ARTHRITIS & RHEUMATISM
Vol. 56, No. 5, May 2007, pp 1512–1520
© 2007, American College of Rheumatology
Longitudinal Performance Evaluation and Validation of
Fixed-Flexion Radiography of the Knee for Detection of
Joint Space Loss
Michael C. Nevitt,1Charles Peterfy,2Ali Guermazi,2David T. Felson,3Jeff Duryea,4
Thasia Woodworth,5Hepei Chen,6Kent Kwoh,7and Tamara B. Harris6
Objective. The ability of nonfluoroscopically
guided radiography of the knee to assess joint space loss
is an important issue in studies of progression and
treatment of knee osteoarthritis (OA), given the practi-
cal limitations of protocols involving fluoroscopically
guided radiography of the knee. We evaluated the ability
of the nonfluoroscopically guided fixed-flexion radiog-
raphy protocol to detect knee joint space loss over 3
Methods. We assessed the same-day test–retest
precision for measuring minimum joint space width
(JSW), the sensitivity for detection of joint space loss
using serial films obtained a median of 37 months
(range 23–47 months) apart, and the relationship of
joint space loss to radiographic and magnetic resonance
imaging (MRI) measures of knee OA. Participants were
men and women (ages 70–79 years) with knee pain who
were participating in the Health, Aging, and Body
Composition Study. We assessed baseline radiographic
OA and measured JSW using a computerized algorithm.
Serial knee MRIs obtained over the same interval were
evaluated for cartilage lesions.
Results. A total of 153 knees were studied, 35% of
which had radiographic OA at baseline. The mean ? SD
joint space loss for all knees over 3 years was 0.24 ? 0.59
mm (P < 0.001 for change). In knees with OA at
baseline, the mean ? SD joint space loss over 3 years
was 0.43 ? 0.66 mm (P < 0.001), and in knees with joint
space narrowing at baseline, joint space loss was 0.50 ?
0.67 mm (P < 0.001). Joint space loss and its standard-
ized response mean increased with the severity of base-
line joint space narrowing and with the presence of
cartilage lesions at baseline and worsening during fol-
Conclusion. Radiography of the knee in the fixed-
flexion view provides a sensitive and valid measure of
joint space loss in multiyear longitudinal studies of knee
OA, without the use of fluoroscopy to aid knee position-
Serial radiographs remain the gold standard (and
regulatory agency–approved) method of assessing carti-
lage loss in osteoarthritis (OA) of the knee. Although
radiography does not permit direct visualization of car-
tilage, the minimum medial tibiofemoral interbone dis-
tance, or joint space width (JSW), parallels articular
cartilage thickness in knees radiographed in a moder-
ately flexed, weight-bearing position (1). Weight-bearing
is essential to displace intervening joint fluid and bring
the opposing cartilage surfaces into contact. Flexion of
Supported in part by the Intramural Research Program of the
NIH (National Institute on Aging contracts N01-AG-6-2101, N01-AG-
6-2103, and N01-AG-6-2106).
1Michael C. Nevitt, PhD, University of California, San Fran-
cisco;2Charles Peterfy, MD, PhD, Ali Guermazi, MD: Synarc, San
Francisco, California;3David T. Felson, MD, MPH: Boston University
Medical Center, and Boston University School of Medicine, Boston,
Massachusetts;4Jeff Duryea, PhD: Brigham and Women’s Hospital,
and Harvard University Medical School, Boston, Massachusetts;5Tha-
sia Woodworth, MD: Roche, Rheinfelden, Switzerland;6Hepei Chen,
MD, Tamara B. Harris, MD: National Institute on Aging, Bethesda,
Maryland;7Kent Kwoh, MD: University of Pittsburgh, and Pittsburgh
Department of Veterans Affairs Health Care System, Pittsburgh,
Dr. Peterfy owns stock and/or holds stock options in Synarc.
Synarc provides centralized image and biochemical analysis services
and subject recruitment for clinical trials to pharmaceutical, biotech-
nology, and medical device companies, including but not limited to
Abbott, Alexion, Allergan, Amgen, Astra-Zeneca, Aventis, Bayer,
Biogen Idec, Bristol-Myers Squibb, Centocor, Elan, Eli Lilly, Genen-
tech, Genzyme, GlaxoSmithKline, Novartis, Pfizer, Procter & Gamble,
Roche, Serono, Servier, and Wyeth. Dr. Woodworth owns stock and/or
holds stock options in Pfizer and Novartis.
Address correspondence and reprint requests to Michael C.
Nevitt, PhD, Department of Epidemiology and Biostatistics, Univer-
sity of California, San Francisco, 185 Berry Street, Lobby 4, Suite 5700,
San Francisco, CA 94107. E-mail: email@example.com.
Submitted for publication November 22, 2006; accepted in
revised form January 24, 2007.
