PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
JAMA Intern Med. Author manuscript; available in PMC 2014 January 14.
Published in final edited form as:
PMCID: PMC3547540
NIHMSID: NIHMS432735

Unintended consequences of eliminating Medicare payments for consultations1

Zirui Song, Ph.D.,1,2,3,4 John Z. Ayanian, M.D., M.P.P.,1,5 Jacob Wallace, B.A.,4 Yulei He, Ph.D., M.S.,1 Teresa B. Gibson, Ph.D.,1,6 and Michael E. Chernew, Ph.D.1,2

Abstract

Background

Prior to 2010, Medicare payments for consultations (commonly billed by specialists) were substantially higher than for office visits of similar complexity (commonly billed by primary care physicians). In January 2010, Medicare eliminated consultation payments from the Part B Physician Fee Schedule and increased fees for office visits. This change was intended to be budget neutral and to decrease payments to specialists while increasing payments to primary care physicians. We assessed the impact of this policy on spending, volume, and complexity for outpatient office encounters in 2010.

Methods

We examined 2007–2010 outpatient claims for 2,247,810 Medicare beneficiaries with Medicare Supplemental (Medigap) coverage through large employers in the Thomson Reuters MarketScan Database. We used segmented regression analysis to study changes in spending, volume, and complexity of office encounters adjusted for age, sex, health status, secular trends, seasonality, and hospital referral region.

Results

“New” office visits largely replaced consultations in 2010. An average of $10.20 (6.5 percent) more was spent per beneficiary per quarter on physician encounters after the policy. The total volume of physician encounters did not change significantly. The increase in spending was largely explained by higher office visit fees from the policy and a shift toward higher complexity visits to both specialists and primary care physicians.

Conclusions

The elimination of consultations led to a net increase in spending on visits to both primary care physicians and specialists. Higher prices, partially due to the subjectivity of codes in the physician fee schedule, explained the spending increase, rather than higher volumes.

INTRODUCTION

In an era of cost control, physician fees are under heavy scrutiny.15 The Medicare Physician Fee Schedule, which lists the fees that Medicare pays for each physician service under Part B of the program, contains large payment differences between primary care and specialty services. Stemming from the resource-based relative value system, these fee differences have been linked to a substantial income gap between primary care physicians (PCPs) and specialists and to the shortage of medical students entering primary care.612 This has motivated efforts to reform physician payment towards increasing fees for primary care.

On January 1, 2010, the Centers for Medicare and Medicaid Services (CMS) instituted a targeted policy to redistribute Medicare payments from specialists to PCPs in a budget-neutral manner. Specifically, CMS eliminated payments for consultations from the Medicare Physician Fee Schedule, which were billed more frequently by specialists, and raised fees for office visits, which were billed more frequently by PCPs.13

Similar to billing for consultations, physicians bill for office visits (“evaluation and management”) using a set of codes that contains 5 levels of clinical complexity. This is true for both “new” patient office visits and “established” patient office visits. Consultations had long commanded higher payments than office visits at each level of clinical complexity. In 2009, for example, Medicare paid $124.79 on average for a consultation of medium complexity, compared to $91.97 for a new patient office visit and $61.31 for an established patient office visit of similar complexity.

In place of consultations, the 2010 policy required physicians to bill all encounters as office visits. While fees for office visits were simultaneously increased, they remained well below prior consultation fees, with new office visits 16–26 percent lower and established office visits 42–61 percent lower than their consultation counterparts from 2009 (e-Table 1). The increase in office visit fees averaged approximately 5–6 percent (e-Table 2).

Table 1
Characteristics of the Study Population.*
Table 2
Average spending, volume, and coded complexity of outpatient physician encounters before and after 2010.*

CMS designed the policy to be budget neutral, implying a transfer of Medicare dollars from specialists to PCPs.13 Budget neutrality was based on an assumption that physicians would bill for office visits of equal complexity as they had done for prior consultations.14 We evaluated the effects of this policy on spending, volume, and intensity of coding for office encounters in the first year of implementation. We hypothesized that cutting physician fees may have unintended consequences for spending on physician encounters through either increased volume or intensity of coding.

