M

CLINICAL RESEARCH STUDY

Interventions to Increase Osteoporosis Treatment inPatients with ‘Incidentally’ Detected Vertebral FracturesSumit R. Majumdar, MD, MPH,a,b Finlay A. McAlister, MD, MSc,a,b Jeffrey A. Johnson, PhD,b Debbie Bellerose, BScN,a

Kerry Siminoski, MD,a David A. Hanley, MD,c Ibrahim Qazi, PhD,a Douglas A. Lier, MA,a Robert G. Lambert, MD,d

Anthony S. Russell, MD,a Brian H. Rowe, MD, MSce

aDepartment of Medicine and bSchool of Public Health, University of Alberta, Edmonton, Alberta, Canada; cDepartment of Medicine,University of Calgary, Calgary, Alberta, Canada; dDepartment of Radiology and Diagnostic Imaging and eDepartment of Emergency

edicine, University of Alberta, Edmonton, Alberta, Canada.

and the accuracy

0002-9343/$ -see fhttp://dx.doi.org/10

ABSTRACT

BACKGROUND: Most vertebral compression fractures are not recognized or treated. We conducted acontrolled trial in older patients with vertebral fractures incidentally reported on chest radiographs,comparing usual care with osteoporosis interventions directed at physicians (opinion-leader-endorsedevidence summaries and reminders) or physicians�patients (adding activation with leaflets and telephonecounseling).METHODS: Patients aged �60 years who were discharged home from emergency departments and who hadvertebral fractures reported but were not treated for osteoporosis were allocated to usual care (control) orphysician intervention using alternate-week time series. After 3 months, untreated controls were re-allocated to physician�patient intervention. Allocation was concealed, outcomes ascertainment blinded,and analyses intent-to-treat. Primary outcome was starting osteoporosis treatment within 3 months.RESULTS: There were 1315 consecutive patients screened, and 240 allocated to control (n � 123) orphysician intervention (n � 117). Groups were similar at baseline (average age 74 years, 45% female, 58%previous fractures). Compared with controls, physician interventions significantly (all P �.001) increasedosteoporosis treatment (20 [17%] vs 2 [2%]), bone mineral density testing (51 [44%] vs 5 [4%]), and bonemineral density testing or treatment (57 [49%] vs 7 [6%]). Three months after controls were re-allocatedto physician�patient interventions, 22% had started treatment and 65% had bone mineral density testingor treatment (P �.001 vs controls). Physician�patient interventions increased bone mineral density testingor treatment an additional 16% compared with physician interventions (P � .01).CONCLUSIONS: An opinion-leader-based intervention targeting physicians substantially improved rates ofbone mineral density testing and osteoporosis treatment in patients with incidental vertebral fractures,compared with usual care. Even better osteoporosis management was achieved by adding patient activationto physician interventions [NCT00388908].© 2012 Elsevier Inc. All rights reserved. • The American Journal of Medicine (2012) 125, 929-936

KEYWORDS: Guidelines; Osteoporosis; Quality improvement; Treatment; Trials

conception and design and analysis and interpretation and providedcritical revision to manuscript drafts. IQ also undertook statisticalanalyses. SRM also wrote the first draft, obtained funding, and super-vised the study.

Requests for reprints should be addressed to Sumit R. Majumdar, MD,MPH, Department of Medicine, University of Alberta, 2F1.24 WalterMackenzie Health Sciences Centre, University of Alberta Hospital, 8440-112th Street, Edmonton, Alberta T6G 2B7, Canada.

Funding: Peer-reviewed operating grants from Knowledge TranslationCanada, the Canadian Institutes of Health Research (MOP #151454), andthe Alberta Heritage Foundation for Medical Research (AHFMR). S.R.M.,F.A.M., J.A.J. hold salary awards from AHFMR, and J.A.J. and B.H.R.hold Canada Research Chairs.