the knee is required to avoid artifactual increases in
apparent cartilage thickness that occur when the knee is
fully extended (2,3). In addition, flexion allows imaging
of a more posterior sectional plane of the femoral
cartilage, the location of peak load during walking and
stair climbing and a location of early cartilage loss in OA
Joint space loss, a key measure of disease pro-
gression in patients with knee OA, is estimated by
comparing measurements of JSW between serial radio-
graphs of the knee. Variability in positioning the knee
during radiography can have a substantial influence on
measured JSW (7–10). Several radiography protocols
use fluoroscopic guidance as an aid to standardize and
reproduce knee positioning (1,9,11), and these protocols
have demonstrated excellent test–retest precision for
measuring medial tibiofemoral JSW. In addition, non-
fluoroscopically guided knee radiography protocols have
been developed that fix the radioanatomic position of
the knee in a manner designed to be reproducible
between examinations, and these have also been shown
to have excellent test–retest precision for measuring
Short-term measurement precision demonstrated
under controlled conditions, however, does not guaran-
tee that a method can precisely and accurately assess
joint space loss in a large, multicenter, longitudinal study
lasting several years. In these circumstances, a variety of
factors can impede reproducible radiographic technique
and positioning. These factors include changes in radiog-
rapher performance due to staff turnover, deterioration
of morale and drift in technique, OA-related changes in
anatomy (deterioration of ligamentous support, increas-
ing mediolateral laxity, flexion contracture), changes in
leg musculature, and the waxing and waning of knee
pain. Consequently, the ability of a radiography tech-
nique to detect joint space loss must be assessed directly
from repeat films obtained over time intervals during
which real changes in joint space and other potentially
confounding changes are likely to occur.
The sensitivity of fluoroscopically positioned
flexed-knee radiography to detect significant joint space
loss in knees with OA has been demonstrated in longi-
tudinal studies of 12–30 months duration (5,11,15–17).
However, there are no similar longitudinal data for the
nonfluoroscopically guided flexed-knee radiography
protocols. Such data would be of great interest, because
the use of fluoroscopy presents substantial practical,
logistic, budgetary, and other challenges, including the
technical difficulty of the procedure, limited availability
of suitable fluoroscopy equipment, and additional radi-
ation exposure of the patient (3,18).
We evaluated the performance of the nonfluoro-
scopically guided fixed-flexion radiography protocol (13)
for assessment of medial tibiofemoral joint space loss.
Our objectives were to evaluate short-term test–retest
precision for measuring JSW, to determine the sensitiv-
ity of this method for detection of joint space loss over 3
years, and to evaluate the association of joint space loss
with measures of radioanatomic positioning and OA
severity, the latter including baseline x-ray findings and
tibiofemoral cartilage damage assessed with magnetic
resonance imaging (MRI).
PATIENTS AND METHODS
Patients. Participants were drawn from the Health,
Aging, and Body Composition (Health ABC) Study, a longi-
tudinal study of weight-related diseases that contribute to
disability in 3,075 men and women, ages 70–79 years, who were
able to walk at least one-quarter of a mile and climb 10 steps
without difficulty at baseline. The study was approved by the
institutional review boards at the University of Tennessee and
the University of Pittsburgh, the 2 clinical sites for the study. In
patients who reported knee OA symptoms in at least 1 knee,
bilateral knee radiographs and knee MRIs were obtained at
the second or third annual clinic visit. Symptoms of OA were
defined as “pain, aching or stiffness on most days for at least 1
month” during the past 12 months, or moderate or worse knee
pain in the last 30 days during any activity, as assessed using the
Western Ontario and McMaster Universities Osteoarthritis
Index pain scale (19). Followup knee radiographs and knee
MRIs were obtained in these patients at the fifth or sixth
annual visit, depending on whether their baseline imaging was
done at the second or third visit. At the followup visit, repeat
knee films (with repositioning) were obtained in 29 patients, in
order to assess short-term test–retest precision for measuring
The present analysis was limited to patients for whom
followup knee radiographs were acquired by October 2002,
and who did not have bilateral end-stage disease (Kellgren/
Lawrence [K/L]  grade 4 or Osteoarthritis Research Soci-
ety International [OARSI] atlas  grade 3 joint space
narrowing) or primarily lateral compartment joint space nar-
rowing in either knee. Three hundred twenty-eight patients
met these criteria, and a random sample of 80 patients was
selected, using a random number generator, for detailed
radiographic measurements as part of a study comparing 3
radiography protocols (22).
Imaging protocols. Radiography. Bilateral posteroante-
rior knee films were obtained with the patient standing, knees
flexed to 20–30 degrees, and feet internally rotated 10 degrees
(13). Knees were imaged together on 14 ? 17–inch film using
a focus-to-film distance of 72 inches. The degree of knee
flexion and foot rotation was fixed for each patient, using a
plexiglass frame (SynaFlexer; Synarc, San Francisco, CA) for
FIXED-FLEXION RADIOGRAPHY FOR ASSESSMENT OF KNEE JOINT SPACE LOSS1513
positioning. The great toes, patellae of both knees, and both
thighs were touching the anterior wall of the frame, which in
turn contacts the bucky tray so that the patellae are within a
few centimeters of the film cassette. The x-ray beam was
angled 10 degrees caudally, centered midway between the 2
knees at the level of the popliteal crease.
Radiography technicians were trained by an experi-
enced research technician using a written operations manual as
instructional material. A half-day training session took place at
each radiology site involved in the study. Technicians were
certified after successfully passing a central review of 10
radiographs obtained with the study protocol. During the
study, 3 certified technicians were at one of the sites, and 5
technicians were at the other site. A half-day on-site refresher
training session was held ?3 months after study start and again
just prior to the start of followup imaging. Radiographs were
reviewed centrally by a trained research assistant, and repeat
films were requested in instances of incomplete anatomic
coverage, poor exposure, poor beam centering, or incorrect
beam angle as indicated by visual assessment of the relative
position of 2 columns of metallic beads on the SynaFlexer
MRI. MRI studies were performed at baseline and
followup, at each clinical center, using a Signa 1.5T whole-body
clinical scanner (General Electric Signa, Milwaukee, WI) with
a standard unilateral, commercial circumferential knee coil.
