Association Between Home-Based Exercise Modality and Improvements in Exercise Capacity During Outpatient Cardiac Rehabilitation Authors: Ryosuke Aotani, Daisuke Tomioka, Shuto Tanaka, Kota Tsutsui, Atsushi Tanabe, Kenji Watanabe, Yuki Uchida, Yukiko Nakata, Sho Yasui, Tatsuya Iwasaki, Naofumi Oyamada, Toshikazu Jinnai, Kazuaki Kaitani Published: 2026 ABSTRACT BackgroundPeak oxygen uptake (V̇O2) is a clinically meaningful indicator of functional capacity and prognosis during cardiac rehabilitation (CR). Although home-based exercise is essential in outpatient CR, the impact of different exercise modalities on exercise capacity remains unclear.Methods and ResultsThis retrospective observational study included 69 patients who underwent outpatient CR and completed cardiopulmonary exercise testing at baseline and follow up. Home-based exercises were classified as ergometer based or walking based. Exercise self-efficacy was assessed as a behavioral factor. When the peak V̇O2 was analyzed as a continuous outcome via analysis of covariance, ergometer-based home exercises were independently associated with a higher post-intervention peak V̇O2 after adjustment for baseline peak V̇O2, age, and sex. This association was attenuated after additional adjustment for changes in exercise self-efficacy. The proportion of patients with ≥6% improvement in peak V̇O2 was non-significantly higher in the ergometer-based exercise group.ConclusionsIn an outpatient CR setting, ergometer-based home exercises were associated with modestly greater improvement in exercise capacity, assessed as peak V̇O2, particularly as a continuous outcome. Although self-efficacy was not an independent predictor, its association with this relationship highlights the potential importance of behavioral engagement in the relationship between prescribed exercises and physiological improvements. The principles underlying ergometer-based exercise – such as structured intensity prescription and objective feedback – may be applicable to broader home-based exercise programs, including walking-based interventions. Central Figure Cardiac rehabilitation (CR) centered on structured exercise therapy has been shown to improve exercise capacity, reduce hospitalizations, and improve the prognosis of patients with cardiovascular diseases, including coronary artery disease and heart failure.1,2 Improvements in exercise capacity, particularly the peak oxygen uptake (V̇O2), are strongly associated with favorable clinical outcomes and are widely used as clinically meaningful indicators of functional capacity during CR.3,4 In Japan, outpatient CR programs typically combine supervised exercise sessions with home-based exercises during the maintenance phase, as supported by both clinical studies and guideline recommendations.5,6 Home-based exercises are critical to sustain training effects beyond supervised settings; however, substantial heterogeneity in their implementation has been reported, including differences in exercise modalities, intensity control, and patient engagement.7–9 Despite the growing interest in home-based CR, the clinical impact of specific home-based exercise modalities on improvements in exercise capacity remains insufficiently characterized.10,11 Exercise modalities may influence the translation of prescribed exercises into physiological adaptations, particularly during unsupervised periods. Ergometer-based exercises allow precise workload prescriptions and objective feedback, whereas walking-based exercises are more accessible but may be associated with greater variability in exercise intensity. In addition to physiological factors, behavioral constructs such as exercise self-efficacy are known to influence exercise adherence and maintenance, potentially modifying the relationship between exercise modalities and training effects.12–14 Recent systematic reviews and scientific statements have confirmed the effectiveness of both center- and home-based CR in contemporary clinical practice.1,2,7,9 However, the interplay between home-based exercise modalities, behavioral engagement, and improvements in exercise capacity during outpatient CR has not been sufficiently explored in real-world clinical settings. Accordingly, we conducted this study to investigate the association between home-based exercise modalities and improvements in exercise capacity, as assessed by changes in peak V̇O2, in patients undergoing outpatient CR. We further examined whether changes in exercise self-efficacy were associated with this relationship. By focusing on standardized cardiopulmonary exercise testing (CPX) and a Japanese CR outpatient population, we sought to provide clinically relevant insights into the translation of home-based exercise prescriptions into physiological improvements during the maintenance phase of CR. Methods Study Design and Participants This retrospective observational study was conducted in accordance with the STROBE statement.15 The study was conducted at a single outpatient CR center at Otsu Red Cross Hospital from April 2019 to October 2025. Patients who participated in outpatient CR in accordance with the criteria of the Japanese Association of Cardiac Rehabilitation were screened for eligibility. Patients were included if they: (1) completed outpatient CR, (2) underwent CPX at baseline and after the intervention period; and (3) documented their home-based exercise modalities during the study period. Patients with missing CPX data or incomplete information regarding home-based exercises were excluded. The study protocol was approved by the Ethics Committee of Otsu Red Cross Hospital (reference no. 962), which waived the requirement for informed consent because of the study’s retrospective nature. The procedures followed were in accordance with the Declaration of Helsinki and the ethical standards of the responsible committee on human experimentation. Outpatient CR Program All patients participated in a standardized outpatient CR program in accordance with the guidelines of the Japanese Circulation Society and the Japanese Association of Cardiac Rehabilitation.5 The program consisted of supervised exercise sessions combined with home-based exercises during the maintenance phase. Home-based exercise prescriptions were individualized according to patients’ clinical status and exercise tolerance. The choice of home-based exercise modality (ergometer or walking based) was determined by the clinician’s judgment and patient preference, reflecting routine clinical practice.6 The program included once-weekly supervised exercise sessions at the outpatient CR center and twice-weekly home-based aerobic