Study population and analytic cohort. CR indicates cardiac rehabilitation; PREMIER, Prospective Registry Evaluating Myocardial Infarction: Events and Recovery; and TRIUMPH, Translational Research Investigating Underlying Disparities in Acute Myocardial Infarction Patients’ Health Status study.
A, Mean SAQ quality of life scores. B, Mean SAQ angina frequency scores. C, Mean SAQ treatment satisfaction scores. D, Mean SAQ physical limitation scores. CR indicates cardiac rehabilitation.
Mean differences at 6 and 12 months after acute myocardial infarction are shown between propensity-matched patients who participated in cardiac rehabilitation (CR) and those who did not participate in CR. A mean difference of ≥5 points is the minimal threshold for a clinically significant difference between participants and nonparticipants. AF indicates angina frequency; MCS, mental component summary; PCS, physical component summary; PL, physical limitation; QoL, quality of life; and TS, treatment satisfaction.
Propensity-matched patients who participated in cardiac rehabilitation (CR) and those who did not participate in CR (hazard rate, 0.59; 95% CI, 0.46-0.75).
eFigure 1. Mean Short Form-12 (SF-12) Physical Component Summary (PCS) and Mental Component Summary (MCS) Scores of Cardiac Rehabilitation Participants (n = 2015) and Nonparticipants (n = 2914) During 12 Months After Acute Myocardial Infarction
eFigure 2. Propensity Scatter Plot Used for Matching Cardiac Rehabilitation Participants and Nonparticipants
eFigure 3. Standardized Differences of Patient Covariates Before and After Propensity Matching
Customize your JAMA Network experience by selecting one or more topics from the list below.
Kureshi F, Kennedy KF, Jones PG, et al. Association Between Cardiac Rehabilitation Participation and Health Status Outcomes After Acute Myocardial Infarction. JAMA Cardiol. 2016;1(9):980–988. doi:10.1001/jamacardio.2016.3458
What is the association between participation in cardiac rehabilitation and health status outcomes of patients after acute myocardial infarction?
In this cohort study of 4929 patients enrolled in 2 acute myocardial infarction registries, mean Seattle Angina Questionnaire domain scores were similar at 6 and 12 months after acute myocardial infarction between patients who did and did not participate in cardiac rehabilitation.
Increased use of validated health status outcome measures are needed to further examine if and how health status can be maximized for participants in cardiac rehabilitation after acute myocardial infarction.
Cardiac rehabilitation (CR) improves survival after acute myocardial infarction (AMI), and referral to CR has been introduced as a performance measure of high-quality care. The association of participation in CR with patients’ health status (eg, quality of life, symptoms, and functional status) is poorly defined.
To examine the association of participation in CR with health status outcomes after AMI.
Design, Setting, and Participants
A retrospective cohort study was conducted of patients enrolled in 2 AMI registries: PREMIER, from January 1, 2003, to June 28, 2004, and TRIUMPH, from April 11, 2005, to December 31, 2008. The analytic cohort was restricted to 4929 patients with data available on baseline health status, 6- or 12- month follow-up health status, and participation in CR. Data analysis was performed from 2014 to 2015.
Participation in at least 1 CR session within 6 months of hospital discharge.
Main Outcomes and Measures
Patient health status was quantified using the Seattle Angina Questionnaire (SAQ) and the 12-Item Short-Form Health Survey (SF-12). The primary outcomes of interest were the mean differences in SAQ domain scores during the 12 months after AMI between patients who did and did not participate in CR. Secondary outcomes were the mean differences in the SF-12 summary scores and all-cause mortality.