The “short” protocol consisted of 3 sequences: 1) axial views
were T2-weighted fast spin-echo (FSE) including the entire
patella (acquisition time 30 seconds), 2) sagittal views were
T2-weighted FSE, including the entire synovial cavity, with
frequency-selective fat suppression (acquisition time 4 min-
utes, 30 seconds), and 3) coronal views were T2-weighted FSE
(acquisition time 4 minutes).
Assessment of knee images. Radiographs. The baseline
hard-copy films were evaluated by a single expert reader at
Boston University for K/L grades and OARSI atlas grades (23)
for osteophytes and joint space narrowing in the medial and
lateral compartments. As part of a study comparing 3 radio-
graph acquisition techniques (22), rulers were used on baseline
and followup films of right knees to manually measure the
minimum medial tibiofemoral JSW and mid–medial tibial
plateau alignment, a measure of radioanatomic positioning
defined by the distance between the anterior and posterior
rims of the medial tibial plateau at its center (1). Measure-
ments were made on baseline and followup films individually,
with the reader blinded to chronologic order and without the
reader viewing the paired film. The intraclass correlation
coefficients (ICCs) (interreader and intrareader reliability) for
manual measurements of JSW were 0.98 and 0.90, respectively,
and for tibial plateau alignment were 0.88 and 0.88 (22).
Hard-copy films were digitized using a Lumisys Lumi-
scan laser digitizer (Sunnyvale, CA) with a 100-?m pixel size.
The minimum medial tibiofemoral JSW was measured from
the digitized images using a computerized edge detection
program (24). Bone margins were drawn by the computer
independently on the baseline and followup films. A single
trained reader then reviewed paired films, side by side but
randomly and blinded to chronology, and made manual cor-
rections to the bone margins as needed. Intrareader reliability
(root mean square error [RMSE] coefficient of variance
percentage [CV%]) of minimum JSW using this method is
2.9% for knees with OA and 5.5% for normal knees (24). On
images of the left knee, measurements of midtibial plateau
alignment were made using an electronic caliper. In 25 ran-
domly selected right knees, agreement (ICC) between ruler
and electronic caliper measurements of tibial plateau align-
ment was 0.88. In these same knees, intrareader reliability for
repeat electronic caliper measurements was 0.96.
MRIs. Magnetic resonance images were evaluated for
cartilage damage by trained readers using the Whole-Organ
MRI Score (WORMS) method (25). Using the WORMS
method, cartilage damage is scored on a 7-level ordinal scale,
assessing the articular surfaces of the medial tibia and femur
and the lateral tibia and femur, with anterior, central, and
posterior subregions of each plate scored separately, for a total
of 10 scores that can be summed for a compartment-specific
composite score. A cartilage score of 1 does not represent a
change in morphology but rather a change in signal in cartilage
of otherwise normal morphology. Scores of 2 and 3 represent
similar types of abnormality of the cartilage, involving focal
defects without overall thinning. Therefore, to create a scale
for evaluation of cartilage morphologic change, as in previous
analyses (26,27), we collapsed the original WORMS values of
0 and 1 to a score of 0, the original values of 2 and 3 were
collapsed to a score of 1, and the original values of 4, 5, and 6
were considered to be scores of 2, 3, and 4, respectively, on a
modified 0–4-point scale. A compartment was defined as
worsening if there was an increase of at least 1 point, reflecting
significant morphologic change, in the cartilage score for any
of its subregions. Interreader agreement for the WORMS
cartilage assessments performed in Health ABC was accept-
able, with ICCs of 0.89–0.92 for the medial and lateral
tibiofemoral compartments (28).
Statistical analysis. Analyses of joint space loss were
limited to knees with at least 0.5 mm of minimum JSW at
baseline according to the computerized measurement, because
knees with little or no joint space remaining are not suitable for
studies of joint space loss. Joint space loss was calculated by
subtracting the followup value of the minimum JSW from the
baseline value in the same knee, so that a positive value
indicates a decrease in JSW.
A random-effects analysis of variance (ANOVA)
model was used to determine the reproducibility of JSW
measurements from same-day test–retest knee radiographs
(29). The RMSE was calculated from the ANOVA model, with
the knee as the explanatory variable and JSW as the dependent
variable; the CV% was calculated as (RMSE/mean JSW) ?
100. For comparison with other studies, CV% was also calcu-
lated using the mean and median of the within-pair SDs.
Agreement between measurements was evaluated using the
The significance of change in the minimum JSW
between baseline and followup was evaluated with paired
t-tests. Sensitivity to detect joint space loss was expressed as
the standardized response mean (SRM), calculated as the
mean change in joint space divided by its SD. One-way
ANOVA and general linear models (GLMs) were used to
assess the association of joint space loss with disease charac-
teristics and positioning indicators. Tests for trend across
ordered categoric variables were performed using GLM.
1514NEVITT ET AL
More than one-third of the 80 patients were men,
39% were African American, and 44% had symptoms in
only 1 knee at baseline (Table 1). Median followup was
37 months (range 23–47 months). Radiographic knee
OA, defined as a K/L grade of ?2 or definite osteo-
phytes (OARSI atlas grade ?1), was present in at least
1 knee in 49% of patients at baseline, while in 39% of
the patients, symptoms and osteophytes were in the
same knee. In 2 right knees and 4 left knees, minimum
JSW at baseline was ?0.5 mm (computerized measure-
ment), and 1 right knee was replaced during followup;
these knees were not further analyzed.
