rate of AIMS data was primarily due to
administrative oversight and/or personnel changes at
the participating clinics.
Subjects in the 3 treatment groups were comparable at study entry in demographic characteristics
including age, sex, and race, and also in measures of
disease severity and duration. The 3 groups were also
Table 4.
Adjusted change scores in health status scales, by treatment group
Treatment group
P*
Health status
scales
Mobility
Physical activity
Dexterity
Activities of daily
living
Household activity
Anxiety
Depression
Pain
Social activity
Arthritis impact
(n = 34)
Oral gold
(n = 50)
Gold
shots
(n = 37)
-0.03t
-0.43
-0.33
-0.21
-0.53
-0.67
-0.59
-1.45
-1.13
0.659
0.005
0.308
0.01
0.31
0.38
0.02
-0.31
0.07
-0.31
-0.24
0.02
-0.55
-0.61
-1.54
-0.15
-1.21
-0.67
-0.04
-0.67
-0.92
-2.72
-0.50
-2.02
0.114
0.324
0.019
0.020
Placebo
Any
Placebo Placebo Shots vs.
difference vs. oral vs. shots
oral
0.738
0.012
0.002
0.0001
0.018
0.060
0.002
0.009
0.006
0.0001
0.594
0.217
0.005
0.254
0.001
0.020
0.001
0.033
* Change scores have been adjusted by regression on initial level since there were differences between
the treatment groups in initial value of some of the health status measures. The overall test for any
difference between the 3 groups was an analysis of variance F-test on ranked data after linear
adjustment for initial score. If this was significant (P < 0.05). pairwise group comparisons were
performed using t-tests on adjusted means of the ranked data.
t Minus sign indicates lower mean score at end of trial; no sign indicates higher score.
OUTCOME ASSESSMENT
1349
Table 5. Adjusted change scores in health status components, by treatment group
P*
Treatment group
Health status
component
Physical
Placebo
Oral
gold
-0.295t
= 27)
0.246
(n = 31)
-0.31
(n = 34)
-0.282
= 44)
-0.579
(n = 50)
- 1.54
(n = 50)
(n
Psychological
Pain
(n
Gold
shots
-0.658
= 30)
-0.781
(n = 35)
-2.72
Any difference
Placebo
vs. oral
Placebo
vs. shots
Shots
vs. oral
0.368
(n
(n
=
0.013
0.025
0.004
0.331
0.o001
0.002
o.Ooo1
0.005
35)
* Change scores have been adjusted by regression on initial level since there were differences between
the treatment groups in initial value of some of the health status measures. The overall test for any
difference between the 3 groups was an analysis of variance F-test on ranked data after linear
adjustment for initial score. If this was significant ( P < 0.05). pairwise group comparisons were
performed using !-tests on adjusted means of the ranked data.
t Minus sign indicates lower mean score at end of trial; no sign indicates higher score.
able gold was significantly more effective than both
placebo and oral gold. Oral gold was more effective
than placebo only in the case of lowering the sedimentation rate. There were no differences in rheumatoid
factor or ANA titer. The greater efficacy of injectable
gold can also be seen in the fact that it produced the
greatest improvement in all 9 of the clinical variables
and all 3 of the laboratory tests.
To ensure that no bias resulted from the failure
to obtain post-treatment AIMS data on some subjects,
the analyses in Tables 2 and 3 were repeated using the
clinical and laboratory measures from the subset of
121 subjects who had completed both pre- and posttreatment questionnaires. The results were very similar, with the exception that the swelling score and the
physician assessment variables no longer showed statistically significant differences across the 3 treatment
groups. This change was caused more by a loss of
statistical power than by any major shift in the mean
change scores for each treatment group, suggesting
that the clinical and laboratory results for the larger
group (n = 161) are generally valid.
The change scores for the 9 health status scales
and the 3 health status components are shown in
Tables 4 and 5 , respectively. Change scores with
negative values represent improvement in any AIMS
scale or component. There were significant differences
between treatment groups in Physical Activity, Anxiety, Depression, Pain, and Arthritis Impact. In each
case, the pairwise comparisons showed that injectable
gold was significantly more effective than placebo.
