STRESS FRACTURES AND TIBIAL BONE WIDTH

A RISK FACTOR

M. GILADI, C. MILGROM, A. SIMKIN, M. STEIN, H. KASHTAN, J. MARGULIES, N. RAND, R. CHISIN,

R. STEINBERG, Z. AHARONSON, R. KEDEM, V. H. FRANKEL

From Ichilov Hospital, Tel-Aviv and Hadassah University Hospital, Jerusalem

A prospective study of 295 infantry recruits has shown that the mediolateral width of the tibia measured
radiographically at each of three different levels in the bone had a statistically significant correlation with the
total incidence of stress fractures as well as with those in the tibia alone or the femur alone. A narrow tibial
width was shown to be a risk factor, but cortical thickness was not found to be significant.

Stress fractures are a problem among military and study at the Osteoporosis Institute of Jerusalem exam-
athletic trainees (Gilbert and Johnson 1966; Belkin med the hypothesis that tibial bone geometry and
1980), who can develop forces of several times their body composition might be risk factors for stress fractures in
weight at the interface between foot and ground during Israeli recruits, and we report here the correlation
running (Mann et al. 1979). These forces, although between tibial bone width and the incidence of such
modified, are transmitted proximally to the tibia and fractures.
beyond. Lanyon et al. (1975) have shown by strain gauge
studies that large compression and tension forces are
developed in the tibia during walking and running. The
magnitude of these forces varies with different surfaces MATERIALS AND METHODS
of the tibia, the phase of the gait cycle, the gait pattern
A group of 295 male infantry recruits were evaluated
and cadence, and which ground-foot interface is studied.
prospectively for possible risk factors for stress fracture.
The repetitive cyclical absorption by bone ofthese forces
All participants gave their informed consent. Each
and those produced by dynamic muscle action is thought
recruit had a pre-training screening which included
to produce stress fractures (Devas 1958).
measurement of weight and height, clinical measure-
We have previously shown that Israeli recruits are
ment of tibial torsion and classification of tibial
similar to athletes in that the most frequent site of their
alignment into genu varum, straight, or genu valgum.
stress fractures is the tibia followed by the femur (Giladi
Tibial bone mineral content was measured 8 cm above
et al. 1985; Milgrom, Giladi et al. 1985). In other armies
each ankle joint by the Cameron photon absorption
different patterns are found. Part of our prospective
technique using a Norland Digital Bone Densitometer
with the beam directed from medial to lateral.
Standard radiographs were taken with a tube-film
M. Giladi, MD, Resident, Department of Internal Medicine
H. Kashtan, MD, Resident, Department of Surgery
distance of 90 cm. For the anteroposterior view the feet
Tel-Aviv Medical Centre, Ichilov Hospital, Tel-Aviv, Israel. were positioned in 15#{176}
of medial rotation; standard
C. Milgrom, MD, Senior Lecturer, Department of Orthopaedic lateral views were used. Measurements were then made,
Surgery
A. Simkin, PhD, Biomechanical Engineer, Department of Orthopae-
using a magnification ruler, of total tibial width and
dic Surgery cortical widths in both anteroposterior and mediolateral
J. Margulies, MD, Resident, Department of Orthopaedic Surgery
R. Chisin, MD, Resident, Department of Nuclear Medicine
planes at three levels, in each bone : at 8 cm above the
R. Steinberg, MS. Director, Osteoporosis Institute ankle joint, at the point of the narrowest mediolateral
Hadassah University Hospital, POB 1200, Em Kerem, Jerusalem,
Israel 91120.
width, and at the point of the narrowest anteroposterior
M. Stein, MD
width (Fig. 1). The six measurements for each bone are
N. Rand, MD laid out and given a number in the heading of Table I.
z. Aharonson, MD
R. Kedem, BSc, Chief Statistician
During their 14-week basic training course the
Medical Corps, Israeli Defense Forces. recruits were reviewed at three-weekly intervals by army
V. H. Frankel, MD, Chairman ofDepartment ofOrthopaedic Surgery doctors in the field and questioned about symptoms
Hospital for Joint Diseases Orthopaedic Institute, New York, USA.
compatible with stress fracture. Those with complaints
Requests for reprints should be sent to Dr C. Milgrom. were examined. The recruits also had free daily access to
© 1987 British Editorial Society of Bone and Joint Surgery medical staff and were encouraged to report symptoms.
030l-620X/87/2008 $2.00
Those with symptoms compatible with stress fracture