exercise sessions, each lasting approximately 30 min. Home-Based Exercise Modalities In addition to supervised outpatient sessions, patients were instructed to perform regular home-based aerobic exercises. Patients were categorized into 2 groups based on the primary prescribed modality: (1) an ergometer-based exercise group, in which patients performed cycling exercises by using a home ergometer; and (2) a walking-based exercise group in which walking was the primary aerobic activity. The prescribed exercise intensity for the home-based exercises was individually determined based on the CPX results. In both groups, the initial exercise intensity was set at the workload corresponding to 1 min before the anaerobic threshold (AT) during the baseline CPX. In the ergometer-based exercise group, this corresponded to a specific wattage target, and the intensity was managed using the workload display (watts) and cadence (revolutions per minute) targets. Because ergometers for home use can have either a digital or analog dial, the precision of workload adjustment varied; however, patients were instructed to align their efforts with the prescribed wattage as closely as possible by monitoring the digital display of the ergometer. In the walking-based exercise group, exercise intensity was guided primarily by heart rate 1 min before the AT, supplemented by walking speed and the Borg rating of perceived exertion. During the weekly supervised sessions, the exercise response was monitored using the Borg scale and heart rate, and the prescribed home-based exercise intensity was progressively adjusted as tolerated. Adherence to the prescribed frequency and intensity of home-based exercises was verbally confirmed during these supervised sessions; however, objective exercise logs were not routinely collected. CPX CPX was performed at baseline and follow up on an electronically braked cycle ergometer using a symptomlimited, incremental exercise protocol. Breath-by-breath gas exchange was measured continuously by using a calibrated metabolic cart (Minato Medical Science Co., Ltd, Osaka, Japan), with simultaneous 12-lead electrocardiographic monitoring and intermittent blood pressure measurements throughout the test, in accordance with established guidelines.16,17 Peak V̇O2 was defined as the highest 30-s average V̇O2 value obtained during the exercise test. The test was terminated once a maximal effort was achieved. Maximal effort was defined based on the European Respiratory Society’s statement on the standardization of CPX,17 as the attainment of at least 2 of the following: a peak respiratory exchange ratio >1.10; achievement of the age-predicted maximal heart rate; or a plateau in V̇O2. Baseline and follow-up CPX were conducted using identical protocols and equipment to ensure comparability.18 Although the anaerobic threshold and ventilatory efficiency parameters, including the minute ventilation-to-carbon dioxide output slope, were assessed as part of routine clinical evaluation, the peak V̇O2 was selected as the primary indicator of exercise capacity because it represents an integrated measure of cardiovascular, pulmonary, and peripheral muscular functions.19 Assessment of Exercise Self-Efficacy Exercise self-efficacy was assessed using a validated questionnaire to measure patients’ confidence in performing and maintaining exercise behaviors.14 The change in self-efficacy was examined as a behavioral factor based on established theoretical and clinical frameworks.12,20 Definition of Responders Responders were defined as patients who achieved a relative improvement of ≥6% in peak V̇O2 from baseline to follow up. This threshold was selected to exceed the expected measurement variability in peak V̇O2 during CPX and reflect a clinically meaningful change in exercise capacity, as supported by previous studies of patients undergoing CR.4,6 Sensitivity analyses were performed using a stricter responder definition of ≥10% improvement in peak V̇O2. Statistical Analysis Categorical data are presented as counts and percentages. Continuous data are reported as means±standard deviation if normally distributed, and as the median (quartile 1, quartile 3) if not normally distributed. Between-group comparisons of continuous variables were performed using the unpaired t-test for normally distributed data and the Mann–Whitney U test for nonparametric data. Categorical variables were compared using the Fisher exact test. Factors associated with a peak V̇O2 improvement of ≥6% or ≥10% were analyzed using univariate logistic regression models. Subsequently, multivariable logistic regression analyses were performed with the selected variables forced into the model. Univariate linear regression analyses were performed to examine the effects of variables on post-intervention peak V̇O2. Multivariable linear regression analyses were also performed, forcing selected variables into the analysis. The primary outcome was the association between home-based exercise modalities and post-intervention peak V̇O2, analyzed as a continuous outcome using analyses of covariance, with adjustment for baseline peak V̇O2, age, and sex. The secondary outcome was the association of factors with responder status, determined using multivariable logistic regression models. Covariates were selected a priori based on clinical relevance and prior studies rather than data-driven procedures. Age, sex, and baseline peak V̇O2 were included as fundamental physiological determinants of exercise capacity. The change in exercise self-efficacy was also included in exploratory models as a behavioral factor to potentially explain the association between exercise modality and improvements in peak V̇O2.21,22 Multicollinearity was assessed using variance inflation factors. Covariates were predefined based on clinical relevance and prior reports to minimize the risk of overfitting.23–25 Sensitivity analyses were also performed using penalized likelihood estimation. All statistical analyses were performed using IBM SPSS Statistics for Windows version 29.0.2.0 (IBM Japan, Tokyo, Japan). A 2-sided P value <0.05 was considered statistically significant. Results Baseline Characteristics In total, 69 patients were included in this study (ergometer group, n=45; walking group, n=24). As shown in Table 1, the groups did not differ in terms of age (66.7±9.7 vs. 69.7±8.4 years; P=0.209), sex distribution (male: 76% vs. 88%; P=0.349), or baseline peak V̇O2 (20.2±4.4 vs. 20.5±5.2 mL/min/kg; P=0.759). In contrast, baseline left ventricular ejection fraction and B-type natriuretic peptide levels differed significantly between the groups. Table 