Oral gold was more effective than placebo in 3 of the 5
measures. When the 2 active treatments were compared, injectable gold was more effective than the oral
treatment in 3 of the 5 health status scales.
In terms of the health status component measures, the treatments differed significantly in the improvements in pain and psychological status that they
produced. Both injectable and oral gold were more
effective than placebo for these 2 components, and
injections were more effective than oral gold in terms
of producing pain relief. Once again, the finding that
injectable gold was the most effective of the 3 treatments was supported by the fact that it produced the
largest improvement scores for all 10 health status
scales as well as each of the 3 health status components.
The change scores produced by the clinical and
health status measures were then compared in the
subset of 121 patients for whom AIMS data were
available, using correlation coefficients and discriminant analysis. The results of the correlation analysis
are shown in Table 6. The physical dimension of health
status was significantly correlated with the 3 clinical
measures, all of which have a physical orientation.
The correlations of these change measures with
Table 6 . Correlation matrix of change scores for grouped clinical
measures and health status dimensions (n = 121)*
Grouped clinical measure
Health status
dimension
Joint
tenderness
Physical
Ps ycholog~cal
Pain
0.33:
0.22i
0.19f
Joint swelling
Grip strength
0.26t
0.17
0.527
0.18
0.32t
0.03
* Simple Pearson correlation coefficients for variables standardized
within treatment group.
+ P < 0.01.
i P < 0.05.
1350
MEENAN ET AL
changes in the psychological dimension of AIMS were
noticeably lower. The pain dimension of health status
was significantly correlated with grip strength and, to a
lesser degree, with joint tenderness, but not with joint
swelling. The self-assessments of patient status generated by the 2 approaches (patient assessment and
Arthritis Impact measure in AIMS) were weakly correlated at the level of 0.14 ( P = 0.13).
The discriminant analyses showed that information on the 4 clinical factors (swelling, tenderness, grip
strength, and overall patient assessment) allowed the
correct assignment of subjects to their true drug category in 42% of cases, while AIMS data did so in 43%.
When the 2 sets of data were combined, the percentage increased to 52%. If only the 2 categories of drug
and no drug were used, accurate assignment was made
in 71% of patients using the clinical factors, in 67% of
patients using the AIMS factors, and in 79% using both
sets together.
DISCUSSION
The results of this study provide interesting
information on both the sensitivity of a self-administered health status measure and the efficacy of oral
gold. The analyses of clinical, laboratory, and health
status measures all indicate that both oral and injectable gold are more effective than placebo in the
treatment of active rheumatoid arthritis. Pairwise
comparisons between auranofin and gold sodium thiomalate suggest that the latter is somewhat more effective, although it apparently has a significantly higher
incidence of adverse effects (22). The efficacy findings
in this trial, including those derived from the health
status measure, are consistent with published studies
comparing oral and injectable gold, which have generally found the 2 treatments to have roughly comparable efficacy with a tendency toward fewer side effects
in the patients treated with oral gold (28-30).
These results provide important, corroborating
evidence for the effectiveness of oral gold. This evidence goes beyond the limited perspective of clinical
measures by indicating that the drug is effective in
terms of the crucial measure of improved overall
health of the individual. More importantly, the results
clearly document the sensitivity of a formal health
status measure in a clinical trial of chronic disease
therapy. A reliable and valid health status instrument
was used in tandem with traditional clinical and labo-
ratory measures, and the 2 approaches produced comparable conclusions about the efficacy of gold treatment for rheumatoid arthritis. The similarity of the
conclusions, as well as the similarity of the AIMS
results to the clinical results from previous studies of
oral and injectable gold, strongly suggest that this
health status measure is able to detect meaningful
improvements in outcome in the clinical trial setting.
