Psychosocial Factors, Exercise Adherence, and Outcomes in Heart Failure Patients: Insights from HF-ACTION

Abstract

Background

Psychosocial factors may influence adherence with exercise training for heart failure patients. We aimed to describe the association between social support and barriers to participation with exercise adherence and clinical outcomes.

Methods and Results

Of patients enrolled in HF-ACTION, 2279 (97.8%) completed surveys to assess social support and barriers to exercise, resulting in the perceived social support score (PSSS) and barriers to exercise score (BTES). Higher PSSS indicated higher levels of social support, while higher BTES indicated more barriers to exercise. Exercise time at 3 and 12 months correlated with PSSS (r= 0.09 and r= 0.13, respectively) and BTES (r= − 0.11 and r= − 0.12, respectively), with higher exercise time associated with higher PSSS and lower BTES (All p <0.005). For CV death or HF hospitalization, there was a significant interaction between randomization group and BTES (p=0.035), which corresponded to a borderline association between increasing BTES and CV death or HF hospitalization in the exercise group (HR 1.25, 95% CI: 0.99, 1.59) but no association in the usual care group (HR 0.83, 95% CI: 0.66, 1.06).

Conclusions

Poor social support and high barriers to exercise were associated with lower exercise time. PSSS did not impact the effect of exercise training on outcomes. However, for CV death or HF hospitalization, exercise training had a greater impact on patients with lower BTES. Given that exercise training improves outcomes in heart failure patients, assessment of perceived barriers may facilitate individualized approaches to implement exercise training therapy in clinical practice.

Clinical Trial Registration

URL: http://www.clinicaltrials.gov. Unique identifier: NCT00047437.

In February 2014, Medicare announced the decision to cover cardiac rehabilitation for patients with chronic heart failure with reduced ejection fraction. Despite the known benefits of cardiac rehabilitation for patients with heart failure and other cardiovascular diseases, it has been shown to be an under-utilized therapy.¹ Among patients who do participate, adherence may be less than recommended.² Even in a clinical trial setting, many patients do not fully adhere to recommended exercise programs.³

It is important to understand the reasons for poor exercise adherence and how low adherence affects clinical outcomes. In addition to physical limitations, psychosocial factors may also influence adherence to exercise. To date, there are limited data characterizing the relationship of psychosocial factors to exercise adherence in heart failure patients. Heart Failure: A Controlled Trial Investigating Outcomes of Exercise Training (HF-ACTION) was the largest clinical trial to assess the safety and efficacy of aerobic exercise training in patients with heart failure with reduced ejection fraction.³ Exercise training was associated with reduced incidence of all-cause mortality or all-cause hospitalization and cardiovascular mortality or heart failure hospitalization.

Using data from HF-ACTION, we aimed to describe characteristics of patients who had poor adherence to exercise training and identify the influence of social support in exercise training and on outcomes in this patient population using a validated tool to assess perceived social support. We also explored a novel tool to assess perceived barriers to participation in an exercise program. We hypothesized that lower levels of social support and more perceived barriers to exercise would be associated with worse adherence to an exercise training program and worse outcomes.

Methods

Trial Overview

The design and primary results of the HF-ACTION study have been published previously.^{3, 4} Briefly, this was a multicenter, randomized controlled trial designed to evaluate the efficacy and safety of exercise training in patients with stable chronic heart failure (New York Heart Association [NYHA] Class II–IV) with reduced ejection fraction (left ventricular ejection fraction [LVEF] ≤ 35%). Patients were randomized to usual care or usual care plus aerobic exercise training. Exercise training consisted of supervised exercise three times per week for 36 sessions over three months, with increasing duration and intensity over time, with a gradual transition to a home exercise program for an additional two years. The exercise training goal was 90 minutes per week for the first three months, then 120 minutes per week thereafter. The protocol was approved by the institutional review board or ethics committee at each institution and the coordinating center. All patients provided written informed consent.

The primary endpoint was all-cause mortality or all-cause hospitalization, and the secondary endpoints were all-cause mortality, cardiovascular mortality or cardiovascular hospitalization, and cardiovascular mortality or heart failure hospitalization. An independent committee blinded to treatment assignment adjudicated clinical events.

