www.medscape.

com

Benefits and Costs of Intensive Lifestyle Modification Programs for Symptomatic Coronary Disease in Medicare Beneficiaries
Wu Zeng, MD, MS, PhD, William B. Stason, MD, MSci, Stephen Fournier, PhD, Moaven Razavi, PhD, Grant Ritter, PhD, Gail K. Strickler, PhD, Sarita M. Bhalotra, MD, PhD, Donald S. Shepard, PhD Am Heart J. 2013;165(5):785-792.

Abstract and Introduction

Abstract

Background This study reports outcomes of a Medicare-sponsored demonstration of two intensive lifestyle modification programs (LMPs) in patients with symptomatic coronary heart disease: the Cardiac Wellness Program of the Benson-Henry Mind Body Institute (MBMI) and the Dr Dean Ornish Program for Reversing Heart Disease (Ornish). Methods This multisite demonstration, conducted between 2000 and 2008, enrolled Medicare beneficiaries who had had an acute myocardial infarction or a cardiac procedure within the preceding 12 months or had stable angina pectoris. Health and economic outcomes are compared with matched controls who had received either traditional or no cardiac rehabilitation following similar cardiac events. Each program included a 1-year active intervention of exercise, diet, small-group support, and stress reduction. Medicare claims were used to examine 3-year outcomes. The analysis includes 461 elderly, fee-for-service, Medicare participants and 1,795 controls. Results Cardiac and non-cardiac hospitalization rates were lower in participants than controls in each program and were statistically significant in MBMI (P < .01). Program costs of $3,801 and $4,441 per participant for the MBMI and Ornish Programs, respectively, were offset by reduced health care costs yielding non-significant three-year net savings per participant of about $3,500 in MBMI and $1,000 in Ornish. A trend towards lower mortality compared with controls was observed in MBMI participants (P = .07). Conclusions Intensive, year-long LMPs reduced hospitalization rates and suggest reduced Medicare costs in elderly beneficiaries with symptomatic coronary heart disease.
Introduction

Cardiovascular diseases, including high blood pressure, stroke, and coronary heart disease (CHD), impose large health and economic burdens in the United States and worldwide through acute illness, disability, lost productivity, and premature death. In the United States in 2008, CHD was responsible for about one in five deaths, 1.2 million new or recurrent heart attacks, and $156 billion in direct and indirect costs. Most CHD deaths (82%) were in individuals aged 65 or older.[1,2] Although major advances have been made in the treatment of CHD, the success of efforts to reduce risk by changing individual behavioral risk factors has been mixed. For example, smoking rates have decreased in adults, while other risk factors for CHD, such as diabetes and obesity, have increased. Medicare has covered cardiac rehabilitation (CR) programs for patients who have had acute myocardial infarctions (AMIs) or have undergone coronary artery bypass graft surgery (CABG) since 1982. [3] Coverage was extended to include percutaneous coronary interventions (PCIs) and heart or heart-lung transplants in 2006.[3] In 1997 however, only 18.7% of eligible Medicare beneficiaries actually received CR following their cardiac events.[4] Cardiac rehabilitation has been shown to reduce mortality in individual randomized controlled trials and in meta-analyses, though most trials were based on care delivered in the 1980s or 1990s and enrolled middleaged white men with moderately severe CHD.[59] More recent studies have largely, though not universally, reinforced these findings. A study of Medicare beneficiaries who received CR following cardiac procedures in 1997, which used propensity-score matching and instrumental variables to control for confounding, verified mortality benefits.[10] Similarly, a study in a national five percent sample of Medicare beneficiaries who received CR between 2000 and 2005 identified a strong dose-response relationship between the number of CR

