JOURNAL OF BONE AND MINERAL RESEARCH

Volume 8, Number 10, 1993

Mary Ann Liebert. Inc., Publishers

Simple Measurement of Femoral Geometry Predicts

Hip Fracture: The Study of Osteoporotic Fractures

KENNETH G. FAULKNER,'.' STEVEN R. CUMMINGS,' DENNIS BLACK,3 LISA PALERMO,'

CLAUS-C. GLUER,' and HARRY K. GENANT'

ABSTRACT
Based on engineering principles, geometric measurements of femoral size should be related to femoral
strength and the risk for hip fracture. To evaluate whether a simple measurement of femoral geometry is as-
sociated with hip fracture risk, we obtained dual x-ray absorptiometry scans of the proximal femur on 8074
white women age 67 or older. During an average of 1.6 years of follow-up, 64 participants suffered hip frac-
tures. In all fracture cases and in a random sample of 134 women who did not subsequently suffer a hip frac-
ture, we measured hip axis length (the distance from greater trochanter to inner pelvic brim), neck width,
and the neck/shaft angle on the scan printout, with the observer blinded to subsequent fracture status of the
participant. Results were analyzed using multiple logistic models, and odds ratios were determined. After ad-
justment for age, each standard deviation decrease in femoral neck bone mineral density increased hip frac-
ture risk 2.7-fold (95% confidence interval 1.7, 4.3), and each standard deviation increase in hip axis length
nearly doubled the risk of hip fracture (odds ratio = 1.8; 95% CI 1.3, 2.5). The relationship between hip
axis length and fracture risk persisted even after adjustment for age, femoral neck density, height, and
weight. A longer hip axis length was associated with an increased risk of both femoral neck (OR = 1.9; 95%
CI 1.3, 3.0) and trochanteric fractures (1.6; 1.0, 2.4). We found no significant association between the neck
width (1.1; 0.8, 1.5) or the neck/shaft angle (1.4; 0.9, 2.2) and risk of hip fracture. In a combined analysis of
the control group with an additional population of younger volunteers, no significant relationship was found
between the hip axis length and age ( r = 0.04, P = 0.60) or femoral neck density ( r = 0.01, P = 0.84) in
225 women from 41 to 92 years of age. We conclude that hip axis length predicts hip fractures independently
of age and bone mineral density in elderly women. If verified by additional studies, this simple measurement
can improve the assessment of hip fracture risk compared to a measurement of femoral neck bone density
alone.

INTRODUCTION physical characteristics that predict hip fracture. Bone

represent the most serious consequence
H IP FRACTURES
of osteoporosis. In the United States, one of every six
white women suffers a hip fracture during her lifetime; up
to 20% of these women die as a result.(I) Among the sur-
vivors of hip fracture, half of the patients are unable to re-
sume normal daily activity for at least a year after the in-
jury, often requiring long-term health care. ( 2 - 4 ) Because of
the magnitude of the problem, it is important to find the
mineral density is one of the primary determinants of bone
strength. From numerous biomechanical studies, the
strength of bone tissue is known to be proportional to the
mass and density of bone mineral it contains.(5-8)Several
prospective studies have demonstrated a relationship be-
tween low bone mass at several skeletal sites and the risk of
subsequent hip fra~ture(~.'O); the relationship is even
stronger for measurements of femoral bone density.'")
Bone density alone, however, is not the sole factor influ-

IDepartment of Radiology, Osteoporosis Research Group, University of California, San Francisco.

'Current address: Providence Center for Osteoporosis Research and Oregon Health Sciences University, Portland, Oregon.
'Department of Epidemiology, Prevention Sciences Group, University of California, San Francisco.