After successfully matching the cohorts of the 4929 patients (3328 men and 1601 women; mean [SD] age, 60.0 [12.2] years) for the propensity to participate in CR and comparing the groups using linear, mixed-effects models, mean differences in the SAQ and SF-12 domain scores were similar at 6 and 12 months between the 2012 patients participating in CR (3 were unable to be matched) and the 2894 who did not participate (20 were unable to be matched). At 6 months, the mean difference was –0.76 (95% CI, –2.05 to 0.52) for the SAQ quality of life score, –1.53 (95% CI, –2.57 to –0.49) for the SAQ angina frequency score, 0.38 (95% CI, –0.51 to 1.27) for the SAQ treatment satisfaction score, –0.42 (95% CI, –1.65 to 0.79) for the SAQ physical limitation score, 0.50 (95% CI, –0.22 to 1.22) for the SF-12 physical component score, and 0.13 (95% CI, –0.53 to 0.79) for the SF-12 mental component score. At 12 months, the mean difference was –0.89 (95% CI, –2.20 to 0.43) for the SAQ quality of life score, –1.05 (95% CI, –2.12 to 0.02) for the SAQ angina frequency score, 0.38 (95% CI, –0.54 to 1.29) for the SAQ treatment satisfaction score, –0.14 (95% CI, –1.41 to 1.14) for the SAQ physical limitation score, 0.17 (95% CI, –0.57 to 0.92) for the SF-12 physical component score, and 0.12 (95% CI, –0.56 to 0.80) for the SF-12 mental component score. In contrast, the hazard rate of all-cause mortality (up to 7 years) associated with participating in CR was 0.59 (95% CI, 0.46-0.75).
Conclusions and Relevance
In a cohort of 4929 patients with AMI, we found that those who did and did not participate in CR had similar reported health status during the year following AMI; however, participation in CR did confer a significant survival benefit. These findings underscore the need for increased use of validated patient-reported outcome measures to further examine if and how health status can be maximized for patients who participate in CR.
Cardiac rehabilitation (CR) is an important component of secondary prevention after acute myocardial infarction (AMI).1,2 Supported by systematic reviews and meta-analyses documenting reduced mortality associated with participation in CR,3-6 the American Heart Association and American College of Cardiology have designated referral to CR as a class IA recommendation and endorsed it as a performance measure for the quality of care of patients with AMI.2,7,8 Although health status improvement is among the commonly cited goals and benefits of CR,8,9 there is sparse evidence on the effect of CR on patient-reported health status (eg, patients’ symptoms, function, and quality of life [QoL]).
Examining the health status benefits of CR can enable clinicians and health care professionals who provide CR to better inform patients about the potential benefits of participating in CR. To better examine the association of CR with health status outcomes after AMI, we used detailed, patient-centered data from 2 large multicenter AMI registries in the United States: the Prospective Registry Evaluating Myocardial Infarction: Events and Recovery (PREMIER)10 and the Translational Research Investigating Underlying Disparities in Acute Myocardial Infarction Patients’ Health Status study (TRIUMPH)11 to compare patient-reported health status between those who did and did not participate in CR.
Details regarding the study designs, patient selection criteria, site characteristics, and follow-up assessments of the PREMIER10 and TRIUMPH11 studies have been described previously. Briefly, both studies were US multicenter, prospective, observational AMI registries with similar study protocols. A total of 2498 patients from 19 medical centers were enrolled into PREMIER (January 1, 2003, to June 28, 2004), while TRIUMPH (April 11, 2005, to December 31, 2008) enrolled 4340 patients across 21 sites (9 sites participated in both registries). The PREMIER and TRIUMPH study sites included large academic medical centers, single-payer systems, inner-city hospitals, and non–university hospitals. Eligible patients were required to present to or be transferred to an enrolling site within 24 hours of presentation. Elevated biomarkers and either prolonged ischemic signs and symptoms or ischemic changes on results of electrocardiography were used to confirm diagnosis of AMI. Data regarding baseline patient demographics, comorbidities, and health status, in addition to hospital course and management, were obtained through abstraction of medical records and detailed, structured interviews conducted by trained research staff during hospital admission. Follow-up data regarding participation in CR and health status were obtained through centralized, standardized telephone interviews at 1, 6, and 12 months after hospital discharge. Each of the 30 participating sites obtained institutional review board approval, and all patients provided written informed consent for study enrollment, data collection, and follow-up.