Conflict of Interest: None.Authorship: The corresponding author (SRM) had full access to all

the data in the study and takes responsibility for the integrity of the data

of the data analysis. All authors were involved in E-mail address: majumdar@ualberta.ca

ront matter © 2012 Elsevier Inc. All rights reserved..1016/j.amjmed.2012.02.021

ttott2tdpat

wrgtsgh

chbtoyv

bmaattrUs

orosi

930 The American Journal of Medicine, Vol 125, No 9, September 2012

Osteoporosis is a common, chronic, and costly problem thatis under-diagnosed and under-treated.1-4 Spinal or vertebralcompression fractures are the most common osteoporosis-related fractures and are present in 12%-25% of people aged60 years or older.2-7 At least two thirds of vertebral frac-ures, however, escape clinical de-ection because they occur withoutvert symptoms.5-7 Whether a ver-ebral fracture causes acute symp-oms or not, it is associated with a- to 5-fold increased risk of fu-ure fractures, as well as physicaleformities, height loss, chronicain, decrements in quality of life,nd increased morbidity and mor-ality.5-7 Despite their importance,

screening for vertebral fracturesusing spinal radiographs is not arecommended approach.3-7

Many older adults, however,require chest radiographs for otherreasons. Most (80%-90%) verte-bral fractures can be visualized ona chest radiograph, and 15%-25%of radiographs in older patientswill show clinically important ver-tebral fractures.8-15 Furthermore,

hen board-certified radiologistseport an “incidental” vertebral fracture on chest radio-raphs, it reliably and accurately identifies clinically impor-ant fractures with 100% specificity.10,11 This has led to theuggestion that vertebral fractures reported on chest radio-raphs could be used as a case-finding tool for identifyingigh-risk patients6-15 who would benefit from treatment

with drugs like the bisphosphonates, which are known todecrease the risk of future fractures by 50%-60% in thisvery population.3,4,7,16 Because previous studies have beenross-sectional in nature,9-15 it is not even currently knownow often an incidental vertebral fracture report triggersone mineral density testing and osteoporosis treatment inhe setting of “usual care.” Given that usual care rates ofsteoporosis testing and treatment are �10%-20% in theear following symptomatic fractures of the wrist, hip, orertebrae,1,4,8 we suspect that the likelihood of incidentally

reported asymptomatic vertebral fractures triggering osteo-porosis management is even lower.

Therefore, we conducted a pragmatic controlled trial toimprove quality of osteoporosis care in older patients withincidental vertebral fractures reported on chest radiographstaken for other reasons. Our goals were to first define therates of usual osteoporosis care following a vertebral frac-ture report and then to compare usual care with 2 differentosteoporosis quality-improvement interventions: one di-rected at primary care physicians (opinion-leader-endorsedevidence summaries and reminders) and the other directedat both physicians�patients (activation of patients with ed-

CLINICAL SIGNIF

● Osteoporotic franosed and undgraphic screenin

● Older patientsthat often incidfractures.

● In a 3-armed coonstrated that,care, a physiciancould lead to 10osteoporosis treing patient activintervention coquality of osteop

ucational leaflets and telephone counseling).

METHODS

Subjects and SettingThe study was conducted in 2 emergency departments inEdmonton, Alberta, Canada. The Edmonton region has a

large integrated health care deliv-ery system that includes approxi-mately 1 million patients with uni-versal health insurance coverageand 1200 primary care physicians.One emergency department wasaffiliated with a large tertiary-careuniversity teaching hospital, whilethe other was a high-volume free-standing community-based clinic.Both sites employed computerizedinformation systems with electronictriage, patient tracking, electronichealth records, and digital radio-graph archives.

All older subjects that had achest radiograph done for any rea-son were potentially eligible. In-clusion criteria were: age 60 yearsor greater and a chest radiographwith a vertebral fracture reportedby a board-certified radiologist.Pilot data from our region demon-

strated that when the terms “compression, deformity, wedg-ing, loss of height, or fracture” are used by a board-certifiedradiologist to describe a vertebrae, it accurately identifiesthe presence of a vertebral fracture with high inter-rateragreement and 100% specificity.10,11 This pattern of verte-ral fracture recognition has been reported by others inany different settings.2,5,6,9,12 Exclusion criteria were: un-

ble or unwilling to participate or provide informed consent;lready taking prescription osteoporosis treatment; admittedo hospital; unable to read or converse in English; inabilityo contact; or already participating in other osteoporosis-elated studies or trials. The study was approved by theniversity of Alberta Health Ethics Research Board, and all

ubjects provided written informed consent.

