Predictors of Osteoporosis Screening Completion
Rates in a Primary Care Practice
Ramona S. DeJesus, MD,1Rajeev Chaudhry, MBBS, MPH,1Kurt B. Angstman, MD,2
Stephen S. Cha, MS,3Sidna M. Tulledge–Scheitel, MD, MPH,1Rebecca L. Kesman, MD,1
Matthew E. Bernard, MD,2and Robert J. Stroebel, MD1
The United States Preventive Services Task Force and the National Osteoporosis Foundation recommend
routine osteoporosis screening for women aged 65 years or older. Previous studies have shown that the use of a
clinical decision-support tool significantly improves screening rates. In a recently published study, a statistically
significant improvement was found in the screening rates for eligible women with use of the tool. To evaluate
whether a clinical decision-support tool independently predicts completion of osteoporosis screening tests and to
identify predictors of screening completion, we examined the records of 2462 female patients who were eligible
for osteoporosis screening but had no prior baseline screening and who were seen in our primary care practices
in 2007 and 2008. Patient and provider characteristics and clinic visit type were identified, and their association
with screening test completion was statistically analyzed using both univariate and multivariate models.
Screening completion rates increased significantly from 2007 to 2008. Factors associated with increased likeli-
hood of screening completion included race, marital status, residence, presence of comorbidity (cancer, rheu-
matologic disease), and the year and type of visit. Screening was less likely for women aged 80 years or older.
The use of a point-of-care decision-support tool not only improved osteoporosis screening rates significantly but
appeared to be an independent predictor of screening completion. It potentially can facilitate the systematic and
effective delivery of preventive health services to patients in the primary care setting. (Population Health Man-
low bone mass and structural deterioration of bone tissue, it
causes increased bone fragility and poses an increased frac-
its debilitating consequences constitute a substantial national
economic burden.1,2To address this growing threat, the 2002
United StatesPreventive ServicesTaskForceandtheNational
Osteoporosis Foundation recommended routine screening for
osteoporosis for all women aged 65 years or older.1,3Bone
mineral density measurement by dual-energy X-ray absorp-
tiometry (DEXA) is the preferred screening test and the best
predictor of fracture.1,2The task of screening eligible patients
for osteoporosis rests primarily with primary care providers.
However, despite these recommendations, the osteopo-
rosis screening rate remains low, at 12% to 56%.4,5The use of
steoporosis is a common condition that affects about
8 million women in the United States. Characterized by
a clinical decision-support system has been shown to im-
prove the screening rate for osteoporosis.6A recently pub-
lished study reviewed records of female patients who were
eligible for osteoporosis screening in a primary care practice,
before and after implementation of a point-of-care clinical
decision-support system; it found a statistically significant
improvement in screening rates with use of the tool.7The
study, however, did not look into factors that predict
screening completion during clinic visits. A previously con-
ducted retrospective review of test ordering for abdominal
aortic aneurysm showed that screening was more likely to be
ordered at a general medical examination than at an acute
care visit or follow-up examination.8
Our study sought to identify patient and provider charac-
teristics that predict completion of osteoporosis screening
within 30 days of a clinic visit among eligible female patients
in 4 primary care clinics during 2007 and 2008. The associa-
tion of clinic visit type with screening test ordering also was
1Division of Primary Care Internal Medicine, Center for Innovation, and Departments of2Family Medicine and3Health Sciences Research,
Mayo Clinic, Rochester, Minnesota.
POPULATION HEALTH MANAGEMENT
Volume 14, Number 5, 2011
ª Mary Ann Liebert, Inc.
analyzed. We hypothesized that use of a point-of-care deci-
sion-support tool, the Generic Data Management System
(GDMS), was an independent factor that would predict
completion of osteoporosis screening in a primary care setting.