1.Clinical Characteristics nErgometer groupnWalking groupP valueAge (years)45 24 Mean±SD 66.7±9.7 69.7±8.40.209b Median [Q1, Q3] 69.0 [61.0, 73.0] 71.5 [60.8, 76.8]0.221cSex45 24 0.349a Male 34 (75) 21 (87.5) Female 11 (24) 3 (12.5) Body weight (kg)45 24 Mean±SD 65.5±11.9 65.6±10.70.978b Median [Q1, Q3] 65.4 [59.1, 73.7] 66.3 [57.8, 75.7]0.738cBMI (kg/m2)45 24 Mean±SD 23.6±3.1 24.0±2.70.609b Median [Q1, Q3] 24.0 [21.0, 25.8] 24.3 [22.0, 26.0]0.623cIndications45 24 0.462a Angina pectoris 10 (22) 5 (21) Myocardial infarction 22 (49) 12 (50) Heart failure 0 (0) 2 (8) Vascular disease 3 (7) 0 (0) Chronic heart failure 5 (11) 3 (13) Post-cardiac surgery 5 (11) 2 (8) Comorbidities45 24 Hypertension 20 (44) 6 (25)0.127a Diabetes 34 (76) 15 (63)0.278a Dyslipidemia 16 (36) 5 (21)0.179a Smoking45 24 0.640a No smoking history 21 (47) 7 (29) Previously smoked 10 (22) 7 (29) Quit after the onset of illness 11 (24) 7 (29) Currently smoking 3 (7) 2 (8) Laboratory data Creatinine (mg/dL)44 24 Mean±SD 1.00±0.33 1.05±0.380.590b Median [Q1, Q3] 0.93 [0.83, 1.08] 0.94 [0.81, 1.17]0.700c Hb (g/dL)44 24 Mean±SD 13.3±1.6 13.8±1.40.262b Median [Q1, Q3] 13.4 [11.9, 14.5] 14.1 [13.3, 14.6]0.227c Na (mmol/L)43 24 Mean±SD 141.4±2.2 141.0±2.30.517b Median [Q1, Q3] 142.0 [140.0, 143.0] 141.0 [140.0, 143.0]0.539c CRP (mg/dL)42 23 Mean±SD 0.12±0.18 0.18±0.270.261b Median [Q1, Q3] 0.10 [0.00, 0.10] 0.10 [0.00, 0.20]0.589c BNP (pg/mL)45 24 Mean±SD 79.3±123.4 162.1±189.20.032b* Median [Q1, Q3] 79.3 [20.0, 80.0] 108.5 [26.5, 175.3]0.036c* EF (%)45 24 Mean±SD 59.5±13.5 50.8±14.30.015b* Median [Q1, Q3] 64.0 [52.0, 68.0] 53.0 [40.6, 59.5]0.008c*Medication use45 24 ACE inhibitors or ARB 24 (53) 12 (50)0.806a β-blockers 16 (36) 8 (33)>0.999a SGLT2 inhibitors 31 (69) 15 (63)0.603a ARNI 40 (89) 19 (79)0.300a MRAs 37 (82) 18 (75)0.537a Diuretics 40 (89) 20 (83)0.709aCardiopulmonary exercise testing Peak V̇O2 (mL/min/kg)45 24 Mean±SD 20.2±4.4 20.5±5.20.759b Median [Q1, Q3] 19.7 [17.4, 22.9] 19.9 [17.1, 23.4]0.985c Peak V̇O2 (%)45 24 Mean±SD 84.4±16.9 86.9±21.30.596b Median [Q1, Q3] 84.0 [73.5, 98.0] 83.0 [74.0, 94.8]0.985c Peak V̇O2/HR (mL/beat)45 24 Mean±SD 9.8±2.4 10.4±2.30.326b Median [Q1, Q3] 10.1 [9.2, 11.4] 10.4 [8.6, 12.2]0.492c Peak V̇O2/HR (%)45 24 Mean±SD 75.3±13.9 79.0±16.90.341b Median [Q1, Q3] 74.0 [67.0, 83.5] 77.5 [69.5, 89.5]0.381c V̇E/V̇CO2 slope45 24 Mean±SD 31.1±5.0 30.1±4.40.443b Median [Q1, Q3] 30.7 [27.9, 33.5] 30.5 [26.2, 33.3]0.641c ΔV̇O2/ΔWL (mL/min/W)45 24 Mean±SD 9.9±1.2 9.7±1.40.437b Median [Q1, Q3] 10.1 [9.3, 10.7] 9.2 [8.4, 11.1]0.319c ΔV̇O2/ΔWL, %45 24 Mean±SD 96.3±12.1 93.7±13.70.427b Median [Q1, Q3] 98.0 [90.0, 103.5] 89.5 [81.3, 107.0]0.313c AT METS (mL/min/kg)45 24 Mean±SD 3.7±0.7 3.7±0.70.925b Median [Q1, Q3] 3.6 [3.2, 4.2] 3.7 [3.3, 4.0]0.990c Peak METS44 24 Mean±SD 5.7±1.2 5.9±1.50.716b Median [Q1, Q3] 5.6 [5.0, 6.6] 5.7 [4.9, 6.7]0.928cPhysical function assessment 6MD (m)45 24 Mean±SD 515.7±77.7 507.9±85.50.703b Median [Q1, Q3] 520.0 [472.5, 570.0] 510.0 [438.8, 562.5]0.549c SPPB score45 24 Mean±SD 11.9±0.4 11.9±0.40.893b Median [Q1, Q3] 12.0 [12.0, 12.0] 12.0 [12.0, 12.0]0.959c Self-efficacy score45 24 Mean±SD 12.8±3.5 13.9±4.40.280b Median [Q1, Q3] 13.0 [10.0, 15.0] 13.5 [11.0, 18.3]0.285c*P<0.05. The data are presented as n, %, unless indicated otherwise. P values were determined using the following tests: a, Fisher’s exact tests; b, unpaired t-tests; c, Mann–Whitney U tests. 6MD, 6-min walk distance; ACE, angiotensin-converting enzyme; ARNI, angiotensin receptor-neprilysin inhibitor; ARB, angiotensin II receptor blocker; AT, anaerobic threshold; BNP, B-type natriuretic peptide; CRP, C-reactive protein; EF, ejection fraction; Hb, hemoglobin; HR, heart rate; METS, metabolic equivalents; MRA, mineralocorticoid receptor antagonist; Na, sodium; Q, quartile; SD, standard deviation; SGLT2, sodium/glucose cotransporter 2; SPPB, Short Physical Performance Battery; V̇O2, oxygen uptake; WL, workload. Table 1.Clinical Characteristics nErgometer groupnWalking groupP valueAge (years)45 24 Mean±SD 66.7±9.7 69.7±8.40.209b Median [Q1, Q3] 69.0 [61.0, 73.0] 71.5 [60.8, 76.8]0.221cSex45 24 0.349a Male 34 (75) 21 (87.5) Female 11 (24) 3 (12.5) Body weight (kg)45 24 Mean±SD 65.5±11.9 65.6±10.70.978b Median [Q1, Q3] 65.4 [59.1, 73.7] 66.3 [57.8, 75.7]0.738cBMI (kg/m2)45 24 Mean±SD 23.6±3.1 24.0±2.70.609b Median [Q1, Q3] 24.0 [21.0, 25.8] 24.3 [22.0, 26.0]0.623cIndications45 24 0.462a Angina pectoris 10 (22) 5 (21) Myocardial infarction 22 (49) 12 (50) Heart failure 0 (0) 2 (8) Vascular disease 3 (7) 0 (0) Chronic heart failure 5 (11) 3 (13) Post-cardiac surgery 5 (11) 2 (8) Comorbidities45 24 Hypertension 20 (44) 6 (25)0.127a Diabetes 34 (76) 15 (63)0.278a Dyslipidemia 16 (36) 5 (21)0.179a Smoking45 24 0.640a No smoking history 21 (47) 7 (29) Previously smoked 10 (22) 7 (29) Quit after the onset of illness 11 (24) 7 (29) Currently smoking 3 (7) 2 (8) Laboratory data Creatinine (mg/dL)44 24 Mean±SD 1.00±0.33 1.05±0.380.590b Median [Q1, Q3] 0.93 [0.83, 1.08] 0.94 [0.81, 1.17]0.700c Hb (g/dL)44 24 Mean±SD 13.3±1.6 13.8±1.40.262b Median [Q1, Q3] 13.4 [11.9, 14.5] 14.1 [13.3, 14.6]0.227c Na (mmol/L)43 24 Mean±SD 141.4±2.2 141.0±2.30.517b Median [Q1, Q3] 142.0 [140.0, 143.0] 141.0 [140.0, 143.0]0.539c CRP (mg/dL)42 23 Mean±SD 0.12±0.18 0.18±0.270.261b Median [Q1, Q3] 0.10 [0.00, 0.10] 0.10 [0.00, 0.20]0.589c BNP (pg/mL)45 24 Mean±SD 79.3±123.4 162.1±189.20.032b* Median [Q1, Q3] 79.3 [20.0, 80.0] 108.5 [26.5, 175.3]0.036c* EF (%)45 24 Mean±SD 59.5±13.5 50.8±14.30.015b* Median [Q1, Q3] 64.0 [52.0, 68.0] 53.0 [40.6, 59.5]0.008c*Medication use45 24 ACE inhibitors or ARB 24 (53) 12 (50)0.806a β-blockers 16 (36) 8 (33)>0.999a SGLT2 inhibitors 31 (69) 15 (63)0.603a ARNI 40 (89) 19 (79)0.300a MRAs 37 (82) 18 (75)0.537a Diuretics 40 (89) 20 (83)0.709aCardiopulmonary exercise testing Peak V̇O2 (mL/min/kg)45 24 Mean±SD 20.2±4.4 20.5±5.20.759b Median [Q1, Q3] 19.7 [17.4, 22.9] 19.9 [17.1, 23.4]0.985c Peak V̇O2 (%)45 24 Mean±SD 84.4±16.9 86.9±21.30.596b Median [Q1, Q3] 84.0 [73.5, 98.0] 83.0 [74.0, 94.8]0.985c Peak V̇O2/HR (mL/beat)45 24 Mean±SD 9.8±2.4 10.4±2.30.326b Median [Q1, Q3] 10.1 [9.2, 11.4] 10.4 [8.6, 12.2]0.492c Peak V̇O2/HR (%)45 24 Mean±SD 75.3±13.9 79.0±16.90.341b Median [Q1, Q3] 74.0 [67.0, 83.5] 77.5 [69.5, 89.5]0.381c V̇E/V̇CO2 slope45 24 Mean±SD 31.1±5.0 30.1±4.40.443b Median [Q1, Q3] 30.7 [27.9, 33.5] 30.5 [26.2, 33.3]0.641c ΔV̇O2/ΔWL (mL/min/W)45 24 Mean±SD 9.9±1.2 9.7±1.40.437b Median [Q1, Q3] 10.1 [9.3, 10.7] 9.2 [8.4, 11.1]0.319c ΔV̇O2/ΔWL, %45 24 Mean±SD 96.3±12.1 93.7±13.70.427b Median [Q1, Q3] 98.0 [90.0, 103.5] 89.5 [81.3, 107.0]0.313c AT METS (mL/min/kg)45 24 Mean±SD 3.7±0.7 3.7±0.70.925b Median [Q1, Q3] 3.6 [3.2, 4.2] 3.7 [3.3, 4.0]0.990c Peak METS44 24 Mean±SD 5.7±1.2 5.9±1.50.716b Median [Q1, Q3] 5.6 [5.0, 6.6] 5.7 [4.9, 6.7]0.928cPhysical function assessment 6MD (m)45 24 Mean±SD 515.7±77.7 507.9±85.50.703b Median [Q1, Q3] 520.0 [472.5, 570.0] 510.0 [438.8, 562.5]0.549c SPPB score45 24 Mean±SD 11.9±0.4 11.9±0.40.893b Median [Q1, Q3] 12.0 [12.0, 12.0] 12.0 [12.0, 12.0]0.959c Self-efficacy score45 24 Mean±SD 12.8±3.5 13.9±4.40.280b Median [Q1, Q3] 13.0 [10.0, 15.0] 13.5 [11.0, 18.3]0.285c*P<0.05. The data are presented as n, %, unless indicated otherwise. P values were determined using the following tests: a, Fisher’s exact tests; b, unpaired t-tests; c, Mann–Whitney U tests. 