Each year, >2 million Americans are hospitalized for an acute cardiac condition or procedure such as acute myocardial infarction (MI), heart failure (HF), percutaneous coronary intervention (PCI), coronary artery bypass graft surgery (CABG), or heart valve surgery (HVS).1 After discharge, attending outpatient cardiac rehabilitation (CR) is a key step toward full recovery for these patients. However, despite strong evidence supporting the importance of outpatient CR,2–4 only 30% to 35% of the patients attend outpatient CR, partly due to poor inpatient referral and weak facilitation of enrollment.5–8
As part of the initial recovery process, some hospitals provide a formal program of inpatient cardiac rehabilitation (ICR). These programs are responsible for providing patient ambulation, risk factor modification education, and motivation and encouragement to attend outpatient CR2 and are distinct from services provided during admissions to inpatient rehabilitation wards or hospitals. Prior studies reported that activities provided by ICR substantially increase participation in outpatient CR,9–11 improve patient satisfaction,12 and may reduce mortality.13 However, a prior survey suggested that ICR is not universally available,14 and based upon limited regional data from 1986 to 1997, was declining in use.15
Consequently, we sought to describe utilization of ICR in a more contemporary, large, and diverse sample of US hospitals. We aimed to describe current trends in ICR utilization, services provided during an ICR visit, and the patient, hospital, and regional factors that influence the receipt of ICR. We hypothesized that, similar to outpatient CR, overall utilization of ICR would be low and that substantial variation in care would exist across the United States.
STUDY DESIGN AND SETTING
We identified patients discharged from US hospitals that contribute to the Premier Healthcare Alliance Inpatient Database, which has been previously described.16 In brief, the database contains detailed administrative data from a geographically and structurally diverse group of >500 US hospitals representing approximately 15% to 20% of inpatient US hospitalizations. Unlike claims databases that contain only sociodemographic, diagnostic, and selected procedure codes assigned at the time of discharge, the Premier inpatient database contains date-stamped hospital service codes for all medications, procedures, diagnostic tests, and therapeutic services. Because the data are fully de-identified, the institutional review board at the Baystate Medical Center determined that this study did not meet the federal definition of research of human subjects.
PATIENT AND HOSPITAL FACTORS
We included all hospitalized patients between January 2007 and June 2011 with an acute cardiac condition who, based upon Medicare insurance guidelines,17 , 18 were eligible to attend outpatient CR (MI, HF, CABG, HVS, and PCI). Although HF was not a covered indication during the time frame of this study, we included patients with a principal diagnosis of HF as a baseline measure because CR for HF became a covered indication for CR by Medicare in February 2014 and the majority of private insurance companies now reimburse CR for patients with HF.18 , 19 We included patients with the following International Classification of Diseases, Ninth Revision (ICD-9) principal diagnosis codes: 410.x for MI and 428.x; 402.01; 402.11; 402.91; 404.03; 404.11; 404.13; 404.91; and 404.94 for HF. Also included were patients with an ICD-9 procedure code (either primary or secondary) of 36.1x; 35.1x; 35.2x; or 36.06, 36.07, or 36.09 for patients with CABG, HVS, or PCI, respectively.
Because principal diagnoses and procedures are not mutually exclusive, we categorized patients into 1 of 3 mutually exclusive groups: surgical, PCI, and medical. The surgical group included patients with CABG, HVS, or combined CABG + HVS regardless of whether they had an MI or PCI. The PCI group included patients with elective PCI, urgent PCI (urgent hospital admission for PCI but without evidence for MI), or PCI + MI. The medical group included patients with medically managed MI and HF who did not have a PCI or cardiac surgery during the hospital admission. We used 3 strata (surgical, PCI, medical) for the main analysis and reporting and also reported on the full 8-level strata (CABG, valve, CABG + valve, elective PCI, urgent PCI, PCI + MI, MI, and HF) in supplemental online tables.
Demographic data were recorded including age, gender, race/ethnicity, and insurance status for each admission and computed a comorbidity score as described by Gagne et al.20 We calculated 29 individual comorbidity indicators on the basis of methods developed by Elixhauser et al21 using software provided by Healthcare Costs and Utilization Project of the Agency for Healthcare Research and Quality. Hospital characteristics included size based on number of beds, teaching status, urban location, census region, and surgical volume.