326 THE JOURNAL OF BONE AND JOINT SURGERY

 STRESS FRACTURES AND TIBIAL BONE WIDTH 327

V made on the basis of either a positive radiograph or a

positive scintigram, using the criteria of Prather et al.
Anterior (1977). Scintigraphy was considered to be diagnostic of a
stress fracture when a focal area of increased uptake was
found in the absence of other bony pathology (Milgrom,
Chisin et al. 1985).
Results were recorded on a standard evaluation
Antero
Posterior form which had been designed before the study began,
Tibial and all data was processed through the IBM computer
Width
facilities of the Israeli Defense Forces Medical Corps.
Lateral Medial
Statistical results including means, standard deviations
and p values were calculated with the Statistical Package
for the Social Sciences (SPSS) using Student’s t-test or
Yates’ correction of the chi-square test.

Principal
RESULTS
Axis
During the course of basic training 91 of the 295 recruits
Posterior
(31%) were found to have one or more stress fractures
V (Milgrom, Giladi et al. 1985). Of the total of 184 stress
Fig. 1 fractures, 51% were in the tibial diaphysis, 5% in the
tibial plateau, 21% in the femoral diaphysis, 9% in the
Diagram of a cross-section through the left tibia.
supracondylar region and 4% in the femoral condyles. All
of the tibial and femoral diaphyseal fractures were in the
medial cortex. Only 9% of the fractures were in the feet.
were given three days’ rest; if symptoms then persisted Both anteroposterior and mediolateral bone widths
the recruit was sent for orthopaedic investigation in were measured in both tibias of 288 of the 295 recruits at
hospital. each of the three levels (Table I). The correlation
Sites of pain were then recorded in relation to between right and left tibial widths varied from 0.81 to
anatomical landmarks, appropriate radiographs were 0.91 depending
, on the measurement site ; the mean
taken and late phase 99mTcMDP scintigraphy was width for each level on right and left legs was used for
performed in all cases. A diagnosis of stress fracture was calculation. The correlation between each of the six

Table I. The mean mediolateral (ML) and anteroposterior (AP) diameters ofthe tibia ± s.d. measured in millimetres at each ofthree levels in 286
recruits with and without stress fractures

Level measured

Narrowest ML5 Narrowest APt 8 cm above ankle

ML AP ML AP ML AP
Stressfractures 1 2 3 4 5 6

Femora/
Present (n = 36) 22.8 ± 1.6 28.0 ± 2.9 24.8 ± 2.0 25.0 ± 2.3 25.3 ± 2.3 25.3 ± 2.3
Absent (n = 250) 24.5 ± 1.9 28.2 ± 2.6 26.4 ± 2.0 25.9 ± 2.0 26.9 ± 2.2 26.2 ± 2.0
Significance of difference in mean width p < 0.001 p 0.609 p < 0.001 p 0.016 p < 0.001 p 0.012

Tibia!
Present (n = 58) 23.8 ± 2.1 28.1 ± 2.7 25.7 ± 2.1 25.4 ± 2.0 26.1 ± 2.4 25.7 ± 2.2
Absent (n = 228) 24.4 ± 1.9 28.3 ± 2.6 26.3 ± 2.0 25.9 ± 2.0 26.9 ± 2.3 26.2 ± 2.0
Significance of difference in mean width p = 0.036 p = 0.578 p 0.027 p 0.085 p 0.036 p 0.101