Mayo Clinic’s Employee and Community Health practice
provides primary care to more than 100,000 patients at 4
practice sites in the Rochester, Minnesota area that are
staffed by 45 internists and 50 family physicians. The GDMS,
a clinical decision-support tool, was adopted at all practice
sites of Employee and Community Health for adults in Jan-
uary 2008. Details of the tool were described in a prior
publication.7The GDMS includes a rule-based application
coded with national guidelines for age- and gender-specific
preventive services and outcome measures for chronic dis-
eases such as diabetes and coronary artery disease. Based on
data from electronic records, GDMS provides point-of-care
decision support regarding services that a patient needs at
the time of the visit and in the next 90 days. It checks for
prior bone density tests among women ages 65 years and
older and identifies those eligible for initial screening. Under
a revised workflow, a paper copy of the GDMS summary
screen is included in the rooming packet for the allied health
staff. Rooming personnel then place the order for DEXA for
women who meet initial osteoporosis screening criteria as
recommended on the GDMS. The provider activates the
order after discussion with the patient.
Table 1. Univariate Analysis of Variables Associated with Test Completion, 2007 and 2008
sum of diseases
sum of diseases
Charlson comorbidity index
Congestive heart failure
Peripheral vascular disease
Coronary vascular disease
Chronic pulmonary disease
Mild liver disease
Renal disease, moderate
Severe liver disease
Olmsted County Resident?, No. (%)
244 DEJESUS ET AL.
An independent data abstractor reviewed all records of
female patients aged 65 years or older who were seen in the
Family Medicine and Primary Care Internal Medicine prac-
tice sites in 2007, prior to GDMS implementation, and in
2008, 1 year into GDMS implementation. Patient character-
istics, which included age, sex, race, marital status, residence,
comorbidity, type of clinic visit (full vs. limited examination),
and provider specialty (primary care internal medicine or
family medicine) were identified.
Pearson’s chi-square test was used for univariate analysis
to compare screening rates between 2007 and 2008 and to
identify variables that predicted test completion each year.
SAS GENMOD statistical analysis was conducted to analyze
if any of the variables studied independently predicted
completion of osteoporosis screening 30 days after a clinic
visit. All statistical analyses were performed with SAS ver-
sion 9.1.3 software (SAS Institute, Inc., Cary, NC). A P value
of <0.05 was considered statistical significant.
A total of 2462 women who were eligible for osteoporosis
screening were seen in both years. Most were white (96%)
and resided in Olmsted County (70%); 59% of the women
were married, and 32% were widowed; 61% were seen for a
full examination, while 35% had a limited examination. Fa-
mily medicine providers saw 67% of patients studied; pri-
mary care internal medicine providers saw 33%.
In 2007, 897 (76.3%) of the 1128 eligible women seen in
both primary care internal medicine and family medicine
clinics completed the screening test for osteoporosis within
30 days of their clinic visit. The remaining 278 were not
screened. A total of 1047 (81.4%) of 1287 eligible patients
completed screening in 2008. The difference in completion
rates between the 2 years was statistically significant (2-tailed
P value of 0.0027 using Pearson’s chi-square test).
In univariate analysis using the Charlson Comorbidity
Index, age, race, marital status, residence, and the presence
of 3 comorbidities (ie, congestive heart failure, uncompli-
cated diabetes, cancer) were statistically significant predic-
tors of osteoporosis screening completion (P value: <0.005).
The likelihood of missed screening was lower in the presence
of these variables (Table 1). The severity-weighted sum of
diseases was not associated with test completion; however,
when age was added to the severity-weighted sum of dis-
eases, there was a statistically significant association with test
ordering, but only in 2008 (P¼0.035).
Using the multivariate model, age, race, marital status,
residence, and the presence of a comorbidity (ie, cancer,
rheumatologic disease) were independent predictors of
screening test completion. Women who were white, married,
and residents of Olmsted County were more likely to be
screened (P¼<0.005). Likewise, women with a diagnosis of
cancer or rheumatologic disease were 50% more likely to
complete screening (odds ratio [OR]: 0.5). The year and type
of clinic visit also significantly predicted test completion.
