Central DXA utilization shifts from office-based to hospital-based settings among medicare beneficiaries in the wake of reimbursement changes

Authors


Abstract

In the United States, Medicare gradually reduced payments for central dual-energy X-ray absorptiometry (DXA) performed at physician offices (or other nonhospital settings) from an average of $139 in 2006 to about $82 in 2007 and 2008 and $72 in 2009. Reimbursement for hospital outpatient DXA service was unchanged. We investigated the utilization of hip and spine (central) DXA in the Medicare population before and after the reduction. We identified individuals from the national 5% random sample of Medicare beneficiaries who were ≥65 years of age and enrolled in Medicare Parts A and B but not in a Medicare Advantage plan from 2002 through 2009. For each calendar year, we calculated the proportion of beneficiaries who submitted claims for DXA, the proportions of DXAs performed in hospitals and in physician offices and the number of physician office-based practices that discontinued or started to provide DXA services. From 2002 to 2006, the proportion of beneficiaries who had at least one central DXA increased from 7.9% to 9.6% at an annual increase of 0.4% and from 2006 to 2009, the annual increase dropped to 0.1%. The number of DXAs performed in physician offices dropped from 1,643,720 (69% of 2,363,500 total DXAs) in 2006 to 1,534,240 (66% of 2,338,240) in 2009. This decline was offset by an increase in the number of DXAs performed in hospitals, which increased from 719,780 (31%) in 2006 to 804,000 (34%) in 2009. Among physician office-based practices, more practices initiated than discontinued DXA service each year from 2002 to 2006. However, the trend was reversed since 2007 such that in 2009, 1876 practices discontinued and only 1394 initiated DXA service. The reduction in DXA reimbursement was associated with a decrease in the number of DXAs performed in physician offices and fewer physician offices that provided DXA services. © 2012 American Society for Bone and Mineral Research.

Introduction

Osteoporosis is associated with significant and rising mortality and morbidity in the United States, and central dual-energy X-ray absorptiometry (DXA) is essential in the diagnosis and management of this disease.1–3 Beginning in 2007, the Centers for Medicare and Medicaid Services (CMS) implemented provisions in the Deficit Reduction Act (DRA) of 2005 and revisions to the Medicare Physician Fee Schedule that resulted in reduced reimbursement for central DXA services performed in physician offices or other nonhospital settings, referred to in this paper as physician offices, from an average of $139 in 2006 to $82 in 2007 and 2008, and then $72 in 2009. In 2006, Medicare covered over 35 million Americans 65 years of age or older4 and an estimated 2.9 million DXAs were performed among Medicare beneficiaries; two-thirds of the DXAs were performed in physician offices, which were differentially affected by these reductions. Experts have voiced concern that these reductions could potentially result in a decline in the number of practices providing DXA services and an increase in the number of undiagnosed and/or untreated individuals with osteoporosis.5 Hayes and colleagues6 surveyed members of the International Society for Clinical Densitometry (ISCD) to assess the impact of the reimbursement cuts on their practices; over a third of the 277 respondents reported that they had reduced the number of DXA tests performed and declined software or hardware upgrades in 2007–2008 compared with 2005–2006. In another study conducted among individuals 50 years of age or older with employer sponsored Medicare supplemental coverage or private insurance under a variety of commercial health plans, the use of central DXA continued to grow from 2007 to 2009 but at a slower rate.7 Although an overall decline in DXA utilization was not observed in the study, the impact of reimbursement cuts on Medicare beneficiaries with fee-for-service coverage, which accounted for a majority of Medicare beneficiaries, was not known because commercial health plans may or may not have adopted a similar change in reimbursement policies, and Medicare supplemental coverage may have lessened the impact of the reimbursement changes.

We hypothesized that, among Medicare beneficiaries, reimbursement cuts beginning in 2007 for central DXAs performed in physician offices would be associated with a reduction both in the volume of central DXAs performed in physician offices and in the number of physician office-based practices that provide DXA services. To evaluate this hypothesis, we examined changes in the utilization of central DXAs among Medicare beneficiaries over time in relation to the implementation of reductions in reimbursement.

