Osteoporos Int (2006) 17: 4145
DOI 10.1007/s00198-005-1887-3
O R I GI N A L A R T IC L E
Rib fractures predict incident limb fractures: results from the European
prospective osteoporosis study
A.A. Ismail A.J. Silman J. Reeve S. Kaptoge
T.W. ONeill
Received: 28 October 2004 / Accepted: 1 March 2005 / Published online: 1 June 2005
International Osteoporosis Foundation and National Osteoporosis Foundation 2005
Abstract Population studies suggest that rib fractures are
associated with a reduction in bone mass. While much is
known about the predictive risk of hip, spine and distal
forearm fracture on the risk of future fracture, little is
known about the impact of rib fracture. The aim of this
study was to determine whether a recalled history of rib
fracture was associated with an increased risk of future
limb fracture. Men and women aged 50 years and over
were recruited from population registers in 31 European
centres for participation in a screening survey of osteo-
porosis (European Prospective Osteoporosis Study).
Subjects were invited to complete an interviewer-
administered questionnaire that included questions
about previous fractures including rib fracture, the age
of their rst fracture and also the level of trauma. Lat-
eral spine radiographs were performed and the presence
of vertebral deformity was determined morphometri-
cally. Following the baseline survey, subjects were
followed prospectively by annual postal questionnaire to
determine the occurrence of clinical fractures. The
subjects included 6,344 men, with a mean age of
64.2 years, and 6,788 women, with a mean age of
63.6 years, who were followed for a median of 3 years
This study was carried out in conjunction with the EPOS Study
Group (see Appendix).
A.A. Ismail (&)
Department of Rheumatology,
Stepping Hill Hospital,
Stockport, UK
E-mail: abbas.ismail@tiscali.co.uk
Tel.: +44-161-4194798
Fax: +44-161-4195548
A.J. Silman T.W. ONeill
ARC Epidemiology Unit,
University of Manchester,
Manchester, UK
J. Reeve S. Kaptoge
University Department of Medicine,
Strangeways Research Laboratory,
Worts Causeway, Cambridge, UK
(range 0.45.9 years), of whom 135 men (2.3%) and 101
women (1.6%) reported a previous low trauma rib
fracture. In total, 138 men and 391 women sustained a
limb fracture during follow-up. In women, after age
adjustment, those with a recalled history of low trauma
rib fracture had an increased risk of sustaining any
limb fracture [relative hazard (RH) =2.3; 95% CI 1.3,
4.0]. When stratied by fracture type the predictive risk
was more marked for hip (RH=7.7; 95% CI 2.3, 25.9)
and humerus fracture (RH=4.5; 95% CI 1.4, 14.6) than
other sites (RH=1.6; 95% CI 0.6, 4.3). Additional
adjustment for prevalent vertebral deformity and pre-
vious (non-rib) low trauma fractures at other sites
slightly reduced the strength of the association between
rib fracture and subsequent limb fracture. In men, after
age adjustment, there was a small though non-signicant
association between recalled history of rib fracture and
future limb fracture. Our data highlight the importance
of rib fracture as a marker of bone fragility in women.
Keywords Epidemiology Incident limb fracture
Prospective study Osteoporosis Rib fracture
Introduction
Rib fractures are common in middle-aged and elderly
men and women with an estimated incidence of 315 per
100,000 person years at ages 60 years and over [1]. The
incidence increases with age and is greater in women
than men [2]. As with vertebral fractures not all rib
fractures come to clinical attention and the true inci-
dence may be higher. Population-based studies suggest
that rib fractures are linked with reduced bone mass [3,
4] and thus those who have sustained a low trauma rib
fracture might be expected to be at increased risk of
further fragility fracture. However, while much is known
about the predictive risk of hip, spine and wrist frac-
ture on the risk of future fracture [5, 6], we know
little about the impact of rib fracture. Such data are
42
importantevidence that rib fractures are linked with
an increased risk of future fracture provides an impor-
tant rationale for considering aected individuals for
assessment and therapy to prevent bone loss and reduce
the risk of further fractures. We used data from the
European Prospective Osteoporosis Study (EPOS) to
determine the relationship between a recalled history of
rib fracture and subsequent risk of limb fracture.
