Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health, vol. 16, no.

4, 482e488, 2013
Ó Copyright 2013 by The International Society for Clinical Densitometry
1094-6950/16:482e488/$36.00
http://dx.doi.org/10.1016/j.jocd.2013.08.003

2013 Position Development Conference on Bone Densitometry

The Official Positions of the International Society for Clinical

Densitometry: Vertebral Fracture Assessment
Harold N. Rosen,*,1 Tamara J. Vokes,2 Alan O. Malabanan,1 Chad L. Deal,3
Jimmy D. Alele,4 Thomas P. Olenginski,5 and John T. Schousboe6
1
Division of Endocrinology, Department of Medicine, Beth Israel Deaconess Medical Center, Boston, MA, USA;
2
Division of Endocrinology, Department of Medicine, University of Chicago, Chicago, IL, USA;
3
Department of Rheumatology, Orthopedic and Rheumatology Institute, Cleveland, OH, USA; 4Division of Endocrinology,
Department of Medicine, St Rita Hospital, Lima, OH, USA; 5Division of Rheumatology, Department of Medicine,
Geisinger Medical Center, Danville, PA, USA; and 6Park Nicollet Institute for Research and Education and
University of Minnesota, Minneapolis, MN, USA

Abstract
Vertebral fracture assessment (VFA) is a low-cost method of accurately identifying individuals who have clini-
cally unrecognized or undocumented vertebral fractures at the time of bone density test. Because prevalent vertebral
fractures predict subsequent fractures independent of bone mineral density and other clinical risk factors, their rec-
ognition is an important part of strategies to identify those who are at high risk of fracture, so that prevention ther-
apies for those individuals can be implemented. The 2007 Position Development Conference developed detailed
guidelines regarding the indications for acquisition of, and interpretation and reporting of densitometric VFA tests.
The purpose of the 2013 VFA Task Force was to simplify the indications for VFA yet keep them evidence based. The
Task Force reviewed the literature published since the 2007 Position Development Conference and developed pre-
diction models based on 2 large cohort studies (the Study of Osteoporotic Fractures and the Osteoporotic Fractures
in Men Study) and the densitometry database of the University of Chicago. Based on these prediction models, in-
dications for VFA were reduced to a simplified set of criteria based on age, historical height loss, use of systemic
glucocorticoid therapy, and self-reported but undocumented prior vertebral fracture.

Key Words: Dual-energy X-ray absorptiometry; fracture; guideline; imaging; vertebral fracture.

Introduction occurrence, and spine imaging is required to document their

Vertebral compression fractures (VCF) occur commonly
among postmenopausal women and older men, such that
the prevalence is estimated to be 10%e26% among women
and men older than 50 yr, the prevalence of moderate to
severe VCF being 5%e15%. Prior VCF are a powerful
predictor of subsequent fractures, particularly of incident
VCF. Unlike fractures at other skeletal sites, however, only
25% of VCF are clinically apparent at the time of their
Received 08/10/13; Accepted 08/14/13.
*Address correspondence to: Harold N. Rosen, MD, CCD, Divi-
sion of Endocrinology, Department of Medicine, Beth Israel Deacon-
ess Medical Center, 330 Brookline Ave, Gryzmish 619, Boston,
MA 02215. E-mail: hrosen@bidmc.harvard.edu
presence. Densitometric lateral spine imaging, called verte-
bral fracture assessment (VFA), can efficiently and quickly
be done at the time of a bone density test and can accurately
detect moderate-to-severe vertebral fractures. Hence, imaging
the lateral spine at the time of a bone density test can substan-
tially improve fracture prediction and identification of those
for whom fracture prevention therapies are appropriate.
The 2007 Position Development Conference (PDC)
produced extensive Position Statements that constitute guide-
lines and standards regarding the indications for, acquisition
of, and interpretation and reporting of VFA tests (1). The
2007 PDC recommended VFA for those who had one of the
following:
1. Age
70 yr for women and
80 yr for men

