Journal of Clinical Densitometry
Volume 9, Issue 1 , Pages 72-77, January 2006

Development and Evaluation of a Vertebral Fracture Assessment Program Using IVA and Its Integration With Mobile DXA

Department of Rheumatology, Geisinger Medical Center, Danville, PA

Received 17 May 2005; received in revised form 11 July 2005; accepted 12 August 2005.

Article Outline

Abstract 

Currently, it is unusual to combine evaluation for vertebral fracture with measurement of bone mineral density in clinical practice. Using Quantitative Morphometric Vertebral Analysis (Instant Vertebral Assessment [IVA]) in our existing Mobile Dual-Energy X-Ray Absorptiometry (DXA) Program, we implemented a testing procedure that examined 5 different IVA protocols focusing on clinical utility and cost. Using small-scale tests of change (PDSA cycles), data from the preceding cycle drives the development of the next cycle. In this article, we describe the process and rationale for selecting patients for the IVA study. In addition, we review the literature on vertebral fracture assessment using DXA and emphasize the clinical utility of point of service testing by providing the needed knowledge for best patient care by simultaneous DXA and IVA testing. The application of this new technology increased identification of the high-risk patient by 11%, with a nominal additional cost per DXA study of $14. This study provides a useful framework for the integration of IVA into a clinical DXA program.

Key Words: Dual-energy X-ray absorptiometry (DXA), dual-energy vertebral assessment (DVA), instant vertebral assessment (IVA), osteoporosis, vertebral fracture, vertebral fracture assessment

 

Back to Article Outline

Introduction 

The profound and devastating health consequences of osteoporosis are well understood and reported 1, 2. Evidence based Food and Drug Administration (FDA) approved medications exist for both osteoporosis prevention and treatment, as well as for glucocorticoid induced bone loss. In addition, newer therapies appear on the horizon, potentially bolstering our therapeutic armamentarium. Dual-energy X-ray absorptiometry (DXA) technology remains the “gold standard” for risk assessment of future fracture and determination of whether intervention is necessary. This assessment should be made in the context of other additional risks, such as whether the patient has an existing vertebral fracture (VF). A prevalent VF significantly increases the risk of future vertebral and hip fractures 3, 4.

Currently it is unusual to combine evaluation for VF with measurement of bone mineral density (BMD), except in clinical osteoporosis trials. Published data clearly document that a significant percentage of patients with low BMD have existing, often unrecognized, VF 5, 6, 7, 8. The majority of these are clinically silent 6, 9, 10, 11. Importantly, a significant percentage of patients classified as normal or osteopenic by BMD alone have existing VF 6, 7. It is these patients who have been shown to especially benefit from therapy (6). Having previously reported on the implementation of our Mobile DXA Osteoporosis Diagnostic Unit and appreciating that knowledge of existing VF is important in classifying osteoporosis risk and recommending treatment, we developed and tested a new process for integrating knowledge of existing vertebral fracture, using Instant Vertebral Assessment™ (IVA), within our existing Mobile DXA Program (12).

Back to Article Outline

Methods 

Images of the thoracic and lumbar vertebrae were generated and processed at the time of the patient's DXA study using a Delphi-C Densitometer and Image-Pro™ (Hologic Inc., Bedford, MA). These images were electronically sent via our health system's intranet using Digital Imaging and Communications in Medicine protocol to the physician reader's workstation. All physician readers were International Society for Clinical Densitometry (ISCD) certified and American Board of Internal Medicine certified clinical rheumatologists with several years experience in DXA interpretation.

