THE AMERICAN JOURNAL OF GASTROENTEROLOGY
1, 2000
© 2000 by Am. Coll. of Gastroenterology
Published by Elsevier Science Inc.
Risk of Fractures in Celiac Disease Patients:
A Cross-Sectional, Case-Control Study
Horacio Vazquez, M.D., Roberto Mazure, M.D., Diana Gonzalez, M.D., Daniel Flores, M.D.,
Silvia Pedreira, M.D., Sonia Niveloni, M.D., Edgardo Smecuol, M.D., Eduardo Mauriño, M.D., and
Julio C. Bai, M.D.
Small Bowel Section, Clinical Service, Hospital de Gastroenterologı́a “Dr. Carlos Bonorino Udaondo,”
Gastroenterology Chair, Universidad del Salvador; and Centro de Osteopatı́as Médicas y Reumatologı́a,
Hospital de Clı́nicas (UBA), Buenos Aires, Argentina
OBJECTIVES: Although osteopenia and osteoporosis are
well-recognized complications of celiac disease, no con-
trolled studies have been done to assess the prevalence of
fractures in a large cohort of patients. The objectives of this
study were to determine the prevalence of bone fractures
and vertebral deformities in celiacs and to analyze the re-
lationship between fractures and clinical data of patients.
METHODS: We studied 165 patients with a well-established
diagnosis of celiac disease. A similar number of age- and
gender-matched control subjects with functional GI disor-
ders were evaluated. The design of the study was cross-
sectional, with a retrospective historical review through a
personal interview of all subjects. All patients underwent
bone mineral density measurement by dual-energy, x-ray
absorptiometry and spinal x-ray. Vertebral deformities were
determined by visual inspection of spinal x-rays and by
morphometric analysis.
RESULTS: Among celiacs, 41 patients (25%) referred have
had from one to five fractures in the peripheral skeleton. On
the contrary, only 14 (8%) control subjects experienced
fractures. This difference was highly significant (odds ratio,
3.5; 95% confidence interval [CI], 1.8 –7.2; p ⬍ 0.0001).
Although inspection of spinal x-rays showed evidence of
vertebral deformities in the lumbar spine in only two pa-
tients, a more detailed examination of lateral x-rays using
morphometric criteria detected lumbar spine vertebral de-
formities in nine (five also had fractures in the peripheral
skeleton) and in four controls (odds ratio, 2.8; 95% CI,
0.7–11.5; p ⫽ NS). Eighty percent of fractures were de-
tected before the diagnosis of celiac disease or in patients
who were noncompliant with the gluten-free diet; only 7%
of patients experienced fractures after starting treatment.
Regression analysis adjusted for multiple comparisons
showed that patients with fractures were diagnosed with
celiac disease later (p ⬍ 0.06) and remained undiagnosed
for more prolonged periods (p ⬍ 0.05). There was a trend,
which did not reach statistical significance, for a lower bone
mineral density in the lumbar spine and total skeleton
among patients with fractures.
Vol. 95, No.
ISSN 0002-9270/00/$20.00
PII S0002-9270(99)00741-8
CONCLUSIONS: This study has demonstrated that patients
with celiac disease had a high prevalence of bone fractures
in the peripheral skeleton. Most of these events occurred
before diagnosis or while patients were noncompliant with
gluten-containing diet. Our results suggest that early diag-
nosis and effective treatment of celiac disease were the most
relevant measures to protect patients from the risk of
fractures. (Am J Gastroenterol 2000;95:183–189. © 2000 by
Am. Coll. of Gastroenterology)
INTRODUCTION
Osteoporotic fractures are a major public health problem
worldwide. It has been calculated that 1.2–1.3 million frac-
tures occur annually in the United States (1, 2). In osteopo-
rosis, the risk of fractures has been shown to be increased
two- to threefold for every z-score decrease in bone mineral
density (BMD) (1). The most common and well-known sites
of osteoporotic fractures are at the distal forearm, spine, hip,
ankle, humerus, and pelvis (1).
Osteopenia and osteoporosis are well-recognized compli-
cations of celiac disease and constitute a major problem
through their association with fractures (3–5). In the last few
years, several studies have determined, using very sensitive
methods, that as many as 75% of untreated patients have a
decreased BMD (4 – 6). Although lower in degree and fre-
quency, osteopenia can also be observed among patients
with subclinical or asymptomatic disease (7–9). The etiol-
ogy of bone loss in celiac disease is likely to be multifac-
torial. Calcium and vitamin D malabsorption, low physical
activity, and malnutrition are, among others, factors impli-
cated in the bone disorders (4, 7–14). All metabolic studies
suggest that osteopenia in celiac disease is a consequence of
secondary hyperparathyroidism (9, 12–14). Recently, proin-
flammatory and antiinflammatory cytokines (interleukin-1␤,
interleukin-6, and interleukin-1 receptor antagonist) were
suggested as playing an active role in celiac disease-asso-
ciated osteopenia (15).
