Full text of DOI:10.1359/jbmr.2002.17.12.2214

JOURNAL OF BONE AND MINERAL RESEARCH
Volume 17, Number 12, 2002
© 2002 American Society for Bone and Mineral Research
The Relationship Between Bone Density and Incident
Vertebral Fracture in Men and Women
THE EUROPEAN PROSPECTIVE OSTEOPOROSIS STUDY (EPOS) GROUP*
ABSTRACT
Bone mineral density (BMD) is an important predictor of future fracture risk in women; however, there are
few prospective data in men. The aim of this analysis was to determine whether there are differences in the
relationship between BMD and incident vertebral fracture in men and women. Men and women were
recruited from population-based registers in 21 European centers. Those recruited were interviewed and had
spinal radiographs performed. The radiographs were assessed morphometrically and prevalent vertebral
deformity was defined using the McCloskey-Kanis method. Repeat spinal radiographs were performed at a
mean of 3.8 years after the baseline radiographs. Incident fractures were defined using a combination of the
point prevalence and 20% reduction in vertebral height (plus a 4-mm reduction in absolute height) criteria.
BMD measurements were made in a subsample of those recruited. Poisson regression was used to explore the
influence of gender, age, prevalent deformity, and BMD on the incidence of vertebral fracture. Thirty-four
hundred sixty-one men and women had both paired spinal radiographs and bone density measurements
performed. BMD at the spine and femoral neck was higher in men than in women. After adjusting for age,
the risk of incident vertebral fracture was greater in women than in men (relative risk [RR] ؍ 2.3; 95% CI,
1.5–3.6) and increased by a factor of 1.4 (95% CI, 1.2–1.8), 1.5 (95% CI, 1.2–1.8), and 1.6 (95% CI, 1.3–1.9)
per decrease of 0.1 g/cm2 in BMD at the spine, femoral neck, and trochanter, respectively. After adjusting for
BMD at the spine or trochanter, the gender difference in the predicted age-specific incidence of vertebral
fracture was no longer significant (RR ؍ 1.1 and 95% CI, 0.6–1.9 at the spine; RR ؍ 1.5 and 95% CI, 0.8–2.7
at the trochanter), although it persisted after adjusting for femoral neck BMD (RR ؍ 1.9; 95%CI, 1.1–3.3).
The presence of a prevalent vertebral deformity was a strong risk factor for future vertebral fracture,
although the strength of the association was reduced after adjustment for age, sex, and spine BMD. However,
adjustment for the presence of a baseline vertebral deformity did not alter the main findings. In conclusion,
at a given age and spine (although not femoral neck) bone density, the risk of incident vertebral fracture is
similar in men and women. Incident vertebral fractures are more common in women than men because at any
age their spine BMD is lower. (J Bone Miner Res 2002;17:2214–2221)
Key words incident vertebral fracture, bone mineral density, sex differences, prevalent vertebral deformity,
population based study
of individuals with osteoporotic fracture, particularly hip
but also vertebral fracture, is set to rise in both men and
women because of demographic changes toward an older
population.(4,5) Early identification of individuals who are at
risk of fracture and targeting interventions at these individ-
uals is an important component of any strategy to reduce the
burden of disease in the population.
INTRODUCTION
ERTEBRAL FRACTURE is an important health problem in
both men and women through its association with back
V
pain and disability.(1–3) Over the next 50 years, the numbers
The authors have no conflict of interest.
*EPOS Study Group Authors: O’Neill TW,† Lunt M, Silman AJ, Felsenberg D, Benevolenskaya LI, Bhalla AK, Cannata JB, Cooper
C, Crabtree N, Dequeker J, Hoszowski K, Jajic I, Kanis JA, Kragl G, Lopes Vaz A, Lorenc R, Lyritis G, Masaryk P, Miazgowski T, Parisi
G, Pols HAP, Poor G, Reid DM, Scheidt-Nave C, Stepan J, Todd C, Weber K, Woolf AD, and Reeve J.
†ARC Epidemiology Unit, Stopford Building, University of Manchester, Manchester, United Kingdom.