sessions and the reduced risk of death or MI after four years of follow-up.[11] An observational study in patients who received PCIs between 1994 and 2008 in Olmsted County, Minnesota found significant relationships between participation in CR and lower all-cause mortality, but no effects on rates of subsequent AMIs or the need for cardiac revascularization.[12] Contrary to these positive results, a multicenter randomized controlled trial performed in the United Kingdom between 1997 and 2000 found that "comprehensive" CR following AMI had no important effects on mortality, morbidity, or cardiac risk factors.[13] The Lifestyle Modification Program Demonstration (LMPD) was begun by the Centers for Medicare & Medicaid Services (CMS) in 1999 to examine the effectiveness of intensive and prolonged lifestyle modification programs. Participants were Medicare beneficiaries with symptomatic CHD who enrolled in one of 2 multisite lifestyle modification programs (LMPs): the Dr. Dean Ornish Program for Reversing Heart Disease (Ornish) or the Cardiac Wellness Program of the Benson-Henry Mind Body Medical Institute (MBMI). The demonstration protocol included a one-year active program and an additional year of active follow-up for all participants.[14] This report describes the impacts of these two programs on mortality, recurrence of cardiovascular events, and health care costs in a matched-pair design study that compares participants in each program with up to four matched controls, two of whom had received traditional CR following the qualifying cardiac event and two of whom had not.

Methods
Demonstration Design

Medicare's demonstration was designed to test the premise that aggressive cardiac risk factor reduction may slow, stop or even reverse the progression of CHD, improve health outcomes, and be a cost-effective or costsaving use of Medicare funds. The participating lifestyle modification programs had multiple, geographically dispersed, clinical sites that were serving non-institutionalized individuals with CHD. Both fee-for-service and managed care Medicare beneficiaries were eligible to participate if they met clinical criteria and lived within a one hour drive of the clinical site. The demonstration included a one-year active treatment program with monitored exercise, nutrition counseling, stress management, and group support beginning with an intensive three-month period followed by nine months of less frequent sessions and more emphasis on home maintenance of healthy behaviors. Participants were contacted over the second year, but program visits were no longer funded under the demonstration. Adherence to protocols was closely monitored by the Delmarva Foundation for Medical Care, Inc. a CMS contracted Quality Improvement Organization headquartered in Easton, MD. The two programs had many similarities, but also important differences. Ornish advocated a vegetarian diet limited to 10% of calories from fat and used yoga, stretching, and relaxation techniques to reduce stress. MBMI recommended the American Heart Association's diet that includes 30% of calories from fat and placed particular emphasis on training and practice in the relaxation response, mindfulness, cognitive-behavioral skills, and positive psychology and spiritual connectedness in small group settings to elicit social support. Ornish was more intensive than MBMI in terms of both the duration and frequency of sessions. Both programs included monitored exercise and group support. summarizes the programs' key components.
Table I. Components of the lifestyle modification programs Component Ornish MBMI

Diet and nutrition counseling Aerobic exercise Stress management Small group support Program sessions months 13

Vegetarian diet with 10% calories from fat Condition-appropriate, monitored exercise at least 3 hours per week

Individualized diets aimed at weight control and reducing cardiac risk factors. Target - AHA diet with 30% calories from fat. Same

Daily practice in stretching, relaxation, Training and practice in the relaxation response imagery, and meditation and cognitive behavioral skills Yes 4-hour group sessions:3 in week 1; 2 in weeks 211, and 3 in week 12 Yes One 3-hour session per week

Program sessions months 412

12 or 24 weekly 2-hour sessions or 40 weekly 4-hour sessions depending on 3-hour sessions twice a month medical risk

Notation: AHA denotes American Heart Association.

Participant Enrollment and Personal Characteristics

Ornish included twelve participating clinical sites, and MBMI included five sites. A total of 580 participants entered the demonstration between June 1, 2000 and February 28, 2006, including 440 from MBMI sites and 140 from Ornish sites. Of these, 79% were enrolled in fee-for-service Medicare and 21% in health maintenance organizations. All participants had at least one qualifying cardiac event, including acute myocardial infarction (AMI), cardiac artery bypass surgery (CABG), or percutaneous coronary intervention (PCI) (angioplasty with or without stent placement) during the 12 months preceding enrollment, or had stable angina pectoris. Negotiated maximum allowed charges per participant for the full program were $5,650 for Ornish and $4,800 for MBMI. Medicare paid 80% of these amounts. Payments by Medicare were linked to continued active participation during each 3-month period. This report, which depends on Medicare claims data, includes only the 461 feefor-service beneficiaries (324 in MBMI and 137 beneficiaries in Ornish).
Matched Controls