Materials and methods
Subjects
The subjects were recruited for participation in a base-
line screening survey of vertebral osteoporosis, the
European Vertebral Osteoporosis Study (EVOS). The
detailed methods have been described elsewhere [7]. In
brief, men and women aged 50 years and over were re-
cruited from population registers in 36 European cen-
tres. Stratied sampling was used with the aim of
recruiting equal numbers of men and women in each of
six 5-year age bands: 5054 years, 5559 years, 60
64 years, 6569 years, 7074 years and 75 years and
over. Subjects were interviewed using a structured
interview, which included questions about previous
fragility fractures. Subjects were asked, Have you ever
suered from a broken bone (fractures). If yes, Analysis was undertaken looking at any limb fracture
subjects were then asked about the site of their previous
fracture (vertebral, hip, rib, forearm or other), the age of
their rst fracture and the level of trauma (spontaneous,
minor or major trauma). Lateral spinal radiographs
were performed. The radiographs were evaluated mor-
phometrically by one of three observers and the presence
of vertebral deformity determined using the McCloskey-
Kanis method [8]. It was therefore possible to identify
subjects at the baseline survey with a history of rib
fracture, history of other fractures and prevalent radio-
graphic vertebral deformity.
Follow-up
The subjects recruited in 29 of the EVOS centres were
followed prospectively by annual postal questionnaire,
and in a further three centres by telephone or personal
interview (European Prospective Osteoporosis Study-
EPOS). Because of low follow-up rate (in part a conse-
quence of large scale population mobility) the data from
one center (Zagreb, Croatia) were excluded from this
analysis. Subjects were asked to record details of any
fractures sustained in the intervening period, including
marking on a body mannequin (included with the postal
questionnaire) the position or site of their fractures.
Fractures reported were conrmed at each of the par-
ticipating centres by review of radiographs, medical re-
cord or subject interview. In addition, the investigator
conrmed the site of the reported fracture. From these
sources, contemporary data to conrm or refute the
occurrence and site of fracture were not available in 9%
of cases. In these cases the site of fracture was deter-
mined from the area marked by the subject on the
mannequin [9]. The data coordination for the study was
in Manchester (UK) and Lubeck (Germany).
Analysis
Subjects contributed follow-up time (person years) from
the date of the baseline survey until limb fracture, death
or the end of the study. A history of rib fracture was
dened as a low trauma fracture of the rib occurring at
any age (minor trauma or less). A history of previous
low trauma fracture (excluding rib fracture) was dened
as a reported fracture occurring after age 50 years
(minor trauma or less) at the vertebrae, hip, forearm or
elsewhere.
Incident limb fractures were classied using the 9th
edition of the International Classication of Diseases
[14] into the following categories: hip (neck of femur),
ICD code 820; distal forearm, ICD codes 813; humerus,
ICD code 812; other limb, ICD codes 814817, 821
826. Cox regression analysis was used to assess whether
a recalled history of rib fracture was associated with an
increased risk of future limb fracture, with time to the
rst limb fracture event as the outcome of interest.
and also hip, humerus, distal forearm and other frac-
tures. Due to small numbers it was not possible to
undertake the analysis separately for other individual
fracture sites. In subjects who sustained more than one
fracture of the same type, the time to the rst fracture
event was used in the analysis. Adjustments were made
for age. Because a history of previous fracture apart
from rib fracture may also inuence future risk of
fracture, additional adjustments were made for previous
low trauma fractures (hip, wrist or other) over the age of
50 years (yes/no) and baseline radiographic vertebral
deformity (yes/no). All analyses were undertaken sepa-
rately in men and women. Analyses were performed
using the statistical package STATA [10].
