Juvenile Idiopathic Arthritis of Peripheral Joints
Transcript of Juvenile Idiopathic Arthritis of Peripheral Joints
Review Article
Juvenile Idiopathic Arthritis of Peripheral Joints:Quality of Reporting of Diagnostic Accuracy of Conventional MRI1
Elka Miller, MD, Andreas Roposch, MD, MSc, Elizabeth Uleryk, BA, MLS, Andrea S. Doria, MD, PhD, MSc
Rationale and Objectives. The aim of this study was to systematically review the quality of papers on the clinimetric properties
of magnetic resonance imaging for the diagnosis of juvenile idiopathic arthritis in peripheral joints.
Materials and Methods. A review of Medline, EMBASE, the Database of Abstracts of Reviews of Effects, and the Cochrane
Library was performed by using a systematic search strategy. Two independent reviewers evaluated selected articles by using
Standards for Reporting of Diagnostic Accuracy (STARD) and Quality Assessment of Diagnostic Accuracy Studies (QUADAS)
tools. Items were reported independently for STARD and QUADAS.
Results. Eighteen studies (validity, n = 18; reliability, n = 3; responsiveness, n = 3) were included. Their overall quality of
reporting of methods was fair. Methodological problems with the STARD system included a lack of reporting of exclusion
criteria (n = 14), partial or no information on operators’ expertise (n = 14) or blinding (n = 18), and deficient information on study
time frames (n = 12), treatments (n = 10), or indeterminate results (n = 18). The distribution of QUADAS scores was hetero-
geneous, with overall scores ranging between 3.5 (poor) and 16.5 (excellent) (maximum score, 17.5).
Conclusions. The quality of reporting of methods in studies on the magnetic resonance imaging assessment of juvenile idiopathic
arthritis is heterogeneous and fair overall. Further methodological refinement of research design should be sought in future
studies to provide stronger evidence for the value of novel techniques in clinical settings.
Key Words. Juvenile idiopathic arthritis; magnetic resonance imaging; systematic review; peripheral joints; children.
ª AUR, 2009
Juvenile idiopathic arthritis (JIA) is the most common rheu-
matic disease of childhood, with six to 19.6 incident cases per
100,000 children yearly in North America (1). Although the
prognosis of this disease in children is generally favorable,
with a majority of patients having no active synovitis in
Acad Radiol 2009; 16:739–757
1 From the Department of Diagnostic Imaging, Hamilton Health Science,
McMaster University, 1200 Main Street West, Room 2SRAD, Hamilton, ON
L8N 3Z5, Canada (E.M.-work developed during Dr. Miller’s fellowship at the
Department of Diagnostic Imaging, Hospital for Sick Children, University of
Toronto, Toronto, Canada); the Department of Diagnostic Imaging, Hospital for
Sick Children, University of Toronto, Toronto, Canada (E.U., A.S.D.); and the
Department of Orthopaedic Surgery, Great Ormond Street Hospital for
Children, Institute of Child Health, University College London, London, United
Kingdom (A.R.). This study was partially funded by a Career Development
Award from the Canadian Child Health Clinician-Scientist Program and from
the Department of Medical Imaging, University of Toronto, to Dr Doria.
Received October 3, 2008; accepted January 8, 2009. Address correspon-
dence to: E.M. e-mail: [email protected] or [email protected]
ª AUR, 2009doi:10.1016/j.acra.2009.01.012
adulthood, persistent disease or progression of the disease is
not uncommon, affecting children’s growth and development
(2,3).
Laboratory indices of synovial inflammation are measures
that are easily quantifiable but fail to provide accurate in-
formation about functional joint outcomes in JIA (4). Ra-
diographs are usually nonspecific in the early stages of the
disease, while magnetic resonance imaging (MRI) is a sensi-
tive imaging tool for the detection of synovial hypertrophy
and cartilage degeneration, which can be evaluated only
indirectly by plain radiography (5).
For a diagnostic test such as MRI to be helpful, it must
fulfill basic diagnostic test standards, including accuracy
(reliability and validity) and responsiveness (sensitivity to
change) (6). Only when these basic measurement properties
are established can the further assessment of results on clin-
ical outcomes, decision making, cost-effectiveness, and risks
associated with a test be evaluated (7–10).
Because of the increasing interest in the role of anatomic
MRI as an outcome measure in clinical trials and in
739
MILLER ET AL Academic Radiology, Vol 16, No 6, June 2009
functional MRI as a predictor in arthritis, it becomes essential
to determine the current quality of reporting of diagnostic
studies of MRI in JIA. A recent unstructured review (5) as-
sessed the role of MRI for the diagnosis of JIA, but to our
knowledge, no prior systematic review has been conducted to
assess the current status of knowledge on the measurement
properties of MRI for the diagnostic assessment of JIA. The
goal of the present systematic review was to assess the quality
of diagnostic accuracy reporting of studies with regard to the
clinimetric (reliability, validity, and responsiveness) proper-
ties of MRI to diagnose JIA.
MATERIAL AND METHODS
Data Sources and Search
An electronic search of the literature was performed by
three investigators (E.M., E.U., A.S.D.), who identified studies
in which the authors reported the diagnostic accuracy of MRI
for the assessment of JIA. Medline (January 1966 to June
2008), EMBASE (January 1980 to June 2008), the Database of
Abstracts of Reviews of Effects of the National Health Service
Center for Reviews and Dissemination, and the Cochrane Li-
brary were searched through OVID using a validated search
strategy (8) that combined Medical Subject Headings and
EMBASE terms with free-text words. These terms included
‘‘juvenile idiopathic arthritis,’’ ‘‘juvenile rheumatoid arthri-
tis,’’ ‘‘arthritis,’’ ‘‘cartilage degeneration,’’ ‘‘magnetic reso-
nance imaging,’’ ‘‘T2 mapping,’’ ‘‘molecular imaging,’’
‘‘diagnostic sensitivity,’’ ‘‘treatment,’’ and ‘‘outcome.’’
Two reviewers (E.M., A.S.D.) independently read the
abstracts of all articles with relevant titles. If the content of
a study was not obvious from the title, key words, and ab-
stract, the original article was retrieved and evaluated by both
reviewers for eligibility. Subsequently, all original articles
that were found to be eligible for inclusion were reviewed
independently. At any stage, disagreements were discussed
and resolved in a consensus.
Inclusion Criteria
Our systematic review included studies on clinimetric
(evaluative, discriminative, and predictive) measurement
properties of any MRI methods that were reported for the
diagnosis of JIA in patients aged <18 years. Specifically, we
included studies in which the conversion criteria (ie, how to
analyze and interpret an MRI examination) used to evaluate
the reliability or validity of MRI assessment of peripheral
joints (knees, hips, ankles and feet, wrists and hands, and
shoulders) in children with JIA was reasonably described.
The minimum criterion for the inclusion of articles with
regard to process criteria (ie, how to perform the MRI scan-
ning) was a one-paragraph report on the proposed MRI pro-
tocol and scanner information. Eligible studies reported
740
primary data obtained at a health care center or research
institute.
Eligible studies reported the methodologic concepts of
either construct or criterion validity, regardless of whether
these terms were or were not mentioned in the article.
Studies excluded from our review included those not
providing patients’ demographic information, case reports,
pictorial essays, opinion letters, and reviews. Studies that
specifically reported axially located joint disease (sacroiliac
joint or temporomandibular joint) were also excluded. Arti-
cles written in languages other than English, French, Ger-
man, Italian, Spanish, and Portuguese were excluded.
Data Extraction and Outcome Measures
Quality assessment of the reporting of included articles was
performed using both the Standards for Reporting of Diag-
nostic Accuracy (STARD; 25 items) (11) and Quality As-
sessment of Diagnostic Accuracy Studies (QUADAS; 14
items) (12) criteria for diagnostic test reporting. The STARD
tool items were rated as adequately described, not described,
or partially described (6). The quality of conversion meth-
odology of each study was evaluated using the QUADAS
criteria (12); for this tool, we indicated whether the items
were or were not adequately described (yes or no). If it was
unclear from the information provided in the article, this item
was rated as ‘‘unclear’’ (6). Reliability studies were not as-
sessed with the QUADAS criteria, because only two items of
this system were applicable (6).