Earlier studies have demonstrated that the most common
clinical manifestations of osteopenia in celiac disease pa-
184 Vazquez et al.
tients are bone pain, altered gait, retardation of growth, bone
deformities, and fractures (10, 11). Pathological bone frac-
tures (minimal trauma fractures) and deformities have been
reported anecdotally in celiac disease patients; most were
detected before diagnosis of the disease was made (10).
However, epidemiological information on fractures in celiac
disease patients is excessively scarce. Thus, up to now, no
controlled studies have been reported assessing the preva-
lence of fractures in a large cohort of celiac patients. As
treatment of these fractures is expensive and often requires
surgery, the aim should be prevention. However, although
there is a general consensus that a gluten-free diet improves
previously affected BMD (4, 9, 12, 13–16), the effect of
treatment on bone fractures has not been assessed.
Our aims in the present case-control study were to deter-
mine the prevalence of fractures and vertebral deformities in
patients with celiac disease and to define the relationship
between bone disturbances and the clinical characteristics of
patients.
MATERIALS AND METHODS
Patients
One hundred and sixty-five patients (143 women; median
age, 40 yr; range, 16 –74 yr) diagnosed with celiac disease
took part in this study. They were part of 183 consecutive
patients who attended the malabsorption clinic of the Hos-
pital de Gastroenterologı́a of Buenos Aires. Patients who
were included completed the study protocol including spinal
x-ray and BMD determination. At entry to the study, 51
patients were considered untreated; 38 were evaluated at the
time of diagnosis and 13 who were not compliant with the
diet despite diagnosis had evaluation performed earlier.
According to the initial interview, 74 of 114 patients were
considered to be on a strict gluten-free diet and the remain-
ing 40 were on partial restriction. The diagnosis of celiac
disease was based on the combination of a compatible
clinical picture (chronic diarrhea with or without malabsorp-
tion, anemia, malnutrition, etc.), a flat intestinal mucosa at
the time of diagnosis, and a positive clinical and/or histo-
logical response to a gluten-free diet. At diagnosis, positive
celiac disease-related serology (antigliadin and/or antien-
domysium antibodies) was observed in 103 patients. Post-
treatment small bowel biopsy demonstrated histological im-
provement in 38 patients. Gluten-free diet produced
negative or reduced titers of celiac disease-related serology
in all those with positive serology. Patients were excluded if
they had evidence of secondary causes of osteoporosis other
than celiac disease (primary biliary cirrhosis, thyroid dys-
function, steroid treatment, chronic treatment with drugs
known to affect bone and mineral metabolisms, etc.). At the
time of entry to the study, 38 patients (23%) were older than
50 yr.
Controls
A cohort of 165 subjects (143 women; mean age, 41 yr;
range, 16 –74 yr) was selected from the outpatient clinic of
AJG – Vol. 95, No. 1, 2000
the Hospital de Gastroentrologı́a of Buenos Aires. They
were included only if their final diagnosis was a functional
gastrointestinal (GI) disorder. Thirty-six subjects (21%)
were older than 50 yr. We also excluded cases with disor-
ders known to affect bone metabolism or those with chronic
use of medications known to affect bones.
Study Design
The design of the study was cross-sectional and was based
on a retrospective historical review through a personal in-
terview by the physician in charge and using a redesigned
form. Data reported in the clinical history of patients
through the follow-up were also considered. Epidemiolog-
ical information evaluated were patient’s age; age at diag-
nosis of celiac disease; estimated age at clinical onset of
celiac disease; number and site of fracture(s); age at diag-
nosis of fractures and relation to diagnosis of celiac disease
and compliance with a gluten-free diet; type and intensity of
trauma producing fractures; site of occurrence of fractures;
number of pregnancies; smoking habit; long-term medica-
tions; and hormone replacement therapy. Trauma was con-
sidered severe if it involved a traffic accident, was sports-
related, or caused by falling from a height; moderate if the
fracture resulted from slipping or stumbling or from a fall on
level ground; and mild when trauma minimal was involved.