1211
1212
encing whether a fracture occurs. According to basic engi-
neering principles, the strength of an object is a function
of other properties besides the mass and density of the ma-
terial present. Strength depends on ( I ) the mechanical
properties of the materials, (2) accurate knowledge of the
object’s geometry, and (3) the loading conditions, in terms
of magnitude, rate, and direction, of force applied to the
object.‘”) Therefore, measurements of hip geometry
should be related to hip strength independently of femoral
bone mineral density. I f the geometry of the hip can be
shown to be related to fracture risk independently of
known risk factors (such as age and bone density), geomet-
ric measurements might be used together with densitornet-
ric measurements for a better assessment of hip fracture
risk than might be obtained from a density measurement
alone.
For this study, we derived several simple geometric mea-
surements from dual x-ray absorptiometry (DXA) scans of
the hip based on the hypothesis that they are related to hip
fracture risk. To test our hypothesis, we measured the fem-
oral dimensions from the DXA scans of participants in a
large multicenter study of elderly women, the Study of
Osteoporotic Fractures (SOF). Scans were obtained on all
participants, and then the subjects were followed for the
occurrence of hip fractures over the next 2 years. The base-
line femoral dimensions of women who suffered subse-
quent hip fracture and a random sample of controls were
compared to determine if hip geometry is a predictor of
hip fracture that is independent of femoral bone mineral
density and age.
MATERIALS AND METHODS
The Study of Osteoporotic Fractures is a multicenter
study of 9704 women at least 65 years of age at the time of
recruitment from four regions of the United States. ‘ l o )
Black women (because of their low incidence of hip frac-
ture), women with bilateral hip replacement, and women
unable to walk without assistance were excluded from the
study. Baseline examinations of these subjects were per-
formed from September 1986 to October 1988. These ini-
tial examinations did not include DXA scans because the
technology was not yet available. From November 1988 to
December 1990, these women were invited for a second ex-
amination that included DXA scans of the spine and hip.
From the 9483 survivors of the original cohort, 8074 at-
tended the second examination and had a DXA scan of the
proximal femur. In addition to the densitometric measure-
ments, height and weight were also recorded for each sub-
ject at the time of the examination.
The femoral DXA measurements were all performed
using Hologic QDR-1000 scanners (Hologic, Inc.,
Waltham, MA). The QDR- lo00 is a pencil beam scanner,
such that the linear dimensions of the scan were unaffected
by variations in the vertical position of the hip between the
source and detector. The right hip was scanned in all cases,
except in the event of hip replacement or severe degenera-
tive change. In these cases, the opposite hip was scanned.
Area (in cm’), bone mineral content (in grams), and bone
FAULKNER ET AL.
mineral density (BMD, in g/cm’) were determined for the
femoral neck, trochanter, intertrochanteric region, Ward’s
triangle, and total hip. The coefficient of variation for the
BMD measurement at the four SOF centers was 1.2% for
the femoral neck measurement based on measurements
performed on research staff scanned on each DXA scan-
ner.(13)
During the average 1.6 year follow-up period after the
DXA scans, 64 hip fractures were reported. Fractures were
verified from preoperative radiographs and reviewed by a
coordinating center radiologist to classify all hip fractures
as either femoral neck or trochanteric. As soon as possible
after the fracture, patients were interviewed about how the
fracture occurred. I f they were unavailable, this informa-
tion was obtained from the closest available friend or rela-
tive. Based on this information, one fracture case was ex-
cluded, the result of a motor vehicle accident. Virtually all
the hip fractures were the result of falls; only a small frac-
tion ( 5 % ) were reported to be “spontaneous” by the partic-
ipant.
For control subjects, 134 women were chosen at random
from the remaining cohort of nonfracture patients who
had a femoral DXA scan. This number was chosen as rep-
resenting roughly twice the number in the fracture group.
The scans for each patient were reviewed by a trained ob-
server for proper analysis technique. Using a goniometer,
hip axis length, neck width, and the neck/shaft angle were
determined directly from the scan printout as defined in
Fig. 1. Because the measurements were performed from
the scan printout, the linear dimensions reported are ap-
proximately 30% less than the true dimensions. To avoid
errors due to differing printer scales. all printouts were
generated using the same thermal printer supplied by the
DXA manufacturer attached to a manufacturer’s analysis
workstation.
The geometric measurements were conducted by one in-
vestigator who was blinded to the fracture status of the
subjects. Precision error (reproducibility) of the geometric
measurements was determined from a group of 48 volun-
teer women scanned three times in succession on the same
DXA scanner with interim repositioning. The triplicate
scans were printed on the same thermal printer and each
measured as described earlier. The precision error (defined
as the average coefficient of variation of the three geomet-
ric measurements) was less than 1.0% for all the geometric
parameters measured from the scan printout.
Odds ratios calculated from a logistic regression analysis
were used to describe the association between DXA mea-
surements and the occurrence of hip fractures. I f the 95%
confidence interval for the odds ratio did not include 1 .O,
the analyzed variable was deemed to be associated with
subsequent hip fracture risk. Multiple logistic models were
used to adjust for potential confounding factors, such as
age and height, and 95% confidence intervals evaluated as
described earlier. Relationships between continuous vari-
ables were described by Pearson correlation coefficients,
with significant relationships defined as those with P <
0.05. All statistical tests were performed using SAS statisti-
cal software (Cary, NC). Because age is a known predictor
of hip fracture in this cohort,(’O)all odds ratios presented
are adjusted for age unless otherwise noted.
FEMORAL GEOMETRY A N D HIP FRACTURE 1213