Statistical Analysis
The level of agreement between the two reviewers in
scoring the STARD and QUADAS criteria was assessed
using k statistics with 95% confidence intervals (CIs) (13,14).
Total agreement between the two readers was considered if
both readers scored a given item as 1 (complete information),
0 (lack of information), or 0.5 (partial information). Partial
interreader agreement was considered if one reader scored
a given item as 1 and the other reader as 0.5, or if one reader
scored the item as 0 and the other reader as 0.5. No agreement
meant that one reader scored a given item as 0 and the other
reader as 1. Kappa coefficients #0.40 indicated poor agree-
ment, $0.40 and #0.60 moderate agreement, $0.60 and
#0.80 good agreement, and $0.80 excellent agreement (13).
We used SAS version 8.2 (SAS Institute Inc, Cary, NC) for
all analyses.
RESULTS
Search and Selection
We retrieved 1,782 citations; 18 studies were found to be
eligible for this systematic review (Fig 1). Reasons for the
exclusion of papers included a lack of information on
Academic Radiology, Vol 16, No 6, June 2009 JIA: QUALITY OF REPORTING OF DIAGNOSTIC ACCURACY OF MRI
conversion criteria for MRI (15,16), pictorial essays (17,18),
review articles (5), a lack of correlation with MRI (19,20),
and axially located joint disease (21–24). Table 1 shows the
demographic characteristic of patients and MRI clinimetric
properties of the selected articles. Details on methodologic
design according to the STARD and QUADAS criteria are
available in Appendices A (STARD criteria) and B
(QUADAS criteria).
Qualitative Assessment of Quality of Reporting(STARD Tool)
The overall interreader agreement for STARD was good
for both total (k = 0.70; 95% CI, 0.64–0.76) and partial (k =
0.77; 95% CI, 0.34–1.0) agreement. Disagreements were the
result of vague description of key issues, such as participant
sampling (item 5), study design (item 6), reference standard
(items 7 and 19), reproducibility (item 13), and statistics
(item 24). Only four studies (22.2%) (25–28) reported the
Medical Subject Heading term ‘‘sensitivity,’’ ‘‘specificity,’’
or ‘‘predictive values.’’
The objectives of the selected studies were (1) assess-
ment of the role of contrast-enhanced MRI in the diagnosis
of long-standing (25,27,29–33) and short-standing disease
(<1 year in duration) (28); (2) evaluation of unaffected
knees as an attempt to identify patients at risk for developing
polyarticular JIA (26); (3) evaluation of novel MRI tech-
niques, including three-dimensional fat-saturated contrast
MRI (34), T2 relaxation (35), synovium volume quantifi-
cation (36), and synovium pharmacokinetic characteristic
analysis (37–39); (4) determination of the efficacy and
toxicity of intra-articular steroids using conventional MRI as
the outcome measure (33,40,41); and (5) examination of the
role of ultrasound in assessing joint inflammation in JIA
compared with MRI (42).
Participants (Items 3–6)In 11 studies (61.1%), the authors mentioned the criteria
used for the diagnosis of JIA (International League of As-
sociations for Rheumatology or American College of Rheu-
matology). Only four studies (22.2%) reported exclusion
criteria, such as previous intra-articular injection of steroids
(28,42), motion artifacts (35), and need for sedation for im-
aging (28,35,36).
Fifteen studies (83.3%) (25,26,28–30,33,35–42) included
information on the recruitment process of patients, and only
eight studies (44.4%) reported sampling of consecutive
cases. Data were collected prospectively in 11 studies
(61.1%) (26,28,33,35–42).
Reference Standard (Item 7)Nine studies (50%) used physical examination of the pa-
tient as a reference standard for MRI examinations
(25,28,30,33,35,36,38,39,41). Four studies (22.2%) used
MRI for discrimination between diseased and healthy joints
(25,26,35,42). In five studies (27.8%), MRI was used a ref-
erence standard, as an internal reference (contrast-enhanced
MRI) for unenhanced MRI (31,32,42), as an internal refer-
ence for different pharmacokinetic models of enhanced MRI
(37), or as an outcome measure for clinical or laboratory
evaluation (27). In two studies (11.1%) (31,32), arthroscopy
was used as the reference standard for MRI.
Test Methods (Items 8–11)Process criteria were reasonably well described in all
but two studies (11.1%) (25,33), which partially described
the imaging acquisition methods and the criteria for inter-
preting magnetic resonance images. Only two studies
(11.1%) (27,28) stated the rationale for using specific units
and cutoff values of the index test (MRI). Four studies
Figure 1. Flow diagram revealing the search and selection pro-
cess used for the identification and quality assessment of articles
on the diagnostic accuracy of juvenile idiopathic arthritis. DARE,
Database of Abstracts of Reviews of Effects.
741
MILLER ET AL Academic Radiology, Vol 16, No 6, June 2009
Table 1Demographic Characteristics of Patients and Corresponding MRI Clinimetric Properties of the Selected Articles in This Review
Study
Number of Patients,
Number of Joints,
Joint Types
Mean
(Range) Age (y) % Female
Research
Design
Construct
Validity Reliability Responsiveness
Herve-Somma et al (31) 24 (24 knees) 10 (3–18) 17 Retrospective Yes
Eich et al (40) 15 (11 knees, 4 hips) 6.3 (3.5–11.8) 7 Prospective Yes Yes
Huppertz et al (41) 21 (18 knees,
2 ankles, 1 elbow)
10.1 (1.4–18.9) 14 Prospective Yes Yes
Murray et al (32) 7 (14 hips) 11 (7–17) 4 Not stated Yes
Remedios et al (33) 11 (13 ankles) 9.7 (5–14) 9 Prospective Yes
Ramsey et al (29) 21 (21 knees) 13 (2–17) 11 Retrospective
Uhl et al (25) 21 (42 knees) 9.5 (5–13) 8 Not stated Yes Yes
Cakmakci et al (34) 38 (38 knees) 8 (2–17) 25 Prospective Yes
Gylys-Morin et al (28) 30 (30 knees) 10.2 (5–16) 21 Prospective Yes Yes
El-Miedany et al (42) 40 (40 knees)
JIA patients,
40 control
patients
11 (3–17) 32 Not stated
Argyropoulou et al (30) 28 (56 hips) 12.5 (2–24) 14 Not stated Yes
Kight et al (35) 18 (18 knees)
JIA patients,
21 (21 knees)
healthy children
8.1 (4.9–10.8) 39 Prospective
Workie et al (38) 13 (13 knees) 10.2 (6–16) 9 Prospective Yes
Workie and Dardzinski (37) 10 (10 wrists) 11.1 (5.2–15.7) 9 Prospective Yes Yes
Graham et al (36) 8 (8 knees) 11 (6–15) 7 Not stated Yes
Gardner-Medwin (26) 10 (10 knees) 9.4 (5.2–14.2) 7 Prospective Yes
Nistala et al (27) 34 (68 hips) 14.4 (4.3–19.7) No information Retrospective
Workie et al (39) 17 (17 knees) 10.3 (6.4–15.5) 13 Prospective Yes
JIA, juvenile idiopathic arthritis; MRI, magnetic resonance imaging.
(22.2%) (26–29) provided information on the expertise of
the professionals who executed and rated the MRI and
reference standards. In only four studies (22.2%) (26–
28,35) was clear information about the blinding of asses-
sors available.
Statistics (Items 12 and 13)Of three studies (16.7%) (27,28,36) that assessed reli-
ability, k statistics were used in two (11.1%) and coefficients
of variation were used in one (5.6%). In two additional
studies (11.1%) (27,34), k statistics were inappropriately
applied for agreement between MRI and clinical results. Only
three studies (16.7%) (25,27,28) provided receiver-operating
characteristic (ROC) curves as measures of validity.