2214

BMD AND RISK OF INCIDENT VERTEBRAL FRACTURE
2215
Bone mineral density (BMD) is one of the most important before or within 1 year of the initial radiograph, 1110
determinants of incident vertebral fracture in women, with measurements were made Ͼ12 months after the first radio-
data from many studies suggesting a twofold increase in risk graph but within a month of the second one, and 424
per SD reduction in BMD.(6) However, there are no pub- measurements were made after the second radiograph (256
lished data from prospective population studies examining measurements within 6 months, 70 measurements between
the relationship between BMD at either the spine or femoral 6 months and 1 year, and 98 measurements Ͼ1 year after).
neck and vertebral fracture in men. We have published a In general, each center measured all of its participants over
retrospective study examining the effect of gender on the a short period; therefore, they were all measured either at
relationship between bone density and prevalent vertebral the time of the first film, during the follow-up, or after the
fracture but there is little other data bearing on this ques- second film. Of subjects with two X-rays, BMD was mea-
tion.(7) As a consequence, at present, it is difficult to apply sured in 69% of subjects in the centers that measured only
a clinical value to the results of bone density measurements spine BMD, 65% of subjects in the centers that measured
in men. If the relationship is similar to that observed in only hip BMD, and 80% of subjects in the centers that
women, then it may be possible to use a single gender- measured both spine and hip BMD.
neutral diagnostic criterion for bone density. If not, gender-
The densitometers in each center were, with one excep-
specific criteria should be used. This has implications in tion (a Sopha fan-beam machine), pencil beam DXA ma-
clinical decision making, particularly because effective chines made by Lunar (Madison, WI, USA), Hologic
therapies have become available for use in treating men (Waltham, MA, USA), or Norland (Fort Atkinson, WI,
with osteoporosis.(8,9)
USA) and were cross-calibrated using the European Spine
In this study, using data from the population-based Eu- Phantom.(11) At least five measurements of the phantom
ropean Prospective Osteoporosis Study (EPOS), our aim were made on each machine and a two-parameter empiri-
was to determine (i) whether there are differences in the cally fitted linear or exponential calibration curve was used
strength of the relationship between BMD and incident to convert measured density values into standardized val-
vertebral fracture in men and women and (ii) if so, whether ues, as described by Pearson.(12) Detailed descriptions of the
the risk of incident vertebral fractures is the same at the densitometry procedures as they applied to the subjects are
same absolute level of BMD in men and women.
presented elsewhere.(13)
Radiographs
MATERIALS AND METHODS
The radiographs at both baseline and follow-up were
taken using a standard protocol. The thoracic spine films
were centered at T7 and lumbar films were centered at L2.
The breathing technique was used to obscure overlying lung
shadows. All of the radiographs were forwarded to the
central coordinating facility in Berlin for evaluation and
were assessed morphometrically by one of four observers.
In each subject, vertebrae from T4 to L4 were assessed and
anterior, middle, and posterior heights were recorded. Using
these measurements, the presence or absence of a prevalent
vertebral deformity in both films was defined using the
McCloskey-Kanis algorithm.(14) An incident vertebral frac-
ture was defined as a vertebra that showed evidence of a
change in absolute height (anterior, middle, or posterior
height) between films of 20% or more (plus at least 4 mm),
together with the requirement that the vertebra satisfy cri-
teria for prevalent vertebral deformity on the follow-up (but
not the baseline) film.(15)
The subjects who took part in this analysis were recruited
for participation in the EPOS. The detailed methods have
been described elsewhere.(10) In brief, men and women were
recruited from population-based registers in 36 centers
across Europe. Stratified sampling was used with the aim of
recruiting equal numbers of men and women in each of six
5-year age bands: 50–54 years, 55–59 years, 60–64 years,
65–69 years, 70–74 years, and Ն75–79 years. All subjects
who agreed to take part had baseline lumbar and spinal
radiographs, which were taken according to standard pro-
tocol, and an interviewer-administered lifestyle question-
naire. A subsample of those recruited to the baseline survey
had BMD assessed at the spine and femoral neck. The
subjects recruited were followed prospectively to ascertain
new clinical fractures, and in 28 centers, subjects were
invited to attend for repeat spinal radiographs, which were
performed at a mean of 3.8 years after the baseline survey.
BMD
Statistical analysis
In 21 of the 36 European Vertebral Osteoporosis Study
(EVOS) centers, subjects had bone densitometry performed
at baseline or during the follow-up period. Eleven centers
measured at both the spine and the hip, 2 centers measured
the spine alone, and 8 centers measured the hip alone. Of the
2204 spine BMD measurements, 1865 measurements were
made before or within 1 year of the initial radiograph, 275
measurements were made Ͼ12 months after the first radio-
graph but within a month of the second one, and 64 mea-
Poisson regression was used to determine the relationship
between incident vertebral fracture and BMD. Poisson re-
gression predicts an incidence rate (IR) as
IR ϭ exp
␤ x ϩ offset ,
i i
͸
ͩ
ͪ
i
surements were made Ͼ1 month after the second radiograph where the xi’s represent any predictor variables and the ␤i’s
(between 1 and 2 years after, all in one center). Of the 2986 represent the corresponding coefficients. The offset repre-
hip BMD measurements, 1452 measurements were made sents the duration of exposure to risk, in this case, the time

2216
O’NEILL ET AL.