These health status findings corroborate previous work which has suggested that patient-generated,
“subjective” health measures could be profitably applied to clinical trials. Researchers, for example, have
demonstrated the utility of visual analog-type pain
rating scales in clinical trials (30-32). A recent study of
behavioral treatment for cancer chemotherapy patients has shown that a subjective measure, such as
anticipatory nausea, can be quite sensitive when used
in a clinical trial (33). These and other studies demonstrate that, in addition to having a high level of
relevance for outcome assessment, health status measures can meet the standards for reliability, validity,
and sensitivity that are required of a good outcome
measure (34).
To our knowledge, however, this is the first
report on the use of a previously validated, multidimensional health status instrument in a randomized,
controlled clinical trial involving an established therapy. Various health status measures have been used in
clinical trials before, but for a variety of reasons these
studies have not provided strong evidence for the
broader applicability of such measures. Measures of
health status components such as physical function or
psychological status are often developed informally by
investigators for a particular trial, without their reliability and/or validity being adequately tested. One
cannot be certain exactly what such tools are actually
measuring.
In the present study, the failure of the patient
assessment variable in the clinical measure set to
demonstrate a significant difference across the treatment groups is a case in point. This particular question
had not been pre-tested, contained only 5 possible
responses, was not self-administered, and could be
asked either before or after various other measures
were obtained. The Arthritis Impact measure in
AIMS, on the other hand, had been thoroughly tested
in our developmental work, was a visual analog-type
item which allowed a full range of responses, was selfadministered to avoid interviewer inconsistencies, and
was always completed in the same sequence at the end
of a full health status questionnaire. These differences
1351
OUTCOME ASSESSMENT
also explain the relatively low correlation between
these 2 apparently similar measures.
Evidence from the broadly-based single-dimension measures used to date is also unsatisfactory,
because such measures provide no information on
which aspects of health may or may not have improved, making it difficult to match the expected and
actual outcomes for certain types of interventions (35).
The multidimensional character of an instrument such
as AIMS permits it to be broadly based and still
provide more specific information on the effectiveness
of a given intervention. The majority of recently
developed health status measures incorporate this
type of approach.
These are 2 other reasons the present study
provides stronger evidence for the sensitivity of health
status measures. One is that, to our knowledge, a
validated health status measure has not previously
been used in a trial that involves a treatment of known
efficacy. When a relatively new outcome measure is
used in a situation where the effectiveness of the
treatment is unknown, it is difficult to determine
whether the findings, be they positive or negative, are
due primarily to the properties of the treatment or the
properties of the measurement approach (36). In the
present trial, a known treatment was used in parallel
with the study treatment. This provided a favorable
opportunity for determining the true properties of the
new outcome instrument because the approximate
effect of 1 intervention was already known.
A second unique aspect of the present study is
that the 2 sets of outcome data were processed and
analyzed independently: the health status data at Boston University and the clinical and laboratory data at
the University of Utah. This approach ensured that the
health status results would be analyzed separately and
not be treated as simply one of many outcome variables,
as has generally been the case. More importantly, this
separate analysis approximates a situation of 2 independent tests in which the traditional clinical findings
serve to validate the newer health status findings. The
change score correlations presented in Table 6 address
this issue of measurement comparison. They show
significant, but not startling, correlations between the
clinical and health status variables, and are generally
similar to the validity results we have previously
reported for static measures of these variables (24).
The study design and analysis approach used
here also facilitate a direct comparison between the 2
sets of measures, and allow one to address the issue of
which set would be “better” as a clinical trial outcome
measure. In the discriminant analyses, the clinical and
health status measures were very similar in their
ability to discriminate among the groups and between
treatment and no treatment. This provides further
evidence that the health status measures performed as
well as the standard measures in this trial, although it
must be pointed out that this conclusion may be
somewhat dependent on the particular drug, disease,
and/or duration factors of the study. Perhaps the most
important finding of all was that treatment discrimination by the 2 types of measures was additive, suggesting that the 2 approaches measure similar but slightly
different things and that they might be used most
profitably in tandem to assess the efficacy of a given
intervention.