Only one study (5.6%) (27) modeled the data using re-
gression analysis. Four studies (22.2%) (29,33,40,42) did not
apply any statistical methods at all. None of the three studies
on responsiveness (34,40,41) used sound statistical methods
to report changes over time.
Time Frame of Study (Item 14)Only six studies (33.3%) included information on when
the studies had been performed.
742
Characteristics of Participants (Item 15 and 16)There was satisfactory reporting of demographic and
clinical characteristics of participants (age, sex, and anatomic
localization of arthritis) in all studies.
Test Results (Items 17–20)Six of 18 diagnostic accuracy studies (33.3%) and two of
three responsiveness studies (66.7%) adequately reported the
treatments used and the time intervals between evaluation of
the index test and the corresponding reference standard. The
distribution of disease severity was well described in seven
studies (38.9%), partially described in seven (38.9%), and not
described at all in four (22.2%).
In nine diagnostic accuracy studies (50%), the authors
reported their results in raw data tables that enabled the re-
calculation of results. Two of the studies (66.7%) on re-
sponsiveness (40) reported patients’ adverse events with the
use of steroids. No studies reported adverse events with ga-
dolinium administration or sedation, if used in the study.
Estimates (Items 21–24)Precision values, such as 95% CIs, were reported in only
three studies (16.7%) (28,32,41), and standard deviation or
Academic Radiology, Vol 16, No 6, June 2009 JIA: QUALITY OF REPORTING OF DIAGNOSTIC ACCURACY OF MRI
standard error measures were reported in five studies
(25,28,30,35,41). Indeterminate results were not reported in
any validity study. Estimates of interreader agreement for
MRI findings reported as k coefficients denoted good (0.40–
0.75) or excellent (>0.75) overall reliability (27,28,43). In the
study of Graham et al (36), the intrareader and interreader
reliability of MRI measured as coefficients of variation
ranged between 7.6% and 13.9% for individual readers and
was 11.2% for interreader assessment.
Semiquantitative Assessment of Quality ofReporting (QUADAS Tool)
Overall, the interreader reliability of interpretation of the
QUADAS items was moderate for total agreement (k = 0.57;
95% CI, 0.47–0.67) and good for partial agreement (k = 0.64;
95% CI, 0.52–0.77). Disagreements resulted from unclear
information regarding the type of reference standard used
(item 3), the timing (item 4) and independence (item 7) be-
tween the performance of the reference standard and the in-
dex test, and verification bias (item 10).
There was substantial heterogeneity in the quality of
methods used for reporting results in the studies included in
this review, as noted by the distribution of QUADAS scores
(Appendix B). Scores for single studies (maximum score,
17.5) ranged between 3.5 (poor) and 16.5 (excellent) (median
score, 10.3 [fair quality]). Scores for individual studies
(maximum score, 14) ranged between 1 to 5 (poor) and 11 to
14 (excellent) (27,28,35) (median score, 6–10 points, 8.3
[fair quality]).
None of the studies reported the method used for sample
size calculation (item 1). In 16 studies (88.9%), the reference
standard measure was assessed regardless of the index test
result (item 6). In 12 studies (66.7%), the authors reported
whether the reference standard measure was assessed in all
patients or in only some patients (item 5). Information on
imaging acquisition was consistently reported in 15 studies
(83.3%) (item 8). In 17 studies (94.4%), clinical data were
available by the time the test results were interpreted (item
12). Uninterpretable test results (item 13) (score, 1.5), with-
drawals from the study (item 14) (score, 3.5), and interpre-
tation of reference standard results without knowledge of
index test results (item 11) (score, 6) were items that were
poorly scored.
DISCUSSION
The results of this review indicate that diagnostic test
standards were fairly fulfilled in studies of the validity, reli-
ability, and responsiveness of MRI in JIA.
Ideally, more clarity in the description of research designs
is desirable in upcoming studies. Suggestions to improve the
methodologic report of studies include the use of the medical
subject terms ‘‘sensitivity,’’ ‘‘specificity,’’ or ‘‘positive [or
negative] likelihood [or predictive value]’’ and the use of
ROC curves as a measure of concurrent validity. The
standardization of protocols and the validation of an MRI
scale for the interpretation of JIA findings are topics to be
pointed out in future studies. Although MRI is able to
discriminate different types of cartilage (articular, epiphy-
seal, and physeal) at distinct stages of development of
growing joints (44), so far, no MRI scale has been validated
for use in children. The investigation of the Outcome
Measures in Rheumatology group has been focused on the
definition and testing of novel imaging tools for the
assessment of rheumatoid arthritis of adults (45). This
limitation with regard to the unavailability of MRI scales
targeted to the pediatric population is demonstrated in
Gylys-Morin et al’s (28) study, which provides cutoff
values for Pettersson radiographic scores but fails to report
cutoff values for MRI findings. This limitation makes it
difficult to assess the reproducibility of ROC curves, which
are considered standard methods for descriptions of diag-
nostic accuracy (46,47).
In this review, only two studies (31,32) provided infor-
mation on the criterion validity of MRI compared with ar-
throscopic results. In Gylys-Morin et al’s (28) study, only one
patient had femoral and tibial cartilage thinning confirmed at
arthroscopy. The evaluation of criterion validity of MRI in
JIA is a challenging issue, because in vivo reference stan-
dards are typically unavailable for the determination of in-
active or remission disease (48).
Clinical examinations cannot be considered as reference
standards to determine the diagnostic accuracy of an imaging
test, because they do not provide an equivalent or superior
quantity of information compared to the diagnostic test (6).
Nevertheless, clinical findings can be used as constructs for
comparison with MRI. Future studies should explain the ra-
tionale for choosing a reference standard.
With regard to construct validity, most of the selected
studies reported internal correlations between MRI findings
or correlations between MRI findings and clinical and lab-
oratory results. In this review, although several studies
evaluated the correlation between synovial contrast en-
hancement and disease activity (38,39), only one study (38)
compared MRI findings to functional outcomes (total joint
scores and Childhood Health Assessment Questionnaire).
The Childhood Health Assessment Questionnaire is the most
widely used instrument for the assessment of functional
status during childhood (age 1–19 years) for musculoskeletal
disorders, and it has demonstrated high validity, reliability,
and responsiveness to changes over time (49). Further in-
vestigation of correlations between MRI findings and
Childhood Health Assessment Questionnaire constructs
should be encouraged.
743
MILLER ET AL Academic Radiology, Vol 16, No 6, June 2009
Very few studies (three of 18) evaluated the responsive-
ness of MRI. The fact that responsiveness studies typically
require a long time frame for the investigation of changes
over time and for monitoring adverse drug events (40,41)
may have contributed to the shortage of studies that ad-
dressed this clinimetric property of MRI. Likewise, only
three studies (72.2%) assessed the interreader reliability of
the interpretation of MRI.
Our review showed that the process criteria were rea-
sonably well reported in most of the selected articles.
However, both in our review and in other reviews (6), the
rationale for using specific cutoff values and units was
omitted. The results of our review also showed that the
articles that failed to report statistical methods were usually
published in the 1990s, likely reflecting a growing focus in
the scientific community on following standard guidelines
for the publication of articles on diagnostic tests (50,51).
On the other hand, with regard to the descriptions of the
conversion criterion (interpretation of findings) in the se-
lected studies, <30% of the articles reported blinding of
reviewers. Blinding is important to generate less biased
results.
With regard to reliability, none of the studies in this review
reported the causes of disagreement between readers or how
the authors handled indeterminate results, similar to what has
been noted in other reviews (6). Reasons for disagreement
should be reported in an effort to provide unequivocal in-
formation for clinical practice (52). None of the studies in-
cluded in this review reported the methods used for sample
size calculation or defined the spectrums of patients selected
for their studies. The lack of this information impaired the
external validity of their results.
Although the earliest changes in JIA are seen in the small
joints of the feet and hands (53,54), most of the studies of this
review (16 [88.9%]) evaluated large joints (the knees and
hips). The limited availability of high-resolution coils for
imaging the small joints may have accounted for the prefer-
ential investigation of large joints.
