Author + information
- Received May 31, 2018
- Revision received September 11, 2018
- Accepted September 20, 2018
- Published online November 26, 2018.
- Mauro F.F. Mediano, PhDa,
- Eric S. Leifer, PhDb,
- Lawton S. Cooper, MD, MPHb,
- Steven J. Keteyian, PhDc,
- William E. Kraus, MDd,
- Robert J. Mentz, MDd and
- Jerome L. Fleg, MDb,∗ ()
- aEvandro Chagas National Institute of Infectious Disease, Oswaldo Cruz Foundation, Manguinhos, Rio de Janeiro, Brazil
- bNational Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland
- cHenry Ford Hospital, Detroit, Michigan
- dDuke University Medical Center, Durham, North Carolina
- ↵∗Address for correspondence:
Dr. Jerome L. Fleg, Division of Cardiovascular Sciences, National Heart Lung, and Blood Institute, 6701 Rockledge Drive, Bethesda, Maryland 20892.
Objectives This study sought to evaluate the influence of baseline physical activity (PA) on responses to aerobic exercise training and clinical events in outpatients with chronic systolic heart failure (HF) from the multicenter HF-ACTION (Exercise Training Program to Improve Clinical Outcomes in Individuals With Congestive Heart Failure) trial.
Background The influence of baseline PA on exercise capacity, responses to exercise training and clinical outcomes in patients with chronic HF is unclear.
Methods Of 2,130 participants who provided consent for this analysis, 1,494 patients (64%) had complete baseline PA data, using a modified version of the International Physical Activity Questionnaire–Short Form questionnaire and were included in the analysis; 742 received usual care and 752 were allocated to the exercise training group. Changes in exercise capacity, all-cause mortality and hospitalization, cardiovascular (CV) mortality and hospitalization, and CV mortality and HF hospitalization were evaluated as a function of baseline PA tertile.
Results At baseline, the highest PA tertile showed greater peak oxygen uptake, cardiopulmonary exercise test duration, and 6-min walk test distance than the other 2 PA tertiles, as well as lower New York Heart Association functional class, lower Beck depression score, and lower atrial fibrillation prevalence than the lowest PA tertile. Compared to the lowest PA tertile, the middle tertile had 18% lower risk of CV death/CV hospitalizations, and the upper tertile showed 23% lower risk of CV death/HF hospitalizations. Exercise capacity and clinical outcome responses to training were similar and largely nonsignificant across baseline PA tertiles with significant benefit of training on exercise test duration for all tertiles.
Conclusions In patients with chronic systolic HF, aerobic exercise training significantly improves exercise test duration to a similar extent across baseline PA tertiles. Although higher baseline PA was associated with lower risk of clinical events, no significant differences in event rates within each PA tertile were seen between subgroups randomized to exercise training versus usual care. (Exercise Training Program to Improve Clinical Outcomes in Individuals With Congestive Heart Failure [HF-ACTION]; NCT00047437)
Although aerobic exercise training has been advocated as an important modality in the treatment of chronic heart failure (HF) to improve functional capacity and possibly reduce clinical events, its effects on mortality reduction in patients with HF remain controversial (1–3). A meta-analysis by Davies et al. (4) that examined exercise training in patients with chronic systolic HF showed reductions in HF-related hospitalizations and improvements in quality of life but no significant influence on all-cause mortality. Lewinter et al. (5) confirmed these previous results, demonstrating significant improvements in exercise capacity and hospitalizations but not in mortality. Moreover, the results of studies examining improvements in peak VO2 after an exercise training program vary widely (6,7).
These variable responses to exercise training in patients with chronic HF may be explained partly by differences in baseline physical activity (PA) levels, as higher PA levels are important clinical and mortality predictors in a variety of populations, including HF patients (8–10). For instance, Miura et al. (11) found that higher baseline PA levels were significantly associated with lower rates of all-cause death, HF hospitalizations, and other cardiovascular (CV) events in a 3-year longitudinal multicenter study of patients with chronic HF.
Although baseline PA level has emerged as an important variable associated with clinical outcomes, its influence on the response to exercise training among patients with HF is not well characterized. Therefore, this study sought to evaluate the influence of baseline PA on the responses to exercise training and clinical events in outpatients with chronic HF with reduced ejection fraction from the multicenter HF-ACTION (Heart Failure: A Controlled Trial Investigating Outcomes of Exercise TraiNing) trial.