Of the 153 knees included in the analysis, 53
(35%) had radiographic OA at baseline. Baseline and
followup computerized measurements of minimum me-
dial tibiofemoral JSW were available for 153 knees, and
paired manual measurements of JSW were available for
77 right knees. In right knees, there was excellent
agreement between the computerized and manual mea-
surements of JSW at baseline (ICC 0.96) and very good
agreement for joint space loss over 3 years (ICC 0.79).
All subsequent results were for the computerized JSW
measurements. Results were nearly identical using man-
ual measurements (data not shown).
For 53 knees with same-day repeat films, the ICC
for JSW was 0.98, the RMSE was 0.23 mm, and the
RMSE CV was 7.4%. The mean and median within-pair
SDs (CVs) were, respectively, 0.15 mm (4.9%) and 0.08
mm (2.6%). Eighty-eight percent of test–retest measure-
ments were within 0.5 mm of each other, and 95% were
within 1.0 mm.
The mean ? SD joint space loss over 3 years in all
knees was 0.24 ? 0.59 mm (Table 2), with a mean ? SD
annual rate of joint space loss of 0.07 ? 0.20 mm.
Pseudo-widening (an increase in JSW of ?0.5 mm over
3 years) occurred in 9 knees (5.9%), while joint space
loss of ?0.5 mm occurred in 44 knees (28.8%). Seven of
the 9 instances of pseudo-widening occurred in knees
with a normal baseline joint space, defined as a JSW
?4.0 mm and joint space narrowing grade 0.
The value for joint space loss over 3 years and its
SRM increased sharply with the degree of baseline
medial compartment joint space narrowing (Table 3).
Compared with the 73% of knees with normal joint
space (grade 0) at baseline, joint space loss was 2.6-fold
greater in knees with grade 1 narrowing at baseline and
nearly 5-fold greater in knees with grade 2 narrowing
(P ? 0.001 for trend across the 3 categories). There was
a trend for increased joint space loss in knees with grade
2 compared with grade 1 narrowing, but this was not
significant. The mean ? SD annual rate of joint space
loss in knees that were narrowed at baseline (n ? 42)
was 0.17 ? 0.23 mm. In knees with K/L grade ?2 or
Characteristics of the 80 study patients*
Age, mean ? SD years
BMI, mean ? SD kg/m2
Radiographic knee OA‡
73.5 ? 3.1
27.8 ? 4.3
* Except where indicated otherwise, values are the percentage. BMI ?
body mass index; OA ? osteoarthritis.
† Pain, aching, or stiffness on most days for at least 1 month in the past
12 months, or moderate or worse knee pain for ?1 activity on the
Western Ontario and McMaster Universities Osteoarthritis Index.
‡ Kellgren/Lawrence grade ?2 or definite osteophytes (Osteoarthritis
Research Society International grade ?1).
and manual measurements of joint space loss over 3 years*
Medial tibiofemoral JSW at baseline, and computerized
(n ? 153)
(n ? 77)
Baseline JSW, mm
Joint space loss, mm
Baseline JSW, mm
Joint space loss, mm
3.66 ? 1.25
0.24 ? 0.59
3.62 ? 1.31
0.26 ? 0.57
3.79 ? 1.40
0.27 ? 0.50
* Except where indicated otherwise, values are the mean ? SD.
Manual measurements were obtained in right knees only. JSW ? joint
space width; SRM ? standardized response mean.
† By paired t-test.
femoral joint space loss over 3 years*
Baseline medial compartment JSN grade and medial tibio-
Baseline JSNJoint space lossSRMP†
Grade 0 (n ? 111)
Grade ?1 (n ? 42)
Grade 1 (n ? 23)
Grade 2 (n ? 19)
0.14 ? 0.53
0.50 ? 0.67‡
0.36 ? 0.76
0.63 ? 0.66§
* Values are the mean ? SD mm (computerized measurement of joint
space loss). JSN ? joint space narrowing; SRM ? standardized
† By paired t-test.
‡ P ? 0.001, grade ?1 versus grade 0.
§ P ? 0.001 for trend, by increasing grade of JSN; P ? 0.15, grade 2
versus grade 1.
FIXED-FLEXION RADIOGRAPHY FOR ASSESSMENT OF KNEE JOINT SPACE LOSS1515
definite osteophytes (n ? 53), the mean ? SD joint
space loss over 3 years was 0.43 ? 0.66 mm (P ? 0.001;
SRM 0.65), significantly (P ? 0.01) greater than that in
knees without radiographic OA, and annual joint space
loss was 0.14 ? 0.22 mm.
In 122 knees with serial MRIs scored for cartilage
lesions, those that did not have medial compartment
cartilage lesions did not have significant joint space loss,
those with baseline lesions that worsened had the high-
est rate of joint space loss, while knees with baseline
lesions that did not worsen had an intermediate rate of
joint space loss (P ? 0.017 for trend across the 3
categories) (Table 4). The difference in joint space loss
between knees with lesions that worsened and those with
lesions that did not worsen was not significant. Higher
baseline composite cartilage scores and increases in
composite cartilage scores during followup were moder-
ately correlated with greater joint space loss (r ? ?0.33,
P ? 0.0002 and r ? ?0.26, P ? 0.01, respectively).