Study DesignWe conducted a nonrandomized controlled trial with con-cealed allocation and blinded outcomes ascertainment. Wefirst allocated patients to control or to the physician inter-vention, and after primary outcomes were ascertained at 3months, we re-allocated untreated controls to thephysician�patient intervention and re-ascertained outcomes3 months later. Allocation was based on alternate weeks(physician intervention “on” or “off”) time series methods,and blinding was carried out by assigning scrambled studyidentification numbers to each patient using a secure Inter-net-based computerized sequence generator housed at theEpidemiology Coordinating and Research Center (Edmon-

CE

s are underdiag-ated, but radio-ot recommended.

hest radiographsy report vertebral

ed trial, we dem-ared with usualcted interventionimprovements int, and that add-to the physicianfurther improves care.

ICAN

ctureertreg is n

get centall

ntrollcomp-dire-foldatmenationuld

ton, Alberta). Allocation was concealed from patients (in

fiti

PvmnpPmpfcm

931Majumdar et al Trial of Osteoporosis Interventions

that all received an osteoporosis intervention of some type),research nurses, outcomes assessors, and investigators. Ra-diologists themselves were not aware of the study, and thereports they generated were part of usual clinical care.Alternate-week time series allocation has been used in thepast and consistently balances patient groups.17-19 The de-sign is of sufficient validity to be included in systematicreviews conducted by the Cochrane Collaboration EffectivePractice and Organization of Care (EPOC) group.20

Study ProceduresTrained research staff accessed computerized emergencydepartment records on a daily basis. Because it takes 1-2days for the official chest radiograph report to become partof the electronic medical record, and patients had alreadyleft the emergency department, recruitment and consenttook place by telephone. Research staff generated a list ofpotentially eligible patients based on age and chest radio-graphs.10,11 Radiograph reports and medical records werescrutinized only for available inclusion (ie, vertebral frac-ture reported) and exclusion (ie, osteoporosis treatment,hospital admission) criteria. During weeks when the physi-cian intervention was “on,” eligible patients were tele-phoned. During “off” weeks, the same recruitment list wasgenerated, but patients were not contacted for 3 months, aperiod of time we considered “usual care.”

Study ArmsControl Group. The control group (enrolled during inter-vention “off” weeks) received usual care for 3 months.Within 1 week of their emergency department visit, weensured that the primary care physician of record was sentall documentation about the visit, planned follow-up ap-pointments, and the official chest radiograph report.

Physician Intervention Group. The physician interventiongroup (enrolled during “on” weeks) received usual care, andtheir physicians received a patient-specific reminder with alocal opinion-leader-endorsed evidence summary deliveredby fax or electronic or regular mail. These evidence sum-maries were a local adaptation of available guidelines andprovided treatment recommendations generated and en-dorsed by locally nominated osteoporosis opinion leaders.We have previously used these methods in studies of osteo-porosis, ischemic heart disease, and heart failure.19,21-23 Theopinion leaders were nominated by surveying all physiciansin the region using validated sociometric instruments.19,21-23

The evidence summaries were brief and emphasized 3points (see Appendix, online): 1) patients with a vertebralfracture and osteoporosis are at 20-fold increased risk offuture fracture compared with those with no vertebral frac-tures and normal bones; 2) this patients’ risk of anotherfracture is as high as 20% over the next year; and 3)evidence-based treatments can reduce risk of fracture by50%. Calcium and vitamin-D supplements were mentioned

as necessary but insufficient background therapy. Bisphos-

phonates were recommended as first-line treatment for pa-tients with low bone mass and vertebral fracture, but na-tional guideline-endorsed and -approved second-linealternatives (ie, calcitonin, estrogen, raloxifene, teripa-ratide) were mentioned.3,4,7,16 The reminder and summary

t on a single page, and because they were labeled withhe patient’s name, were intended to be part of the med-cal record and act as point-of-care reminders.

hysician�Patient Intervention. Physician�patient inter-ention (former controls re-allocated to an intervention 3onths after emergency department visit if they were still

ot treated for osteoporosis) consisted of usual care plus thehysician intervention, plus a patient activation strategy.atient activation was achieved by providing educationalaterials (Osteoporosis Canada pamphlets and a single-

age lay summary of the relationship between vertebralractures and osteoporosis) by mail and telephone-basedounseling by a nurse practitioner that reinforced printedaterials and suggested a visit with their physicians.

