: To describe trends in out-of-pocket healthcare costs, including insurance premiums, for older Medicare beneficiaries living in the community.

: Medicare Current Beneficiary Survey data were analyzed for community-dwelling beneficiaries 65 years or older between 1992 and 2004.

: The primary focus of the analysis was out-of-pocket healthcare costs and out-of-pocket costs as a percentage of income. Descriptive statistics are presented for 1992, 1996, 2000, and 2004.

: Inflation-adjusted median out-of-pocket costs were stable between 1992 and 2000 and then rose by 21.7% between 2000 and 2004. Median costs as a percentage of income declined between 1992 and 1996 but increased from 12.6% in 2000 to 15.5% in 2004. Between 1992 and 2004, out-of-pocket costs increased fastest at the upper percentiles of the distribution. High out-of-pocket costs tended to persist from year to year, exacerbating the financial burden for some beneficiaries.

: Following a period of declining burden between 1992 and 1996, out-of-pocket healthcare costs rose significantly between 2000 and 2004, increasing the financial burden for many older Medicare beneficiaries. These data provide a baseline for evaluating Medicare reform proposals that affect beneficiary spending.

Following a period of relative stability between 1992 and 2000, the median out-of-pocket healthcare costs grew by 21.7% between 2000 and 2004 after adjustment for inflation. 

The relative burden of out-of-pocket costs, measured as a percentage of income, declined between 1992 and 1996 and then rose between 2000 and 2004. 

High costs tended to persist over multiple years, significantly exacerbating the long-term financial effect of chronic illness on many beneficiaries.

Medicare costs have risen rapidly in recent years and are projected to do so for the foreseeable future.

 Cost increases potentially create a burden for beneficiaries, who must pay for a portion of their care through Part B premiums, deductibles, and coinsurance payments. Increases in the costs of noncovered services also have a significant effect. The introduction of the Medicare Part D prescription drug benefit in 2006 was expected to ease the financial burden on beneficiaries, but the effect seems to be modest.

Given recent and projected increases in healthcare costs, it is useful to examine the current level of financial burden on Medicare beneficiaries, the distribution of that burden, and trends over time. Previous research has shown that out-of-pocket spending as a share of income has increased in recent years.

 Out-of-pocket costs tend to be highly skewed and are higher for beneficiaries in poor health and for those who have an individually purchased private insurance supplement (Medigap) or no supplement.

 Beneficiaries with higher incomes tend to spend more on healthcare, but such costs represent a substantially greater burden for low-income beneficiaries.

This article describes trends in out-of-pocket costs among older community-dwelling beneficiaries between 1992 and 2004 using data from the Medicare Current Beneficiary Survey (MCBS). All out-of-pocket healthcare costs are examined (vs just those associated with Medicarecovered services) to obtain a complete picture of the financial burden of healthcare on beneficiaries. The following questions are addressed: (1) How much have out-of-pocket costs increased absolutely and relative to income? (2) Has the distribution of out-of-pocket costs changed over time? (3) How do costs vary by beneficiary characteristics such as income and health status? (4) To what extent do high out-of-pocket costs persist from year to year?

The MCBS is a longitudinal multipurpose survey of a nationally representative sample of the Medicare population.10 It has been conducted by the Centers for Medicare & Medicaid Services (formerly the Health Care Financing Administration) continuously since 1991. The survey uses a rotating panel design, with each panel consisting of about 4000 respondents who participate in 12 interviews that produce data for 3 complete calendar years. Medicare administrative records are routinely linked to survey data.

Respondents are asked about their use of healthcare services, including Medicare-covered and non&#8211;Medicare-covered services. Costs and sources of payment are determined for each service received by the respondent. Survey-reported services are matched to Medicare claims records to improve the accuracy of utilization and cost data.

 Information on health and functional status, specific medical conditions, supplemental health insurance, and demographics is also collected.

The MCBS collects information about income of the respondent and spouse (if married). Respondents are asked to provide total income only, with prompts to include several specific types of income in the total. In cases of missing data, hot-deck procedures are used to generate responses. The MCBS income data are subject to underreporting and misreporting,

 with systematic underreporting primarily occurring among higher-income beneficiaries (D Waldo, MA, unpublished data, 2005). Despite the limitations of income as captured in the MCBS, the consistency of the data collection procedures over time should yield a reasonably accurate picture of trends. Because income is not collected separately for respondents and their spouses, reported income was divided by 2 for married respondents in the analyses reported herein. 

The median out-of-pocket costs (after adjustment for inflation) rose from $2107 in 1992 to $2581 in 2004, an increase of 22.5% (

). The increase was not uniform; the median out-of-pocket costs decreased by 6.5% between 1992 and 1996, followed by increases of 7.6% between 1996 and 2000 and 21.7% between 2000 and 2004. Out-of-pocket costs were higher for older beneficiaries. The median out-of-pocket costs were lowest for low-income beneficiaries, for whom costs did not increase during 12 years of the study after adjustment for inflation.

In 1992, half of the beneficiaries reported spending 14.8% or more of their income on healthcare costs; 10% reported spending 42.4% or more (Table 2). Between 1996 and 2004, the burden at the upper percentiles of the distribution increased at a faster rate than the burden at the lower percentiles. By 2004, half of the beneficiaries reported spending 15.5% or more of income on healthcare costs, which was not significantly different from 1992. However, 10% reported spending 48.5% or more of income, which was significantly higher than in 1992.

For beneficiaries with chronic conditions, high out-ofpocket costs may persist over multiple years, exacerbating the total financial burden of illness. For example, combining data from 2003 and 2004 for respondents who were in the survey both years, the 90th percentile of out-of-pocket costs as a percentage of income was 43.8%, indicating a high 2-year burden. Among respondents who had high out-of-pocket costs as a percentage of income in 2003 (defined as those at the 90th percentile or above), 33.8% also had high out-ofpocket costs in 2004 (data not shown).