T o further investigate the potential influence of age on
the geometric measurements, an additional group of 91
DXA scans were obtained from women a&:ed 41 to 67
(average age 56 + 6.9 years) who are part of an ongoing
clinical drug trial. None of these women had suffered hip
fractures. Geometric measurements were obtained in the
same manner as described, and the results were combined
with those for the SOF control group, who are 67 years of
age and older. The relationship of the geometric parame-
ters with age and with femoral neck bone mineral density
was assessed by a regression analysis in this combined sam-
ple. Significant correlations were defined as those with P
< 0.05.

RESULTS

Of the geometric measurements, only the hip axis length

FIG. 1. Definition of geometric measurements from
femoral DXA scans. A A ‘ , hip axis length, defined as the
length along the femoral neck axis as defined by the DXA
analysis software, from below the lateral aspect of the
greater trochanter, through the femoral neck, to the inner
pelvic brim. Angle A B C , neck/shaft angle, defined as the
angle formed between the femoral neck and the shaft of
the femur. DD’,neck width, defined as the shortest dis-
tance within the femoral neck region of interest, as defined
by the DXA analysis software, perpendicular to the
femoral neck axis.
was significantly different between fracture and nonfrac-
ture groups (Table 1). After age adjustment, a hip axis
length 1 standard deviation longer than average was associ-
ated with almost a twofold increase in the risk for subse-
quent hip fracture (Table 2). In Fig. 2, the distribution of
hip axis length with age is shown for the hip fracture and
control groups used in the analysis. The trochanter area as
determined by the analysis software showed a significant
association with hip fracture risk (odds ratio, OR, per
standard deviation increase 1.4; 95% confidence interval,
CI, 1.0, 2.0); however, the trochanter area and hip axis
length as defined for this study were correlated ( r = 0.60,
P < 0.001). After adjustment for hip axis length, the asso-
ciation between trochanter area and fracture risk vanished
(OR = 1.0; 95% CI 0.7, 1.5).
The bone density of all femoral regions was strongly as-
sociated with the risk of hip fracture. For example, each
standard deviation decrease in age-adjusted femoral neck
density was associated with a 2.7-fold increase in risk
(Table 2). This result is in agreement with previously pub-
lished data on the entire SOF cohort.(11B Figure 3 shows
the distribution of femoral neck density with age for this
study.
In the multiple logistic analysis, hip axis length remained
a strong predictor of hip fracture after adjustment for
both femoral neck density and age (OR = 2.3; 95% C1
1.6, 3.4). Of the 198 women measured in the SOF cohort,