Outcomes and MeasurementsThe primary study outcome was starting any effectiveprescription osteoporosis treatment within 3 months. Thiswas based on self-report to ensure that information re-lated to the use of free samples and calcium and vita-min-D supplements were captured. All prescription med-ication starts were verified with dispensing pharmacies.As in prior studies,19,21-23 100% agreement between self-report and dispensing records for new medication startswas documented. Secondary outcomes included newbone mineral density tests confirmed by ordering physi-cians and the composite endpoint of new bone mineraldensity test or new osteoporosis treatment. Other patientcharacteristics were based on self-report and collectedusing standardized data collection forms. In a randomsample of 50 patients, we undertook time-motion studiesand determined intervention costs.

Sample Size ConsiderationsWe used the patient as the unit of analysis, osteoporosistreatment at 3 months as the primary outcome, assumedcontrol treatment rates of 10%, defined a 20% absoluteincrease in treatment attributable to intervention as the min-imal clinically worthwhile difference (as per external ex-perts19,22), alpha � 0.05, and beta � 0.20, and calculated atotal sample size of 150 patients. This was inflated to 240patients to allow analysis of the physician�patient interven-tion and to account for attrition.

AnalysesThe patient was the unit of allocation, analysis, and causalinference; based on prior trials, we did not expect anystatistical clustering of patients within physicians.19,21-23

Descriptive data included proportions, means with standard

deviations (SD), or medians with interquartile range, as

articip

932 The American Journal of Medicine, Vol 125, No 9, September 2012

appropriate. Analyses were conducted according to the in-tention-to-treat principle and compared the proportion ofpatients achieving the primary outcome in the physicianintervention group with the control group using a chi-squared test for statistical significance and relative riskswith 95% confidence intervals (CI) to quantify strength ofassociation. Comparisons between the physician�patientintervention and controls were a within-group matched-pairs analysis, and McNemar’s test was used.

RESULTSFrom 2006 to 2010, we screened 19,915 chest radiographs,identified 1315 potentially eligible patients with a vertebralfracture, and enrolled 240 subjects (Figure 1). Four reasons

Eli

Subjects sc

ALLO

Allocated to physicianinterven�on: 117

At 3 months:

116 subjects1 lost: study dropout

Analyzed: 117

At 6 months:

108 subjects8 lost: 2 died, 6 dropouts

Analyzed: 117

Figure 1 Flow of study p

accounted for 97% of exclusions: already treated for osteo-

porosis (480 [45%]), unwilling to consent (277 [26%]),admitted to hospital (224 [21%]), or nonfluent in English(50 [5%]). Of note, 410 of 480 (85%) subjects were ex-cluded from the trial because they were already treated forosteoporosis were women. We allocated 117 patients to thephysician intervention and 123 patients to control. By studycloseout, only 9 (4%) patients had been lost: 7 withdrew and2 died. All 240 patients were included in the primary anal-ysis (Figure 1).

Patient CharacteristicsAverage age of the 240 study patients was 74 (SD 9) years,56% were male, 96% were white, 32% lived alone, andmedian number of comorbidities was 4 (Table 1). The most

0

1315

240

Allocated to usual care(controls): 123

123 subjectsNone lostAnalyzed: 123 for main comparison,2 achieved primary outcome;121 RE-ALLOCATED to physician+pa�entinterven�on

121 subjectsNone lost

Analyzed: 121

Excluded: 1075• 480 (45%) already treated for osteoporosis• 277 (26%) unable or unwilling to consent• 224 (21%) admi�ed to hospital• 50 (5%) non-fluent in English• 31 (3%) unable to contact• 13 (1%) other reasons

ants (CONSORT diagram).

