Article Text

Download PDFPDF

Original articles
Six-year follow-up of a randomised controlled trial examining hospital versus home-based exercise training after coronary artery bypass graft surgery
  1. Kelly M Smith1,
  2. Robert S McKelvie2,
  3. Kevin E Thorpe3,
  4. Heather M Arthur4,5
  1. 1Department of Medicine, University of Illinois College of Medicine at Chicago, Section of Health Promotion Research, Chicago, Illinois, USA
  2. 2David Braley Cardiac, Vascular, and Stroke Research Institute, McMaster University, Hamilton, Ontario, Canada
  3. 3Dalla Lana School of Public Health, University of Toronto, Toronto, Ontario, Canada
  4. 4Faculty of Health Sciences, McMaster University, Hamilton, Ontario Canada
  5. 5Chief Scientific Officer, Hamilton Health Sciences, Hamilton, Ontario, Canada
  1. Correspondence to Dr Kelly M Smith, University of Illinois College of Medicine at Chicago, Department of Medicine, Section of Health Promotion Research, Department of Medical Education, 982 College of Medicine East Tower (M/C 591), 808 South Wood Street, Chicago, Illinois 60612-7309, USA; kellys{at}


Objective To compare the long-term effectiveness of hospital versus telephone-monitored home-based exercise training during cardiac rehabilitation (CR) on exercise capacity and habitual physical activity.

Design Six-year follow-up of patients who participated in a randomised controlled trial of hospital versus monitored home-based exercise training during CR after coronary artery bypass graft surgery.

Setting Outpatient CR centre in Central-South Ontario, Canada.

Participants 196 Patients who participated in the original randomised controlled trial and who attended an evaluation 1 year after CR.

Interventions 6 months of home or hospital-based exercise training during CR.

Main outcome measures Peak oxygen uptake (peak Vo2), Physical Activity Scale in the Elderly (PASE) to assess habitual activity, semi-structured interviews to assess vital status, demographic and descriptive information.

Results Of the 196 eligible patients, 144 (75.5%; 74 Hospital, 70 Home) were available for participation. Patients were predominantly male (n=120; 83.3%) aged 70±9.5 years. Clinical and sociodemographic outcomes were similar in both groups. While exercise performance declined over time, there were significant between-group differences in peak Vo2 (1506±418 ml/min vs 1393±341 ml/min; p=0.017) and PASE scores (166.7±90.2 vs 139.7±66.5; p=0.001) at 6-year follow-up in favour of the home group.

Conclusions Home and hospital-based exercise training maintained exercise capacity above pre-CR levels 6 years after CR. Exercise training initiated in the home environment in low-risk patients undergoing coronary artery bypass graft surgery conferred greater long-term benefit on Vo2 and persistent physical activity compared with traditional hospital-based CR.

  • Coronary artery bypass graft surgery
  • cardiac rehabilitation
  • exercise training
  • randomised controlled trial
  • long-term follow-up
  • exercise training
  • cardiac rehabilitation
  • delivery of care

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Both cardiovascular fitness and persistent regular physical activity have been shown to result in mortality benefits.1–3 Patients enter cardiac rehabilitation (CR) with diminished exercise capacity, which improves by 10–60% after CR in most patients.4 5 Home-based exercise training during CR confers similar benefits in exercise capacity as traditional hospital-based CR.6–10 Evidence suggests a gradual return of exercise capacity towards baseline levels early after discharge from CR.8 11–15 Results indicate that home-based exercise training during CR sustained improvements in exercise capacity to a greater extent than hospital-based training up to 1 year after discharge from CR.7–9 16–19 Research is needed to extend these findings over the long term.19 The present study compares the effectiveness of monitored home-based versus hospital-based exercise training during CR on the long-term sustainability of functional capacity, habitual physical activity and exercise maintenance outcomes 6 years after the cessation of CR.



A 6-year follow-up study of a randomised controlled trial (RCT) of monitored home-based (Home) versus hospital-based (Hospital) exercise training during CR after coronary artery bypass graft (CABG) surgery at the Hamilton Health Sciences-General Site was performed.7 Outcomes were assessed upon entry to the study (baseline), after 6 months of CR (discharge), 1 year after the cessation of CR (1 year) and again 6 years later (long-term follow-up, LTF). Results on changes between discharge and 1 year have been reported previously.8 There was no contact between the study investigators and the study participants during the time between the 1 year and the LTF study periods. The period of interest is between 1 year and LTF. Baseline and discharge data are presented for comparison purposes (figure 1).

