Author + information
- Received January 17, 2015
- Accepted February 26, 2015
- Published online May 19, 2015.
- Krishna G. Aragam, MD, MS∗∗ (, )
- Dadi Dai, PhD†,
- Megan L. Neely, PhD†,
- Deepak L. Bhatt, MD, MPH‡,
- Matthew T. Roe, MD, MHS†,
- John S. Rumsfeld, MD, PhD§ and
- Hitinder S. Gurm, MBBS‖
- ∗Cardiology Division, Massachusetts General Hospital, Boston, Massachusetts
- †Duke Clinical Research Institute, Duke University Medical Center, Durham, North Carolina
- ‡Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts
- §Denver Veterans Affairs Medical Center, Denver, Colorado
- ‖Division of Cardiovascular Medicine, University of Michigan Medical Center, Ann Arbor, Michigan
- ↵∗Reprint Requests and correspondence:
Dr. Krishna G. Aragam, Cardiology Division, Massachusetts General Hospital, 55 Fruit Street, GRB 8-852, Boston, Massachusetts 02114.
Background Rates of referral to cardiac rehabilitation after percutaneous coronary intervention (PCI) have been historically low despite the evidence that rehabilitation is associated with lower mortality in PCI patients.
Objectives This study sought to determine the prevalence of and factors associated with referral to cardiac rehabilitation in a national PCI cohort, and to assess the association between insurance status and referral patterns.
Methods Consecutive patients who underwent PCI and survived to hospital discharge in the National Cardiovascular Data Registry between July 1, 2009 and March 31, 2012 were analyzed. Cardiac rehabilitation referral rates, and patient and institutional factors associated with referral were evaluated for the total study population and for a subset of Medicare patients presenting with acute myocardial infarction.
Results Patients who underwent PCI (n = 1,432,399) at 1,310 participating hospitals were assessed. Cardiac rehabilitation referral rates were 59.2% and 66.0% for the overall population and the AMI/Medicare subgroup, respectively. In multivariable analyses, presentation with ST-segment elevation myocardial infarction (odds ratio 2.99; 95% confidence interval: 2.92 to 3.06) and non–ST-segment elevation myocardial infarction (odds ratio: 1.99; 95% confidence interval: 1.94 to 2.03) were associated with increased odds of referral to cardiac rehabilitation. Models adjusted for insurance status showed significant site-specific variability in referral rates, with more than one-quarter of all hospitals referring <20% of patients.
Conclusions Approximately 60% of patients undergoing PCI in the United States are referred for cardiac rehabilitation. Site-specific variation in referral rates is significant and is unexplained by insurance coverage. These findings highlight the potential need for hospital-level interventions to improve cardiac rehabilitation referral rates after PCI.
Participation in cardiac rehabilitation after myocardial infarction and/or myocardial revascularization is associated with better clinical outcomes, including lower all-cause mortality, cardiac-specific mortality, re-hospitalization, revascularization, and cardiovascular disease–related functional disability and mood disorders in patients with coronary artery disease (CAD) (1–5). Numerous national and international guidelines endorse cardiac rehabilitation for chronic stable angina, and after acute coronary syndromes, percutaneous coronary intervention (PCI), and cardiac surgery (6–13). In addition, referral to cardiac rehabilitation after acute myocardial infarction (AMI) or PCI represents a newly reported performance measure instituted by the Centers for Medicare and Medicaid Services (CMS) in 2014 (14).
Lack of referral to cardiac rehabilitation is an important impediment to participation in rehabilitation programs (15,16). Although previous regional analyses demonstrated low rates of referral after PCI, it is unclear whether post-PCI referral to cardiac rehabilitation remains deficient in contemporary interventional practice across the United States (17). We used the National Cardiovascular Data Registry (NCDR) to assess cardiac rehabilitation referral rates and both patient- and hospital-level factors associated with referral in patients who underwent PCI in the United States. In addition, we evaluated insurance coverage, which is a frequently cited barrier to cardiac rehabilitation referral, in relation to national referral patterns.