TABLE 1 . MEANSA N D STANDARD DEVLATIONSOF MEASUREMENTS

IN HIP FRACTURE
A N D CONTROL
GROUPS

Mean f standard deviation

Measurement (units) Fracture cases Controls P valuea

~~ ~ ~

Age, years 76.5 f 6.2 73.4 f 5.4 O.OOO4

Femoral neck BMD, g/cm’ 0.556 f 0.094 0.652 f 0.118 O.OOO1
Hip axis length, cm 6.93 f 0.40 6.70 f 0.38 O.OOO1
Neck width, cm 2.0 f 0.16 1.98 f 0.16 0.41
Neckishaft angle, degrees 127 f 5.5 126 f 4.7 0.42
Trochanter area, cm’ 11.8 f 1.6 11.3 f 1.5 0.035
Height at examination, cm 158 f 6.9 160 f 6.4 0.14
Height at age of 25, cm 164 f 6.0 163 f 6.1 0.33
Weight at examination, kg 65.0 f 13.5 66.3 f 11.0 0.50
aProbability that no difference etists between participants with and without hip fractures by a two-
tailed group /-test.
1214 FAULKNER ET AL.

TABLE
2. ASWMnONS BETWEEN FEMORALMEASUREMENTS A N D HIP
RISK (LOGISTIC
FRACTURE REGRESSION
RESULTS)

Age-adjusted odds ratio

Measurement per standard deviation (95% Cf)
Femoral neck BMD 2.7 (1.7, 4.3)a
Hip axis length 1.8 (1.3, 2.5)
Neck width 1.1 (0.8, 1.5)
Neck/shaft angle 1.1 (0.8, 1.5)
Trochanter area 1.4 ( I .o, 2.0)
Height at examination 1 .O (0.8, 1.4)a
Height at age of 25 1.3 (1.0, 1.8)
Weight at examination 1 .O (0.7, 1.4)
Hip axis length, adjusted for
Neck BMD 2.3 (1.6, 3.4)
Height 2.2 (1.5, 3.3)
Weight 1.9 (1.4, 2.8)
Neck BMD, height, weight 2.3 (1.5, 3.6)
aThese odds ratios are per standard deviation decreuse in measurement. Re-
maining odds ratios are based on a standard deviation increase in the given mea-
surement.

85 .
Controls

0 . Hip Fractures

-5 7 5 0

-
c
'

e e
n

n
P e n . . *
n n
n
e -

3 7 n e 0 .

s
0 6 5 .
n

e
e
0
n
e
n
e

e
n

n
e n
n
o
n
e
e
e
e
n o
n
e

n o e e e
n
e
* *
e e n
n
e

I
0 .
e * e e e * . * *
0 m e n
* e o n 0
6 .

5.5 .

5 '
65 70 75 80 85 90 95 100
Age (years)

FIG. 2. Hip axis length as a function of age for the hip fracture and control subjects.

71% of those with femoral neck bone density in the lowest
third and a hip axis length in the upper third suffered a hip
fracture. None of the women in the upper third of femoral
neck density or the lowest third of hip axis length suffered
hip fractures (Fig. 4).
There was a low but statistically significant correlation
between the hip axis length, neck width, and the various
body size parameters (Table 3). However, the association
between hip axis length and age-adjusted risk remained un-
changed after adjustment for height (OR = 2.2; 95% CI
1.5, 3.3) and weight (OR = 1.9; 95% C1 1.4, 2.8). Hip
axis length was associated with both subsequent femoral
FEMORAL GEOMETRY AND HIP FRACl'URE 1215

1.4 i
1.2 '
Controls
. . Hip Fractures

I
iT
5rn
% * * a *
e m
0' 0.8 * *
5 1::
X
0

0.6
a 1 ' 0 I
X
* .
t
. 0
E :
- 7!
E
Y
3 :-
:.
A

0.4

0.2 '

0 .
65 70 75 80 a5 90 95 100

Age (yeam)

FIG. 3. Femoral neck bone mineral density (BMD) as a function of age for the hip fracture and control subjects.

neck fracture (1 .9-fold increase per standard deviation in-

crease in hip axis length) and trochanteric fractures (1.6-
fold increase per standard deviation increase in hip axis
length; Table 4). Neither neck width nor neck/shaft angle
was significantly associated with femoral neck or trochan-
teric fractures.
In the combined analysis of the 91 younger volunteer
women with the sample of 134 participants from SOF con-
trol group, no statistically significant relationship was
found between hip axis length and age from 41 to 92 years
r = 0.04, P = 0.60). Likewise, the hip axis length was
found to be unrelated to femoral neck bone density in the
same population ( r = 0.01, P = 0.84).