Figure 1

Randomisation flow diagram.

Setting and participants

The Cardiac Health and Rehabilitation Center (CHRC), Hamilton Health Sciences—General Campus in Central South Ontario, Canada served as the study site (1997–2005). The CHRC is a multidisciplinary CR clinic which offers services that comply with standardised guidelines for exercise prescription, smoking cessation, patient counselling for anxiety and depression, as well as nutritional guidelines for diabetes mellitus and lipid management.20 Eligibility criteria for the RCT have been reported previously and included low-risk patients after CABG.7–9 All patients eligible to participate in the RCT and who completed the 1-year evaluation (n=196) were contacted for participation in the LTF study. Recruitment began in February 2005 and was completed in October 2005.

Randomisation and intervention

The randomisation protocol and exercise training intervention have been previously reported.7 Briefly, following the informed consent process, patients were randomly assigned to home or hospital exercise training. Participants exercised in the home or hospital environment for a period of 6 months. Exercise sessions were approximately 30–50 min in length, three times per week at an intensity of 60–80% of target heart rate reserve (maximum heart − resting heart rate).20 21 Exercises for the hospital group included treadmill walking, stationary cycling, arm ergometry and stair climbing. Home exercise consisted primarily of walking but was tailored to include any exercise equipment possessed by the patient at the time of the study. Resistance training was not performed in either group. In addition to the study exercise prescription, participants in both groups were given guidelines for exercising that were consistent with national clinical practice guidelines (5–7 days per week for a minimum of 20 min).20 21

At LTF, patients were scheduled for one 3-hour session at the CHRC. The session included the informed consent process, review of health status, anthropometric measurements and habitual physical activity questionnaire. Demographic information such as employment status (full-time, part time, retired, sick leave, employment insurance, disability, other), living arrangements (spouse/partner, children, alone, other) and smoking status (current, quit, quit date) were collected during the interview. Hypertension was defined as a resting blood pressure exceeding 140 mm Hg systolic and 90 mm Hg diastolic on three consecutive readings. Dyslipidaemia was defined as total lipoprotein cholesterol >5.2 mmol/l and requiring treatment. Vital status was obtained through a combination of patient self-report and electronic health records. Patient self-report was verified through the primary care provider if an electronic record was unavailable. Patients completed a graded symptom-limited exercise test (GXT) supervised by a cardiologist not associated with the study. The study coordinator, a Master's prepared kinesiologist, reviewed GXT results with the patient, provided them with a summary of their results and gave the patient general advice on continued exercise.20 21 The patient was thanked for his/her participation in the research study, was informed that no further study follow-up was planned and given reimbursement for parking expenses.

Outcomes and follow-up

The primary outcome was exercise capacity (peak Vo2, ml/min) assessed by a symptom-limited GXT with direct measurement of Vo2 using a SensorMedics Ergo-metrics 800S cycle ergometer and a SensorMedics Vmax 229 metabolic cart (Yorba Linda, California, USA), calibrated prior to each test. Peak Vo2 was defined as the average value of three measurements recorded in the last minute of exercise. Work rate increments were 15 watts per min until symptom limits were reached. Blood pressure was measured at the end of every minute of exercise using a sphygmomanometer. Heart rate and 12-lead ECG (Marquette case 15-Electrocardiograph; Milwaukee, Wisconsin, USA) were monitored continuously and reported at the end of every minute. Exercise tests were terminated if the patient had a fall in systolic blood pressure of >10 mm Hg from baseline, ST elevation >1 mm in leads without diagnostic Q waves (other than V1 or AVR), sustained ventricular tachycardia and/or signs of poor perfusion. Reasons for patients' termination of the test were leg fatigue, dyspnoea, dizziness or angina. The protocol and staff for the GXT were consistent throughout the study (1997–2005). Peak METs is also reported to allow for comparison with the CR literature and was calculated using the formula (Vo2 in ml/min/body weight in kg)/3.5 ml/kg/min.

Exercise maintenance was a secondary outcome. Exercise maintenance was defined as a decline in peak Vo2 of <10% between 1 year and LTF or an exercise capacity ≥80% of age and sex predicted maximum. Thresholds for exercise maintenance were calculated based on Fleg et al22 who indicated a decline in peak Vo2 of 2% per year with natural ageing.22 23 A decline in peak Vo2 <10% between 1 year and LTF was defined as successful exercise maintenance. Only changes between 1 year and LTF were explored.