Some recent trends in out-of-pocket healthcare costs for older Medicare beneficiaries may be of concern to policy makers. Following a period of relative stability between 1992 and 2000, the median costs grew by 21.7% between 2000 and 2004 after adjustment for inflation. Although the relative burden of out-of-pocket costs, measured as a percentage of income, declined significantly between 1992 and 1996, it rose by almost 3 percentage points between 2000 and 2004. Increases in financial burden tended to be greatest at the upper end of the distribution. High costs also tended to persist over multiple years, significantly exacerbating the long-term financial effect of chronic illness on many beneficiaries. A key question for policy makers is whether the increase in burden of out-of-pocket costs since 2000 will continue, especially for the segment of the beneficiary population with the highest out-of-pocket costs.

A more favorable trend was that the median out-of-pocket costs did not change between 1992 and 2004 for low-income beneficiaries. This finding is attributable to increases in Medicaid coverage during the study period. Among respondents in the lowest-income quartiles, Medicaid coverage increased from 28% in 1992 to 39% in 2004. Out-of-pocket costs were much lower for Medicaid recipients than for low-income respondents without Medicaid (data not shown). On the other hand, the largest increase in financial burden over 12 years of the study occurred among beneficiaries in the second lowest&#8211;income quartile, for whom Medicaid coverage was much less common.

Medicare&#8217;s long-term financial health continues to be of great concern.

 Policy makers have considered several proposals aimed at containing costs such as raising the age of eligibility, increasing taxes, restructuring Medicare benefits, implementing a defined contribution or premium support model, and encouraging private savings.

Underlying any proposals for program reform are the fundamental questions of how much additional cost burden should be borne by beneficiaries and how such burdens should be distributed. Some proposals call for increased beneficiary contributions in the form of increased premiums or greater cost sharing.

 Future increases in beneficiary financial burden could also be tied to wealth or income,

and means testing of the Part B premium and Part D benefits has already been introduced through the Medicare Modernization Act of 2003. The findings reported herein provide baseline information for evaluating the beneficiary effect of reform proposals, especially among vulnerable subgroups of beneficiaries.

Although the Part D prescription drug benefit is expected to significantly reduce beneficiary out-of-pocket costs, it may be of limited help for beneficiaries with the highest out-ofpocket costs. Among such beneficiaries in 2004, prescription drugs accounted for only 15% of their out-of-pocket costs. It should be noted that the full effect of Part D will not be limited to direct out-of-pocket costs for prescription drugs; many beneficiaries may also experience lower health insurance premiums under Part D because these premiums should cover fewer prescription drug benefits.

This study was limited to the community-dwelling population. Institutionalization is often associated with high out-of-pocket costs, sometimes resulting in spend-down to eligibility for Medicaid coverage. The rate of nursing home use has declined substantially among older persons in recent years.

 This decline has reduced the beneficiary financial burden associated with institutionalization.

The Medicare program will continue to attract proposals for reform. The amount that beneficiaries contribute to the costs of their healthcare and the way that burden is distributed will be central to the debate. Current trends in the level and distribution of out-of-pocket costs will be important factors in planning for the future of Medicare.

: I thank James Lubitz, MPH, and Brigid Goody, ScD, for their valuable comments on early drafts of the manuscript.

: Office of Research, Development, and Information, Centers for Medicare & Medicaid Services, Baltimore, MD.

: The author reports no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article. The statements contained in this article are those of the author and do not necessarily reflect the views or policies of the Centers for Medicare & Medicaid Services.

: Concept and design (GFR); acquisition of data (GFR); analysis and interpretation of data (GFR); drafting of the manuscript (GFR); critical revision of the manuscript for important intellectual content (GFR); statistical analysis (GFR); and provision of study materials or patients (GFR).

: Gerald F. Riley, MSPH, Office of Research, Development, and Information, Centers for Medicare & Medicaid Services, 7500 Security Blvd, Rm C3-21-27, Baltimore, MD 21244. E-mail: gerald. 

1. Boards of Trustees, Federal Hospital Insurance and Federal Supplementary Medical Insurance Trust Funds.

2008 Annual Report of the Boards of Trustees of the Federal Hospital Insurance and Federal Supplementary Medical Insurance Trust Funds

. Washington, DC: Federal Hospital Insurance and Federal Supplementary Medical Insurance Trustees; 2008.

3. Neuman P, Cubanski J, Desmond KA, Rice TH

. How much â€˜skin in the gameâ€™ do Medicare beneficiaries have? The increasing financial burden of health care spending, 1997-2003. 

5. Hwang W, Weller W, Ireys H, Anderson G

. Out-of-pocket medical spending for care of chronic conditions. 

7. Gross DJ, Alecxih L, Gibson MJ, Corea J, Caplan C, Brangan N

. Out-of-pocket health spending by poor and near-poor elderly Medicare beneficiaries. 

9. Federal Interagency Forum on Aging-Related Statistics

Older Americans 2008: Key Indicators of Well-being

. Washington, DC: US Government Printing Office; March 2008.

. Matching MCBS (Medicare Current Beneficiary Survey) and Medicare data: the best of both worlds. 

. Patterns of health care and disability for Medicare beneficiaries under 65. 

. A moving target: financing Medicare for the future. 

17. National Bipartisan Commission on the Future of Medicare Web site

. Final Breaux-Thomas Medicare reform proposal. March 16, 1999. 

 2004;suppl Web exclusives:W4-546-W4-557. 

Articles

TRENDS FROM THE FIELD

Out-of-pocket healthcare costs rose significantly between 2000 and 2004; these data provide a baseline for evaluating Medicare reform proposals that affect beneficiary spending.

Trends in Out-of-Pocket Healthcare Costs Among Older Community-Dwelling Medicare  Beneficiaries

: To examine whether physicians who sought and received Bridges to Excellence (BTE) recognition performed better than similar physicians on a standardized set of population-based performance measures.

: Cross-sectional comparison of performance data.