Four matched controls were sought for each participant from Medicare claims data, 2 of whom had received traditional CR within 12 months following their cardiac events (CR controls) and 2 of whom had not (non-CR controls). Matching was based on the type of cardiac event, the time between the event and enrollment into the demonstration, age, gender, location of residence, and diagnostic cost group (DxCG) risk scores. The DxCG score is a continuous case-mix adjustment score based on the patient's inpatient diagnoses, including comorbidities over the past year. The DxCG score is used by CMS for risk adjustment in Medicare+Choice plans. In addition, controls had to live within a 25-mile radius of the program site attended by the participant if it was in an urban area or within 50 miles if it was in a rural areahence, controlling for geographic location. Controls, like participants, had to have been enrolled in fee-for-service Medicare Parts A and B during the year preceding the demonstration and the following three years. Matching was exact on gender and qualifying event, within a decade on age, and to the nearest month for the time lag between the onset of the qualifying event and enrollment into the LMPD or CR program. To find the closest matches for each participant, we used a combined indicator of DxCG scores and age, where a difference of one year of age was weighted equivalent to a difference of 0.1 point in DxCG score (ie, a 10% change in risk). This weighting was chosen to approximate the increase in risk of hospitalization[15] and death[16] per year of increasing age in the elderly. The pool for matching was much larger for non-CR controls than CR controls. Each participant was matched to two CR and two non-CR controls in 421 cases (91%).
Data Sources

Primary data sources were Medicare's National Claims History files, standard analytic files, and Medicare master enrollment files for years 1998 to 2008. Medicare's master enrollment database included information on date of birth, sex, date of death (where applicable), residence zip code, Medicare eligibility over time, and group health plan membership. CMS's payment file for the LMPD was used to sum quarterly payments to sites for each LMPD participant.
Outcome Measures

Outcome measures were (1) mortality rates during the three post-enrollment years; (2) total hospitalizations; (3) hospitalizations with a cardiac-related principal discharge diagnosis; and (4) Medicare-paid costs of care. Survival data were unavailable in Medicare files for 22 beneficiaries (12 MBMI participants, 1 Ornish participant, and 9 controls). In our analysis, we made the conservative assumption that participants without vital information had died and that their deaths were equally distributed among the three years of follow-up. Controls without vital information were assumed to be alive at end of the follow-up. For hospitalizations, we assumed that patients who had more than one claim for inpatient services within a 24-hour period had been transferred from another hospital and classified them as having had a single hospitalization. Reported costs are amounts paid by Medicare for medical services from 1 year prior to enrollment (or the pseudo-enrollment date for controls) and up to 3 years following enrollment or until death. For LMPD participants, total costs include payments made by CMS to the demonstration sites where lifestyle modification services were performed. Because of limitations in retrieving data from the CMS Data Extraction System, data for skilled nursing facilities

and home health agencies were available only through 2004 and durable medical equipment expenditures were imputed. Thereafter, costs were estimated using regression models based on the costs of inpatient, outpatient, and physician visits during each year.
Data Analysis

Mortality rates and total and cardiac hospitalizations were first examined descriptively. Then, multiple regressions using hierarchical linear models were used to determine the effects of the MBMI and Ornish programs on hospitalizations, controlling for historical hospitalizations, type of qualifying event, gender, and age at the time of enrollment. Since most lifestyle participants had 2 CR and two non-CR controls, models were used to address the clustering of these five observations. The effects of lifestyle programs on the numbers and types of hospitalizations were calculated from descriptive results and through coefficients to compare participants with their respective CR and non-CR controls. Results are presented as average annual differences. The differences derived from regression models were tested by linearly combining the associated coefficients from the model. 2 tests were used to examine differences in hospitalization rates and mortality rates among groups. Costs were compared between participants and CR and non-CR control groups adjusting for key independent variables. These were (1) medical costs in the year before enrollment; (2) the participant's treatment group; (3) type of qualifying cardiac event; and (4) the participant's gender and age. Interactions among variables were also included. Cost results are presented as average annual differences between each lifestyle program and its respective controls.
Funding Sources