Assessments

Patients enrolled in HF-ACTION were asked to assess sources of social support and anticipate potential barriers to participation in the study. The Perceived Social Support Scale (PSSS) is a validated tool to asses feelings of social support (for example: My family really tries to help me, I can talk about my problems with my friends) in which patients rate 12 statements on a 7 point scale (i.e.: very strongly disagree [1], strongly disagree [2], mildly disagree [3], neutral [4], mildly agree [5], strongly agree [6], very strongly agree [7]).⁵ PSSS scores range from 12 to 84 with higher scores indicating a higher level of perceived social support. The Barrier Scale is a novel scale developed for this study which asked patients to evaluate the extent to which 10 potential barriers (for example: finances, child care, weather) might interfere with their participation in an exercise program, on a 3-point scale (i.e.: not at all [1], somewhat [2], a lot [3]). Barriers to exercise score (BTES) range from 10 to 30 with higher scores indicating more barriers to exercise. Surveys were completed at baseline and 12 months. (See Supplemental Appendix for complete Perceived Social Support Scale and Barriers to Exercise Scale). Patients also completed the Beck Depression Inventory II (BDI-II) and the Kansas City Cardiomyopathy Questionnaire (KCCQ) at baseline, every 3 months for the first year, and yearly thereafter.

Exercise adherence was assessed in patients randomized to exercise therapy and was measured by minutes per week of exercise, described as poor adherence (< 90 min/week), partial adherence (≥ 90 min/week for 3 months, then < 120 min/week thereafter), or full adherence (≥ 90 min/week for 3 months, then ≥ 120 min/week). Cardiopulmonary exercise (CPX) testing was completed at baseline, 3 months, 12 months, and 24 months. CPX testing was performed using a modified Naughton treadmill protocol or a leg ergometer, and exercise continued until sign- or symptom-limited maximal exertion was reached. Peak oxygen uptake (VO₂) was defined as the VO₂ at peak exercise, either within the last 90 seconds of exercise or the first 30 seconds of recovery, whichever was higher. CPX data were analyzed at a core laboratory.

Statistical Analyses

Baseline characteristics were reported by level of exercise adherence throughout the duration of the study among patients in the exercise arm of the study, alive at 12 months, and with adherence measured at both 3 and 12 months. Continuous variables were reported as medians with 25^th and 75^th percentiles and compared using the Kruskal-Wallis test. Categorical variables were reported as frequencies and percentages and compared using Pearson Chi-square or Fisher’s exact tests.

We used unadjusted Spearman correlation to assess whether baseline PSSS was associated with adherence to exercise at 3 months or 12 months. Similar methods were used to assess whether baseline PSSS was associated with baseline depression, as measured by the BDI-II, or baseline or change in quality of life, as measured by the KCCQ. We also used Spearman correlation to assess whether baseline PSSS was associated with change in exercise capacity from baseline to 12 months, as measured by peak VO₂. To visualize the data, we provide descriptive statistics by quintile of PSSS. Kaplan-Meier methods were used to compare outcomes by quintiles of PSSS; log-rank test p-values compare survival across quintiles. We used paired t-test to compare scores at baseline and 12 months.

Cox proportional hazards models assessed the association between baseline PSSS and clinical outcomes. PSSS was evaluated as a continuous variable; in additional analyses, we also evaluated PSSS dichotomized as above the median or at or below the median in Cox models. Linearity of the relationship between continuous PSSS and the log hazard rate was assessed by a likelihood ratio test comparing the linear fit to the fit of a restricted cubic spline transformation. The proportional hazards assumption was assessed by plotting Schoenfeld residuals against functions of time and calculating the correlation of these. For both assumptions, no violations were detected. The unadjusted model included baseline PSSS score, while the adjusted model included baseline score and adjustment variables that were previously used in other post-hoc HF-ACTION analyses.^6–10 We assessed whether belonging to certain sub-groups modified the association between PSSS and outcomes using interactions. When significant, the association between PSSS and outcome was reported by sub-group. The analysis was repeated for BTES.

Cox proportional hazards models were also used to assess the association between exercise adherence and clinical outcomes after 1 year. We used weighting methods to account for missing exercise adherence data. Weights were calculated as the inverse of a patient’s estimated probability to have non-missing endpoint data.¹¹ A sensitivity analysis was done without the inverse weighting method, with similar results.