The chief limitation of this systematic review is the het-
erogeneity of scores on research methodology items noted
with the QUADAS tool. The use of different strength field
MRI scanners (0.5 vs 1.5 T), coils, and sequences in the se-
lected studies contributed to the heterogeneity. Finally, nei-
ther the STARD nor the QUADAS tool incorporates a quality
score; therefore, the scores in this review tended to ignore the
importance of individual items and the direction of potential
bias (11,12).
In conclusion, although most studies had prospective de-
signs, which enables better planning of research design, the
overall reporting of the diagnostic accuracy of MRI in as-
sessing JIA was fair, with several methodologic flaws noted.
The standardization of MRI protocols and scales for the in-
terpretation of findings in growing joints with JIA is clearly
744
needed. This may facilitate future reporting of sound diag-
nostic test statistics providing sensible cutoff values for the
calculation of ROC curves. Future studies in JIA should
focus on the assessment of small joints (hands, wrists, and
feet) rather than large joints for the detection of early
changes in JIA. Improvement of reporting should empha-
size conversion criteria. Reports should include information
on the blinding of reviewers, the number and expertise of
readers, indeterminate results, diagnostic accuracy in terms
of ROC curves, precise estimates (95% CIs, standard de-
viations, and standard errors) of values, assessments of the
reliability of interpretation of MRI results and responsive-
ness of MRI, and investigation of the value of MRI as
a predictive index.
GLOSSARY
QUADAS (Quality Assessment of DiagnosticAccuracy Studies) (12): A scoring system developed in an
attempt to produce standards for reporting evaluations of
diagnostic tests (55).
Reliability: Obtaining the same result when a phenome-
non is measured by the same clinician or by different clini-
cians on the same occasion or on different occasions (56).
Responsiveness: The ability of a scale to detect change in
outcomes when change is present (sensitivity to change)
(57,58).
STARD (Standards for Reporting of Diagnostic Ac-curacy) (11): A scoring system developed in an attempt to
produce standards for reporting evaluations of diagnostic
tests (55).
Validity: The degree to which the result of a measure-
ment corresponds to the true state of the phenomenon being
measured (56). Four types of validity are recognized: face,
content, construct, and criterion (59). Construct validity
specifies the factors, or constructs, that account for variance
in the proposed measures as well as the hypothesized re-
lations among them (ie, convergent or discriminate rela-
tionships) (59). Criterion validity is the correspondence
between a proposed measure and a reference standard
variable (59).
REFERENCES
1. Cassidy JT, Petty RE. Juvenile rheumatoid arthritis. In: Textbook of pe-
diatric rheumatology. Philadelphia, PA: Saunders; 1995; 135.
2. Oen K. Long-term outcomes and predictors of outcomes for patients
with juvenile idiopathic arthritis. Best Pract Res Clin Rheumatol 2002; 16:
347–360.
3. Oen K, Malleson PN, Cabral DA, et al. Disease course and outcome of
juvenile rheumatoid arthritis in a multicenter cohort. J Rheumatol 2002;
29:1989–1999.
4. Prakash K. Pispati. Evidence-based practice in rheumatology. APLAR J
Rheumatol 2003; 6:44–49.
Academic Radiology, Vol 16, No 6, June 2009 JIA: QUALITY OF REPORTING OF DIAGNOSTIC ACCURACY OF MRI
5. Graham TB. Imaging in juvenile arthritis. Curr Opin Rheumatol 2005; 17:
574–578.
6. Roposch A, Moreau NM, Uleryk E, Doria AS. Developmental dysplasia of
the hip: quality of reporting of diagnostic accuracy for US. Radiology
2006; 241:854–860.
7. Lijmer JG, Mol BW, Heisterkamp S, et al. Empirical evidence of design-
related bias in studies of diagnostic tests. JAMA 1999; 282:1061–1066.
8. Deville WL, Bezemer PD, Bouter LM. Publications on diagnostic test
evaluation in family medicine journals: an optimal search strategy. J Clin
Epidemiol 2000; 53:65–69.
9. Smidt N, Rutjes AW, van der Windt DA, et al. Quality of reporting of di-
agnostic accuracy studies. Radiology 2005; 235:347–353.
10. Reid MC, Lachs MS, Feinstein AR. Use of methodological standards in
diagnostic test research. Getting better but still not good. JAMA 1995;
274:645–651.
11. Bossuyt PM, Reitsma JB, Bruns DE, et al. Towards complete and ac-
curate reporting of studies of diagnostic accuracy: the STARD initiative.
Clin Chem 2003; 49:1–6.
12. Whiting P, Rutjes AW, Reitsma JB, et al. The development of QUADAS:
a tool for the quality assessment of studies of diagnostic accuracy in-
cluded in systematic reviews. BMC Med Res Methodol 2003; 3:25.
13. Altman DG. Practical statistics for medical research. London: Chapman
& Hall; 1991.
14. Fleiss JL. Statistical methods for rates and proportions; New York: John
Wiley, 1981.
15. Tynjala P, Honkanen V, Lahdenne P. Intra-articular steroids in radiologi-
cally confirmed tarsal and hip synovitis in juvenile idiopathic arthritis. Clin
Exp Rheumatol 2004; 22:643–648.
16. Johnson K, Wittkop B, Haigh F, et al. The early magnetic resonance im-
aging features of the knee in juvenile idiopathic arthritis. Clin Radiol 2002;
57:466–471.
17. Yulish BS, Lieberman JM, Newman AJ, et al. Juvenile rheumatoid
arthritis: assessment with MR imaging. Radiology 1987; 165:149–152.
18. Senac MO Jr, Deutsch D, Bernstein BH, et al. MR imaging in juvenile
rheumatoid arthritis. AJR Am J Roentgenol 1988; 150:873–878.
19. Adib N, Silman A, Thomson W. Outcome following onset of juvenile
idiopathic inflammatory arthritis: I. frequency of different outcomes.
Rheumatology (Oxford) 2005; 44:995–1001.
20. Adib N, Silman A, Thomson W. Outcome following onset of juvenile
idiopathic inflammatory arthritis: II. predictors of outcome in juvenile
arthritis. Rheumatology (Oxford) 2005; 44:1002–1007.
21. Kuseler A, Pedersen TK, Gelineck J, Herlin T. A 2 year followup study of
enhanced magnetic resonance imaging and clinical examination of the
temporomandibular joint in children with juvenile idiopathic arthritis.
J Rheumatol 2005; 32:162–169.
22. Arabshahi B, Cron RQ. Temporomandibular joint arthritis in juvenile
idiopathic arthritis: the forgotten joint. Curr Opin Rheumatol 2006; 18:
490–495.
23. Scolozzi P, Bosson G, Jaques B. Severe isolated temporomandibular
joint involvement in juvenile idiopathic arthritis. J Oral Maxillofac Surg
2005; 63:1368–1371.
24. Neidel J, Boehnke M, Kuster RM. The efficacy and safety of intraarticular
corticosteroid therapy for coxitis in juvenile rheumatoid arthritis. Arthritis
Rheum 2002; 46:1620–1628.
25. Uhl M, Krauss M, Kern S, et al. The knee joint in early juvenile idiopathic
arthritis. An ROC study for evaluating the diagnostic accuracy of con-
trast-enhanced MR imaging. Acta Radiologica 2001; 42:6–9.
26. Gardner-Medwin JM, Killeen OG, Ryder CAJ, et al. Magnetic resonance
imaging identifies features in clinically unaffected knees predicting ex-
tension of arthritis in children with monoarthritis. J Rheumatol 2006; 33:
2337–2343.
27. Nistala K, Babar J, Johnson K, et al. Clinical assessment and core
outcome variables are poor predictors of hip arthritis diagnosed by
MRI in juvenile idiopathic arthritis. Rheumatology (Oxford) 2007; 46:
699–702.
28. Gylys-Morin VM, Graham TB, Blebea JS, et al. Knee in early juvenile
rheumatoid arthritis: MR imaging findings. Radiology 2001; 220:
696–706.
29. Ramsey SE, Cairns RA, Cabral DA, et al. Knee magnetic resonance im-
aging in childhood chronic monarthritis. J Rheumatol 1999; 26:
2238–2243.