Our primary outcome was the presence of ≥1 inpatient service codes for any kind of ICR, which we grouped by their content, differentiating them by CR exercise, CR education, and CR other (see Supplemental Digital Content 1, available at: http://links.lww.com/JCRP/A90). We also recorded the first hospitalization day in which a patient received ICR, the total number of days with ICR, and the proportion of total hospital days (days of ICR/hospital length of stay) that had a code for ICR. In cases in which patients had multiple service codes, we counted only 1 code for ICR each day. Notably, the primary role of these codes is for tracking internal hospital process of care that influences hospital costs rather than for external billing or insurance claims. As such, these codes represent professionally delivered and tracked ICR associated with formal services, rather than informal education or ambulation offered by a nurse, physician, or other non-CR professional.
Recognizing that physical therapy (PT) sometimes partially fills the role of ICR at some hospitals, especially at hospitals without an ICR program, we separately assessed all patients for ≥1 inpatient service code(s) for PT. We included only codes that seemed likely to have some component of exercise, ambulation, or risk-factor education and selected 52 of 236 available PT codes (see Supplemental Digital Content 1, available at: http://links.lww.com/JCRP/A90). We noted the type, timing, and dose of PT delivered and reported this separately and distinctly from ICR utilization.
We considered a hospital to provide ICR if ≥1 eligible patients received ICR service codes at the hospital. Similarly, hospital capability for cardiac surgery and PCI were determined by the presence of any patients with these procedures. Characteristics of patients and hospitals with and without ICR were compared using χ2 tests for categorical factors and Kruskal-Wallis tests for continuous factors. Restricting focus to patients receiving ICR, we then described the variation in type (education/evaluation/exercise), timing, and dose of ICR across patient strata (surgical, PCI, medical).
To assess trends in the use of ICR, we restricted our analysis to hospitals that participated in the database during the entire 4.5-yr study period and computed proportion of patients receiving ICR during each 6-mo interval of the study period. Temporal trends of ICR were assessed with the Cochran-Armitage trend test for overall group and within each stratum. To identify patient characteristics associated with the use of ICR, we restricted analysis to hospitals that provided ICR, defined as having ≥1 patient with ICR within each 6-mo interval. We developed a hierarchical generalized linear model with a random hospital effect, using a logit link, that included patient demographics, the Gagne comorbidity score,20 selected comorbidities, and hospital characteristics. Age, race, and sex were forced into all models; all other factors with P > .05 were removed. Finally, intraclass correlation coefficients were used to compare variation in ICR use across hospitals, taking into account patient characteristics in surgical, PCI, and medical groups.22 All analyses were performed using SAS version 9.3 (SAS Institute).
During the study period, 1 343 537 patient admissions met inclusion criteria. Of these, 209 932 (15.6%) had a cardiac surgical procedure; 370 324 had a PCI (27.6%); 171 729 had MI without PCI (12.8%); and almost half of the total number of admissions 592 551 (44.1%) were for HF. Within each patient stratum, patient characteristics varied significantly by the receipt of ICR (Table 1). In general, patients with ICR were younger, had fewer comorbidities, were more often male, and were more likely to have non-Medicare insurance. Results for the full 8-stratum analysis are provided in Supplemental Digital Content 2, available at: http://links.lww.com/JCRP/A91.
Overall, patient-level utilization of ICR at all hospitals was 20.8%, but this varied greatly by indication, ranging from 43.3% (surgical patients) to 11.7% (medically treated MI and HF) (Table 2). Between 2007 and 2011, utilization of ICR decreased, but only slightly, from 21.2% to 19.1%, P < .001. There was a more sizable, but still small, decline in ICR among surgical patients from 45.1% to 42.0%, P < .001 (see Supplemental Digital Content 3, available at: http://links.lww.com/JCRP/A92).