Two-tailed P-values <0.05 were considered statistically significant. SAS version 9.2 or higher (SAS Institute Inc., Cary, North Carolina) was used for all analyses.

Results

Of the 2331 patients enrolled in HF-ACTION, 2279 (97.8%) patients completed baseline surveys to assess for feelings of social support and perceived barriers to exercise with 1090 randomized to exercise training (Figure 1). For patients randomized to exercise training with completed surveys and adherence data, baseline PSSS weakly correlated with exercise time (minutes/week) at 3 and 12 months (r = 0.09 and r = 0.13, respectively), as did baseline BTES (r = −0.11 and r = −0.12, respectively, all p < 0.01) (Figure 2). For the highest quintile PSSS compared to the lowest quintile, median exercise time at 3 months was 79 min/week vs 71 min/week, and at 12 months was 118 min/week vs 92 min/week, respectively. For the lowest quintile BTES compared to the highest quintile, median exercise time at 3 months was 89 min/week vs 70 min/week, and at 12 months was 169 min/week vs 86 min/week, respectively.

Figure 1.

Study Patient Population

Figure 2.

Relationship Between Baseline Perceived Social Support Scores and Barriers to Exercise Scores and Exercise Adherence.

Panel A shows the minutes per week of exercise at 3 months by baseline PSSS, N=1073.

Panel B shows the minutes per week of exercise at 12 months by baseline PSSS, N=909.

Panel C shows the minutes per week of exercise at 3 months by baseline BTES, N=1073.

Panel D shows the minutes per week of exercise at 12 months by baseline BTES, N=909.

Table 1 describes the baseline characteristics of the participants in the exercise training group with completed baseline surveys, grouped by adherence to exercise. While 1090 patients in the exercise training group had completed baseline surveys, only 935 had complete exercise adherence data, and were included in Table 1. Overall, more patients were poorly adherent than fully adherent. Patients with poor adherence were more likely to be younger, female, and black, with higher body mass index. At baseline, patients with poor adherence were more likely to have NYHA Class III–IV symptoms but the frequency of prior heart failure hospitalization was similar between groups. Patients with poor adherence also had significantly worse baseline exercise capacity as measured by peak VO₂ and 6 minute walk test distance. Patients with poor adherence had lower quality of life as measured by the KCCQ and higher levels of depression as measured by the BDI-II Score, compared to patients with full adherence and partial adherence. Furthermore, poorly adherent patients were less likely to be married or living with a partner than fully adherent patients.

Table 1.

Baseline Patient Characteristics by Adherence Groups

Characteristic	Full Adherence* (N=238)	Partial Adherence* (N=143)	Poor Adherence* (N=554)	P-Value
Patient Characteristics
Age, years (median, 25th–75th)	62 (52–70)	62 (55–71)	58 (50–67)	<.001
Female Sex	59 (24.8%)	40 (28.0%)	190 (34.3%)	0.02
Race, White	175 (74.2%)	95 (66.4%)	310 (56.9%)	<.001
Body Mass Index, kg/m^2 (median, 25th–75th)	29 (25–34)	29 (25–33)	31 (26–36)	0.002
Higher Education†	144/230 (62.6%)	97/142 (68.3%)	345/540 (63.9%)	0.52
Employed‡	159 (67.4%)	103 (73.6%)	324 (59.8%)	0.004
Higher Income§	145/210 (69.0%)	87/126 (69.0%)	285/503 (56.7%)	0.001
Partnered Status‖	163 (69.1%)	97 (67.8%)	318 (57.7%)	0.003
Heart Failure Hospitalizations prior 6 months				0.98
None	181 (76.7%)	107 (74.8%)	417 (76.0%)
1	43 (18.2%)	28 (19.6%)	97 (17.7%)
2	9 (3.8%)	5 (3.5%)	25 (4.6%)
3 or more	3 (1.3%)	3 (2.1%)	10 (1.8%)
Left Ventricular EF, % (median, 25th–75th)	25 (22–30)	26 (21–31)	24 (20–29)	0.02
NYHA III–IV vs II	74 (31.1%)	38 (26.6%)	240 (43.3%)	<.001
Baseline Exercise Capacity
Peak VO2, mL/kg/min (median, 25th–75th)	15.4 (12.8–18.7)	14.4 (11.9–17.8)	13.7 (10.9–17.2)	<.001
Peak Heart Rate with Exercise, bpm (median, 25th–75th)	120 (108–136)	115 (104–130)	120 (103–136)	0.14
VE/VCO2 Slope (median, 25th–75th)	32 (28–38)	33 (29–39)	33 (28–40)	0.43
CPX Duration, min (median, 25th–75th)	10 (8–13)	10 (7–12)	9 (6–12)	<.001
6 Minute Walk Distance, m (median, 25th–75th)	391 (326–454)	378 (312–431)	357 (279–424)	<.001
Patient Quality of Life
KCCQ Overall Summary Score (median, 25th–75th)	70 (58–86)	69 (54–85)	67 (47–81)	0.003
Beck Depression Score (median, 25th–75th)	8 (4–14)	8 (5–14)	9 (5–16)	0.08
Euro Thermometer (median, 25th–75th)	70 (55–80)	70 (60–80)	66 (50–80)	0.05