: Using a claims dataset of all commercially insured members from 6 health plans in Massachusetts, we examined population-based measures of quality and resource use for physicians recognized by the BTE programs Physician Office Link and Diabetes Care Link, compared with nonrecognized physicians in the same specialties. Differences in performance were tested using generalized linear models.

: Physician Office LinkÃ¢â‚¬â€œrecognized physicians performed significantly better than their nonrecognized peers on measures of cervical cancer screening, mammography, and glycosylated hemoglobin testing. Diabetes Care LinkÃ¢â‚¬â€œrecognized physicians performed significantly better on all 4 diabetes process measures of quality, with the largest differences observed in microalbumin screening (17.7%). Patients of Physician Office LinkÃ¢â‚¬â€œrecognized physicians had a significantly greater percentage of their resource use accounted for by evaluation and management services (3.4%), and a smaller percentage accounted for by facility (-1.6%), inpatient ancillary (-0.1%), and nonmanagement outpatient services (-1.0%). After adjustment for patient age and sex, and case mix, Physician Office LinkÃ¢â‚¬â€œrecognized physicians had significantly fewer episodes per patient (0.13) and lower resource use per episode ($130), but findings were mixed for Diabetes Care LinkÃ¢â‚¬â€œrecognized physicians.

: Our findings suggest that the BTE approach to ascertaining physician quality identifies physicians who perform better on claims-based quality measures and primary care physicians who use a less resource-intensive practice style.

Bridges to Excellence (BTE) rewards physicians based on the results of site surveys and retrospective review of clinical data. Physicians recognized as high quality performed better than their peers on claims-based measures of quality and, in some cases, resource use measures.

Physicians recognized by BTE appear to rely more on evaluation and management, and less on tests and procedures.

Higher performance for recognized physicians may have predated the recognition program.

Further research is needed to examine whether the recognition by BTE is associated with patient flow or performance improvement.

During the past several years, both public and private payers have adopted pay for performance in a wide variety of contexts with the hope of prompting more evidence-based and higher value patterns of care.

 When it was launched in 2002 by a group of large employers collaborating with several health plans and provider organizations, Bridges to Excellence (BTE) was one of the first multistakeholder pay-for-performance programs.

 Its programs have relied on certification of performance based on self-reported (subject to audit) medical record and practice systems data. Physicians are invited by BTE to seek certification and begin receiving payments for all patients covered by the program’s sponsors. Incentives for physicians who meet or exceed the performance criteria were initially determined from actuarial analyses that estimated savings to employers from better quality of care

 and are paid by participating employers and health plans directly to physicians, according to a specified bonus schedule that can vary between implementation regions.

Now operated through an independent not-for-profit entity with a board comprised of employer, health plan, and physician representatives, BTE’s model is in operation nationwide through licenses to health plans, employer coalitions, and state agencies.

 Bridges to Excellence was first implemented in Massachusetts in 2003 with 2 major physician reward components: the Physician Office Link (POL) and the Diabetes Care Link (DCL). Attainment of the quality standards set by BTE is associated with both financial rewards and public recognition.

In 2003, approximately 20 physicians in Massachusetts were recognized by BTE under what was then its only program: DCL. By the end of 2006 there were more than 1000 BTE-recognized physicians in Massachusetts across all BTE programs and total payments to these physicians reached $2.4 million. (In 2004 the Cardiac Care Link was introduced. Because that program is newer, we do not examine it in our analysis.) Nationally, 8500 physicians have been recognized and a total of $10 million paid out. Although the BTE model continues to expand, comparative studies of BTE-recognized physicians and their practice patterns are lacking. In this article, we examine the quality, care delivery patterns, and resource use of BTE-recognized physicians compared with nonrecognized physicians in Massachusetts.

On quality measures alone, this comparison is of interest because of BTE’s unusual reliance on voluntary certification through a site survey and chart review that yields data regarding quality measures similar to those used in the Healthcare Effectiveness Data and Information Set (HEDIS). For the POL in particular, which is intended to reward overall population health management, substantial weight is placed on the structural measures of quality derived from the site survey. Thus, it is not clear that physicians with BTE certification would necessarily get better-than-average scores on the quality measures alone. Moreover, there is no comparative aspect to BTE recognition—the standards are set nationally and fixed (ie, they do not reflect relative rankings within a market), so it is unknown whether the physicians seeking certification are those whose performance is above average or simply those with larger numbers of eligible patients (and thus with more to gain). Finally, because BTE certifies physicians for 3 years based on a single retrospective review of 25 sampled charts for the quality measures, we also considered it important to ask whether a contemporaneous comparison between recognized and nonrecognized physicians would show better performance for BTE physicians. Even if recognized physicians performed better than average with the selected patients, these differences might not have persisted.

We examined population-based measures of the quality of care and resource use for Massachusetts physicians participating in BTE’s first 2 rewards programs, POL and DCL. Using an administrative dataset comprised of claims from all commercially insured members of 6 Massachusetts plans, we compared the performance of recognized physicians with the performance of nonrecognized physicians.

The POL program focuses on promoting the office practice’s use of systems to enhance the quality of patient care. Today, this assessment is commonly being referred to as a core characteristic of a Medical Home.

 Practices that can demonstrate specific processes and systems of care can earn up to $50 for each patient covered by a participating employer. (The rewards were designed to last 3 years, the period of time during which the recognition is valid.) To obtain the rewards available through the POL program, eligible physicians must pass the National Committee for Quality Assurance (NCQA) Physician Practice Connections assessment program or meet a comparable standard set by the Massachusetts Quality Improvement Organization: MassPRO. The standards for the POL, which are described in Table 1,

 include maintenance of patient registries for the purpose of identifying and following up with at-risk patients, provision of educational resources to patients, and use of electronic systems to maintain patient records, provide decision support, enter orders for prescriptions and lab tests, and provide patient reminders.

The DCL program rewards physicians with $80 per year for each diabetic patient covered by a participating employer. To obtain the rewards, eligible physicians must demonstrate that they provide high levels of diabetes care by passing NCQA’s Diabetes Physician Recognition Program, which was developed in collaboration with the American Diabetes Association.