Bone Mineral Density Measurement
We measured BMD of the lumbar spine, total skeleton, and
femoral neck by dual-energy x-ray absorptiometry. The
reproducibility of measurements and the coefficient of vari-
ation of the method have been reported previously (16).
Results are reported as z-scores that represent the number of
standard deviations (SDs) by which an individual value is
separated from the corresponding mean normal value, cor-
rected for gender and age. z-score is calculated by the
following formula:
Measured BMD ⫺ Normal mean BMD
(age and sex-matched controls)
SD of the normal BMD
Clinical and Nutritional Assessment
At the time of the patient interview, a clinical assessment
and biochemical studies were performed. Data on the clin-
ical outcome and dietary compliance with a gluten-free diet
were obtained by the physician in charge, who is experi-
enced in malabsorption. The interview protocol included
precise reference to the age of appearance of diarrhea,
weight loss, anemia, or other symptoms related to celiac
disease. Body weight and body mass index (BMI) were
determined in all patients. Body mass index was determined
by the following formula: Body weight/(height)2 ⫽ (kg/m2)
Radiological Assessment
Vertebral fractures and/or deformities, common findings in
osteoporotic subjects, can be present in otherwise asymp-
tomatic subjects. Therefore, we aimed to detect such abnor-
AJG – January, 2000 Risk of Fracture in Celiac Patients 185

Table 1. Epidemiological Data of Untreated and Treated Celiac Disease Patients at the Time of Entry to the Study
Celiac Patients
Treated
Untreated Strict GFD Partial GFD
Number 51 74 40
Median age, yr 34 44 32
(Range) (16–73) (21–74) (16–70)
Age at diagnosis, median, yr 32 39 20
(Range) (1–74) (1–74) (1–66)
Estimate time from onset of symptoms to diagnosis, months
Median 119 167 86
(Range) (0–780) (0–708) (0–660)
Median time on gluten restriction, months 29 78
Bone mineral density (z-score), median (95% CI)
Lumbar spine ⫺1.3 (⫺2.2 to ⫺1.0) ⫺1.2 (⫺1.3 to ⫺0.5) ⫺1.0 (⫺1.7 to ⫺0.5)
Total skeleton ⫺2.0 (⫺2.9 to ⫺1.6) ⫺1.6 (⫺1.9 to ⫺1.0) ⫺0.9 (⫺1.4 to ⫺0.1)*
Body mass index (X ⫾ SEM) 20.3 ⫾ 0.4 21.6 ⫾ 0.5 22.6 ⫾ 0.5
Versus untreated patients p ⬍ 0.02; *p ⬍ 0.002.
Treated patients were discriminated according to the degree of compliance with GFD (strict gluten restriction and partial adherence).
CI ⫽ confidence interval; SEM ⫽ standard error of the mean; GFD ⫽ gluten-free diet.

malities by performing a radiological assessment of the Statistics

lumbar spine of patients and controls. Each patient and 93
(62%) controls subjects had anteroposterior and lateral ra-
diographs of the lumbar spine. Although all x-rays were
evaluated for anatomical deformities (visual inspection by
an experienced radiologist), only radiographs from 68 pa-
tients and 78 control subjects were considered of adequate
quality for the morphometric assessment. A comparison of
the demographic and clinical data of those patients and
controls who were documented with the morphometric anal-
ysis with those who did not showed no significant differ-
ences. Films were taken in a standardized manner centered
on vertebra L2 and vertebral deformities were defined (by an
experienced radiologist) using a morphological criterion and
by the morphometric algorithm of Eastell et al. (17). In this
method the anterior (ha), posterior (hp), and mid (hm)
heights of each vertebra from D12 to L5 were measured by
a single blinded observer from the lateral radiograph. Using
these parameters, we defined % wedge, % biconcavity, and
% compression using the following formulas:
% wedge ⫽ (hp ⫺ ha)/hp ⫻ 100
% biconcavity ⫽ (hp ⫺ hm)/hp ⫻ 100
% compression ⫽ (hp⬘ ⫺ hp)/hp⬘ ⫻ 100
where hp⬘ is the posterior height of the vertebra below or the
vertebra above.
The intra- and interobserver variability coefficients of
these measurements were 3.2% and 9.1%, respectively. A
patient was considered as having a fracture if she or he had
at least two minor deformities (deformities of two vertebra
by at least ⫺2.5 SD) or a single vertebral deformity greater
than ⫺3 SD below the mean normal value for that vertebra
(obtained from measurements in healthy controls) (18).