30. Argyropoulou MI, Fanis SL, Xenakis T, et al. The role of MRI in the eval-
uation of hip joint disease in clinical subtypes of juvenile idiopathic ar-
thritis. Br J Radiol 2002; 75:229–233.
31. Herve-Somma CM, Sebag GH, Prieur AM, et al. Juvenile rheumatoid arthritis
of the knee: MR evaluation with Gd-DOTA. Radiology 1992; 182:93–98.
32. Murray JG, Ridley NTF, Mitchell N, Rooney M. Juvenile chronic arthritis of
the hip: Value of contrast-enhanced MR imaging. Clin Radiol 1996; 51:
99–102.
33. Remedios D, Martin K, Kaplan G, et al. Juvenile chronic arthritis: diag-
nosis and management of tibio-talar and sub-talar disease. Br J Rheu-
matol 1997; 36:1214–1217.
34. Cakmakci H, Kovanlikaya A, Unsal E. Short-term follow-up of the juvenile
rheumatoid knee with fat-saturated 3D MRI. Pediatr Radiol 2001; 31:
189–195.
35. Kight AC, Dardzinski BJ, Laor T, Graham TB. Magnetic resonance im-
aging evaluation of the effects of juvenile rheumatoid arthritis on distal
femoral weight-bearing cartilage. Arthritis Rheum 2004; 50:901–905.
36. Graham TB, Laor T, Dardzinski BJ. Quantitative magnetic resonance
imaging of the hands and wrists of children with juvenile rheumatoid ar-
thritis. J Rheumatol 2005; 32:1811–1820.
37. Workie DW, Dardzinski BJ. Quantifying dynamic contrast-enhanced MRI
of the knee in children with juvenile rheumatoid arthritis using an arterial
input function (AIF) extracted from popliteal artery enhancement, and the
effect of the choice of the AIF on the kinetic parameters. Magnc Reson
Med 2005; 54:560–568.
38. Workie DW, Dardzinski BJ, Graham TB, et al. Quantification of dynamic
contrast-enhanced MR imaging of the knee in children with juvenile
rheumatoid arthritis based on pharmacokinetic modeling. Magn Reson
Imaging 2004; 22:1201–1210.
39. Workie DW, Graham TB, Laor T, et al. Quantitative MR characterization of
disease activity in the knee in children with juvenile idiopathic arthritis:
a longitudinal pilot study. Pediatr Radiol 2007; 37:535–543.
40. Eich GF, Halle F, Hodler J, et al. Juvenile chronic arthritis: imaging of the
knees and hips before and after intraarticular steroid injection. Pediatr
Radiol 1994; 24:558–563.
41. Huppertz HI, Tschammler A, Horwitz AE, Schwab KO. Intraarticular
corticosteroids for chronic arthritis in children: efficacy and effects on
cartilage and growth. J Pediatr 1995; 127:317–321.
42. El-Miedany YM, Housny IH, Mansour HM, et al. Ultrasound versus MRI in
the evaluation of juvenile idiopathic arthritis of the knee. Joint Bone Spine
2001; 68:222–230.
43. Rosner B. Fundamentals of biostatistics. Belmont, CA: Duxbury; 1995.
44. Doria AS, Babyn PS, Feldman B. A critical appraisal of radiographic
scoring systems for assessment of juvenile idiopathic arthritis. Pediatr
Radiol 2006; 36:759–772.
45. Ostergaard M, McQueen F, Bird P, et al. The OMERACT Magnetic Res-
onance Imaging Inflammatory Arthritis Group—advances and priorities.
J Rheumatol 2007; 34:852–853.
46. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver
operating characteristic (ROC) curve. Radiology 1982; 143:29–36.
47. Moses LE, Shapiro D, Littenberg B. Combining independent studies of
a diagnostic test into a summary ROC curve: data-analytic approaches
and some additional considerations. Stat Med 1993; 12:1293–1316.
48. Wallace CA, Ravelli A, Huang B, Giannini EH. Preliminary validation of
clinical remission criteria using the OMERACT filter for select categories
of juvenile idiopathic arthritis. J Rheumatol 2006; 33:789–795.
49. Singh G, Athreya BH, Fries JF, Goldsmith DP. Measurement of health
status in children with juvenile rheumatoid arthritis. Arthritis Rheum 1994;
37:1761–1769.
50. Jaeschke R, Guyatt G, Sackett DL. Evidence-Based Medicine Working
Group. Users’ guides to the medical literature. III. How to use an article
about a diagnostic test. A. Are the results of the study valid? JAMA 1994;
271:389–391.
51. Jaeschke R, Guyatt GH, Sackett DL. Evidence-Based Medicine Working
Group. Users’ guides to the medical literature. III. How to use an article
about a diagnostic test. B. What are the results and will they help me in
caring for my patients? JAMA 1994; 271:703–707.
745
52. O’Dowd TC. Informing patients about clinical disagreement. Lancet
1989; 2:744.
53. Scott DL, Coulton BL, Popert AJ. Long term progression of joint damage
in rheumatoid arthritis. Ann Rheum Dis 1986; 45:373–378.
54. Sharp JT. An overview of radiographic analysis of joint damage in rheu-
matoid arthritisand its use in metaanalysis. J Rheumatol 2000;27:254–260.
55. Rennie D. Improving reports of studies of diagnostic tests: the STARD
initiative. JAMA 2003; 289:89–90.
56. Feinstein AR. Clinimetric perspectives. J Chronic Dis 1987; 40:
635–640.
57. Kirshner B, Guyatt G. A methodological framework for assessing health
indices. J Chronic Dis 1985; 38:27–36.
58. Nunnally J. Psychometric theory. New York: McGraw-Hill; 1978.
59. Patrick DL, Erickson P. Health status and health policy allocation
resources to health care. New York: Oxford University Press; 1993:
198–202.