METHODS

Study Population

Our population consisted of Medicare beneficiaries drawn from the 2007–2010 Thomson Reuters MarketScan Medicare Supplemental and Coordination of Benefits claims Database. This database consists of a large convenience sample of U.S. Medicare beneficiaries with Medicare Supplemental (Medigap) coverage through large employers. For these beneficiaries, Medicare is the primary payer and all of their Medicare claims are reported in the data.15

We used a database containing 2.9 million beneficiaries with any diagnosis related to diabetes or cardiovascular conditions or prescriptions for any cardiovascular or cholesterol-reducing drugs (e-Methods). From this population, we excluded 326,491 beneficiaries who were not enrolled for at least 1 calendar year. In addition, we excluded 311,271 beneficiaries who were enrolled in health maintenance organization or other plans paid through capitation. Physicians for beneficiaries in capitated plans may not be directly affected by the fee schedule changes. Our final analytic sample comprised 2.2 million beneficiaries who were enrolled for at least 1 year. In sensitivity analyses, we further restricted the sample to 798,262 beneficiaries who were continuously enrolled for all 4 years in the data. We compared characteristics of the population before and after the intervention using t-tests for continuous variables and the Pearson Chi-squared test for categorical variables.

Study Design

We used an interrupted time series (segmented regression) design to estimate changes in spending, volume, and intensity of coding associated with the Medicare policy for all 3 types of outpatient encounters (consultations, new patient office visits, and established patient office visits). The pre-intervention period was 2007–2009. The post-intervention period was 2010.

In the Medicare Physician Fee Schedule, determination of the clinical complexity of a patient-physician encounter is somewhat subjective. For example, a level 1 new patient office visit is characterized by a “problem-focused” history and physical examination involving “straightforward” medical decision-making, typically requiring 10 minutes. In contrast, a level 5 new patient office visit entails a “comprehensive” history and examination involving decision making of “high complexity,” typically requiring 60 minutes. These rough time guidelines are not binding. Any given level of complexity could be appropriate if certain elements of a visit (e.g., comprehensive history) are met (e-Table 3). We examined changes in the coding of complexity associated with the policy, adjusting for health status, baseline trends, and other covariates.

We also assessed the policy’s effect by specialty category. Specifically, we assigned claims to the PCP category if the physician’s specialty was internal medicine, family medicine, geriatric medicine, or preventive medicine. These physicians accounted for about half of the claims. The remaining half was assigned to the specialist category, with a small fraction of non-physicians and facility providers excluded. Our analyses used all outpatient claims and included both professional and facility fees for all encounters.

Variables

Spending (in 2010 U.S. dollars) was computed from claims-level total payments, including employer and beneficiary cost-sharing. Volume was the number of encounters. Each encounter was a unique instance of a consultation or office code for a unique beneficiary on a unique day. Complexity was defined as the encounter’s level of complexity.

Potential confounding variables included age, sex, health risk score, secular trends, indicators for quarter of the year (to account for seasonality), and indicators for the beneficiary’s hospital referral region (HRR). We calculated health risk scores using the CMS hierarchical condition categories (CMS-HCC) model, which is used for formal risk adjustment of payments to Medicare Advantage plans.16 This model generates a risk score based on contemporaneous age, sex, and clinical diagnoses. Adjusting for risk helps reduce bias in the non-randomly drawn sample of Medicare beneficiaries. HRRs are hospital markets constructed based on where patients receive care and are commonly used in the study of geographic variations in spending and utilization.1719 Since cost-sharing also affects utilization of services, we included each beneficiary’s cost-sharing as a percentage of total spending in our model as a sensitivity analysis. Cost-sharing is a variable unique to the MarketScan Medicare claims database and allows us to test for the effect of patient-side financial incentives. This is an advantage over the traditional Medicare Part B claims data.