Among patients who received ICR, the median number (interquartile range of days with ICR services per admission was 1 (1, 2), ranging from 3 (1, 4) sessions for surgical patients to 1 (1, 2) session per PCI admission (Table 2). For surgical admissions, the initial evaluation by ICR typically occurred on hospital day 4, which was about halfway through the hospitalization. For all other admissions, ICR occurred during the latter half of the hospital stay. Exercise was the most common modality of treatment by ICR (61.8%); education and evaluation services accounted for 20.8% and 27.1% of services rendered, respectively. Although 87.7% of surgical patients were treated with ICR and/or PT, most of all patients (51.5%) did not have a service code for either ICR or PT. Results for the full 8-stratum analysis are provided in Supplemental Digital Content 4, available at: http://links.lww.com/JCRP/A93.
Of the 458 hospitals, 223 (49%) had at least 1 hospital admission with an ICR code, suggesting the availability of ICR at these hospitals. Several hospital characteristics were associated with the presence of an ICR program. Hospitals that were located in urban areas (52.3 vs 47.7%, P = .01) and those that offered cardiac surgery (67.1 vs 32.9%, P < .001) or PCI services (63.3 vs 36.7%, P < .001) were significantly more likely to offer ICR (Table 3). Hospitals located in the Midwest and particularly the west north central region of the Midwest (Missouri, Kansas, Iowa, Minnesota, Nebraska, and North and South Dakota) were most likely to provide ICR.
RESULTS AT HOSPITALS WITH ICR PROGRAMS
When the analysis was restricted to hospitals that provided ICR, 30.6% of patients received ICR and hospital median utilization was 18.8% (interquartile range: 0.3%, 51.9%). Use of ICR varied substantially by indication and was highest among surgical patients (median [interquartile range] = 86% [45%, 96%]) and lowest among patients with HF or medically managed MI (7%, [0.45, 30%]). Use of ICR among patients who had undergone PCI showed the greatest variation between hospitals (mean = 48%, range = 4.0%, 83%) (Figure). The presence of a surgical program with high ICR utilization rates was strongly associated with use of ICR among other patient groups (see Supplemental Digital Content 5, available at: http://links.lww.com/JCRP/A94).
Multivariable hierarchical modeling revealed that several patient factors associated with lower likelihood of ICR utilization (Table 4). Among surgical patients, older age (OR = 0.88; 95% CI, 0.81-0.96) for those aged ≥80 yr versus those aged 55 to 64 yr; “other” racial category (OR = 0.88; 95% CI, 0.81-0.94) versus whites; and female sex (OR = 0.91; 95% CI, 0.87-0.95), Medicare insurance and higher comorbidity burden were independently associated with lower odds of receiving ICR. Results were similar among patients with PCI procedures or medically managed MI or HF. Tobacco abuse was the only factor examined that was consistently associated with significantly greater ICR utilization (OR range across conditions was 1.07-1.14). In general, PT use was associated with lower ICR use (OR range across conditions was 0.51-0.98) (Table 4).
Model discrimination was excellent with C-statistics for all patient strata ranging from 0.90 to 0.92 (Table 4). In models including a hospital random effect, there was substantial variation in ICR use associated with the hospital only, 83.2%, 79.7%, and 69.7% for surgical, PCI, and medical groups, respectively. Addition of patient characteristics to these models accounted for a nominal addition to the explained variation (Table 4).
Using a large, detailed, national database of >450 hospitals and 1.3 million admissions, we found that 21% of eligible patients received formal ICR before hospital discharge and that the percentage of patients receiving ICR gradually declined over time. In addition, <50% of hospitals treating cardiac patients offered formal ICR, and, among the hospitals that offered ICR, <33% patients received ICR services. We also found that only approximately 60% of hospitals that perform cardiac surgery offer formal ICR and, overall, <50% of CABG patients received ICR. Given the important role both inpatient CR and outpatient CR play in recovery from an acute cardiac condition, our findings suggest that there is a large opportunity to increase hospital-initiated, lifestyle-focused efforts in the secondary prevention of heart disease.
Including PT as an ICR surrogate improved rates of use particularly among the surgical population. However, even when considering both of these services together, both PT and ICR still did not reach even half of the potentially eligible population. Moreover, there are often substantial differences between the goals and purposes of ICR and PT, so it is unclear whether PT can really be considered a full surrogate for ICR.