Results
Subjects
The subjects included 6,344 men, with a mean age of
64.2 years (SD=8.4), and 6,788 women, with a mean
age of 63.6 years (SD=8.4), who were followed for a
median of 3 years (range 0.45.9 years), a total of 40,348
person years of follow-up. The baseline characteristics of
these subjects are shown in Table 1. A previous low
trauma rib fracture was reported by 135 (2.3%) of men
and 101 (1.6%) of women. In contrast, there was a
higher proportion of women who reported a previous
low trauma non-rib fragility fracture (12.9 vs. 4.3%). As
previously reported [13] the prevalence of vertebral
 43

Table 1 Baseline characteristics of study subjects fracture and 1.1 per 1,000 pys in those without. The
Men Women incidence of humerus fracture was 10.2 per 1,000 pys in
(n=6,344) (n=6,788) those with a history of rib fracture and 1.9 per 1,000 pys
in those without.
Mean age (SD) 64.2 (8.4) 63.6 (8.4) In men, after age adjustment, there was a small
Prevalence of vertebral 743 (11.7%) 799 (11.8%) though not signicant association between recalled his-
deformity
History of low trauma 135 (2.3%) 101 (1.6%) tory of rib fracture and future limb fracture: age-ad-
rib fracture (%) justed RH=2.0 (95% CI 0.94.5). Additional
History of low trauma 275 (4.3%) 873 (12.9%) adjustment for BMI and centre did not alter the results
fractures (>50 years)* (%) in both men and women (results not presented).
*Excluding rib fracture

deformity was similar in men and women (11.7 vs.
11.8%).
Limb fractures at follow-up
In total, 138 men and 391 women sustained a limb
fracture during the follow-up period (Table 2). Of these
fractures the most frequent individual fracture site was
the distal forearm. Ten subjects sustained more than one
type of limb fracture.
Rib fracture and incident limb fractures
In women the incidence of any limb fracture was 47.4
per 1,000 patient years (pys) in those with a history of
rib fracture and 18.9 per 1,000 pys in those without. For
men the corresponding gures were 14.5 per 1,000 pys
and 7.2 per 1,000 pys. Table 3 presents the adjusted
relative hazards (RH) for the association between
baseline rib fracture and future limb fracture in men and
women. In women, after age adjustment, those with a
history of low trauma rib fracture had a modestly in-
creased risk of sustaining any limb fracture: age-ad-
justed relative hazard (RH)=2.3 (95% CI 1.34.0).
When stratied by fracture type the risk was more
marked for the hip (RH=7.7; 95% CI 2.3, 25.9) and
humerus fracture (RH=4.5; 95% CI 1.4, 14.6). Addi-
tional adjustment for prevalent vertebral deformity and
previous low trauma fractures slightly reduced the
strength of the association between rib fracture and
subsequent limb fracture. The incidence of hip fracture
was 10.2 per 1,000 pys in those with a history of rib
Table 2 Number (%) of subjects with incident limb fracture by
gender
Type of incident Men Women
limb fracture
Hip 19 (14%) 28 (7%)
Distal forearm 32 (23%) 150 (38%)
Humerus 15 (11%) 43 (11%)
Other limb 74 (53%) 178 (45%)
Any limb fracture* 138 (100%) 391 (100%)
*Numbers in cells do not add up because ten subjects incurred
more than one type of limb fracture
Discussion
In this study, women with a recalled history of low
trauma rib fractures had a signicantly increased risk of
future limb fracture, particularly hip and humerus
fracture. Our study had several advantagesit was
population based and included both men and women.
There are, however, several limitations that need to be
considered when interpreting the results. Classication
of rib fracture at the baseline survey was based on self
report, which is subject to errors of recall both of under-
and over-reporting. Under-reporting may have occurred
if aected individuals did not come to medical attention,
or alternatively if they did and the diagnosis was missed.
Over-reporting may have occurred if individuals or their
medical attendants ascribed chest pain, usually follow-
ing an episode of trauma or coughing episode, as due to
rib fracture without conrmation by chest radiograph.