The present study is a secondary analysis of the HF-ACTION study, a multicenter, international, randomized clinical trial investigating the effects of exercise training on morbidity and mortality in patients with chronic HF with reduced ejection fraction. The full description and the main results of the trial have been published elsewhere (1,12,13).
HF-ACTION enrolled stable outpatients from the United States, Canada, and France with left ventricular ejection fraction <35% and NYHA functional classes II to IV symptoms despite optimal HF therapy. The study excluded individuals who reported exercising more than once per week at a moderate-to-vigorous intensity at any time during the previous 6 weeks. All participants provided signed informed consent, and the institutional review board or ethics committee for each participating center approved the study. The study was funded by the U.S. National Heart, Lung, and Blood Institute/National Institutes of Health.
Participants were randomly assigned to an exercise training or usual care group. Both groups received general advice regarding symptoms and clinical treatment, but only the exercise training group was provided with a formal exercise intervention. During the first 36 sessions, exercise training consisted of 3 sessions per week of supervised treadmill walking or stationary cycling, initially for 35 min and gradually increased to 55-min duration, at 60% to 70% of heart rate reserve. After the completion of 18 supervised training sessions, patients began concomitant home-based exercise twice per week until the 36 supervised exercise sessions were completed. After that point, they were fully transitioned to a 5-day-per-week home-based exercise program consisting of 40 min per session at 60% to 70% of heart rate reserve until the end of the trial, typically 2 to 4 years.
Endpoints considered in the present analysis were those for the HF-ACTION trial, consisting of the primary outcome (a composite of all-cause mortality or all-cause hospitalization), secondary outcomes (composite of CV mortality or CV hospitalization and the composite of CV mortality or HF hospitalization), and changes in exercise capacity during the follow-up as measured by peak oxygen uptake (VO2), cardiopulmonary exercise test (CPET) duration, and 6-min walk test (6MWT) distance. A clinical endpoint committee blinded to the treatment assignment adjudicated deaths and hospitalizations. The CPET and 6MWT were administered at baseline and at 3 and 12 months of assigned treatment to assess functional exercise tolerance, as described previously (14–16).
Baseline physical activity
Self-reported baseline PA was evaluated using the International Physical Activity Questionnaire–Short Form (IPAQ-SF). This instrument was developed by a group of experts in 1998 to facilitate surveillance of PA based on global standards and was validated in 12 countries from 6 different continents (17). The IPAQ has since become the most widely used physical activity questionnaire, with 2 versions available: the IPAQ-long form (LF) and the IPAQ-SF). The IPAQ-SF consists of 7 questions about the frequency, duration, and intensity (vigorous or moderate and walking activities) of 4 different PA domains: work, transportation, and home and leisure time (18). Both categorical and continuous indicators of PA are possible from IPAQ-SF. In the present study, PA data collected were reported as a continuous variable, weighting each type of activity by its energy requirements defined in terms of metabolic equivalents of task (MET) to yield a score in METs minutes per week, according to the IPAQ guideline for data processing and analysis (19).
Clinical and demographic characteristics of patients are mean ± SD for continuous variables and frequencies (percentages) for categorical variables. Participants were allocated to 1 of 3 groups according to their tertiles of baseline PA: low, middle, and high. The follow-up time was measured from the time of study entry (time zero) until occurrence of mortality or hospitalization endpoints or the time of final participant contact, including those who withdrew consent or were lost to follow-up, at which point follow-up was censored.
Linear mixed models were used to assess the relationship between baseline PA tertile and exercise training on exercise capacity changes (VO2 peak, CPET duration, and 6MWT distance) during follow-up. First, changes within each PA tertile were examined by using a model which included fixed effect terms for time (baseline, 3 months, 12 months), treatment group, and time × treatment group interaction. This model was fitted without adjusting for other covariates as well as adjusted for age, sex, HF cause, and known HF-ACTION prognostic variables (history of atrial fibrillation, left ventricular ejection fraction, Beck depression score), as previously described (1). Adjustments were not made for baseline CPET duration because duration was 1 of the exercise outcomes in this analysis. The terms of interest were the time × treatment group interactions at 3 and 12 months. The interaction terms estimate, respectively, the exercise training versus usual care differences in the baseline-to-3-month and baseline-to-12-month change in exercise capacity. Next, changes in exercise capacity was examined between PA tertiles, from baseline to 3 months and baseline to 12 months, using a model that included fixed effect terms for time, baseline PA tertile, treatment group, and time × baseline PA tertile × treatment group interaction. This model was fitted both without and after adjusting for other covariates as described above. For both time points, each of the 2 interaction terms estimated differences between the exercise training versus usual care in the baseline-to-3-month and baseline-to-12-month change, respectively, in the exercise outcome relative to the lowest tertile of baseline PA. At both 3 and 12 months, 1 interaction term compares the middle tertile to the lowest tertile; the other interaction term compares the highest tertile to the lowest tertile. Residual plots of all models were visually examined and did not demonstrate substantial deviations from the regression assumptions.