gible: 24

reened:

CATED:

common reasons for emergency department visits were car-

[

m

933Majumdar et al Trial of Osteoporosis Interventions

diovascular problems (eg, chest pain, heart failure, syncope[44%]), shortness of breath (eg, asthma, chronic obstructivepulmonary disease, pneumonia [18%]), and abdominalsymptoms (eg, nausea, vomiting, pain [20%]). While nopatients were taking osteoporosis medications at baseline bydesign, 58% reported at least one previous fracture since theage of 40 years. There were no statistically significant dif-ferences between the physician intervention and controlpatients (Table 1).

Physician Intervention versus Usual CareControlsDuring 3 months of usual care, the rate of new osteopo-rosis treatment triggered by identification of a vertebralfracture was 2% and the rate of bone mineral densitytesting was 4%. Compared with usual care, the physicianintervention increased rates of new osteoporosis treat-ment (20 of 117 [17%] vs 2 of 123 [2%]; 15% absolutedifference, P �.001), bone mineral density testing (5144%] vs 5 [4%]; 40% difference, P �.001), and the

composite endpoint of bone mineral density testing ortreatment (57 [49%] vs 7 [6%]; 43% difference, P �.001)within 3 months of study entry (Table 2, Figure 2). Thenumber-needed-to-treat for these outcomes ranged from

Table 1 Characteristics of 240 Study Patients

Variables

SociodemographicMedian age, years (interquartile range)MaleWhiteLess than high school educationRetiredLives alone

ComorbiditiesMedian number (interquartile range)Heart diseaseCOPDAcid-peptic diseaseDiabetesDepressionNeeds assistance to walk or rise fromchair

Osteoporosis-relatedPostmenopausal womanAny previous fracture as an adultWeight �57 kgCurrent smokingMore than 2 alcoholic drinks dailyThyroid diseasePast steroidsCalcium supplementsVitamin D supplements

COPD � chronic obstructive pulmonary disease.

2 to 7 (Table 2). From 3 to 6 months, rates of testing and

treatment returned to baseline: 5 (4%) additional patientsstarted osteoporosis treatment and 8 (7%) more had bonemineral density tests.

Impact of the Physician�Patient InterventionFor all 3 study outcomes, the physician�patient interven-tion was significantly and substantially better than usualcare (P �.001 for all comparisons; Table 2). Comparedwith the physician intervention, the physician�patient in-tervention further increased rates of new osteoporosis treat-ment (26 of 121 [22%] vs 20 of 117 [17%]; P � .39), bone

ineral density testing (69 [57%] vs 51 [44%]; P � .038),and new bone mineral density testing or treatment (79[65%] vs 57 [49%]; P � .010; Table 2).

Intervention CostsThe physician intervention, including the initial screeningof 83 chest radiograph reports per enrolled patient, took anaverage of 34 minutes per patient at a cost of $1 per minute($34 total) based on a registered nurses’ average salary($37.33 per hour) with 18% benefits and 33% overhead. Thephysician�patient intervention required another 8 minutes

ing to Allocation Status

Intervention117 (%)

Controlsn � 123 (%)

P Value forDifference

66-82) 75 (66-83) .7853) 72 (59) .3996) 94 (96) 1.0036) 38 (31) .7673) 92 (75) .3931) 44 (33) .83

2-6) 4 (1-7) .2943) 64 (52) .1515) 17 (14) .7340) 48 (39) .8621) 27 (22) .7815) 28 (23) .1521) 29 (23) .91

47) 51 (41) .3962) 67 (54) .2110) 14 (11) .8011) 11 (11) .943) 1 (1) .3522) 31 (25) .5921) 17 (14) .1735) 49 (40) .4447) 60 (49) .78

Accord

MDn �

74 (62 (87 (42 (85 (36 (

4 (50 (18 (47 (24 (18 (24 (

55 (73 (12 (10 (3 (

26 (24 (41 (55 (

and cost a total of $42.