Data were analyzed from the NCDR’s CathPCI registry program. The program is jointly sponsored by the American College of Cardiology and the Society for Cardiovascular Angiography and Interventions, and has been described previously (18,19). Data on consecutive patients who underwent PCI, as collected from cardiac catheterization laboratories across the United States, include baseline demographic and clinical characteristics, such as admission symptoms and risk factors, as well as detailed procedural and angiographic information. Discharge medications and in-patient referral to cardiac rehabilitation are recorded for all patients who are discharged. Data collected using CathPCI registry version 4 were used for this analysis (20).
Waiver of written informed consent and authorization for this study was granted by Chesapeake Research Review Incorporated.
The study population was composed of all consecutive patients who underwent PCI at the participating hospitals between July 1, 2009 and March 31, 2012. Patients were excluded if they died in-hospital, or were discharged to nursing home facilities, acute care hospitals, hospice, or who left against medical advice.
The primary outcome was referral to cardiac rehabilitation. Overall rates of referral to cardiac rehabilitation were determined for the total study period, and temporal trends were assessed quarterly.
Baseline characteristics were compared for patients who were referred and not referred to cardiac rehabilitation using Pearson’s chi-square tests for categorical variables and Wilcoxon rank sum tests for continuous and ordinal variables. We also assessed other quality measures at hospital discharge, including discharge on aspirin, beta-blockers, angiotensin-converting enzyme (ACE) inhibitors/angiotensin-receptor blockers (ARBs) if the ejection fraction (EF) was <40%, statins, and any P2Y12 antagonist. McNemar’s test for paired proportions was used to compare rates of referral to cardiac rehabilitation with rates of the previously mentioned quality-of-care measures.
Multivariable logistic regression was used to estimate the effect of each statistically significant univariate predictor on the primary outcome of referral to cardiac rehabilitation. A random effects model was fit to assess the association between patient- and hospital-level covariates and rehabilitation referral rates by deconstructing each covariate into its within- and across-center components.
To assess for interhospital variation in cardiac rehabilitation referral rates, a logistic regression model was fit with a random effect on the hospital, and the variance component of the model was assessed. A variance significantly different from zero denoted statistically significant variation in referral rates between sites. To evaluate the association between insurance status and observed interhospital variation in referral rates, the model was adjusted for insurance type, and the variance was reassessed. A 95% bootstrap confidence interval (CI) was constructed for the relative change in variance between the adjusted and unadjusted models to determine the statistical significance of any noted difference. The model was further adjusted for available hospital-level characteristics (number of beds, PCI volume, urban or rural location, teaching or nonteaching status, and geographic region) with bootstrap analysis to assess for differences in variance.
Analyses were performed using SAS version 9.2 (SAS Institute, Cary, North Carolina).
Acute myocardial infarction/medicare subgroup analysis
To further evaluate the relationship between insurance status and cardiac rehabilitation referral rates, an analysis was performed that evaluated the trends and predictors of referral to cardiac rehabilitation among PCI patients with Medicare and who presented with AMI as a representative subset of patients with definitive insurance coverage for cardiac rehabilitation. Univariate and multivariate analyses, as outlined previously, were repeated for this subset of the study population.
Rates of referral to cardiac rehabilitation and other acute myocardial infarction quality-of-care measures
The sample study population consisted of 1,432,399 patients who underwent PCI between July 1, 2009 and March 31, 2012 at 1,310 hospitals across the United States. The overall cardiac rehabilitation referral rate for all patients during the 2-year study period was 59.2%. There was minimal change in referral rates over the study period, with quarterly rates ranging from 57.9% to 61.2% (Figure 1). In contrast, rates were significantly higher for other AMI quality-of-care indexes, including discharge on aspirin (97.5%), statins (89.8%), beta-blockers (84.8%), ACE inhibitors/ARBs if EF was <40% (79.6%), and P2Y12 antagonists (97.0%) (all p < 0.001) (Figure 2).