Patients seen in 2008 for full or general examinations were
more likely to be screened (Table 2). Interestingly, age was
inversely correlated to test completion; screening was more
likely to be missed in older persons, specifically those aged
80 years or older (OR: 1.059). The provider’s specialty did not
predict test completion (P¼0.16).
In uinvariate analysis, age and presence of chronic illness
(eg, cancer, diabetes) were significant predictors of screening
test completion during both 2007 and 2008. Older women
with comorbidities are probably seen more often in their
primary care clinic, which increases the opportunity to cap-
ture gaps in preventive care. The actual number of clinic
visits that each patient made during each year was not ex-
amined and potentially would constitute a variable in oste-
oporosis screening completion rates.
With multivariate analysis, age remains an independent
predictor of screening completion. While it is an established
risk factor for osteoporosis and would generally trigger
screening,9this study showed advancing age to be associated
with a decreased likelihood of completing the screening test;
women aged 80 years or older are less likely to be screened.
Indeed, increasing age by itself has been identified as a
barrier to effective osteoporosis care, along with the pres-
ence of dementia, lack of treatment adherence, and inade-
quate social support among vulnerable elderly patients.10
Table 2. Multivariate Analysis of Variables Significantly Associated with Test Completion
95% CI95% CI
Estimate SELowerUpper Odds ratioLowerupperP value
Rheumatologic disease, No. (%)
Cancer, No. (%)
CI, confidence interval; SE, standard error.
PREDICTORS OF OSTEOPOROSIS SCREENING 245
Likewise, it is common practice for women in this age group
who present with vertebral or hip fracture to be clinically
treated for osteoporosis without a screening test. Decreased
life expectancy may also limit screening. One study identified
greater concern about developing osteoporosis and better
knowledge of bone mineral density (BMD) testing as 2 de-
terminants of the readiness to undergo osteoporosis screening
in older, high-risk patients.11A closer look at factors that ac-
count for decreased screening among elderly patients in this
study would be useful in designing effective practice inter-
ventions to improve osteoporosis diagnosis and management.
Race and residence are also significantly associated with
increased screening rates. Because more than two thirds of
the women in this study are white and reside in predomi-
nantly white Olmsted County, the results could have been
skewed. However, the data were unchanged even in the
final, stable multivariate model. There was a trend toward
increased test completion among Asian women in 2008, but
the number is too low to determine statistical significance.
Two diseases, cancer and rheumatologic illness, are asso-
ciated with higher screening rates. Other studies have ob-
served the correlation between osteoporosis and the presence
of these conditions because of the underlying disease pa-
thology and/or treatment (eg, corticosteroids).12Therefore,
their presence is likely to trigger an order for a DEXA test.
Patients with these comorbidities also require frequent clinic
visits, allowing greater opportunity to screen.
There was a statistically significant increase in the osteo-
porosis screening rate from 2007 to 2008. This result con-
firmed the finding of a previous study, which showed that
use of a clinical decision-support tool significantly improved
the osteoporosis screening rate among eligible women seen
in the primary care setting.7
Screening of eligible women is likely to be completed after
a full examination. This finding was also observed in a pre-
vious study that looked at screening orders for abdominal
aortic aneurysm in a primary care practice; the test was more
likely to be ordered at a general or full examination than at
an acute care or follow-up examination.8More time is usu-
ally allotted for a full examination, which provides a greater
opportunity to address preventive health screening.
The characteristics and number of women seen for full and
limited examinations in both years were similar; the only
factor that would account for the statistically significant as-
sociation of the year 2008 with screening completion was the
availability of GDMS. This finding reflects the utility of the
tool in improving preventive care practice across primary
care population groups and supports the hypothesis that it
appears to be an independent predictor of osteoporosis
More eligible patients were seen in family medicine clinics
than in primary care internal medicine; however, test or-
dering did not differ statistically between the two specialty
areas. This also was observed in the retrospective study on
abdominal aortic aneurysm screening in which no difference
was seen in the ordering rates among providers with dif-
ferent roles or of different sexes.8
This study identified factors that predict osteoporosis
screening completion by primary care providers in an aca-
demic institution in the Midwestern United States, and re-
sults may not be generalizable to community-based primary
care practices in other geographic areas. Most of the women
in the study were white, which limits the application of re-
sults to other ethnic and minority groups. It would be in-
teresting to see results from a larger, multicenter study with
a more diverse ethnic population.