Results are reported as median and range or 95% confidence
interval (CI) or mean ⫾ SEM when appropriate. Compari-
son of results among the different groups where performed
using the t test or Mann-Whitney test. The proportion of
patients and controls with fractures were compared by cal-
culation of odds ratios and 95% CI. Uni- and multivariate
linear regression analyses were performed.
RESULTS
Clinical Aspects
Table 1 shows some epidemiological data of patients at
entry to the study. Patients on a strict gluten-free diet were
10 and 12 yr older than untreated celiacs and those who
partially adhered to the diet, respectively. The estimated
time from onset of symptoms to diagnosis was similar in the
three groups. The median time on gluten restriction was
longer in partially compliant patients. Untreated patients had
a nonsignificant lower lumbar spine BMD z-score, com-
pared with treated patients. At the total skeleton level, mean
BMD z-score of untreated patients was significantly reduced
with respect to treated celiacs (p ⬍ 0.05). Untreated patients
presented a lower but nonsignificant BMI compared with
treated patients.
Frequency of Fractures in the Peripheral Skeleton
A total of 41 patients with celiac disease (25%) and 14
healthy controls (8%) reported fractures in the peripheral
skeleton (odds ratio, 3.5; 95% CI, 1.8 –7.2; p ⬍ 0.0001). A
total of 51 and 15 fractures were identified in celiac patients
and controls, respectively. The wrist and radius were the
commonest sites of fracture (13 cases) among patients. Due
to the retrospective characteristic of our interview and to the
fact that no medical records of the trauma events were
available, the intensity of trauma was only partially consid-
186 Vazquez et al. AJG – Vol. 95, No. 1, 2000

Table 2. Demographic and Clinical Characteristics of Celiac Disease Patients and Gender- and Age-Matched Controls According to
the Presence or Absence of Fractures (Including Vertebral Deformities)
Celiac Disease Patients Control Population
With Without With Without
Fracture Fracture Fracture Fracture
Number 45 120 18 147
Gender (F/M) 38/7 105/15 15/3 129/19
Median age, yr 45 38* 50 39†
Range (16–74) (16–70) (19–74) (16–70)
Age at diagnosis, median, yr 38 30‡
(Range) (1–74) (1–68)
Estimated time from onset of symptoms to diagnosis, yr
Median 7 1
(Range) (0–65) (0–59)
Number of women in menopause at the time of the study 10 28 9 32
Number of subjects with fractures caused by severe trauma 3 4
Total number of pregnancies 30 68
BMD (z-score)
L2–L4, median ⫺1.4 ⫺1.2
(95% CI) (⫺2.0 to ⫺0.9) (⫺1.4 to ⫺0.7)
Total skeleton, median ⫺1.7 ⫺1.5
(95% CI) (⫺2.3 to ⫺1.1) (⫺1.9 to ⫺1.1)
BMI, kg/m2
X ⫾ SEM 21.4 ⫾ 0.5 21.4 ⫾ 0.4
*p ⬍ 0.007; †p ⬍ 0.004; ‡p ⬍ 0.005.
BMD ⫽ bone mineral density; BMI ⫽ body mass index; CI ⫽ confidence interval; SEM ⫽ standard error of the mean.

ered in this analysis. Thus, three patients and four controls fractures (211 pregnancies in 68 patients; odds ratio, 1.06;
presented fractures as a consequence of severe trauma. No 95% CI, 0.62–1.79). Mean lumbar spine and total skeleton
difference was detected on comparing age at the time of BMD of patients with fractures was lower than that assessed
fractures between patients and controls (median age, 31 yr). in those celiacs without fractures. Table 3 shows the uni-
Thirty-two percent of men and 27% of women had fractures. variate logistic regression analysis comparing fractures in
In the control group, the prevalence of fractures in men was celiac disease patients with gender, age at entry to the study,
also slightly greater than that observed in women (14% and age at onset of symptoms, age at diagnosis, age at menarche,
10%, respectively). Thirty-four patients (75%) and 12 con- presence of menopause, time on gluten-free diet, anthropo-
trols (67%) had fractures before age 50 yr. In 36 patients metric parameters (weight, height, and BMI), serum anti-
(80%), peripheral fractures were produced before diagnosis gliadin antibodies titers (IgG and IgA), and BMD (absolute
or in diagnosed patients who had never adhered to specific values and z-scores) at the lumbar spine and total skeleton
treatment. Only nine of 117 patients (8%) who had started levels, This analysis showed a significant correlation with
treatment at the time of the study experienced fractures age of patients (p ⬍ 0.01), age of clinical onset of symptoms
while on a gluten-free diet. Interestingly, in three of these (p ⬍ 0.06), and treatment with a gluten-free diet (p ⬍ 0.05).
patients who fractured while treated, the events occurred as
a consequence of severe trauma.