Materials and Methods

Data source and study population

We obtained Medicare data from the Chronic Conditions Warehouse (CCW) that included enrollment information and claims for a 5% random sample of beneficiaries from 1999 to 2009. Eligible beneficiaries included those who were ≥65 years of age and who were enrolled in Medicare Parts A and B but not in Medicare Advantage plans. The study was approved by the University of Alabama at Birmingham institutional review board and CMS.

Identification of DXA use and DXA providers

We identified the use of hip and spine DXAs using Health Care Current Procedure Classification System (HCPCS) codes (76075 and 77080) as described previously.7 Other forms of bone mass measurement (eg, peripheral DXA, quantitative computed tomography) were not considered as they are less commonly used and do not directly relate to any U.S. screening or treatment guidelines for osteoporosis. In cases where the technical and professional components of DXA claims were billed separately, DXA claims occurring within 15 days of each other were aggregated as a single DXA test. The provider (either a physician or a hospital system) billing for the technical component of the DXA was identified based on the source of the claim and the HCPCS modifier (–TC, for technical component). We grouped claims submitted by a particular physician office-based practice, or nonfacility provider, using Tax Identification Numbers (IDs) under which DXA claims were billed. If a DXA was performed at a hospital outpatient clinic, or facility provider, the hospital was identified by linking provider codes in the hospital outpatient DXA claims to Provider of Service file records.

Data analysis

From 2002 through 2009, for each calendar year, we identified beneficiaries who had fee-for-service coverage for the entire calendar year and calculated the proportion of beneficiaries who had at least one DXA. Among those who received at least one DXA, we determined the proportions of DXAs (for those with more than one DXAs in 1 year, only the first DXA was included) performed in facility (hospital) and nonfacility (physician office) settings.

We assessed changes over the time period from 2002 to 2009 in the utilization of screening and monitoring DXAs among beneficiaries who had at least 4 years of fee-for-service coverage. We defined “screening” DXAs as DXAs performed in beneficiaries who had not had a central DXA in the previous 3 years. “Monitoring” DXAs were defined as DXAs performed in beneficiaries who had a central DXA in the previous 3 years. We further examined whether the time trend in utilization differed by various beneficiary characteristics, including age, gender, race/ethnicity, income (imputed from the census tract of residence), and geographic location (rural versus urban classified using the Rural–Urban Commuting Area Codes).8 For monitoring DXAs, we examined the proportion of monitoring DXAs that were performed at the same practice, and if performed at different practices, we compared the setting (facility versus nonfacility).

To identify physician office-based practices that ceased to provide DXA services or reduced the volume of DXAs performed over time, we examined the number of practices that billed for at least one DXA in each calendar year from 2002 to 2009. We defined practices that discontinued providing DXA services as those that billed for at least one DXA in the previous year but none in the current year. Practices initiating DXA services were defined as those that had not billed for a DXA in the previous year but billed for at least one in the current year. We calculated the quartiles of the number of DXAs performed annually at each physician office-based practice from 2002 to 2009. For each calendar year, we then used these quartiles to categorize practices that provided DXA services according to volume of tests performed, and we then examined changes in the total numbers of practices, and in those starting and discontinuing DXA services, by volume, over time.

Results

The average age of all Medicare beneficiaries in 2009 was 75 years (standard deviation = 8); 58% were female, 88% were Caucasians, 7% were African Americans, and the remaining 5% reported other races. From 2002 to 2006, the proportion of beneficiaries who had at least one central DXA increased from 7.9% to 9.6% at an annual rate of 0.4% (Fig. 1). The proportion continued to climb after 2006, peaked in 2008 at 9.9% and remained essentially unchanged in 2009 (9.9%). However, the rate at which it increased declined to 0.1% annually from 2006 to 2009. Compared with 2006 (with extrapolation to estimate nationally), in which 1,643,720 DXAs (69% of 2,363,500 total DXAs) were performed in physician offices, by 2009 both the absolute number and the relative proportion performed in physician offices declined to 1,534,240 and 66%, respectively (Fig. 2). This decline was offset by an increase in the number of facility DXAs, which increased from 719,780 (31% of all DXA tests) in 2006 to 804,000 (34% of all DXA tests) in 2009.

Figure 1.

Proportion of beneficiaries who received at least one central DXA by calendar year.

Figure 2.

Proportions of facility and nonfacility central DXA by calendar year.