Financial support for our evaluation of the LMPD came from the Centers for Medicare & Medicaid Services under contracts 500-95-0060 to Brandeis University and 500-02-0012 to the Delmarva Foundation for Medical Care and from grant 7R01 DP000339-03 from the Centers for Disease Control to the Benson-Henry Mind Body Program at the Massachusetts General Hospital to support data analysis.

Results
Baseline Characteristics

Of the 580 Medicare beneficiaries who participated in the demonstration, 461 (79%) were enrolled in fee-forservice Medicare and are the focus of this report. Sociodemographic and clinical characteristics of fee-forservice participants are summarized in . Their mean age was 72 years, two thirds were men, more than 90% were white, 60% had received at least some college education, and 38% had completed a college degree or more.
Table II. Characteristics of fee-for-service participants in the lifestyle modification programs and their controls MBMI Non-CR CR MBMI(N = controls (N = controls(N = 324) 626) 627) P Non-CR controls(N = 272) Ornish CR Ornish(N = controls(N = 137) 270) P

Characteristic

Age (mean SD) Gender (% male) Race/ethnicity White Non-white Qualifying event AMI AP CABG

72.6 5.5 65.0%

72.7 6.1 65.4%

72.2 5.2 64.8%

.38 .98

71.8 5.2 65.8%

72.4 5.6 65.6%

71.3 5.3 65.7%

.11 1.00

89.4% 10.6%

93.4% 6.6%

94.2% 5.8%

.02

97.3% 2.7%

97.0% 3.0%

98.3% 1.7%

.75

14.5% 17.7% 31.6%

14.4% 16.8% 32.7%

14.2% 17.6% 32.7%

1.00 19.1% 25.7% 18.4%

18.9% 25.6% 18.5%

19.0% 26.3% 18.2%

1.00

PCI

36.1%

36.2%

35.5%

36.8%

37.0%

36.5%

Note: Missing observations are not included in statistical comparisons of participants in the Benson-Henry Mind Body Institute (MBMI) or Ornish participants. AP, Angina pectoris. More than one-third of participants (35.8%) had received percutaneous coronary interventions (PCIs); 28.4% had undergone CABG; 15.6% had had an AMI with no subsequent surgical procedure; and 20.2% had stable angina pectoris. Baseline cardiovascular disease diagnoses differed between participants in the two programs. Ornish included larger proportions of participants with stable angina pectoris (26.3% vs. 17.6%) and AMIs (19.0% vs. 14.2%), while MBMI included more who had received CABG surgery as their qualifying event (32.7% vs. 18.2%). Over 91% of participants were matched with two CR and two non-CR controls, and the remainder were matched with at least one of each. At baseline, 87% of participants were receiving an antilipemic agent and nearly 80% were receiving a beta-blocker.
Hospitalizations

Hospitalization rates are summarized in . The high rates of cardiac-related hospitalizations during the year before the demonstration reflect the study's eligibility and matching criteria. During the active intervention and follow-up years, total, cardiac, and non-cardiac hospitalizations were lower in MBMI participants than their controls for each comparison (P < .001). Ornish showed a similar pattern with significantly lower hospitalization rates for non-cardiac diagnoses (P = .002) and all causes (P = .02), and a non-significant reduction in hospitalizations for cardiac diagnoses (P = .21). The difference-in-difference analysis showed that participants in MBMI had 46% fewer total and 50% fewer cardiac hospitalizations than controls, while corresponding reductions in Ornish participants were 30% and 15%.
Table III. Annual cardiac and non-cardiac hospitalization rates per person in lifestyle program participants and controls MBMI Year Non-CR control CR control MBMI F value Non-CR control Ornish CR control Ornish F value