Table 2 shows demographic and clinical data of patients Table 3. Multivariate Regression Analysis Adjusted for Multiple
Comparisons on the Presence of Fractures With Several Clinical
and controls segregated according to whether or not they Parameters
presented fractures. The median age for patients and con-
trols with fractures was greater than those of subjects with- Variable r p Value
out fractures by 7 (p ⬍ 0.007) and 11 (p ⬍ 0.004) yr, Gender 0.057 0.46
respectively. Age at the time of diagnosis of celiac patients Age at onset of symptoms 0.028 0.73
Age at diagnosis 0.145 0.06*
was significantly greater in patients with fractures than in Age at the time of starting GFD 0.152 0.05*
those without bone lesions. Furthermore, the median time Age at menarche 0.016 0.85
from onset of symptoms to diagnosis was greater in patients Age at the entry of the study 0.198 0.01*
with fractures. This difference was not significant. There Time on gluten-free diet 0.028 0.72
was a difference in the proportion of pregnant women be- Lumbar spine z-score 0.126 0.19
Total skeleton z-score 0.062 0.51
tween patients with fractures (30 of 38) and those without AGA-A serum titers 0.045 0.60
fractures (68 of 104; Yates corrected ␹2, 1.8; p ⫽ NS). BMI 0.031 0.73
Patients with fractures presented a similar number of preg- *Statistical significance.
nancies (88 pregnancies in 30 patients) than those without GFD ⫽ gluten-free diet; AGA ⫽ antigliadin antibodies; BMI ⫽ body mass index.
AJG – January, 2000
Figure 1. Cumulative percentage of fractures (Kaplan-Mayer life
table). By the age of 70 yr, 43% of celiac disease patients (䊉) and
20% of controls (E) have presented fractures (␹2 11.1; p ⬍
0.0001).
Figure 1 shows data on the cumulative proportion of patients
and controls presenting fractures. The figure shows that
patients had a significantly higher prevalence of fractures
than controls. At the end of the eighth decade of life, 43%
of patients and 20% of controls were estimated to present
fractures (␹2 ⫽ 11.1; p ⬍ 0.001). Figure 2 shows data on the
incidence of fractures according to age. Patients presented
two peaks of incidence that, compared to the incidence
observed in controls, were statistically different: between
the third and four decades of life (p ⬍ 0.03) and between the
sixth and seventh decades (p ⬍ 0.05).
Assessment of Vertebral Deformities
The radiological inspection of lumbar spine radiographs
revealed Schmorl’s nodules in two patients. Interestingly,
according to the morphometric analysis, four of 78 (5%)
control subjects and nine of 68 (13%) celiac patients (eight
Figure 2. Incidence of fractures and vertebral deformities among
celiac patients (䊉) and age- and gender-matched controls (E)
according to the age of bone fracture. Patients presented two
significant peaks of incidence, one between the third and fourth
decades of life (compared with controls: *p ⬍ 0.03) and the other
between the sixth and seventh decades (**p ⬍ 0.05).
Risk of Fracture in Celiac Patients 187
women) presented a single or multiple vertebral deformities
(odds ratio, 2.8; 95% CI, 0.7–11.5; p ⫽ NS). Three of the
control subjects were older than 60 yr at the time of diag-
nosis of vertebral deformities and only one of them had two
vertebrae compromised. Six patients exhibited vertebral de-
formities in only one vertebra (four patients had compres-
sion in L3, one in D12, and one presented wedge in L5), two
others were affected in two vertebrae (one had wedge in L1
and L2 and another in L3 and L5), and one had deformities
in three different locations (wedge in D12 and L1, and
compression in L2). Biconcavity was associated with other
types of vertebral deformities in four patients. Three celiacs
were diagnosed with vertebral fractures at ages older than 60
yr. Five of the patients with vertebral deformities also re-
ported fractures in the peripheral skeleton. Patients with
vertebral fractures had a nonsignificant lower BMD z-score
at the lumbar spine compared with those without this type of
fractures (⫺1.8 ⫾ 0.7 vs ⫺1.2 ⫾ 0.1, respectively). Celiac
disease patients had a significantly lower percentage of
wedge in L3 (p ⬍ 0.03) and L5 (p ⬍ 0.02), and a greater
percentage of biconcavity in D12 (p ⬍ 0.03), compared with
the values for controls. Patients with fractures also presented
a significantly greater percentage of deformity (p ⬍ 0.03)
and percentage of biconcavity (p ⬍ 0.01) than patients
without fractures.