About 5% of eligible beneficiaries received screening DXAs each year from 2002 to 2009. Although receipt of screening DXA differed by age, gender, race, income, and geographic location (rural versus urban), the trend that the proportion of beneficiaries that received screening DXAs essentially remained unchanged over time was consistently observed in each subgroup of patients (Table 1). From 2002 to 2009, the proportion of beneficiaries who received a “Monitoring” DXA increased continuously (Table 2). Again, although there were significant differences in the proportion of beneficiaries who received a monitoring DXA when stratified by beneficiaries' characteristics, the proportion uniformly increased over time within each subpopulation. Over time, a higher proportion of monitoring DXAs were performed at the same practice. However, among beneficiaries who received their monitoring DXA at different practices, we observed a significant increase among those who had their initial DXAs at physician offices and the monitoring DXA at hospitals from 2006 to 2009. In contrast, the number of beneficiaries in all other groups (received both DXAs at hospitals, both at physician offices, or received initial DXAs at hospitals and monitoring DXAs at physician offices) declined over the same time period (Table 2).

Table 1. Numbers and Proportions of Beneficiaries Who Received a Screening DXA by Calendar Year
 2002 N (%)2003 N (%)2004 N (%)2005 N (%)2006 N (%)2007 N (%)2008 N (%)2009 N (%)
No. eligible786,607773,201775,843776,903754,077725,313696,155678,815
All41,977 (5.3)42,257 (5.5)42,059 (5.4)42,736 (5.5)42,206 (5.6)40,293 (5.6)38,359 (5.5)37,076 (5.5)
Age
 65–695,203 (6.0)5,351 (6.2)5,243 (5.8)5,376 (5.8)5,503 (6.0)5,474 (6.2)5,199 (6.2)4,802 (6.1)
 70–7413,966 (6.2)13,530 (6.2)12,960 (6.1)13,039 (6.1)12,814 (6.3)12,104 (6.2)11,397 (6.0)11,420 (6.1)
 75–7911,991 (5.9)11,818 (5.9)11,832 (6.0)11,951 (6.2)11,342 (6.1)10,880 (6.2)10,104 (6.2)9,433 (6.1)
 80–847,319 (5.0)7,611 (5.2)7,800 (5.3)8,077 (5.4)8,098 (5.5)7,442 (5.4)7,202 (5.4)7,071 (5.5)
 85+3,498 (2.8)3,947 (3.2)4,224 (3.3)4,293 (3.4)4,449 (3.5)4,393 (3.4)4,457 (3.0)4,350 (3.4)
Gender
 Male4,458 (1.2)5,194 (1.4)5,719 (1.5)6,382 (1.7)6,351 (1.7)6,165 (1.7)6,032 (1.7)6,026 (1.7)
 Female37,519 (8.8)37,063 (9.0)36,340 (9.0)36,354 (9.2)35,855 (9.5)34,128 (9.5)32,327 (9.5)31,050 (9.5)
Race
 African American1,925 (3.1)2,173 (3.5)2,291 (3.6)2,438 (3.9)2,412 (4.1)2,253 (4.2)2,204 (4.)2,167 (4.4)
 Caucasian38,688 (5.6)38,615 (5.7)38,106 (5.6)38,598 (5.6)38,112 (5.7)36,454 (5.7)34,587 (5.6)33,289 (5.6)
 Other1,364 (4.6)1,469 (4.9)1,662 (5.5)1,700 (5.5)1,682 (5.6)1,586 (5.4)1,568 (5.2)1,620 (5.4)
Annual Income, $
 <29,9997,773 (4.4)8,112 (4.7)7,836 (4.7)7,843 (4.8)7,508 (4.9)7,035 (4.)6,425 (4.8)6,239 (4.8)
 30,000–44,99916,087 (5.2)16,182 (5.3)15,961 (5.2)15,944 (5.3)15,793 (5.4)14,703 (5.3)13,945 (5.3)13,473 (5.3)
 45,000–59,9999,592 (5.9)9,372 (5.8)9,536 (5.8)9,692 (5.8)9,587 (5.9)9,277 (5.8)9,028 (5.9)8,376 (5.6)
 65,000–74,9994,492 (6.0)4,551 (6.1)4,608 (6.0)4,904 (6.1)4,974 (6.2)4,844 (6.2)4,721 (6.1)4,650 (6.1)
 >75,0003,970 (6.6)3,997 (6.6)4,067 (6.5)4,308 (6.6)4,300 (6.5)4,323 (6.5)4,091 (6.2)4,154 (6.3)
Urban versus Rural
 Rural14,575 (4.7)14,953 (4.9)14,571 (4.8)14,730 (4.8)14,821 (5.0)14,128 (5.0)13,530 (4.9)13,240 (4.9)
