Journal Logo


Baseline and Exercise Predictors of V˙O2peak in Systolic Heart Failure Patients: Results from SMARTEX-HF


Author Information
Medicine & Science in Sports & Exercise: April 2020 - Volume 52 - Issue 4 - p 810-819
doi: 10.1249/MSS.0000000000002193


Peak oxygen uptake (V˙O2peak) is a strong prognostic factor in heart failure with reduced ejection fraction (HFrEF) (1). Endurance exercise training has a positive effect on V˙O2peak (2,3), left ventricular function (4), quality of life (5), mortality, and morbidity (3,6,7). Studies evaluating dose and intensity of exercise training show variability in exercise responses from moderate to large (2–4,8,9). Absence of improvement in V˙O2peak after a systematic exercise program was a strong and independent predictor of adverse cardiac events that were not associated with traditional risk factors (10), whereas a modest increase in 3-month V˙O2peak was associated with less all-cause mortality and fewer hospitalizations in the large HF-ACTION trial (3,11).

In general, multicenter exercise studies produce smaller outcome effects than do single-center studies (2,3,8,12). In the HF-ACTION multicenter trial, adherence to target training volume was less than optimal, with only 40% of the patients at or above target exercise minutes per week at 3-month follow-up (3,11). In the SMARTEX Heart Failure Study multicenter trial (SMARTEX-HF), adherence to the number of exercise sessions was excellent (96%) during the supervised training period in both the high-intensity interval training (HIIT) group and in the moderate continuous training (MCT) group, whereas self-report of exercise training in the recommendation of regular exercise (RRE) group gave less data precision. Despite excellent adherence to exercise sessions, moderate exercise response, and no differences in comparative effectiveness were observed between HIIT and MCT for improvement in V˙O2peak (13). Hence, it is currently unclear how the magnitude of improvement in V˙O2peak with exercise training is modified by patient characteristics, adherence, disease severity, comorbidity, exercise follow-up, or simply motivation to exercise.

To investigate baseline and exercise training predictors of ΔV˙O2peak from baseline to 12-wk follow-up in HFrEF patients, we performed a post hoc analysis of data from SMARTEX-HF to address if ΔV˙O2peak was associated with the following: 1) one or more of the baseline characteristics and 2) exercise training characteristics (e.g., workload and heart rate during training sessions), exercise testing characteristics, or clinical characteristics known to affect physical performance (e.g., heart failure pathogenesis, age, and smoking). We considered the study too small to investigate whether baseline variables have different effects depending on the three specific training interventions.


Details of the SMARTEX-HF study protocol and the intervention results on primary and secondary end points have been published previously (14,15).


In nine European study centers, 261 clinically stable HFrEF patients were randomized from outpatient heart failure clinics, hospital registries, cardiac rehabilitation referrals, and public announcements. After withdrawals and appropriate exclusions, 231 started training, and 215 patients completed 12 wk of exercise and clinical baseline and follow-up assessments. Patient flow in the study has been detailed elsewhere (15). At baseline, all subjects had stable, symptomatic HFrEF with left ventricular ejection fraction (LVEF) ≤35%. All subjects were in New York Heart Association (NYHA) functional classes II–III and were on optimal medical treatment. Further details of inclusion and exclusion criteria have been described in the rationale and design article (14).

National ethics committees for medical research approved the study in all countries. All patients gave written informed consent. The study was registered in the clinical trial database before the start (NCT00917046) and conducted in conformity with the policy statement for the use of human subjects of the Declaration of Helsinki and Medicine & Science in Sports & Exercise.


Patients were randomized 1:1:1 to a 12-wk program of HIIT, MCT, or a control group given a recommendation of mainly home-based regular exercise (RRE), stratified by study center, sex, and disease pathogenesis (ischemic vs nonischemic heart failure). Randomization was performed by a web-based randomization system developed and administered by the Unit of Applied Clinical Research, Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology, Trondheim, Norway. Patients in the HIIT and MCT groups performed three weekly sessions of supervised exercise training. Briefly, the HIIT group performed a training program with 4 × 4 min of interval training aiming for a target heart rate of 90%–95% of peak heart rate (HRpeak; 38-min workout including warm-up, active breaks, and cool-down), and the MCT group performed a program with 47 min of moderate continuous training aimed at 60%–70% of HRpeak, designed to be isocaloric. RRE patients were advised to exercise at home according to current exercise guidelines, that is, 30 min 5 d·wk−1 (16), and attended a session of moderate-intensity training every 3 wk (50%–70% HRpeak) (14). The exercise training was performed either on a stationary bicycle ergometer or a treadmill (2,14).