482
2013 Official Positions on Vertebral Fracture Assessment
2. Historic height loss (HHL) of O4 cm for women or
O6 cm for men
3. Prospective height loss of O2 cm for women or O3 cm
for men
4. Glucocorticoid use
5. Self-reported but unconfirmed VCF
In addition, combinations of
2 lesser risk factors likewise
gave patients high enough a risk of prevalent VCF to warrant
VFA. These included the following:
6. Age 60e70 yr for women or 70e80 yr for men
7. HHL of 2e4 cm for women or 3e6 cm for men
8. Self-reported nonvertebral fractures
9. Chronic systemic diseases such as chronic obstructive pul-
monary disease, rheumatoid arthritis, or Crohn’s disease
10. Orchiectomy or androgen deprivation therapy (men)
Although these statements are highly evidence based, they
appear to be too complex for practitioners to remember and
apply. For 3 of the 5 single criteria (age, HHL, and prospec-
tive height loss), the criterion cutpoint is different for men
compared with women. The combinations of lesser risk fac-
tors that give yet different cutoffs for the previous criteria
adds additional complexity. Lastly, the chronic disease crite-
rion is confusing, in that it included certain risk factors that
are included in FRAX such as rheumatoid arthritis but in-
cluded other risk factors that are not included in FRAX
such as chronic obstructive pulmonary disease. Additionally,
FRAX included certain chronic diseases that were not in-
cluded in the VFA criteria, such as type I diabetes mellitus.
Informal polling of colleagues with a serious commitment
to the field of bone densitometry revealed that few were
able to recall these complex criteria, which does not auger
well for generalists to be able to remember and apply these
guidelines.
The 2013 VFA Task Force’s charge was to revisit the indi-
cations for VFA, review the literature published since the
2007 PDC, and to use heretofore unpublished cohort data
to develop evidence-based indications for VFA. Moreover,
in light of the fact that many health-care organizations do
not have bone densitometers with lateral spine imaging capa-
bility, the Task Force guideline indications for VFA were
shaped with the intention that they apply for use of lateral
spine radiography to detect clinically unapparent vertebral
fracture.
This article will describe the methodology of the Task
Force and questions posed to the Task Force, the Statement
addressing those questions that were voted as appropriate
without disagreement by the 2013 International Society for
Clinical Densitometry (ISCD) PDC Expert Panel and ap-
proved by ISCD Board of Directors, and explain the rationale
behind the statement. Separate articles will describe the de-
velopment of and comparison between prediction models
for prevalent VCF for women and men using, respectively,
data from the Study of Osteoporotic Fractures (SOF) and Os-
teoporotic Fractures in Men (MrOS) cohort studies.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
483
Methodology
The methods used to develop and grade the Official Posi-
tion Statement for VFA presented in this document are pre-
sented in the Executive Summary of the 2013 PDC
regarding bone densitometry that is also in this issue. In brief,
the Position Statement presented here was rated as appropri-
ate without disagreement by the Expert Panel of the 2013
ISCD PDC. This position was also rated by the Expert Panel
on quality of evidence, strength of recommendation, and ap-
plicability. Quality of evidence is rated as good, fair, or
poor, where good is evidence that includes results from
well-designed, well-conducted studies in representative popu-
lations; fair is evidence sufficient to determine effects on out-
comes, but the strength of the evidence is limited by the
number, quality, or consistency of the individual studies;
poor is evidence that is insufficient to assess the effects on
outcomes because of limited number or power of studies, im-
portant flaws in their design or conduct, gaps in the chain of
evidence, or information. Strength of the recommendation is
rated as A, B, or C, where A is a strong recommendation sup-
ported by the evidence, B is a recommendation supported by
the evidence, and C is a weak recommendation supported pri-
marily by expert opinion. Applicability is rated as worldwide
or local; local statements may vary in their applicability ac-
cording to local requirements.
What Are the Appropriate Indications
for VFA?
2013 ISCD Official Position
Lateral spine imaging with standard radiography or densi-
tometric VFA is indicated when T-score is less than 1.0 and
of one or more of the following is present:
a. Women age
70 yr or men age
80 yr
b. Historical height loss O4 cm (O1.5 inches)
c. Self-reported but undocumented prior vertebral fracture
d. Glucocorticoid therapy equivalent to
5 mg of predni-
sone or equivalent per day for
3 mo
Grade: Fair-B-W
Rationale
The Task Force thought that the 2007 PDC recommenda-
tions for VFA were, and continue to be, valid. The published
2007 Task Force article on VFA (1) adequately summarized
the literature on the topic until that point, including studies
that had estimated the multivariable adjusted strength of asso-
ciation of predictors with prevalent vertebral fracture on lat-
eral spine imaging (2e5). The Task Force identified 7
additional articles that identified mulitivariable-adjusted pre-
dictors of prevalent vertebral fractures (identified either on
standard radiographs or lateral densitometric lateral images),
but these articles by and large confirmed previously appreci-
ated risk factors (6e13). The Task Force considered 2
Volume 16, 2013
484
approaches to make evidence-based criteria for VFA easier to
implement in clinical practice:
1. Retaining the simple univariate criteria for selecting pa-
tients for VFA and eliminating the combinations
2. Developing a regression-based prediction model that
could be programmed into bone densitometers (as the
FRAX fracture prediction equation is) to indicate to
