David W. Bates, Dennis M. Black, Steven R. Cummings. Clinical Use of Bone DensitometryClinical Applications. JAMA. 2002;288(15):1898–1900. doi:10.1001/jama.288.15.1898
Author Affiliations: Division of General Internal Medicine and Primary Care, Department of Medicine, Brigham and Women's Hospital, and the Center for Applied Medical Information Systems Research, Partners Healthcare Systems, and Harvard Medical School, Boston, Mass (Dr Bates); and the UCSF Coordinating Center, Departments of Medicine and Epidemiology and Biostatistics, University of California, San Francisco (Drs Black and Cummings).
Scientific Review and Clinical Applications Section
Editor: Wendy Levinson, MD, Contributing Editor. We encourage authors
to submit papers to "Scientific Review and Clinical Applications." Please
contact Wendy Levinson, MD, Contributing Editor, JAMA; phone: 312-464-5204;
fax: 312-464-5824; e-mail: firstname.lastname@example.org.
Osteoporosis represents a difficult problem for physicians because,
although many diagnostic tests are available, interpreting their results is
not straightforward. As a result, many patients, even those with clear indications
such as long-term steroid therapy or vertebral fractures on radiography, do
not get screened or treated. Current evidence-based guidelines recommend screening
for all white women older than 65 years and not already receiving an osteoporosis
treatment and for many nonwhite women. For postmenopausal women who are younger
than 65 years and have strong risk factors for osteoporosis, screening may
also be beneficial. The optimal testing strategy depends on what is available
locally. The best role for follow-up testing is still being defined, and interpretation
of such testing is tricky. Reports of results can be hard to understand; a
randomized controlled trial of clearer reports increased testing and decreased
confusion about the meaning of test results. Densitometry might be more effectively
used in practice if strategies such as having patients fill out a short questionnaire
to assess for risk factors or creating a nurse-based system were used to identify
patients. Clinicians need better approaches for identifying patients most
likely to benefit from screening, systems that facilitate their application,
and test results that are easy to interpret.
Osteoporosis is complex. It is clinically silent until fractures occur,
making it easy to ignore. There are now several relatively complicated guidelines
regarding screening.1- 3 Furthermore,
results of the many types of bone densitometry can be difficult to translate
into risk of fracture and potential benefit of treatment. There are also several
treatment choices whose magnitude and profile of protection against fractures
differ; some (in particular, raloxifene and hormone replacement therapy) also
have important effects on other systems besides the skeleton. However, the
benefits of treating patients with osteoporosis are substantial, and identifying
these patients need not be difficult.
In the simplest terms, the goal of screening is to find patients who
have osteoporosis and offer them effective treatment to reduce their risk
of fractures, which means making measurement of bone mass a routine part of
preventive care for appropriate patients.
A 68-year-old Hispanic woman weighing 49.5 kg asks whether she should
have a bone mineral density (BMD) screening.
The National Osteoporosis Foundation and the US Preventive Services
Task Force recommend that all white women aged 65 years or older and not already
receiving an effective treatment for osteoporosis be offered a screening test.1,2 Algorithms developed for deciding which
postmenopausal white women to refer for densitometry (Figure 1)1 triage more than 80% of
them to densitometry.
Although there is no consensus about which patients to assess who are
younger than 65 years, it is reasonable to recommend densitometry to postmenopausal
women who have strong and well-established risk factors for fracture
or other uncommon conditions known to cause osteoporosis.
Maternal hip fracture after age 50 years
Body weight less than 56.3 kg
History of fracture after age 50 years
Needs to use arms to rise from a chair
Receiving or about to start long-term (>3 months) oral corticosteroid
*See Black et al.12
Nonwhite women and men have a lower prevalence of osteoporosis and lower
risk of fracture, and the evidence for making recommendations about densitometry
is less solid. Nevertheless, it would be reasonable to consider testing any
patient older than 65 years who requests a measurement of bone density and
also to recommend a densitometry test to individuals who are older than 65
years and have a history of a fracture as an adult, those who are receiving
or about to start long-term (≥3 months) oral corticosteroid treatment,
and those who have other strong risk factors for fracture (parental history
of hip fracture, current smoking, body weight <56.3 kg, or serious long-term
illness known to substantially increase fracture risk). In men and nonwhite
women younger than 65 years, testing those who have had fractures as adults
and those who are already receiving or about to start long-term oral corticosteroid
therapy is also probably worthwhile.
A 49-year-old sedentary white woman whose menses have become irregular
but who has no risk factors for fractures requests a measurement of bone density.
Patients commonly request a measurement of bone density even if they
are not at high risk of osteoporosis. In general, tests are useful only if
they lead to changes in treatment or behavior. Women whose bone density is
reported as low are more likely to make changes in behaviors that might decrease
fracture risk.4 On the other hand, some become
unnecessarily anxious about a low result. It is often helpful to tell women
that osteopenia is not a disease and show them an estimate of their risk of
fracture. Women who have osteoporosis can be reassured that there are effective
treatments to reduce their risk.
You suspect osteoporosis in a 77-year-old white woman and want to screen—but
which test to choose?
Hip BMD is generally considered the best test for osteoporosis screening,
especially in patients 65 years or older,5 although
all the tests help assess a patient's risk of fracture. The optimal testing
approach will depend on what modalities are available locally. If a patient
has osteoporosis according to forearm, heel, or finger densitometry, it may
be worthwhile to confirm this diagnosis with hip densitometry before starting
long-term treatment with drugs. Drug therapy reduces the risk of fractures
in women with osteoporosis or low BMD at dual-energy x-ray absorptiometry
at the hip or spine; the value of treatments in women with low BMD at other
sites is less certain. For this reason, patient 3 should have a hip BMD. Generally,
this measurement also includes a measurement of spine BMD at no extra cost.