repthwrd

934 The American Journal of Medicine, Vol 125, No 9, September 2012

Other Analyses and OutcomesThere were no statistically significant differences in clinicalevents between the 2 intervention groups. Within 3 monthsof exposure to either intervention, no patient had died orbeen admitted to the hospital, but almost all (90%) hadvisited their primary care physician. In terms of osteoporo-sis, no patient sustained a new symptomatic fracture; all 48new medication initiations were for an oral bisphosphonate;and 83 of 125 (66%) bone mineral density tests obtaineddocumented low bone mass at one or more skeletal sites.Last, there was no sex disparity with respect to initiatingosteoporosis treatment after exposure to the interventionswe tested. Specifically, the rate of newly starting osteopo-rosis treatment after the interventions was 17% (23 of 133)among men, compared with 21% (23 of 107) among women(P � .34 for the difference).

Table 2 Effect of 2 Interventions to ImproveFracture Identification on Chest Radiographs

Physician intervention vs usual careOsteoporosis treatmentBMD testingBMD testing or treatment

Physician�patient intervention vs usual careOsteoporosis treatmentBMD testingBMD testing or treatment

Physician�patient vs physician interventionOsteoporosis treatmentBMD testingBMD testing or treatment

BMD � bone mineral density.

Figure 2 Osteoporosis treatment and bone mineral density(BMD) testing 3 months after interventions compared with

usual care.

DISCUSSIONWhen vertebral fractures are reported on chest radiographsin the setting of usual care, their presence triggers osteopo-rosis testing or treatment in only about 1 in 20 patients. Thefact that 58% of the study subjects already had clinicallyrecognized fractures before their chest radiographs weretaken and none were treated for osteoporosis confirms thatwe identified a very high-risk population. In a controlledtrial we found that, compared with usual care, interventionsdirected at physicians alone (opinion-leader-endorsed evi-dence summaries and reminders) or physicians�patients(adding a patient activation strategy with educational leaf-lets and telephone counseling) led to statistically significant8- to 10-fold increases in osteoporosis treatment, bone min-eral density testing, and the composite quality measure oftesting or treatment within 3 months. The addition of apatient activation strategy to a physician intervention wasinexpensive, and it led to even greater improvements inosteoporosis care. The number-needed-to-treat to improveosteoporosis testing or treatment after vertebral fractureidentification, for either intervention we studied, was 2.

Nine controlled trials using 10 different interventions toimprove quality of osteoporosis care after 1522 symptom-atic fractures have been reported.24 These interventionsange from generic patient letters to decision support withlectronic reminders to nurse case-managers.24 Indeed, allostfracture osteoporosis interventions studied in these 9rials were effective to some extent, although the effectsave been modest compared with controls: pooled increasesith any intervention of 20% (95% CI, 10-30) for osteopo-

osis treatment and 36% (95% CI, 21-50) for bone mineralensity testing.24 Although it remains uncertain whether and

to what degree point-of-care reminders and evidence sum-maries can improve processes-of-care and clinical outcomes

orosis Testing and Treatment after Vertebral

luteease

CI)

Relative RiskImprovement(95% CI)

NumberNeeded ToTreat

(8-23) 10.5 (2.5-44.0) 7(30-49) 10.7 (4.4-25.9) 3(33-52) 8.6 (4.1-17.2) 2

(12-28) 13.0 (3.1-54.8) 5(42-61) 13.8 (5.6-34.2) 2(48-67) 11.3 (5.3-24.2) 2

(�6-14) 1.3 (0.8-2.1) 20(1-26) 1.3 (1.0-1.7) 8(4-28) 1.3 (1.1-1.7) 6

Osteop

AbsoIncr(95%

15%40%43%

20%53%59%

5%13%16%

in other areas,25 our earlier studies and now this trial suggest

tmc

935Majumdar et al Trial of Osteoporosis Interventions

that the inclusion of local opinion leader endorsements canaugment the efficacy of quality improvement interventionsdirected at physicians.19,21-23