Total hospitalizations Year 0 Average(years 1 3) 0.97 0.46 1.03 0.46 0.87 0.25 3.19* 17.58 0.96 0.42 0.84 0.40 0.72 0.29 2.36 4.05*

Cardiac-related hospitalizations Year 0 Average(years 1 3) 0.79 0.20 0.85 0.19 0.72 0.10 4.38* 11.83 0.66 0.16 0.73 0.19 0.58 0.15 1.74 1.57

Non-cardiac-related hospitalizations Year 0 Average(years 1 3) 0.18 0.27 0.18 0.27 0.16 0.15 0.27 9.49 0.30 0.26 0.11 0.21 0.13 0.14 6.68 6.21

CR denotes cardiac rehabilitation. *P < .05. P < .01. P < .001.

Multiple regression analyses showed similar patterns (). Compared with controls, MBMI participants had significantly fewer total, cardiac, and non-cardiac hospitalizations than controls. Findings in Ornish participants were similar in direction but less significant statistically due, in part, to smaller sample sizes. Overall, annual absolute reductions in hospitalizations per participant compared with CR and non-CR controls, respectively, were 0.19 and 0.20 for MBMI and 0.10 and 0.11 for Ornish.
Table IV. Effects of lifestyle modification programs on hospitalization rates per person adjusted for differences in patient characteristics Lifestyle program compared with controls Total hospitalizations per year Cardiac-related hospitalizations per year Non cardiac-related hospitalizations per year

MBMI program Non-CR control CR control Ornish program Non-CR control CR control 0.11 0.10* 0.01 0.04 0.10 0.07 0.20 0.19 0.09 0.08 0.11 0.11

CR denotes cardiac rehabilitation. *P < .05. P < .01. P < .001.

Medicare Costs

Mean Medicare payments per person-year, excluding the costs of the lifestyle modification programs, were similar in participants and controls for each program during the year before enrollment, indicating that matching was successful with respect to prior care received (Figure 1). Participants in MBMI had higher average baseline costs than Ornish participants (t = 3.28, P < .001) due, principally, to the higher proportion who had received CABG surgery as their qualifying cardiac event. Mean annual Medicare costs during the program year were lower in participants than controls for each program. Unadjusted cost reductions were $2,849 ( P < .05) and $3,290 (P < .001) in matched CR and non-CR controls, respectively, for MBMI and $2,332 (P = .19) and $869 (P = .25) in Ornish. Corresponding mean cost reductions during the two follow-up years after the intervention year were $5,076 and $3,600 for MBMI and $2,789 and $3,818 for Ornish.

Figure 1.

Medicare cost per participant per year of MBMI and Ornish participants and their controls. Mean program costs were $3,801 for MBMI and $4,441 for Ornish program participants and exceeded Medicare cost savings in Year 1. Net savings were achieved by both programs; however, over 3 years of follow-up, with cumulative savings of $4,122 and $3,087 compared with CR and non-CR controls, respectively, in MBMI participants and $551 and $245 in Ornish participants. Cost savings in Ornish were not statistically significant, however, because of high variations in costs. Multiple regression analyses of subsequent Medicare costs, controlling for Medicare costs in the year prior to participation in the lifestyle program, found significant positive relations with prior year medical care costs in both programs, patient age in MBMI, and a diagnosis of angina in Ornish. Lower subsequent costs occurred in patients receiving CABG in MBMI. Controlling for patient-level variables, lifestyle program participants had lower costs compared with non-CR controls, but not CR controls, in both programs. This finding was statistically significant only for MBMI, however. Three-year cumulative, regression-adjusted, net savings were $3,509 compared with CR controls and $3,562 compared with non-CR controls for MBMI and $1,047 and $1,025, respectively, for Ornish. Three-year savings for each program are statistically significant (P < .001 for MBMI and P < .05 for Ornish).
Mortality