Visual inspection and quantitative morphometric fracture assessment was performed without knowledge of the patient's DXA results using CADfx™ software (Hologic Inc.) on the same day. The thoracolumbar images were brought into focus on the reader's workstation computer screen. Using the visual software tools (size magnification, contrast, brightness, invert, flip, etc.), each vertebrae (T7–L4) was individually outlined by marking the superior and inferior endplates with 6 available markers (3 on the superior and 3 on the inferior endplates). The software performed quantitative morphometric assessment on each vertebra based on marker placement. Abnormal vertebrae were graded as wedge, biconcave, or crush deformities. Each specific deformity was further subcategorized as mild, moderate, or severe based on the percentage endplate compression in anterior height (wedge), middle height (biconcave), or posterior height (crush) (mild: 20–25%; moderate: 25–40%; severe: >40%). If the reader was unsure of the vertebral endplates or could not accurately outline the endplate margins, then that specific vertebra was excluded from IVA analysis. Once the IVA analysis was completed, a hard copy was printed on each patient. On the subsequent day, the patient's questionnaire, the DXA results, and the IVA results were reviewed as a set, and the interpretation was templated into the electronic medical record and the results were electronically sent to the requesting physician within 48 hours of the time of the study. The IVA interpretation was incorporated into the body of the DXA report rather than being reported separately to promote a simple integrated consultative report for the ordering physician.

A protocol for deciding on the optimal testing strategy for the IVA was designed based on simple measurements that could be determined by the DXA technologist at the time of the study (Fig. 1). The initial protocol was tested on 100 consecutive patients meeting inclusion/exclusion criteria. After the first 100 IVAs were completed, the data was analyzed and the protocol changed according to the previous results. Each of these changes was performed using a series of “Plan–Do–Study-Act” (PDSA) rapid change cycles. The rapid PDSA model utilizes a “trial and learning” approach that allows one to define the problem, implement a change, analyze the results of that change, modify the process, and repeat the cycle within a short time frame (13). Because the PDSA model is a process improvement design, it was not felt that informed consent was necessary. However, approval for the review of the data obtained was granted by our institutional review board.

To measure the benefit of IVA, clinical and financial outcomes were assessed. The clinical outcome of interest was whether IVA changed the risk stratification from low or moderate risk to high risk. Within our DXA program, patients are stratified by relative risk based on the Geisinger Health System and National Osteoporosis Foundation Guidelines (see Table 1) 14, 15. Prescription osteoporosis therapy is only recommended to the high-risk group. We defined “risk changed to high” as either a low or moderate risk patient, which was classified based on the DXA analysis and patient questionnaire, in which the detection of VF on IVA, by itself, changed the relative future fracture risk assessment from low or moderate to high. The risk changed to high group would then be recommended prescription osteoporosis therapy in accordance with our web-based treatment guidelines based on the detection of VF by IVA alone. A cost-analysis was performed for each PDSA cycle, based on $135 per DXA and $30 per IVA. We defined the following cost-analysis items:

(1)
(2)
(3)

Table 1. Relative Fracture Risk
High riskT-score: below −2.0, T-score: −1.6 to −2.0+major risk factora, T-score: below −1.0+chronic glucocorticoid therapy, or nontraumatic VF, or hip fracture
Moderate riskT-score: −1.0 to −1.5 or T-score: −1.6 to −2.0 without a major risk factora
Low riskT-score:−1.0 or above

Abbr: VF, vertebral fracture.

aMajor risk factors included: (1) low body weight less than 127 lbs, (2) current smoker, (3) family history of osteoporotic fracture, (4) personal history of osteoporotic fracture.

Back to Article Outline

Results 

Based on the results of the original PDSA cycle, 5 cycles were accomplished as follows: (1) age 70 protocol: IVA is performed if age ≥70 or height loss ≥1½ inches (first 100 consecutive DXAs); (2) age 60 protocol: IVA is performed if age ≥60 or height loss ≥1½ inches (second 100 consecutive DXAs); (3) model age 65 protocol: data analysis of the age 60 protocol ignoring the IVAs done in those patients aged 60–64; (4) age 65 protocol: IVA performed if age ≥65 or height loss ≥1½ inches (third 100 consecutive DXAs); (5) model age 65 protocol+age 65 protocol: combination of protocols 3 and 4.