Independent predictors of increased likelihood of osteo-
porosis screening completion among women aged 65 years
or older include ethnicity (white), marital status (married),
and the presence of certain comorbidities (ie, cancer,
rheumatologic disease). There is a decreased likelihood of
screening completion with advancing age. Screening is more
likely to be done during full examination visits and with use
of a point-of-care clinical decision-support tool. Use of such a
tool significantly improved osteoporosis screening rates for
an eligible patient group seen in a primary care practice and
appears to independently predict screening completion.
The authors acknowledge the Mayo Foundation for
funding this study and manuscript preparation.
Author Disclosure Statement
Dr. Chaudhry is an employee of Mayo Clinic and the in-
ventor of the GDMS software referenced in this publi-
cation. Mayo Clinic has licensed this technology to a
commercial entity (VitalHealth Software) but to date has
received no royalties. Dr. Chaudhry receives no royalties
from the licensing of this technology. Drs. DeJesus, Ang-
stman, Tulledge–Scheitel, Kesman, Bernard, and Stroebel,
and Mr Cha disclosed no conflicts of interest.
1. National Osteoporosis Foundation. Clinician’s Guide to
Prevention and Treatment of Osteoporosis. Washington,
DC: National Osteoporosis Foundation; 2008.
2. US Department of Health and Human Services. Bone Health
and Osteoporosis: A Report of the Surgeon General. Rock-
ville, MD: US Dept of Health and Human Services, Public
Health Service, Office of the Surgeon General; 2004.
3. US Preventive Services Task Force (USPSTF). Screening for
osteoporosis in postmenopausal women: Recommendations
and rationale. Ann Intern Med. 2002;137:526–528.
4. Gallagher TC, Geling O, Comite F. Missed opportunities for
prevention of osteoporotic fracture. Arch Intern Med.
5. Cohen K, Maier D. Osteoporosis: evaluation of screening
pattern in a primary-care group practice. J Clin Densitom.
6. Kastner M, Straus SE. Clinical decision support tools
for osteoporosis disease management: A systematic review
of randomized controlled trials. J Gen Intern Med. 2008;23:
7. DeJesus RS, Angstman KB, Kesman R, et al. Use of a clinical
decision support system to improve osteoporosis screening.
J Eval Clin Pract. Epub ahead of print August 10, 2010. doi:
8. Eaton J, Reed D, Angstman K, et al. Effect of visit length and
a clinical decision support tool on abdominal aortic screen-
ing rates in a primary care practice: A retrospective review.
J Eval Clin Pract. In press.
246DEJESUS ET AL.
9. Barrett–Connor E, Weiss TW, McHorney CA, Miller PD,
Siris ES. Predictors of falls among postmenopausal women:
Results from the National Osteoporosis Risk Assessment
(NORA). Osteoporosis Int. 2009;20:715–722.
10. Switzer JA, Jaglal S, Bogoch ER. Overcoming barriers to
osteoporosis care in vulnerable elderly patients with hip
fractures. J Orthop Trauma. 2009;23:454–459.
11. Polinski JM, Cadarette SM, Arnold M, et al. High-risk pa-
tients’ readiness to undergo BMD testing for osteoporosis
diagnosis in Pennsylvania. Int Q Community Health Educ.
12. Sinigaglia L, Nervetti A, Mela Q, et al. A multicenter cross
sectional study on bone mineral density in rheumatoid ar-
thritis: Italian Study Group on Bone Mass in Rheumatoid
Arthritis. J Rheumatol 2000;27:2582–2589.
Address correspondence to:
Ramona S. DeJesus, M.D.
Division of Primary Care Internal Medicine
200 First St. S.W.
Rochester, MN 55905
PREDICTORS OF OSTEOPOROSIS SCREENING 247