Univariate and multivariate predictors of referral
Baseline characteristics of PCI patients who were and were not referred to cardiac rehabilitation are presented in Table 1. There were several statistically significant, but clinically small differences between the 2 groups. Patients referred to cardiac rehabilitation were generally younger, men, white, had private insurance carriers, were current or recent smokers, and had higher body mass index values. Referred patients were less likely to have various medical co-morbidities, including hypertension, dyslipidemia, cerebrovascular disease, peripheral artery disease, chronic lung disease, and diabetes mellitus. A history of MI, congestive heart failure, previous PCI, previous coronary artery bypass graft (CABG), and previous valve surgery were all slightly less prevalent among patients referred to rehabilitation programs. Notably, a higher proportion of referred patients presented with AMI (non–ST-segment elevation myocardial infarction [NSTEMI] or STEMI) or experienced cardiogenic shock or cardiac arrest within 24 h of the PCI procedure. Among hospital-level characteristics, patients referred to cardiac rehabilitation were more likely to be from hospitals with more beds and that performed a greater volume of PCIs (Table 1).
Factors independently associated with referral to cardiac rehabilitation in the final multivariate model are displayed in Table 2. Among patient-level covariates, presentation with STEMI (odds ratio [OR]: 2.99; 95% CI: 2.92 to 3.06) and NSTEMI (OR: 1.99; 95% CI: 1.94 to 2.03) conferred the highest likelihood of referral to cardiac rehabilitation, followed by presentation with unstable angina (OR: 1.12; 95% CI: 1.10 to 1.14) and peri-procedural MI (OR: 1.42; 95% CI: 1.37 to 1.47). Older age and several medical co-morbidities, including diabetes mellitus, peripheral artery disease, previous congestive heart failure, and chronic lung disease, as well as a history of PCI, CABG, or valve surgery were all associated with a decreased likelihood of referral to cardiac rehabilitation, although the absolute effects of these associations were small (Table 2).
Among institutional characteristics, a greater volume of PCIs, larger hospital size (denoted by number of hospital beds), private/community hospital designation, and particular geographic locations were associated with increased referral to cardiac rehabilitation. Notably, certain hospital-level covariates (i.e., Midwest region [OR: 7.36; 95% CI: 5.08 to 10.67] and private/community hospital [OR: 2.33; 95% CI: 1.34 to 4.05]) demonstrated robust effect sizes compared with most patient-level factors.
Subgroup analysis: Medicare patients presenting with AMI
An analysis was performed, restricted to patients who presented with AMI and who had Medicare coverage (n = 196,214). The 2-year referral rate for this subset of patients was 66.0%, with stable quarterly rates over the study period ranging from 64.4% to 67.4% (Figure 1). Rates of other AMI quality-of-care determinants were the same or higher in the AMI/Medicare subset compared with the overall population.
The predictors of referral in this population were broadly similar to those in the overall cohort and are listed in Online Tables 1 and 2.
Interhospital variation in cardiac rehabilitation referral rates
Two logistic regression models with a random effect on the hospital (one unadjusted and the other adjusted for insurance type) were constructed to assess interhospital variation in cardiac rehabilitation referral rates. For each model, the variance was significantly different from zero (p < 0.0001), which suggested significant variation in hospital-level referral rates whether or not insurance type was considered. In an analysis of 400 bootstrap samples, the ratio of variance components between the adjusted and unadjusted models was 1.002 (95% CI: 1.001 to 1.003), which denoted an increased variation in hospital-level referral rates after adjustment for insurance type, and indicated insurance type as a confounder for hospital-specific effects on referral rates. However, the negligible increase in variation (0.2% on average) suggested that insurance type did not heavily influence the variation in cardiac rehabilitation referral rates observed across hospitals. Figure 3 demonstrates the comparable distributions of referral rates across hospital sites before and after adjustment for insurance type, with the majority of hospital sites displaying referral rates >80% or <20% in each scenario. Site-specific variability in referral rates persisted even after further adjustment for hospital-level characteristics, with more than one-quarter of all hospitals demonstrating referral rates of <20%. The ratio of variance between the unadjusted model and the model adjusted for both insurance status and hospital characteristics was 0.994 (95% CI: 0.980 to 1.003), which suggested that hospital characteristics assessed in our study were not significantly associated with the observed variability in referral rates.
Our analyses demonstrated low cardiac rehabilitation referral rates in the United States between 2009 and 2012, with minimal improvement over the defined 2-year study period and significant variability across hospital sites. Under-referral was widespread across patient types and not predicated on patient risk, as most comorbidities demonstrated only a weak association with referral rates. Similarly, insurance coverage for cardiac rehabilitation had a minimal effect on overall referral rates and interhospital variability in referral patterns. Despite the general notion that lack of insurance coverage is a major barrier to rehabilitation referral, our data suggest that other, unidentified factors are the primary determinants of decreased referral.