^*Full Adherence: ≥90minutes/week for 3 months then ≥120minutes/week. Partial Adherence: ≥90minutes/week for 3 months then <120minutes/week. Poor Adherence: <90minutes/week

^†Dichotomized comparison of education by <high school vs ≥high school

^‡Dichotomized comparison of current or prior employment (employed, volunteer, student, homemaker, retired) vs No employment (unemployed, disabled)

^§Dichotomized comparison of income by <$25,000 vs ≥$25,000

^‖Dichotomized comparison of current or prior partner (married, living with partner, widowed) vs No partner (single, divorced, separated)

Abbreviations: CPX: cardiopulmonary exercise testing, EF: ejection fraction, KCCQ: Kansas City Cardiomyopathy Questionnaire, NYHA: New York Heart Association, VE: minute ventilation, VCO₂ carbon dioxide production, VO₂ oxygen consumption

For all patients with baseline and 12 month PSSS and BTES data, regardless of randomization group, PSSS and BTES were not significantly different between baseline and 1 year, with a median PSSS of 72 (IQR 62, 80) at baseline and 72 (IQR 63, 81) at follow-up, and a median BTES of 12 (IQR 11, 14) at baseline and 12 (IQR 10, 14) at follow up. Figure 3 shows the distribution of scores at baseline and follow-up.

Figure 3.

Distribution of Perceived Social Support and Barriers to Exercise Scores at Baseline and 12 Months

Panel A shows the distribution of PSSS at baseline.

Panel B shows the distribution PSSS at 12 months.

Panel C shows the distribution of BTES at baseline.

Panel D shows the distribution of BTES at 12 months.

For all patients, baseline PSSS score correlated with baseline measures of depression and quality of life (Table 2). As PSSS increased, KCCQ scores also increased (r = 0.29, p < 0.0001) and BDI-II score decreased (r = −0.36, p < 0.0001). Baseline BTES also correlated modestly with baseline BDI-II score (r = 0.29, p < 0.0001) and KCCQ scores (r = −0.22, p < 0.0001). Baseline PSSS and BTES were not significantly correlated with change in KCCQ from baseline to 12 months.

Table 2.

Correlation between Baseline Scores and Baseline Depression and Baseline and Change in Quality of Life

	Baseline PSSS		Baseline BTES
	Spearman Correlation Coefficient	P-Value	Spearman Correlation Coefficient	P-Value
Baseline BDI-II Score	−0.36	<.001	0.29	<.001
Baseline KCCQ Score	0.29	<.001	−0.22	<.001
Change in KCCQ Score	−0.05	0.05	0.03	0.22

Abbreviations: BDII: Beck Depression Inventory; BTES: Barriers to Exercise Score, KCCQ: Kansas City Cardiomyopathy Questionnaire, PSSS: Perceived Social Support Score

For all patients, regardless of whether they were randomized to exercise training or usual care, neither baseline PSSS nor baseline BTES was significantly correlated with change in peak VO₂ from baseline to 12 months (r = − 0.005, p=0.86 and r = 0.04, p=0.10, respectively).