bone densitometry technicians when a VFA is indicated
along with standard bone density measurement
The former approach would presumably weaken the ability
to identify patients with combinations of lesser risk factors as
being at risk. To compensate, the Task Force considered the
latter approach, advocating the use of a regression that used
bone mineral density (BMD), HHL, age, gender, race, prior
non-VCF, and steroid use to give an estimate of risk of prev-
alent VCF on the supposition that a regression approach may
do even better at identifying patients at high risk for VCF than
the combinations previously recommended. However, a re-
gression-based approach would be prohibitively burdensome
for primary care providers (PCPs) or densitometrists to per-
form. To address this concern, there was interest in identify-
ing a robust regression that would allow us to predict risk
of prevalent VCF and to have the manufacturers add this to
the densitometer software as an option, so that this risk can
automatically be printed out by the densitometer. This way,
by the time a bone densitometry technologist finishes measur-
ing BMD, the prevalent vertebral fracture risk can be printed
out based on information already collected by the densitome-
ter. The technologist can do the VFA if the likelihood of prev-
alent VCF printed out on the BMD is
10%, once this feature
is available.
Predictors of Prevalent Vertebral Fracture:
Summary of Prior Studies
There are a number of studies that have identified predic-
tors of prevalent vertebral fracture that are significant after
multivariable adjustment (Table 1). They vary considerably
in sample size, target populations, and the method by which
they determined whether a vertebra was fractured. All stud-
ies that included bone mineral density or height loss in their
multivariable models found these 2 variables to be associ-
ated with prevalent vertebral fracture. All but 2 of the stud-
ies found age to be an independent risk factor for prevalent
vertebral fracture: one that did not had study population re-
stricted to the oldest of the elderly (10) and the other had
a small study population (n 5 216) of men (8). Only 3 stud-
ies included systemic glucocorticoid use in their models,
and 2 of these confirmed a strong association with prevalent
vertebral fracture. Prior nonvertebral fracture was associated
with prevalent radiographic vertebral fracture in most but
not all of these studies (Table 1). Five of these studies spe-
cifically used densitometric lateral spine imaging to detect
prevalent vertebral fractures, and by and large, the
multivariable-adjusted associations of predictors with prev-
alent vertebral fracture were similar to those studies that
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Rosen et al.
used standard radiographs for vertebral fracture detection
(6,8,9,12,14).
Many of these published regressions could be used ‘‘off
the shelf’’ for the earlier mentioned purposes but have weak-
nesses that would make them difficult to use ‘‘as is’’ for prac-
titioners of clinical densitometry. By far, the largest of these
studies was that of Vogt et al (5), who studied a large group
of women (N 5 13,051) who had been screened for the Frac-
ture Intervention Trial of alendronate vs placebo and had
BMD, spine X-rays, and had completed questionnaires. The
weaknesses of the study included that this study included
no men, did not provide the racial breakdown of the women,
and did not include BMD as a predictor. Because densitomet-
ric VFA is done at the time of bone mass measurement, BMD
values are of course readily available to aid decision making
as to whether lateral spine imaging should be done, and
hence, the lack of BMD in the decision rule of Vogt et al
limits its usefulness.
All these studies were restricted to 1 or 2 ethnic groups or
were limited to 1 geographic region. Some studies used full
quantitative morphometry as the method of adjudicating
whether a vertebra was fractured or not (2,3,5,12). However,
these methods do not distinguish nonosteoporotic deformities
with height loss from true fractures and are more difficult to
execute in clinical practice. Perhaps most important, none of
these studies examined whether the simpler prediction models
might be as good as more complex models discriminating be-
tween subjects with and without fractures. Variables that are
independently associated with an outcome do not necessarily
improve the ability of a simpler model to discriminate those
with from those without the outcome. The Task Force felt
that if a simpler prediction rule or decision model were
used to determine for whom lateral spine imaging is indicated
instead of the complex decision rule of the 2007 PDC, testing
the ability of simpler models or decision rules to discriminate
those who have compared with those who do not have prev-
alent vertebral fracture would be an essential step.
The Task Force decided to access very large databases that
could be queried to identify predictors of prevalent VCF, and
a robust regression generated could be recommended for in-
clusion in the densitometer software.
Development and Testing of Prediction Models
for Men
Through a coinvestigator (JTS) at the Minneapolis site of
the SOF and the MrOS study, we were approved by both co-
horts to develop prediction equations based on logistic regres-
sion for prevalent vertebral fractures in women (using SOF
data) and men (using MrOS data). Our purpose was to com-
pare the ability of regression-based prediction models and
nonregression-based decision rules to discriminate those
who have from those who do not have prevalent radiographic
vertebral fracture, using both area under receiving operating
characteristics curves (AUROC) analyses and the Net Reclas-
sification Method of Pepe (15e17). We restricted our analy-
ses to those with a femoral neck T-score of 1 or less, for
Volume 16, 2013
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health