 Urban27,401 (5.8)27,304 (5.9)27,488 (5.9)28,006 (5.9)27,385 (6.0)26,164 (5.9)24,828 (5.9)23,835 (5.8)
Table 2. Numbers and Proportions of Beneficiaries Who Received a Monitoring Central DXA by Calendar Year
 2002 N (%)2003 N (%)2004 N (%)2005 N (%)2006 N (%)2007 N (%)2008 N (%)2009 N (%)
No. eligible158,832183,214206,214223,825229,883232,902233,807235,662
Total31,397 (19.8)36,567 (20.0)42,392 (20.6)47,896 (21.4)50,866 (22.1)51,538 (22.1)52,959 (22.7)52,738 (22.4)
Age
 65–693,869 (21.3)4511 (21.1)5,459 (21.9)6,247 (22.7)6,650 (23.7)6,886 (23.6)6,966 (24.1)6,857 (24.0)
 70–7410,799 (21.1)12,283 (21.5)13,686 (21.9)15,328 (23.1)15,882 (23.7)16,058 (24.0)16,519 (24.4)16,770 (24.1)
 75–799,349 (20.5)10,786 (20.6)12,409 (21.3)13,704 (22.1)14,578 (23.1)14,525 (23.2)14,489 (23.6)14,174 (23.7)
 80–845,260 (18.4)6,273 (18.5)7,546 (19.4)8,702 (20.0)9,416 (21.0)9,459 (20.8)9,862 (21.3)9,741 (21.0)
 85+2,120 (13.9)2,714 (14.8)3,292 (15.1)3,915 (16.0)4,340 (16.2)4,610 (16.1)5,123 (17.3)5,196 (16.7)
Gender
 Male1,464 (14.4)1,819 (14.2)2,345 (14.8)2,822 (15.1)3,251 (15.6)3,379 (15.3)3,564 (15.5)3,572 (15.4)
 Female29,933 (20.1)34,748 (20.4)40,047 (21.0)45,074 (22.0)4,7615 (22.8)48,159 (22.8)49,395 (23.4)49,166 (23.1)
Race
 African American796 (15.9)1,093 (16.8)1,309 (16.0)1,613 (16.9)1,774 (17.8)1,753 (17.4)1,934 (18.8)2,040 (18.8)
 Caucasian29,621 (19.9)34,354 (20.1)39,648 (20.7)44,606 (21.6)47,317 (22.3)47,968 (22.4)49,122 (22.9)48,668 (22.6)
 Other980 (20.8)1,120 (19.8)1,435 (21.4)1,677 (21.9)1,775 (22.2)1,817 (21.5)1,903 (21.4)2,030 (22.1)
Annual Income, $
 <29,99911,129 (18.9)12,975 (19.1)15,005 (19.7)16,810 (20.6)17,602 (21.3)17,441 (21.0)18,162 (22.1)17,615 (21.4)
 30,000–44,9997,425 (20.2)8,694 (20.4)10,097 (21.1)11,349 (21.6)12,292 (22.8)12,508 (22.7)12,645 (22.9)12,881 (22.9)
 45,000–59,9993,886 (21.1)4,644 (21.7)5,388 (22.1)6,121 (22.6)6,786 (23.6)6,988 (23.5)7,212 (23.7)7,232 (23.3)
 65,000–74,9994,938 (18.1)5,664 (18.1)6,585 (18.8)7,392 (19.8)7,570 (20.2)7,494 (20.4)7,414 (20.4)7,400 (20.6)
 >75,0003,979 (23.0)4,549 (23.1)5,257 (23.6)6,160 (24.6)6,558 (24.6)6,925 (25.0)7,268 (25.4)7,338 (25.2)
Urban versus rural
 Rural8,898 (17.7)10,727 (18.1)12,610 (18.8)14,506 (19.9)15,356 (20.4)15,543 (20.4)15,959 (20.8)16,265 (20.9)
 Urban22,499 (20.7)25,840 (20.8)29,781 (21.4)33,390 (22.1)35,509 (22.9)35,995 (23.0)36,999 (23.5)36,470 (23.1)
Performed at the same practice
 Yes20,653 (65.8)24,451 (66.9)28,797 (67.9)32,691 (68.3)35,188 (69.2)36,546 (70.9)37,853 (71.5)37,870 (71.8)
 Both at facility5,493 (26.6)6,741 (27.6)7,801 (27.1)8,982 (27.5)9,532 (27.1)10,016 (27.4)10,858 (28.7)11,243 (29.7)
 Both at nonfacility15,160 (73.4)17,710 (72.4)20,996 (72.9)23,709 (72.5)25,656 (72.9)26,530 (72.6)26,995 (71.3)26,627 (70.3)
 No10,742 (34.2)12,111 (33.1)13,595 (32.1)15,204 (31.7)15,677 (30.8)14,989 (29.1)15,105 (28.5)14,867 (28.2)
 From facility to nonfacility2,216 (20.6)2,630 (21.7)2,861 (21.0)3,275 (21.5)3,313 (21.1)3,083 (20.6)2,725 (18.0)2,461 (16.6)
 From nonfacility to facility2,164 (20.1)2,508 (20.7)2,592 (19.1)2,739 (18.0)2,912 (18.6)3,068 (20.5)3,501 (23.2)3,926 (26.4)
 Both at facility752 (7.00)765 (6.32)983 (7.23)1,126 (7.41)1,255 (8.01)1,078 (7.19)1,019 (6.75)1,083 (7.28)
 Both at nonfacility5,610 (52.2)6,208 (51.3)7,159 (52.7)8,064 (53.0)8,197 (52.3)7,760 (51.8)7,860 (52.0)7,397 (49.8)