Habitual physical activity was assessed at 1 year and LTF using the Physical Activity Scale for the Elderly (PASE).24 25 The PASE is simple to administer; it contains items related to inside and outside the home activities, as well as items suitable for both men and women. Scores from the PASE can range from 0 to 360, with 0 being a very limited level of physical activity and 360 being a very high level of physical activity. The PASE was originally developed and validated in a sample of community-dwelling older adults.24 25 Mean PASE score in the original study was 102.9±64.1 in participants with a mean age of 73. Since that time, Washburn et al have continued to further develop the tool in a RCT on the effect of aerobic conditioning on psychological function.25 The latter study involved patients with chronic medical conditions such as cardiovascular disease and correlated PASE scores with peak Vo2 as further evidence of construct validity. Among participants aged ≥65 years, the mean PASE score was 118.9±63.9. Bianchi et al also found the PASE to be a valid tool for assessment of physical activity in patients with the metabolic syndrome.26

Anthropometric data were collected throughout the study. Each patient was assessed for weight (kg), height (m), waist and hip circumference (cm) to calculate waist-to-hip ratio (WHR), and body mass index (BMI, kg/m2). WHR of >0.9 for men and >0.8 for women demonstrate increased cardiovascular risk. BMI was assessed using criteria for overweight (BMI 25–29.9 kg/m2) and obesity (BMI ≥30 kg/m2) as defined by the Centers for Disease Control.27

Statistical analysis

Analyses were performed based on an intention-to-treat principle. Categorical variables were compared using χ2 or Fisher exact tests. Differences between the two groups at 1 year were compared for continuous outcomes using a two-sided t test for independent samples. For analysis over time, continuous variables were analysed using a two-way repeated measures analysis of variance or analysis of covariance. An a priori decision to impute missing data for the primary outcome was established. A conservative strategy towards imputation of the missing data was conducted. Missing values were imputed using a random assignment based on mean peak Vo2 values (±1SD) of similar age- (5-year age cohorts) and sex-matched LTF participants. Group assignment was not a factor in the imputation strategy and the imputed value could not be greater than the GXT results from the 1-year follow-up. We adjusted significance levels for multiple testing using Bonferroni for pairwise comparisons over time.28 Variables significantly different between groups at baseline or 1 year were used as covariates.

Statistical analysis was conducted using SPSS version 11.2 for Mac OSX. A one- or two-sided α ≤0.05 was considered statistically significant.



Vital status was obtained on 96.4% of eligible participants. Seventeen (8.7%) deaths were observed between 1 year and LTF, 23 deaths among the original 242 RCT participants.7 8

Of the 196 eligible participants, 144 (73.5%) agreed to participate in the LTF (figure 1). Participation rates were similar among Home (n=70; 72.9%) and Hospital (n=74; 74%) groups (p=0.873).

The reasons for non-participation are outlined in table 1 and were similar between the groups (p=0.588). Reported causes of death included cancer (n=9), cardiac (n=4), aneurysm (n=2), renal failure (n=1) and unknown (n=1). Additionally, 22 patients in the Home group (31.4%) and 14 in the Hospital group (18.9%) were unwilling or unable to perform a GXT. Questionnaire and/or anthropometric data for these patients are included.

Table 1

Reasons for non-participation in the long-term follow-up (LTF) study

Participant characteristics are reported in table 2. LTF occurred 6.6±0.3 years after discharge from CR and 7.3±0.6 years after CABG. LTF participants were primarily male (84.7%), living with a spouse or partner (80%), with a mean age of 70 years. The sex distribution was stable throughout the study.

Table 2

Subject characteristics

Similar numbers of Hospital (n=14) and Home (n=13) patients transitioned to retirement between 1 year and LTF. There were 16 new cases of diabetes diagnosed between 1 year and LTF (no significant difference between groups).

Clinical events

No significant differences were observed between Hospital and Home groups in the number or proportion of clinical events (table 3). New cancer diagnoses were observed in 19 patients (13.2%). The total number (p=0.001) and the distribution (p=0.025) of hospitalisations were significantly different between the Home (n=42) and Hospital (n=79) groups, but the median time to first hospitalisation was similar (4.49 years, p=0.616). Reasons for hospitalisations were similarly distributed between the two groups and included cancer (n=32), cardiac catheterisation (n=13), percutaneous coronary intervention (n=6), myocardial infarction (MI) (n=13), angina (n=5), cardiac arrest (n=2), heart failure (CHF) (n=5), other cardiac (n=4), stroke (n=13), peripheral vascular disease (n=4), orthopaedic surgery (n=14) and other (n=5).