MILLER ET AL Academic Radiology, Vol 16, No 6, June 2009
746
APPENDIX A. REPORT OF QUALITY IN REPORTING USING STARD ITEMS FOR 18 STUDIES OF JIAOF THE PERIPHERAL JOINTS USING MRI
Quality Study Response
1. Identify the article as a study of
diagnostic accuracy
Herve-Somma et al (31) Yes (construct validity)
Eich et al (40) Yes (construct validity and responsiveness)
Huppertz et al (41) Yes (construct validity and responsiveness)
Murray et al (32) Yes (construct validity)
Remedios et al (33) Yes (construct validity)
Ramsey et al (29) Yes (criterion validity)
Uhl et al (25) Yes (construct validity)
Cakmakci et al (34) Yes (construct validity and responsiveness)
Gylys-Morin et al (28) Yes (construct validity and reliability); one case of
cartilage thinning was confirmed at arthroscopy
(criterion validity)
El-Miedany et al (42) Yes (construct validity)
Argyropoulou et al (30) Yes (construct validity)
Kight et al (35) Yes (construct validity)
Workie et al (38) Yes (construct validity)
Workie and Dardzinski (37) Yes (construct validity); parameters (Ktrans, Vp, Kep)
of three models for arterial input function were
evaluated
Graham et al (36) Yes (construct validity) and reliability
Gardner-Medwin (26) Yes (construct validity)
Nistala et al (27) Yes (construct validity and reliability)
Workie et al (39) Yes (construct validity)
2. State the research questions or study
aims, such as estimating diagnostic
accuracy or comparing accuracy
between tests or across participant
groups
Herve-Somma et al (31) To evaluate the role of contrast MRI in diagnosis,
staging, and planning treatment of JRA
Eich et al (40) To evaluate the use of radiography, ultrasound, and
MRI in assessment of affected knee and hips before
and after intra-articular joint injection
Huppertz et al (41) To assess efficacy and potential toxicity of
intra-articular corticosteroids therapy
Murray et al (32) To determine the value of contrast enhancement in
MRI diagnosis of hip joint disease with JIA
Remedios et al (33) To compare clinical evaluation of hind foot synovitis
with enhanced MRI in children with JIA
Ramsey et al (29) To compare clinical evaluation of hind foot synovitis
with contrast-enhanced MR and evaluate the
efficacy of intra-articular steroid injection
Uhl et al (25) To determine sensitivity and specificity of MR in
diagnosis of JIA
Cakmakci et al (34) To establish correlation between clinical status and
three-dimensional fast-spin contrast MRI in
response to treatment
Gylys-Morin et al (28) To determine MRI findings in early JRA
El-Miedany et al (42) To assess the role of ultrasound vs MRI in inflammation
in JIA of the knee
Argyropoulou et al (30) To establish the role of MRI in the assessment of hip
joint involvement in clinical subtypes with JIA
Academic Radiology, Vol 16, No 6, June 2009 JIA: QUALITY OF REPORTING OF DIAGNOSTIC ACCURACY OF MRI
747
APPENDIX A (Continued)
Quality Study Response
Kight et al (35) To examine MRI T2 relaxation times in the
weight-bearing cartilage of the distal femur in
healthy children and children with JRA
Workie et al (38) Not reported
Workie and Dardzinski (37) Not reported
Graham et al (36) To assess feasibility of measuring synovial volume in
the hand/wrist in polyarticular JIA
Gardner-Medwin (26) To evaluate if MRI of clinical unaffected joints is more
sensitive than clinical assessment in identifying risk
patients
Nistala et al (27) To assess diagnostic performance of clinical vs hip
MRI and to determine clinical and serological
predictors of MRI diagnosed hip arthritis
Workie et al (39) Utility of dynamic contrast-enhanced MRI based on
pharmacokinetic modeling to evaluate disease
activity in the knee and to correlate with clinical
findings
3. The study population: inclusion and
exclusion criteria, setting, and location
where data were collected
Herve-Somma et al (31) Attending pediatric rheumatology clinic as outpatient;
study population was well reported (age and
gender); setting and location were not reported
Eich et al (40) EULAR criteria for JIA and one of the following: (1)
failure of systemic treatment and physiotherapy at 3
mo, (2) local growth, (3) popliteal cyst; study
population was well reported (age and gender);
setting and location were not reported
Huppertz et al (41) Consent for intra-articular corticosteroids
administration; population age and gender were
reported; location was partially reported
Murray et al (32) Diagnose according to juvenile chronic arthritis criteria;
age and gender reported; location not reported
Remedios et al (33) Age and gender reported; no inclusion criteria or
location reported
Ramsey et al (29) Country hospital in Canada; included all patients
referred with presumptive diagnosis of clinical
monoarthritis; study population was well described
(age, gender)
Uhl et al (25) Children included were followed for $1 y and
diagnosed to have JIA of the knee by an experienced
pediatric rheumatologist; age and gender reported;
location of the study not included
Cakmakci et al (34) Patients attending pediatric rheumatology clinic as
outpatients; study population (age, gender)
included; location of the study was not reported
Gylys-Morin et al (28) Included according to ACR criteria, clinical evident
arthritis in at least one knee, disease duration of <1 y;
children were excluded if needed sedation for
imaging or if injection of intra-articular steroids in the
affected knee; age and gender reported; referred
from rheumatology clinic
MILLER ET AL Academic Radiology, Vol 16, No 6, June 2009
748
APPENDIX A (Continued)
Quality Study Response
El-Miedany et al (42) Included according to ILAR criteria; excluded if
injection of intra-articular steroids in the affected
knee; study population (age and gender); location of
the study reported
Argyropoulou et al (30) Included according to ILAR criteria; study population
(age and sex) reported; location not reported
Kight et al (35) Included according to ACR criteria, active arthritis
documented by rheumatologist in at least one
examination, disease duration 2–7 y, girls, and age
4.9–10.8 y; excluded girls who needed sedation for
imaging and if motion; location reported
Workie et al (38) Inclusion criteria: history of JRA; study population (age
and gender) reported; location not reported
Workie and Dardzinski (37) Included children with history of JRA and good
popliteal artery signal enhancement; study
population (age and gender) reported; location not
reported
Graham et al (36) Inclusion criteria: polyarticular disease, ACR criteria (at
least three active joints, one hand); excluded
children who needed sedation; study population
(age and gender) and location of the study reported
Gardner-Medwin (26) Arthritis criteria for monoarthritis; Age and gender
reported; tertiary center
Nistala et al (27) ILAR criteria; disease duration >6 mo; study (age and
gender) reported; location not reported
Workie et al (39) ILAR criteria for active arthritis; study population (age
and gender) reported; location not reported
4. Participant recruitment: was
recruitment based on presenting symptoms,
results from previous test, or the fact that the
participants had received the index test or the
reference standard?
Herve-Somma et al (31) Not reported
Eich et al (40) Presenting symptoms
Huppertz et al (41) Need for administration of intra-articular steroids
Murray et al (32) Not reported
Remedios et al (33) Presenting symptoms (pain and swelling)
Ramsey et al (29) Presenting symptoms
Uhl et al (25) Partial reported
Cakmakci et al (34) Not reported
Gylys-Morin et al (28) Presenting symptoms
El-Miedany et al (42) Presenting symptoms
Argyropoulou et al (30) Partial reported
Kight et al (35) Patients with JIA recruited by chart review and mail;
controls were children of hospital personnel and
healthy children who respond from an advertisement
Workie et al (38) History of JRA
Workie and Dardzinski (37) Not reported
Graham et al (36) Presenting symptoms
Gardner-Medwin (26) Presenting symptoms (monoarthritis)
Nistala et al (27) Presenting symptoms (hip pain)
Workie et al (39) Presenting symptoms
Academic Radiology, Vol 16, No 6, June 2009 JIA: QUALITY OF REPORTING OF DIAGNOSTIC ACCURACY OF MRI
749
APPENDIX A (Continued)
Quality Study Response
5. Participant sampling: was the study population
a consecutive series of participants defined by
selection criteria in items 3 and 4?
Herve-Somma et al (31) Not reported
Eich et al (40) Consecutive patients
Huppertz et al (41) Consecutive patients
Murray et al (32) Not reported
Remedios et al (33) Consecutive patients
Ramsey et al (29) Consecutive patients
Uhl et al (25) Not reported
Cakmakci et al (34) Consecutive patients
Gylys-Morin et al (28) Not mentioned but likely consecutive patients
El-Miedany et al (42) Consecutive patients
Argyropoulou et al (30) Consecutive patients
Kight et al (35) Not reported
Workie et al (38) Not reported
Workie and Dardzinski (37) Not reported
Graham et al (36) Not reported
Gardner-Medwin (26) Consecutive patients
Nistala et al (27) Partial (no information about consecutive patients)
Workie et al (39) Partial (no information about consecutive patients)
6. Data collection: was data collection planned
before the index test and reference standard
were performed (prospective study) or after
(retrospective study)?