Clinical measurements

Cardiopulmonary exercise testing (CPET), medical history, anthropometrics, physical examination, fasting blood sampling, quality of life questionnaires, and echocardiography were performed at baseline and after 12 wk of training (14,15). V˙O2peak was measured by CPET performed either on a treadmill or a bicycle ergometer, corresponding to the preferred training mode at each study center, and was similar at baseline and 12 wk for each participant. An incremental protocol with 10- or 20-W increase in workload approximately every minute was used. V˙O2peak was measured using standard equipment for indirect calorimetry. The mean of the three highest 10-s consecutive measurements was used to calculate V˙O2peak. HRpeak and other related values are reported from the time point when this value was reached. Echocardiography data were acquired according to standard operation procedures of the study (15).

Statistical analysis

In the first post hoc analysis, data were analyzed using logistic regression comparing the highest versus the lowest tertile of ΔV˙O2peak (high tertile >1.5 mL·kg−1·min−1 and low tertile ≤−1.5 mL·kg−1·min−1). In the second analysis, we used multivariate linear regression with ΔV˙O2peak as a continuous dependent variable. Data are given as frequencies with percentage in parenthesis, or median with 95% confidence interval of the median in parenthesis, if otherwise is not stated. P values <0.05 were considered significant.


To investigate whether the overall moderate changes after exercise training in the SMARTEX-HF study was due to demographics or other characteristics at baseline, we compared the highest versus lowest tertile of ΔV˙O2peak. The middle tertile was not included in the analysis to increase the contrast between groups, thereby better permitting differences to be identified. The analysis was done for the patient population as a whole, without considering treatment group (i.e., RRE, MCT, or HIIT). V˙O2peak at baseline and treatment group were included as adjustment variables in the analysis.

Additional variables were selected, applying no additional a priori hypothesis for an unbiased selection of predictors and to avoid overfitting the analysis model. To this end, a predefined selection of baseline variables (see discussion hereinafter) was prescreened using random forest analysis with bootstrapping (n = 2000), using the “party” package in the R statistical environment (version 3.0.2, R Foundation,

The baseline variables screened included the following: study center, heart failure pathogenesis (ischemic vs nonischemic), height, sex, age, LVEF, NYHA class, V˙O2peak, sinus rhythm, systolic and diastolic blood pressures, body mass index, duration of HFrEF, cardiac device therapy, chronic obstructive pulmonary disease, smoking (never vs ever smoker), concentrations of N-terminal brain natriuretic peptide, high-sensitive C-reactive protein, and thyroxin. The following baseline variables were identified as giving a strong signal of association in the random forest model: NYHA class, LVEF, age, smoking, and treatment group (MCT, RRE or HIIT). In addition, creatinine clearance and left ventricular end diastolic diameter (LVEDD) were included in an additional sensitivity analyses.

The final main end point analysis was logistic regression modeling using the selected baseline variables indicated previously, as well as baseline V˙O2peak. The SE values of the final logistic regression model were bootstrapped (n = 1000) to get less biased results. Linearity of logits was tested using restricted cubic splines. Because a sensitivity analysis to examine whether omittance of the middle delta V˙O2peak tertile influenced the results, a linear regression model including all patients was also fitted, using ΔV˙O2peak as a dependent variable and the same predictors as in the logistic regression model.


We then investigated whether exercise test– and training-related variables were associated with the variability in V˙O2peak, adjusting for relevant baseline variables. ΔV˙O2peak was analyzed as a continuous variable using multivariate linear regression. Training and exercise test values in the model each represent measures of test and training quality, which are expected to be associated with ΔV˙O2peak. For instance, significant improvements in both change in exercise training work load (ΔWatt) and ΔV˙O2peak are typically seen after HIIT (2,17). Only data from MCT and HIIT patients were included in this analysis, as training data were recorded to a limited degree in the mainly home-based RRE group.