The total number of physician office based practices that provided DXA services increased from 6800 in 2002 to 9486 in 2006 (peak year) and then declined to 8547 in 2009. This decline from 2006 to 2009 resulted from a continued decrease in the number of physician office based practices that started to offer DXA services each year from 2082 in 2006 to 1394 in 2009, and a moderate increase in the number of practices that discontinued providing DXA services. We summarized the net effect of these two trends by calculating the difference between initiating and discontinuing practices each year (Fig. 3). Before 2007, more practices started than discontinued DXA services and since 2007, the pattern has reversed.

Figure 3.

The difference between the numbers of physician office-based practices initiating and discontinuing DXA services by calendar year.

When we stratified practices by the number of DXAs performed, we found that regardless of the volume of DXAs, more practices discontinued providing DXA services from 2007 to 2009 compared with 2002 to 2006 (Table 3). Indeed, 2007 was the first year in which more physician office-based practices discontinued than started providing DXA services. The number of practices that discontinued DXA services peaked in 2008 for those performing 80 DXAs or less annually. However, for those that performed more than 80 DXAs a year, the number of discontinued practices was greatest in 2009 (Table 3). From 2008 and 2009, among practices that performed on average ≥220 DXAs a year among Medicare beneficiaries who had fee-for-service coverage, the number of practices discontinuing DXA services increased by 50%. The number of practices starting to provide DXA services declined from 2007 to 2009 across practices of all sizes; the greatest decline (50% decrease from 2006–2009) occurred among those performing more than 80 DXAs annually (Table 3).

Table 3. Temporal Trends of Physician Office-Based Practices That Provided DXA Services From 2002 to 2009 by Volume of DXAs Performed Annually
 Calendar Year
20022003200420052006200720082009
  • a

    Based upon data from the random 5% sample.

  • b

    Compared with the prior year: new practices were defined as those that did not perform a DXA in the previous year but in the current year; discontinued practices as those performed at least one DXA in the previous year but not in the current year.