There was no significant association of joint
space loss with the degree of alignment of the tibial
plateau rims at baseline and followup (Table 5). The
greatest joint space loss (and SRM value) was observed
in knees with parallel alignment at both time points, but
only 12% of knees were in this group, and the values
were not significantly different from those in knees with
This study evaluated assessment of medial tib-
iofemoral minimum joint space loss, a primary outcome
measure for knee OA, with the nonfluoroscopically
guided fixed-flexion radiography protocol, a posteroan-
terior view that fixes the degree of knee flexion and
rotation using a positioning frame and angles the x-ray
beam caudally 10 degrees (13). Our results demonstrate
the sensitivity of this protocol for detection of joint space
loss in a 3-year longitudinal study of knees with OA. We
observed a significant decrease in joint space in a sample
that included knees with and those without radiographic
OA and knees with and those without pain. Knees with
radiographic evidence of OA at baseline, either osteo-
phytes or joint space narrowing, and those with MRI
evidence of morphologic lesions in cartilage had signif-
icantly greater joint space loss than knees without these
We confirmed results of a previous study (30)
showing that the short-term test–retest precision for the
measurement of JSW for repeat fixed-flexion radio-
graphs acquired in a field setting is comparable with that
achieved in similar settings with the fluoroscopically
space loss over 3 years*
Medial tibiofemoral compartment MRI scores and joint
Cartilage lesion scoreJoint space lossSRMP†
Baseline grade 0 (n ? 56)
Baseline grade ?1 (n ? 66)
Baseline grade ?1, no
worsening (n ? 47)
Baseline grade ?1,
worsening (n ? 19)
0.08 ? 0.49
0.34 ? 0.71‡
0.30 ? 0.64
0.42 ? 0.86§0.49 0.044
* Values are the mean ? SD mm (computerized measurement of joint
space loss). Cartilage lesion scores are based on the modified Whole-
Organ Magnetic Resonance Imaging (MRI) Score method (0–4 scale),
where worsening within a compartment is defined as a score increase
of ?1 in any subregion. SRM ? standardized response mean.
† By paired t-test.
‡ P ? 0.016, baseline grade ?1 versus grade 0.
§ P ? 0.017 for trend across categories (no baseline lesions; versus
baseline lesions and no worsening; versus baseline lesions and wors-
ening); P ? 0.522 for baseline grade ?1 with worsening versus baseline
grade ?1 with no worsening.
(FU) on medial tibiofemoral joint space loss over 3 years*
Effect of mid–medial compartment tibial plateau rim alignment at baseline (BL) and followup
Medial tibial rim alignmentJoint space loss SRMP†
Both BL and FU alignment ?1.5 mm (n ? 18)
Either BL or FU alignment ?1.5 mm, difference
between BL and FU ?1.5 mm (n ? 93)
Either BL or FU alignment ?1.5 mm, difference
between BL and FU ?1.5 mm (n ? 42)
0.41 ? 0.70
0.23 ? 0.58
0.21 ? 0.590.36 0.028
* Values are the mean ? SD mm (computerized measurement of joint space loss). Mid–medial
compartment tibial rim alignment is defined as the absolute value of the distance between the anterior and
posterior rims of the medial tibial plateau at its midpoint. There were no significant differences in joint
space loss by category of tibial rim alignment (P ? 0.49 for overall analysis of variance, all pairwise
comparisons, and tests for trend across the 3 categories; P ? 0.20 for aligned [n ? 18] versus not aligned
[n ? 135]). Results were no different after adjustment for baseline joint space narrowing. SRM ?
standardized response mean.
† By paired t-tests.
1516NEVITT ET AL
guided semiflexed view (31) and Lyon schuss view (5)
radiography protocols and the nonfluoroscopically
guided metatarsophalangeal (MTP) protocol (14,32,33).
Short-term test–retest precision does not differentiate
between these alternative knee radiography protocols.
Many factors that influence the ability to reproduce
radioanatomic positioning of the knee, such as pain,
muscle weakness, and ligamentous laxity, can change
substantially over the extended time interval needed to
detect joint space loss. Therefore, acceptable short-term
test–retest precision is an essential but not a sufficient
condition for good performance of knee radiography in
detecting loss of joint space, and this must be assessed
directly in longitudinal studies.
The performance of the fixed-flexion view radi-
ography protocol in terms of the rate of medial tib-
iofemoral joint space loss and sensitivity to change
(SRM) in knees with OA was within the range seen in
longitudinal studies of 12 months or longer using fluo-
roscopically assisted radiography protocols in knees with
OA (Table 6). Because both joint space loss and the
SRM value may increase with duration of followup, the
most informative comparison with our findings is the
30-month study using the fluoroscopically assisted semi-
flexed view protocol (15). To our knowledge, there are
no studies of at least 12 months duration using other
nonfluoroscopically assisted flexed-knee radiography
protocols that demonstrate a loss of joint space. How-
ever, without the benefit of a head-to-head comparison
between different protocols used in the same knees over
the same time interval, it is not possible to definitively
compare their performance. Differences in patient char-
acteristics, duration of followup, and other factors be-
tween studies that use different radiography protocols
may confound differences in the rate and variability of
joint space loss observed with different protocols.
Our study supports the construct validity of as-
sessing joint space loss with the fixed-flexion view radi-
ography protocol, by demonstrating the association of
joint space loss with several factors with which it should
be related. First, in this sample from the general popu-
lation, the rate of joint space loss and SRM value were
substantially greater in knees with baseline osteophytes
(34% of knees) or in knees with joint space narrowing
(29% of knees) compared with knees without these
findings, and both joint space loss and its SRM value
increased with the severity of baseline joint space nar-
rowing. Second, to further validate the radiographically
determined estimates of structural progression, we ex-
amined the relationship between joint space loss and
MRI-based measures of cartilage morphologic lesions.
The rate of joint space loss and SRM value were
associated with the presence of cartilage lesions and with
worsening of cartilage lesion scores during the study.
