The Role of Exercise Self-Efficacy in Exercise Participation Among Women With Persistent Fatigue After Breast Cancer: A Mixed-Methods Study

Abstract

Objective

Survivors of breast cancer with persistent cancer-related fatigue (CRF) report less exercise participation compared with survivors of breast cancer without CRF. Although CRF predicts other domains of self-efficacy among survivors, the effect of CRF on exercise self-efficacy (ESE)—an important predictor of exercise participation—has not been quantified. This study examined the relationship between CRF, ESE, and exercise participation and explored the lived experience of engaging in exercise among survivors of breast cancer with persistent CRF.

Methods

Fifty-eight survivors of breast cancer (3.7 [SD = 2.4] years after primary treatment) self-reported CRF, ESE, and exercise participation (hours of moderate-intensity exercise per week). Regression and mediation analyses were conducted. Survivors who reported clinically significant CRF and weekly exercise were purposively sampled for 1-on-1 interviews (N = 11). Thematic analysis was performed across participants and within higher versus lower ESE subsets.

Results

Greater CRF predicted lower ESE (β = −0.32) and less exercise participation (β = −0.08). ESE mediated the relationship between CRF and exercise participation (β = −0.05, 95% CI = −0.09 to −0.02). Qualitative data showed that survivors of breast cancer with higher ESE perceived exercise as a strategy to manage fatigue, described self-motivation and commitment to exercise, and had multiple sources of support. In contrast, survivors with lower ESE described less initiative to manage fatigue through exercise, greater difficulty staying committed to exercise, and less support.

Conclusions

Survivors of breast cancer with persistent CRF may experience decreased ESE, which negatively influences exercise participation. Clinicians should screen for or discuss confidence as it relates to exercise and consider tailoring standardized exercise recommendations for this population to optimize ESE. This may facilitate more sustainable exercise participation and improve outcomes.

Impact

This study highlights the behavioral underpinnings of CRF as a barrier to exercise. Individualized exercise tailored to optimize ESE may facilitate sustainable exercise participation among survivors of breast cancer with CRF. Strategies for clinicians to address ESE are described and future research is suggested.

Lay Summary

Women with fatigue after breast cancer treatment may have lower confidence about their ability to engage in exercise. Individually tailoring exercise to build confidence as it relates to exercise may result in more consistent exercise and better health-related outcomes.

Introduction

Cancer-related fatigue (CRF), defined as “a sense of physical, emotional, and/or cognitive tiredness or exhaustion that is not proportional to recent activity and interferes with usual functioning,”¹ affects up to 85% of survivors of cancer during active treatment.² For most survivors, CRF declines in the months posttreatment, but up to one-third of survivors of breast cancer experience persistent CRF, which can last for years beyond treatment.^3,4 Compared with survivors of breast cancer without CRF, those with persistent CRF demonstrate worse performance on functional tests, report poorer physical function, and experience twice the rate of falls.⁵ Systematic reviews and meta-analyses have established exercise as an effective intervention to reduce CRF^6–9; however, exercise levels among survivors of breast cancer often decline following diagnosis and are significantly lower among survivors with persistent CRF compared with those without.^5,10–12 Low exercise levels may have profound consequences, leading to further deconditioning, poorer outcomes, and increased risk of cancer recurrence.^7,13

Exercise self-efficacy (ESE), or the belief that one can successfully engage in exercise,^14,15 is an important predictor of initiating and adhering to exercise behaviors among survivors of breast cancer.¹⁶ Although CRF predicts several domains of self-efficacy among survivors such as self-care, fatigue self-management, and perceived ability to work,^17–19 the effect of CRF on ESE has not been quantified. Considering the numerous barriers to exercise reported by survivors with CRF,^20,21 as well as emerging evidence of post-exertional impairments (eg, loss of muscle force/power, changes in memory/concentration) associated with persistent CRF,^22,23 ESE may be reduced in this population. Because ESE is a barrier to and predictor of exercise, the relationship between CRF and ESE may influence exercise participation among survivors of breast cancer with persistent CRF and therefore warrants investigation.