Table 2 summarizes the major characteristics and findings of each PDSA cycle. In the age 70 protocol, which included our first 100 consecutive patients studied with IVA capability, 35 IVAs were performed with 8 VFs detected (23%). Importantly, 14% of the IVAs resulted in the DXA interpretation changing to high relative fracture risk (i.e., 14% of IVAs in which the risk changed to high).

Table 2. Characteristics and Findings in Each PDSA Cycle
Age 70 protocolAge 60 protocolModel age 65Age 65 protocolModel+real combined 65
No. of DXAs100100100100200
No. of IVAs3565494190
No. of VFs8981422
% of IVAs with VF23%14%16%34%24%
No. of IVAs where “risk changed to high”a565510
% of IVAs where “risk changed to high”a14%9%10%12%11%
No. of patients with 1 fracture6871320
No. of patients with 2 fractures11112
No. of patients with 3 or more fractures10000
Cost of add IVA to DXA$11$20$15$12$14
Cost of detect 1 VF$131$217$184$88$123
Cost to detect high-risk patient$210$325$294$246$270
Cost of DXA for 100 patients$13,500$13,500$13,500$13,500$13,500
Additional cost of IVA$1,050$1,950$1,470$1,230$1,350

Abbr: DXA, dual-energy X-ray absorptiometry; IVA, instant vertebral assessment; VF, vertebral fracture.

a“Risk changed to high” was defined as either low or moderate risk patient (based on DXA and patient questionnaire) in which the presence of VF on IVA itself changed the fracture risk assessment to high.

Based on this cycle, the next cycle was designed to determine if lowering the age increased VF detection. The age 60 protocol included the next 100 consecutive patients who underwent DXA with IVA capability. As might be expected, because of the lower age cutoff, more IVAs were done (65), adding to the overall costs (an additional $900 for 100 patients). With this lower age cutoff, VFs were detected in only 14% of IVAs, and 9% of the IVAs resulted in risk changed to high. However, 5 patients aged <70 were found to have a VF, verifying that lowering the age did increase VF detection.

The next cycle (model age 65) simply utilized the age 60 protocol data ignoring any data on patients aged 60–64. This model only increased costs by $420 for 100 patients (compared with the age 70 protocol) and would have missed only 1 VF found in the age 60 protocol. Based on this modeling, the next cycle chosen was age 65 protocol performed on the next 100 consecutive patients. This group resulted in the highest percentage VF detection (34 %), with a similar percentage of risk changed to high compared with the age 70 protocol. The final cycle combined data from the model age 65+age 65 protocol (n=200) and demonstrated an increased yield in VF case finding compared with the age 60 protocol with a nominal increase in overall cost compared with the age 70 protocol ($300 per 100 DXAs).

Looking at our entire population studied, 42% of vertebral fractures were mild and 58% were moderate or severe. Table 3 outlines the breakdown of vertebral fracture severity based on PDSA cycle protocol.

Table 3. Vertebral Fracture Classification
Study group% of IVAs with only mild fracture% of IVAs with moderate/severe fracture
Age 70 protocol50%50%
Age 60 protocol33%67%
Model age 6525%75%
Age 65 protocol43%57%
Model+real combined 6536%64%

Abbr: IVAs, instant vertebral assessments.

Using data from the final chosen cycle (model+real combined age 65), the cost to add IVA to DXA was only $14 per study, yielding a cost of $123 to detect one VF and $270 to detect one risk changed to high patient. In comparison, in our entire Mobile DXA Program cohort, in which one-third of the patients were high risk (n=7,368), the cost to detect a high-risk patient by DXA alone was $405 (3 DXAs needed to find one high-risk patient; 3×$135=$405).