Table 3

Clinical events between 1 year and long-term follow-up

Exercise capacity

Of the 144 participants, 108 (75.0%) completed a GXT (Home: n=48 (68.6%); Hospital: n=60 (81.1%); p=0.09). Reasons for not completing a GXT included arthritis (n=6; 16.7%), back pain (n=1; 2.8%), hypertension (n=1; 2.8%), CHF (n=4; 11.1%), cancer (n=3; 8.3%), cataract surgery (n=2; 5.6%), elbow surgery (n=1; 2.8%), a severe fall (n=1; 2.8%), peripheral vascular disease (n=1; 2.8%), problem walking (n=1; 2.8%), stroke (n=4; 11.1%), unable to attend in person (n=4; 11.1%), work (n=3; 8.3%) or refused (n=3; 8.3%) (p=0.388). Patients who did not complete a GXT were older (73.6 vs 69.1 years; p=0.013) and had a lower peak Vo2 at 1 year (1376 vs 1675 ml/min; p<0.0001). There were no significant differences in peak Vo2 at 1 year follow-up between patients in the Hospital (1384±405 vs 1618±396 ml/min) and Home (1333±398 vs 1762±392 ml/min) groups who did not perform a GXT at LTF (p=0.224).

Mean peak Vo2 in the Home group (1621±472 ml/min, n=48) was significantly greater than in the Hospital group (1418±373 ml/min, n=60; p=0.01) at LTF. Change scores and annualised estimates of decline presented were calculated using the raw data (non-imputed results) for patients who attended the LTF and completed the GXT (n=108). Table 4 outlines the imputed exercise outcomes for the study. There was no significant difference between the imputed and non-imputed results.

Table 4

Exercise and anthropometric profiles over the 6-year study

Peak Vo2 declined significantly in both groups between 1 year and LTF (p<0.0001). The relative rate of decline between 1 year and LTF was greater for the Hospital group (10.4±14.9%, n=60) than for the Home group (6.0±11.1%, n=48; p=0.045). Despite this decline, peak Vo2 remained significantly elevated above baseline in both the Home and Hospital groups (p<0.0001).

Exercise maintenance

Exercise maintenance was estimated for participants who completed a GXT at the LTF (n=108). Results are reported separately. A total of 45.8% (n=27) of Hospital participants and 58.3% (n=28) of Home participants met the criteria for exercise maintenance at the LTF (p=0.244). Additionally, 58.6% of Hospital participants and 66.7% of Home participants achieved >80% of predicted peak Vo2 at LTF (p=0.427).

Habitual physical activity

PASE scores were significantly higher among patients in the Home group than in the Hospital group at the 1-year follow-up (mean difference 56.4 points, 95% CI 31.4 to 85.5; p<0.0001). This difference persisted at LTF (p=0.042). Between 1 year and LTF, both groups showed a decline in PASE scores. The decline was greater among Home participants (26% decline) than in Hospital participants (19%; p=0.042). Despite the greater decline, PASE scores in the Home group were significantly greater at both 1 year and LTF compared with the Hospital group (42.3, 95% CI 16.7 to 67.9; p=0.001).


The Home group weighed approximately 3.1 kg less (95% CI −0.673 to 6.91) than patients in the Hospital group at baseline (p=0.066; table 4). Both groups gained weight during CR, a trend that continued at the 1 year and LTF (p<0.0001). On average, weight increased about 2 kg from 1 year to LTF (p=0.006) with no significant between-group differences. Home participants trended towards improved WHR profiles throughout the study, but this failed to reach significance. At LTF, 17.5% (n=13) of the Hospital group and 27.1% (n=19) of the Home group participants were within therapeutic targets for WHR (p=0.247). Home group participants also reported lower BMI throughout the study (p=0.041). BMI increased significantly between discharge and 1 year and LTF (table 4). Importantly, approximately 50% of Home and Hospital patients were considered overweight throughout the study and 20% met the criterion for obesity.


This is one of the longest follow-up studies of patients who participated in a RCT of monitored home versus hospital-based exercise training during CR. The investigation revealed several important findings. First, both home and hospital-based exercise training resulted in improvements in exercise capacity that were maintained above baseline (CR entry) up to 6.6 years after CR. Second, low-risk patients after CABG who initiated exercise training in the home environment preserved exercise capacity to a greater extent than hospital-based CR. Third, home-based exercise participants reported significantly higher levels of habitual physical activity than the hospital-based group. These findings extend our previous research8 on the sustainability of the physical outcomes of CR after CABG surgery and provide further evidence supporting monitored home-based exercise training for selected patients.