A previously published model, which suggested that increased physical activity improves ESE and, in turn, reduces CRF,²⁴ may have limited clinical utility due to its postulation of ESE solely as an outcome of physical activity. This model does not consider the role of ESE as a barrier to exercise among survivors of breast cancer^16,25–28 or the potential impact of CRF on ESE. Without these considerations, promotion of exercise among survivors with persistent CRF may have limited success due to poor adherence—a known moderator of the effect of exercise on CRF (ie, high adherence yields a large effect, low adherence yields a negligible effect).⁶ For this reason, identification of modifiable factors that may optimize exercise adherence among survivors of breast cancer living with CRF is imperative. Furthermore, greater insight into the experience of engaging in exercise while living with CRF, including the role of behavioral factors such as ESE, is necessary to support the promotion of exercise in this population using a person-centered approach. This approach may lead to more optimal exercise adherence, participation, and improved outcomes among survivors with CRF.

Therefore, the purposes of this study were twofold: (1) to quantify the relationship between persistent CRF, ESE, and exercise participation among survivors of breast cancer; and (2) to describe the lived experience of exercise participation in survivors with persistent CRF. We hypothesized that higher levels of CRF would be associated with lower ESE and less exercise participation, that ESE would mediate the relationship between CRF and exercise participation, and that qualitative data would enrich our understanding of the role of ESE in exercise participation among survivors of breast cancer with persistent CRF.

Methods

Study Design

A sequential, mixed-methods design was employed to, first, quantify the relationship between CRF, ESE, and exercise participation, and, second, to describe the lived experience of exercise participation for survivors of breast cancer with persistent CRF. We selected this design for the purposes of illustration and completeness; that is, qualitative findings would illuminate and provide a more comprehensive account of the relationship between CRF, ESE, and exercise participation among survivors of breast cancer.²⁹ All study procedures were approved by the institutional review board (IRB).

Participants and Recruitment

Participants were recruited from a high-volume breast cancer center at a large urban hospital in the Northeast region of the United States. Potential participants were identified via electronic medical record review if they met all the following criteria: female, aged 18 to 85 years, and had completed chemotherapy with or without radiation therapy for stage I-III breast cancer at least 12 months before study participation (ongoing anti-hormone therapies permitted). To minimize confounding of the effect of CRF, potential participants were excluded if they reported presence of any of the following conditions that typically include a component of fatigue: precancer diagnosis of fibromyalgia, posttreatment Lyme Disease syndrome, chronic fatigue syndrome, hypothyroidism without replacement therapy, anemia with hemoglobin levels <12 g/dL, or positive screens for major depression or an anxiety disorder based on responses on the Patient Health Questionniare-4.³⁰

Eligible participants who agreed to participate provided consent through an IRB-approved electronic consent form via a browser-based Research Electronic Data Capture platform.³¹

Sample Size Calculation

We calculated a priori sample size estimates using 2 models. Because the relationship between CRF and ESE has not previously been quantified, to our knowledge, we estimated sample size for a range of plausible effect sizes using G^*Power 3.1,³² assuming a linear regression model with ESE as the primary outcome and CRF as the predictor, power set at 0.8, and a conventional statistical significance (α = .05). To detect a medium effect with 3 covariates included in the model, the estimated sample size was 85. For our mediation hypothesis, a Monte Carlo power analysis simulation was performed, which indicated that, with power set at 0.8, a sample of 85 would be required to detect small to medium correlations.³³ Thus, we aimed to recruit 85 participants to adequately power all quantitative analyses. Interim analyses were planned to follow initial recruitment to determine ongoing recruitment needs and strategies.³⁴

For the qualitative aim, a minimum sample size of 6 is recommended to explore the lived experience of a given phenomenon (ie, for a phenomenological study),^35–37 though the adequacy of the sample size is ultimately determined by data saturation (ie, similar data emerge/are repeated by multiple participants).^35,36 Thus, a priori, we estimated recruiting 6 to 20 interview participants, but sample size was ultimately determined by data saturation.^38,39

Data Collection

Participants completed all quantitative study questionnaires online via Research Electronic Data Capture.³¹ Participants who reported current presence of clinically significant CRF and any amount of weekly participation in moderate-intensity exercise-related activities were purposively sampled as key informants to participate in a qualitative interview.