Although we found that older patients, females, and those with more comorbidities and non-Medicare insurance were less likely to receive ICR, these factors were dwarfed by the much larger role of the hospital in predicting use of ICR. This is not necessarily surprising because, unlike outpatient CR, hospitalized patients do not actively decide to participate in ICR and hospital protocols and policies likely play a stronger role than physician referral patterns. However, this pattern of ICR use strongly suggests that any initiatives seeking to improve either ICR or outpatient CR will be more successful if they focus on hospital-level interventions rather than on patient or physician interventions.
Although the exact reason for low ICR utilization is unknown, 2 key factors are probably at play. First, declining hospital length of stay23 , 24 over the past few decades has made some cardiac hospitalizations so brief (1-2 d in this PCI group) that it is now difficult to ensure that every patient receives ICR. Second, because ICR has not been a billable service since the 1980s, hospitals have had no direct financial incentive to retain these prevention-focused programs. As a result, it is understandable that such programs might be inadequately staffed, poorly supported, never initiated, or have been discontinued after initial implementation. However, because ongoing health care reforms are increasingly using bundled payments and population risk management, we suspect that ICR will become more appealing to hospital executives as a mechanism to improve outpatient CR rates and thereby improve both financial and patient outcomes.4 , 25 , 26
Although ideal secondary prevention and referral to outpatient CR can be accomplished in the outpatient setting, there are several reasons why the inpatient setting is an optimal place and time to initiate CR. First, prior research has suggested that only 1 in 4 eligible hospitalized cardiac patients receives ideal care, composed of outpatient CR referral, smoking cessation advice, lifestyle counseling, and proper medications prior to discharge.27 As ICR should be providing at least 2 of these 4 services, a robust ICR program could help decrease this gap and potentially decrease mortality.13 Second, counseling hospitalized patients on lifestyle issues improves patient experience12 and substantially increases the chance that important lifestyle and behavior counseling will occur.15 Third, inpatient liaison-facilitated CR referral is an effective tool to increase outpatient CR participation rates.10 Fourth, making an appointment for CR prior to hospital discharge hastens enrollment into outpatient CR and improves CR participation.25 As the vast majority of patients are hospitalized before commencing outpatient CR, improving ICR utilization is likely to increase outpatient CR participation and improve secondary prevention care.
Notably, several features of ICR are similar to the findings of prior publications concerning the epidemiology and utilization of outpatient CR. First, our finding of overall low ICR utilization is consistent with data from the 1995-2005 era, which show low national utilization of outpatient CR.5 , 7 Second, patients undergoing procedures were more likely to receive ICR and this is consistent with prior studies which demonstrated that patients with CABG or PCI were more likely to be referred to and attend outpatient CR than patients with MI.7 , 28 , 29 Third, we found the highest utilization of ICR in the Midwest (specifically the West-North-Central areas), consistent with prior publications in which outpatient CR utilization was highest in Nebraska and other North-Central states.7 , 30 Fourth, we found that active smoking was associated with increased ICR utilization, consistent with a prior publication showing that smokers are more likely to be referred for CR.31
Several study limitations deserve to be mentioned. First, it is possible that our reported ICR utilization rate is artificially low because of incomplete ascertainment of ICR. However, given that ICR is usually provided by trained professionals in hospital service lines where productivity is monitored and reported, we believe that our database captures the vast majority of formal ICR delivered in these hospitals. In addition, prior studies have reported very high correlations between treatment rate estimates produced using the Premier database and those obtained through chart review.32 , 33 Second, physicians, nurses, or other non-ICR staff members at hospitals without ICR programs may provide at least some ICR equivalent services to patients without such efforts being recorded in the database. Depending on how frequently this occurred, efforts at CR referral and secondary prevention of heart disease at hospitals without ICR programs might be substantially higher than suggested by our data. However, the size of the national CR referral gap8 , 29 and other known secondary prevention quality gaps27 , 34 suggests that current US hospital-based efforts in the secondary prevention of heart disease are suboptimal. Third, although national ICR guidelines recommend that each patient be ambulated, educated, and referred to outpatient CR,2 we cannot verify which of these 3 activities were performed as part of the database-recorded ICR visit. Consequently, this limits our ability to know whether a patient was referred to outpatient CR, but this limitation seems unlikely to affect whether or not he or she actually had an ICR visit.