All
Present (n = 86) 23.8 ± 2.1 28.2 ± 2.8 25.6 ± 2.2 25.5 ± 2.1 26.2 ± 2.5 25.8 ± 2.2
Absent (n = 200) 24.6 ± 1.8 28.2 ± 2.5 26.4 ± 1.9 25.9 ± 2.0 26.9 ± 2.2 26.9 ± 2.0
Significance of difference in mean width p = 0.001 p = 0.980 p = 0.003 p = 0.088 p = 0.010 p = 0.098

a, mean I 3.7 cm above the ankle joint

t, mean 8.7 cm above the ankle joint

VOL. 69-B, NO. 2, MARCH 1987

328 M. GILADI, C. MILGROM, A. SIMKIN, ET AL.

measurements of mean tibial bone widths and the is proportional to a factor of the fourth power of the
incidence ofstress fractures is given in Table I. The mean radius in the mediolateral plane (Frankel and Burstein
mediolateral width at all three levels (measurements 1 3 , 1970). Therefore wider tibias in this plane should have
and 5) showed a statistically significant difference in greater resistance to bending in this direction. Resis-
recruits with and without any stress fractures. This tance to compression and tension should also be greater
difference was also significant for the smaller groups since a larger cross-sectional area allows these forces to
sustaining either femoral or tibial stress fractures alone. be distributed over a greater unit area (Frankel and
Anterolateral measurements 4 and 6 showed a statistical- Burstein 1970). The fact that all of the stress fractures in
ly significant difference only in relation to femoral stress the tibial and femoral shafts in this series occurred in the
fractures, while measurement 2 was found to have no medial cortex helps explain why the tibial bone width in
statistical significance. a mediolateral direction correlates best with the
No statistical correlation was found between the incidence of such fractures; the distribution of bone in
incidence of stress fractures and height, weight, tibial this plane is one determinant of the resistance of the
torsion, tibial alignment, tibial bone mineral content, or medial cortex to bending, compression and tension.
the radiographic width of the tibial cortex. Use of the Moreover, most of the tibial stress fractures occurred in
ratio of tibial bone width to the height of the recruit the middle third of the bone ; this implicates bending
rather than tibial bone width alone did not increase the forces, since such forces are maximal in this region.
statistical correlation with the incidence of stress A wide tibia is normally associated with wide
fracture, and other ratios related to the height and weight tubular bones in general and this explains why total and
ofthe recruit as well as to tibial width were shown to have femoral stress fracture incidence also correlated with
no statistical significance. tibial bone width. We found, however, more correlation
of tibial bone width with femoral than with tibial stress
fractures. The reason for this is not readily apparent and
requires further study, but the mechanism of femoral
DISCUSSION
stress fractures may well differ from that of the majority
Stress fractures are a problem in sports medicine (Belkin of tibial stress fractures. The number of metatarsal stress
1980), but the true incidence in athletes andjoggers is not fractures in this study was too small to establish any
known. Brudvig, Gudger and Obermeyer (1983) suggest statistical correlation with tibial bone width.
an incidence ofless than 2% in US Army recruits, but we The finding that tibial bone width is related to stress
have reported a 31% incidence of stress fractures in a fracture morbidity may help to explain Protzman and
group of special infantry recruits in the Israeli Army Griffis’ (1977) finding that women sustained many more
(Milgrom, Giladi et al. 1985). One of the possible stress fractures than men during the same training.
explanations for this difference could be the existence of Miller and Purkey (1980) studied paired human tibias
a group with high risk for stress fractures within the and found that women had relatively narrower bones
Israeli Army (Giladi et al. 1985). Brudvig et al. (1983) with lower moments of inertia.
noted sub-populations with a higher risk factor in the US The statistical correlation found between tibial bone
Army; white men had twice the incidence of stress width and the incidence of stress fractures is important,
fractures as black men doing the same training at West since it is the first physical parameter of the bone to be
Point. Protzman and Griffis (1977) showed that women identified as a risk factor. Attempts can now be centred
had ten times as many stress fractures as men on the on determining the biomechanical basis for its signifi-
same course. cance and on identifying risk groups for stress fractures
We found that recruits with stress fractures had which will be clinically useful.
tibias which were narrower mediolaterally at all three Improved correlation might result if tibial widths
measured levels (1, 3 and 5) than recruits without stress were measured along axes other than the simple
fractures. Measurement 5 was taken at a fixed distance anteroposterior and mediolateral ones used in this study.
above the ankle joint rather than at a point of narrowing The principal axis of the tibia (see Fig. 1), that in which
and this may be one reason why it was less statistically the tibial diameter is widest at any given cross-section,
significant than measurements I and 3. These two may be important, but its direction is not constant and
measurements give the mediolateral diameter ofthe tibia varies at different levels of the tibia. Another possible
at the levels of the narrowest mediolateral and axis for measurement is the predominant axis of
anteroposterior tibial widths respectively. occurrence of stress fractures but this axis has not yet
The tibia has varying cross-sectional shapes, but it been identified.
may be idealised as a cylinder whose cross section is an Ifclinically useful risk groups can be identified, they
eccentric ellipse within an ellipse. For an oblate cylinder, will help with the screening of athletes or recruits before
bending strength in the mediolateral plane is related to training. High-risk subjects could be identified and
the area moment of inertia of the cylinder about the y- either given a modified training programme or be
axis of rotation (see Fig. 1). This area moment of inertia excluded entirely from certain courses.