The benefits of comprehensive cardiac rehabilitation programs in CAD patients have been well noted, including associated reductions in morbidity and mortality, and improvements in functional status and quality of life; these are presumed to be a result of exercise training, psychological counseling, and a consistent emphasis on preventive strategies (e.g., medication compliance and cardiovascular risk factor modification) (21–26). A number of studies have demonstrated better outcomes in clinical trial and sicker registry populations (2–5). Furthermore, in a recent study of 2,395 PCI patients, participation in cardiac rehabilitation was associated with a 45% to 47% reduction in 5-year all-cause mortality (p < 0.001), which further supports the benefits of cardiac rehabilitation for post-procedural care (27).
Despite the considerable evidence regarding the benefits of cardiac rehabilitation for secondary prevention in CAD, previous studies have demonstrated low rates of participation in rehabilitation programs in the United States (28,29). An analysis of 267,427 elderly Medicare beneficiaries reported that only 13.9% of patients with AMI and 31.0% of those who had undergone CABG surgery participated in cardiac rehabilitation after their index hospitalizations (30). An assessment of potential barriers to participation in cardiac rehabilitation in post-AMI patients cited inpatient referral to cardiac rehabilitation as one of the strongest predictors of program enrollment (OR: 6.14; 95% CI: 2.12 to 17.74) (31). A combined analysis of 10 observational studies with a total of 30,000 CAD patients highlighted insurance coverage, history of previous MI, being English speaking, and admission to a hospital with existing cardiac rehabilitation programs as the four primary determinants of referral to cardiac rehabilitation (32). A regional analysis of post-PCI referral patterns in Michigan displayed a decreased likelihood of referral for most comorbidities, and increased odds of referral among men, younger patients, patients self-identified as white, and patients with more severe or acute presentations, including STEMI (17).
Our contemporary analysis of PCI patients across the United States extends these previous studies in several ways. Compared with previous analyses, we observed that older patients, women, those with the most comorbidities, and those with previous MI or revascularization were less likely to be referred to cardiac rehabilitation, despite data that suggest that these populations might benefit most from rehabilitation programs (28,33,34). However, in our large national sample, the effect sizes of these negative predictors were small, which suggested minimal overall influence of patient factors on rates of referral to cardiac rehabilitation.
Conversely, hospital-level characteristics in our analysis demonstrated robust associations with referral patterns. In particular, Midwest location and private/community hospital designation were associated with marked increases in referral to cardiac rehabilitation. However, a limited number of hospital characteristics were available for inclusion in our analysis. Therefore, it is likely that the aforementioned hospital-level predictors of referral were confounders for other, unmeasured institutional characteristics such as the presence of automated discharge sets, which have been associated with increased cardiac rehabilitation referral rates in previous studies; unfortunately, these data were unavailable for the present analysis (35,36).
Lack of insurance coverage has been reported as a negative predictor of referral to cardiac rehabilitation and has been cited as a major reason for continued low referral rates (37). In our analysis, insurance coverage for rehabilitation programs had a negligible effect on referral patterns. Despite a slight increase in overall referral rates among Medicare patients who presented with AMI, predictors of referral were relatively unchanged. Notably, concurrent Medicaid status conferred a decreased likelihood of referral, whereas concurrent private insurance was associated with increased referral. These data suggested that insurance coverage for cardiac rehabilitation had, at best, a modest effect on referral rates, but that type of insurance might be a slightly stronger predictor of referral patterns. However, insurance type might also be a proxy for socioeconomic status, but these data were unavailable for this study. Referral patterns might be affected by unmeasured confounders such as low income, which was previously reported as a negative predictor of participation in rehabilitation programs (30). Our observation that non-white race was associated with decreased referral might also be confounded by the unavailable data on income and socioeconomic status. However, although participation in cardiac rehabilitation might be affected by income-related factors such as time away from work and distance of rehabilitation center from home, theoretically, referral should not be affected in the setting of available insurance coverage. It is possible that physicians preferentially refer those patients presumed likely to participate in and benefit from rehabilitation programs, instead of referring all PCI patients to cardiac rehabilitation and encouraging it as an integral component of post-procedural, post-MI care (38,39).