6MD, 6-min walk distance; ACE, angiotensin-converting enzyme; ARNI, angiotensin receptor-neprilysin inhibitor; ARB, angiotensin II receptor blocker; AT, anaerobic threshold; BNP, B-type natriuretic peptide; CRP, C-reactive protein; EF, ejection fraction; Hb, hemoglobin; HR, heart rate; METS, metabolic equivalents; MRA, mineralocorticoid receptor antagonist; Na, sodium; Q, quartile; SD, standard deviation; SGLT2, sodium/glucose cotransporter 2; SPPB, Short Physical Performance Battery; V̇O2, oxygen uptake; WL, workload. Clinical Characteristics *P<0.05. The data are presented as n, %, unless indicated otherwise. P values were determined using the following tests: a, Fisher’s exact tests; b, unpaired t-tests; c, Mann–Whitney U tests. 6MD, 6-min walk distance; ACE, angiotensin-converting enzyme; ARNI, angiotensin receptor-neprilysin inhibitor; ARB, angiotensin II receptor blocker; AT, anaerobic threshold; BNP, B-type natriuretic peptide; CRP, C-reactive protein; EF, ejection fraction; Hb, hemoglobin; HR, heart rate; METS, metabolic equivalents; MRA, mineralocorticoid receptor antagonist; Na, sodium; Q, quartile; SD, standard deviation; SGLT2, sodium/glucose cotransporter 2; SPPB, Short Physical Performance Battery; V̇O2, oxygen uptake; WL, workload. Primary Outcome: Peak V̇O2 as a Continuous Outcome When peak V̇O2 was analyzed as a continuous outcome via analysis of covariance, ergometer-based home exercises were independently associated with a higher post-intervention peak V̇O2 after adjustment for baseline peak V̇O2, age, and sex (Table 2, model 1). This association remained directionally consistent after additional adjustment for changes in exercise self-efficacy; however, it was attenuated and did not remain statistically significant (Table 2A, model 2). The adjusted mean post-intervention peak V̇O2 values according to home-based exercise modality are shown in Figure 1. In additional models further adjusted for baseline left ventricular ejection fraction and plasma B-type natriuretic peptide levels, the associations between exercise modality and post-intervention peak V̇O2 remained directionally consistent but were attenuated and not statistically significant (Table 2B). Table 2.Examination of Factors Affecting Post-Intervention Peak V̇O2 (Linear Regression Analysis) Univariate analysisMultivariable model 1Multivariable model 2Coeff.SEβtP valueCoeff.SEβtP valueToleranceVIFCoeff.SEβtP valueToleranceVIFA. Examination of factors affecting post-intervention peak V̇O2 (Linear regression analysis) Home exercises, walking (vs. ergometer)−1.5701.335−0.142−1.1760.244−1.4720.664−0.133−2.2190.030*0.9441.060−1.2370.664−0.112−1.8620.0670.9091.101 Age (per 1 year)−0.3400.056−0.599−6.123<0.001*−0.1500.038−0.265−4.008<0.001*0.7821.278−0.1480.037−0.260−4.011<0.001*0.7811.280 Sex, female (vs. male)−3.7281.531−0.285−2.4350.018*−0.4460.815−0.034−0.5470.5860.8781.139−0.3260.803−0.025−0.4060.6860.8721.146 Baseline peak V̇O2 (per 1 mL/min/kg)0.9490.0770.83412.358<0.001*0.8120.0770.71310.519<0.001*0.7421.3480.7970.0760.70110.467<0.001*0.7341.363 Change in self-efficacy0.3060.1310.2742.3300.023*– 0.1220.0660.1091.8350.0710.9321.073B. Examination of factors affecting post-intervention peak V̇O2 (Linear regression analysis; EF and BNP included) Home exercises, walking (vs. ergometer)−1.5701.335−0.142−1.1760.244−1.2590.728−0.114−1.7300.0890.8011.248−1.0700.722−0.097−1.4820.1440.7841.275 Age (per 1 year)−0.3400.056−0.599−6.123<0.001*−0.1570.039−0.277−4.056<0.001*0.7451.342−0.1550.038−0.272−4.050<0.001*0.7441.344 Sex, female (vs. male)−3.7281.531−0.285−2.4350.018*−0.4680.827−0.036−0.5650.5740.8691.151−0.3640.815−0.028−0.4470.6560.8651.156 Baseline peak V̇O2 (mL/min/kg)0.9490.0770.83412.358<0.001*0.7880.0840.6929.391<0.001*0.6401.5610.7780.0830.6849.426<0.001*0.6381.568 Change in self-efficacy0.3060.1310.2742.3300.023*– 0.1220.0680.1091.8010.0770.9141.094 Baseline EF (per 1%)0.0930.0440.2492.1080.039*0.0200.0250.0540.7950.4300.7491.3350.0210.0250.0550.8280.4110.7491.335 Baseline BNP (per 1 pg/mL)−0.0100.004−0.290−2.4840.015*0.0000.002−0.004−0.0630.9500.7761.2890.0000.0020.0120.1810.8570.7621.313*P<0.05. Model 1: all variables except change in self-efficacy forced into the model. Model 2: all variables forced into the model. R-squared=0.784. Adjusted R-squared=0.763. β, standardized regression coefficient; BNP, B-type natriuretic peptide; Coeff., regression coefficient; EF, ejection fraction; SE, standard error; VIF, variance inflation factor. Figure 1.Adjusted post-intervention peak oxygen uptake (V̇O2) according to home-based exercise modality. Adjusted mean values and 95% confidence intervals for post-intervention peak V̇O2 are shown for the ergometer- and walking-based exercise groups. Values were adjusted for age, sex, and baseline peak V̇O2 using analysis of covariance. Table 2.Examination of Factors Affecting Post-Intervention Peak V̇O2 (Linear Regression Analysis) Univariate analysisMultivariable model 1Multivariable model 2Coeff.SEβtP valueCoeff.SEβtP valueToleranceVIFCoeff.SEβtP valueToleranceVIFA. Examination of factors affecting post-intervention peak V̇O2 (Linear regression analysis) Home exercises, walking (vs. ergometer)−1.5701.335−0.142−1.1760.244−1.4720.664−0.133−2.2190.030*0.9441.060−1.2370.664−0.112−1.8620.0670.9091.101 Age (per 1 year)−0.3400.056−0.599−6.123<0.001*−0.1500.038−0.265−4.008<0.001*0.7821.278−0.1480.037−0.260−4.011<0.001*0.7811.280 Sex, female (vs. male)−3.7281.531−0.285−2.4350.018*−0.4460.815−0.034−0.5470.5860.8781.139−0.3260.803−0.025−0.4060.6860.8721.146 Baseline peak V̇O2 (per 1 mL/min/kg)0.9490.0770.83412.358<0.001*0.8120.0770.71310.519<0.001*0.7421.3480.7970.0760.70110.467<0.001*0.7341.363 Change in self-efficacy0.3060.1310.2742.3300.023*– 0.1220.0660.1091.8350.0710.9321.073B. Examination of factors affecting post-intervention peak V̇O2 (Linear regression analysis; EF and BNP included) Home exercises, walking (vs. ergometer)−1.5701.335−0.142−1.1760.244−1.2590.728−0.114−1.7300.0890.8011.248−1.0700.722−0.097−1.4820.1440.7841.275 Age (per 1 year)−0.3400.056−0.599−6.123<0.001*−0.1570.039−0.277−4.056<0.001*0.7451.342−0.1550.038−0.272−4.050<0.001*0.7441.344 Sex, female (vs. male)−3.7281.531−0.285−2.4350.018*−0.4680.827−0.036−0.5650.5740.8691.151−0.3640.815−0.028−0.4470.6560.8651.156 Baseline peak V̇O2 (mL/min/kg)0.9490.0770.83412.358<0.001*0.7880.0840.6929.391<0.001*0.6401.5610.7780.0830.6849.426<0.001*0.6381.568 Change in self-efficacy0.3060.1310.2742.3300.023*– 0.1220.0680.1091.8010.0770.9141.094 Baseline EF (per 1%)0.0930.0440.2492.1080.039*0.0200.0250.0540.7950.4300.7491.3350.0210.0250.0550.8280.4110.7491.335 Baseline BNP (per 1 pg/mL)−0.0100.004−0.290−2.4840.015*0.0000.002−0.004−0.0630.9500.7761.2890.0000.0020.0120.1810.8570.7621.313*P<0.05. Model 1: all variables except change in self-efficacy forced into the model. Model 2: all variables forced into the model. R-squared=0.784. Adjusted R-squared=0.763. β, standardized regression coefficient; BNP, B-type natriuretic peptide; Coeff., regression coefficient; EF, ejection fraction; SE, standard error; VIF, variance inflation factor. Examination of Factors Affecting Post-Intervention Peak V̇O2 (Linear Regression Analysis) *P<0.05. Model 1: all variables except change in self-efficacy forced into the model. Model 2: all variables forced into the model. R-squared=0.784. Adjusted R-squared=0.763. β, standardized regression coefficient; BNP, B-type natriuretic peptide; Coeff., regression coefficient; EF, ejection fraction; SE, standard error; VIF, variance inflation factor. Adjusted post-intervention peak oxygen uptake (V̇O2) according to home-based exercise modality. Adjusted mean values and 95% confidence intervals for post-intervention peak V̇O2 are shown for the ergometer- and walking-based exercise groups. Values were adjusted for age, sex, and baseline peak V̇O2 using analysis of covariance. Distribution of Changes in Peak V̇O2 The distribution of changes in peak V̇O2 from baseline to follow up is shown in Figure 2. Although the median increase in peak V̇O2 tended to be greater in the ergometer-based exercise group, considerable overlap was observed between the 2 groups, indicating substantial interindividual variability in training responses. Figure 2.Changes in peak oxygen uptake (V̇O2) following home-based exercise training. Box-and-whisker plots show the distribution of changes in peak V̇O2 from baseline to follow up in the ergometer- and walking-based exercise groups. The boxes represent the interquartile range, the horizontal line indicates the median, and the whiskers indicate the range. Dots indicate individual outliers. Changes in peak oxygen uptake (V̇O2) following home-based exercise training. Box-and-whisker plots show the distribution of changes in peak V̇O2 from baseline to follow up in the ergometer- and walking-based exercise groups. The boxes represent the interquartile range, the horizontal line indicates the median, and the whiskers indicate the range. Dots indicate individual outliers. Secondary Outcome: Responder Status Based on ≥6% Improvement in Peak V̇O2 Using the responder definition of ≥6% improvement in peak V̇O2, 52 (75%) patients were classified as responders. The responder rate was higher in the ergometer-based exercise group than in the walking-based exercise group, although the difference was not statistically significant. In the multivariable logistic regression model adjusted for age, sex, and baseline peak V̇O2, ergometer-based home exercise was positively associated with responder status; however, this association was attenuated after further adjustment for changes in exercise self-efficacy (Table 3A). Additional models further adjusted for baseline left ventricular ejection fraction and B-type natriuretic peptide levels yielded similar results (Table 3B). Table 3.Logistic Regression Analysis for Responder Status Based on Peak V̇O2 Improvement Univariate analysisMultivariable model 1Multivariable model 2OR95% CIP valueOR95% CIP valueOR95% CIP valueA. Logistic regression analysis for peak V̇O2 ≥6% (vs. <6%; n=69) Home exercises, walking (vs. ergometer)0.3600.117, 1.1110.0760.4010.113, 1.4270.1580.4950.132, 1.8510.296 Age (per 1 year)0.9410.873, 1.0140.1090.9070.832, 0.9890.027*0.9100.834, 0.9920.032* Sex, female (vs. male)0.7740.208, 2.8850.7020.3940.080, 1.9400.2520.4440.082, 2.3910.344 Baseline peak V̇O2 (per 1 mL/min/kg)0.8960.794, 1.0100.0730.8000.676, 0.9470.009*0.7780.653, 0.9280.005* Change in self-efficacy (per 1 point)1.1040.980, 1.2440.102n.c. 1.1320.977, 1.3120.099B. Logistic regression analysis for peak V̇O2 ≥6% (vs. <6%; EF and BNP added; n=69) Home exercises, walking (vs. ergometer)0.3600.117, 1.1110.0760.3620.089, 1.4750.1560.4400.103, 1.8710.266 Age (per 1 year)0.9410.873, 1.0140.1090.9070.830, 0.9920.033*0.9100.831, 0.9950.039* Sex, female (vs. male)0.7740.208, 2.8850.7020.4760.089, 2.5390.3850.5350.090, 3.1610.490 Baseline peak V̇O2 (per 1 mL/min/kg)0.8960.794, 1.0100.0730.8040.672, 0.9620.017*0.7880.653, 0.9500.012* Change in self-efficacy (per 1 point)1.1040.980, 1.2440.102n.c. 1.1270.972, 1.3070.113 Baseline EF (per 1%)0.9790.939, 1.0220.3330.9740.917, 1.0340.3840.9740.914, 1.0390.421 Baseline BNP (per 1 pg/mL)1.0000.996, 1.0030.8380.9990.994, 1.0030.5150.9990.994, 1.0030.586*P<0.05. Chi-squared test for multivariate model coefficients: P=0.007 [model 1]. Chi-squared test for multivariate model coefficients: P=0.005 [model 2]. BNP, B-type natriuretic peptide; CI, confidence interval; EF, ejection fraction; n.c., not calculated; OR, odds ratio. Table 3.Logistic Regression Analysis for Responder Status Based on Peak V̇O2 Improvement Univariate analysisMultivariable model 1Multivariable model 2OR95% CIP valueOR95% CIP valueOR95% CIP valueA. Logistic regression analysis for peak V̇O2 ≥6% (vs. <6%; n=69) Home exercises, walking (vs. ergometer)0.3600.117, 1.1110.0760.4010.113, 1.4270.1580.4950.132, 1.8510.296 Age (per 1 year)0.9410.873, 1.0140.1090.9070.832, 0.9890.027*0.9100.834, 0.9920.032* Sex, female (vs. male)0.7740.208, 2.8850.7020.3940.080, 1.9400.2520.4440.082, 2.3910.344 Baseline peak V̇O2 (per 1 mL/min/kg)0.8960.794, 1.0100.0730.8000.676, 0.9470.009*0.7780.653, 0.9280.005* Change in self-efficacy (per 1 point)1.1040.980, 1.2440.102n.c. 1.1320.977, 1.3120.099B. Logistic regression analysis for peak V̇O2 ≥6% (vs. <6%; EF and BNP added; n=69) Home exercises, walking (vs. ergometer)0.3600.117, 1.1110.0760.3620.089, 1.4750.1560.4400.103, 1.8710.266 Age (per 1 year)0.9410.873, 1.0140.1090.9070.830, 0.9920.033*0.9100.831, 0.9950.039* Sex, female (vs. male)0.7740.208, 