In summary, this study has examined the sensitivity of a self-administered health status measure by
comparing it with standard clinical outcome approaches in a clinical drug trial in an important chronic
disease. The results demonstrate that such a health
status measure is able to detect clinically meaningful
differences between drug treatments in terms of their
effects on previously validated health status scales and
components. Our findings extend the conceptual
promise of health status instruments as outcome measures by providing empirical evidence of their utility.
These findings offer important new justification for the
further application of health status measures to clinical
trials of chronic disease. Such applications should
provide better insights about the efficacy of a variety
of health care interventions.
REFERENCES
1. Bull JP: The historical development of clinical therapeutic trials. J Chronic Dis 10:218-248, 1959
2. Pet0 R, Pike C, Armitage P, Breslow NE, Cox DR,
Howard S V , Mantel N , McPherson K , Pet0 J , Smith
PG: Design and analysis of randomized clinical trials
requiring prolonged observation of each patient: introduction and design. Br J Cancer 34585-612, 1976
3. Pet0 R, Pike C, Armitage P, Breslow NE, Cox DR,
Howard SV, Mantel N, McPherson K , Pet0 J, Smith
PG: Design and analysis of randomized clinical trials
requiring prolonged observation of each patient: analysis and examples. Br J Cancer 35:l-39, 1977
4. DerSimonian R, Charett LJ, McPeek B, Mosteller F:
Reporting on methods in clinical trials. N Engl J Med
306:1332-1337, 1982
5 . Feinstein AR: Clinical biostatistics. XLI. Hard science,
soft data and the challenges of choosing clinical variables in research. Clin Pharmacol Ther 22:485-498, 1977
6. Meenan RF: Assessing therapeutic effectiveness in the
1352
rheumatic diseases: the case for functional and health
status measures, Controversies in the Clinical Evaluation of Analgesic-Anti-Inflammatory-Antirheumatic
Drugs. Edited by HE Paulus, GE Ehrlich, E Lindenlaub. New York, Schattauer Verlag, 1981, pp 353-364
7. Eberl R, Fashing V, Rahfles V, Schleyer I, Wolf R:
Repeatability and objectivity of various measurements
in rheumatoid arthritis: a comparative study. Arthritis
Rheum 19:1278-1286, 1976
8. Mewa A, Law J, Cockshott WP, Park W, Pui M,
Buchanan WW: Is erosion count a valid outcome measure in RA? (abstract). Clin Res 31:158A, 1983
9. McDermott HW: Absence of indicators of the influence
of its physicians on a society’s health: impact of physician care on society. Am J Med 703333-843, 1981
10. Rogers DE: Who should give primary care: the continuing debate. N Engl J Med 205577-578, 1981
11. Kaplan RM, Bush JW, Berry CC: Health status: types of
validity and the Index of Well-Being. Health Serv Res
11:478-507, 1976
12. Brook RH, Ware JE, Davies-Avery A, Stewart AL,
Donald CA, Rogers WH, Williams KN, Johnston SA:
Overview of adult health status measures fielded in
Rand’s Health Insurance Study. Med Care (suppl) 17:l131, 1979
13. Bergner M, Bobbitt RA, Carter WB, Gilson BS: The
Sickness Impact Profile: development and final revision
of a health status measure. Med Care 19:787-805, 1981
14. Meenan KF: The AIMS approach to health status measurement: conceptual background and measurement
properties. J Rheumatol 9:785-788, 1982
15. Fries JF, Spitz PW, Young DY: The dimensions of
health outcomes: the Health Assessment Questionnaire
pain and disability scales. J Rheumatol 9:789-793, 1982
16. Jette AM: Functional Status Instrument: reliability of a
chronic disease evaluation instrument. Arch Phys Med
Rehabil 61:395-401, 1980
17. Chambers LW, MacDonald LA, Tugwell P, Buchanan
WW, Kraag G: The McMaster Health Index Questionnaire as a measure of quality of life for patients with
rheumatoid disease. J Rheumatol 9:780-784, 1982
18. Koran LM: The reliability of clinical methods, data and
judgements. N Engl J Med 293:642-646, 695-702, 1975
19. Bombardier C, Tugwell P, Sinclair A, Dok C, Anderson
G, Buchanan WW: Preference for endpoint measures in
clinical trials: the results of structured workshops. J
Rheumatol 9:798-801, 1982
20. Liang MH, Cullen K, Larson M: In search of a more
perfect mousetrap (health status or quality of life instrument). J Rheumatol 9:775-779, 1982
21. Kaye RL: The current clinical status of gold therapy in
rheumatic diseases. J Rheumatol (suppl) 9:124-131, 1982
22. Ward JR, Williams HJ, Egger MJ, Reading JC, Boyce E,
Altz-Smith M, Samuelson CO Jr, Willkens RF, Solsky
MEENAN ET AL
MA, Hayes SP, Blocka KL, Weinstein A, Meenan RF.