In a previous study in elderly women 23% of self-re-
ported rib fractures were false positives [11]. In our study
there were more self-reported fractures in men compared
to women (2.3 vs. 1.6%). A possible explanation is that
men, due to sports or occupational hazard, may have
had more frequent chest trauma causing rib fracture
than women. Alternatively chest wall pain in men may
have been more likely to be attributed to rib fracture
when actually no fracture had occurred. We repeated the
analysis excluding those subjects with rib fractures that
occurred under the age of 40 years and the results were
broadly similar though condence intervals were wide.
Errors may also have occurred in the classication of
incident fractures. To reduce the risk of over-reporting,
fractures were where possible conrmed by either review
of the radiograph or contemporary medical records or
subject interview. In a small proportion of cases (9%) it
was not possible to conrm fracture by any of these
methods. It is possible therefore that subjects who were
more likely to over-report rib fractures at baseline may
also have been more likely to over-report incident limb
fractures during follow-up (systematic bias). Restricting
the analysis, however, to those individuals in whom
incident limb fractures were conrmed (by radiograph,
medical records or subject interview) did not aect the
results (data not shown). Given the study design it was
not possible to assess the degree of under-reporting.
However, in a separate study at three of the participat-
ing centers, amongst 174 subjects with a known history
44
Table 3 Relationship between history of rib fracture and incident limb fracture in men and women.RH relative hazard; CI condence
interval
Men (n=6,344)
Age adjusted
n RH (95% CI)
Any limb fracture* 138 2.0 (0.94.5)
Hip 19


Humerus 15


Colles 32 1.4 (0.29.9)
Other 74 2.4 (0.96.5)
*Numbers do not add up because ten subjects incurred more than one type of limb fracture.
Too few fracture events in men. Multi-
variate: adjusted for age, baseline vertebral deformity, history of low trauma fracture (excluding rib fracture)
of previous fracture, only 12 (7%) did not recall the
event, and only 3% of subjects did not recall a hip or
distal forearm fracture [9]. The eect of any under-
reporting would again tend to reduce the chance of
nding any signicant association between rib fracture
and future limb fracture. In our study we did not include
fractures that occurred following rib fracture and prior
to the baseline survey. The eect of this would be to
underestimate the strength of the associations. Finally
our results were derived from a predominantly Cauca-
sian population in Europe, and the data should be
extrapolated beyond this population with caution.
It is well established that hip, wrist and vertebral
fractures are associated with an increased risk of future
limb fracture [6, 12, 13, 14]. Studies have also suggested
an increased risk of fracture among those with a history
of any previous fracture [5, 15]. To our knowledge,
however, there are no data concerning the relationship
between rib fracture and future limb fracture. Our re-
sults suggest that women with rib fracture are at in-
creased risk of future limb fracture, particularly of the
hip and humerus though the condence limits are wide.
The strength of the association between rib and future
limb fracture was broadly similar in men though the
condence intervals around the eect estimates included
unity. This almost certainly reects reduced statistical
power because of the smaller number of incident limb
fractures.
What is the explanation for the association between
prevalent rib fracture and future limb fractures? At
least part of the explanation is that both fractures are
associated with low bone mass [3, 4]. In our study bone
mass measurements were undertaken in a sub sample of
the study cohort at baseline; however, because of the
relatively small numbers of individuals with complete
data concerning incident limb fractures and bone den-
sity, we were unable to determine whether adjustment
for bone mass inuenced or explained our ndings. In
studies of fractures at other skeletal sites, adjustment
for bone mass did not signicantly inuence the risk of
future fracture [16, 17]. It is possible that the occur-
rence of a previous rib fracture may also be a marker of
impaired bone quality, or increased bone turnover,
which inuences fracture risk. Alternatively, the risk
Women (n=6,788)
Multivariate Age adjusted Multivariate
RH (95% CI) n RH (95% CI) RH (95% CI)
1.9 (0.84.2) 391 2.3 (1.34.0) 2.0 (1.13.5)
28 7.7 (2.325.9) 5.5 (1.618.8)
43 4.5 (1.414.6) 4.0 (1.213.2)
1.3 (0.29.3) 150 1.7 (0.64.5) 1.6 (0.64.3)
2.3 (0.86.2) 178 1.6 (0.64.3) 1.4 (0.53.7)
may be explained by non-skeletal factors such as an
increased tendency to fall among those with prevalent
rib fracture. Further studies are required in order to
tease out the relative contribution of bone mass, qual-
ity, bone turnover and falls in explaining the associa-
tion.