A total of 6838 patients with AMI were enrolled in the TRIUMPH and PREMIER studies. For our primary analysis, we restricted our cohort to patients who had data available on baseline health status, follow-up health status at 6 or 12 months, and participation in CR. Accordingly, excluded patients were those who died before 6-month follow-up (n = 323), died between the 6- and 12-month follow-up without any follow-up data (n = 97), were missing data on baseline health status (n = 50), were missing data on 6- and 12- month follow-up health status (n = 1094), and were missing data on participation in CR (n = 345), leaving a final analytic cohort of 4929 patients (Figure 1).
Information about participation in CR was obtained through centralized, standardized telephone interviews at 1 and 6 months after hospital discharge. Cardiac rehabilitation programs at participating sites from both registries were institutional based. During the 1- and 6-month follow-up telephone interviews, patients were asked if they had attended and participated in a CR program since hospitalization for their “heart attack or heart problem.” CR participants were defined as those who confirmed participating in a CR program (minimum of 1 session) within 6 months of hospital discharge following AMI.
Patients’ health status was assessed at baseline and at each follow-up interview using the Seattle Angina Questionnaire (SAQ) and the 12-Item Short-Form Health Survey (SF-12). The SAQ is a validated, disease-specific instrument consisting of 19 items that measure 5 clinically relevant domains of health status in patients with coronary artery disease.12 For our analysis, the following 4 domains were included: QoL, angina frequency (AF), treatment satisfaction (TS), and physical limitation (PL). Domain scores range from 0 to 100, with lower scores indicating poorer health status (eg, more angina, more limitation in physical activities owing to angina, and poorer QoL). Prior work has suggested that a change of approximately 5 points in the SAQ QoL, AF, TS, and PL domain scores is clinically significant.12,13 The SF-12 is a generic health status measure consisting of 12 items derived to provide overall summary scales of the 36-Item Short-Form Survey physical component summary and mental component summary.14 Scoring is norm based and standardized to a mean (SD) of 50 (10), with lower scores indicating poorer health status.
The primary outcomes of interest were the mean differences in SAQ QoL, AF, TS, and PL domain scores at 6 and 12 months after hospital discharge between patients who did and did not participate in CR. Secondary outcomes of interest were the mean differences in the SF-12 physical component summary and mental component summary scores and all-cause mortality (up to 7 years after AMI). Mortality was determined through follow-up interviews and querying the Social Security Death Index.
Data analysis was performed from 2014 to 2015. We first measured the distribution of baseline patient characteristics between patients who did and did not participate in CR using standardized differences, defined as the mean difference between groups divided by the pooled SD of the 2 groups. This measure of distribution is not as sensitive to sample size as traditional tests and provides a sense of the relative magnitude of differences, with a standardized difference of more than 10% typically considered to represent a meaningful imbalance.15 Next, to ensure an appropriate balance of patient characteristics between groups before comparing outcomes, we derived propensity scores for the likelihood of participating in CR within 6 months of hospital discharge after AMI. We constructed nonparsimonious logistic regression models using 45 variables to examine the likelihood of participation in CR.9,16,17 Explanatory variables included in the propensity model (Table) were among larger conceptual domains that occurred before hospital discharge (and before the opportunity to participate in CR) and that were hypothesized to potentially be associated with participation in CR. These variables included demographic, psychosocial, and socioeconomic characteristics; clinical status at presentation; comorbidities; revascularization during hospitalization for AMI; new events during hospitalization for AMI; health status at time of AMI; and medications prescribed at hospital discharge. Multiple imputation data sets were created to handle missing patient variables when constructing the propensity scores. We then conservatively matched (caliper width of 0.2 times the pooled SD of the logit propensity score) patients participating and not participating in CR by using an optimal strategy allowing many-to-many matching.18 After matching, standardized differences of all covariates were calculated to ensure that each was less than 10%, indicating an adequate balance of covariate distribution between groups. For the analysis of health status outcomes, we used a linear mixed-effects model to estimate the mean difference of SAQ and SF-12 scores at 6 and 12 months between propensity-matched patients participating and not participating in CR. This model allowed for the incorporation of all eligible patients with baseline and follow-up health status scores at 6 or 12 months while assessing for interactions between groups and time. Mortality differences (up to 7 years after AMI) between propensity-matched pairs of patients participating and not participating in CR were examined using a Cox proportional hazards regression model to estimate hazard rates.