Statistical Analyses

All statistical analyses were conducted at the beneficiary-quarter level. We used a multivariate interrupted time series (segmented regression) model. Independent variables included age categories, interactions between age categories and sex, CMS-HCC risk score, indicators for quarter and HRR, secular trend, indicator for year 2009, and indicator for year 2010. The indicator for year 2010 produced our estimate of the policy effect. We added an indicator for year 2009 because there was a small legislative payment increase in the 2009 Medicare Physician Fee Schedule that affected all encounters. The coefficient on the year 2010 indicator reflects the change in the average level of the outcome after the policy took effect, controlling for pre-intervention trends. Standard errors were clustered at the HRR level, which produces more conservative standard errors than individual-level models.20,21

We conducted a number of sensitivity analyses, including alterations to the statistical model (such as omitting indicators for HRR or quarter of the year) and restricting the sample to continuous 4-year enrollees (e-Table 4). All analyses were conducted using STATA software, version 11 (Statacorp; College Station, TX). Results are reported with two-tailed P values. The Harvard Medical School Office for Research Subject Protection approved the study protocol.

RESULTS

Study Population

There were 2,247,810 unique Medicare beneficiaries who were continuously enrolled for at least one year during the study period. Table 1 describes the age distribution, gender, CMS-HCC risk scores, and region of residence for the population before and after the 2010 policy.

Spending

In the first year, the policy eliminated an average of $18.52 per beneficiary per quarter (95% confidence interval [CI], −$19.48 to −$17.55; p<0.001) in consultation spending. At the same time, spending increased by $13.64 (95% CI, $12.89 to $14.40; p<0.001) on new patient office visits (131 percent) plus $15.08 (95% CI, $13.56 to $16.60; p<0.001) on established patient office visits (12 percent) per quarter for each beneficiary. Established visits comprised the majority of physician encounters. On net, spending on all physician encounters was higher by $10.20 per beneficiary per quarter after the policy (6.5 percent) controlling for pre-existing trends (95% CI, $8.58 to $11.82, p<0.001) (Table 2).

Figures 1A and 1B show unadjusted spending by specialty. Prior to the elimination of consultations, average spending on specialist consultations was 3-fold higher than on PCP consultations. As consultations were converted to new office visits, spending on new office visits to specialists saw a larger increase than that for PCPs in 2010 (Figure 1A). Spending on established office visits increased for both PCPs and specialists after the policy (Figure 1B), driving an increase in overall spending for both groups. Our statistical model showed that 58 percent of the increase in spending was attributable to PCP encounters and 42 percent to specialist encounters (p<0.001).

Figure 1Figure 1
Spending on outpatient physician encounters*

Volume

After Medicare’s policy took effect, the volume of consultations decreased by 11.4 per 100 beneficiaries per quarter (p<0.001) (essentially 100 percent). At the same time, the volume of new office visits rose by 8.7 per 100 beneficiaries per quarter (p<0.001) (a 92 percent increase), substituting for the majority of consultations that were eliminated (Table 2). This was consistent with unadjusted trends (Figure 2). The volume of established office visits increased by a statistically insignificant 1.3 per 100 beneficiaries per quarter (p=0.19) (1 percent). On net, there was no significant change in the total volume of encounters (−1.4 per 100 beneficiaries per quarter, p=0.18) (Table 2). Subgroup findings for PCPs and specialists were consistent with overall results.

Figure 2
Volume of outpatient physician encounters*

Complexity

Figure 3 displays the discrete jump in coded complexity of encounters before and after January 2010. Because consultations were largely replaced by new patient office visits (Figure 2), these two types of encounters were combined in Figure 3. In adjusted analyses, our statistical estimates show a discontinuous rise of 0.03 (p<0.001) in the average level of complexity for consultations and office visits (1–5 scale) in 2010. We also found an increase of 0.01 (p<0.001) in the average complexity of established office visits. On net, the magnitude of this coding effect for all encounters was 0.01 (p<0.001), driven by the larger volume of established visits. These estimates control for the underlying rise in the secular trend of complexity over this 4-year period (Figure 3). Similar results were found among the continuously enrolled subsample.