TABLE 1. SUBJECT CHARACTERISTICS
increased exposure to heavy activity during earlier adult
life.(17) Compared with those with any BMD measurement
(3461 subjects), those without (1189 subjects) were of a
similar age (62.6 years vs. 62.5 years) and had a similar
prevalence of vertebral deformity (10% vs. 9.1%) with no
statistically significant difference between them.
Men
Women
(n ϭ 1537)
(n ϭ 1924)
Age (years)
63.3 (7.9)
1.06 (0.22)
0.83 (0.14)
0.77 (0.14)
10.9
62.2 (7.5)
0.92 (0.20)*
0.73 (0.14)*
0.63 (0.12)*
9.7
Spine BMD (g/cm2)
Femoral neck BMD (g/cm2)
Trochanteric BMD (g/cm2)
Prevalent deformity (%)
Risk factors for incident vertebral fracture
In total, 26 men and 69 women had evidence of one or
more incident vertebral fractures, an incidence of 9.3/1000
person-years in women (95% CI, 7.3–11.8) and 4.5/1000
person-years in men (95% CI, 3.0–6.6). Increasing age, low
BMD, and the presence of a vertebral deformity at baseline
were significant predictors of incident fracture in both men
and women (Table 2). There was no evidence that the
association between BMD and incident fracture risk de-
pended on the time the BMD measurement was made (p Ͼ
0.3 for all three sites). The differences in the strength of
these relationships between men and women were not sta-
tistically significant. After age adjustment, women had a
twofold increased risk of vertebral fracture compared with
men (RR ϭ 2.3; 95% CI, 1.5–3.6). After adjusting for BMD
at the spine or trochanter, the gender difference in the
predicted age-specific incidence of vertebral fracture was no
longer significant (RR ϭ 1.1 and 95% CI, 0.6–1.9 at the
spine; RR ϭ 1.5 and 95% CI, 0.8–2.7 at the trochanter),
although it persisted after adjusting for femoral neck BMD
(RR ϭ 1.9 ; 95% CI, 1.1–3.3). After adjusting for age and
gender, the risk of incident vertebral fracture increased by a
factor of 1.4 per 0.1 g/cm2 reduction in spine BMD (95%
CI, 1.2–1.7), 1.5 per 0.1 g/cm2 reduction in femoral neck
BMD (95% CI, 1.2–1.8), and 1.6 per 0.1 g/cm2 reduction in
trochanteric BMD (95% CI, 1.3–1.9). Further adjustment
for the presence of baseline vertebral deformity reduced
these estimates slightly, but they remained statistically sig-
nificant (Table 3). Among those with femoral neck mea-
surements, after adjusting for age and sex, the predictive
risk of a baseline vertebral deformity for future vertebral
fracture was 4.9 (95% CI, 3.1–7.8). Further adjustment for
femoral neck BMD reduced this risk to 4.3 (95% CI, 2.7–
6.8). Among those with spine measurements, the corre-
sponding RRs were 3.9 (95% CI, 2.3–6.6) and 2.9 (95% CI,
1.7–5.1), respectively. There was no evidence in any of
these analyses that the timing of the BMD measurement
influenced the results.
* p Ͻ 0.05.
between the two radiographs. Thus Poisson regression can
allow for different durations of follow-up in different indi-
viduals, which is why it was preferred to the more conven-
tional logistic regression. The equation may be used to
determine the relative risk (RR) for a unit change in xi
(RR ϭ exp(␤i)), and a 95% CI can be calculated from the
SE of ␤i. The foregoing equation can also be used to predict
the IR as a function of the predictor variables. In particular,
the IR can be calculated as a function of a particular variable
when the other predictor variables are kept fixed at partic-
ular values. For example, if the IR is calculated as a function
of BMD with age kept fixed at 65 years, this is referred to
as the predicted IR adjusted to age 65 years.
In the analysis we first determined the strength of the
relationship (RR) in men and women between incident
vertebral fracture, age, spine BMD, femoral neck BMD, and
prevalent vertebral deformity. To permit direct comparisons
between men and women, the BMD values were standard-
ized to the same unit—per 0.1 g/cm2 (for comparison pur-
poses we also looked at the effect of using National Health
and Nutrition Examination Survey [NHANES] SDs(16)).
The influence of gender on these relationships was deter-
mined in a separate model including gender as a covariate.
To allow for the possibility that the strength of the associ-
ation between BMD and incidence of vertebral fracture was
influenced by the timing of the measurement of the BMD,
the interaction between BMD and the delay between first
X-ray and the BMD measurement was included in the
regression models. We then determined the effect of the
various risk factors identified on the predicted incidence of
vertebral fracture separately in men and women.