 To qualify for the 3-year recognition, physicians submit medical record data on glycosylated hemoglobin (A1C), blood pressure, and lipid testing, as well as data on eye, foot, and nephropathy exams, for a random sample of 25 diabetes patients. (The 25 charts are selected retrospectively for patients who have been in the practice’s care for at least 1 year. A date is selected at random from a set window of time, and the next 25 charts for patients with a diagnosis for diabetes are pulled for analysis. Both the sampling method and the actual charts are reviewed and subject to NCQA audit.) The NCQA assesses the data submitted by the physician and attributes points based on the level of disease management demonstrated by the physician. All information submitted for POL and DCL certification is subject to audit.

We obtained data from BTE on the identities of physicians recognized for the POL and DCL programs between 2003 and 2006. Administrative data, including claims and enrollment files, were obtained from the Massachusetts Group Insurance Commission (GIC), which has created a database that includes all claims from all commercially insured members (not just those for GIC members) from the 6 health plans that serve the GIC (Harvard Pilgrim Healthcare, Tufts Health Plan, Unicare, Fallon, Neighborhood Health Plan, and Health New England). The GIC’s data cover approximately 50% of privately insured residents and more than 7000 physicians in Massachusetts including all but 2 BTE-recognized physicians. Physician-level measures of quality and resource use per episode were computed by Mercer Human Resources Consulting using profiling software licensed from Resolution Health, Inc, and Ingenix, respectively.

To ascertain differences between recognized and nonrecognized physicians, we examined claims-based quality measures and average resource use per standardized episode for several categories of services and the total. (Episodes were created using Symmetry’s Episode Treatment Groupers, a unit of Ingenix. These episodes are defined using proprietary algorithms that identify in claims data clinical events that trigger, break, and end an episode. Similarly, algorithms based on clinical logic are used to group all claims related to that episode. For a more complete description, see reference 11.) For the POL cohort, we examined a set of measures typically used for examining primary care quality, including cervical cancer screening, mammography, A1C testing, cholesterol screening for individuals with coronary heart disease, and cholesterol screening for individuals with hypertension. For the DCL cohort, we examined 4 widely used claims-based measures of diabetes care quality. These were A1C testing, cholesterol testing, microalbumin testing, and diabetic retinal exams. Measure specifications were adapted by Resolution Health, Inc, from HEDIS and other national evidence-based guidelines to conform to the provider profiling context. The most recent 18 months of outpatient claims and pharmacy data were used for evaluation and calculation of each quality score (eg, A1C testing rate).

Eligible patients were attributed to multiple physicians based on the following steps. In the first step of the analysis, a member’s claims were evaluated to determine which physicians had the greatest number of claims. In the second step, the rules (eg, diabetic patients should have A1C testing at least annually) that are applicable to an individual member are then assigned to all the physicians identified in step 1 who could be responsible for giving that type of care based on their specialty (eg, internal medicine or endocrinology could be responsible for diabetes care, but dermatology could not). The quality score for a physician was then calculated by determining what percentage of the patients for whom the physician was deemed “responsible” through this process met the criteria for the rule (eg, annual A1C testing).

The second set of measures we examined includes standardized measures of spending that we refer to as resource use (we used Medicare rates rather than the actual transaction prices, which are unique to each payer-provider pair). We decomposed resource use in 2 ways. First, we use the Episode Treatment Grouper software from Ingenix to tally resource use in 6 categories: evaluation and management, surgery (claims submitted by physicians for surgical interventions), facility (room and board only), inpatient ancillary (inpatient claims for services other than room and board), outpatient (claims for services provided in the outpatient setting other than evaluation and management), and prescription drugs.

 Eligible patients were attributed to physicians who were responsible for the majority of the clinician fees in the episode. We compared average percentages of total patient resource use in each category for recognized and nonrecognized physicians. Then, we examined the number of episodes per patient and the total resource use per episode.

We compared average performance for physicians recognized by the 2 BTE programs with the performance of similar physicians who were not recognized. All physicians who were eligible for BTE rewards in Massachusetts were included in the comparison groups. The initial list was refined by excluding physicians who did not have at least 200 episodes and did not have enough data to receive a quality score. Additionally, comparison groups were selected from the same specialties as those of the recognized physicians: internal medicine and family practice for the POL and those 2 specialties plus endocrinology for the DCL. For the POL analysis, the comparison group was further restricted by excluding physicians who were outliers in their resource use (ie, whose average resource use per episode of care was more than 6 standard deviations above the mean resource use for all POL physicians). Analyses were conducted using the physician as the unit of analysis. The statistical significance of differences in average performance for each measure of resource use and quality was determined by using generalized linear models in which the explanatory variables were specialty, mean patient age, the percentage of patients who were female, and a case mix indicator, which is a measure of the resource intensiveness of the Episode Treatment Groupers of a physician’s entire patient population compared with Episode Treatment Groupers for other physicians in the same specialty. We report predicted values from these regressions for both sets of physicians and used a bootstrapping (ie, resampling) method

 to compute 95% confidence intervals (CIs) for the differences.

We identified 405 physicians who were recognized by BTE’s POL program between 2003 and 2006 and 3916 nonrecognized physicians in the GIC’s data (Table 2). Physicians recognized through the POL were more likely to be specialists in internal medicine than nonrecognized physicians and were less likely to be family practitioners. Patient sex was similar between POLrecognized and nonrecognized physicians, but patients were younger and case mix was substantially higher for nonrecognized physicians. Between 2003 and 2006, BTE’s DCL program recognized 91 Massachusetts physicians; GIC’s data showed 1204 nonrecognized physicians treating diabetic patients during the same time period. Physicians recognized through the DCL were much more likely to be specialists in endocrinology than nonrecognized physicians and much less likely to be family practitioners. Patient age and sex were similar between DCL-recognized and nonrecognized physicians, but case mix was substantially lower for nonrecognized physicians. Comparison of Quality Measures On all process measures of quality, recognized physicians performed better than nonrecognized physicians (Table 3). Differences were statistically significant for cervical cancer screening, mammography, and A1C testing for the comparisons of POL-recognized physicians with nonrecognized physicians. The DCL-recognized physicians performed statistically significantly better on all 4 diabetes process measures of quality, with the largest differences observed in microalbumin screening (17.7%; 95% CI = 14.0%, 21.4%). This second finding seems to affirm that the NCQA’s Diabetes Physician Recognition Program concurs with claims-based measures of high-quality care for patients with diabetes.