ΔV˙O2peak was analyzed as a continuous variable using a multivariate linear regression model including the following explanatory/adjustment variables selected per protocol: V˙O2peak at baseline (CPET1), difference in peak heart rate between baseline and follow-up test at 12 wk (ΔHRpeak), peak respiratory ratio at CPET2, change in ΔWatt after 12 wk of exercise training, and training group (MCT or HIIT). Based on clinical knowledge on suspected influence, heart failure pathogenesis, age, and smoking were also included in the model for adjustment. Robust SE values were used, and model fit was evaluated using residual plots. The analysis was performed in 106 patients (data for ΔWatt missing in n = 20 (31%) in MCT and n = 15 (19%) in HIIT).

As a supplementary secondary analysis, we removed ΔWatt from the model to avoid case loss due to missing exercise work load data. This analysis was performed in 134 patients (HRpeak missing in 3 MCT patients and 4 HIIT patients, i.e., 5% missing in both groups).


Changes in V˙O2peak

One patient in the MCT group had missing values for the baseline CPET and was excluded from the analysis, leaving 214 patients for investigation. Characteristics of these patients are shown in Table 1 and in Supplemental Table 1 (Supplemental Digital Content 1, additional patient characteristics,

Baseline characteristics.

There was large variability in ΔV˙O2peak after the 12-wk intervention (from −8.50 to +11.30 mL·kg−1·min−1). The distribution of ΔV˙O2peak in each intervention group is illustrated in Figure 1.

Distribution of ΔV˙O2peak after 12 wk of exercise training in the HIIT, MCT, and RRE groups. The dotted line marks zero change in V˙O2peak, with positive and negative changes in V˙O2peak to the right and left sides of zero.

The percentages of patients in the high versus the low tertile were 39% versus 31% in the HIIT group, 40% versus 25% in the MCT group, and 19% versus 49% in the RRE group. The numbers of responders in the two training groups were significantly higher than that in the RRE group (P = 0.003). The median change in V˙O2peak in each of the tertiles is displayed in Figure 2.

Median ΔV˙O2peak (mL·kg−1·min−1) after 12 wk of exercise training in the three tertiles of high (H), medium (M), and low (L) V˙O2peak responders (all patients). The medium tertile: −1.5 to 1.5 mL·kg·min−1. Open bars: range. Gray shading: 95% confidence interval of the medians.

Associations of ΔV˙O2peak with baseline values

In the final logistic regression model, NYHA class, age, LVEF, and treatment group were significantly associated with ΔV˙O2peak. V˙O2peak at baseline (P = 0.34 or ever being a smoker (P = 0.09) was not associated with ΔV˙O2peak. Table 2 shows the multivariate model (as well as univariate associations, even if they were not used for explanatory variable selection).

Logistic regression model for associations of delta V˙O2peak with baseline values. a

The analysis indicated 7.1 higher odds for an exercise response (highest ΔV˙O2peak tertile) if classified in NYHA II vs NYHA III at baseline. In the SMARTEX-HF data set (i.e., without bootstrapping), 58 of 70 (82.9%) of the patients with a positive change in V˙O2peak (above the tertile cutoff) were in NYHA class II (mean ± SD baseline V˙O2peak values were 18.7 ± 4.8 mL·kg−1·min−1 for NYHA II and 15.0 ± 3.8 mL·kg−1·min−1 for NYHA III). Compared with control (RRE), the proportion of responders (i.e., highest ΔV˙O2peak tertile) was higher in the two exercise groups (HIIT and MCT), with no statistically significant difference between HIIT and MCT (P = 0.71).

The sensitivity analysis using ΔV˙O2peak as a continuous dependent variable and including all patients Table 3 confirmed the direction and significance of the associations from the main model for NYHA class (P = 0.002), age (P = 0.001), and training group (HIIT or MCT vs RRE: P < 0.01, HIIT vs MCT: P = 0.93), but not for LVEF (P = 0.10). Sensitivity analyses including estimated creatinine clearance (P = 0.84) or LVEDD (P = 0.17) showed that these variables were not significant.