The development and implementation of performance measures to improve health care accountability has increased dramatically in the past decade, particularly in the management of AMI, as greater compliance with established care processes has been associated with enhanced clinical outcomes (40,41). Our study corroborated previous work that demonstrated that rates of more established, publicly reported AMI quality-of-care measures were significantly higher than rates of referral to cardiac rehabilitation. This suggested that deficiencies in cardiac rehabilitation referral rates did not reflect a broader lack of quality care for AMI patients (15,17).
The incorporation of referral to cardiac rehabilitation as a new, publicly reported CMS performance measure in 2014 may provide the impetus to improve rates of referral across the country. As with other performance measures of accountability, the prospect of incentives for improved performance and penalties for poor compliance will likely motivate prompt identification of sites with deficient referral rates, and concerted efforts (e.g., incorporation of automated discharge checklists) to improve referral rates at underperforming hospitals (35,42).
First, the CathPCI Registry contained only in-hospital data, such as referral to cardiac rehabilitation before discharge; no follow-up data were available to assess participation in rehabilitation programs or subsequent outcomes. Second, lack of socioeconomic data might confound the noted associations between referral to cardiac rehabilitation and insurance status or race. Third, as outlined previously, the limited availability of institutional characteristics precluded a more detailed assessment of hospital-level predictors of referral. Fourth, geographic proximity to cardiac rehabilitation sites was associated with increased referral patterns; unfortunately, such data were unavailable for geospatial analyses (43). Finally, the study cohort was derived from hospitals that participated in a quality improvement initiative and were more likely to implement evidence-based measures. Accordingly, cardiac rehabilitation referral rates reported in this study might not be representative of the general population, although they likely reflect a conservative overestimate of true national referral rates.
We report low rates of referral to cardiac rehabilitation after PCI in the United States between 2009 and 2012, with >40% of patients not referred by the time of hospital discharge. Our study highlights significant interhospital variability in referral rates and suggests that unidentified, hospital-level factors may have a stronger influence on referral rates than most patient-specific factors, including insurance coverage for rehabilitation programs (Central Illustration). The data provide a useful baseline prior to public reporting by CMS on hospital-specific rates of referral to cardiac rehabilitation. Moreover, the findings support a call to identify and assist underperforming institutions through hospital-level interventions and ongoing efforts to monitor for future improvements in national cardiac rehabilitation referral rates.
COMPETENCY IN MEDICAL KNOWLEDGE: Presentation with AMI and certain institutional variables are more strongly associated with referral of patients for cardiac rehabilitation after PCI than specific patient characteristics or insurance status, but across the United States, rates of referral are relatively low (approximately 60%).
TRANSLATIONAL OUTLOOK: Further investigations are required to identify additional hospital-level factors associated with referral for cardiac rehabilitation of patients undergoing PCI.
The CathPCI Registry is an initiative of the American College of Cardiology Foundation and the Society for Cardiovascular Angiography and Interventions.
Dr. Bhatt is on the advisory board for Elsevier Practice Update Cardiology, Medscape Cardiology, and Regado Biosciences; is on the Board of Directors of the Boston VA Research Institute, Society of Cardiovascular Patient Care; is the Chair of the American Heart Association Get With The Guidelines Steering Committee; has received honoraria from the American College of Cardiology (Editor, Clinical Trials, Cardiosource), Belvoir Publications (Editor in Chief, Harvard Heart Letter), Duke Clinical Research Institute (clinical trial steering committees), Population Health Research Institute (clinical trial steering committee), Slack Publications (Chief Medical Editor, Cardiology Today’s Intervention), and WebMD (Continuing Medical Education steering committees); is the Senior Associate Editor, Journal of Invasive Cardiology; is a member of the Data Monitoring Committees of Duke Clinical Research Institute, Harvard Clinical Research Institute, Mayo Clinic, and the Population and Health Research Institute; has received research grants from Amarin, AstraZeneca, Bristol-Myers Squibb, Eisai, Ethicon, Medtronic, Sanofi, and The Medicines Company; and has taken part in unfunded research at FlowCo, PLx Pharma, and Takeda. Dr. Roe has received research funding from Eli Lilly, Revalesio, Sanofi, American College of Cardiology, American Heart Association, and the Familial Hyperlipidemia Foundation; and has received consulting or honoraria from Eli Lilly, AstraZeneca, Sanofi, Janssen Pharmaceuticals, Merck, Regeneron, and Daiichi-Sankyo. Dr. Gurm has received research funding from the National Institutes of Health and the Agency for Health Care Research and Quality. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- angiotensin-converting enzyme
- acute myocardial infarction
- angiotensin-receptor blocker
- coronary artery bypass graft
- coronary artery disease
- confidence interval
- Centers for Medicare and Medicaid Services
- ejection fraction
- National Cardiovascular Data Registry
- non–ST-segment elevation myocardial infarction
- odds ratio
- percutaneous coronary intervention
- Received January 17, 2015.