Mortality rate trends were lower in MBMI program participants than each control group at the end the active intervention year and after each year of follow-up, even using the conservative assumption that persons with missing vital status data were dead. After year 1, the mortality rate was 1.5% in MBMI program participants compared with 2.5% and 4.2%, respectively, in CR and non-CR controls; after year 3, comparable figures were 6.2% in MBMI participants, 10.5% in CR controls, and 11.0% in non-CR controls. These mortality differences for MBMI reached borderline significance (P = .08) (Figure 2). In Ornish, mortality trends were in favorable directions but not statistically significant. Residual case-mix differences between participants and controls may contribute to these mortality results.

Figure 2.

Mortality in lifestyle modification program participants and controls. Note: MBMI is on left, Ornish on right. Vital status data were missing for 12 MBMI participants, one Ornish participant, and 9 controls. Conservatively, participants with missing vital status were assumed to be dead, and controls were assumed to be alive.

Discussion
Our study examined the clinical and economic benefits of two intensive, year-long, lifestyle modification programs in elderly Medicare beneficiaries who had had a cardiac event (AMI, CABG, or PCI) within the preceding 12 months or stable angina pectoris with documented myocardial ischemia. We assessed 3-year hospitalization rates, Medicare costs, and mortality. Comparisons are between MBMI and Ornish participants and matched controls who received either Medicare-supported CR or no CR. Both programs emphasized monitored physical exercise and nutritional counseling. Major differences from traditional CR programs were the greater intensity and longer durations of active interventions and emphases on the importance of stress reduction and small group support. Major findings were reductions in cardiac and non-cardiac hospitalizations, modest net savings in Medicare costs for participants compared with matched controls in each program, and trends toward lower mortality rates in MBMI participants. These findings are present after adjusting for risk factor differences between participants

and controls using the data available in Medicare claims records. Incomplete matching cannot be fully excluded, however, because of the limitations of Medicare claims data in assessing the severity of cardiac disease and the possible exclusion of high-risk patients by the demonstration's eligibility criteria, and differences in motivation and adherence to medical regimens. Benefits of the lifestyle modification programs are supported by the favorable changes in cardiac risk factors and cardiac function that were achieved.[17] Systolic blood pressure, total serum cholesterol, and cardiac functional capacity all improved during the intense period of the intervention, and improvements were maintained or improved further at 12 and 24 months in participants with active follow-up. Reductions in hospitalizations may reflect risk factor reductions, increased exercise, and the emphasis placed on stress reduction and small group support. Stress, depression, and lack of social support have been associated with higher risks of complications from CHD in several studies.[1822] MBMI emphasized practice of the relaxation response during formal sessions plus daily practice at home, while Ornish included regular yoga and stress reduction sessions while encouraging practice at home. Small group support and spiritual connectedness were important objectives of both programs. Participation in these lifestyle programs was associated with statistically significant net savings in Medicare costs over the three-year period of the study of about $3,500 and $1,000 per participant for MBMI and Ornish, respectively. The study's findings are consistent with meta-analyses of randomized controlled trials that have demonstrated lower mortality and better functional outcomes in CR participants than non-CR controls.[5,6,812,23] Mortality trends shown in Figure 2 favor CR recipients over non-CR controls, even though 3-year follow-up results are not statistically significant. Important differences from previous studies are that this Medicare demonstration was limited to elderly Medicare beneficiaries (mean 72 years), included a larger proportion of women (35 percent), and represented advancements in cardiac care provided in recent years (20022008). Most trials reported in meta-analyses focus on care delivered before 2000 that included younger men and few women and were conducted before recent improvements in cardiac procedures and reductions in risks due to the widespread use of medications such as statins and -blockers. The limitations of the study need to be acknowledged. First, matching of lifestyle program participants to controls was based on factors available in Medicare claims data including age, gender, race, cardiac procedures and diagnosis, and the presence of comorbidities. We used the DxCG score, a well-standardized and widely used approach by Medicare to adjust costs of care for the presence and severity of clinical conditions. This approach does not fully adjust for the severity of underlying cardiac or comorbid diseases, cardiac risk factors, or the intensity of concurrent medical therapy. Nevertheless, the large pool of non-CR controls for selection to match LMPD participants allowed for close matching on characteristics in claims data. However, supplemental data for a subset of participants and controls suggested that participants were more highly educated than controls, with 38% having graduated from college compared with 32% of controls with CR and 22% of those with no CR.[24] However, there is no unanimous agreement that higher education contributes to positive behavior changes for better survival among patients with cardiovascular diseases. [25] Second, Medicare's eligibility criteria for the demonstration excluded high-risk individuals who had involvement of the left main coronary artery, severe congestive heart failure, or impaired cognitive function. It is unlikely, however, that the matched controls included significant proportions of such cases. Third, the statistical power of the analysis was limited by the relatively small size of the study population (324 fee-for-service beneficiaries in MBMI and 137 in Ornish). Data from participants who were enrolled in health maintenance organizations could not be included because Medicare claims data were not available for them. The policy implications of our findings emphasize the high priority that needs to be given to innovative approaches to improving health outcomes while controlling or reducing health care costs. Intensive lifestyle modification programs such as MBMI and Ornish fall into this category. The Medicare Improvements for Patients and Providers Act of 2008[26] extended coverage under Medicare Part B beginning on January 1, 2010 to include intensive cardiac rehabilitation and provide reimbursement for substantially more hours than standard comprehensive rehabilitation. Our study's findings support the benefits of extending enhanced insurance coverage to additional qualified programs that can be shown to reduce hospitalization rates, reduce costs, and, possibly, reduce mortality.
References