Back to Article Outline

Discussion 

It is exceedingly uncommon in clinical practice to combine an assessment for vertebral fracture with measurement of BMD in the clinical evaluation of the individual patient suspected to be at risk from complications of osteoporosis. Ross et al (3) demonstrated that the combination of BMD and VF detection yields an even stronger prediction of the risk for subsequent VF. Genant et al (6) reviewed the available methods for current identification of VF, which included: standard 4-view thoracic and lumbar x-ray with analysis by either qualitative, semi-quantitative method of Genant, or pure quantitative method; lateral spine absorptiometry (e.g., IVA) with fan-beam densitometry systems with either visual/qualitative analysis or new upgraded software techniques (Cadfx™ and others) that allow morphometric analysis. In comparing these techniques for VF detection, image resolution in the upper thoracic spine remains superior with conventional x-rays 6, 8. However, there is an additional cost and radiation exposure when compared with IVA. A very important benefit of IVA is point of service testing, adding to patient convenience in having DXA and IVA performed simultaneously, and creating the ability to integrate the results of DXA and IVA into a seamless, single consultative report.

Several previous studies have assessed the benefits of DXA-based techniques for assessing VF. Greenspan et al (7) evaluated 482 women with no prior known VF who were being recruited for an osteoporosis prevention and treatment trial. The IVA lateral spine imaging (L4–T4) with Hologic QDR-4500A was performed, and the image was displayed on a monitor with adjustable brightness, contrast, and size functionality. Qualitative visual VF assessment was performed by an investigator blinded to subject data. The VFs were detected in 18.3% of asymptomatic subjects. Surprisingly as many as 18.7% of subjects classified as normal based on BMD alone had asymptomatic VFs. When VF was added to BMD classification, the prevalence of osteoporosis changed from 10.6%, based on total hip BMD to 26.1% when VF detected by IVA was considered, and from 25.1% based on spine BMD to 38.6% when VF detected by IVA was considered. The utility of IVA in better capturing osteoporotic patients in need of treatment was emphasized (7).

Similarly, Rea et al (8) compared spinal images obtained via DXA to conventional x-rays in assessing similar unrecognized vertebral deformities. In their study, 2 lateral spine DXA images were obtained at the same time as DXA with the Hologic QDR-4500A, and patients were then sent for anteroposterior and lateral lumbar and thoracic x-rays. Three groups were recruited from the general practice along with a fourth group from a metabolic bone clinic in which patients had low BMD and VF. Plain x-rays (XSQ) were analyzed by the Genant semi-quantitative method, and vertebral images via DXA (VXA) were interpreted by visual qualitative analysis. In this study, 8.4% of patients had one or more VF detected, whereas 5.1% of vertebrae analyzed by VXA versus only 0.5% vertebrae analyzed by XSQ could not be interpreted. Many of the vertebrae that were difficult to analyze were in the upper thoracic spine where VXA/IVA resolution is known to be inferior to standard x-rays (8). Most visualized deformities were seen at T12/L1 or T7/T8. The agreement between XSQ and XVA improved when analysis was confined to L4–T7, with VXA sensitivity of 86%, specificity of 92.7%, a positive predictive value of 84.3%, and a negative predictive value of 93.6% (8).

Most recently, Durosier et al (16) compared automated VF analysis software (CADfx™) with conventional x-rays. Eighty-one subjects (age: 71.3±9.7 yr) had lateral spine imaging (T4–L4) performed on a Hologic Delphi and conventional thoracic and lumbar x-rays. Using conventional x-rays as the standard, the sensitivity, specificity, and negative predictive value of adjusted CADfx™ were 83%, 96.5%, and 98.1%, respectively. Genant et al (6) reported that the agreement between conventional x-rays and DXA-based spine assessment (such as IVA) is as good as that obtained among different radiologists and different VF analysis methods (6).

Although most reported studies evaluating DXA absorptiometry have utilized Hologic technology, Vokes et al (17) described the Lunar Prodigy™ Densitometer (GE Medical Systems, Madison, WI) and their lateral spine imaging system, termed dual-energy vertebral assessment (DVA). They reported on 297 subjects (272 women: aged 64 ± 13 yr) with DVA imaging, and a subset of 66 patients who also had standard thoracic and lumbar spine x-rays. Reviewing the 231 who had only DVA analysis, 14% had VF detected. Including all 297 (66 with DVA and x-rays), 19% had VF detected. Compared with x-rays, DVA had a 95% sensitivity and an 82% specificity, similar to other reports (17).