RESULTS
Influence of age and BMD on the incidence of
vertebral fracture
Subjects
In the 21 participating EPOS centers who had performed
BMD measurements, 4650 men and women had paired
The incidence of vertebral fracture as a function of spine
spinal radiographs of adequate quality to be used for the BMD was similar in men and women (Fig. 1). This shows
identification of incident vertebral fractures. In these 21 how the predicted incidence varies with spine BMD ad-
centers, 3461 subjects (74%) had bone mineral measure- justed to age 65 years. The incidence of vertebral fracture as
ments of either the hip or the spine or both. The character- a function of femoral neck BMD was greater in women than
istics of subjects who had any bone density measurements men (Fig. 2). Figures 3 and 4 present the predicted inci-
are shown in Table 1. Mean age and prevalence of vertebral dence of vertebral fracture by age separately in men and
deformity were similar in men and women. However, for all women, before and after adjusting for spine and femoral
skeletal regions, BMD values were significantly greater in neck BMD, respectively. After adjusting for spine bone
men than in women. The higher than expected prevalence of density, at all ages the incidence of vertebral fracture was
deformity in men at this age in part may be related to an similar in men and women (Fig. 3). This was true after

BMD AND RISK OF INCIDENT VERTEBRAL FRACTURE
2217
TABLE 2. RR OF INCIDENT VERTEBRAL FRACTURE IN MEN AND WOMEN
Men RR
Women RR
(95% CI)
(95% CI)
Age (per decade)
2.5 (1.5–4.3)
1.4 (1.1–1.9)
1.5 (1.0–2.1)
1.7 (1.0–2.8)
1.3 (0.9–1.9)
1.4 (0.9–2.1)
5.9 (2.7–12.9)
2.0 (1.5–2.8)
1.5 (1.3–1.9)
1.8 (1.4–2.2)
2.0 (1.5–2.6)
1.9 (1.5–2.4)
1.9 (1.5–2.4)
5.2 (3.1–8.5)
Spine BMD (per 0.1 g/cm2)
Femoral neck BMD (per 0.1 g/cm2)
Femoral neck (per SD,a aged 20–29 years)
Trochanteric BMD (per 0.1 g/cm2)
Trochanteric BMD (per SD,a aged 20–29 years)
Prevalent deformity
Age, BMD, and prevalent vertebral deformity as univariate predictors.
a NHANES SDs.(16)
TABLE 3. RR OF INCIDENT VERTEBRAL FRACTURE
IN MEN AND WOMEN
vertebral fracture. This was true for individuals with and
without a baseline prevalent vertebral deformity. Although
the increase in risk of incident vertebral fracture per unit
decrease in BMD at the femoral neck was similar in men
and women, the absolute risk at a given age and given level
of BMD was greater in women. The presence of a baseline
vertebral deformity was a stronger predictor of incident
vertebral fracture than gender.
Our study had several strengths. It was both population
based and used standardized approaches in the study design,
conduct, and analysis. However, there are a number of
methodological issues to be considered in interpreting the
results. The data were collected as part of a multicenter
study. Variation in survey methods between centers may
have lead to variation in data quality. The effect of this
though would be to dilute the strength of any observed
associations toward no effect. It seems unlikely, given that
similar considerations would have applied to men and
women, that it would have altered our main findings in
relation to the gender comparison.
RR (95% CI)
Gender (female vs. male)a
Age (per decade)b
1.2 (0.6–2.1)
1.8 (1.3–2.6)
1.3 (1.2–1.6)
1.4 (1.1–1.7)
1.4 (1.2–1.7)
2.9 (1.7–5.1)
Spine BMD (per 0.1 g/cm2)c
Femoral neck BMD (per 0.1 g/cm2)c
Trochanteric BMD (per 0.1 g/cm2)c
Prevalent deformityd
Gender, age, BMD, and prevalent vertebral deformity as multi-
variate predictors.
a Adjusted for age, spine BMD, and prevalent deformity.
b Adjusted for gender, spine BMD and prevalent deformity.
c Adjusted for age, gender, and prevalent deformity.
d Adjusted for age, gender, and spine BMD.
adjusting for trochanteric BMD, but not after adjusting for
femoral neck BMD (Fig. 4).
In the 21 centers who participated in this study, follow-up
X-rays were available in ϳ50% of those subjects recruited
to the baseline survey, although there was considerable
between-center variability in the proportion with follow-up
measurements. The predominant reasons were because of
resource constraints rather than subject refusal. In a separate
analysis, although there were differences in the baseline
characteristics between those with and without repeat spinal
radiographs, these had little effect on the overall IR ob-
served.(15) However, even if there were genuine differences
in fracture risk between those with and without follow-up
radiographs, this should not have impacted on the compar-
ison of BMD effects in men and women.