Cox regression models were used to investigate the influence of baseline PA tertile on mortality/hospitalization outcomes. For each outcome, both the unadjusted and adjusted models were fitted as described above for linear mixed models, except that CPET duration was included as a covariate in the adjusted model. Separate Cox models were used to investigate the effect of group assignment (exercise training vs. usual care) response to mortality/hospitalization outcomes within each tertile. For each outcome and each tertile, a model was fitted that included a treatment group term, either unadjusted or adjusted, for the prognostic variables. To investigate the heterogeneity between the baseline PA tertiles of the exercise training response to each outcome, a separate model was fitted that included terms for PA tertile, treatment group, and PA tertile × treatment group interaction. The Schoenfeld residuals test did not demonstrate substantial deviation from proportional hazard assumptions. Data analyses were performed using Stata version 13.0 software (College Station, Texas), and statistical significance was set at a 2-tailed p value of <0.05 for all analyses.
Of the 2,331 participants enrolled in the HF-ACTION study, 201 did not give consent for the use of de-identified data and were not included in this secondary analysis. From those 2,130 who remained, 636 did not have complete data from the baseline PA questionnaire and were excluded from data analysis, leaving a sample of 1,494 participants (Figure 1). The baseline clinical and demographic and exercise capacity characteristics of participants with missing and nonmissing data for the IPAQ physical activity questionnaire are shown in Online Table 1. The group with missing data had a higher proportion of women and a lower treadmill exercise duration. No further differences were observed between groups.
Table 1 describes the baseline characteristics of all participants included in the present analysis according to their baseline PA tertiles. The median (5th to 95th percentiles) baseline PA was 0 (0 to 66), 278 (99 to 594), and 1,560 (693 to 7,173) MET ∙ min ∙ week−1 for low, middle, and high baseline PA level groups, respectively. Overall, the highest PA tertile had greater peak VO2, CPET duration, and 6MWT distance than the other 2 tertiles at baseline, plus lower NYHA functional class, lower Beck depression score, and less atrial fibrillation than the lowest PA tertile.
The influence of baseline PA tertile on response to exercise training is depicted in Table 2. There is a trend toward exercise training benefit within all 3 tertiles for all 3 exercise endpoints. However, only CPET duration showed significant improvement within all 3 tertiles. There were no significant differences in exercise training response between tertiles except for peak VO2 at 3 months, where the middle and lowest tertiles had significantly different exercise responses at 3 months in the unadjusted analysis (−0.05 vs. +0.7, respectively; p = 0.04), but not in the fully adjusted analysis (p = 0.06). In addition, a separate analysis was performed to determine whether baseline peak VO2 predicted subsequent exercise responses to training. No differential responses between tertiles of baseline peak VO2 were observed (p > 0.10 for all comparisons).
Tables 3 and 4⇓⇓ show the relationships between baseline PA and mortality/hospitalization outcomes over the mean follow-up duration of 30.5 ± 13.0 months. In Table 3, which is not stratified as to assigned treatment group, participants in both the middle and the highest PA tertiles experienced lower rates of all 3-event outcomes than those in the lowest tertile in the unadjusted model. The unadjusted survival curves for CV mortality/HF hospitalization according to baseline PA tertiles are presented in Figure 2. However, in the fully adjusted model, only the 18% lower CV death/CV hospitalization rate in the middle tertile and the 23% lower CV death/HF hospitalization rate in the highest tertile were significantly different from the lowest tertile.