DISCUSSION

FIG. 4. Hip fracture incidence in terms of number of hip
fractures per lo00 person-Years according to age-adjusted
tertile of hip axis length (HAL) and femorid neck bone
mineral density (BMD). Fracture incidence was calculated
based on the 8074 participants in the Study of Osteopo-
rotic Fractures, from which the sample for this study was
drawn. F~~ this analysis, the distribution of the hip =is
length measurement in the entire cohort was assumed to be
the same as that in the 134 randomly selected control sub-
jects.
This study has shown that hip axis length, a simple geo-
metric measurement from DXA scans of the proximal
femur, is associated with hip fracture risk independently of
age and bone density. For the participants in this study, a
hip axis length measurement was able significantly 10 en-
hance the ability of femoral DXA measurements to predict
hip fracture. The hip axis length can be easily derived from
DXA obtained on commercially available systems,
with little or no modification to the scan protocol. An
automated determination of the hip axis length as defined
for this study could be incorporated with existing DXA
analysis software. Normative data could be generated
rapidly based on existing normative DXA scans and the
1216 FAULKNER ET AL.

3. CORRELATION
TABLE ( P VALUETHATN o CORRELATION
COEFFICIENTS EXISTS

Age Hip axis length Neck width Neck/shaft angle Neck B M D Height Height at 25

Hip axis length 0.08

(0.27)
Neck width 0.05 0.43
(0.49) (0.001)
Necklshaft angle 0.04 0.16 0.12
(0.57) (0.02) (0.10)
Neck BMD -0.36 0.05 0.08 0.004
(0.001) (0.47) (0.27) (0.95)
Height -0.35 0.42 0.40 0.09 0.36
(0.001) (0.001) (0.001) (0.20) (0.001)
Height at 25 -0.18 0.48 0.43 0.08 0.24 0.85
(0.01) (0.001) (0.001) (0.26) (0.001) (0.001)
Weight -0.23 0.29 0.32 0.11 0.43 0.44 0.41
(0.001) (0.001) (0.001) (0.11) (0.001) (0.001) (0.001)