A number of studies have documented that patients with low BMD or even “normal BMD” have evidence of “silent,” asymptomatic VFs (18). Schousboe et al (18) offered lateral vertebral imaging (i.e., IVA analysis) to 456 consecutive patients scheduled for BMD testing on their Hologic Delphi C densitometer. Compared with prior IVA analyses utilizing either visual qualitative assessment or CADfx™ morphometric assessment, these spinal images were interpreted by the semi-quantitative method of Genant or the pure quantitative morphometric method. The 342 of 456 patients who consented had a 14.6% VF detection rate. Significantly, 9.3% of those aged <60 with normal BMD had one or more VF detected. Of the 73 patients aged >60 with “osteopenia” on BMD, 27.4% had 1 or more VF detected, again documenting the ability to powerfully stratify patients by combining knowledge of BMD with concomitant knowledge of vertebral fracture.

While DXA remains the gold standard for diagnostic classification, risk stratification, and monitoring, there are numerous instances in which the lumbar spine DXA is difficult to interpret. These include VF, spinal degenerative disc disease, posterior facet osteoarthritis, scoliosis, lumbar laminectomies, spinal instrumentation, aortic or pancreatic calcifications, renal or biliary tract stones, contrast agents, or ingested calcium tablets (19). The additional information provided by IVA could be helpful in this population as well.

Using PDSA cycle analysis, we have presented data on the implementation of IVA within our existing Mobile DXA Program. We expected to detect VF, and that in some patients, such VF detection would change the fracture risk beyond what was indicated by DXA analysis alone. Without the VF knowledge that IVA provides, such patients would be at higher risk of future fracture, but would not have had prescription therapy recommended. Analysis of our protocols shows that VF is commonly found by IVA (14–34% of the time) and not uncommonly, changes the fracture risk to high (9–14 % of the time). Lindsay et al (11) report that this fracture population is at significantly increased risk for additional fracture within the next year. Given the availability of a number of therapies that have been shown to reduce the risk of vertebral and hip fractures, it seems paramount to better stratify those patients most at risk with a point of service application, such as IVA.

There are yet no formal published guidelines on the clinical application of IVA by the ISCD or other respected organizations (20). Our study is timely in this clinical regard and it also provides evidence that IVA is a cost-effective test. The final accepted protocol adds only US $14 to the cost of an average DXA, and is less expensive in detecting a high-risk patient than DXA itself.

The direct economic cost of osteoporotic fractures was estimated at $14 billion dollars in 1995 and will rise with our aging population 1, 2. In fact, it is estimated that by the year 2040, the economic burden of such osteoporosis care will reach $240 billion dollars (2). Men and women suffer a twofold to fivefold increase in mortality after sustaining a hip fracture. Only one-third of those who sustain a hip fracture regain their pre-fracture status (1). Appreciating the enormity of the osteoporosis challenge, it does appear that several conclusions can be drawn from our study: (1) it was relatively easy to implement an IVA protocol into an existing and busy clinical osteoporosis service that already services a high-risk population; (2) using age 65 as a guide for determining IVA is reasonably supported by our PDSA protocol cycle analysis, and it seems to offer the best balance between VF detection and additional cost; (3) there are instances in which IVA VF detection by itself, independent of DXA information or major historical risk factors, determines that prescription osteoporosis therapy is recommended; and (4) the potential health economic osteoporosis savings with this strategy needs further study.

Back to Article Outline

Acknowledgments 

We would like to thank Tanna Culp for her expert assistance in assembling the manuscript.