Table 4 compares the treatment group responses to mortality/hospitalization outcomes both within and between baseline PA tertiles. Within tertiles, there were no significant differences between the exercise training and usual care groups with the exception of less event risk in the lowest tertile unadjusted analysis for CV death/HF hospitalization. Despite the lack of statistical significance, the estimated hazard ratios were typically in the beneficial direction. The effects of exercise training on CV mortality/HF hospitalization within each tertile of baseline PA are illustrated in Figure 3. For the fully adjusted model, exercise training had no significant effect within each tertile.
The major findings of this secondary analysis from HF-ACTION are as follows. Stable outpatients with chronic systolic HF with higher baseline PA levels demonstrated a more favorable clinical profile and higher exercise capacity than those with lower PA. However, a program of supervised and home-based exercise training significantly improved exercise test duration similarly across baseline PA tertiles. Higher baseline PA was associated with lower risk of clinical events during long-term follow-up. No significant differences were seen in event rates within each PA tertile between subgroups randomized to exercise training versus those receiving usual care, although most hazard ratios were in the direction of benefit from training.
Prior studies in the general population have suggested greater benefits from increasing PA levels on both exercise capacity and clinical events in the subset with lowest baseline activity levels (20). For example, a study by Hautala et al. (21) showed a significant inverse correlation between habitual PA levels and changes in VO2 max among healthy volunteers following an 8-week training period. Regarding the effect of exercise training on clinical outcomes, a meta-analysis published by Woodcock et al. (22), including almost 1 million adults from 22 prospective cohort studies, suggested greater mortality reduction by increasing PA levels among sedentary individuals in the general population. Those results are in accordance with a recent modeling study, indicating that the least active people experienced the greatest reductions in mortality from increasing PA levels by initiating a formal exercise program (23). In contrast, similar improvements were found in exercise capacity and no significant differences in clinical outcomes across baseline PA tertiles in patients with chronic HF in response to exercise training.
To the best of our knowledge, this study is the first to evaluate the influence of different baseline PA levels on exercise training responses among patients with chronic systolic HF. Differences observed between the results of the present study in comparison to those consisting of healthy participants may be attributed in part to the marked differences between the clinical profile of HF patients and those of the general population (24,25). The observation that aerobic capacity improved similarly in the 3 PA tertiles reinforces the message that exercise training should be included as a therapeutic strategy for all capable patients with clinically stable systolic HF to improve exercise tolerance and health status (26).
The current analysis confirms prior findings that higher habitual levels of PA are associated with lower risk of mortality/hospitalization in patients with HF (27). In a sample of 84 patients with chronic HF, Walsh et al. (28) observed an increased risk of death in the subset with low pedometer-derived PA levels. More recent studies in cohorts with HF and preserved ejection fraction have shown similar findings as well as higher rates of HF hospitalizations in the groups with low PA levels (29). In addition, studies in patients with coronary artery disease have demonstrated that sedentary patients experienced increased CV mortality risk compared to more active patients, even after adjustments for potential confounders (30). Although participation in any formal exercise program prior to the study was considered an HF-ACTION exclusion criterion, the habitual baseline PA levels of patients included in the study varied widely. Overall, the highest baseline PA tertile experienced lowest CV mortality/HF hospitalization (Table 3), suggesting a dose-response relationship between PA and that outcome among patients with HF as previously demonstrated (31). A similar graded benefit of higher PA for reduction in mortality and incidence of coronary heart disease has been observed in the general population (32,33). The consistency of these findings can serve as a motivating strategy to increase PA levels among sedentary HF patients.
The HF-ACTION trial provided a unique opportunity to assess the effect of a prolonged exercise training intervention on clinical events as a function of their baseline PA levels. The current analysis demonstrated a nonsignificant 33% lower risk for the composite secondary endpoint of CV death/HF hospitalizations within the highest PA tertile subgroup randomized to exercise training versus controls. Directionally similar nonsignificant trends of lesser magnitude were observed for CV deaths/CV hospitalizations. These findings parallel the overall findings of HF-ACTION showing trends toward lower clinical event rates in the group randomized to exercise training.