Figure 3
Average coded complexity of encounters*

Converting this coding effect to dollars, our model suggests that coding accounted for about one-third of the overall increase in spending. Given no significant changes in volume, our results suggest that the other two-thirds of the overall spending effect was attributable to the fee increase for each of the office visit codes embedded in the 2010 policy.

DISCUSSION

Medicare’s elimination of consultations was associated with a 6.5 percent increase in overall spending for outpatient encounters in 2010. This increased spending was explained by higher fees paid for office visits and by increased intensity of coding. Our results suggest that the policy did not achieve its goal of budget neutrality in the first year. However, it did appear to narrow the gap in Medicare payments for office encounters between PCPs and specialists. Of the overall increase in spending, PCPs accounted for a greater proportion (58% vs. 42% for specialists). Thus, since specialists received more income from consultations than PCPs did before 2010, the relative income gap was effectively narrowed. This gap applies only to income from office encounters. Whether the gap in total Medicare reimbursements between PCPs and specialists changed depends on all other physician services. The overall discrepancy in Medicare payments derives largely from procedural services, which this policy does not address.

In contrast to previous studies of Medicare fee cuts,2226 we did not observe a volume increase by specialists in response to the fee cut. There are several potential explanations for this finding. First, relative to their fees for procedures and other services, office visit fees for specialists may be low enough that extra volume is not profitable. This explanation may be more likely for procedural specialists or surgeons as opposed to cognitive specialists whose portfolio is dominated by office-based care. Any pressure to increase volume may have been directed to other, perhaps more profitable, services. We were unable to systematically test for such potential “spillover” effects given other policies that were simultaneously in place in 2010, such as Medicare fee cuts in certain radiology and cardiology imaging services, which were directly impacting utilization in those service areas.

Second, volume may have remained unchanged because coding for higher complexity visits allowed physicians specialists to make up for about one-third of the fee cut through changes in coding. Importantly, this change in coding complexity need not reflect coding at a higher than appropriate level. Given the variety of definitions governing the complexity of encounters (e-Table 3), a physician who codes based on time may justifiably determine that in place of a level 3 consultation (40 minutes), an appropriate substitute is a level 4 new patient office visit (45 minutes) rather than a similar level 3 new patient office visit (30 minutes).

Third, volume may not have responded because most physicians are already at capacity with office visits. Thus, it would be difficult to increase the number of visits without working longer hours or shortening the length of visits, and shorter visits would earn a lower fee

These explanations highlight the importance of the interpretation of coding definitions in the Medicare Physician Fee Schedule, which are also widely used by commercial insurers. Policies that aim to achieve savings by targeting only the price (fees) component of health care spending may be susceptible to coding changes that (at least partially) offset its intended effects. Increased coding intensity can be difficult to separate from inappropriate “upcoding” even after adjusting for health risk scores, due to the flexibility in code definitions.

Our study has several potential limitations. First, the population contains only Medicare beneficiaries who have employer-sponsored supplemental (Medigap) coverage. Thus, results may not generalize to Medicare beneficiaries without such coverage. Second, the MarketScan Medicare Supplemental population is derived mostly from large employers. Former employees of small firms and formerly self-employed beneficiaries are not represented. However, our comparisons with the Medicare Expenditure Panel Survey Medicare Supplemental population yielded similar age, sex, geography, and utilization distributions. Additionally, our sample was not randomly selected from the MarketScan database; inclusion criteria captured about half of available beneficiaries (e-Methods). However, all analyses were adjusted for risk using the nationally representative CMS-HCC risk score system. Our results were also robust to sensitivity analyses within each age subcategory (not shown).

The policy’s effects in the first year may not generalize to the long-run. For example, the increase in office visits fees could have encouraged PCPs to see more complex patients and refer them less often to specialists. In addition, the relatively high volume of “new” office visits to specialists may decrease as patients make repeat visits and become “established.”