DISCUSSION
Osteoporotic fractures increase in frequency with age, are
more prevalent in women than men, occur in sites where
trabecular bone predominates, and are associated with min-
imal or moderate trauma. Although skeletal deformities,
fractures, osteoporosis, and osteomalacia have been re-
ported in celiac disease patients, the true magnitude of this
problem remains unknown. To our knowledge, this is the
first controlled survey on the prevalence of bone fractures in
celiac disease patients. The present study, based on a retro-
spective review, clearly shows that celiac patients presented
a significantly increased prevalence of fractures in the pe-
ripheral skeleton compared with gender- and age-matched
healthy controls. While inspection of spinal x-rays showed
evidence of vertebral deformities in the lumbar spine in only
two patients, a more detailed examination of lateral x-rays
using morphometric criteria detected a higher number of
cases with vertebral deformities, in patients and in controls.
However, the prevalence of fractures in the axial skeleton of
celiacs was not significantly different than in controls. Most
fractures were produced by minimal or moderate trauma. In
contrast to age-related osteoporotic fractures, three-quarters
of fracture cases were observed in patients younger than 50
yr. Our study estimated that by the end of the eighth decade
of life the cumulative prevalence of fractures was 43% for
celiac patients and 20% for the control population. Twenty-
seven percent of fractures of peripheral bones were located
in the wrist, radius, and ulna. Most fractures reported by
celiac patients occurred before diagnosis or in patients who
188 Vazquez et al.
were noncompliant with the gluten-free diet. Based on the
fact that only 7% of patients developed fractures after the
start of treatment, we suggest that a gluten-free diet seems
to be the most relevant single factor protecting patients from
this risk. This fracture rate was very similar to that of the
control population. Finally, patients who experienced frac-
tures had a nonsignificantly lower BMD than those who had
no history of fractures.
An important aspect that deserves further comment is
related to the retrospective characteristic of our study. In
clinical research, a prospective evaluation is the best way to
confirm a finding; however, several data from our study
allows us to suggest that conclusions from a prospective
protocol would not give the true prevalence of bone frac-
tures in celiac disease. Such prospective evaluation would
only be possible in diagnosed patients who, by medical and
ethical reasons, should be treated with a gluten-free diet. As
was shown by our study, patients on a gluten-free diet have
a similar prevalence of fractures as the control population.
Our data on the prevalence of bone fractures in celiacs
showed a similar prevalence as that previously reported by
two noncontrolled studies (3, 7) and in other autoimmune
inflammatory processes such as rheumatoid arthritis (18, 19)
and ankylosing spondylitis (20). Because celiac disease is
largely underdiagnosed, it seems reasonable to analyze
whether the data presented in the present study were over-
estimated or whether they represent the true magnitude of
the problem. Our celiac disease population consisted of
patients with either overt malabsorption or subclinical forms
of celiac disease. According to recent epidemiological stud-
ies, both subgroups would only represent approximately
50% of all gluten-sensitive cases (21). Therefore, it seems
realistic to believe that a degree of selection bias could be
present. However, assuming that all those existing cases not
included in this study presented with fractures at the same
rate as that estimated for control population, the theoretical
prevalence of fractures occurring in this larger population
would still remain almost double that reported for controls.
On the other hand, underestimation of the true prevalence of
bone fractures could also be argued by the fact that the
present cross-sectional study was based on predominantly
young adult populations (almost 80% were younger than 50
yr old). Thus, it seems possible that we were omitting a
number of age-related osteoporotic fractures. However, our
calculations on the cumulative number of cases with frac-
tures estimated that the probability of being free of fractures
at the end of the seventh decade of life was significantly
lower in celiacs, compared with controls. Interestingly, pa-
tients and controls with fractures exhibited peaks of inci-
dence at the third and sixth decades of life; however, pa-
tients presented a significantly greater number of affected
members in both periods. This finding would suggest that
bones of patients would intrinsically be more susceptible to
fracture as a consequence of similar mechanical require-
ments.
Our study differs from the only case-control study on
AJG – Vol. 95, No. 1, 2000
fractures in celiac patients, recently reported in abstract
form. Based on a mailing questionnaire, Thomason et al.