The exclusion of patients without baseline PA data may have introduced some selection bias, as those excluded were more often women and had lower baseline exercise capacity. Furthermore, the resulting 35% decrease in sample size (1,494 vs. 2,331 patients), may have contributed to the lack of statistical significance for some associations. The small changes for exercise capacity variables observed in this secondary HF-ACTION study were similar to those observed in the primary study (1). The suboptimal adherence in the exercise training group likely blunted differences compared to the usual care group with respect to exercise capacity and clinical events during follow-up. In addition, the use of questionnaires rather than direct measurements to evaluate PA levels may have introduced recall bias and/or measurement error (34). However, questionnaires are the most common method for obtaining PA data (35). The IPAQ-SF used in HF-ACTION is the most widely used PA questionnaire, presenting acceptable measurements properties to estimate PA levels in different populations, although it has not been previously validated for patients with HF (18). Finally, it should be emphasized that by excluding individuals who reported regular moderate or vigorous exercise in the 6 weeks prior to study enrollment, the range of self-reported baseline PA was truncated, thereby reducing the likelihood of observing the positive effects of greater PA on clinical outcomes.
In stable outpatients with chronic systolic HF, higher baseline PA was associated with a more favorable clinical profile and greater exercise capacity. However, a program of supervised and home-based exercise training significantly improved exercise test duration similarly across baseline PA tertiles. Although higher baseline PA was associated with a lower risk of clinical events during long-term follow-up, no significant differences in event rates within each PA tertile were seen between subgroups randomized to exercise training versus usual care, although most hazard ratios were in the direction of benefit. These findings reinforce the concept that the salutary effects of exercise training in patients with chronic HF are similar regardless of baseline PA level.
COMPETENCY IN MEDICAL KNOWLEDGE: Higher baseline PA levels were associated with similar training-induced responses but lower clinical event rates in patients with chronic HF. These results reinforce the need for motivating patients with HF to participate in exercise training programs, regardless of their habitual PA level.
TRANSLATIONAL OUTLOOK: Identification of characteristics that are associated with different exercise training responses is of paramount importance to the development of tailored strategies to improve clinical and exercise capacity outcomes in patients with HF. Further studies are necessary to examine the influence of baseline characteristics on responses to exercise training.
The authors thank Drs. Adrian Coles and Stephen Ellis for help in understanding the physical activity dataset.
Supported by National Heart, Lung, and Blood Institute/National Institutes of Health grant 5U01HL063747. The contents of this paper are the sole responsibility of the authors and do not necessarily reflect the views of the National Heart, Lung, and Blood Institute, the National Institutes of Health, or the U.S. Department of Health and Human Services. All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- 6-min walk test
- cardiopulmonary exercise test
- heart failure
- New York Heart Association
- physical activity
- Received May 31, 2018.
- Revision received September 11, 2018.
- Accepted September 20, 2018.
- 2018 American College of Cardiology Foundation
- Belardinelli R.,
- Georgiou D.,
- Cianci G.,
- Purcaro A.
- Fleg J.L.,
- Cooper L.S.,
- Borlaug B.A.,
- et al.
- Davies E.J.,
- Moxham T.,
- Rees K.,
- et al.
- Pina I.L.,
- Apstein C.S.,
- Balady G.J.,
- et al.
- Tucker W.J.,
- Lijauco C.C.,
- Hearon C.M. Jr..,
- et al.
- da Silva V.Z.,
- Lima A.C.,
- Vargas F.T.,
- Cahalin L.P.,
- Arena R.,
- Cipriano G. Jr..
- Keteyian S.J.,
- Patel M.,
- Kraus W.E.,
- et al.
- Balady G.J.,
- Arena R.,
- Sietsema K.,
- et al.
- ↵International Physical Activity Questionnaire. Guidelines for data processing and analysis of the International Physical Activity Questionnaire (IPAQ), 2005. Available at: https://sites.google.com/site/theipaq/scoring-protocol. Accessed April 4, 2017.
- Minton J.,
- Dimairo M.,
- Everson-Hock E.,
- Scott E.,
- Goyder E.
- Roger V.L.
- Writing Committee M.,
- Yancy C.W.,
- Jessup M.,
- et al.
- Hegde S.M.,
- Claggett B.,
- Shah A.M.,
- et al.
- Keteyian S.J.,
- Leifer E.S.,
- Houston-Miller N.,
- et al.
- Arem H.,
- Moore S.C.,
- Patel A.,
- et al.
- Sattelmair J.,
- Pertman J.,
- Ding E.L.,
- Kohl H.W.,
- Haskell W.,
- Lee I.M.
- Silsbury Z.,
- Goldsmith R.,
- Rushton A.