In the majority of subjects, BMD was assessed at the time
of the baseline survey. However, because of practical con-
straints a proportion of subjects had BMD assessed after the
baseline survey. However, there was no evidence that the
timing of the BMD measurement influenced the results. In
addition, when the data were reanalyzed, using BMD data
from the baseline survey only, the results remained un-
changed. Finally, our data were obtained from a predomi-
nantly white group and the results may not be extrapolated
with full confidence beyond this group.
Influence of prevalent deformity on the incidence of
vertebral fracture
The predicted incidence of vertebral fracture in men and
women with and without a baseline vertebral deformity by
age, after adjusting for spine BMD, is shown in Fig. 5.
There was no interaction between age and prevalent verte-
bral deformity on the incidence of vertebral fracture. The
presence of a prevalent vertebral deformity significantly
increased the risk of incident vertebral fracture in both men
and women. There was no significant difference in the
strength of the risk between men and women. A man with
a prevalent vertebral fracture was 2.2 (95% CI, 1.5–3.4)
times more likely to sustain an incident vertebral fracture
than a woman without a prevalent deformity and 2.5 (95%
CI, 1.2–5.5) times more likely to sustain an incident verte-
bral fracture than a woman without a prevalent deformity
after adjusting for age and spine BMD.
DISCUSSION
In this study men and women of the same age and spine
Prospective studies suggest that BMD is an important risk
bone density had the same risk of sustaining an incident factor for future fracture in women.(6) In contrast, relatively

2218
O’NEILL ET AL.
FIG. 1 Predicted incidence (%/year) of verte-
bral fracture with spine BMD in men and
women.
FIG. 2. Predicted incidence (%/year) of verte-
bral fracture with femoral neck BMD in men and
women.
FIG. 3. Predicted incidence (%/year) of verte-
bral fracture with age in men and women, unad-
justed and adjusted to a spine BMD of 1.0 g/cm2.

BMD AND RISK OF INCIDENT VERTEBRAL FRACTURE
2219
FIG. 4. Predicted incidence (%/year) of verte-
bral fracture with age in men and women, unad-
justed and adjusted to a femoral neck BMD of
0.75 g/cm2.
FIG. 5. Predicted incidence (%/year) of verte-
bral fracture with age in men and women, with
and without evidence of baseline prevalent ver-
tebral deformity, and adjusted to a spine BMD of
1.0 g/cm2.
little is known about the strength of the relationship between change in BMD (at both the spine and the femoral neck) the
fractures and BMD in men. The evidence from cross- change in the risk of incident vertebral fracture is similar in
sectional studies is somewhat conflicting. Lunt et al. found men and women.
that spine BMD was an important determinant of radio-
In addition to the gradient of risk being similar in men
graphic vertebral deformity in men and women Ͼ50 years and women, several studies support our findings in relation
old and that the gradient of risk was similar in men and to spine BMD—that the absolute risk of fracture at a given
women.(7) In contrast, Melton et al. reported that the risk of level of BMD is the same in men and women. Thus, Cheng
fragility fractures increased with decreasing BMD in both et al. assessed calcaneal BMD in a group of men and women
men and women though the risk increased more rapidly in aged 75 years and 80 years and followed the cohort during
women.(18)
a 5-year period for the occurrence of clinical fractures.(21)
There are data from several prospective studies. In the Fractures were more common in women and occurred at
Rotterdam study, De Laet et al. reported that the increase in lower bone density; however, for a given level of bone
risk of hip fracture (per SD decrease in femoral neck BMD) density the probability of fracture was the same. Using data
was similar in men and women,(19) while in the prospective from the Hawaii Osteoporosis Study, Ross et al. reported
Hawaii Osteoporosis Study, the increase in risk of vertebral that incident vertebral fractures occurred at the same level
fracture (per SD change in distal radius BMD and also of calcaneal BMD in men and women.(20) While using
calcaneal BMD) was similar in men and women.(20) Our national data concerning the incidence of hip fracture in The
results support these findings, indicating that for a given Netherlands and cross-sectional BMD data from the Rotter-

2220
O’NEILL ET AL.
dam study and assuming a similar gradient of risk, De Laet
In conclusion, in this population-based prospective study,
et al. reported that the risk of hip fracture by age and men and women of the same age and with the same spine
femoral neck bone density is similar in men and women.(22) (but not femoral neck) bone density had a similar risk of
The similarity in the relationship between BMD and incident sustaining an incident vertebral fracture. Incident vertebral
vertebral fractures in men and women observed in our study fractures are more common in middle-aged and elderly
appears at first sight somewhat surprising because it is known women than in men because at any age their spine bone
that men have larger vertebrae than women, even after density is lower.