(22) demonstrated a similar prevalence of fractures in celi-
acs and controls. They determined that 46% of male and
28% of female patients reported any type of fracture. Sim-
ilar figures were observed in controls (42% and 30%, re-
spectively). As can be observed, the overall rate of fractures
in the study from England (34%) was greater than we
observed. However, our most significant difference from the
study of Thomason et al. (22) was related to the very high
prevalence of fractures that they reported for the control
population. Although we observed fractures in the periph-
eral skeleton in 8% of age- and gender-matched controls,
their controls reported an excessively high prevalence. Our
data on the prevalence of fractures in controls were similar
to those reported for healthy populations investigated in
other studies. Our study also differs from that of Thomason
et al. (22) in that, although these authors only evaluated
mailing replies, we performed in-person interviews, with
lumbar spine radiographs in which a morphometric analysis
of vertebral deformities was considered. There is no other
obvious explanation for the differences between these stud-
ies. Data on age at entry to the study, age at diagnosis, and
time of overt symptoms without specific diagnosis in both
cohorts should be compared to obtain clues for a better
understanding of these differences.
Most fractures of peripheral bones are easily reported by
patients. In contrast, vertebral deformities are frequently
underestimated and not commonly reported by patients be-
cause they frequently produce mild symptoms or nonspe-
cific complaints. Our study also aimed, for the first time in
celiac patients, to analyze the proportion of vertebral defor-
mities on spinal x-rays using morphological and morpho-
metric criteria. This analysis did not show a significant
difference in the prevalence of vertebral deformities be-
tween patients and controls. However, the concomitant pres-
ence of vertebral deformities (sometimes affecting more
than one bone) and peripheral fractures was only detected in
celiac disease patients, suggesting that some of them can be
affected by a more generalized deterioration of plane and
trabecular bones.
Although we and others have identified risk factors for
osteoporosis and fractures, we still cannot determine why
some patients show a marked reduction in bone mass and
are prone to multiple fractures, whereas other patients with
similar risk factors do not have such propensity despite a
very low BMD. Bone fractures are the consequence of the
counterbalance between aggressive factors, such as trauma
and its intensity on the one hand, and bone resistance and
degree of protection provided by the muscle structure, on
the other (10). Although decreased BMD is a constant
finding among celiacs, the poor correlation between the
presence of fractures and bone mass suggests that additional
factors should be concurrent. Based on these facts, we
should explore the etiology of such propensity, evaluating
other factors in the equation. Skeletal muscles are dependent
AJG – January, 2000
on their function; therefore, physical activity gives additional
protection to bones to resist trauma. It has been argued that a
low muscle mass and strength in celiacs could be responsible
for the increased prevalence of bone fractures (23). However,
the chronic disease state of patients and its concomitant low
physical activity may often counterbalance such risk.
In our opinion, another aspect that deserves consideration
is related to the fact that some patients with severe clinical
compromise and a very low BMD at diagnosis show a
striking clinical improvement induced by treatment, allow-
ing a rapid normalization of physical activity. Although they
normalize weight, fat, and lean masses very rapidly, bone mass
increment is a slow process such that, after 1 yr of treatment,
is very rarely a ⬎10% increase over the baseline mass (11, 12,
24). Normal physical activity could expose patients to unex-
pected mechanical requirements and, therefore, to an additional
increased theoretical risk of fractures. However, the prevalence
of fractures in patients who are on gluten restriction is similar
to that of the control population. Bone mineral density mea-
surement is believed to reflect approximately 75% of the vari-
ability in bone strength; bone quality contributes to the remain-
ing 25% (18). The mechanical efficacy of bone structure as a
protective factor from fractures has received attention only
very recently. We suggest that the quality of bone matrix in
patients with celiac disease should be better understood to
define bone fragility appropriately.
In summary, we have demonstrated a considerable in-
crease in the risk of fractures in the peripheral skeleton of
celiac disease patients. Such propensity was not observed in
the axial skeleton. Early diagnosis and effective treatment
were the most relevant measures protecting patients from
fractures. Although a lower BMD was observed in patients
with fractures, it does not seem to explain the increased risk.
We speculate that there must be additional factors that
should be explored (for example, a bone quality defect),
which may be participating in the fracture propensity ob-
served in patients with celiac disease.
ACKNOWLEDGMENT
We are grateful to Dr. Christina Surawicz from the Univer-
sity of Washington, Harbor View Medical Center, Seattle,
Washington, and Dr. Eamonn Quigley from the University
of Nebraska Medical Center, Omaha, Nebraska, for their
generous help and critical reading of our manuscript.
Reprint requests and correspondence: Julio C. Bai, M.D., Hos-
pital de Gastroenterologı́a, “Dr. Carlos Bonorino Udaondo,” Ca-
seros 2061, (1436) Buenos Aires, Argentina.