- Accepted February 26, 2015.
- 2015 American College of Cardiology Foundation
- Heran B.S.,
- Chen J.M.,
- Ebrahim S.,
- et al.
- Suaya J.A.,
- Stason W.B.,
- Ades P.A.,
- Normand S.L.,
- Shepard D.S.
- Leon A.S.,
- Franklin B.A.,
- Costa F.,
- et al.
- Thomas R.J.,
- King M.,
- Lui K.,
- et al.
- Anderson J.L.,
- Adams C.D.,
- Antman E.M.,
- et al.
- Levine G.N.,
- Bates E.R.,
- Blankenship J.C.,
- et al.
- Eagle K.A.,
- Guyton R.A.,
- Davidoff R.,
- et al.
- ↵Phurrough SE, Salive M, Baldwin J, McClain S, Schott L, Chin J. Decision Memo for Cardiac Rehabilitation Programs (CAG-00089R). Baltimore, MD; Centers for Medicare and Medicaid Services; Baltimore MD, 2006. Available at: http://www.cms.gov/medicare-coverage-database/details/nca-decision-memo.aspx?NCAId=164&NcaName=Cardiac+Rehabilitation+Programs&NCDId=36&ncdver=3&IsPopup=y&bc=AAAAAAAACAAAAA%3D%3D&. Accessed November 23, 2014.
- Fifth Joint Task Force of the European Society of Cardiology
- Stone J.A.,
- Arthur H.M.,
- Suskin N.
- ↵Medicare and Medicaid Programs: Hospital Outpatient Prospective Payment and Ambulatory Surgical Center Payment Systems and Quality Reporting Programs; Electronic Reporting Pilot; Inpatient Rehabilitation Facilities Quality Reporting Program; Revision to Quality Improvement Organization Regulations. Available at: https://www.federalregister.gov/articles/2012/11/15/2012-26902/medicare-and-medicaid-programs-hospital-outpatient-prospective-payment-and-ambulatory-surgical. Accessed on December 20, 2014.
- Brown T.M.,
- Hernandez A.F.,
- Bittner V.,
- et al.
- Grace S.L.,
- Chessex C.,
- Arthur H.,
- et al.
- Roe M.T.,
- Messenger J.C.,
- Weintraub W.S.,
- et al.
- Dehmer G.J.,
- Weaver D.,
- Roe M.T.,
- et al.
- ↵National Cardiovascular Disease Registry (NCDR) CathPCI Registry version 4.4. Available at: https://www.ncdr.com/WebNCDR/docs/public-data-collection-documents/cathpci_v4_datacollectionform_4-4.pdf. Accessed October 12, 2014.
- Marchionni N.,
- Fattirolli F.,
- Fumagalli S.,
- et al.
- Martin B.J.,
- Hauer T.,
- Arena R.,
- et al.
- Goel K.,
- Lennon R.J.,
- Tilbury R.T.,
- Squires R.W.,
- Thomas R.J.
- Witt B.J.,
- Jacobsen S.J.,
- Weston S.A.,
- et al.
- Suaya J.A.,
- Shepard D.S.,
- Normand S.L.,
- Ades P.A.,
- Prottas J.,
- Stason W.B.
- Clark A.M.,
- King-Shier K.M.,
- Duncan A.,
- et al.
- Eagle K.A.,
- Montoye C.K.,
- Riba A.L.,
- et al.