Our evaluation of Medicare’s elimination of consultations offers potential lessons for policymakers. First, the volume effects associated with fee cuts will depend on the nature of the service. Office visits may be less subject to volume changes due to capacity issues. In contrast, the recent ten-year proposal of the Medicare Payment Advisory Commission to cut specialist fees while freezing PCP fees, which would affect all physician services, would more likely be associated with volume changes.27

Second, the inherent flexibility and subjectivity of code definitions could lead to potentially undesirable coding behavior in response to fee-based policies, as numerous areas in the physician fee schedule feature a gradient of service intensities captured by a set of closely related codes.

Supplementary Material

Supplementary Appendix

Acknowledgments

Supported by The Commonwealth Fund (Dr. Chernew, Principal Investigator). Dr. Song is supported by a Pre-doctoral M.D./Ph.D. National Research Service Award from the National Institute on Aging (F30-AG039175) and a Pre-doctoral Fellowship in Aging and Health Economics from the National Bureau of Economic Research (T32-AG000186). We thank Joseph Newhouse, Thomas McGuire, Michael McWilliams, and Michael McKellar for helpful comments and guidance. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute on Aging or the NIH.

None of the funders were involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, and approval of the manuscript. Zirui Song, Teresa Gibson, and Michael Chernew had full access to the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Footnotes

1Zirui Song, Jacob Wallace, and Yulei He have no potential conflicts of interest to report. John Z. Ayanian is a consultant to RTI International on the development of statistical risk adjustment models for CMS to adjust payments to Medicare Advantage plans. Teresa B. Gibson is Director of Health Outcomes at Thomson Reuters, the distributor of data used for this study. Michael E. Chernew is a commissioner on the Medicare Payment Advisory Commission and a health advisor to the Congressional Budget Office. He also discloses membership, consultancy, and grants from various public and private entities. ICMJE forms are available for all authors.

Disclosures

Zirui Song, Jacob Wallace, and Yulei He have no potential conflicts of interest to report. John Z. Ayanian is a consultant to RTI International on the development of statistical risk adjustment models for CMS to adjust payments to Medicare Advantage plans. Teresa B. Gibson is Director of Health Outcomes at Thomson Reuters, the distributor of data used for this study. Michael E. Chernew is Vice Chairman of the Medicare Payment Advisory Commission and a health advisor to the Congressional Budget Office. He also discloses membership, consultancy, and grants from various public and private entities. ICMJE forms are available for all authors.