There was no association between PSSS and the primary outcome of all-cause death or hospitalization (p=0.98). There was, however, a trend towards an association between BTES and all-cause death or hospitalization events, with higher quintiles associated with more events than lower quintiles (P=0.05). The unadjusted Kaplan Meier event curves for the primary end-point of all-cause death or hospitalization based on baseline PSSS and BTES are included in the Supplemental Appendix, Figure 1.

In unadjusted and adjusted Cox regression models, there was no association between baseline PSSS and all-cause death or hospitalization (Table 3). For BTES, however, in Cox regression models, when BTES was divided at the median, there was a trend toward a significant association between baseline BTES and all-cause death or hospitalization (BTES above median: HR 1.11, 95% CI: 1.00, 1.25, p=0.080); however, this association did not persist after multivariable adjustment. No significant association was observed for continuous BTES and all-cause death or hospitalization in unadjusted (HR 1.06, 95% CI: 0.94, 1.18, p=0.36) or adjusted (HR 0.99, 95% CI: 0.88, 1.11, p=0.83) analyses. The secondary outcome of cardiovascular death or heart failure hospitalization was not associated with baseline PSSS or baseline BTES in either unadjusted or adjusted analyses (Table 3).

Table 3.

Cox Proportional Hazard Models for Outcomes by Baseline Scores

Outcome	Score	Hazard Ratio (95% CI) Unadjusted	P Value	Hazard Ratio (95% CI) Adjusted	P Value
All-cause death/All-cause hospitalization	Baseline BTES: Per 5 unit increase	1.06 (0.94, 1.18)	0.36	0.99 (0.88, 1.11)	0.83
	Baseline BTES >12 vs. ≤12	1.11 (0.99, 1.25)	0.08	1.07 (0.95, 1.21)	0.24
	Baseline PSSS: Per 10 unit increase	1.01 (0.97, 1.05)	0.68	1.01 (0.98, 1.05)	0.50
	Baseline PSSS >72 vs. ≤72	0.93 (0.82, 1.04)	0.19	0.92 (0.82, 1.04)	0.19
Cardiovascular death/Heart Failure hospitalization	Baseline BTES: Per 5 unit increase	1.03 (0.87, 1.22)	0.73	0.94 (0.79, 1.12)	0.47
	Baseline BTES >12 vs. ≤12	1.04 (0.88, 1.23)	0.64	0.96 (0.81, 1.14)	0.62
	Baseline PSSS: Per 10 unit increase	0.99 (0.94, 1.05)	0.70	1.00 (0.95, 1.06)	0.90
	Baseline PSSS >72 vs. ≤72	0.90 (0.76, 1.06)	0.20	0.91 (0.76, 1.08)	0.26

Abbreviations: BTES: Barriers to Exercise Score, PSSS: Perceived Social Support Score

We tested the interaction between scores and randomization group or demographic sub-groups for long-term outcomes. There were no significant interactions between baseline PSSS and randomization group, age, gender, race, etiology of heart failure, NYHA Class, or depression score for all-cause mortality or hospitalization (Supplemental Appendix, Table 1). There were also no significant interactions with baseline PSSS and sub-groups for cardiovascular death or heart failure hospitalization. There was, however, a signal for interaction between baseline BTES and baseline Beck Depression Score (p=0.056) and a significant interaction between baseline BTES and randomization arm for cardiovascular death or heart failure hospitalization (p=0.035) (Supplemental Appendix, Table 1). The latter interaction was due to a borderline association between increasing BTES and increased cardiovascular death or heart failure hospitalization for the exercise group (HR 1.25, 95% CI: 0.99, 1.59) that was not seen in the usual care group (HR 0.83, 95% CI: 0.66, 1.06). No other interactions were found for baseline BTES and cardiovascular death or heart failure hospitalization by defined sub-groups.

There was no association between exercise adherence and the primary outcome of all-cause death or hospitalization (Table 4). Compared to full adherence, poor exercise adherence was associated with a higher risk of cardiovascular death or heart failure hospitalization (HR 1.47, 95% CI: 1.02, 2.11, p=0.04). This association remained after controlling for adjustment variables (HR 1.83, 95% CI: 1.15, 2.92, p=0.01).

Table 4.