Comparison of Practice Patterns and Resource Use Measures

Differences in practice patterns between physicians recognized by the POL program and the comparison group shared a common pattern but had mixed degrees of statistical significance (Table 4). In particular, patients of POL-recognized physicians had a significantly greater percentage of their resource use accounted for by management services (3.4%; 95% CI = 2.4%, 4.4%) and a smaller percentage accounted for by facility (−1.6%; 95% CI = −3.4%, −0.02%), inpatient ancillary services (−0.1%; 95% CI = −0.1%, −0.03%), and nonmanagement outpatient services (−0.1%; 95% CI = −1.7%, −0.3%). Patients of POL-recognized physicians also had a lower percentage of their total resource use accounted for by prescription drugs (−1.6%; 95% CI = −2.8%, −0.3%). The distribution of expenditures for DCL-recognized physicians compared with nonrecognized physicians treating diabetic patients were more similar, with smaller, but significant differences only in the share of standardized resource use accounted for by management (2.0%; 95% CI = 0.5%, 3.6%) and surgery (−0.3%; 95% CI = −0.5%, −0.1%).

After adjustment for specialty, patient age and sex, and case mix, POL-recognized physicians had significantly fewer episodes per patient (0.13; 95% CI = 0.13, 0.15) and lower resource use per episode ($130; 95% CI = $119, $140) (Table 5). For diabetes care, however, DCL-recognized physicians had more episodes per patient (0.17; 95% CI = 0.14, 0.20 for primary care physicians; and 0.08; 95% CI = 0.03, 0.13 for endocrinologists). Resource use per episode was lower for DCLrecognized primary care physicians ($26; 95% CI = $5, $47) and higher, but not significantly so, for DCL-recognized endocrinologists ($137; 95% CI = −$45, $320).

Bridges to Excellence is a national pay-for-performance program that relies on chart review and structural survey data to assess the quality of care in physician practices.

Although analyses conducted during the design phase of BTE suggested that recognized physicians would not only deliver higher quality care but also would do so at lower total cost, ours is the first formal analysis comparing the practice patterns of recognized physicians with those of their peers using a standardized methodology that relied on a large claims database. Our findings suggest that recognition by BTE is associated with systematically better performance on claims-based measures of quality. Moren over, recognized physicians appear to use a different style of practice than their peers, emphasizing patient management over ancillary services and procedures. Patterns of relative resource use for recognized versus nonrecognized physicians were more varied. Possibly because of better management, primary care physicians recognized by the POL had nearly 20% lower resource use per episode with approximately 5% fewer episodes per patient. The DCL-recognized primary care physicians had slightly lower resource use per episode for their patients, but DCL-recognized endocrinologists did not. Both primary care physicians and endocrinologists recognized by the DCL, however, had more episodes per patient than nonrecognized physicians in their respective specialties.

Bridges to Excellence offers a potential complement to claims-data profiling of physicians for identifying and rewarding high-quality care. The advantages of this measurement approach derive primarily from its ability to capture clinical information and the associated greater credibility with physicians. Our analysis inadvertently highlighted another advantage compared with claims-based profiling methods as well. Even though the GIC’s data cover 6 major payers, for some quality measures a segment of physicians could not be profiled because of an insufficient number of patients in the denominator (even using a relatively low cut-off of 10 patients per measure). By design, however, the BTE chart review approach can be applied to virtually all practices because it considers the universe of a practice’s patients. Making “all-payer” data truly include all payers—including Medicare and Medicaid&mdash; and standardizing physician identifiers to maximize linkages among payers would reduce this advantage and dramatically increase the usefulness of claims data for profiling physicians, however.

The primary limitations of the BTE model for profiling physicians are twofold. First, because assessments are voluntary, BTE may only get the attention of already high-performing physicians and do little to encourage improvement among the least prepared. Second, the collection and auditing of both the clinical and site survey data are expensive, particularly for small practices, relative to analysis of existing administrative data.

The association between POL recognition and lower resource use is consistent with findings from the actuarial model underpinning BTE. Each BTE program had been originally subjected to an actuarial analysis to determine ex ante whether cost savings would be realized by recognized physicians, and it is from these analyses that financial reward targets also were established (eg, $80 yearly bonus per diabetic patient). Although we are unable to say at this point whether the relationship between recognition status and lower resource use is a causal one (such that more recognition will increase cost-efficiency), the knowledge of this association is important for payers and consumers who are seeking ways to increase the value obtained in the healthcare system. In particular, moving patients to these physicians might increase value, although this result may be limited to the extent that unmeasured heterogeneity of patients treated by recognized versus nonrecognized physicians is affecting the results. If nonrecognized physicians responded to such a loss of market share by adopting similar practice patterns and improving the quality and efficiency of their practice, the value of care delivered throughout the healthcare system would improve.

Our findings should be viewed in light of several limitations. First, our analysis is cross-sectional in nature, because time-series data were not available. Thus, we cannot make any claims about causal relationships between the BTE recognition program and quality or between quality and resource use. Second, because chartreview data were available only for recognized physicians, we did not have access to clinical data in our analysis for risk adjustment. Although our comparisons were adjusted for patient age and sex, and an indicator of case mix, residual patient heterogeneity correlated with recognition status may be confounding our findings. This is a particular issue for the resource use comparisons. Finally, we chose to examine Massachusetts physicians because of the availability of a large multipayer database, but our results may not generalize to markets where BTE might be launched later. Factors such as overall performance on quality measures, the organization of practice, and the intensity of use, all of which vary regionally, may be important mediators of the relationships we describe.