Quantitative Measures

CRF was measured via self-report using the 36-Item Short Form Health Survey Vitality Subscale (SF-36-VS).⁴⁰ The SF-36-VS is commonly used to assess fatigue in cancer populations^41–44 and has been validated in studies of both diseased and general populations.^45,46 The 4-item VS assesses fatigue during the past 4 weeks, and responses are scored on a 0 to 100 scale, with higher scores corresponding to less fatigue. Scores ≤50 indicate clinically significant fatigue.⁴⁶ The SF-36-VS was reverse-scored to improve interpretability (see Data Analysis).

ESE was measured using the Self-Efficacy for Exercise scale.⁴⁷ Among older adults, this scale has been shown to have excellent internal consistency (Cronbach α = .92), and scores significantly predict exercise activity.⁴⁷ The Self-Efficacy for Exercise scale consists of 9 potential exercise barriers and asks the individual to rate their confidence from 0 (not confident) to 10 (very confident) that they could exercise for 20 minutes, 3 times per week, given each barrier (maximum score of 90 indicates highest ESE). Although this scale has not been validated in samples of survivors with CRF, a validated measure of ESE in this population does not currently exist, and we felt that the 9 barriers appropriately represent scenarios that survivors of breast cancer may regularly encounter.

Exercise participation was assessed using the Community Health Activity Model Program for Seniors questionnaire.⁴⁸ This questionnaire is able to discriminate between inactive, somewhat active, and active groups (P < .001) by asking participants to report weekly frequency and time spent in sedentary and low-, moderate-, and vigorous-intensity activities during the last 4 weeks.⁴⁸ We operationally defined exercise participation as hours of moderate-intensity exercise per week given the current recommendation of 90 to 150 min/wk of at least moderate-intensity exercise to effectively reduce CRF.⁴⁹

Demographic and Clinical Data

A standard demographic survey was used to collect data, including age, ethnicity and racial background, education, employment, and income. Cancer treatment details, including time since treatment, were extracted from the medical record. The presence of chronic medical conditions was measured using the Functional Comorbidity Index, an 18-item self-report measure designed to examine the impact of comorbidities on physical function.⁵⁰

Covariates

Given the prevalence of chemotherapy-induced peripheral neuropathy (CIPN) and pain among survivors of breast cancer with CRF and their potential impact on physical function and exercise participation,⁵ CIPN severity and average pain were measured as covariates.

CIPN (severity of sensory symptoms in the feet) was measured using the sensory neuropathy items on the Functional Assessment of Cancer Therapy/Gynecologic Oncology Group Neurotoxicity (FACT/GOG-Ntx-4 v4).^51,52 The FACT/GOG-Ntx-4 v4 is highly recommended by the American Physical Therapy Association’s (APTA) Academy of Oncologic Physical Therapy (APTA Oncology) task force on breast cancer outcomes due to its good internal consistency (Cronbach α = 0.62–0.90) and moderate-to-strong convergent validity with objective measures of peripheral neuropathy (r = 0.39–0.64).^51,52 Symptom severity in the feet is scored on a 0 to 8 range, with higher scores corresponding to more severe symptoms.