Cox Proportional Hazard Models for Outcomes by Exercise Adherence

Outcome	Exercise Adherence	Hazard Ratio (95% CI) Unadjusted	P Value	Hazard Ratio (95% CI) Adjusted	P Value
All-cause death/All-cause hospitalization	Diminished vs Full	1.21 (0.85, 1.71)	0.29	1.23 (0.81, 1.88)	0.33
	Poor vs Full	1.15 (0.87, 1.53)	0.32	1.41 (1.00, 1.99)	0.05
Cardiovascular death/Heart Failure hospitalization	Diminished vs Full	1.15 (0.73, 1.81)	0.56	1.48 (0.83, 2.65)	0.18
	Poor vs Full	1.47 (1.02, 2.11)	0.04	1.83 (1.15, 2.92)	0.01

Discussion

Exercise programs, while proven to be safe and effective for heart failure patients, are often poorly followed.^{1, 2} This study examined heart failure patient characteristics associated with poor exercise adherence, and the influence of perceived social support on exercise adherence and long-term outcomes. This study also explored the association of barriers to exercise with exercise adherence and outcomes, and the association of exercise adherence with outcomes. Social support and barriers to exercise correlated with exercise adherence, consistent with our hypothesis; however they were not independently associated with clinical outcomes, though barriers to exercise impacted the effect of exercise training on CV death or HF hospitalization.

Both physical and psychosocial factors were associated with exercise adherence. Patients with poor exercise adherence throughout the study period had worse functional class and exercise capacity at baseline, as measured by peak VO₂ and 6 minute walk distance. Patients with higher perceived social support exercised more than those with lower perceived social support. Similarly, patients with fewer barriers to exercise exercised more than those with more barriers to exercise.

The perceived social support scale and barriers to exercise scale are two tools that can be used to assess patients’ psychosocial health status which may impact adherence to treatment recommendations. Prior studies have shown the KCCQ score correlates with the PSSS, and our study also shows a weak correlation between these two scores as well.¹² KCCQ was also weakly correlated with BTES. Importantly, exercise training has been shown to result in improvement in KCCQ score, and KCCQ score has been shown to be associated with outcomes including functional measures and clinical endpoints.^13–17 However, in our study neither the baseline PSSS nor BTES correlated with change in KCCQ from baseline to 12 months. Furthermore, both the PSSS and BTES correlate modestly with measures of depression. However, neither PSSS nor BTES changed over time. Thus, while exercise training has been shown to alleviate symptoms of depression, it does not appear to change the perception of social support or barriers to exercise.^{6, 18–21} The results of this study, and other studies of psychosocial interventions in heart failure patients, highlight how difficult it is to affect meaningful psychosocial changes in these patients.²²

The HF-ACTION study showed that exercise training for patients with chronic heart failure improved peak VO₂.at 3 months and 12 months.³ Subsequent analyses demonstrated the increase in peak VO₂ at three months was associated with a lower risk of all-cause mortality or all-cause hospitalization, and cardiovascular mortality or heart failure hospitalization.²³ Our study examined VO₂ at 12 months and found that baseline PSSS and BTES were not associated with a change in peak VO₂ from baseline to 12 months.

Furthermore, the HF-ACTION study showed that exercise training reduced all-cause mortality and hospitalizations in chronic heart failure patients.³ However, even in the clinical trial setting, only one-third of the patients randomized to exercise training were fully adherent to the recommended exercise program. In clinical practice, the rates of adherence with recommended exercise are likely to be even lower, potentially limiting the effectiveness of exercise training for improving clinical outcomes. Although heart failure patients may have physical and cardiopulmonary limitations to exercise, it is important to recognize that psychosocial factors may also limit a patient’s ability or perceived ability to exercise. Regardless of treatment arm, we sought to determine whether psychosocial factors influenced outcomes in heart failure patients. We found that perceived social support was not associated with all-cause death or hospitalization or with cardiovascular death or heart failure hospitalization, consistent with a prior study that did not show an association between PSSS and cardiovascular mortality.²⁴ However, prior studies have shown an association between PSSS and treatment adherence in various conditions, including fluid restriction in renal failure, immunosuppression in organ transplantation, and medication compliance in heart failure.^25–27 This study adds to that literature showing an association between PSSS and exercise adherence in heart failure. While a direct link between perceived social support and outcomes was not observed in this study, higher levels of social support tended to be associated with higher levels of exercise adherence.