Radioanatomic positioning of the knee and its
variability between exams is an important factor in
determining the ability of a radiography protocol to
assess knee OA progression (7–9). Conventional stand-
ing radiographs obtained with the knees in full extension
do not provide reproducible measurements of JSW nor
necessarily accurate depictions of cartilage thickness
(2,5,8,11,34). In the fully extended knee, interposition of
the meniscus and contact between bony prominences of
the articulation may cause apparent JSW to be discrep-
ant with cartilage thickness (3). In addition, femoral
cartilage loss in OA varies in the anteroposterior direc-
tion (4,6). Changes in the degree of knee extension
attained may alter the area of the femoral condyle in
contact at the load-bearing axis of the knee as well as the
effect of the meniscus and bony prominences, potentially
resulting in changes in JSW despite no corresponding
change in cartilage thickness.
In nonfluoroscopically guided fixed-flexion and
MTP (12) views and fluoroscopically positioned semi-
flexed (8) and Lyon schuss views (5), the knees are
positioned in flexion in a manner designed to be repro-
knees from the present study*
Joint space loss in OA knees from studies using fluoroscopically guided radiography protocols and
Fluoro Lyon schuss (16)
Fluoro semiflexed (17)
Fluoro semiflexed (31)
Fluoro Lyon schuss (11)
Fluoro semiflexed (15)
Non-fluoro fixed-flexion (present study)
0.19 ? 0.48
0.12 ? 0.42
0.09 ? 0.31
0.24 ? 0.50
0.45 ? 0.70
0.43 ? 0.66
* Except where indicated otherwise, values are the mean ? SD mm. OA ? osteoarthritis; SRM ? standardized
response mean; fluoro ? fluoroscopically guided; ACR ? American College of Rheumatology criteria.
FIXED-FLEXION RADIOGRAPHY FOR ASSESSMENT OF KNEE JOINT SPACE LOSS1517
ducible over time. The approach to this varies, as does
the resulting degree of knee flexion. The protocols for
fixed-flexion and Lyon schuss views use variants of a
positioning frame to attempt to standardize and repro-
duce knee flexion, and for these views the amount of
flexion attained averages ?25–30 degrees (5,13). The
protocol for MTP views attempts to achieve this by
positioning the feet and knees in a standardized location
relative to the cassette (or x-ray detector), with resulting
knee flexion reported to average ?7–10 degrees (12,35).
The protocol for semiflexed views, in contrast, attempts
to standardize and reproduce both knee flexion and the
sectional plane of the joint space imaged, by having the
patient flex and extend the knee under fluoroscopic
guidance until the anterior and posterior rims of the
medial tibial plateau appear to be parallel with the x-ray
beam, resulting in average flexion of 5–7 degrees (3,8).
Due to biologic variability in the inclination of the tibial
plateau (36,37), the amount of knee flexion needed to
align the tibial rims varies from knee to knee and
includes minimal or no flexion.
The degree of knee flexion typically achieved
with a given protocol may influence sensitivity to joint
space loss, since a more posterior region of the femoral
condyle corresponds to the region of peak load on the
femoral cartilage during walking and stair climbing and
of early cartilage loss in OA (4,5,6). Despite the poten-
tial importance of this variable, no study has compared
the amount, variability, and reproducibility of knee
flexion attained with different protocols when used in
the same knees or examined their effects on sensitivity
The fixed-flexion protocol does not attempt to
align the x-ray beam with the tibial plateau rims of each
knee, as is done in the 2 fluoroscopy protocols, but
rather fixes the angulation of the tibia using the posi-
tioning frame and sets the x-ray beam at a constant
10-degree caudal angle, based on fluoroscopy studies
showing that a mean beam angle of 9–11 degrees caudal
is required to align the mid–medial tibial plateau rims
with fixed-flexion positioning (13,36). As a result, the
x-ray beam is aligned in parallel with the tibial plateau in
some knees but not in others. As expected, fixed flexion
did not achieve parallel alignment of the tibial plateau in
a high proportion of knees in this study but did achieve
consistent alignment over 3 years to an extent similar to
that seen in some studies with fluoroscopically guided
Some studies have shown that sensitivity to joint
space loss is greater in knees with serial radiographs that
show parallel alignment of the tibial plateau rims com-
pared with other knees with serial radiographs, acquired
with the same protocol, that do not exhibit parallel
alignment (11,16). Although we observed a trend for
increased joint space loss in knees with parallel com-
pared with nonparallel alignment, the difference was not
statistically significant, and there was a small number of
knees with parallel alignment on both films. It is possible
that varying the beam angle under fluoroscopic guidance
(as is done in the Lyon schuss protocol) to achieve
parallel alignment in a larger number of knees would
have provided greater sensitivity to joint space loss in
our subjects. Head-to-head comparison studies are
needed to determine the independent effect of parallel
alignment of the tibial plateau rims. Such information
will facilitate weighing all the advantages and disadvan-
tages of the alternative protocols, including radiation
exposure from fluoroscopy, training and quality assur-
ance burdens, the need for magnification correction of
JSW, and costs (3).