Sensitivity analysis: linear regression model for associations of delta V˙O2peak with baseline values.

Associations of ΔV˙O2peak with test- or training-related variables (HIIT and MCT groups)

In a multivariate linear regression model with ΔV˙O2peak as a continuous outcome variable, the significant variables were as follows: ΔHRpeak between baseline and 12-wk test (P = 0.007), change in training workload between baseline and follow-up (P = 0.003), age (negative coefficient, P < 0.001), and ever smoker (P = 0.001). R2 for this model was 0.34. The following variables were not significant: HIIT versus MCT (P = 0.47), peak respiratory quotient (RQ) at 12-wk test (P = 0.53), heart failure pathogenesis (P = 0.92), and V˙O2peak at baseline (P = 0.55). The model is given in Supplemental Table 2 (Supplemental Digital Content 2, linear regression model for associations of delta V˙O2peak with test- or training-related variables: primary model, and illustrated in Figure 3A, showing results for an increase or decrease in HRpeak of 20 bpm.

Prediction of ΔV˙O2peak differences after 12 wk of supervised exercise training (data from HIIT and MCT) vs the following. A, Effect of change in exercise training work load in patients with either a positive or a negative ΔHRpeak from CPET1 to CPET2. The multivariable linear regression model also includes delta workload, age, ever smoking, exercise training group, peak RQ at 12 wk, heart failure pathogenesis, and V˙O2peak at baseline. B, Effect of ΔHRpeak from CPET1 to CPET2 in ever vs never smokers. The multivariable linear regression model also includes age, delta HRpeak from CPET1 to CPET2, exercise training group, peak RQ at 12 wk, heart failure pathogenesis, and V˙O2peak at baseline. C, Effect of change in ΔHRpeak from CPET1 to CPET2 in HIIT vs MCT, same model as panel B. Data are means with 95% confidence intervals.

In the secondary model given in Supplemental Table 3 (Supplemental Digital Content 3, linear regression model for associations of delta V˙O2peak with test- or training-related variables: secondary model,, excluding ΔW (due to lower n for this variable), 29% of the variation in ΔV˙O2peak was explained, and the significant variables were as follows: ΔHRpeak from baseline to 12-wk test (P < 0.001), age (negative coefficient, P = 0.002), and ever smoker (P = 0.02; Fig. 3B). There were still no differences between HIIT and MCT (P = 0.42; Fig. 3C). The initial model explained more of the variance in the V˙O2peak response than the second model (34% vs 29%). When including the same patients in the two models (n = 106), the explained variations were 34% and 29% for the initial and secondary models, respectively.

Both logistic regression and linear regression analyses excluding the RRE group gave the same results as analyses reported in the article (unpublished data).


Associations of ΔV˙O2peak with baseline values

The main finding of this study was that the baseline characteristics NYHA class, LVEF, age, and treatment group were associated with ΔV˙O2peak after 12 wk of exercise training. Older age, poorer left ventricular function, and higher NYHA class were associated with a less favorable 12-wk change in V˙O2peak. As illustrated in Figure 2, a large part of the study participants in all three groups had neutral or negative changes in V˙O2peak over the 12-wk intervention. This does not necessarily mean that they were negative responders to exercise. It could also be due to a negative fitness trajectory caused by advancing severity of heart failure. V˙O2peak and NYHA class are closely related, with higher V˙O2peak (18,19) and lower number of long-term cardiac events (10) in NYHA II versus NYHA III–IV HFrEF patients (18). We confirmed that baseline NYHA class and ΔV˙O2peak are associated as well, with the ΔV˙O2peak response independent of baseline V˙O2peak.

Each 1% higher baseline LVEF was associated with 10% greater odds of being in the highest delta V˙O2peak tertile, independent of exercise intensity or exercise group. The overall group response in LVEF at 12 wk was moderate (15). Our logistic regression analysis shows that baseline LVEF might indicate the left ventricular exercise recovery potential in HFrEF patients. To the best of our knowledge, the baseline LVEF–exercise response association adds new knowledge about individual exercise responses, with improved exercise recovery prognosis in HFrEF patients with higher baseline contractile function.