Although the evidence of impact is mixed,

pay-forperformance programs like BTE have engaged payers and providers in an important dialogue about how to improve the quality and cost-efficiency of care. Our analysis demonstrated a positive relationship between BTE recognition and population-based measures of quality. It remains unknown, however, whether the program motivated physicians to reorganize their practices or adopt new treatment patterns. Although providing a means for high-performing physicians to identify themselves to payers and patients may be a worthwhile goal in itself, it also is critical to understand whether BTE and similar programs create a business case for poor performers to change.  Diffusion of BTE’s program in markets around the country should create opportunities to address this question.

: From the Department of Health Policy and Management, Harvard School of Public Health (MBR, ADS), Boston, MA; Bridges to Excellence (FSD), Newtown, CT; Mercer Health Benefits (MF), Cleveland, OH; Mercer Health Benefits (RDR), Norwalk, CT; and Health Data Management Solutions (SY), Shaker Heights, OH.

: This work was supported by general institutional funds.

: The authors (MBR, ADS, MF, RDR, SY) report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article. Mr de Brantes is the CEO of Bridges to Excellence, the organization whose pay-for-performance programs are discussed in this study.

: Concept and design (MBR, FSD, MF); acquisition of data (FSD, MF); analysis and interpretation of data (MBR , ADS, MF, RDR, SY); drafting of the manuscript (MBR, FSD, ADS, RDR); critical revision of the manuscript for important intellectual content (MBR , FSD, RDR); statistical analysis (MBR, ADS); administrative, technical, or logistic support (MBR); and supervision (MBR).

: Meredith B. Rosenthal, PhD, Department of Health Policy and Management, Harvard School of Public Health, 677 Huntington Ave, Boston, MA 02115. E-mail: 

Pay-for-Performance in State Medicaid Programs: A Survey of State Medicaid Directors and Programs

 Health plan pay-for-performance strategies. 

3. Rosenthal MB, Landon BE, Normand SL, Frank RG, Epstein AM

. Paying for pay-for-performance in commercial HMOs. 

. Paying physicians for high-quality care. 

5. Rosenthal MB, Fernandopulle R, Song HR, Landon B

. Paying for quality: providersÃ¢â‚¬â„¢ incentives for quality improvement. 

http://www.bridgestoexcellence.org/Content/ContentDisplay.aspx?ContentID=81

. The Patient Centered Medical Home. History, Seven Core Features, Evidence and Transformational Change. November 2007. 

http://www.adfammed.org/documents/grahmcentermedicalhome.pdf

9. National Committee for Quality Assurance

http://www.ncqa.org/tabid/141/Default.aspx

10. National Committee for Quality Assurance

. Diabetes Physician Recognition Program. 

http://www.ncqa.org/tabid/139/Default.aspx

Symmetry Episode Treatment Groups. A Condition Classification and Episode Building System

http://www.ingenix.com/content/attachments/ETG%206.0%20White%20Paper_01-17-07.pdf

. Bootstrap critical values for tests based on generalized-method-of-moments estimators. 

14. Petersen LA, Woodard LD, Urech T, Daw C, Sookanan S

. Does pay-for-performance improve the quality of health care? 

15. Lindenauer PK, Remus D, Roman S, et al

. Public reporting and pay for performance in hospital quality improvement. 

. Paying for performance at the tipping point. 

17. Young GJ, Meterko M, Beckman H, et al

. Effects of paying physicians based on their relative performance for quality.

POLICY

Physicians recognized as high quality by Bridges to Excellence performed better than their peers on claimsbased quality measures and, in some cases, on resource use measures.

Bridges to Excellence&mdash;Recognizing High-Quality Care: Analysis of Physician Quality and Resource Use

: To examine financial implications of the Centers for Medicare & Medicaid Services Hierarchical Condition Categories (CMS-HCC) risk-adjustment model on Medicare payments for individuals with comorbid chronic conditions. 

: The study used 1992-2000 data from the Medicare Current Beneficiary Survey and corresponding Medicare claims. Pairs of comorbidities were formed based on prior evidence about possible synergy between these conditions and activities of daily living (ADLs) deficiencies, and included heart disease and cancer, lung disease and cancer, stroke and hypertension, stroke and arthritis, congestive heart failure (CHF) and osteoporosis, diabetes and coronary artery disease, and CHF and dementia.

: For each beneficiary, we calculated the actual Medicare cost ratio as the ratio of the individualâ€™s annualized costs to the mean annual Medicare cost for all people in the study. The actual Medicare cost ratios, by ADLs, were compared with HCC ratios under the CMS-HCC payment model. Using multivariate regression models, we tested whether having the identified pairs of comorbidities affected the accuracy of CMS-HCC model predictions.

: The CMS-HCC model underpredicted Medicare capitation payments for patients with hypertension, lung disease, CHF, and dementia. The difference between the actual costs and predicted payments was partially explained by beneficiary functional status and less-than-optimal adjustment for these chronic conditions. 

: Information about beneficiary functional status should be incorporated in reimbursement models. Underpaying providers who care for populations with multiple comorbidities may provide severe disincentives for managed care plans to enroll such individuals and to appropriately manage their complex and costly conditions.

Our findings indicate that information about beneficiary functional status should be incorporated in Medicare reimbursement models.

Underpaying providers who care for populations with multiple comorbidities may provide severe disincentives for managed care plans to enroll such individuals and to appropriately manage their complex and costly conditions.

Chronic conditions such as heart disease, hypertension, arthritis, cancer, and diabetes are the leading causes of disability and death in the United States for people older than 65 years.

 Medicare beneficiaries with 5 or more chronic conditions account for 68% of the program&#8217;s spending.

 Co-occurrence of diseases increases markedly with age, with two-thirds of noninstitutionalized Medicare beneficiaries over the age of 65 years reporting 2 or more chronic conditions,

 with the prevalence of multiple comorbidities being even higher among the Medicare population overall. Approximately 25% of those who experience chronic illness have some limitations in functional activity, and the percentage of those with disability increases with the number of coexisting conditions.