Our study has several strengths. The radiographs
were obtained in the field setting of a multicenter study
with an intensity of radiographer training and ongoing
quality assurance that is attainable in most studies;
therefore, our results should be generalizable to typical
uses of this protocol in clinical research settings. Our
study included knees with and knees without OA at
baseline, and our results are relevant to both clinical
studies of knee OA and epidemiologic studies of the
general population. Independent computerized and
manual measurements of joint space width gave nearly
Our study also has features that limit its general-
izability. We included knees with less baseline minimum
joint space (from 0.5 mm to ?2.0 mm) compared with
that in many clinical trials of knee OA, which typically
require minimum joint space of at least 2.0 mm. The
duration of followup was 3 years, and no interim radio-
graphs were obtained; our results may not apply to
shorter periods of followup. Our results also may not
apply to unilateral radiographs acquired using the fixed-
flexion view, due to differences in centering of the x-ray
beam (13). The number of knees with OA in our study is
relatively small, although it is comparable with that in
several studies evaluating the sensitivity of other meth-
ods (11,16,33); a larger sample would provide more
precise estimates of joint space loss. Finally, our study
was not a head-to-head comparison of alternative pro-
tocols, which is the only way to definitively compare
In conclusion, fixed-flexion knee radiography
provides a sensitive and valid measure of joint space loss
1518NEVITT ET AL
in long-term (3-year) longitudinal studies of knee OA,
without the use of fluoroscopy to aid in knee positioning.
We thank Burton Sack, MD, Boston University.
Dr. Nevitt had full access to all of the data in the study and
takes responsibility for the integrity of the data and the accuracy of the
Study design. Nevitt, Peterfy, Felson, Duryea, Woodworth, Harris.
Acquisition of data. Nevitt, Guermazi, Kwoh, Harris.
Analysis and interpretation of data. Nevitt, Peterfy, Guermazi, Dur-
yea, Woodworth, Chen, Kwoh.
Manuscript preparation. Nevitt, Peterfy, Guermazi, Felson, Duryea,
Woodworth, Kwoh, Harris.
Statistical analysis. Chen.
1. Buckland-Wright JC, Macfarlane DG, Lynch JA, Jasani MK,
Bradshaw CR. Joint space width measures cartilage thickness in
osteoarthritis of the knee: high resolution plain film and double
contrast macroradiographic investigation. Ann Rheum Dis 1995;
2. Buckland-Wright JC. Radiographic assessment of osteoarthritis:
comparison between existing methodologies. Osteoarthritis Carti-
3. Buckland-Wright C. Review of the anatomical and radiological
differences between fluoroscopic and non-fluoroscopic positioning
of osteoarthritic knees. Osteoarthritis Cartilage 2006;14 Suppl
4. Messieh SS, Fowler PJ, Munro T. Anteroposterior radiographs of
the osteoarthritic knee. J Bone Joint Surg Br 1990;72:639–40.
5. Piperno M, Hellio Le Graverand MP, Conrozier T, Bochu M,
Mathieu P, Vignon E. Quantitative evaluation of joint space width
in femorotibial osteoarthritis: comparison of three radiographic
views. Osteoarthritis Cartilage 1998;6:252–9.
6. Andriacchi TP, Mundermann A, Smith R, Alexander EJ, Dyrby
CO, Koo S. A framework for the in vivo pathomechanics of
osteoarthritis at the knee. Ann Biomed Eng 2004;32:447–57.
7. Fife RS, Brandt KD, Braunstein EM, Katz BP, Shelbourne KD,
Kalasinski LA, et al. Relationship between arthroscopic evidence
of cartilage damage and radiographic evidence of joint space
narrowing in early osteoarthritis of the knee. Arthritis Rheum
8. Buckland-Wright JC, Macfarlane DG, Williams SA, Ward RJ.
Accuracy and precision of joint space width measurements in
standard and macroradiographs of osteoarthritic knees. Ann
Rheum Dis 1995;54:872–80.
9. Ravaud P, Auleley GR, Chastang C, Rousselin B, Paolozzi L,
Amor B, et al. Knee joint space width measurement: an experi-
mental study of the influence of radiographic procedure and joint
positioning. Br J Rheumatol 1996;35:761–6.
10. Brandt K, Mazzuca S, Conrozier T, Dacre JE, Peterfy CG,
Provvedini D, et al. Which is the best radiographic protocol for a
clinical trial of a structure modifying drug in patients with knee
osteoarthritis? J Rheumatol 2002;29:1308–20.
11. Vignon E, Piperno M, Hellio Le Graverand MP, Mazzuca SA,
Brandt KD, Mathieu P, et al. Measurement of radiographic joint
space width in the tibiofemoral compartment of the osteoarthritic
knee: comparison of standing anteroposterior and Lyon schuss
views. Arthritis Rheum 2003;48:378–84.
12. Buckland-Wright JC, Wolfe F, Ward RJ, Flowers N, Hayne C.
Substantial superiority of semiflexed (MTP) views in knee osteo-
arthritis: a comparative radiographic study, without fluoroscopy, of
standing extended, semiflexed (MTP), and schuss views. J Rheu-
13. Peterfy C, Li J, Zaim S, Duryea J, Lynch J, Miaux Y, et al.
Comparison of fixed-flexion positioning with fluoroscopic semi-
flexed positioning for quantifying radiographic joint-space width in
the knee: test-retest reproducibility. Skeletal Radiol 2003;32:
14. Buckland-Wright JC, Ward RJ, Peterfy C, Mojcik F, Leff RL.
Reproducibility of the semiflexed (metatarsophalangeal) radio-
graphic knee position and automated measurements of tibiofemo-
ral joint space width in a multicenter clinical trial of knee
osteoarthritis. J Rheumatol 2004;31:1588–97.
15. Brandt K, Mazzuca S, Katz BP, Lane KA, Buckwalter KA, Yocum
DE, et al. Effects of doxycycline on progression of osteoarthritis:
results of a randomized, placebo-controlled, double-blind trial.
Arthritis Rheum 2005;52:2015–25.