In HFrEF, older age is associated with lower V˙O2peak (18,20), more severe symptoms, and worse prognosis compared with younger patients (20). Our study confirms an age-dependent effect in ΔV˙O2peak as well, with higher odds for increasing V˙O2peak in the youngest HFrEF patients (median age, 56 and 65 yr in high and low V˙O2peak tertiles, respectively). In comparison, some have reported a larger training response in HFrEF patients older than 70 yr (2), whereas others report an age-independent response in HFrEF patients younger than 65 yr and older than 65 yr (5,21,22). The differences between studies could be due to patient selection; physiological aging, which reduces HRpeak and V˙O2peak (20); clustering of comorbidities, medication; age-dependent deteriorating heart failure that may affect the ability or motivation to exercise (11); different training quality; or continuous versus categorical statistical analysis. The age-dependent exercise response was confirmed in the secondary analyses as well. HFrEF duration was classified above and below 12 months in our study, making interaction analysis between age and years with symptomatic HFrEF impossible. In addition, the study sample was too small to study this association; however, heart failure duration was far from significant in the main logistic regression model.

Associations of ΔV˙O2peak with test- or training-related characteristics (HIIT and MCT groups)

According to the multivariable linear regression analysis, a total of 34% of the variability in ΔV˙O2peak was explained by the test and training quality variables ΔHRpeak (CPET2 minus CPET1) and ΔWatt (exercise training workload from exercise weeks 1 to 12), in addition to the baseline variables age and ever being a smoker.

Challenges for long-term adherence to exercise training in patients with chronic symptomatic heart failure include dyspnea, medication, muscle, and physiological deconditioning (3). Peak heart rate rarely changes in apparently healthy individuals, and ΔHRpeak seldom changes from baseline to follow-up testing in HIIT studies (2,23,24). In HFrEF patients, both no change and increasing HRpeak are reported after exercise training (2,25–27). A positive ΔHRpeak and ΔV˙O2peak could indicate a transition from peripheral (muscle) to central (heart) limitations to maximal exercise performance throughout the training period (9,28). A negative ΔHRpeak and ΔV˙O2peak may indicate deteriorating heart failure and decreased exercise tolerance (11), or could indicate some variability in test quality in the study. Maximal RQ values indicated similar levels of effort during testing at all time points (13). Because there were only minor changes in medication throughout the training intervention, change in medication does not explain ΔHRpeak from CPET1 to CPET2.

In addition to the moderate increase in exercise training workload (Δ workloads were 21 and 15 W in HIIT and MCT, respectively), the lack of difference in intensities (mean training intensities in HIIT and MCT were 88% and 80%, respectively) between groups is most likely also responsible for the V˙O2peak response (15).

In CVD patients, superior exercise response was found in the higher part of the HIIT workload zone (29). In comparison to Wisløff et al. (2) (Δ workload HIIT = 95 W) and Iellamo et al. (17) (Δ workload HIIT = 70 W), the increase in exercise training workload and the ability to maintain exercise intensity within the target range were moderate in the SMARTEX-HF study (9). Maintaining target exercise intensity is challenging (30), and the limited increase in exercise training workload may be due to physiological, pathological, psychological factors, or patient and/or coaching motivation (9). Heart failure deterioration is associated with a negative exercise response (11,31) and may explain part of the modest improvement in V˙O2peak and LVEDD in the SMARTEX-HF study (14,15). Similarly, others have reported a moderate exercise outcome even in coronary patients, with a neutral outcome of HIIT versus MCT in a large multicenter study (32), whereas combining endurance and strength training was not associated with improved cardiac function (4). A subgroup of patients with advanced chronic heart failure improved exercise capacity and reversed left ventricular remodeling after daily, long-term moderate exercise training (6 and 12 months) (33). Because patients with the poorest left ventricular function responded the least to exercise training in our study, further investigation of whether daily exercise and longer duration of the intervention is necessary to gain a positive exercise response, or if this may lead to deterioration of congestive heart failure. With both positive and negative exercise responders in our study, tailor-made programs and follow-up may be highly warranted in deconditioned congestive heart failure patients. The findings in the primary statistical model suggest that both physiological and pathological factors may limit the ability to exercise at moderate and high intensity, and we acknowledge that our model leaves 66% of the variability in the exercise response unexplained. Because the change in V˙O2peak is influenced by several central and peripheral factors (7,26,27,34) that were not measured in the present study, we are unable to conclude which of them are the most important, except to confirm the importance of chronotropic incompetency. It may be argued that inclusion of nonbaseline variables precludes prediction of the exercise response, but this was not the focus of the secondary analyses. Because we have no data on exercise motivation, this factor could also not be discussed.