Seventeen (8.7%) deaths were observed between the 1 year and LTF (23 deaths among 242 randomised patients over 6.6 years).7 8 This is similar to other studies following patients after CR where rates ranged from 2.4% at 2 years to 14.6% up to 6 years after CR.16 19 29 30 A recent systematic review reported no significant difference in total mortality between home or centre-based CR within 3–12 months of follow-up.19 Jolly et al extended these findings to 24 months after CR.31 Additionally, in the present study cardiac mortality was 2.1%, which is similar to other published reports of CR participants.29 30 Collectively, these findings suggest that cardiovascular mortality is low among post-CABG CR participants up to 6.6 years after participation.

The type and distribution of non-fatal clinical events were similar among Home (55.7%) and Hospital patients (66.2%) at LTF. Cardiac events were observed in 40.5% of Hospital patients and 34.3% of Home patients during the 6.6 years after CR. Reid et al30 found that 44 of 392 (11.2%) CR participants experienced cardiac events within 24 months after CR. Home patients also reported fewer hospitalisations within the follow-up period (42 Home vs 79 Hospital, p<0.05). Of these, 5.1% were hospitalised for myocardial infarction. Contrary to this finding, Dalal et al32 and Jolly et al31 reported no significant differences in cardiac events between home and centre-based CR during up to 24 months of follow-up. The length of follow-up or the level of comorbidity among the patient populations may account for these differences. The present findings are consistent with other controlled trials.16 17 However, the novel finding is that home-based exercise training resulted in fewer hospitalisations than standard hospital-based CR.

A successful long-term outcome of CR is to achieve sustained lifestyle changes including an improvement in exercise capacity. Despite the significant decline in peak Vo2 observed between the 1 year and LTF, peak Vo2 remained 19.5% and 14% above baseline levels in the Home and Hospital groups, respectively. While it is acknowledged that peak Vo2 is reduced following CABG due to deconditioning and spontaneously improves during early recovery even without CR intervention,33 the maintenance of peak Vo2 above baseline levels remains important. Together with the observation that the decline in peak Vo2 between 1 year and LTF was less than expected, this suggests that enhanced exercise maintenance is a long-term benefit of CR. This sustained improvement in peak Vo2 has been observed by others with a shorter period of follow-up.11–15 19 In a recent systematic review Dalal et al19 reported that exercise capacity among home-based CR participants was better (but not significantly) than centre-based CR, although only three studies met the criteria for inclusion. In a RCT following 394 patients for a mean 4.1 years after CR, Lear et al11 reported a 5.6% decline in peak METs, which was similar in patients randomised to CR maintenance or standard care. Bjarnason-Wehrens et al34 reported a slight (4.6%) but significant increase in peak workload (Watts) 2 years after CR. A large proportion of patients (61%) in this study reported daily physical activity within the past year.

The decline in peak Vo2 was greater in the Hospital group (∼10%) than in the Home group (6%). This is important and extends our previous findings.8 In patients after MI, home-based CR resulted in maintained total work capacity but declined among hospital-based CR participants 1 year after CR.9 A recent study demonstrates that CR improves stress myocardial blood flow, which may relate to the mechanism of action underlying our findings.35 While this has been shown in post-MI patients, the mechanism is likely to apply to patients after CABG as well. Thus, it may be a partial explanation for the better outcomes in the group that continued to sustain long-term activity. Mechanisms underlying the decline in peak Vo2 and strategies to mitigate them require further exploration.

Few studies have objectively examined the improvement in and sustainability of habitual physical activity after CR. Habitual physical activity declined between 1 year and LTF in both Home and Hospital groups. The decline in the Hospital group (approximately 19%) was less than that of the Home group (26%). Several factors may account for the difference in decline while maintaining peak Vo2. The most likely reason relates to exercise intensity. Walking was the prescribed exercise for patients in the home group. The PASE measures walking as a single item rated on length of time in the activity and its weekly frequency, but not its intensity. For instance, an individual walking at 2 mph would receive the same score for this activity as someone walking 3 mph given time and frequency as constants. This explanation may be supported by the Home group reporting a lower rate of decline in peak Vo2 during the follow-up period in comparison to the Hospital group.