In examining the novel BTES, we found a trend toward an association between higher baseline barriers score and increased all-cause death or hospitalization though this association did not persist after multivariable adjustment. Thus, barriers to exercise training may capture psychosocial factors associated with worse outcomes but the BTES may not offer independent prognostic information above and beyond recognized prognostic variables. There was a significant interaction between baseline BTES and randomization arm with respect to cardiovascular death or heart failure hospitalization. This was due to a borderline association in the exercise training group between higher baseline BTES and increased cardiovascular death or heart failure hospitalizations, which was not seen in the usual care group. This suggests that exercise training may be more beneficial for those with lower BTES, which may be related to exercise adherence, as those with fewer barriers to exercise had higher levels of exercise adherence. Continued study is warranted to further explore and validate the BTES. Moreover, because neither PSSS nor BTES changed during the course of this study, future studies should assess whether exercise programs that incorporate personalized strategies to reduce patient perceived barriers to exercise training may translate into improved psychosocial factors, exercise adherence, and clinical outcomes.

We showed that exercise adherence, as measured as minutes of exercise per week, is independently associated with cardiovascular death or heart failure hospitalization. These findings add to results from a prior HF-ACTION analysis which included exercise intensity along with exercise time, and showed that moderate exercise was associated with decreased risk of cardiovascular death or heart failure hospitalization.²⁸ Since exercise volume has been associated with outcomes, the assessment of and intervention on support structures for heart failure patients may offer a way to facilitate improved health behaviors and potentially improve outcomes.²⁸ These hypotheses warrant future prospective study.

Although both physical limitations and lack of psychosocial support may limit a patient’s ability to adhere to an exercise training program, it is important to note that even low levels of exercise have been shown to be beneficial compared to no exercise at all.^{29, 30} As cardiac rehabilitation referrals for heart failure patients increase, clinicians must recognize the factors that limit a patient’s ability to participate and thus realize the full benefits of this therapy. These results highlight the need to assess for social support and barriers to exercise, and recognize when these social factors may limit the patient’s participation in an exercise training program. By identifying patients at high risk for exercise non-adherence, providers can address these limitations in an effort to increase adherence.

Our study has several limitations. First, there is a selection bias inherent in clinical trials. HF-ACTION only enrolled patients who were deemed able to participate in an exercise training program; therefore, patients with very poor social support or seemingly insurmountable barriers to exercise could not be enrolled, limiting the more extreme examples of inability to exercise due to psychosocial factors. Furthermore, men and younger patients were overrepresented compared to the general heart failure population, potentially limiting generalizability. Second, while the PSSS is a validated tool to measure perceived social support, the BTES is a new assessment tool developed for the HF-ACTION protocol and has not been used or validated in other studies. Third, we did not adjust for multiplicity of statistical testing, thus these data should be viewed as exploratory.

In conclusion, psychosocial limitations adversely affect adherence to an exercise program in patients with heart failure, but social support and barriers to exercise were not independently associated with clinical outcomes in HF-ACTION. However, a higher barriers to exercise score reduced the exercise training effect on the outcome of CV death or HF hospitalization. Because exercise training has been shown to improve outcomes in this population, increasing exercise adherence is an important goal in the care of heart failure patients. Assessment of patient-reported social support and perceived barriers may facilitate individualized approaches to implement and sustain exercise training therapy in clinical practice.

Clinical Perspective.

As cardiac rehabilitation referrals for heart failure patients increase, clinicians must recognize the factors that limit a patient’s ability to participate and thus benefit from this therapy. In addition to physical limitations, psychosocial factors may also influence adherence to exercise. To date, there are limited data characterizing the relationship of psychosocial factors to exercise adherence in heart failure patients. Using data from HF-ACTION, we described characteristics of patients who had poor adherence to exercise training and examined the influence of social support and barriers to exercise on adherence. We found that social support and barriers to exercise correlated with exercise adherence. For CV death or HF hospitalization, exercise training had a greater impact on patients with fewer barriers to exercise. Given that exercise training improves outcomes in heart failure patients, assessment of perceived barriers may facilitate individualized approaches to implement exercise training therapy in clinical practice.