Average pain was measured using the Brief Pain Inventory-Short Form (BPI-SF), recommended by the APTA Oncology Task Force on Cancer as a clinical measure of pain.⁵³ The BPI-SF has been extensively used in cancer pain research,^54–58 and in a sample of survivors of breast cancer, the BPI-SF had excellent internal consistency (Cronbach α = .89).⁵⁹ Construct validity has been reported to be high for pain severity (r = 0.70–0.91), and reliability for average pain rating is high (r = 0.78).^60,61 Average pain is scored on a 0 to 10 range, with higher scores corresponding to worse pain.

Qualitative Interview

The principal investigator (S.W.) conducted 1-on-1 semi-structured interviews via secure online platform. The research team developed an interview guide using a descriptive phenomenological method,⁶² and S.W. pilot tested this guide through an interview with a male survivor of non-breast cancer who otherwise met the inclusion criteria. The guide was developed to elicit discussion about (1) the personal experience and management of symptoms of CRF in the context of performing regular exercise; (2) how persistent CRF shapes survivors’ ability to perform regular exercise; and (3) how survivors with persistent CRF sustain a regular exercise routine while maintaining life roles. The interview guide was iteratively revised based on quantitative data collection and initial interviews. S.W. recorded field notes during and after each interview to aid in data analysis. Audio from all interviews was digitally recorded and transcribed verbatim before analysis to ensure accuracy of each participant’s experience.

Role of the Funding Source

The funders played no role in the design, conduct, or reporting of this study.

Data Analysis

Quantitative analyses were conducted in R.⁶³ Frequency distributions, and descriptive statistics were examined for all study variables. All data met assumptions and therefore did not require transformation. To improve interpretability of self-reported CRF, SF-36-VS scores were centered and reversed so that higher values represent higher fatigue levels. To improve interpretability and ability to compare ESE scores, percentages were calculated by dividing participants’ self-efficacy for exercise scale scores by the max score of 90. Correlation coefficients were examined for all independent variables to assess for multicollinearity. To determine whether CRF predicts lower ESE and exercise participation, we separately regressed ESE scores and hours of moderate-intensity exercise per week on CRF level controlling for covariates. Covariates were included if at least moderately correlated with CRF and the respective outcome or if considered potentially influential based on prior research.⁵ Regression assumptions and diagnostics were assessed for all final models. The final model results were not significantly different with multivariate outliers and influential cases removed, so these cases were included to preserve sample size.

Our conceptual mediation approach followed the criteria outlined by Baron and Kenney and MacArthur.⁶⁴ To determine if ESE mediates the relationship between CRF and exercise participation, a bootstrap test of mediation was performed, controlling for covariates. The bootstrap technique of calculating standard errors of indirect effects allows for non-normal distributions and reduces type II error compared with the traditional Sobel test of mediation.^65,66 Indirect effects were calculated via repeated random sampling (N = 10,000) with replacement.⁶⁷ Mediation would be determined to be present if the resultant CI of the indirect effect did not contain zero.⁶⁶ Post-hoc power analyses were conducted for regression and mediation results.

Qualitative Interviews

An iterative, multi-step inductive process was used to examine the data, compare codes, challenge interpretations, and inductively develop themes while ensuring the credibility and rigor of qualitative data analysis.^38,39 This process was first performed across all participants and then within subsets of higher (>70%) versus lower (<70%) self-reported ESE scores to explore potential differences in experience by ESE level. First, 2 authors (S.W. and M.R.F.) independently read transcripts to achieve dataset immersion, gain a broad understanding of the text, and identify key quotations. Second, these authors discussed key codes related to the research questions and iteratively developed a coding scheme to describe and summarize relevant content. Coded quotations were then combined into a single file to formulate meaning. Initial themes identified through clustering of key coded quotations were evaluated by the other researchers with consensus reached and major themes further developed through active dialogue. The final themes retained were those that generated consensus from all team members and were supported by the data.