2.8850.7020.4760.089, 2.5390.3850.5350.090, 3.1610.490 Baseline peak V̇O2 (per 1 mL/min/kg)0.8960.794, 1.0100.0730.8040.672, 0.9620.017*0.7880.653, 0.9500.012* Change in self-efficacy (per 1 point)1.1040.980, 1.2440.102n.c. 1.1270.972, 1.3070.113 Baseline EF (per 1%)0.9790.939, 1.0220.3330.9740.917, 1.0340.3840.9740.914, 1.0390.421 Baseline BNP (per 1 pg/mL)1.0000.996, 1.0030.8380.9990.994, 1.0030.5150.9990.994, 1.0030.586*P<0.05. Chi-squared test for multivariate model coefficients: P=0.007 [model 1]. Chi-squared test for multivariate model coefficients: P=0.005 [model 2]. BNP, B-type natriuretic peptide; CI, confidence interval; EF, ejection fraction; n.c., not calculated; OR, odds ratio. Logistic Regression Analysis for Responder Status Based on Peak V̇O2 Improvement *P<0.05. Chi-squared test for multivariate model coefficients: P=0.007 [model 1]. Chi-squared test for multivariate model coefficients: P=0.005 [model 2]. BNP, B-type natriuretic peptide; CI, confidence interval; EF, ejection fraction; n.c., not calculated; OR, odds ratio. Sensitivity Analysis With a ≥10% Improvement Threshold The sensitivity analysis, which we conducted using a stricter responder definition (a ≥10% improvement in peak V̇O2), showed a similar trend, with a higher responder rate in the ergometer-based exercise group than in the walking-based exercise group. In the fully adjusted model, changes in exercise self-efficacy were independently associated with responder status, whereas exercise modality was not (Table 4A). Additional models, further adjusted for baseline left ventricular ejection fraction and B-type natriuretic peptide levels, yielded similar results (Table 4B). Table 4.Sensitivity Analysis: Logistic Regression Analysis for Peak V̇O2 ≥10% Improvement Univariate analysisMultivariable model 1Multivariable model 2OR95% CIP valueOR95% CIP valueOR95% CIP valueA. Logistic regression analysis for peak V̇O2 ≥10% (vs. <10%; n=69) Home exercises, walking (vs. ergometer)0.4060.145, 1.1350.0860.4440.135, 1.4650.1830.5760.161, 2.0560.396 Age (per 1 year)0.9320.871, 0.9960.039*0.8900.821, 0.9640.004*0.8860.815, 0.9630.005* Sex, female (vs. male)1.0290.303, 3.4960.9640.5640.126, 2.5360.4560.6220.120, 3.2210.571 Baseline peak V̇O2 (per 1 mL/min/kg)0.8980.802, 1.0050.0600.7850.666, 0.9260.004*0.7560.636, 0.9000.002* Change in self-efficacy (per 1 point)1.1331.011, 1.2700.031*n.c. 1.1771.026, 1.3500.020*B. Logistic regression analysis for peak V̇O2 ≥10% (vs. <10%; EF and BNP added; n=69) Home exercises, walking (vs. ergometer)0.4060.145, 1.1350.0860.4630.126, 1.6970.2450.5880.153, 2.2600.440 Age (per 1 year)0.9320.871, 0.9960.039*0.8890.820, 0.9640.005*0.8860.814, 0.9640.005* Sex, female (vs. male)1.0290.303, 3.4960.9640.6070.130, 2.8280.5250.6420.120, 3.4310.604 Baseline peak V̇O2 (per 1 mL/min/kg)0.8980.802, 1.0050.0600.7810.655, 0.9300.006*0.7540.624, 0.9110.003* Change in self-efficacy (per 1 point)1.1331.011, 1.2700.031*n.c. 1.1741.023, 1.3480.023* Baseline EF (per 1 %)0.9890.954, 1.0250.5330.9930.945, 1.0450.7980.9970.943, 1.0550.925 Baseline BNP (per 1 pg/mL)1.0000.997, 1.0030.8910.9990.995, 1.0030.5950.9990.995, 1.0040.799*P<0.05. Chi-squared test for multivariate model coefficients: P=0.001 [model 1]. Chi-squared test for multivariate model coefficients: P<0.001 [model 2]. Abbreviations as in Table 3. Table 4.Sensitivity Analysis: Logistic Regression Analysis for Peak V̇O2 ≥10% Improvement Univariate analysisMultivariable model 1Multivariable model 2OR95% CIP valueOR95% CIP valueOR95% CIP valueA. Logistic regression analysis for peak V̇O2 ≥10% (vs. <10%; n=69) Home exercises, walking (vs. ergometer)0.4060.145, 1.1350.0860.4440.135, 1.4650.1830.5760.161, 2.0560.396 Age (per 1 year)0.9320.871, 0.9960.039*0.8900.821, 0.9640.004*0.8860.815, 0.9630.005* Sex, female (vs. male)1.0290.303, 3.4960.9640.5640.126, 2.5360.4560.6220.120, 3.2210.571 Baseline peak V̇O2 (per 1 mL/min/kg)0.8980.802, 1.0050.0600.7850.666, 0.9260.004*0.7560.636, 0.9000.002* Change in self-efficacy (per 1 point)1.1331.011, 1.2700.031*n.c. 1.1771.026, 1.3500.020*B. Logistic regression analysis for peak V̇O2 ≥10% (vs. <10%; EF and BNP added; n=69) Home exercises, walking (vs. ergometer)0.4060.145, 1.1350.0860.4630.126, 1.6970.2450.5880.153, 2.2600.440 Age (per 1 year)0.9320.871, 0.9960.039*0.8890.820, 0.9640.005*0.8860.814, 0.9640.005* Sex, female (vs. male)1.0290.303, 3.4960.9640.6070.130, 2.8280.5250.6420.120, 3.4310.604 Baseline peak V̇O2 (per 1 mL/min/kg)0.8980.802, 1.0050.0600.7810.655, 0.9300.006*0.7540.624, 0.9110.003* Change in self-efficacy (per 1 point)1.1331.011, 1.2700.031*n.c. 1.1741.023, 1.3480.023* Baseline EF (per 1 %)0.9890.954, 1.0250.5330.9930.945, 1.0450.7980.9970.943, 1.0550.925 Baseline BNP (per 1 pg/mL)1.0000.997, 1.0030.8910.9990.995, 1.0030.5950.9990.995, 1.0040.799*P<0.05. Chi-squared test for multivariate model coefficients: P=0.001 [model 1]. Chi-squared test for multivariate model coefficients: P<0.001 [model 2]. Abbreviations as in Table 3. Sensitivity Analysis: Logistic Regression Analysis for Peak V̇O2 ≥10% Improvement *P<0.05. Chi-squared test for multivariate model coefficients: P=0.001 [model 1]. Chi-squared test for multivariate model coefficients: P<0.001 [model 2]. Abbreviations as in Table 3. Exploratory Analysis of Behavioral Factors Across both responder-based and continuous outcome analyses, adjustment for changes in exercise self-efficacy consistently attenuated the association between home-based exercise modalities and improvements in peak V̇O2, suggesting that behavioral engagement may partially be associated with the relationship between exercise modality and physiological improvements. Discussion In this single-center observational study of outpatient CR, ergometer-based home exercises were associated with modestly greater improvement in exercise capacity than walking-based exercises, particularly when the peak V̇O2 was analyzed as a continuous outcome. Improvement in peak V̇O2 is a well-established indicator of the functional capacity and prognosis of patients undergoing CR. The present findings extend prior evidence7,9 by highlighting the potential relevance of home-based exercise modalities during the maintenance phase of outpatient CR. Role of Modalities in Exercise Capacity Improvement The present findings suggest that home-based exercises may influence the translation of prescribed exercises into physiological adaptations during the CR maintenance phase. Ergometer-based exercise allows for precise workload prescription and objective feedback, both of which are emphasized as core components of contemporary CR programs.6 In contrast, walking-based exercises, although