Despite some novelties and strengths, our work has severallimitations. First, we did not use random allocation. Second,we did not re-review chest radiographs to confirm or refute thepresence of vertebral fractures, but instead relied on the officialreports. Third, process-of-care measures were examined asstudy endpoints rather than clinical outcomes such as newfractures. The goal of this study was not to re-affirm treatmentefficacy; rather, the goal was to identify ways to accelerate thetranslation of this poorly adopted evidence into clinical prac-tice. Indeed, several organizations have endorsed the measureswe used as markers of high quality postfracture osteoporosiscare,26,27 and when such measures are evidence based andightly linked to clinical outcomes such as recurrent fracture orortality, they are very sensitive indicators of the quality of

are delivered.27-29

Lastly, even with the interventions studied, most (80%)study subjects remained untreated for osteoporosis, and ourstudy is limited by the fact that we do not know what theachievable rates of treatment ought to be or the reasons thatphysicians or patients might have chosen to forgo treatment.The subjects in our study had very high rates of potentiallylife-limiting cardiopulmonary disease and multiple comor-bidities, and may have had foreshortened life expectancy; 3months postdischarge might not be long enough for conva-lescence that would permit other health issues such as os-teoporosis to start to take priority; bisphosphonates (in par-ticular) might be withheld from elderly patients with activeor ongoing abdominal complaints, chronic kidney disease,or risk factors for osteonecrosis of the jaw or atypicalfemoral fractures; and finally, there might be concerns re-lated to poly-pharmacy and drug interactions in this elderlycohort. Collectively, such issues might lead pragmatic phy-sicians and their informed patients to appropriately forgo ordelay osteoporosis treatment, and so we estimate that bench-mark rates of osteoporosis treatment in our study settingshould be at least 60%-70%—and certainly far greater thanthe 2% rates currently achieved with usual care.

In conclusion, even when vertebral compression fracturesare actually reported on chest radiographs, osteoporosis man-agement is rarely initiated. An opinion-leader-based interven-tion targeting physicians substantially improved rates of bonemineral density testing and osteoporosis treatment comparedwith usual care, and even better management was achieved byadding a patient activation strategy. Both interventions weresimple and inexpensive and, until future studies are completedto determine methods that are more effective, we believe thatthe interventions tested here could be safely and effectivelyimplemented in other settings.

References1. Giangregorio L, Papaioannou A, Cranney A, Zytaruk N, Adachi JD.

Fragility fractures and the osteoporosis care gap: an international

phenomenon. Semin Arthritis Rheum. 2006;35:293-305.

2. Delmas PD, Langerjit LV, Watts NB, Eastell R, Genant H; IMPACTInvestigators. Underdiagnosis of vertebral fractures is a worldwideproblem. J Bone Miner Res. 2005;20:557-563.

3. National Osteoporosis Foundation. Clinicians’ Guide to Preventionand Treatment of Osteoporosis [2008]. Available at: http://www.nof.org/professionals/clinical-guidelines. Accessed January 11, 2012.

4. Papaioannou A, Leslie WD, Morin S; Scientific Advisory Council ofOsteoporosis Canada. 2010 clinical practice guidelines for the diag-nosis and management of osteoporosis in Canada. CMAJ. 2010;182:1864-1873.

5. Schousboe JT, Vokes T, Broy SB, et al. Vertebral fracture assessment:the 2007 ISCD official positions. J Clin Densitom. 2008;11(1):92-108.

6. Lentle BC, Brown JP, Khan A, Leslie WD, Levesque J, Lyons DJ;Scientific Advisory Council of Osteoporosis Canada; Canadian Asso-ciation of Radiologists. Recognizing and reporting vertebral fractures:reducing the risk of future osteoporotic fractures. Can Assoc Radiol J.2007;58:27-36.

7. Ensrud KE, Schousboe JT. Vertebral fractures. N Engl J Med. 2011;364:1634-1642.

8. Majumdar SR. Under-appreciation of osteoporosis-related vertebralfractures on chest radiographs. Adv Osteoporotic Fract Manag. 2006;4:114-118.

9. Gehlbach SH, Bigelow C, Heimisdottir M, May S, Walker M, Kirk-wood JR. Recognition of vertebral fracture in clinical setting. Osteo-poros Int. 2000;11:577-582.

10. Kim N, Rowe B, Raymond G, et al. Underreporting of vertebralfractures on routine chest radiography. AJR Am J Roentgenol. 2004;182:297-301.