If the T score at the spine is higher, it should usually be ignored because
spine BMD is often artifactually increased in elderly patients by spinal degenerative
A 55-year-old white woman with no risk factors for osteoporosis has
a T score of −1.8 on an ultrasound heel measurement that was done at
a local health fair.
Bone density of the hip or spine cannot be reliably predicted from measurements
made at peripheral sites. There is substantial variability in such results,
and the best approach to using peripheral densitometry tests has not been
established or adequately studied. If the patient's BMD as measured by a peripheral
device is below a T score of −1.0, the patient should generally have
a follow-up test using hip or spine BMD, if it is available. Even if the T
score is less than −2.5 on the peripheral test, bone density of the
hip or spine may be quite different, and densitometry should generally be
recommended before the decision to start drug therapy is made. Women with
values above a T score of −1.0 on a peripheral test are unlikely to
have osteoporosis (T score ≥−2.5) at the hip or spine, and they probably
do not need additional testing or pharmacological treatment. For technical
reasons, T scores tend to be higher for ultrasound than other types of densitometry,
so it may be worthwhile obtaining hip and spine densitometry results on women
who, as in this case, have a T score below 0 on ultrasound.
A 66-year-old woman has been receiving alendronate for a year for documented
osteoporosis, and you repeat a densitometry, which shows a decrease of 3%
from her baseline measurement.
Follow-up measurements of bone density during treatment can reassure
physicians and patients that treatment is working. Many believe that follow-up
tests improve adherence to treatment, but little evidence supports this belief.
Physicians sometimes use the results of densitometry to decide whether the
patient has responded to treatment. In this case, the bone density seems to
have decreased, and the physician might conclude that the patient has not
responded to treatment. However, it is difficult to interpret a decrease such
as this5; the patient may indeed be responding
to treatment, since the apparent decrease might well have been even greater
without the medication. Sometimes, a decrease indicates that the patient has
not been receiving treatment or, in the case of bisphosphonates, may be taking
it with food or medications that interfere with absorption of the drug. Therefore,
it is important to make certain that the patient is adhering to treatment
and taking her bisphosphonate properly, to ensure that the test was done on
the same machine, since changes in brands or models can cause artifactual
differences in bone density, and to realize that densitometry cannot reliably
detect small (<5%) changes in bone density in individual patients. Even
large changes between 2 tests are often due to variability in the test rather
than a real change in the patient's bone density; when the test is repeated,
these changes are often smaller and the measurements more similar to the first
test result ("regression to the mean"). Therefore, if a patient is adhering
to her treatment, it is reasonable to simply continue it.
A 58-year-old woman weighing 51.8 kg but who has no other risk factors
for osteoporosis has a femoral neck bone density T score of −0.8. She
asks when she needs to have another measurement.
Few data are available regarding how frequently osteoporosis patients
should be followed up; it depends on the expected change over time and the
precision of the test. Medicare and some insurance plans pay for follow-up
measurement of bone density every 2 years, but spine densitometry tests should
be repeated earlier (within 3 to 6 months) after patients start oral corticosteroid
treatment because they can lose bone rapidly. In patients who are not taking
corticosteroids, do not lose weight, do not have other severe illnesses, are
not receiving treatment, and reached the end of menopause at least 3 years
earlier, the average rate of bone loss is less than 0.1 T score units annually.
In this case, the patient has normal bone density (T score >−1.0), and
if her health and weight remain unchanged, it is highly unlikely that her
bone density will fall below a T score of −2.5 for at least 10 years.
It would be reasonable for her to wait 5 years or longer to have a repeat
At a quality meeting, you realize that only a small proportion of elderly
women with hip fractures in your organization have received densitometry or
been treated for osteoporosis.
In one study6 of US primary care practitioners,
72% never used densitometry. Barriers to use included cost, unfamiliarity
with guidelines, uncertainty regarding clinical applicability, minimal impact
on treatment decisions, and limited availability. Yet another issue has been
that the reports of densitometry results are not readily comprehensible to
primary care physicians. A randomized trial7 assessing
the impact of providing longer clinical reports to physicians showed that
the longer reports nearly doubled the use of testing and resulted in much
less confusion about reports (confusion fell from 36% to 1%). Patients with
fractures often do not get screened; for example, a study of patients with
vertebral fractures found incidentally on chest radiography showed that this
group rarely got screened or received appropriate therapy.8 Even
many patients receiving long-term steroid therapy have not been screened.9
You come to your group's quality circle, excited about doing more screening,
but find that your group does not want the so-called burden.
Most primary care physicians feel enormous time pressure and need to
meet productivity standards. There are many ways to deal with screening. One
approach is to have patients fill out a questionnaire to assess risk factors
for fracture, which can be done before or at the beginning of a visit. Although
this questionnaire can be paper-based,10 patients
with Internet access can readily complete it online,11,12 which
in the future will likely represent an important time-saver. Another approach
is to shift some of this assessment to a nurse in the practice. To achieve
substantial improvement in osteoporosis detection and care, most practices
will need to develop better systems.
These data point to a number of clinician needs, including tools to
identify patients most likely to benefit from screening and systems that facilitate
routine application of such tools. In particular, orthopedists and internists
should develop routine systems for regularly conducting densitometry in postmenopausal
patients and older men who present with fractures. Physicians and patients
need more informative densitometry reports that include an estimate of a patient's
fracture risk. Better evidence is needed about the potential value of monitoring
bone density, and better guidelines about testing frequency that take into
account the bone density, treatment, age, and clinical conditions of each
patient are needed. As these needs are met, the gaps between evidence and
practice for this important clinical problem will narrow.