2013 Official Positions on Vertebral Fracture Assessment

Table 1
Multivariable-Adjusted Predictors of Prevalent Vertebral Fractures on Lateral Spine Radiographs or Densitometric Images

Association of predictors with prevalent vertebral fracture

Vertebral fracture Height Glucocorticoid Prior Grip Back
Citation Study population Image modality definitiona Age BMD loss use Weight fracture Smoking strength pain

Vogt et al (5),*,y,z Women age
50 yr Standard X-ray Full QM Yes n/ab Yes n/ab n/ab n/ab n/ab n/ab n/ab
Ling et al (3) Women age 50
yr Standard X-ray Full QM Yes Yes Yes n/ab No No No No No
Kaptoge et al (2) Women and men age Standard X-ray Full QM Yes Yes Yes n/ab Yes Yes n/ab n/ab n/ab

50 yr
Tobias et al (4) Women age 65e75 yr Standard X-ray ABQ No Yes Yes n/ab n/ab Yes n/ab n/ab Yes
Middleton et al (14) Women age
65 yr DXA SQ Yes Yes Yes n/ab No Yes n/ab n/ab n/ab
Vokes and Gillen (6),z Women and men age DXA SQ Yes Yes Yes Yes n/ab Yes n/ab n/ab n/ab
19e95 yr
El Maghraoui et al (8) Men age 50e79 yr DXA SQ No Yes n/ab n/ab No n/ab No n/ab n/ab
El Maghraoui et al (9) Women age
50 yr DXA SQ Yes Yes n/ab n/ab Yes Yes n/ab n/ab n/ab
Kwok et al (13),x Women and men age Standard X-ray SQ Yes Yes Yes n/ab n/ab M: No n/ab M: Yes Yes

65 yr F: Yes F: No
Waterloo et al (12) Women and men age DXA Full QM Yes Yes n/ab n/ab n/ab n/ab n/ab n/ab n/ab
38e87 yr
van der Jagt-Willems Women and men Standard X-ray SQ No n/ab n/ab Yes No Yes No n/ab n/ab
et al (10),k mean age 81 yr
Sanfelix-Gimeno Women age
50 yr Standard X-ray SQ Yes Yes n/ab No No No No n/ab n/ab
et al (11),{
Note: Additional factors associated with prevalent vertebral fracture (multivariable adjusted): *falls, yself-reported osteoporosis, zself-reported vertebral fracture, xself-reported
physical activity, klow serum albumin, and {educational level.
Abbr: ABQ, algorithm-based method; BMD, bone mineral density; F, female; M, male; QM, quantitative morphometry; SQ, Genant semiquantitative method.
a
Methods of adjudicating/defining vertebral fracture: full QM, ABQ, and SQ.
b
Predictor not included in regression models.
Volume 16, 2013