Herve-Somma et al (31) Retrospective
Eich et al (40) Prospective
Huppertz et al (41) Prospective
Murray et al (32) Not stated
Remedios et al (33) Prospective
Ramsey et al (29) Retrospective
Uhl et al (25) Not stated
Cakmakci et al (34) Not stated
Gylys-Morin et al (28) Prospective
El-Miedany et al (42) Prospective
Argyropoulou et al (30) Not stated
Kight et al (35) Prospective
Workie et al (38) Prospective
Workie and Dardzinski (37) Not stated
Graham et al (36) Prospective
Gardner-Medwin (26) Prospective
Nistala et al (27) Retrospective
Workie et al (39) Prospective
7. The reference standard and its rationale Herve-Somma et al (31) Unenhanced MRI, test measure; enhanced MRI,
reference standard; or x-ray, test measure;
unenhanced and enhanced MRI, reference
standards
Eich et al (40) MRI, US, and clinical findings, tests; follow-up,
reference standard
Huppertz et al (41) For responsiveness: MRI and clinical assessment
(tests), follow-up (reference standard); for construct
validity: MRI (test), clinical assessment (reference
standard)
Murray et al (32) Unenhanced MRI, test measure; enhanced MRI,
reference standard
MILLER ET AL Academic Radiology, Vol 16, No 6, June 2009
750
APPENDIX A (Continued)
Quality Study Response
Remedios et al (33) MRI, test measure (scoring system); clinical findings,
reference standard
Ramsey et al (29) Responsiveness of clinical findings to NSAIDs and
intra-articular steroids: clinical response confirmed
by MRI n = 1/4; accuracy of MRI: MRI diagnosis
confirmed by arthroscopy n = 5/5; accuracy of x-rays
compared by MRI false-negatives = 620; false-
positives = 1/20
Uhl et al (25) MRI, test measure; clinical findings, reference standard
for discrimination of JIA vs non-JIA knees
Cakmakci et al (34) MRI and clinical findings, tests; follow-up, reference
standard
Gylys-Morin et al (28) Evaluative role among internal items: MRI, test
measure (40 items) and between MRI items and
clinical synovitis; discriminative role: MRI for
discrimination between JIA and control knees
El-Miedany et al (42) Evaluative role comparing unenhanced (test) vs
enhanced MRI (reference standard): extent of
pannus, joint effusion; evaluative role comparing
x-rays, US, unenhanced MRI (tests) and enhanced
MRI (reference standard): cartilage destruction;
discriminative role: MRI vs US for discrimination
between JIA and control subjects: 9 structural items
Argyropoulou et al (30) MRI, test measure (scoring system); clinical findings,
reference standard
Kight et al (35) Discriminative role of MRI (test); clinical assessment
(JIA vs healthy children): reference standard
Workie et al (38) Evaluative role of MRI, test measure (signal
enhancement patterns) compared with clinical
measures (CHAQ-Disability by parents and total
knee scores [swelling, tenderness and limitation]):
functional measure, reference standard, and
evaluative role (internal relationship) of MRI
components (enhancement rates of synovium vs
femoral physis)
Workie and Dardzinski (37) Not applicable: assessment of differences/relationship
between parameters of three models
Graham et al (36) MRI, test measure (total synovial volume); clinical
scores, reference standard
Gardner-Medwin (26) Follow-up: reference standard
Nistala et al (27) MRI, outcome measure, reference standard; clinical
and laboratory findings, predictor measures
Workie et al (39) MRI, test measure (scoring system); clinical findings,
reference standard
8. Technical specifications of material and
methods involved, including how and when
measurements were taken, and/or cite
references for index test and reference
standard
Herve-Somma et al (31) Methods (MRI) well described
Eich et al (40) Methods (MRI) well described
Huppertz et al (41) Methods (MRI) well described
Murray et al (32) Methods (MRI) well described
Remedios et al (33) Methods (MRI) partially described
Academic Radiology, Vol 16, No 6, June 2009 JIA: QUALITY OF REPORTING OF DIAGNOSTIC ACCURACY OF MRI
751
APPENDIX A (Continued)
Quality Study Response
Ramsey et al (29) Methods (MRI) well described
Uhl et al (25) Methods (MRI) partially described
Cakmakci et al (34) Methods (MRI) well described
Gylys-Morin et al (28) Methods (MRI) well described
El-Miedany et al (42) Methods (MRI) well described
Argyropoulou et al (30) Methods (MRI) well described
Kight et al (35) Methods (MRI) well described
Workie et al (38) Methods (MRI) well described
Workie and Dardzinski (37) Methods (MRI) well described
Graham et al (36) Methods (MRI) well described
Gardner-Medwin (26) Methods (MRI) well described
Nistala et al (27) Methods (MRI) well described
Workie et al (39) Methods (MRI) well described
9. Definition of and rationale for the units, cutoffs,
and/or categories of the results of the index
tests and the reference standard
Herve-Somma et al (31) Not reported
Eich et al (40) Not applicable
Huppertz et al (41) Not applicable
Murray et al (32) Not reported
Remedios et al (33) Not reported
Ramsey et al (29) Not reported
Uhl et al (25) Not reported
Cakmakci et al (34) Not applicable
Gylys-Morin et al (28) Pettersson score (radiographs)
El-Miedany et al (42) Not reported
Argyropoulou et al (30) Not reported
Kight et al (35) Not reported
Workie et al (38) Not reported
Workie and Dardzinski (37) Not reported
Graham et al (36) Not reported
Gardner-Medwin (26) Not reported
Nistala et al (27) Global assessment of overall disease activity (VAS-
PGA), CHAQ, and VAS global
Workie et al (39) Not reported
10. The number, training, and expertise of the
persons executing and reading the index test
and the reference standard
Herve-Somma et al (31) Independent observer review the MRI (no further
information was given)
Eich et al (40) Not reported
Huppertz et al (41) Not reported
Murray et al (32) Two radiologists review all the images; no information
about the clinicians; no details on reviewers
expertise given
Remedios et al (33) Two radiologists review all the images; no information
about the clinicians; no details on reviewers
expertise given
Ramsey et al (29) Two pediatric radiologists with experience in
musculoskeletal MRI, one blinded to clinical history;
differences were review in consensus
Uhl et al (25) Five independent radiologists without knowledge of
the clinical findings; no information about the
clinicians
Cakmakci et al (34) An independent observer; no further information given
MILLER ET AL Academic Radiology, Vol 16, No 6, June 2009
752
APPENDIX A (Continued)
Quality Study Response
Gylys-Morin et al (28) Two radiologists blinded to clinical details read the
MRI; in case of disagreement, a third radiologist
provided independent interpretation; no further
information given
El-Miedany et al (42) Not reported
Argyropoulou et al (30) One radiologist (specified) review blinded to clinical
evaluation; no further information given
Kight et al (35) Operators involved in data analysis and image
evaluation were blinded to study group; no further
information given
Workie et al (38) Not reported
Workie and Dardzinski (37) Not reported
Graham et al (36) Rheumatologist for clinical inclusion, radiologist for
MRI interpretation
Gardner-Medwin (26) MRI was reviewed by two consultant pediatric
radiologist (expertise specified)
Nistala et al (27) Patients examined by a pediatric rheumatologist or an
experienced pediatric rheumatologist in trainee (not
otherwise specified); MRI review by two pediatric
radiologist (names specified)
Workie et al (39) Not reported
11. Whether or not the readers of the index
tests and reference standard were blind to
the results of the other test and describe any
other clinical information available to
the readers
Herve-Somma et al (31) Not reported
Eich et al (40) Not reported
Huppertz et al (41) Not reported
Murray et al (32) Partial
Remedios et al (33) Not reported
Ramsey et al (29) Partial
Uhl et al (25) Partial
Cakmakci et al (34) Not reported
Gylys-Morin et al (28) Radiologist were blinded to details of the clinical data
El-Miedany et al (42) Not reported
Argyropoulou et al (30) One radiologist review blinded to clinical evaluation
Kight et al (35) Operators involved in data analysis and image
evaluation were blinded to clinical information
Workie et al (38) Not reported
Workie and Dardzinski (37) Not reported
Graham et al (36) Rheumatologist were blinded to result of imaging;
radiologists were blinded to the results of clinical
assessment
Gardner-Medwin (26) Clinical examiners were blinded to MRI results
Nistala et al (27) Pediatric radiologists were blinded to clinical details
Workie et al (39) Not mentioned
12. Methods for calculating or comparing
measures of diagnostic accuracy, and the
statistical methods used to quantify uncertainty
Herve-Somma et al (31) Degree of cartilage destruction (MRI); c2 and
Wilcoxon tests
Eich et al (40) Changes over time; no statistical methods reported
Huppertz et al (41) Not applicable
Murray et al (32) Visualization of pannus; Wilcoxon test
Remedios et al (33) Correlation between MRI signal and clinical outcome
measures; no statistical methods reported
Academic Radiology, Vol 16, No 6, June 2009 JIA: QUALITY OF REPORTING OF DIAGNOSTIC ACCURACY OF MRI
753
APPENDIX A (Continued)
Quality Study Response
Ramsey et al (29) Clinical (MRI: reference standard; n = 4), x-rays (MRI:
reference standard; n = 20), and MRI (arthroscopy
and biopsy: reference standard; n = 5); no statistical
methods reported
Uhl et al (25) Not mentioned, partial, presence or absence of JIA;
terms sensitivity, specificity, positive and negative
likelihood; ROC curves (no cutoff values reported);
standard error
Cakmakci et al (34) Both MRI and clinical scores over time; Spearman
correlation: association between MRI and clinical
results
Gylys-Morin et al (28) Pettersson score (radiographs), MRI findings; t tests
(continuous variables); nonparametric tests (ordinal
variables); Spearman rank correlation: if at least one
variable was ordinal; Pearson correlation: two
continuous variables; ROC curves: nonparametric
methods (Hanley and McNeil method): cut-offs
reported only for x-rays (Pettersson scores), not for
MRI
El-Miedany et al (42) US and MRI findings in affected and control subjects;
no statistical methods reported
Argyropoulou et al (30) Clinical criteria for disease activity; test of normality of
distribution; two-tailed t tests: differences in MRI
grades between active and inactive patients.