THE JOURNAL OF BONE AND JOINT SURGERY

 STRESS FRACTURES AND TIBIAL BONE WIDTH 329

REFERENCES

Belkin SC. Stress fractures in athletes. Orthop C/in North Am Mann RA, Hagy JL, White V, Liddell D. The initiation of gait. J Bone
1980:1 1 :735-41. Joint Surg [Am] 1979:61-A:232-9.
Brudvig TJ, Gudger TD, Obermeyer L. Stress fractures in 295 trainees : Milgrom C, Chisin R, Giladi M, Stein M, Kashtan H, Margulies J, Atlan
a one-year study of the incidence as related to age, sex, and race. H. Multiple stress fractures : a longitudinal study of a soldier with
Mi/it Med 1983:148:666-7.
13 lesions. C/in Orthop 1985:192:174-9.
Devas MB. Stress fractures of the tibia in athletes or “shin soreness”.
J Bone Joint Surg [Br] 1958 :40-B :227-39. Milgrom C, Giladi M, Stein M, et al. Stress fractures in military
recruits : a prospective study showing an unusually high incidence.
Frankel VH, Burstein AH. Orthopaedic biomechanics: the application of Bone Joint Surg [Br] 1985 :67-B :732-5.
engineering to the muscu/oske/etal si’stem. Philadelphia : Lea and
Febiger, 1970 :40-76. Miller GJ, Purkey WW Jr. The geometric properties of paired human
Giladi M, Ahronson Z, Stein M, Danon YL, Milgrom C. Unusual tibiae. J Biomech 1980:13:1-8.
distribution and onset of stress fractures in soldiers. C/in Orthop
1985:192:142-6. Prather JL, Nusynowitz ML, Snowdy HA, Hughes AD, McCartney
Gilbert RS, Johnson HA. Stress fractures in military recruits: a review WH, Bagg RJ. Scintigraphic findings in stress fractures. J Bone
Joint Surg [Am] 1977:59-A:869-74.
of twelve years’ experience. Mi/it Med 1966:131 :716-21.
Lanyon LE, Hampson WGJ, Goodship AE, Shah JS. Bone deformation Protzman RR, Griffis CG. Stress fractures in men and women
recorded in vivo from strain gauges attached to the human tibial undergoing military training. J Bone Joint Surg [Am] 1977:59-
shaft. Acta Orthop Scand 1975:46:256-68. A:825.

VOL. 69-B. NO. 2. MARCH 1987