485
486
2 reasons. First, there is no data regarding the effectiveness of
currently available pharmacologic fracture prevention thera-
pies in those with normal bone density. Second, lateral spine
imaging is needed to detect clinically unrecognized vertebral
fractures. When discovered for the first time on lateral spine
images, the age of these fractures is unknown, and their inter-
pretation in those with normal BMD is problematic. The frac-
tures could have occurred in the distant past with trauma
(especially for men).
These analyses are reported in detail for men in a separate
article in this issue. In summary, we found that the prevalence
of radiographic grade 2 or grade 3 (by the Genant semiquan-
titative criteria) vertebral fracture among those who self-
reported a prior spine fracture was 50%. Among the remaining
men, a simple nonregression-based set of criteria (age
80 yr,
historical height loss O 4 cm, and glucocorticoid use) discrim-
inated men with from those without prevalent radiographic ver-
tebral fracture and a complex regression-based model. By these
criteria, an estimated 54% of men aged 65 yr and older would
undergo lateral spine imaging and 73% of men with one or
more prevalent moderate-to-severe vertebral fractures would
be detected.
Development and Testing of Prediction Models
for Women
Analysis of the subset of SOF participants attending the
third SOF visit with a femoral neck T-score of 1.0 or less
revealed similar findings to those seen in men. Vertebrae
were evaluated for fracture using full quantitative morphom-
etry, the details of which are previously published (18,19).
These analyses are reported in substantial detail in another ar-
ticle in this issue. Women in SOF with self-reported VCF had
a very high likelihood of having a confirmed VCF (59%), so
patients with self-reported VCF were excluded from analysis,
and it was understood that self-reported VCF but undocu-
mented vertebral fracture would be considered a stand-alone
indication for lateral spine imaging. The youngest woman at-
tending the third SOF visit was 68 yr old, and hence, the over-
all prevalence of radiographic vertebral fracture among the
subset with a femoral neck T-score of 1.0 or less was
20% even after excluding those with a self-reported prior
spine fracture. Even complex prediction models could not
identify a subgroup of women in this group for whom lateral
spine imaging would not be reasonable. These data strongly
support both self-reported (but undocumented) prior spine
fracture and age
70 yr being stand-alone indications for lat-
eral spine imaging to identify those with prevalent vertebral
fracture.
Testing of Prediction Models and Nonregression-
Based Decision Rules Using the University of
Chicago Densitometry Database
The Task Force’s next step was to externally validate the
most parsimonious regression-based prediction model devel-
oped in the SOF in a separate data set that included younger
women and test the performance of the prediction model
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Rosen et al.
against nonregression-based decision rules. University of
Chicago database (TJV) consisted of the previously reported
population of 815 women (6) and an additional 368 women
enrolled later for a total of 1228 women who were recruited
to have VFA when they were referred for clinically indicated
bone density testing. The prevalence of vertebral fractures
was 17%. The results of this analysis are reported in detail
in a separate article in this issue. Briefly, we first compared
the 2 prediction models developed in SOF. These were the
simple model that included age, femoral neck BMD, and his-
torical height loss and the complex model that included age,
femoral neck BMD, historical height loss, prior nonvertebral
fracture, and body mass index. As was true in the SOF data,
the complex model did not discriminate those with from
those without prevalent vertebral fracture better than the
simple model. Because the parameter coefficients were de-
rived in a subset of the SOF population all of whom were
aged 68 yr and older, we examined model discrimination us-
ing AUROC analyses. The AUROC for simple model among
451 women aged 50e67 yr was actually higher (0.64, 95%
confidence interval of 0.58e0.71) than for 466 women
aged 68 yr and older (0.52, 95% confidence interval
0.50e0.55).
We then compared the prediction of prevalent vertebral
fractures using the 2007 and 2013 ISCD Official Positions
for indications for VFA. Compared with 2007 indication,
2013 indications resulted in similar sensitivity (89.9% vs
89.7%) and marginally better specificity (41.2% vs 37.0%).
There were no significant differences in the AUROC that
was 0.65 (0.63, 0.68) for 2013 and 0.63 (0.61, 0.66) for
2007 indications. Using the 2013 PDC indications for VFA,
there would be 64.4% screened (vs 67.6% with the 2007 in-