Received Mar. 29, 1999; accepted Aug. 18, 1999.
REFERENCES
1. Lips P. Epidemiology and predictors of fractures associated
with osteoporosis. Am J Med 1997;103:3S–11S.
Risk of Fracture in Celiac Patients 189
2. Melton J III, Eddy D, Johnston C Jr. Screening for osteopo-
rosis. Ann Int Med 1990;112:516 –28.
3. Caraceni M, Molteni N, Bardella M, et al. Bone and mineral
metabolism in adult celiac disease. Am J Gastroenterol 1988;
83:274 –7.
4. Gonzalez D, Mazure R, Mautalen C, et al. Body composition
and bone mineral density in untreated and treated patients with
celiac disease. Bone 1995;16:231–234.
5. Valdimarsson T, Toss G, Ross I, et al. Bone mineral density in
coeliac disease. Scand J Gastroenterol 1994;29:457– 61.
6. Walters JRF. Bone mineral density in celiac disease. Gut
1994;35:150 –1.
7. Mcfarlane X, Bhalla AK, Reeves D, et al. Osteoporosis in
treated adults coeliac disease. Gut 1995;36:710 – 4.
8. Mazure R, Vázquez H, Gonzalez D, et al. Bone mineral
affectation in asymptomatic adult patients with celiac disease.
Am J Gastroenterol 1994;89:2130 – 4.
9. Bode S, Haasager C, Gudmand-Hoyer E, et al. Body compo-
sition and calcium metabolism in adult treated coeliac disease.
Gut 1991;32:1342–5.
10. Marsh MN. Bone disease and gluten sensitivity. Time to act,
to treat and to prevent. Am J Gastroenterol 1994;89:2105–7.
11. Cook WT, Holmes GKT. Clinical presentation. In: Cook WT,
Holmes GKT, eds. Coeliac disease. Edinburgh: Churchill Liv-
ingstone, 1985:81–105.
12. Mautalen C, Gonzalez D, Mazure R, et al. Effect of treatment
on bone mass, mineral metabolism, and body composition in
untreated celiac disease patients. Am J Gastroenterol 1997;92:
313– 8.
13. Valdimarsson T, Lofman O, Toss G, et al. Reversal of os-
teopenia with diet in adult coeliac disease. Gut 1996;38:322–7.
14. Corazza DR, Di Sario A, Cecchettti L, et al. Bone mass and
metabolism in patients with celiac disease. Gastroenterology
1995;109:122– 8.
15. Fornari MC, Pedreira S, Niveloni S, et al. Pre- and post-
treatment serum levels of cytokines IL-1␤, IL-6 and IL-1
receptor antagonist in celiac disease. Are they related to the
associated osteopenia? Am J Gastroenterol 1998;93:413– 8.
16. Bai JC, Gonzalez D, Mautalen C, et al. Long-term effect of
gluten restriction on bone mineral density of patients with
coeliac disease. Aliment Pharmacol Ther 1997;11:157– 64.
17. Eastell R, Cedel S, Wahner H, et al. Classification of vertebral
fractures. J Bone Miner Res 1991;6:207–15.
18. Peel NFA, Moore DJ, Barrington NA, et al. Risk of vertebral
fracture and relationship to bone mineral density in steroid
treated rheumatoid arthritis. Ann Rheum Dis 1995;54:801– 6.
19. Spector TD, Hall GM, McCloskey EV, et al. Risk of vertebral
fracture in women with rheumatoid arthritis. Br Med J 1993;
306:558.
20. Ralston SH, Urquhart GDK, Brzesky Met al. Prevalence of
vertebral compression fractures due to ankylosing spondylitis.
Br Med J 1990;300:563–5.
21. Marsh MN. Gluten histocompatibility complex, and the small
intestine: A molecular and immunobiologic approach to the
spectrum of gluten sensitivity (“celiac sprue”). Gastroenterol-
ogy 1992;102:230 –54.
22. Thomason R, Coupland CA, Holms GKT, et al. Do coeliacs
suffer more fractures than the general population: A popula-
tion-based survey of the fracture experience of older celiacs.
Gut 1997;41:A72.
23. Marsh MN. Bone disease and gluten sensitivity. Time to act,
to treat and to prevent. Am J Gastroenterol 1994;89:2105–7.
24. Smecuol E, Gonzalez D, Mautalen C, et al. Longitudinal study
on the effect of treatment on body composition and anthro-
pometry of celiac disease patients. Am J Gastroenterol 1997;
92:639 – 43.