 The presence of chronic disease has been consistently shown to be associated with functional dependence,

 with combinations of diseases showing different influence on physical functioning than would be expected with the sum of the individual conditions.

Recognizing the increasing prevalence of chronic comorbid conditions in the Medicare population, as well as the need to adequately compensate Medicare managed care plans for the care they provide to this segment of the population, beginning in 2004 the Centers for Medicare & Medicaid Services (CMS) started to phase in a new riskadjusted payment model. Known as the CMS-Hierarchical Condition Categories (CMS-HCC), this risk-adjustment model relies on demographic and diagnostic information available from administrative data to predict resource use. The model uses a selected subset of International Classification of Diseases, Ninth Revision, Clinical Modification (

) diagnostic codes from hospital and physician encounters to place beneficiaries into 70 disease groups: the HCCs.

(The original model was developed using 1999-2000 claims. Starting in 2007, the HCC model was recalibrated using 2002-2003 data.) Each disease group includes conditions that are related clinically and have similar cost implications. In addition, the model accounts for the fact that having certain combinations of diseases may result in higher medical expenditures that simply are the sum of the 2 conditions. For instance, such disease interaction coefficients are allowed for diabetes and congestive heart failure (CHF); diabetes and cerebrovascular disease; diabetes, CHF, and renal failure; and a limited number of others.

There have always been concerns regarding the accuracy of the HCC model in predicting Medicare payment.

Understanding the relationship between functional limitations and cost of medical care in patients with multiple chronic conditions is currently gaining importance and recognition. In 2003, CMS created Special Needs Plans (SNPs), allowing healthcare providers to accept full risk from CMS for all medical and pharmacy health expenses for enrollees with specific chronic diseases.

 By the beginning of 2008, 775 plans enrolled nearly 1 million beneficiaries.

The HCC model does not (except for the Program of All Inclusive Care for the Elderly [PACE]) include adjustment for functional impairment. Studies have shown that this lack of adjustment results in underestimation of payments for enrollees with disabilities. We also know that people with comorbid conditions tend to be more functionally disabled; hence, our interest in examining the extent to which the HCC model may not pay appropriately for this segment of the population, very significantly represented in the SNPs. Furthermore, the HCC model does not account at all for a number of prevalent chronic conditions (eg, dementia, osteoporosis); hence, our interest in including those conditions in the analyses.

The effect of multiple comorbidities on disability and cost of care is poorly understood. Ettinger and colleagues explored synergy between arthritis and 4 comorbidities (heart disease, pulmonary disease, obesity, and hypertension) and proposed a mechanism explaining increased disability resulting from multiple comorbidities.

 They suggested that an impairment from one disease (eg, inactivity resulting from arthritis) may exacerbate the impairment from another comorbid condition (eg, low work capacity caused by heart disease), thus modifying the disease-disability relationship. Prior studies also identified additional specific diseases such as cerebrovascular disorders, diabetes, cancer, osteoporosis, atherosclerosis, and neurologic problems that may exacerbate disability resulting from other conditions.

Based on this evidence, we identified the following 11 target comorbidities to be examined in this study, taking into account the level of functional impairment: arthritis, hypertension, heart disease, cancer, lung disease, stroke, osteoporosis, diabetes, coronary artery disease (CAD), CHF, and dementia.

Furthermore, we hypothesized that certain combinations of the 11 target conditions may have synergistic effects with respect to physical and cognitive functioning when evaluated longitudinally. This in turn would affect patient performance in activities of daily living (ADLs) and the subsequent cost of medical care. These predictions were based on the previous cross-sectional (arthritis and hypertension, heart disease and cancer, lung disease and cancer, and stroke and hypertension)

studies. In addition, dementia may accelerate functional decline and mortality, and may exacerbate other chronic conditions as well. Osteoporosis could lead to more fractures and trauma in older patients, which would increase temporary and permanent disability and might limit people&#8217;s ADL performance as they tried to minimize their risks.

The purpose of this study is to assess the accuracy of the CMS-HCC Medicare capitation model in predicting Medicare expenditures for community-based beneficiaries with at least 2 target comorbidities identified above and various degrees of functional impairment. The population of Medicare beneficiaries with coexisting chronic conditions represents a good case for testing the accuracy of the CMS-HCC Medicare risk-adjustment model, which (1) does not account for patient functional limitations that may exacerbate the inaccuracy of predictions for patients with multiple chronic conditions and (2) does not account for all chronic conditions, which also may result in underestimation of payments.

Beneficiaries with the pairs of target conditions were compared with the general Medicare population on characteristics such as sex, race, frequency of each ADL, and place of residence using chi-square tests. Student 

 tests were used to identify significant differences between these groups of patients by age and number of ADLs. Survey weights were incorporated into the comparisons to represent the entire Medicare population. All statistical tests were 2 tailed and were performed using a significance level of 5%.

Comparing Actual Medicare Costs With the CMS-HCC Model Predictions

Using multiple regressions, we tested whether having the identified pairs of comorbidities affected the accuracy of CMSHCC model predictions. The dependent variable was the residual Medicare expenditures ratio, defined as the difference between the actual cost ratio and the predicted cost ratio (the HCC score) for each individual (similar to the approach used by Kautter and Pope

) and based on the work of Temkin-Greener and colleagues

 The residual ratio reflects the accuracy of CMS-HCC model prediction. The independent variables included the dummy variables for the different levels of physical disability (ADLs), target comorbidities, and the interactions between these comorbidities. Survey sampling weights were incorporated in the multiple regression analysis. The analyses were conducted using STATA Statistical Software for Windows, Release 8.0 (College Station, TX: StataCorp) and SAS for Unix, Version 9 (Cary, NC: SAS Institute).