adjusting for body size, which provides greater strength for
a given level of true (volumetric) bone density.(23,24) How-
ever, because DXA measures areal BMD, any increase in
vertebral size will result in a higher areal BMD for a given
volumetric density (the bone mineral content increases as
the cube of the change in size, while the area increases as its
square). The greater strength due to the larger vertebral size
in men may be offset in part by the apparent overestimation
of bone density (and thus strength) in larger vertebrae when
assessment is made using DXA. It is possible also that the
presence of osteophytes/disc degeneration, which are more
frequent in men than women, may influence the level of
bone mass at which fractures occur.(25)
In our study, adjusting for femoral neck BMD reduced the
difference in the predicted incidence of vertebral fracture
between men and women though the difference persisted. It
is possible that any degree of overestimation of bone density
in men compared with women may differ at the spine and
hip because of the different shapes of the bones and meth-
ods of selecting regions of interest at these sites.
Other factors that influence bone strength and that may
differ in men and women and at different skeletal sites
include microarchitecture,(26) bone quality, and bone turn-
over. However, the clinical relevance of these factors in
independently determining vertebral fracture risk is unclear.
Our data confirm the importance of a previous vertebral
fracture in determining susceptibility to further frac-
tures.(27,28) In our study the risk of incident vertebral frac-
ture was increased by a factor of fivefold in those with a
baseline vertebral deformity with the risk persisting after
adjusting for BMD. This was true in men and women and
indeed the risk for men with a previous vertebral fracture
was 2.5 times greater than for women who have not yet
sustained any fractures after adjusting for age and BMD.
In 1994, the World Health Organization (WHO) estab-
lished thresholds of BMD in women to define osteoporosis
that is widely accepted.(29) Osteoporosis in women is de-
fined as a BMD or bone mineral content of Ͼ2.5 SD below
the young average value in women. Previous investigators
have variously used thresholds for the diagnosis of osteo-
porosis in men derived from either male or female popula-
tions. Recently, it was suggested that the same absolute
diagnostic threshold be used in men as in women using the
reference standard of femoral BMD.(30) Our results, which
indicate that men and women of the same age and spine
bone density have the same risk of sustaining an incident
vertebral fracture, suggest that the same absolute threshold
of spine bone density should be used for diagnosis in men
and women. However, the data in relation to femoral neck
BMD suggest a different threshold might be appropriate.
Further data from large-scale prospective studies are re-
quired to clarify the relationship between the occurrence of
fractures and absolute BMD measured at different skeletal
sites.
ACKNOWLEDGMENTS
We thank the following individuals: Aberdeen, UK: Rita
Smith; Cambridge & Harrow, UK: Uday Bhonsle, Anna Mar-
tin, Judith Walton, and Bridget Wardley-Smith; Truro, UK:
Joanna Parsons; Oviedo, Spain: Manuel Naves Diaz, J. Ber-
nardino Diaz Lopez, and Ana Rodriguez Rebollar. We also
thank the individuals who took part in the study and the many
individuals who helped access our population samples. This
study was financially supported by a European Union (EU)
Concerted Action grant under Biomed-1 (BMH1CT920182)
and also by EU grants C1PDCT925102, ERBC1PDCT
930105, and 940229. The central coordination was also sup-
ported by the UK Arthritis Research Campaign, the Medical
Research Council (G9321536), and the European Foundation
for Osteoporosis and Bone Disease. The EU’s PECO program
linked to BIOMED-1 funded in part the participation of the
Budapest, Warsaw, Prague, Piestany, Szczecin, and Moscow
centers. Data collection from Croatia was supported by a grant
from the Wellcome Trust. The central X-ray evaluation was
generously sponsored by the Bundesministerium fur For-
schung and Technologie, Germany. The remaining funding
was provided by or through the following centers: Radiologi-
cal Evaluation Centre: Department of Radiology and Nuclear
Medicine, Free University, Berlin, Germany (DF); Co-
ordination and Data Evaluation Centers: University Institute
of Public Health, Cambridge, UK (JR, CT, ML, NC) and ARC
Epidemiology Unit, University of Manchester, UK (AJS,
TON, ML); Participating Investigative Centers: Institute of
Rheumatology, Moscow, Russia (LIB); Royal National Hos-
pital for Rheumatic Diseases, Bath, UK (AKB); Asturias Gen-
eral Hospital, Oviedo, Spain (JBC); University of Southamp-
ton, UK (CC); University Hospital, Leuven, Belgium (JD);
Medical Centre, Warsaw, Poland (KH); Clinical Hospital,
Zagreb, Croatia (IJ); University of Sheffield, UK (JAK); Med-
ical Academy, Erfurt, Germany (GK); Hospital de San Joao,
Oporto, Portugal (ALV); The Childrens Memorial Health In-
stitute, Warsaw, Poland (RL); University of Athens, Greece
(GL); Institute of Rheumatic Diseases, Piestany, Slovakia
(PM); Academy of Medicine, Szczecin, Poland (TM); Univer-
sity of Siena, Italy (GP); Erasmus University, Rotterdam, The
Netherlands (HAP); National Institute of Rheumatology and
Physiotherapy, Budapest, Hungary (GP); University of Aber-
deen, UK (DMR); University of Heidelberg, Germany (CS-N);
Charles University, Prague, Czech Republic (JS); University
Hospital, Graz, Austria (KW); Royal Cornwall Hospital,
Truro, UK (ADW).