dications) with the yield (percent of women found to have
vertebral fractures among those selected for screening) of
26.3% for 2013 (vs 23.0% for 2007 indications). Based on
these findings, we conclude that the much simpler 2013 indi-
cations and the more complex ones perform similarly in iden-
tifying who should be selected for VFA imaging among
women referred for BMD testing.
Limitations
There are several limitations inherent in the Position
Statement. First, although we have confirmed in the popula-
tions for whom we did our analyses that other criteria do not
add predictive power, that in part may be because of a low
prevalence in these populations. For example, rheumatoid
arthritis, ankylosing spondylitis, inflammatory bowel dis-
ease, and use of antiandrogen therapy occur in relatively
small proportions of the population at large, and hence in
prediction models developed in general populations, they
may have little predictive power. However, in subspecialty
referral populations, prediction models including covariates
indicating the presence or absence of these conditions might
perform better than models without those covariates. The
Task Force strongly believes that approved Position State-
ment should not be construed to mean that lateral spine
Volume 16, 2013
2013 Official Positions on Vertebral Fracture Assessment
imaging should not be done in these individuals, but rather
that whether lateral spine imaging is indicated for individ-
uals with these conditions needs to be determined on
a case-by-case basis.
Second, we considered the question of whether to define
glucocorticoid use as in FRAX that is cumbersome or some
more simple definition such as ‘‘current use of steroids.’’
The FRAX calculator is now embedded in the software of
the densitometers of the major manufacturers, and hence,
some densitometrists are collecting FRAX risk factors so
that fracture risk scores are included in the densitometer out-
put. Therefore, we preferred to use the FRAX definition to
avoid forcing densitometrists to remember a second ‘‘gluco-
corticoid use’’ criterion.
Third, we considered the issue of including race in select-
ing patients for VFA. Initially, it seemed like a good idea be-
cause it is widely appreciated that race affects the risk of all
fractures, including VCF in men (7,20,21). However, there is
some literature suggesting that once patients are selected for
densitometry, we have selected a higher risk population in
whom race is no longer predictive (6,22). Because our
goal was to simplify criteria for VFA, and adding race would
complicate this without clear evidence of benefit, the deci-
sion was made to recommend retaining the current practice
of not including race in a decision about whether to recom-
mend VFA.
Finally, implementation of vertebral fracture risk assess-
ment either densitometrically or using lateral spine radiogra-
phy requires that the readers of lateral spine images have the
requisite training to discern fractured vertebrae from normal
vertebrae and from nonfracture vertebral deformities. Such
training is available through the VFA course offered by ISCD.
Questions for Future Research
All the prediction models and decision rules the Task
Force considered are modest in their power to discriminate
those with from those without prevalent vertebral fracture.
Future research is needed to determine if other predictors
can be identified that may improve the efficiency of lateral
spine imaging to identify those with clinically unrecognized
vertebral fractures. Moreover, studies will be needed on
new care processes within health-care delivery organizations
to identify those who should have lateral spine imaging. For
densitometric VFA, having the bone densitometry order en-
compass a contingency order for the densitometry technolo-
gist to perform VFA if certain criteria are met has been
shown to be feasible in clinical practice. Bone densitometrists
were able to apply the guidelines (which in that study were
close to the indications of the ISCD 2013 Position Statement),
and this study showed that a VFA positive for prevalent ver-
tebral fracture did increase physician prescription of fracture
prevention medication (23,24). Moreover, there is evidence
that a VFA study that documents that a prevalent vertebral
fracture is present does affect a patient’s perceived need for
fracture prevention medication and, hence, could affect
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
487
patient’s willingness to accept and adherence to recommen-
ded fracture prevention therapy (25). Much more research is
needed regarding the adoption of care processes to identify
those with prevalent vertebral fracture within health-care de-
livery organizations and the impact of those care processes on
patient and provider behavior and patient outcomes.
References
1. Schousboe JT, Vokes T, Broy SB, et al. 2008 Vertebral fracture