To examine the robustness of our findings associated with health status outcomes from our primary analysis, we conducted several additional sensitivity analyses. First, to assess for any survivor bias, patients who died within the first year and did not have health status data from the 6- or 12-month follow-up were included (n = 420) and assigned a score of zero (worst possible score) for all SAQ and SF-12 domain scores at that time. Second, for evaluation of any bias resulting from excluding surviving patients who were missing health status data from the 6- and 12-month follow-up, we used inverse probability weighting. We first compared the characteristics of the analytic cohort vs living patients excluded owing to missing health status data from the 6- and 12-month follow-up (n = 1094). A logistic regression model was then constructed in which patients were weighted on the inverse of the probability to be missing health status data from the 6- and 12-month follow-up so as to enable greater weight in the outcomes to patients who were most like those who were living but missing health status data from the follow-up. Last, to examine for any potential selection bias associated with referral to CR, we reran our primary analysis after restricting our analytic cohort to only patients who were referred to CR (n = 3957). The results of all sensitivity analyses were comparable with those from our original analysis, so only the primary analysis is reported. All analyses were conducted with SAS, version 9.3 (SAS Institute), and R, version 2.15.3 (R Foundation for Statistical Computing), and 95% CIs were calculated for all point estimates.
Among 4929 eligible patients (3328 men and 1601 women; mean [SD] age, 60.0 [12.2] years) with baseline health status data, 3743 (75.9%) had 6- and 12-month health status data available, while 711 (14.4%) and 475 (9.6%) only had 6- and 12-month follow-up health status data available, respectively. A total of 2015 patients (40.9%) reported participation in CR within 6 months of hospitalization for AMI. Before propensity score matching, several important baseline patient covariates were not well balanced between the groups (standardized differences >10%), although the mean (SD) age of patients who did and did not participate in CR was similar (60.2 [11.7] vs 59.9 [12.5] years) (Table). Compared with those who did not participate in CR, a higher proportion of participants were male (70.8% vs 65.2%), white (83.6% vs 63%), married (68.3% vs 52%), employed full-time (45.8% vs 35.1%), had health insurance coverage for medications (85.1% vs 71.1%), and received coronary revascularization during hospitalization for AMI (86% vs 70.1%). Moreover, patients who participated in CR avoided seeking health care owing to cost less often (14.7% vs 25.1%) and were generally healthier than those who did not participate in CR, with fewer participants having a prior MI (14.8% vs 23.3%) or history of smoking (55.4% vs 62.2%).
Although the standardized differences for all mean baseline SAQ and SF-12 scores were greater than 10%, only the standardized difference of the SAQ PL score between patients who did and did not participate in CR was clinically significant (≥5 points), with participants having a mean (SD) higher score (90.5 [17.4] vs 83.4 [24.2]; standardized difference, 33.7%). During the year after AMI, the mean health status scores of patients who did and did not participate in CR improved in all domains except for SAQ TS, with small differences observed between groups (Figure 2 and eFigure 1 in the Supplement).
After using propensity score methods to assemble a matched cohort of patients who participated in CR with characteristics similar to those who did not participate, 3 participants and 20 nonparticipants were unable to be matched and were excluded from further analyses (eFigure 2 in the Supplement). The distribution of observed patient characteristics were similar between the matched cohort of patients who participated in CR (n = 2012) and those who did not participate (n = 2894), as all standardized differences were less than 10% (eFigure 3 in the Supplement).