Back to Article Outline

References 

  1. Lindsay R, Meunier PJ. Osteoporosis: review of the evidence for prevention, diagnosis and treatment and cost effectiveness analysis. The basis for a guideline for the medical management of osteoporosis. Osteoporosis Int. 1998;8(Suppl 4):1–88
  2. Lindsay R. The burden of osteoporosis: cost. Am J Med. 1995;98(Suppl 2A):9S–11S
  3. Ross PD, Davis JW, Epstein RS, Wasnich RD. Pre-existing fractures and bone mass predict vertebral fracture incidence in women. Ann Intern Med. 1991;114:919–923
  4. Cummings SR, Nevitt MC, Browner WS, et al. Risk factors for hip fracture in white women. New Engl J Med. 1995;332:767–773
  5. Ensrud KE, Nevitt ME, Palermo L, et al. What proportion of incident morphometric vertebral fractures are clinically diagnosed and vice versa?. J Bone Miner Res. 1999;14(S1):S138
  6. Genant HK, Li J, Wu CY, Shepherd JA. Vertebral fractures in osteoporosis: a new method for clinical assessment. J Clin Densitom. 2000;3(3):281–290
  7. Greenspan SL, van Stetten E, Emond EK, Jones L, Parker RA. Instant vertebral assessment: a noninvasive dual x-ray absorptiometry technique to avoid misclassification and clinical management of osteoporosis. J Clin Densitom. 2001;4(4):373–380
  8. Rea JA, Blake GM, Sterger P, Genant HK, Fogelman I. Visual assessment of vertebral deformity by x-ray absorptiometry: a highly predictive method to exclude vertebral deformity. Osteoporosos Int. 2000;11:660–668
  9. Ross PD. Osteoporosis: frequency, consequences and risk factors. Arch Med. 1996;156:1399–1411
  10. Lenchik L, Rogers LF, Delmas PD, Genant HK. Diagnosis of osteoporotic vertebral fractures: importance of recognition and description by radiologist. AJR. 2004;183:949–958
  11. Lindsay R, Silverman S, Cooper C. Risk of new vertebral fracture in the year following fracture. JAMA. 2001;285:320–323
  12. Newman ED, Olenginski TP, Perruquet JL, Hummel JH, Indeck C, Wood GC. Using mobile DXA to improve access to osteoporosis care: unit design, program development, implementation and outcomes. J Clin Densitom. 2004;7(1):71–75
  13. Berwick D. Developing and testing changes in delivery of care. Ann Inter Med. 1998;128:651–656
  14. Lindsay R, Meunier PJ. Osteoporosis: review of the evidence for prevention, diagnosis and treatment and cost effectiveness analysis. The basis for a guideline for the medical management of osteoporosis. Osteoporos Int. 1998;8(Suppl 4):1–88
  15. Newman ED, Ayoub WT, Starkey RH, Diehl JM, Wood GC. Osteoporosis disease management in a rural health care population: hip fracture reduction and reduced costs in postmenopausal women after 5 years. Osteoporos Int. 2003;14:146–151
  16. Durosier C, Duboeuf F, Conicella G, et al. Vertebral fracture assessment (VFA): evaluation of a new automatic DXA software for better management of osteoporotic patients. J Clin Densitom. 2004;8(1):S14
  17. Vokes TJ, Dixon LB, Favus MJ. Clinical utility of dual-energy vertebral assessment (DVA). Osteoporosos Int. 2003;14:871–878
  18. Schousboe JT, DeBold CR, Bowles C, Glickstein S, Rubino RK. Prevalence of vertebral compression deformity by x-ray absorptiometry of lateral thoracic and lumbar spines in a population referred for bone densitometry. J Clin Densitom. 2002;5(3):239–246
  19. Bonnick SL. Bone densitometry in clinical practice. 2nd ed.. New Jersey: Humana Press; 2004;39–48
  20. Leib, E. Technical standardization for dual-energy x-ray absorptiometry: the writing group for the ISCD position development conference. J Clin Densitom 7(1):27–36.

PII: S1094-6950(06)00019-9

doi:10.1016/j.jocd.2005.08.002

Journal of Clinical Densitometry
Volume 9, Issue 1 , Pages 72-77, January 2006

Article Tools

* Image Usage & Resolution