Results

Participants

Interim analyses revealed a larger effect size between CRF and ESE (f² = .45) than the anticipated effect size used for a priori sample size estimates as well as adequately powered mediation analyses. Therefore, recruitment was ended at 58 participants. Table 1 provides baseline data for the 58 survivors of breast cancer. The average age of participants was 54.9 (11.2) years, and the average time since treatment was 3.7 (2.4) years. All participants had been treated with chemotherapy, and 82.8% of the sample also received radiation therapy; 63.8% were still taking an antiestrogen at the time of study involvement. The average fatigue level (after reversal) was 41.9 (21.7), with scores ranging from 0 to 93.75. Most participants (69%) reported either 0 or 1 comorbidity.

Table 1.

Participant Demographic and Clinical Characteristics^a

Study Variables	Full Sample (n = 58)	Interview Subsample (n = 11)
		ESE > 70% (n = 5)	ESE < 70% (n = 6)
Fatigue levelb, mean (SD)	41.9 (21.7)	60 (12.2)	64.6 (14.1)
Exercise self-efficacy
Raw score, mean (SD)	62.7 (23.4)	78.4 (11.8)	37.3 (11.7)
Percentage, mean (SD)	69.6 (26)	87.1 (13.1)	41.5 (13.0)
Exercise participation
Hours of moderate-intensity exercise per week, mean (SD)	6.2 (5.3)	6.7 (3.6)	4.3 (2.8)
Sensory symptoms of CIPN
CIPN +	53.4	60	50
Severity if +, mean (SD)	3.3 (2.2)	2.7 (2.1)	4.3 (3.1)
Pain
Average, mean (SD)	2.5 (2.3)	2.6 (2.5)	4.2 (2.5)
Characteristics
Age, mean (SD), y	54.9 (11.2)	48.2 (10.1)	50 (8.1)
Time since treatment, mean (SD), y	3.7 (2.4)	3.2 (1.5)	2.1 (1.2)
Treatment type
Chemotherapy only	17.2	0	0
Chemotherapy and radiation	82.8	100	100
Antiestrogens
None	31	20	60
Tamoxifen	17.2	20	20
Aromatase inhibitors	51.7	60	40
Still taking antiestrogens	63.8	80	60
No. of comorbid conditions, mean (SD)	1.0 (1.1)	1.8 (0.8)	0.7 (0.8)

^a Data are reported as number (percentage) of participants unless otherwise indicated. CIPN = chemotherapy-induced peripheral neuropathy; ESE = exercise self-efficacy.

^b As measured on the SF-36 Vitality Scale (0–100, scores reversed so higher scores indicate higher fatigue).

Effect of CRF on ESE and Exercise Participation

Regression and mediation results are presented in Table 2.

Table 2.

Regression and Mediation Results^a

Linear Regression Results	Exercise Self-Efficacyb β (95% CI)	Exercise Participationc β (95% CI)
Intercept	65.28d (50.71 to 79.85)	8.75 (1.75 to 15.75)e
Fatigue	−0.32e (−0.58 to −0.06)	−0.08 (−0.16 to −0.004)e
Exercise participation	2.00d (0.92 to 3.08)	—
Time since treatment, y	−0.48 (−2.98 to 2.02)	0.92 (0.34 to 1.5)f
Age, y	—	−0.07 (−0.19 to 0.05)
CIPN severity	—	0.12 (−0.52 to 0.76)
Average pain	—	0.44 (−0.25 to 1.13)
Adjusted R2	0.31	0.19
F statistic	9.60d	3.68f
Mediation Results	Estimate	95% CI
Indirect effect (CRF-ESE-Exercise participation)	−0.05d	(−0.09 to −0.02)
Direct effect (CRF-Exercise participation)	−0.03	(−0.08 to 0.02)
Total effect	−0.08f	(−0.14 to −0.03)
Proportion mediatedg	0.60f	(0.26 to 1.44)

^a Values are standardized β coefficients. β = estimate; CIPN = chemotherapy-induced peripheral neuropathy; CRF = cancer-related fatigue; ESE = exercise self-efficacy.

^b Estimates are based on self-efficacy for exercise scale scores.