In the propensity score–matched analysis of health status outcomes, mean SAQ QoL, AF, TS, and PL scores at 6 and 12 months were clinically similar for both patients who participated in CR and those who did not participate. Although mean SAQ AF scores at 6 months were statistically significantly higher in those who did not participate than in CR patients who did participate (between-group difference, 1.53 points), this difference was well below the generally accepted threshold of clinical importance (≥5 points). For generic health status measurement, mean SF-12 physical component summary and mental component summary scores at 6 and 12 months were also similar (Figure 3). The results of the sensitivity analyses examining for any referral, survivor, or loss to follow-up bias were comparable with the above results. In the analysis of all-cause mortality during 7 years of follow-up in the propensity-matched patients who did and did not participate in CR, the participants had a 41% lower hazard rate of mortality compared with nonparticipants (hazard rate, 0.59; 95% CI, 0.46-0.75) (Figure 4).
In 4929 patients with AMI across 31 US centers, we found that, on average, both those who participated in at least 1 CR session and those who did not participate had improvements in their disease-specific and overall health status during the 12 months after hospital discharge. After propensity matching, the disease-specific and generic health status scores of patients who did and did not participate in CR were similar at both 6 and 12 months after AMI. Although health status did not clinically differ between groups, the hazard rate of long-term mortality was 41% lower in patients who participated in CR. To our knowledge, this is the largest effectiveness study to examine the association between CR and health status outcomes in patients after AMI.
Although many studies have examined the effect of CR on morbidity and mortality in patients after AMI,4,19,20 only a limited number of studies have examined the effect of CR on health status using validated instruments, with most using generic measures. For instance, among 63 trials included in a 2016 Cochrane review of exercise-based CR in patients with coronary artery disease, most of whom had experienced AMI, only 20 trials used a validated health status measure.21 Although most of these studies found an improvement in health status scores in both the treatment and control arms, the significant heterogeneity in the measuring and reporting of QoL precluded the Cochrane Collaborative from performing a meta-analysis of the health status benefits of CR.
Among the 20 trials that used validated health status measures, only 3 studies in the post-AMI setting used disease-specific measures with heterogeneity in the type of intervention. The first, a study by Oldridge et al,22 used the interviewer-administered Quality of Life after Myocardial Infarction questionnaire and found similar emotional and physical limitation domain scores at 4, 8, and 12 months in 201 patients with AMI who were randomized to 8 weeks of CR or usual care. The second, a multicenter Australian study by Heller et al,23 used the MacNew Heart Disease Health-Related Quality of Life questionnaire, the self-administered modification of the Quality of Life after Myocardial Infarction questionnaire, in 450 patients with AMI who were randomized to usual care or an intervention, which consisted of education on diet and exercise; they found that the intervention group had statistically higher emotional domain scores at the 6-month follow-up compared with patients who received usual care (5.4 vs 5.2; P = .04), although the mean difference of 0.2 points between groups is not considered clinically significant. Last, a China-based study by Wang et al24 used the Chinese Myocardial Infarction Dimensional Assessment Scale25,26 to assess disease-specific QoL in 160 patients with AMI from 2 centers who were randomized to receive home-based CR, consisting of a self-help manual, or usual care. Although the study did find statistically significant higher scores in 3 of 7 domains in those who participated in CR at the 6-month follow-up, the point differences between groups were only 4.2, 4.9, and 8.3 points in the domains of dependency, physical activity, and concern regarding medications, respectively, and the clinical significance of these differences is not known. Our findings, although observational, expand prior work by examining the association between CR and health status outcomes using well-accepted measures of disease-specific and generic health status in a much larger contemporary, real-world cohort of patients with AMI.
Only 2 prior observational studies have used the SAQ to examine the effect of CR on health status in patients with coronary artery disease; however, neither study was conducted exclusively in the post-AMI setting. The first, by Goss et al,27 was a prospective, observational multicenter study examining the association between participation in CR and health status in 691 patients during 12 months after coronary artery bypass grafting surgery. This 13-center study conducted in Washington state found that the SAQ and 36-Item Short-Form Survey domain scores improved for all patients 1 year after coronary artery bypass grafting surgery independent of participation in CR. Tavella and Beltrame28 conducted a smaller, more contemporary prospective, single-center Australian study that examined the effect of participation in CR on health status in 150 patients with coronary artery disease for 6 months following coronary angiography. They found that all patients had improvements in SAQ and 36-Item Short-Form Survey domain scores independent of participation in CR. Our findings of comparable changes in disease-specific and generic health status of both patients who did and did not participate in CR are similar to the results of these 2 studies, although our analyses focused exclusively on patients with a recent AMI.