Total No. of Practices6,8007,5898,5508,9909,4969,4659,0298,547
 No. DXAs annuallya
  ≤402,087 (30.7)2,332 (30.7)2,760 (32.3)2,823 (31.4)3,183 (33.5)3,238 (34.2)3,061 (33.9)2,900 (33.9)
  >40 and ≤801,073 (15.8)1,279 (16.9)1,466 (17.1)1,585 (17.6)1,647 (17.3)1,780 (18.8)1,674 (18.5)1,516 (17.7)
  >80 and ≤2201,863 (27.4)2,045 (26.9)2,339 (27.4)2,523 (28.1)2,617 (27.6)2,498 (26.4)2,356 (26.1)2,302 (26.9)
  >2201,777 (26.1)1,933 (25.5)1,985 (23.2)2,059 (22.9)2,049 (21.6)1,949 (20.6)1,938 (21.5)1,829 (21.4)
No. of Practices Initiating DXA Service,b %1,732 (25.5)2,041 (26.9)2,193 (25.6)2,082 (23.2)2,082 (21.9)1,741 (18.4)1,573 (17.4)1,394 (16.3)
 No. DXAs annuallya
  ≤401,028 (59.4)1,183 (58.0)1,322 (60.3)1,280 (61.5)1,340 (64.4)1,207 (69.3)1,082 (68.8)1,009 (72.4)
  >40 and ≤80284 (16.4)384 (18.8)393 (17.9)374 (18.0)362 (17.4)306 (17.6)277 (17.6)217 (15.6)
  >80 and ≤220322 (18.6)353 (17.3)378 (17.2)325 (15.6)303 (14.6)198 (11.4)174 (11.1)135 (9.7)
  >22098 (5.7)121 (5.9)100 (4.6)103 (4.9)77 (3.7)30 (1.7)40 (2.5)33(2.4)
No. of Practices Discontinuing DXA Service,b %1,207 (17.8)1,252 (16.5)1,232 (14.4)1,642 (18.3)1,576 (16.6)1,772 (18.7)2,009 (22.3)1,876 (21.9)
 No. DXAs in prior yeara
  ≤40884 (73.2)887 (70.8)903 (73.3)1,202 (73.2)1,109 (70.4)1,259 (71.0)1,422 (70.8)1,312 (69.9)
  >40 and ≤80168 (13.9)200 (16.0)173 (14.0)238 (14.5)269 (17.1)305 (17.2)364 (18.1)301(16.0)
  >80 and ≤220119 (9.9)135 (10.8)117 (9.5)159 (9.7)156 (9.9)178 (10.0)181 (9.0)200 (10.7)
  >22036 (3.0)30 (2.4)39 (3.2)43 (2.6)42 (2.7)30 (1.7)42 (2.1)63 (3.4)

Discussion

We found that despite a series of scheduled cuts in the Medicare reimbursement for central DXA services provided in nonfacility settings beginning in 2007, the annual proportion of Medicare beneficiaries who received a central DXA continued to climb during 2007 and 2008 and did not change from 2008 to 2009. However, the rate of increase in DXA utilization slowed significantly from 0.4% before to 0.1% after the reductions were enacted. We also found a continued decrease in the total number of DXAs performed in physician office-based practices, which paralleled a reduction in the number of physician office practices that provided DXA services.

The finding that the growth in the utilization of DXAs stalled has significant implications. Although many osteoporotic fractures may be prevented through early diagnosis, lifestyle changes, and treatment, screening and management of osteoporosis are inadequate in the United States. In the 2004 U.S. Surgeons General's report on bone health and osteoporosis, osteoporosis was identified as a major public health issue that is underdiagnosed and undertreated.9 From 1999 to 2005, among Medicare beneficiaries, only 30% of eligible women 65 and older had a bone density test.10 In another study conducted among women 50 to 85 and men 65 to 85 years of age using data from a health maintenance organization, less than half of those who sustained a fragility fracture received treatment.11 Central DXAs are the most commonly used tool for the diagnosis of osteoporosis and have been shown to be associated with initiating treatment and lifestyle changes.12–14 Therefore, maintaining the status quo is not an optimal target, and our finding suggests that the slowing increase in the use of DXAs may negatively affect quality of care.

Although the total number of DXAs performed each year continued to climb, the number of DXAs performed at a nonfacility setting (ie, in physician offices) declined from 2007 to 2009. Our findings suggest that this reduction may have resulted from a decrease in the number of physician office-based practices that started providing DXA services and from an increase in the number of practices that discontinued providing DXA services over time, thus creating a negative balance that kept enlarging (Fig. 3). In contrast to the finding reported by Hayes and colleagues6 that practices with fewer DXA scanners experienced a greater amount of decline in the provision of DXA services, we found that reimbursement cuts were associated with discontinuation of DXA services even among practices that provided large volume of DXAs.