practical and accessible, may be more susceptible to variability in exercise intensity owing to environmental factors, such as terrain and weather conditions, and because they are self-paced. Such heterogeneity in training stimuli was highlighted in a recent systematic review as an important limitation of home-based exercise interventions.9 Importantly, the analysis of peak V̇O2 as a continuous outcome demonstrated that ergometer-based exercise was associated with higher post-intervention exercise capacity across the cohort rather than only with strong responders. This observation suggests that the exercise modality may influence the overall distribution of training effects, resulting in modest yet consistent improvements across a broad patient population. From a clinical perspective, such a shift in distribution may be particularly relevant in outpatient settings, where sustained and reproducible improvements are desirable to support long-term functional maintenance and secondary disease prevention. An important consideration is whether the observed differences reflect the exercise modality itself or differences in actual training volume. In the present study, the prescribed exercise frequency, duration, and intensity were standardized across both groups (twice-weekly, 30-min sessions at the AT level). However, because objective exercise logs were not collected, actual adherence and achieved training volume could not be directly compared between the groups. Ergometer-based exercise can be performed indoors regardless of weather conditions and may therefore facilitate more consistent exercise adherence than walking, which is subject to environmental variability. Furthermore, the method of intensity monitoring differed between the groups; the ergometer group used objective wattage targets, whereas the walking group relied on heart rate and perceived exertion, which may have introduced greater variability in the exercise intensity achieved. In addition, exercise frequency and exercise capacity during home-based CR have been reported to differ by sex and age,26 suggesting that patient-level factors may also contribute to the variability in training responses. Consequently, the observed association may partly reflect differences in actual training volume or achieved intensity rather than modality-specific effects alone. Future studies incorporating objective monitoring of home-based exercise are needed to disentangle these factors. Role of Self-Efficacy in Exercise Capacity Improvement Although changes in exercise self-efficacy were not independently associated with responder status in the primary multivariable analyses, adjustment for self-efficacy consistently attenuated the association between exercise modality and improvements in peak V̇O2. Self-efficacy is a well-established behavioral construct that influences exercise adherence and maintenance rather than directly causing improvement.14 In the present findings, self-efficacy may represent a behavioral factor associated with both exercise modality and improvements in exercise capacity.27 Importantly, self-efficacy should be considered a dynamic construct that may evolve in response to patients’ experiences during CR rather than a fixed baseline characteristic. Exercise modalities that provide immediate and objective feedback, such as ergometer-based exercises, may reinforce patients’ confidence in their ability to exercise at a prescribed intensity, which may be related to sustained engagement during unsupervised periods. This interpretation is consistent with behavioral theories that emphasize the role of mastery experiences in the shaping of self-efficacy and exercise behavior.11,20 However, the possibility of reverse causality must also be acknowledged. Patients with inherently higher health consciousness or motivation may have preferentially chosen ergometer-based exercise, which allows for more structured and quantifiable training, and these patients may have been more likely to improve their exercise capacity regardless of the modality used. Although baseline self-efficacy scores did not differ significantly between the groups (12.8±3.5 vs. 13.9±4.4; P=0.28), this does not fully exclude the possibility that unmeasured motivational or behavioral traits influenced both the choice of modality and the subsequent improvement in exercise capacity. The retrospective design of this study did not allow us to establish a temporal sequence between changes in self-efficacy and improvements in peak V̇O2; therefore, prospective studies with repeated assessments of self-efficacy are needed to clarify the direction of this relationship. Clinical Implications The present findings may have important implications for outpatient and home-based CR programs. Although walking-based exercises remain accessible and widely used, the choice of exercise modality may influence the consistency and effectiveness of home-based training. Ergometer-based exercises may facilitate more reproducible workload control and feedback, aligning with the core components of contemporary CR programs. Moreover, the attenuation of the association between exercise modality and improvement in peak V̇O2 after adjustment for self-efficacy suggests that behavioral engagement may play a role in the relationship between prescribed exercises and physiological adaptations. These findings suggest that optimal CR strategies should consider not only the physiological characteristics of exercise modalities but also their capacity to support behavioral engagement during unsupervised exercise. From a practical standpoint, the clinical applicability of these findings may extend beyond the choice of specific equipment. Because only a subset of patients can afford or accommodate a home ergometer, the key clinical message may lie not in recommending a specific device but in incorporating the principles that characterize ergometer-based exercise, namely, quantifiable workload management, structured intensity monitoring, and self-efficacy-enhancing feedback, into the prescription of any home-based exercise, including walking. For example, the use of wearable heart rate monitors and structured exercise diaries may allow walking-based exercise programs to achieve levels of intensity control and behavioral reinforcement similar to those associated with ergometer-based training. A recent multicenter randomized controlled trial demonstrated that community-based walking programs could produce sustained improvements in physical activity, well-being, and social capital among older patients with cardiovascular disease,28 supporting the potential of walking-based