1. 2.

Roger VL, Go AS, Lloyd-Jones DM, et al. Heart disease and stroke statistics2011 update: a report from the American Heart Association. Circulation 2011;123(4):e18e209. Rosamond W, Flegal K, Furie K, et al. Heart disease and stroke statistics 2008 update: a report from the American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circulation 2008;117(4):e25e146. Centers for Medicare and Medicaid Services. Medicare national coverage determinations manual. www.cms.gov/manuals/downloads/ncd103c1_Part1.pdf, 2010 Accessed: 22 Aug 2011. Suaya JA, Shepard DS, Normand SL, et al. Use of cardiac rehabilitation by Medicare beneficiaries after myocardial infarction or coronary bypass surgery. Circulation 2007;116(15):1653 62. Clark AM, Hartling L, Vandermeer B, et al. Meta-analysis: secondary prevention programs for patients with coronary artery disease. Ann Intern Med 2005;143(9):65972. Jolliffe JA, Rees K, Taylor RS, et al. Exercise-based rehabilitation for coronary heart disease. Cochrane Database Syst Rev 2001;1:CD001800. O'Connor GT, Buring JE, Yusuf S, et al. An overview of randomized trials of rehabilitation with exercise after myocardial infarction. Circulation 1989;80(2):23444. Oldridge NB, Guyatt GH, Fischer ME, et al. Cardiac rehabilitation after myocardial infarction. Combined experience of randomized clinical trials. JAMA 1988;260(7):94550. Taylor RS, Brown A, Ebrahim S, et al. Exercise-based rehabilitation for patients with coronary heart disease: systematic review and metaanalysis of randomized controlled trials. Am J Med 2004;116(10):68292.

3. 4. 5. 6. 7. 8. 9.