What are the implications of our ndings? The
treatment for rib fracture is largely symptomatic and
supportive and among individuals with low trauma rib
fracture, radiographs are sometimes not performed. Our
data highlight the importance of rib fracture as a marker
of bone fragility in women. Women presenting with
musculoskeletal chest pain following relatively minor
trauma or in whom rib fracture is suspected should be
referred for chest radiography to conrm whether or not
a rib fracture has occurred. Furthermore, women with
low trauma rib fracture should be considered for bone
mineral densitometry to assess their underlying osteo-
porosis risk and if osteoporosis is conrmed given
appropriate treatment.
In conclusion this study shows that women with low
trauma rib fracture are at signicantly increased risk of
future limb fracture, particularly hip and humerus
fractures. The detection of rib fracture in women may
help identify those at risk of future limb fracture.
Acknowledgements The study was nancially supported by a
European Union Concerted Action Grant under Biomed-1
(BMH1CT920182), and also EU grants C1PDCT925102,
ERBC1PDCT 930105 and 940229. The central coordination was
also supported by the UK Arthritis Research Campaign, the
Medical Research Council (G9321536) and the European Foun-
dation for Osteoporosis and Bone Disease. The EUs PECO pro-
gram linked to BIOMED 1 funded in part the participation of the
Budapest, Warsaw, Prague, Piestany, Szczecin and Moscow cen-
ters. Data collection from Croatia was supported by a grant from
the Wellcome Trust. The central X-ray evaluation was generously
sponsored by the Bundesministerium fuer Forschung and Tech-
nologie, Germany. Individual centers acknowledge the receipt of
locally acquired support for their data collection. Dr. Abbas Ismail
was a Wellcome Trust Clinical Research Fellow. We would like to
thank the following individuals: Rita Smith, Aberdeen, UK; Uday
Bhonsle, Anna Martin, Judith Walton, Bridget Wardley-Smith,
Cambridge and Harrow, UK; Mrs. Joanna Parsons, Truro, UK; J.
Bernardino Diaz Lopez, Ana Rodriguez Rebollar, Oviedo, Spain.
We would like also to thank the individuals who took part in the
study and the many individuals who helped access our population
samples.

Appendix
Members of the EPOS Study Group: Project Coordi-
nators: J. Reeve and A.J. Silman, Cambridge and
Manchester, U.K. Austria (Graz) K. Weber; Belgium
(Leuven) J. Dequeker; Croatia (Zagreb) I. Jajic; Czech
Republic (Prague) J. Stepan; Slovakia (Piestany) P.
Masaryk; France (Montceau-Les-Mines) P.D. Delmas,
F. Marchand; Germany (Berlin, Steglitz) D. Felsenberg,
(Berlin, Potsdam) D. Banzer, (Berlin, Charite) W. Rei-
singer, (Bochum) J. Janott, (Erfurt) G. Kragl, (Heidel-
berg) C. Scheidt-Nave, (Jena) B. Felsch, (Lubeck) H.
Raspe, C. Matthis; Greece (Athens) G. Lyritis; Hungary
(Budapest) G. Poor; Italy (Siena) C. Gennari; The
Netherlands (Rotterdam) H.A.P. Pols; Norway (Oslo)
J.A. Falch; Poland (Szczecin) T. Miazgowski; (Warsaw)
K. Hoszowski, R. Lorenc; Portugal (Azores) J. Bruges
Armas, (Oporto) A. Lopez Vaz; Russia (Moscow) L.I.