In summary, we found that most patients did not receive formal ICR during a hospitalization for a cardiac condition that would make them eligible for outpatient CR. We also found wide variations in care across diagnoses, procedures, and a host of additional patient- and hospital-level factors. These findings highlight the large opportunity hospitals and providers have to use ICR programs to optimize the care provided to their patients in the secondary prevention of heart disease.
This work was supported by the National Center for Advancing Translational Sciences of the National Institutes of Health, award number KL2TR001063. Drs Lagu and Lindenauer were supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under award numbers K01HL114745 and 1K24HL132008, respectively. Dr Ades was supported by the Vermont Center on Behavior and Health (NIH-NIGMS P20GM103644). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
1. Benjamin EJ, Blaha MJ, Chiuve SE, et al Heart disease and stroke statistics—2017 update: a report from the American Heart Association. Circulation. 2017;135(10):e146–e603.
2. American Association of Cardiovascular and Pulmonary Rehabilitation. Guidelines for Cardiac Rehabilitation
and Secondary Prevention Programs. 5th ed. Champaign, IL: Human Kinetics; 2013.
3. Ades PA. Cardiac rehabilitation
and secondary prevention of coronary heart disease. N Engl J Med. 2001;345(12):892–902.
4. Heran BS, Chen JM, Ebrahim S, et al Exercise-based cardiac rehabilitation
for coronary heart disease. Cochrane Database Syst Rev. 2011;(7):CD001800.
5. Receipt of outpatient cardiac rehabilitation
among heart attack survivors—United States, 2005. MMWR Morb Mortal Wkly Rep. 2008;57(4):89–94.
6. Receipt of cardiac rehabilitation
services among heart attack survivors—19 states and the District of Columbia, 2001. MMWR Morb Mortal Wkly Rep. 2003;52(44):1072–1075.
7. Suaya JA, Shepard DS, Normand SL, Ades PA, Prottas J, Stason WB. Use of cardiac rehabilitation
by Medicare beneficiaries after myocardial infarction or coronary bypass surgery. Circulation. 2007;116(15):1653–1662.
8. Aragam KG, Dai D, Neely ML, et al Gaps in referral to cardiac rehabilitation
of patients undergoing percutaneous coronary intervention in the United States. J Am Coll Cardiol. 2015;65(19):2079–2088.
9. Grace SL, Leung YW, Reid R, et al The role of systematic inpatient cardiac rehabilitation
referral in increasing equitable access and utilization
. J Cardiopulm Rehabil Prev. 2012;32(1):41–47.
10. Grace SL, Russell KL, Reid RD, et al Effect of cardiac rehabilitation
referral strategies on utilization
rates: a prospective, controlled study. Arch Intern Med. 2011;171(3):235–241.
11. Gurewich D, Prottas J, Bhalotra S, Suaya JA, Shepard DS. System-level factors and use of cardiac rehabilitation
. J Cardiopulm Rehabil Prev. 2008;28(6):380–385.
12. Jackson EA, Krishnan S, Meccone N, Ockene IS, Rubenfire M. Perceived quality of care and lifestyle counseling among patients with heart disease. Clin Cardiol. 2010;33(12):765–769.
13. Auer R, Gaume J, Rodondi N, Cornuz J, Ghali WA. Efficacy of in-hospital multidimensional interventions of secondary prevention after acute coronary syndrome: a systematic review and meta-analysis. Circulation. 2008;117(24):3109–3117.
14. Pack QR, Squires RW, Lopez-Jimenez F, et al Participation rates, process monitoring, and quality improvement among cardiac rehabilitation
programs in the United States: a national survey. J Cardiopulm Rehabil Prev. 2015;35(3):173–180.
15. Spencer FA, Salami B, Yarzebski J, Lessard D, Gore JM, Goldberg RJ. Temporal trends and associated factors of inpatient cardiac rehabilitation
in patients with acute myocardial infarction: a community-wide perspective. J Cardiopulm Rehabil. 2001;21(6):377–384.