TABLE 4. ASSOCIATIONS BETWEEN FEMORAL MEASUREMENTS dependent on the loading conditions. In addition, the hip
A N D RISK FOR FEMORAL NECKOR TROCHANTERIC axis length was positively correlated with neck width in our
HIP FRACTURE study. An increased neck width serves lo compensate for
an increased moment arm; as such the role of the hip axis
Age-adjusted odds ratio
length as a surrogate measure for the femoral moment arm
per standard deviation
(95% el) is highly questionable.
A second possibility is that a longer hip axis might cause
Measurement Femoral neck Trochanter the greater trochanter to extend beyond the pelvis to a
larger degree, thus creating a more vulnerable target for
Femoral neck BMD 2.9 (1.6,5.2)a 2.5 (1.4,4.4)a impact. Because the hip axis length measurement includes
Hip axis length 1.9 (1.3,3.0) 1.6(1.0,2.4) the tissues of the hip joint, it may be related to joint de-
Neck width 1.1 (0.8,1.7) 1.1 (0.7,1.6) generation in the hip, which is in turn associated with de-
Neck/shaft angle 1.0 (0.7,1.5)a 1.4 (0.9,2.2) creased mobility and increased fracture risk. I t is also pos-
Trochanter area 1.8 (1.2,2.7) 1.1 (0.7,1.7) sible that a hip axis length measurement is a marker for
Height at examination 1 . 1 (0.7,1.6)a 1.0 (0.7,1.6)” some other geometric characteristic we did not measure.
Height at age of 25 1.3 (0.9,2.0) 1.3 (0.8,2.0) However, adjustments for participant size (height and
Weight at examination 1.1 (0.7,1.7)a 1.1 (0.6,1.7) weight) did not alter our findings.
The precise mechanism by which the hip axis length is
aThese odds ratios are per standard deviation decrease in mea-
surement. Remaining odds ratios are based on a standard devia- associated with fracture risk cannot be determined from
tion increase in the given measurement. this study. However, additional studies based on detailed
measurements of pelvic films from the SOF cohort are cur-
rently From these measurements we hope to
understand better the failure mechanisms that may con-
tribute to the association of hip axis length and hip frac-
hip axis length evaluated together with femoral bone den- ture.
sity for identifying patients at the greatest risk for hip frac- Our study has some limitations. Although the partici-
ture. pants were recruited from population-based lists, these
The physical significance of the hip axis length to hip white volunteers may be somewhat healthier than average;
fracture risk can be interpreted in several ways. The length in particular, the results may not apply to men, other
of the hip axis may approximate the femoral moment arm races, or nursing home residents. Nevertheless, this nested
(i.e., the perpendicular distance from an applied force to case-control study avoids some of the most important pit-
the center of pivot). The longer the moment arm, the less falls of other case-control studies of hip fracture. We had
is the force required to produce a fracture. However, the a complete sample of cases and a random sample of con-
hip axis length as defined in this study, although based on trols, scans were acquired before the hip fractures oc-
engineering principles, was chosen primarily for simplicity curred, and all measurements were blinded to the case or
and was not intended to represent the “true” femoral mo- control status of the subjects.
ment arm. A precise measurement of the femoral moment This new measurement of hip axis length has several ad-
arm during walking or during a fall would not include the vantages. The measurement can be obtained along with a
pelvic bone or acetabular tissues. It would also be highly femoral bone density measurement without additional
FEMORAL GEOMETRY AND HIP FRACTURE
cost, time, or inconvenience. When combined with a femo-
ral bone density reading, a hip axis length measurement
significantly improved the evaluation of hip fracture risk
in the SOF participants. It can be automated for use with
existing DXA analysis software, which would increase the
accuracy and precision of the hip axis length measurement.
An accurate and precise measurement is important because
very small variations in hip axis length (a few millimeters)
are related to significant differences in hip fracture risk.
Whereas bone density is known to be related to age, the
hip axis length shows no correlation with age or with femo-
ral bone mineral density. Therefore, if confirmed by addi-
tional data from the Study of Osteoporotic Fractures and
other prospective fracture studies, this simple measure-