Study strengths and limitations

Study strengths include the explorative statistical design using random forest–based analysis to select among a substantial number of potential explanatory factors without overfitting the model, close supervision of exercise training, and thorough documentation of clinical and physiological patient data. Patient adherence to exercise training sessions was excellent. In addition, the multicenter study probably reflects a wider and more representative patient selection compared with single-center studies. The patients included in the present study represented approximately 10% of the heart failure population screened for inclusion. We believe that the study participants are representative for stable HFrEF with LVEF ≤35% under optimal medical care. However, a majority of the screened patients had LVEF greater than 35%, indicating less representativeness of the overall HFrEF population.

It is a limitation that exercise-related data on intensity and duration could not be studied in the RRE group because of their per-protocol unsupervised and unrecorded home-based exercise. Furthermore, we did not assess training motivation and thus could not tell whether there were differences between the intervention groups. Of note, the confidence intervals for the exercise group effects were wide and the precision of the odds ratio should be interpreted with caution.


Exercise training response (ΔV˙O2peak) correlated negatively with age, LVEF, and NYHA class. The ability to increase workload during the training period and a positive ΔHRpeak between baseline and 12-wk test were associated with a positive outcome.


Exercise training is an important and recommended treatment for heart failure, and this study indicates that individualized approaches may be warranted, as different patients experience exercise tolerance and “exercise intolerance” with a limited or negative response to exercise training. Our analyses suggest that age, LVEF, NYHA classification, and the ability to improve V˙O2peak might be considered when advising exercise training and evaluating exercise response in HFrEF, as data point to a gradient toward a poor exercise response in the oldest and most symptomatic HFrEF patients. An exercise response evaluation by exercise testing might indicate if exercise is an individual treatment of choice or not. Furthermore, it is important to focus on a systematic increase in exercise workload and maintaining exercise target exercise intensity, as individual patients have different ability and/or motivation to increase exercise workload during a training period.

We greatly acknowledge the time and the effort of the participating patients and the entire study staff. Jennifer Adam, Elena Bonanomi, Silvia Colombo, Christian Have Dall, Ingrid Granøien, Kjersti Gustad, Anne Haugland, Julie Kjønnerød, Marit Kristiansen, Jorunn Nilsen, Maren Redlich, Anna Schlumberger, and Kurt Wuyts performed exercise testing and training; Rigmor Bøen, Marianne Frederiksen, Eli Granviken, Loredana Jakobs, Adnan Kastrati, Nadine Possemiers, Hanne Rasmusen, Liv Rasmussen, and Johannes Scherr performed patient screening, inclusion, and clinical assessments; Volker Adams, Ann-Elise Antonsen, Wim Bories, Nadine Possemiers, Malou Gloesner, Vicky Hoymans, and Hielko Miljoen collected data; Hanna Ellingsen and Maria Henningsen performed data monitoring; Lars Køber, Christian Torp-Pedersen, John Kjekshus, Rainer Hambrecht, and Stephan Gielen monitored safety. Together these contributors and the authors comprise the SMARTEX Heart Failure Study Group.

This work was supported by St. Olavs Hospital; Faculty of Medicine and Health Sciences, NTNU – Norwegian University of Science and Technology; Norwegian Health Association; Danish Research Council; Central Norwegian Health Authorities/NTNU; Western Norway Health Authorities; Simon Fougner Hartmanns Familiefond; Else-Kröner-Fresenius-Stiftung; and Société Luxembourgeoise pour la recherche sur les maladies cardio-vasculaires.