The observed decrease in the number of practices providing DXA service raises the concern that certain subpopulations, such as those residing in areas without easy access to hospital-based DXA may be more affected than others. With fewer facilities providing DXA services, some beneficiaries may have to travel longer distance for DXA services, and travel distance has previously been show to be associated with the utilization of DXA services.15 Reduction in the number of facilities providing DXA services may also lead patients to undergo DXA at different locations over time, creating a discontinuity that adversely affect quality of care because DXAs performed on different machines are difficult to directly compare with one another. To address this concern, we examined the proportion of beneficiaries who had their previous DXAs performed at a physician office-based practice and had their next DXA performed at a hospital out of all beneficiaries who had their second DXAs performed at a different facility. Our finding confirmed that the proportion who had their first DXA in a physician office and second DXA in a hospital increased from 18.6% in 2006 to 26.4% in 2009 (Table 2).

The results regarding the number of physician office-based practices that discontinued and started providing DXA services need to be interpreted carefully. Because data from a random 5% sample were used in this analysis, for a practice to be identified as a DXA provider for a single year, we required at least one beneficiary in the Medicare random 5% sample to have received a DXA at the practice to identify that practice. Some practices may not have been included in the sample, especially those that performed fewer than 20 DXAs per year in the Medicare fee-for-service population. However, because we examined the trend over time using the same method to identify the practices, this issue does not affect the validity of the trend we observed. In addition, because small practices were more likely to be omitted from our data, the increased number of practices that discontinued DXA services may reflect a reduction in the number of DXAs performed at these practices over time. This is consistent with results from a survey of physician office-based practices in which over a third of the practice reported that they have reduced the number of DXAs provided.6

We acknowledge several other limitations of the study. We defined screening DXAs as DXAs performed among beneficiaries who had not had a DXA in the previous 3 years, whereas the optimal interval between two screening DXAs has been suggested to vary based on the T-score from the first screening test.16 Our results may not apply to those enrolled in managed care, which accounted for approximately 24% of total Medicare enrollees in 2009,17 because their claims are not represented within fee-for-service Medicare data.

In summary, we found that despite a series of reductions in payment for DXAs performed in nonfacility settings, the overall proportion of beneficiaries who received a central DXA each year continued to increase through 2008 and remained unchanged in 2009. However, from 2006 to 2009, the annual rate of increase declined and DXAs were more likely to be performed in hospitals and not in physician offices compared with from 2002 to 2006. We also observed a trend that began in 2007 whereby more practice groups discontinued providing DXA services and fewer started providing these services. The long-term impact of the changes in the reimbursement of central DXAs on quality of osteoporosis care for at-risk patients remains unclear and has yet to be assessed with additional years of data.

Disclosures

JZ was supported by Grant Number T32HS013852 from the Agency for Healthcare Research and Quality (AHRQ). JRC receives support from the NIH (AR053351) and AHRQ (R01HS018517). JZ receives salary support from Amgen. ED receives research support from Amgen. HZ receives salary support from Amgen. AJL receives consulting fees or other remuneration from Amgen, Eli Lilly, Novartis, Roche/Genentech, HGS/GSK. KGS receives consulting fees or other remuneration and research support from Eli Lilly, Novartis, Merck, and Amgen. MLK receives research funding from Amgen. MAM receives research funding from Amgen. NCW receives salary support from Amgen. HY receives salary support from Amgen. JRC is a consultant and has received research support from Amgen, Centocor, CORRONA, Pfizer, Roche/Genentech, Abbott, and UCB.

Acknowledgements

This research was supported by a contract between UAB and Amgen, Inc. The analysis, presentation, and interpretation of the results are solely the responsibility of the authors. Only the authors from UAB had access to the Medicare data used.

Authors' roles: Study design: JZ, ED, JRC. Study conduct: JZ, ED, JRC. Data collection: ED, KGS, KLK, MAM, JRC. Data analysis: JZ, ED, HZ, JRC. Data interpretation: JZ, ED, HZ, JRC. Drafting manuscript: JZ, HZ. Revising manuscript content: JZ, ED, AJL, KGS, KLK, MAM, NCW, HY, JRC. Approving final version of manuscript: JZ, ED, HZ, AJL, KGS, KLK, MAM, NCW, HY, JRC. JZ takes responsibility for the integrity of the data analysis.

Ancillary