ment can potentially enhance the ability of I)XA scans to
evaluate hip fracture risk in elderly women.
ACKNOWLEDGMENTS
Supported by Public Health Service Grants I-ROI-AG-
05407, I-R01-AR35582, 5-ROlAG05394, I-ROI-AM-
35584, and 1-R01-AR35583. Investigators in the Study of
Osteoporotic Fractures research group include University
of California, San Francisco (coordinating center): S.R.
Cummings (principal investigator), M.C. Nevitt (project
director). D. Black (study statistican), C . Arnaud, W.
Browner, K.G. Faulkner, C. Fox, C. Gliier, S. Harvey,
S.B. Hulley, L. Palermo, D. Seeley, and P. Steiger; Uni-
versity of Maryland: R . Sherwin (principal investigator), J .
Scott (project director), K. Fox (study coordinator), J.
Lewis (clinic supervisor), M. Bahr, S. Trusty, B. Hohman,
L. Emerson, D. Rebar, and E. Oliner; University of Min-
nesota: R. Grimm, Jr. (principal investigator). K. Ensrud
(coinvestigator), C. Bell (project director), D Thomas, K.
Jacobson, S. Jackson, E. Mitson, L. Stocke, and P.
Frank; University of Pittsburgh: J.A. Cauley (principal in-
vestigator), L.H. Kuller (coprincipal investigator), L. Har-
per (project director), M. Nasim (clinic coordinator), C.
Bashada, L. Buck, A. Githens, D. Medve, and S. Rudov-
sky; Kasier Permanente Center for Health Research, Port-
land, Oregon: T.M. Vogt (principal investig.ator). W.M.
Vollmer, H. Glauber, E. Orwoll (coinvestigators), J. Blank
(project director), B. Mastel-Smith (clinic coordinator), R.
Bright, and J. Downing. We also acknowledge Dr.
Michael McClung of Providence Medical Center in Port-
land, Oregon for providing additional scan data for this
study. Presented at the 14th Annual Meeting o f the Arneri-
can Society for Bone and Mineral Research.
REFERENCES
1. Cummings SR. Black D, Rubin SM 1989 Lifetime risks of
hip, Colles' or vertebral fracture, and coronary heart disease
1217
among white postmenopausal women. Arch Intern Med 149:
2445-2448.
2. Cummings SR. Osteoporotic fractures: The magnitude of the
problem. In: Christiansen C , Johansen JS, Riis BJ (eds.)
Osteoporosis 1987.Osteopress, Kdbenhavn K. Denmark, pp.
1193-1 1%.
3. Mossey J M , Mutran E, Knott K , Craik R 1989 Determinants
of recovery 12 months after hip fracture: The importance of
psychosocial factors. Am J Public Health 79:279-286.
4. Magaziner J, Simonsick EM, Kashner TM, Hebel JR, Ken-
zora JE 1989 Survival experience of aged hip fracture pa-
tients. Am J Public Health 79:274-278.
5 . Beck TJ, Christopher BR, Warden KE, Scott WW, Rao GU
1990 Predicting femoral neck strength from bone mineral
data: A ctructural approach. Invest Radiol 25:6-18.
6. Alho A. Hdiseth A. Torstein H 1989 Bone-mass distribution
in the femur. Acta Orthop Scand 60(1):101-104.
7. Esses SI, Lotz JC, Hayes WC 1989 Biomechanical properties
of the proximal femur determined in vitro by single-energy
quantitative computed tomography. J Bone Miner Res 4(5):
71 5722.
8. Lotz JC, Gerhart T N , Hayes W C 1990 Mechanical properties
of trabecular bone from the proximal femur: A quantitative
CT study. J Comput Assist Tomogr l4(l):l07-114.
9. Hui SL. Slemenda CW, Johnston CC 1989 Baseline measure-
ment of bone mass predicts fracture in white women. Ann
Intern Med 111:355-361.
10. Cummings SR, Black DM, Nevitt MC, Browner WS. Cauley
JA, Genant HK. Mascioli SR, Scott JC, Seeley Dc;, Steiger
P, Vogt T. SOF Research Group 1990 Appendicular bone
density and age predict hip fracture in women. J A M A
263(5):665-668.
I I . Cummings SR, Black DM, Nevitt MC. Browner W, Cauley
J . Ensrud K. Genant HK, Palermo L, Scott J, Vogt TM 1993
Bone density at various sites for prediction of hip fractures.
Lancet 341:72-75.
12. Hayes WC, Piazza SJ, Zysset PK 1991 Biomechanics of frac-
ture risk prediction of the hip and spine by quantitative com-
puted tomography. Radiol Clin North Am 29:l-18.
13. Steiger P. Cummings SR, Black DM, Spencer NE, Genant
HK 1992 Age-related decrements in bone mineral density in
women over 65. J Bone Miner Res 7(6):625-632.
14. Gliier CC. Cummings SR, Li 1, Gluer K, Pressman A,
Grampp S. Faulkner KG, Genant HK 1992 Prediction of hip
fracture from measurements made on pelvic x-ray radio-
graphs. J Bone Miner Res 7(Suppl. I):S142.
Address reprint requests to:
Kenneth G. Faulkner, Ph.D.
Providence Center for Osteoporosis Research
5050 N.E. Hoyl, Suite 651
Portland, OR 97213
Received in original form September 21, 1992; in revised form
March 16, 1993; accepted March 20, 1993.