References

1. Iglehart JK. Medicare payment reform — proposals for paying for an SGR repeal. N Engl J Med. 2011;365:1859–1861. [PubMed]
2. Aaron HJ. The central question for health policy in deficit reduction. N Engl J Med. 2011;365:1655–1657. [PubMed]
3. Ginsburg PB, White C. Health care’s role in deficit reduction — Guiding Principles. N Engl J Med. 2011;365:1559–1561. [PubMed]
4. Chernew ME, Baicker K, Hsu J. The specter of financial armageddon--health care and federal debt in the United States. N Engl J Med. 2010;362:1166–8. [PubMed]
5. Congressional Budget Office. [Accessed 18 Dec 2011];Changes in Payment to Physicians. 2011 Jun; ( http://www.cbo.gov/doc.cfm?index=12240)
6. Centers for Medicare and Medicaid Services. Physician Fee Schedule: PFS Relative Value Files. ( https://www.cms.gov/PhysicianFeeSched/PFSRVF/)
7. Hsiao WC, Braun P, Dunn D, Becker ER. Resource-based relative values. An overview. JAMA. 1988 Oct 28;260(16):2347–53. [PubMed]
8. Newhouse JP. Pricing the Priceless: A Health Care Conundrum. Massachusetts Institute of Technology Press; Cambridge: 2002.
9. Leigh JP, Tancredi D, Jerant A, Kravitz RL. Physician wages across specialties: informing the physician reimbursement debate. Arch Intern Med. 2010 Oct 25;170(19):1728–34. [PubMed]
10. Lasser KE, Woolhandler S, Himmelstein DU. Sources of U.S. physician income: the contribution of government payments to the specialist-generalist income gap. J Gen Intern Med. 2008;23(9):1477–1481. [PMC free article] [PubMed]
11. Bodenheimer T, Berenson RA, Rudolf P. The primary care-specialty income gap: why it matters. Ann Intern Med. 2007;146(4):301–306. [PubMed]
12. Salsberg E, Rockey PH, Rivers KL, Brotherton SE, Jackson GR. US residency training before and after the 1997 Balanced Budget Act. JAMA. 2008 Sep 10;300(10):1174–80. [PubMed]
13. Centers for Medicare and Medicaid Services, Department of Health and Human Services. Medicare Program; Payment Policies Under the Physician Fee Schedule and Other Revisions to Part B for CY 2010. Federal Register. 2009 Nov 25;74(226):61737–62206. [PubMed]
14. Centers for Medicare and Medicaid Services. CMS Manual System, Pub 100–04 Medicare Claims Processing. Transmittal. 2009 Dec 14;1875
15. Hansen LG, Chang S. Health Research Data for the Read World: the Thomson Reuters MarketScan Databases. Thomson Reuters White Paper. 2010 Jan; ( http://MarketScan.thomsonreuters.com/MarketScanportal/)
16. Pope GC, Kautter J, Ellis RP, Ash AS, Ayanian JZ, Lezzoni LI, Ingber MJ, Levy JM, Robst J. Risk adjustment of Medicare capitation payments using the CMS-HCC model. Health Care Financ Rev. 2004 Summer;25(4):119–41. [PubMed]
17. Fisher ES, Wennberg DE, Stukel TA, Gottlieb DJ, Lucas FL, Pinder EL. The implications of regional variations in Medicare spending. Part 1: the content, quality, and accessibility of care. Ann Intern Med. 2003 Feb 18;138(4):273–87. [PubMed]
18. Welch WP, Miller ME, Welch HG, Fisher ES, Wennberg JE. Geographic variation in expenditures for physicians’ services in the United States. N Engl J Med. 1993;328:621–7. [PubMed]
19. Wennberg JE. The Dartmouth Atlas of Health Care 1996. Chicago: American Hospital Publishing; 1996.
20. White H. A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica. 1980;48:817–30.
21. Huber PJ. The behavior of maximum likelihood estimates under non-standard conditions. Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability; Berkeley: University of California Press; 1967. pp. 221–33.
22. Rice TH. The impact of changing Medicare reimbursement rates on physician-induced demand. Med Care. 1983;21(8):803–15. [PubMed]
23. Nguyen NX, Derrick FW. Physician behavioral response to a Medicare price reduction. Health Serv Res. 1997;32(3):283–98. [PMC free article] [PubMed]
24. Yip WC. Physician response to Medicare fee reductions: changes in the volume of coronary artery bypass graft (CABG) surgeries in the Medicare and private sectors. J Health Econ. 1998;17:675–99. [PubMed]
25. Zuckerman S, Norton SA, Verrilli D. Price controls and Medicare spending: assessing the volume offset assumption. Med Care Res Rev. 1998 Dec;55(4):457–78. [PubMed]
26. Jacobson M, Earle CC, Price M, Newhouse JP. How Medicare’s payment cuts for cancer chemotherapy drugs changed patterns of treatment. Health Aff (Millwood) 2010 Jul;29(7):1391–9. [PubMed]
27. Hackbarth GM. Letter to Senators Max Baucus and Orrin G. Hatch and Representatives Dave Camp, Sander M. Levin, Fred Upton, and Henry A. Waxman. Washington, DC: Medicare Payment Advisory Commission; Oct 14, 2011. Moving forward from the sustainable growth rate (SGR) system.