Guttadauria M, Kaplan SB, Klippel J: Comparison of
auranofin, gold sodium thiomalate, and placebo in the
treatment of rheumatoid arthritis: a controlled clinical
trial. Arthritis Rheum 26: 1303-1315, 1983
23. Ropes MW, Bennett GA, Cobb S, Jacox R, Jessar RA:
1958 revision of diagnostic criteria for rheumatoid arthritis. Bull Rheum Dis 9: 175-176, 1958
24. Meenan RF, Gertman PM, Mason JH, Dunaif R: The
Arthritis Impact Measurement Scales: further investigations of a health status measure. Arthritis Rheum
25:1048-1053, 1982
25. Brown JH, Kazis LE, Spitz PW, Gertman PM, Fries JF,
Meenan RF: The dimensions of health outcome: a cross
validated examination of health status measurement
using rheumatoid arthritis as a model. Am J Public
Health 74:159-161, 1984
26. Winer BJ: Statistical Principals of Experimental Design.
New York, McGraw-Hill, 1980
27. Conover WJ: Practical Non-Parametric Statistics. New
York, John Wiley 8L Sons, 1980
28. Menard HA, Beaudet F, Davis P, Harth M , Percy JS,
Russel AS, Thompson JM: Gold therapy in rheumatoid
arthritis: interim report of the Canadian multicenter
projective trial comparing sodium aurothiomalate and
auranofin. J Rheumatol (suppl) 9: 179-183, 1982
29. Schattenkirchner M, Kaik B, Muller-Faubender H, Rau
R, Zeidler H: Auranofin and sodium aurothiomalate in
the treatment of rheumatoid arthritis: a double-blind,
comparative, multicenter study. J Rheumatol (suppl)
9:184-189, 1982
30. Smith PR, Brown GM, Meyers OL: An open comparative study of auranofin vs. gold sodium thiomalate. J
Rheumatol (suppl) 9: 190-196, 1982
31. Huskisson EC: Measurement of pain. J Rheumatol
9:768-769, 1982
32. Downie WW, Leatham PA, Rhind VM, Wright B,
Banco JA, Anderson JA: Studies with pain rating scales.
Ann Rheum Dis 37:378-387, 1978
33. Morrow GR, Morrell C: Behavioral treatment for the
anticipatory nausea and vomiting induced by cancer
chemotherapy. N Engl J Med 307:1476-1480, 1982
34. Liang MH, Jette AM: Measuring functional ability in
chronic arthritis: a critical review. Arthritis Rheum
24230-86, 1981
35. Karnovsky DA, Burchenal JH: The clinical evaluation
of chemotherapeutic agents in cancer, Evaluation of
Chemotherapeutic Agents. Edited by CM McLeod.
New York, Columbia University Press, 1949, pp 191205
36. Nocturnal Oxygen Therapy Trial Group: Continuous or
nocturnal oxygen therapy in hypoxemic chronic obstructive lung disease: a clinical trial. Ann Intern Med
93:391-398, 1980
Документ
Категория
Без категории
Просмотров
0
Размер файла
828 Кб
Теги
outcomes, statue, health, clinical, sensitivity, evidence, trials, measures, assessment
1/--страниц
Пожаловаться на содержимое документа