Although improvement in health status is a commonly cited benefit of participation in CR after AMI,8,9 our results and prior work underscore the paucity of data to support this statement.3,5 Much of the uncertainty surrounding this association is owing to the lack of adequately powered, high-quality studies capturing the type, intensity, and frequency of CR while using validated, disease-specific health status measures. Despite no differences in health status, the association of participation in CR with survival after an AMI, and the reported underuse of CR,16 are compelling reasons to refer patients to, and ensure their participation in, CR. Nevertheless, it is important that patients are provided with accurate, evidence-based information that CR is associated with better survival but not better health status, given that prior work has suggested that many patients primarily expect improvement in health status from participation in CR.29-32 Moreover, future work should explore whether or not the type, intensity, frequency, duration, or any other additional components of CR may improve health status, as improved angina control and better QoL are primary goals of treatment after AMI. Our findings also underscore the need for increased use of validated cardiovascular health status measures in both clinical studies and the real-world setting to provide opportunities to further examine if and how these patient-centered outcomes can be maximized for patients who participate in CR.
Our study findings should be interpreted in the context of potential limitations. First, participation in CR was self-reported using structured follow-up interviews, and we did not verify patients’ reported participation at each site. However, prior work found a nearly perfect agreement between self-reported and site-verified participation in CR.33 Second, we did not capture the type, intensity, frequency, and length of patients’ participation in CR. However, our results are similar to those of prior randomized clinical trials and observational studies using disease-specific health status measures.22,27,28,34 Moreover, our demonstration of a survival benefit with participation in CR, similar to that in prior studies, further supports our characterization of participation in CR, given our replication of the previously established benefits of this intervention.3 Nevertheless, future research efforts may be able to better quantify the quantity and quality of participation in CR and identify a dose-response association that was missed with our crude categorization of participation. Third, the 2 groups in our analytic cohort had significant baseline differences in several characteristics, and while we were able to successfully match our cohorts using propensity-based methods in addition to several sensitivity analyses, we cannot exclude unmeasured confounding.
In a large, contemporary, multicenter cohort of 4929 patients with AMI, we found that patients who did and did not participate in CR had similar disease-specific and generic health status during the year following hospital discharge; however, participation in CR did confer a significant survival benefit. Our results underscore the need for further investigation of the effect of participation in CR on health status to identify if and how CR programs can better maximize health status outcomes for patients after AMI.
Accepted for Publication: August 8, 2016.
Corresponding Author: Faraz Kureshi, MD, MSc, Saint Luke’s Mid America Heart Institute, 4401 Wornall Rd, Kansas City, MO 64111 (email@example.com).
Published Online: October 19, 2016. doi:10.1001/jamacardio.2016.3458
Author Contributions: Drs Kureshi and Spertus had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Kureshi, Kennedy, Jones, Spertus.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Kureshi.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Kureshi, Kennedy, Jones.
Administrative, technical, or material support: Kureshi, Buchanan, Spertus.
Study supervision: Spertus.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Drs Kureshi, Qintar, and Fendler reported receiving supported Award Number T32HL110837 from the National Heart, Lung, and Blood Institute of the National Institutes of Health. Dr Spertus reported holding a copyright of the Seattle Angina Questionnaire. No other conflicts were reported.
Funding/Support: The Prospective Registry Evaluating Outcomes After Myocardial Infarction: Events and Recovery (PREMIER) study was funded by CV Therapeutics, Palo Alto, California. The Translational Research Investigating Underlying Disparities in Acute Myocardial Infarction Patients’ Health Status (TRIUMPH) study was funded by grant P50 HL 077113 from the National Heart, Lung, and Blood Institute. This study was also funded in part by CV Outcomes, Inc, Kansas City, Missouri.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.