REFERENCES
1. Ettinger B, Black DM, Nevitt MC, Rundle AC, Cauley JA,
Cummings SR, Genant HK, and the Study of Osteoporotic

BMD AND RISK OF INCIDENT VERTEBRAL FRACTURE
2221
Fractures Research Group 1992 Contribution of vertebral de-
formities to chronic back pain and disability. J Bone Miner Res
European Prospective Osteoporosis Study (EPOS). J Bone
Miner Res 17:716–724.
7:449–455.
16. Looker AC, Wahner HW, Dunn WL, Calvo MS, Harris TB,
Heyse SP, Johnston CC, Lindsay R 1998 Updated data on
proximal femur bone mineral levels of US adults. Osteoporos
Int 8:468–489.
2. Matthis C, Weber U, O’Neill TW, Raspe H, the European
Vertebral Osteoporosis Study Group 1998 Health impact as-
sociated with vertebral deformities: Results from the European
Vertebral Osteoporosis Study (EVOS). Osteoporos Int 8:364– 17. Silman AJ, O’Neill TW, Cooper C, Kanis J, Felsenberg D, the
372.
European Vertebral Osteoporosis Study Group 1997 Influence
of physical activity on vertebral deformity in men and women:
Results from the European Vertebral Osteoporosis Study.
J Bone Miner Res 12:813–819.
3. Nevitt MC, Ettinger B, Black DM, Stone K, Jamal SA, Ensrud K,
Segal M, Genant HK, Cummings SR 1998 The association of
radiographically detected vertebral fractures with back pain and
function: A prospective study. Ann Intern Med 128:793–800.
4. Cooper C, Campion C, Melton LJ III 1992 Hip fractures in the
elderly: A world wide projection. Osteoporos Int 2:285–289.
18. Melton LJ III, Atkinson EJ, O’Connor MK, O’Fallon WM,
Riggs BL 1998 Bone density and fracture risk in men. J Bone
Miner Res 13:1915–1923.
5. European Commission 1998 Report on osteoporosis in the 19. De Laet CEDH, Van Hout BA, Burger H, Weel AEAM,
European community—Action for prevention. Office for Of-
ficial Publications of the European Communities, Luxem-
bourg, Belgium.
Hofman A, Pols HAP 1998 Hip fracture prediction in elderly
men and women: Validation of the Rotterdam Study. J Bone
Miner Res 13:1587–1593.
6. Marshall D, Johnell O, Wedel H 1996 Meta-analysis of how 20. Ross PD, Lombardi A, Freedholm D 1999 The assessment of
well measures of bone mineral density predict occurrence of
osteoporotic fractures. BMJ 312:1254–1259.
7. Lunt M, Felsenberg D, Reeve J, Benevolenskaya L, Cannata J,
bone mass in men. In: Orwoll ES (ed.) Osteoporosis in Men:
The effect of Gender on Skeletal Health. Academic Press, San
Diego, CA, USA, pp. 505–525.
Dequeker J, Dodenhof C, Falch JA, Masaryk P, Pols HAP, 21. Cheng S, Suominen H, Sakari-Rantala R, Laukkanen P, Avi-
Poor G, Reid DM, Scheidt-Nave C, Weber K, Varlow J, Kanis
JA, O’Neill TW, Silman AJ 1997 Bone density variation and
its effects on risk of vertebral deformity in men and women
kainen V, Heikkinen E 1997 Calcaneal bone mineral density
predicts fractures occurrence: A five year follow up study in
elderly people. J Bone Miner Res 12:1075–1082.
studied in thirteen European centres: The EVOS study. J Bone 22. De Laet CEDH, van Hout BA, Burger H, Hofman A, Pols
Miner Res 12:1883–1894.
8. Saag KG, Emkey R, Schnitzer TJ, Brown JP, Hawkins F,
HAP 1997 Bone density and risk of hip fracture in men and
women: Cross sectional analysis. BMJ 315:221–225.