Benevolenskaya; Spain (Madrid) A. Rapado, (Oviedo)
J.B. Cannata, M. Naves; Sweden (Malmo) O. Johnell;
Turkey (Istanbul) G. Dilsen; United Kingdom (Aber-
deen) D.M. Reid, (Bath) A.K. Bhalla, (Cambridge)
C.Todd, (Harrow) J. Reeve, (Sheeld) J.A. Kanis,
(Southhampton) C. Cooper, (Truro) A..D.Woolf.
References
1. Jones G, Nguyen T, Sambrook PN, Kelly PJ, Gilbert C, Eis-
man JA (1994) Symptomatic fracture incidence in elderly men
and women: the Dubbo Osteoporosis Epidemiology Study
(DOES). Osteoporos Int 4:277282
2. Baron JA, Barrett JA, Karagas MR (1996) The epidemiology
of peripheral fractures. Bone 18:209213
3. Davie MWJ (1996) Fractures at specic sites indicate low bone
mineral density at lumbar spine and femoral neck in women. J
Orthop Rheumatol 9:4145
4. Honkanen R, Kroger H, Tuppurainen M, Alhava E, Saariko-
ski S (1995) Fractures and low axial bone density in peri-
menopausal women. J Clin Epidemiol 48:881888
45
5. Klotzbuecher CM, Ross PD, Landsman PB, Abbott TA, Ber-
ger M (2000) Patients with prior fractures have an increased
risk of future fractures: a summary of the literature and sta-
tistical analysis. J Bone Miner Res 15:721727
6. Ismail AA, Cockerill W, Cooper C, Finn JD, Abendroth K,
Parisi G, et al (2001) Prevelent vertebral deformity predicts
incident hip though not distal forearm fracture: results from the
European Prospective Osteoporosis Study. Osteoporos Int
12:8590
7. ONeill TW, Felsenberg D, Varlow J, Cooper C, Kanis JA,
Silman AJ (1996) The prevalence of vertebral deformity in
European men and women: The European Vertebral Osteo-
porosis Study. J Bone Miner Res 11:10101017
8. McCloskey EV, Spector TD, Eyres KS, Fern ED, ORourke N,
Vasikaran S, et al (1993) The assessment of vertebral deformity:
A method for use in population studies and clinical trials. Os-
teoporos Int 3:138147
9. Ismail AA, ONeill TW, Cockerill W, Finn JD, Cannata JB,
et al (2000) Validity of self report of fractures :results from a
prospective study in men and women across Europe. Osteo-
poros Int 11:248254
10. Stata Corporation. Stata Reference Manual: Release 3.1, 6th
edn. College Station, TX
11. Nevitt MC, Cummings SR, Browner WS, Seeley DG, Cauley
JA et al (1992) The accuracy of self report of fractures in elderly
women: evidence from a prospective study. Am J Epidemiology
135:490499
12. Melton III LJ, Atkinson EJ, Cooper C, OFallon WM, Riggs
BL (1999) Vertebral fractures predict subsequent fractures.
Osteoporos Int 10:214221
13. Owen RA, Melton III LJ, Ilstrup DM, Johnson KA, Riggs BL
(1982) Colles fracture and subsequent hip fracture risk. Clin
Orth Rel Res 171:3743
14. Cuddihy MT, Gabriel SE, Crowson CS, OFallon WM, Melton
III LJ, et al (1999) Forearm fractures as predictors of sub-
sequent osteoporotic fractures. Osteoporos Int 9:469475
15. Gunnes M, Mellstrom D, Johnell O (1998) How well can a
previous fracture indicate a new fracture? A questionnaire
study of 29,802 postmenopausal women. Acta Orthop Scand
69:508512
16. Burger H, Daele van PLA, Algra D, Hofman A, Grobbee DE,
Schutte HE, et al (1994) Vertebral deformities as predictors of
non-vertebral vertebral fractures. BMJ 309:991992
17. Black DM, Arden NK, Palermo L, Pearson J, Cummings SR,
for the Study of Osteoporotic Fractures Research Group (1999)
Prevalent vertebral deformities predict hip fractures and new
vertebral deformities but not wrist fractures. J Bone Miner Res
14:821828