M. H. reports grants from the Else-Kröner-Fresenius Foundation for the present work and is on the advisory board of Novartis, Sanofi-Aventis, and MSD, outside the present study. A. L. reports grants and personal fees from Medtronic and from Claret Medical, and personal fees from Edwards, SJM, Bard, and Symetis, all outside the present study. The results of this study do not constitute endorsement by the American College of Sports Medicine. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation.


1. Corra U, Giordano A, Mezzani A, et al. Cardiopulmonary exercise testing and prognosis in heart failure due to systolic left ventricular dysfunction: a validation study of the European Society of Cardiology Guidelines and Recommendations (2008) and further developments. Eur J Prev Cardiol. 2012;19(1):32–40.
2. Wisloff U, Stoylen A, Loennechen JP, et al. Superior cardiovascular effect of aerobic interval training versus moderate continuous training in heart failure patients: a randomized study. Circulation. 2007;115(24):3086–94.
3. O’Connor CM, Whellan DJ, Lee KL, et al. Efficacy and safety of exercise training in patients with chronic heart failure: HF-ACTION randomized controlled trial. JAMA. 2009;301(14):1439–50.
4. Haykowsky MJ, Liang Y, Pechter D, Jones LW, McAlister FA, Clark AM. A meta-analysis of the effect of exercise training on left ventricular remodeling in heart failure patients: the benefit depends on the type of training performed. J Am Coll Cardiol. 2007;49(24):2329–36.
5. Flynn KE, Pina IL, Whellan DJ, et al. Effects of exercise training on health status in patients with chronic heart failure: HF-ACTION randomized controlled trial. JAMA. 2009;301(14):1451–9.
6. Vanhees L, Rauch B, Piepoli M, et al. Importance of characteristics and modalities of physical activity and exercise in the management of cardiovascular health in individuals with cardiovascular disease (part III). Eur J Prev Cardiol. 2012;19(6):1333–56.
7. Tucker WJ, Lijauco CC, Hearon CM Jr, et al. Mechanisms of the improvement in peak VO2 with exercise training in heart failure with reduced or preserved ejection fraction. Heart Lung Circ. 2018;27(1):9–21.
8. Belardinelli R, Georgiou D, Cianci G, Purcaro A. Randomized, controlled trial of long-term moderate exercise training in chronic heart failure: effects on functional capacity, quality of life, and clinical outcome. Circulation. 1999;99(9):1173–82.
9. Karlsen T, Aamot IL, Haykowsky M, Rognmo O. High intensity interval training for maximizing health outcomes. Prog Cardiovasc Dis. 2017;60(1):67–77.
10. Tabet JY, Meurin P, Beauvais F, et al. Absence of exercise capacity improvement after exercise training program: a strong prognostic factor in patients with chronic heart failure. Circ Heart Fail. 2008;1(4):220–6.
11. Swank AM, Horton J, Fleg JL, et al. Modest increase in peak VO2 is related to better clinical outcomes in chronic heart failure patients: results from heart failure and a controlled trial to investigate outcomes of exercise training. Circ Heart Fail. 2012;5(5):579–85.
12. Gomes Neto M, Duraes AR, Conceicao LSR, Saquetto MB, Ellingsen O, Carvalho VO. High intensity interval training versus moderate intensity continuous training on exercise capacity and quality of life in patients with heart failure with reduced ejection fraction: a systematic review and meta-analysis. Int J Cardiol. 2018;261:134–41.
13. Ellingsen O, Halle M, Conraads V, et al. High-intensity interval training in patients with heart failure with reduced ejection fraction. Circulation. 2017;135(9):839–49.
14. Stoylen A, Conraads V, Halle M, Linke A, Prescott E, Ellingsen O. Controlled study of myocardial recovery after interval training in heart failure: SMARTEX-HF—rationale and design. Eur J Prev Cardiol. 2012;19(4):813–21.
15. Ellingsen O, Halle M, Conraads VM, et al. High intensity interval training in heart failure patients with reduced ejection fraction. Circulation. 2017.