Goemaere S, Thamsborg G, Liberman UA, Delmas PD, Mal- 23. Gilsanz V, Boechat MI, Gilsanz R, Loro ML, Roe TF, Good-
ice MP, Czachur M, Daifotis AG, for the Glucocorticoid-
man WG 1994 Gender differences in vertebral sizes in adults:
Induced Osteoporosis Intervention Study Group 1998 Alen-
Biomechanical implications. Radiology 190:678–682.
dronate for the prevention and treatment of glucocorticoid 24. Ross PD, Huang C, Davis JW, Wasnich RD 1995 Vertebral
induced osteoporosis. N Engl J Med 339:292–299.
9. Orwoll E, Ettinger M, Weiss S, Miller P, Kendler D, Graham
dimension measurements improve prediction of vertebral frac-
ture incidence. Bone 16:257S–262S.
J, Adami S, Weber K, Lorenc R, Pietschmann P, Vandormael 25. O’Neill TW, McCloskey EV, Kanis JA, Bhalla AK, Reeve J,
K, Lombardi A 2000 Alendronate for the treatment of osteo-
porosis in men. N Engl J Med 343:604–610.
10. O’Neill TW, Felsenberg D, Varlow J, Cooper C, Kanis JA,
Reid DM, Todd C, Woolf AD, Silman AJ 1999 The distribution,
determinants, and clinical correlates of vertebral osteophytosis: A
population based survey. J Rheumatol 26:842–848.
Silman AJ, the European Vertebral Osteoporosis Study Group 26. Aaron JE, Makins NB, Sagreiya K 1987 The microanatomy of
1996 The prevalence of vertebral deformity in European men
and women: The European Vertebral Osteoporosis Study.
J Bone Miner Res 11:1010–1018.
trabecular bone loss in normal aging men and women. Clin
Orthop 215:260–271.
27. Ross PD, Davis JW, Epstein RS, Wasnich RD 1991 Pre-
existing fracture and bone mass predict vertebral fracture in-
cidence in women. Ann Intern Med 114:919–923.
11. Kalender WA, Felsenberg D, Genant HK, Fischer M, De-
queker J, Reeve J 1995 The European Spine Phantom—A tool
for standardisation and quality control in spinal bone mineral 28. Klotzbuecher CM, Ross PD, Landsman PB, Abbott TA III,
measurements by DXA and QCT. Eur J Radiol 20:83–92.
12. Pearson J, Dequeker J, Henley M, Bright J, Reeve J, Kalender
W, Laval-Jeantet AM, Ruegsegger P, Felsenberg D, Adams J,
Berger M 2000 Patients with prior fractures have an increased
risk of future fractures: A summary of the literature and
statistical synthesis. J Bone Miner Res 15:721–739.
Birkenhager JC, Braillon P, Diaz Curiel M, Fischer M, Galan 29. World Health Organization (WHO) 1994 Assessment of frac-
F, Geusens P, Hyldstrup L, Jaeger P, Jonson R, Kalef-Ezras
J, Kotzki P, Kroger H, van Lingen A, Nilsson S, Osteaux
M, Perez Cano R, Reid DM, Reiners C, Ribot C, Schneider
ture risk and its application to screening for post-menopausal
osteoporosis. WHO Technical Report, series 843, WHO, Ge-
neva, Switzerland.
P, Slosman DO, Wittenberg
G 1995 European semi- 30. Kanis JA, Gluer CC, for the Committee of Scientific Advisors,
anthropomorphic spine phantom for the calibration of bone
densitometers: Assessment of precision, stability and accu-
racy. The European Quantitation of Osteoporosis Study
Group. Osteoporos Int 5:174–184.
International Osteoporosis Foundation 2000 An update on the
Diagnosis and assessment of osteoporosis with densitometry.
Osteoporos Int 11:192–202.
13. Lunt M, Felsenberg D, Adams J, Benevolenskaya L, Cannata
J, Dequeker J, Dodenhof C, Falch JA, Johnell O, Khaw KT,
Masaryk P, Pols H, Poor G, Reid D, Scheidt-Nave C, Weber
K, Silman AJ, Reeve J 1997 Population-based geographic
variations in DXA bone density in Europe: The EVOS study.
Osteoporos Int 7:175–189.
Address reprint requests to:
Terence O’Neill, M.D., F.R.C.P.
ARC Epidemiology Research Unit
University of Manchester
14. McCloskey E, Spector TD, Eyres KS, Fern ED, O’Rourke N,
Vasikaran S, Kanis JA 1993 The assessment of vertebral
deformity: A method for use in population studies and clinical
trials. Osteoporos Int 3:138–147.
Manchester M13 9PT, UK
15. The European Prospective Osteoporosis Study Group 2002 Received in original form August 17, 2001; in revised form March
Incidence of vertebral fracture in Europe: Results from the 6, 2002; accepted June 17, 2002.