16. Lindenauer PK, Pekow P, Gao S, Crawford AS, Gutierrez B, Benjamin EM. Quality of care for patients hospitalized for acute exacerbations of chronic obstructive pulmonary disease. Ann Intern Med. 2006;144(12):894–903.
17. Centers for Medicare & Medicaid Services. Decision Memo for Cardiac Rehabilitation
Programs (CAG-00089R). Baltimore, MD: Centers for Medicare & Medicaid Services; 2006.
18. Centers for Medicare & Medicaid Services. Proposed Decision Memo for Cardiac Rehabilitation
(CR) Programs—Chronic Heart Failure (CAG-00437N). Baltimore, MD: Centers for Medicare & Medicaid Services; 2013.
19. Thirapatarapong W, Thomas RJ, Pack Q, Sharma S, Squires RW. Commercial insurance coverage for outpatient cardiac rehabilitation
in patients with heart failure in the United States. J Cardiopulm Rehabil Prev. 2014;34(6):386–389.
20. Gagne JJ, Glynn RJ, Avorn J, Levin R, Schneeweiss S. A combined comorbidity score predicted mortality in elderly patients better than existing scores. J Clin Epidemiol. 2011;64(7):749–759.
21. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8–27.
22. Larsen K, Petersen JH, Budtz-Jorgensen E, Endahl L. Interpreting parameters in the logistic regression model with random effects. Biometrics. 2000;56(3):909–914.
23. Berger AK, Duval S, Jacobs DR Jr, et al Relation of length of hospital stay in acute myocardial infarction to postdischarge mortality. Am J Cardiol. 2008;101(4):428–434.
24. Chin CT, Weintraub WS, Dai D, et al Trends and predictors of length of stay after primary percutaneous coronary intervention: a report from the CathPCI registry. Am Heart J. 2011;162(6):1052–1061.
25. Dunlay SM, Pack QR, Thomas RJ, Killian JM, Roger VL. Participation in cardiac rehabilitation
, readmissions, and death after acute myocardial infarction. Am J Med. 2014;127(6):538–546.
26. Wong WP, Feng J, Pwee KH, Lim J. A systematic review of economic evaluations of cardiac rehabilitation
. BMC Health Serv Res. 2012;12:243.
27. Redfern J, Hyun K, Chew DP, et al Prescription of secondary prevention medications, lifestyle advice, and referral to rehabilitation among acute coronary syndrome inpatients: results from a large prospective audit in Australia and New Zealand. Heart. 2014;100(16):1281–1288.
28. Thomas RJ, Miller NH, Lamendola C, et al National survey on gender differences in cardiac rehabilitation
programs. Patient characteristics and enrollment patterns. J Cardiopulm Rehabil. 1996;16(6):402–412.
29. Brown TM, Hernandez AF, Bittner V, et al Predictors of cardiac rehabilitation
referral in coronary artery disease patients: findings from the American Heart Association's Get With The Guidelines Program. J Am Coll Cardiol. 2009;54(6):515–521.
30. Curnier DY, Savage PD, Ades PA. Geographic distribution of cardiac rehabilitation
programs in the United States. J Cardiopulm Rehabil. 2005;25(2):80–84.
31. Gaalema DE, Cutler AY, Higgins ST, Ades PA. Smoking and cardiac rehabilitation
participation: associations with referral, attendance and adherence. Prev Med. 2015;80:67–74.
32. Rothberg MB, Lahti M, Pekow PS, Lindenauer PK. Venous thromboembolism prophylaxis among medical patients at US hospitals. J Gen Intern Med. 2010;25(6):489–494.
33. Jennings LA, Auerbach AD, Maselli J, Pekow PS, Lindenauer PK, Lee SJ. Missed opportunities for osteoporosis treatment in patients hospitalized for hip fracture. J Am Geriatr Soc. 2010;58(4):650–657.
34. Tang L, Patao C, Chuang J, Wong ND. Cardiovascular risk factor control and adherence to recommended lifestyle and medical therapies in persons with coronary heart disease (from the National Health and Nutrition Examination Survey 2007-2010). Am J Cardiol. 2013;112(8):1126–1132.