ANOVA: differences in MRI grades between patients
with oligo, poly, and systemic forms
Kight et al (35) T2 values of JIA vs healthy knees; standard deviations
for normalized distances; two-tailed t tests for
analyses of differences in mean T2 relaxation times
Workie et al (38) Correlation between MRI signal and clinical outcome
measures; Pearson correlation coefficient for
parameter between distal femoral physis and
synovium; unpaired t tests
Workie and Dardzinski (37) Poorly described; t tests for differences in signal
enhancement using Ktrans, Kep, and Vp
Graham et al (36) Clinical measures (total hand swelling scores); Pearson
correlation (parametric data) and Spearman
correlation (nonparametric data)
Gardner-Medwin (26) Future clinical outcome; Fisher’s exact test and
Mann-Whitney test
Nistala et al (27) MRI, core outcome variables; Pearson correlation:
association between total MRI scores;
Mann-Whitney test: comparison between hip MRI
scores and clinician-defined active and inactive
groups; c2 test: effect of damage on concordance
between clinical and MRI scores
MILLER ET AL Academic Radiology, Vol 16, No 6, June 2009
754
APPENDIX A (Continued)
Quality Study Response
Workie et al (39) MRI parameters (Ktrans, Kep, Vp); Spearman’s rank
correlation: between MRI and clinical/laboratory
parameters; Wilcoxon test: comparison of results at
different time points
13. Methods for calculating
test reproducibility
Herve-Somma et al (31) Not applicable
Eich et al (40) Not done
Huppertz et al (41) Not applicable
Murray et al (32) Not done
Remedios et al (33) Not done
Ramsey et al (29) Not done
Uhl et al (25) Not done
Cakmakci et al (34) Kappa statistics: not appropriate for agreement
between MRI and clinical results.
Gylys-Morin et al (28) k (interobserver reliability): cutoff values for excellent,
good, and marginal reliability
El-Miedany et al (42) Not applicable
Argyropoulou et al (30) Not done
Kight et al (35) Not done
Workie et al (38) Not applicable
Workie and Dardzinski (37) Not done
Graham et al (36) Interoperator reliability of synovial volume calculation;
coefficient of variation (variation within and between
observers)
Gardner-Medwin (26) Not done
Nistala et al (27) k: intraobserver agreement; not appropriate for
concordance between clinician’s assessment and
MRI results
Workie et al (39) Not applicable
ACR, American College of Radiology; ANOVA, analysis of variance; CHAQ, Childhood Health Assessment Questionnaire; EULAR, European
League Against Rheumatism; ILAR, International League of Associations for Rheumatology; JIA, juvenile idiopathic arthritis; JRA, juvenile
rheumatoid arthritis; MRI, magnetic resonance imaging; NSAID, nonsteroidal anti-inflammatory drug; PGA, patient global assessment;
ROC, receiver-operating characteristic; STARD, Standards for Reporting of Diagnostic Accuracy; VAS, visual analog scale.
Academic Radiology, Vol 16, No 6, June 2009 JIA: QUALITY OF REPORTING OF DIAGNOSTIC ACCURACY OF MRI
755
APPENDIX B. REPORT OF THE QUADAS CRITERIA OF 18 ARTICLES ON ASSESSMENT OF JIA OF PERIPHERAL JOINTS USINGMRI
Item
Herve-
Somma
et al (31)
Eich
et al
(40)
Huppertz
et al
(41)
Murray
et al
(32)
Remedios
et al
(33)
Ramsey
et al
(29)
Uhl
et al
(25)
1. Was the spectrum of patients
representative of the patients in who
will receive the test in practice?
U/C U/C Yes No U/C U/C No
2. Were selection criteria clearly
described?
U/C Yes Yes U/C No Yes No
3. Is the reference standard likely to
correctly classify the target condition?
Yes Yes No Yes Yes Yes Yes
4. Is the time period between reference
standard and index test short enough
to be reasonably sure that the target
condition did not change between the
two tests?
U/C U/C Yes U/C No Yes N/A
5. Did the whole sample or a random
selection of the sample, receive
verification using a reference standard
of diagnosis?
Yes Yes Yes Yes Yes Yes No
6. Did patients receive the same
reference standard regardless of the
index test result?
Yes Yes Yes Yes Yes Yes Yes
7. Was the reference standard
independent of the index test?
No No N/A Yes Yes Yes Yes
8. Was the execution of the index test
described in sufficient detail to permit
replication of the test?
Yes Yes Yes Yes No No U/C
9. Was the execution of the reference
standard described in sufficient detail
to permit replication of the test?
Yes Yes N/A Yes U/C Yes No
10. Were the index test results
interpreted without knowledge of the
results of the reference standard?
No No N/A Yes U/C U/C Yes
11. Were the index reference standard
results interpreted without knowledge
of the results of the index test?
No No N/A No No U/C U/C
12. Were the same clinical data available
when test results were interpreted as
would be available when the test is
used in practice?
Yes Yes Yes Yes Yes Yes Yes
13. Were uninterpretable test results
reported?
No U/C No No No No No
14. Were withdrawals from the study
explained?
N/A Yes U/C No No No No
JIA, juvenile idiopathic arthritis; MRI, magnetic resonance imaging; N/A, not available; QUADAS, Quality Assessment of Studies of
Diagnostic Accuracy Included in Systematic Reviews; U/C, unclear.Items were rated ‘‘yes’’ if adequately described, ‘‘no’’ if not adequately described, ‘‘unclear’’ if the information in the article was unclear, and
‘‘not available’’ if no information was given in the article.
MILLER ET AL Academic Radiology, Vol 16, No 6, June 2009
756
Cakmakci
et al(34)
Gylys-Morin
et al(28)
El-Miedany
et al(42)
Argyropoulou
et al(30)
Kight
et al(35)
Workie
et al(38)
Workie and
Dardzinski(37)
Graham
et al(36)
Gardner-
Medwin(26)
Nistala
et al(27)
Workie
et al(39)
U/C U/C U/C U/C U/C U/C No U/C U/C U/C U/C
U/C Yes U/C No Yes No No Yes No Yes Yes
Yes Yes Yes No Yes No N/A Yes Yes Yes Yes
U/C U/C No No N/A N/A N/A Yes U/C Yes No
Yes Yes U/C No No No No No Yes Yes Yes
Yes Yes Yes Yes Yes N/A N/A Yes Yes Yes Yes
No Yes No U/C Yes No N/A Yes Yes Yes Yes
Yes Yes U/C Yes Yes Yes Yes Yes Yes Yes Yes
Yes Yes Yes Yes N/A N/A N/A Yes Yes Yes Yes
No Yes No No Yes No N/A U/C Yes U/C U/C
No Yes No No Yes No No Yes Yes Yes No
Yes Yes Yes U/C Yes Yes No Yes Yes Yes Yes
No No No No Yes No No No No No No
No N/A No No Yes No No No No Yes No
APPENDIX B. (Continued)
Academic Radiology, Vol 16, No 6, June 2009 JIA: QUALITY OF REPORTING OF DIAGNOSTIC ACCURACY OF MRI
757