interventions when appropriately structured. Recent studies in Japan have demonstrated the feasibility and clinical relevance of structured, home-based, multidisciplinary CR programs in real-world settings.29,30 Although specific exercise modalities were not directly compared in these studies, they provide important contextual support and reinforce the external validity of the present findings in routine outpatient practice. Baseline differences in the left ventricular ejection fraction and B-type natriuretic peptide levels between the groups may reflect differences in disease severity or the clinical selection of home-based exercise modalities in real-world outpatient practice. Additional adjustments for baseline left ventricular ejection fraction and B-type natriuretic peptide levels did not materially alter the association between exercise modality and post-intervention peak V̇O2. Importantly, adjustments for baseline exercise capacity and sensitivity analyses that accounted for these variables yielded consistent results, suggesting that these differences are unlikely to explain the observed associations fully. These findings should therefore be interpreted primarily as hypothesis-generating observations regarding the translation of prescribed exercise into reproducible training stimuli during unsupervised phases of cardiac rehabilitation. Study Limitations This study had several limitations. First, its retrospective, single-center design limited the generalizability of the results and precluded causal inference. Because the choice of exercise modality was determined by clinician judgment and patient preference, confounding by indication is inherent in the study design. Patients who selected ergometer-based exercise may have differed from those who chose walking in terms of socioeconomic status, health literacy, residential environment, and access to equipment. Patients with higher socioeconomic status may be more likely to purchase a home ergometer and may also exhibit greater adherence to health behaviors in general, potentially confounding the observed association. These unmeasured confounders could not be addressed in the present analyses. Second, the relatively small sample size (n=69) limited the statistical power of the study. Based on the observed effect size from the primary analyses of covariance model (partial η2=0.072), a post hoc power analysis indicated that the analyses achieved approximately 63% power at a 2-sided significance level of 0.05. Given the limitations of post hoc power analyses in observational studies, we primarily interpreted the findings based on effect sizes and 95% confidence intervals. With only 69 patients and multiple covariates in the regression models, the stability of the estimated coefficients may be limited, and the results should be interpreted with appropriate caution. Third, although we adjusted our models for key clinical covariates, residual confounding could not be excluded. We did not have data on the New York Heart Association functional class or baseline physical activity levels. However, the Short Physical Performance Battery scores (11.9±0.4 vs. 11.9±0.4; P=0.89) and 6-min walk distances (516±78 vs. 508±86 m; P=0.70) were comparable between the groups, suggesting that baseline physical function and frailty status did not differ substantially. Baseline peak V̇O2 values were also similar (20.2±4.4 vs. 20.5±5.2 mL/min/kg; P=0.76), further supporting the comparability of functional capacity at baseline. Fourth, the exercise intensity and training volume during home-based exercises were not directly quantified; differences in these factors might have influenced the observed association between exercise modality and improvement in peak V̇O2. Although the prescribed exercise frequency (twice weekly), duration (30 min), and intensity (AT level) were standardized, actual adherence was confirmed only verbally during supervised sessions, and objective exercise logs were not collected. The method of intensity monitoring also differed between the groups: the ergometer group used objective wattage targets, whereas the walking group relied on heart rate and perceived exertion, which may introduce greater day-to-day variability. Therefore, we cannot exclude the possibility that the ergometer group achieved a higher actual training volume or more consistent exercise intensity, which may have contributed to the observed differences. Nevertheless, standardized CPX protocols were used, and covariates were predefined based on clinical relevance, thereby minimizing the risk of overfitting and enhancing the robustness of the findings. Last, because changes in self-efficacy and improvements in peak V̇O2 were assessed over the same time period, the temporal sequence between these variables cannot be determined. It remains unclear whether changes in self-efficacy preceded or followed improvements in exercise capacity. Furthermore, formal mediation analysis (e.g., estimation of indirect effects) was not performed. Therefore, the observed attenuation of the modality–peak V̇O2 association after adjustment for self-efficacy should be interpreted as suggesting an association, rather than as evidence of a mediating role of self-efficacy. Conclusions In an outpatient CR setting, ergometer-based home exercises were associated with modestly greater improvement in exercise capacity, as assessed by peak V̇O2, particularly when analyzed as a continuous outcome. Although self-efficacy was not an independent predictor, its association with this relationship suggests the potential importance of behavioral engagement in the relationship between prescribed exercises and physiological improvements. These findings should be interpreted in the context of the retrospective study design and unmeasured confounders, including potential differences in actual training volume and patient characteristics. Rather than advocating for a specific exercise modality, the present results suggest that the principles underlying ergometer-based exercise, structured intensity prescription, objective performance feedback, and self-efficacy-enhancing approaches, may be applicable to a broader range of home-based exercise programs, including walking-based interventions. Future studies conducted at multiple centers with larger cohorts and standardized home-based ergometer devices are warranted to confirm the observed associations and clarify their causal relationships. Disclosures The authors declare that there are no conflicts of interest. IRB Information This study was approved by the Ethics Committee of Otsu Red Cross Hospital, Shiga, Japan (reference no. 962).