11. Majumdar SR, Kim N, Colman I, et al. Incidental vertebral fracturesdiscovered with chest radiography in the emergency department: prev-alence, recognition, and osteoporosis management in a cohort of el-derly patients. Arch Intern Med. 2005;165(8):905-909.

12. Mui LW, Haramati LB, Alterman DD, Haramati N, Zelefsky MN, Man-erman D. Evaluation of vertebral fractures on lateral radiographs of innercity postmenopausal women. Calcif Tissue Int. 2003;73:550-554.

13. Neuner JM, Zimmer JK, Hamel MB. Diagnosis and treatment ofosteoporosis in patients with vertebral compression fractures. J AmGeriatr Soc. 2003;51:483-491.

14. Kroth PJ, Murray MD, McDonald CJ. Undertreatment of osteoporosisin women based on detection of vertebral compression fractures onchest radiography. Am J Geriatr Pharmacother. 2004;2(2):112-118.

15. Morris CA, Carrino JA, Lang P, Solomon DH. Incidental vertebralfractures on chest radiographs: recognition, documentation, and treat-ment. J Gen Intern Med. 2006;21:352-356.

16. MacLean C, Newberry S, Maglione M, et al. Systematic review:comparative effectiveness of treatments to prevent fractures in menand women with low bone density or osteoporosis. Ann Intern Med.2008;148(3):197-213.

17. Weingarten SR, Riedinger MS, Conner L, et al. Practice guidelinesand reminders to reduce duration of hospital stay for patients withchest pain—an interventional trial. Ann Intern Med. 1994;120:257-263.

18. Sin DD, Bell NR, Man SFP. Effects of increased primary careaccess on process of care and health outcomes among patients withasthma who frequent emergency departments. Am J Med. 2004;117:479-483.

19. Majumdar SR, Rowe BH, Folk D, et al. A controlled trial to increasedetection and treatment of osteoporosis in older patients with a wristfracture. Ann Intern Med. 2004;141(5):366-373.

20. Cochrane Effective Practice and Organization of Care Review Group.Available at: epoc.cochrane.org. Accessed January 11, 2012.

21. Majumdar SR, Tsuyuki RT, McAlister FA. Impact of opinion leader-endorsed evidence summaries on the quality of prescribing for patientswith cardiovascular disease: a randomized controlled trial. Am Heart J.2007;153:22-29.

22. Majumdar SR, Johnson JA, McAlister FA, et al. Multifaceted inter-

vention to improve osteoporosis diagnosis and treatment in patients

2

2

2

2

2

2

2

936 The American Journal of Medicine, Vol 125, No 9, September 2012

with recent wrist fracture: a randomized controlled trial. CMAJ.2008;178(5):569-575.

3. McAlister FA, Fradette M, Majumdar SR, et al. The enhancing sec-ondary prevention in coronary artery disease (ESP-CAD) trial. CMAJ.2009;181:897-904.

4. Little EA, Eccles MP. A systematic review of the effectiveness ofinterventions to improve post-fracture investigation and managementof patients at risk of osteoporosis. Implement Sci. 2010;5:80-88.

5. Shojania KG, Jennings A, Mayhew A, Ramsay C, Eccles M,Grimshaw J. Effect of point-of-care computer reminders on physi-cian behaviour: a systematic review. CMAJ. 2010;182:216-

225.

6. National Council on Quality Assurance. HEDIS and Quality Measure-ment. Available at: http://www.ncqa.org/tabid/59/Default.aspx. Ac-cessed January 11, 2012.

7. Higashi T, Shekelle PG, Adams JL, et al. Quality of care isassociated with survival in vulnerable older patients. Ann Intern Med.2005;143:274-281.

8. Peterson ED, Roe MT, Mulgund J, et al. Association between hospitalprocess performance and outcomes among patients with acute coro-nary syndromes. JAMA. 2006;295:1912-1920.

9. Beaupre LA, Morrish D, Hanley D, et al. Oral bisphosphonates areassociated with reduced mortality after hip fracture. Osteoporos Int.

2011; 22:983-991.