10. Suaya JA, Stason WB, Ades PA, et al. Cardiac rehabilitation and survival in older coronary patients. J Am Coll Cardiol 2009;54(1):2533. 11. Hammill BG, Curtis LH, Schulman KA, et al. Relationship between cardiac rehabilitation and long-term risks of death and myocardial infarction among elderly Medicare beneficiaries. Circulation 2010;121:6370. 12. Goel K, Lennon RJ, Tilbury RT, et al. Impact of cardiac rehabilitation on mortality and cardiovascular events after percutaneous coronary intervention in the community. Circulation 2011;123:2344 52. 13. West RR, Jones DA, Henderson AH. Rehabilitation after myocardial infarction trial (RAMIT): multicentre randomized controlled trial of comprehensive cardiac rehabilitation in patients following acute myocardial infarction. Heart 2011, http://dx.doi.org/10.1136/heartjrl-2011300302. 14. Shepard DS. Lifestyle Modification to Control Heart Disease: Evidence and Policy. Sudbury, MA: Jones & Bartlett; 2010. 15. Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality (US). HCUP Facts and Figures: Statistics on Hospital-based Care in the United States, 2007. http://www.ncbi.nlm.nih.gov/books/NBK53003, 2009 Accessed: 8 June 2012. 16. US Centers for Disease Control, National Center for Health Statistics. Worktable 23R. Death Rates by 10-Year Age Groups. http://www.cdc.gov/nchs/data/dvs/MortFinal2007_Worktable23r.pdf, 2007 Accessed: 8 June 2012. 17. Shepard DS, Stason WB, Strickler G, et al. Evaluation of the Medicare Lifestyle Modification Program Demonstration. Centers for Medicare and Medicaid Services. http://people.brandeis.edu/~shepard/LMPD_FER.zip, 2009 Accessed: 6 Jan 2013. 18. Hearn MD, Murray DM, Luepker RV. Hostility, coronary heart disease, and total mortality: a 33-year follow-up study of university students. J Behav Med 1989;12(2):10521.

19. Hemingway H, Marmot M. Evidence based cardiology: psychosocial factors in the aetiology and prognosis of coronary heart disease. Systematic review of prospective cohort studies. BMJ 1999;318(7196):14607. 20. Jin H, Folsom D, Sasaki A, et al. Increased Framingham 10-year risk of coronary heart disease in middle-aged and older patients with psychotic symptoms. Schizophr Res 2011;125(23):2959. 21. Kubzansky LD, Park N, Peterson C, et al. Healthy psychological functioning and incident coronary heart disease: the importance of self-regulation. Arch Gen Psychiatry 2011;68(4):4008. 22. Stansfeld SA, Fuhrer R, Shipley MJ, et al. Psychological distress as a risk factor for coronary heart disease in the Whitehall II Study. Int J Epidemiol 2002;31(1):24855. 23. Lavie CJ, Milani RV. Cardiac rehabilitation and exercise training in secondary coronary heart disease prevention. Prog Cardiovasc Dis 2011;53(6):397403. 24. Shepard DS, Stason WB, Strickler G, et al. Evaluation of Lifestyle Modification and Cardiac Rehabilitation in Medicare Beneficiaries Centers for Medicare and Medicaid Services. http://www.cms.gov/Medicare/DemonstrationProjects/DemoProjectsEvalRpts/downloads/LMPD_Summary_of_Findings.pdf, 2009 Accessed: 6 Jan 2013. 25. Chan R, Gordon N, Chong A, et al. Influence of socioeconomic status on lifestyle behavior modifications among survivors of acute myocardial infarction. Am J Cardiol 2008;102(12):1583 8. 26. Government Printing Office. Public Law No. 110275;74. Reg. 61,738,520,62,004. 2008.

Acknowledgements The authors thank Clare L. Hurley for assistance in preparing this manuscript. The principal funding organization (CMS) implemented the demonstration and approved the design of its evaluation by Brandeis University. Data collection and adherence to lifestyle modification protocols were monitored by the Delmarva Foundation for Medical Care, Inc. a CMS-contracted Quality Improvement Organization in Easton, MD. We are particularly grateful to Roxanne Rodgers, PMP, RN and William J. Oetgen, MD, MBA for ensuring the high quality of collected data. The findings and conclusions of the paper are those of the authors alone. Am Heart J. 2013;165(5):785-792. 2013 Mosby, Inc. Copyright Mosby-Year Book, Inc. All rights reserved.