assessment: the 2007 ISCD Official Positions. J Clin Densitom
11(1):92e108.
2. Kaptoge S, Armbrecht G, Felsenberg D, et al. 2004 When
should the doctor order a spine X-ray? Identifying vertebral frac-
tures for osteoporosis care: results from the European Prospec-
tive Osteoporosis Study (EPOS). J Bone Miner Res 19(12):
1982e1993.
3. Ling X, Cummings SR, Mingwei Q, et al. 2000 Vertebral frac-
tures in Beijing, China: the Beijing Osteoporosis Project.
J Bone Miner Res 15(10):2019e2025.
4. Tobias JH, Hutchinson AP, Hunt LP, et al. 2007 Use of clinical
risk factors to identify postmenopausal women with vertebral
fractures. Osteoporos Int 18(1):35e43.
5. Vogt TM, Ross PD, Palermo L, et al. 2000 Vertebral fracture
prevalence among women screened for the Fracture Intervention
Trial and a simple clinical tool to screen for undiagnosed verte-
bral fractures. Fracture Intervention Trial Research Group. Mayo
Clin Proc 75(9):888e896.
6. Vokes TJ, Gillen DL. 2010 Using clinical risk factors and bone
mineral density to determine who among patients undergoing
bone densitometry should have vertebral fracture assessment.
Osteoporos Int 21(12):2083e2091.
7. Cauley JA, Palermo L, Vogt M, et al. 2008 Prevalent vertebral
fractures in black women and white women. J Bone Miner
Res 23(9):1458e1467.
8. El Maghraoui A, Mounach A, Gassim S, Ghazi M. 2008 Verte-
bral fracture assessment in healthy men: prevalence and risk
factors. Bone 43(3):544e548.
9. El Maghraoui A, Rezqi A, Mounach A, et al. 2013 Systematic
vertebral fracture assessment in asymptomatic postmenopausal
women. Bone 52(1):176e180.
10. van der Jagt-Willems HC, van Hengel M, Vis M, et al. 2012
Why do geriatric outpatients have so many moderate and severe
vertebral fractures? Exploring prevalence and risk factors. Age
Ageing 41(2):200e206.
11. Sanfelix-Gimeno G, Sanfelix-Genoves J, Hurtado I, et al. 2013
Vertebral fracture risk factors in postmenopausal women over
50 in Valencia, Spain. A population-based cross-sectional study.
Bone 52(1):393e399.
12. Waterloo S, Nguyen T, Ahmed LA, et al. 2012 Important risk
factors and attributable risk of vertebral fractures in the
population-based Tromso study. BMC Musculoskelet Disord
13:163.
13. Kwok AW, Gong JS, Wang YX, et al. 2013 Prevalence and risk
factors of radiographic vertebral fractures in elderly Chinese
men and women: results of Mr. OS (Hong Kong) and Ms. OS
(Hong Kong) studies. Osteoporos Int 24(3):877e885.
14. Middleton ET, Gardiner ED, Steel SA. 2009 Which women
should be selected for vertebral fracture assessment? Comparing
different methods of targeting VFA. Calcif Tissue Int 85(3):
203e210.
15. Janes H, Pepe MS, Gu W. 2008 Assessing the value of risk pre-
dictions by using risk stratification tables. Ann Intern Med
149(10):751e760.
Volume 16, 2013
488
16. Pencina MJ, D’Agostino RB Sr, D’Agostino RB Jr, Vasan RS.
2008 Evaluating the added predictive ability of a new marker:
from area under the ROC curve to reclassification and beyond.
Stat Med 27(2):157e172. discussion 207e212.
17. Pepe MS, Feng Z, Huang Y, et al. 2008 Integrating the predic-
tiveness of a marker with its performance as a classifier. Am J
Epidemiol 167(3):362e368.
18. Black DM, Cummings SR, Stone K, et al. 1991 A new approach
to defining normal vertebral dimensions. J Bone Miner Res 6(8):
883e892.
19. Black DM, Arden NK, Palermo L, et al. 1999 Prevalent
vertebral deformities predict hip fractures and new vertebral
deformities but not wrist fractures. Study of Osteoporotic
Fractures Research Group. J Bone Miner Res 14(5):
821e828.
20. Tracy JK, Meyer WA, Grigoryan M, et al. 2006 Racial differ-
ences in the prevalence of vertebral fractures in older men: the
Baltimore Men’s Osteoporosis Study. Osteoporos Int 17(1):
99e104.
Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health
Rosen et al.
21. Naing S, Benjamin A, Barengolts E. 2009 Prevalence and
follow-up of vertebral osteoporotic fractures in white and Afri-
can American men with back pain. Endocr Pract 15(5):
499e500.
22. Lansdown D, Bennet B, Thiel S, et al. 2011 Prevalence of ver-
tebral fractures on chest radiographs of elderly African Ameri-
can and Caucasian women. Osteoporos Int 22(8):2365e2371.
23. Schousboe JT. 2011 Does identification of prevalent vertebral
fracture on densitometric vertebral fracture assessment (VFA)
in clinical practice influence physician prescribing behavior?.
[abstract]. Arthritis Rheum 63(10 Suppl):1634.
24. Schousboe JT, McKiernan FE, Binkley N. 2012 A performance
algorithm improves appropriate vertebral fracture assessment
use among those referred for DXA and improves utilization of
fracture prevention medication for those with prevalent vertebral
fracture. J Bone Miner Res 27(1 Suppl).
25. Schousboe JT, Davison ML, Dowd B, et al. 2011 Predictors of
patients’ perceived need for medication to prevent fracture.
Med Care 49(3):273e280.
Volume 16, 2013