Mean PASE scores were higher in both groups compared with the study by Washburn et al of participants ≥65 years of age (mean PASE 118.9±63.9) participating in a 6-month randomised controlled exercise trial to evaluate the effect of aerobic conditioning on psychological function. This finding suggests that CR, whether home- or hospital-based, may provide benefit to habitual physical activity.25 Mean PASE scores were higher in the Home group (166.7±90.2; p<0.0001) but not in the Hospital group (139.7±66.5; p=0.861), compared with community dwelling Canadians (130.6±59.9) aged 65–79 years with no chronic conditions,36 suggesting a positive long-term benefit of home CR. Lear et al11 reported a slight decline in habitual physical activity in both the intervention (27.8%) and control (21.9%) groups 4 years after CR. Similar to our findings, habitual physical activity assessed as the number of weekly exercise sessions (∼4.5 vs 2.5)15 or as the energy expenditure (3127 kcal/wk vs 977 kcal/wk)37 was higher than pre-CR levels 1 and 2 years after CR. Together with the results of the present study, these studies suggest that, while habitual activity may decline modestly after CR, participation results in sustained improvements in habitual activity.

Study limitations

While this represents one of the longest RCTs in the literature on CR, some limitations need to be addressed. Patients who agreed to and were available for participation in the LTF may be systematically different from those unable to participate, which could limit the generalisability of the findings. Our patient population was, however, similar to both our current population of patients undergoing CABG surgery and entering CR after CABG, thus mitigating the impact of this limitation. Since patients were not followed up at regular intervals, the rate of decline in the primary outcome may be overestimated. This trajectory of decline in peak Vo2 requires further study. This was a single institution study. Patients at this institution are automatically referred to CR after CABG, which improves the overall generalisability of the results to tertiary care centres within a socialised medicine framework. Finally, the PASE has not been validated specifically in patients undergoing cardiac surgery. However, since the development of the PASE, researchers including Washburn et al, the original developers of the scale, have used the tool in patient populations with characteristics including cardiovascular disease that are very similar to those who participated in this study. Despite these limitations, this study represents a comprehensive follow-up of outcomes after CR and adds new important information on the benefits of CR over the long term.


To our knowledge, this is the longest follow-up of low-risk patients after CABG surgery using direct measurement of peak Vo2 and validated tools to assess habitual physical activity, more than 6 years after the completion of an RCT of exercise training during CR. This is an important study that contributes several new pieces of information to the literature on CR. Taken together, monitored exercise training initiated in the home environment within 10 weeks after CABG resulted in greater long-term benefits compared with exercise training during CR within a hospital setting. Peak exercise capacity was maintained above pre-CR levels up to 6 years after the conclusion of CR with either home or hospital-based training, demonstrating that CR does improve the acquisition and long-term sustainability of exercise capacity long after CR has been completed. Habitual physical activity, measured using the PASE, was higher in the Home group and similar in the Hospital group compared with population norms for healthy community dwelling elderly Canadians.


The authors would like to thank the staff of the Cardiac Health and Rehabilitation Center and the Medical Diagnostic Unit of the Hamilton Health Sciences—General Site for their invaluable contributions to the study.



  • Funding This research was supported in part by the Heart and Stroke Foundation of Ontario (grant # T4004). KMS was funded by a Canadian Graduate Scholarship from the Canadian Institutes of Health Research, the Margaret McWilliams Fellow of the Canadian Federation of University Women and a doctoral research award from the Canadian Association of Cardiac Rehabilitation. HMA holds the Heart and Stroke Foundation of Ontario/Michael G DeGroote Endowed Chair in Cardiovascular Nursing Research. The study investigators were independent of the sponsors. The sponsors played no role in the design, analysis/interpretation or dissemination of the study findings.

  • Competing interests KMS, RSM, KET and HMA have no relationships that might have an interest in the submitted work in the previous 3 years. KMS was the first author of the previous 1-year follow-up of home versus hospital-based exercise training during CR after CABG; she was also a co-author of the original RCT. HMA was the first author of the original RCT of home versus hospital-based exercise training during CR after CABG and a co-author of the 1-year follow-up study. RSM was a co-author on both the RCT and 1-year follow-up studies related to the present work.

  • Ethics approval This study was conducted with the approval of the Joint Research Ethics Board of McMaster University and the Hamilton Health Sciences, Hamilton, Ontario, Canada. The study complied with the Declaration of Helsinki and was approved by the joint institutional review board of McMaster University and the HHS (Protocol #05-034).

  • Provenance and peer review Not commissioned; externally peer reviewed.