16. Ponikowski P, Voors AA, Anker SD, et al.; ESC Scientific Document Group. 2016 ESC guidelines for the diagnosis and treatment of acute and chronic heart failure: the Task Force for the Diagnosis and Treatment of Acute and Chronic Heart Failure of the European Society of Cardiology (ESC) developed with the special contribution of the Heart Failure Association (HFA) of the ESC. Eur Heart J. 2016;37(27):2129–200.
17. Iellamo F, Manzi V, Caminiti G, et al. Matched dose interval and continuous exercise training induce similar cardiorespiratory and metabolic adaptations in patients with heart failure. Int J Cardiol. 2013;167(6):2561–5.
18. Russell SD, Saval MA, Robbins JL, et al. New York Heart Association functional class predicts exercise parameters in the current era. Am Heart J. 2009;158(4 Suppl):S24–30.
19. Pina IL, Lin L, Weinfurt KP, et al. Hemoglobin, exercise training, and health status in patients with chronic heart failure (from the HF-ACTION randomized controlled trial). Am J Cardiol. 2013;112(7):971–6.
20. Forman DE, Clare R, Kitzman DW, et al. Relationship of age and exercise performance in patients with heart failure: the HF-ACTION study. Am Heart J. 2009;158(4 Suppl):S6–15.
21. Wielenga RP, Huisveld IA, Bol E, et al. Safety and effects of physical training in chronic heart failure. Results of the Chronic Heart Failure and Graded Exercise study (CHANGE). Eur Heart J. 1999;20(12):872–9.
22. Wielenga RP, Huisveld IA, Bol E, et al. Exercise training in elderly patients with chronic heart failure. Coron Artery Dis. 1998;9(11):765–70.
23. Storen O, Helgerud J, Saebo M, et al. The effect of age on the VO2max response to high-intensity interval training. Med Sci Sports Exerc. 2017;49(1):78–85.
24. Helgerud J, Høydal K, Wang E, et al. Aerobic high-intensity intervals improve VO2max more than moderate training. Med Sci Sports Exerc. 2007;39(4):665–71.
25. Jonsdottir S, Andersen KK, Sigurosson AF, Sigurosson SB. The effect of physical training in chronic heart failure. Eur J Heart Fail. 2006;8(1):97–101.
26. Hambrecht R, Gielen S, Linke A, et al. Effects of exercise training on left ventricular function and peripheral resistance in patients with chronic heart failure: a randomized trial. JAMA. 2000;283(23):3095–101.
27. Hambrecht R, Niebauer J, Fiehn E, et al. Physical training in patients with stable chronic heart failure: effects on cardiorespiratory fitness and ultrastructural abnormalities of leg muscles. J Am Coll Cardiol. 1995;25(6):1239–49.
28. Keteyian SJ, Levine AB, Brawner CA, et al. Exercise training in patients with heart failure. A randomized, controlled trial. Ann Intern Med. 1996;124(12):1051–7.
29. Moholdt T, Madssen E, Rognmo O, Aamot IL. The higher the better? Interval training intensity in coronary heart disease. J Sci Med Sport. 2014;17(5):506–10.
30. Aamot IL, Forbord SH, Karlsen T, Stoylen A. Does rating of perceived exertion result in target exercise intensity during interval training in cardiac rehabilitation? A study of the Borg scale versus a heart rate monitor. J Sci Med Sport. 2014;17(5):541–5.
31. Jones LW, Douglas PS, Khouri MG, et al. Safety and efficacy of aerobic training in patients with cancer who have heart failure: an analysis of the HF-ACTION randomized trial. J Clin Oncol. 2014;32(23):2496–502.
32. Conraads VM, Pattyn N, De Maeyer C, et al. Aerobic interval training and continuous training equally improve aerobic exercise capacity in patients with coronary artery disease: the SAINTEX-CAD study. Int J Cardiol. 2015;179:203–10.
33. Hollriegel R, Winzer EB, Linke A, et al. Long-term exercise training in patients with advanced chronic heart failure: sustained benefits on left ventricular performance and exercise capacity. J Cardiopulm Rehabil Prev. 2016;36(2):117–24.
34. Weston KS, Wisloff U, Coombes JS. High-intensity interval training in patients with lifestyle-induced cardiometabolic disease: a systematic review and meta-analysis. Br J Sports Med. 2014;48(16):1227–34.


Supplemental Digital Content

Copyright © 2019 by the American College of Sports Medicine