16. Conrozier T, Mathieu P, Piperno M, Favret H, Colson F, Vignon
M, et al. Selection of knee radiographs for trials of structure-
modifying drugs in patients with knee osteoarthritis: a prospective,
longitudinal study of Lyon schuss knee radiographs with the
definition of adequate alignment of the medial tibial plateau.
Arthritis Rheum 2005;52:1411–7.
17. Spector TD, Conaghan PG, Buckland-Wright JC, Garnero P,
Cline GA, Beary JF, et al. Effect of risedronate on joint structure
and symptoms of knee osteoarthritis: results of the BRISK ran-
domized, controlled trial [ISRCTN01928173]. Arthritis Res Ther
18. Nevitt MC, Sharma L. OMERACT workshop radiography session.
Osteoarthritis Cartilage 2006;14 Suppl A:4–9.
19. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW.
Validation study of WOMAC: a health status instrument for
measuring clinically important patient relevant outcomes to anti-
rheumatic drug therapy in patients with osteoarthritis of the hip or
knee. J Rheumatol 1988;15:1833–40.
20. Kellgren JH, Lawrence JS. Radiological assessment of osteoar-
throsis. Ann Rheum Dis 1957;16:494–502.
21. Buckland-Wright CB. Protocols for precise radio-anatomical po-
sitioning of the tibiofemoral and patellofemoral compartments of
the knee. Osteoarthritis Cartilage 1995;3 Suppl A:71–80.
22. Hellio Le Graverand MP, Mazzuca S, Lassere M, Guermazi A,
Pickering E, Brandt K, et al. Assessment of the radioanatomic
positioning of the osteoarthritic knee in serial radiographs: com-
parison of three acquisition techniques. Osteoarthritis Cartilage
2006;14 Suppl A:A37–43.
23. Altman R, Hochberg M, Murphy W, Wolfe F, Lequesne M. Atlas
of individual radiographic features in osteoarthritis. Osteoarthritis
24. Duryea LJ, Li J, Peterfy CG, Gordon C, Genant HK. Trainable
rule-based algorithm for the measurement of joint space width in
digital radiographic images of the knee. Med Phys 2000;27:580–91.
25. Peterfy CG, Guermazi A, Zaim S, Tirman PF, Miaux Y, White D,
et al. Whole-OrganMagnetic
(WORMS) of the knee in osteoarthritis. Osteoarthritis Cartilage
26. Hunter DJ, LaValley M, Li J, Zhang Y, Bauer D, Nevitt M, et al.
Urinary pentosidine does not predict cartilage loss among subjects
with symptomatic knee OA: the BOKS Study. Osteoarthritis
FIXED-FLEXION RADIOGRAPHY FOR ASSESSMENT OF KNEE JOINT SPACE LOSS 1519
27. Neogi T, Nevitt M, Niu J, LaValley MP, Hunter DJ, Terkeltaub R, Download full-text
et al. Lack of association between chondrocalcinosis and increased
risk of cartilage loss in knees with osteoarthritis: results of two
prospective longitudinal magnetic resonance imaging studies. Ar-
thritis Rheum 2006;54:1822–8.
28. Wildy KS, Zaim S, Peterfy CG, Newman AB, Kritchevsky S, Nevitt
MC. Reliability of the Whole-Organ Review MRI Scoring
(WORMS) method for knee osteoarthritis (OA) in a multicenter
study [abstract]. Arthritis Rheum 2001;44 Suppl 9:S155.
29. Bland JM, Altman DG. Measurement error. Br Med J 1996;313:
30. Kothari M, Guermazi A, von Ingersleben G, Miaux Y, Sieffert M,
Block JE, et al. Fixed-flexion radiography of the knee provides
reproducible joint space width measurements in osteoarthritis. Eur
31. Mazzuca SA, Brandt KD, Buckland-Wright JC, Buckwalter KA,
Katz BP, Lynch JA, et al. Field test of the reproducibility of
automated measurements of medial tibiofemoral joint space width
derived from standardized knee radiographs. J Rheumatol 1999;
32. Mazzuca SA, Brandt KD, Buckwalter KA, Lane KA, Katz BP.
Field test of the reproducibility of the semiflexed metatarso-
phalangeal view in repeated radiographic examinations of sub-
jects with osteoarthritis of the knee. Arthritis Rheum 2002;46:
33. Mazzuca S, Brandt K, Buckwalter KA. Detection of radiographic
joint space narrowing in subjects with knee osteoarthritis: longitu-
dinal comparison of the metatarsophalangeal and semiflexed
anteroposterior view. Arthritis Rheum 2003;48:385–90.
34. Duddy J, Kirwan JR, Szebenyi B, Clark S, Granell R, Volkov S. A
comparison of the semiflexed (MTP) view with the standing
extended view (SEV) in the radiographic assessment of knee
osteoarthritis in a busy routine x-ray department. Rheumatology
35. Buckland-Wright JC, Bird CF, Ritter-Hrncirik CA, Cline GA,
Tonkin C, Hangartner TN, et al. X-ray technologists’ reproduc-
ibility from automated measurements of the medial tibiofemoral
joint space width in knee osteoarthritis for a multicenter, multi-
national clinical trial. J Rheumatol 2003;30:329–38.
36. Conrozier T, Mathieu L, Piperno M, Provvedini D, Taccoen A,
Colson F, et al. Lyon-Schuss radiographic view of the knee: utility
of fluoroscopy for the quality of tibial plateau alignment. J Rheu-
37. Yoshioka Y, Siu DW, Scudamore RA, Cooke TD. Tibial anatomy
and functional axes. J Orthop Res 1989;7:132–7.
1520 NEVITT ET AL