Nearly three-quarters (72.55%) of all Medicare beneficiaries in our study had 2 or more target comorbidities, with the prevalence of different target comorbidities varying substantially. In 

, we compared the characteristics of the general Medicare population with those of beneficiaries who had pairs of target chronic conditions. Although more than a third of all beneficiaries had arthritis and hypertension, only about 1% of people had either CHF and osteoporosis or CHF and dementia. Patients with chronic illnesses were significantly older than the study population overall (age 72.75 years). Proportion of women was greater among patients with osteoporosis and CHF (87.88% women, 

 <.01), and arthritis and stroke (60.54%, women 

 <.01) compared to the general Medicare population (57.56% women). Among patients with cancer and heart (45.98% men, 

 <.01) proportion of men was greater compared to the general Medicare population (42.44% male). Except for the beneficiaries with cancer and heart disease, patients with the pairs of target comorbidities had lower income and were more likely to be on Medicaid than the general Medicare population.

Functional Status of Medicare Beneficiaries With Chronic Conditions

Patients with multiple comorbid conditions had a much greater level of ADL deficiencies than Medicare beneficiaries overall (

). The profiles of disability also varied substantially between patients with different chronic illnesses. Patients with CHF and dementia reported the highest level of deficiency across all ADL categories: 14.38% relied on others&#8217; help with eating (feeding), and more than 50% used help or assisted devices for bathing. Other groups with a high ADL deficiency level included patients with stroke combined with hypertension or arthritis, CHF and osteoporosis, and CAD and diabetes. However, the ranking of the prevalence of individual ADLs was consistent among all patient groups, with eating being the least common and bathing being the most common function for which beneficiaries received help.

Comparing Actual Medicare Costs With CMS-HCC Predicted Payments 

Overall, the CMS-HCC model significantly underpredicted medical expenses of patients with target single comorbidities, except for arthritis (

). We found that for beneficiaries without functional limitations (0 ADLs), the CMS-HCC predicted expenses were no different from the actual cost ratios except for patients with CHF (underpredicted by 18.47%; 

 = .01). As the disability level increased, the model increasingly underpredicted the expenses&#8212;up to 43.65% (

The discrepancy between the actual and predicted cost ratios was larger for beneficiaries with multiple comorbidities than for those with a single target condition. For example, the CMS-HCC model underpredicted the expenses of the beneficiaries with CHF and osteoporosis by 30.02% (

), while the predictions were 20.60% lower (

 <.001) for patients with CHF only and no different from actual costs (

 = .32) for osteoporosis only (Tables 2-3). The model also underpredicted medical expenses for the beneficiaries with arthritis and hypertension by 7.08% (

 = .01), while underpredicting expenses by 5.70% for the patients with hypertension (

 = .01) only; expenditures were underpredicted by 18.70% for patients with diabetes and CAD (

 <.001) for patients with diabetes and 10.40% (

 <.001) for patients with CAD. Moreover, the magnitude of the prediction error was greater for the pairs that included conditions without corresponding HCCs than for the conditions with corresponding HCCs (eg, CHF, cancer) or those accounted for by other HCCs (eg, hypertension, heart disease) (Tables 4-5).

The 95% CIs around the error estimates demonstrated that the study sample size was generally sufficient to make robust predictions. In some cases where the predicted error was not statistically significantly different from zero, the analysis of CIs illustrated clinically or practically substantial error (eg, for osteoporosis with 3 ADLs: 

 = .08; 95% CI = &#8722;5.69, 92.68; for lung disease and cancer with 1 ADL: 

Effect of Functional Status and Comorbidity on Medical Expenses

Because the majority of beneficiaries in our sample had more than 1 of the target comorbidities and various levels of functional impairment, we examined the joint impact of the multiple comorbidities and disability on the accuracy of the CMS-HCC capitation model (Table 3).

Among the pairs of comorbid conditions, having arthritis and hypertension (0.079, 

 = .01) led to substantial underpayments as calculated by the CMS-HCC model. However, these differences were mainly due to the underpayment for the single conditions (CHF and dementia) rather than additional error due to having multiple comorbidities, because adding single conditions improved the explanatory power of the model (R

 values (>.05) for the variables identifying pairs of conditions. Functional status helped explain even more of the difference between the actual costs and the predicted amount based on the capitation model (R

 = 0.46). The number of ADLs was highly significant 

P < .01) in explaining the variation between actual costs and predicted payment, and so was the presence of hypertension, lung disease, and CHF (

Our findings indicate that information about beneficiary functional status should be incorporated in Medicare reimbursement models because without functional-status adjustment such models are likely to underestimate costs of caring for patients with disability and multiple comorbidities. Underpaying providers who care for populations with multiple comorbidities may provide severe disincentives for managed care plans to enroll such individuals and to appropriately manage their complex and costly conditions.

: From the Department of Community and Preventive Medicine, University of Rochester (KN, HL, HT-G), Rochester, NY; and RAND Corporation (HL), Pittsburgh, PA.

: This publication was supported in part by a K01 AG 20980 grant from the National Institute on Aging (KN, HL). The use of the Medicare Current Beneficiary Survey was covered by the Data Use Agreement #12874. The contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Institute on Aging or the National Institutes of Health.

: The authors (KN, HL, HT-G) report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

This study of financial implications of risk adjustment for Medicare payments to individuals with comorbidities and functional impairment demonstrates the importance of controlling for disability.

October 2008

Trends in Out-of-Pocket Healthcare Costs Among Older Community-Dwelling Medicare Beneficiaries

Bridges to Excellence—Recognizing High-Quality Care: Analysis of Physician Quality and Resource Use

Medicare Capitation Model, Functional Status, and Multiple Comorbidities: Model Accuracy

Are Primary Care Physicians Ready to Practice in a Consumer-Driven Environment?

The Changing Effect of Managed Care on Physician Financial Incentives

Ezetimibe 5 and 10 mg for Lowering LDL-C: Potential Billion-Dollar Savings With Improved Tolerability

Health System Correlates of Receipt of Radiation Therapy After Breast-Conserving Surgery: A Study of Low-Income Medicaid-Enrolled Women