^c Estimates are based on hours of moderate-intensity exercise per week.

^d P < .001.

^e P < .05.

^f P < .01.

^g Proportion of CRF’s effect on exercise participation mediated by ESE.

Effect of CRF on ESE

In our sample, the Self-Efficacy for Exercise scale demonstrated excellent internal consistency (Cronbach α = 0.95). Controlling for exercise participation and time since treatment, level of CRF was a statistically significant predictor of ESE (P = .02), with greater CRF predicting lower ESE. Exercise participation was also a significant predictor of ESE (P < .001), with greater hours of exercise predicting higher ESE. Time since treatment was not a significant predictor of ESE. Based on this model, a survivor of breast cancer with self-reported CRF 1 SD above average was predicted to report approximately 8% lower ESE than a survivor of breast cancer with average CRF. Post-hoc power analysis revealed a power of .99 for the final regression model.

Effect of CRF on Exercise Participation

Controlling for time since treatment, age, CIPN severity, and average pain, level of CRF was a statistically significant predictor of exercise participation (P = .04), with greater CRF predicting fewer hours spent engaging in moderate-intensity exercise per week. Time since treatment was also a statistically significant predictor of exercise participation (P = .003), with greater time since treatment predicting more hours of moderate-intensity exercise. Age, CIPN severity, and average pain were not significant predictors of exercise participation. Based on this model, a survivor of breast cancer with self-reported CRF 1 SD above average was predicted to perform approximately 100 fewer minutes of moderate-intensity exercise per week than a survivor with average CRF. Post-hoc power analysis revealed a power of .81 for the final regression model.

ESE as a Mediator of the Effect of CRF on Exercise Participation

When regressed on exercise participation, the interaction between CRF and ESE was not statistically significant (P = .77), whereas the main effect of ESE was P = .05, satisfying the criteria of the MacArthur approach and suggesting evidence of mediation.⁶⁴ Bootstrapping procedures, controlling for covariates, revealed a statistically significant indirect effect (β = −0.05, 95% CI = −0.09 to −0.02, P < .001), confirming that ESE mediated the relationship between CRF and exercise participation (Figure). Post-hoc power analysis revealed a power of mediation of .94.

Figure.

Exercise self-efficacy’s mediation of the effect of cancer-related fatigue on exercise participation. Mediation model with standardized β coefficients. ADE = average direct effects; Total effect = direct and indirect effect of cancer-related fatigue on exercise participation.

Qualitative Results

Participants

Of our 58 participants, 34% (N = 20) qualified for participation in the qualitative interview. Data saturation was achieved with 11 interviews. Characteristics for these 11 participants are provided in Table 1. Per self-report, all qualitative participants met minimum recommended exercise levels, with hours of moderate-intensity exercise per week ranging from 1.75 to 9.5. Six interviewees reported ESE < 70% and 5 reported ESE > 70%. Compared with women with higher ESE, those with lower ESE were closer to completion of treatment and reported slightly worse CRF, pain, and symptoms of CIPN as well as fewer hours of moderate-intensity exercise per week.

Role of ESE in Sustaining Exercise

Qualitative data provided insight into how survivors of breast cancer with CRF engaged in exercise in everyday life and the role of ESE in sustaining regular exercise. Survivors with higher ESE perceived exercise as a strategy to manage fatigue, described self-motivation and commitment to exercise, and had multiple sources of support. In contrast, survivors with lower ESE described less initiative to manage fatigue through exercise, greater difficulty staying committed to exercise, and less support.

Perceiving Exercise as a Strategy to Manage Fatigue

Women with higher ESE expressed a clear intent to use exercise to manage their fatigue: “I don’t think there’s a lot of gray … I know that either I do it and feel better, or I don’t, and I won’t.” (Participant 3) This clarity stemmed from both the transient short-term benefits of exercise as well as the longer-term psychological benefits of sustained exercise that women with higher ESE perceived. “[After exercise I feel] more alert, ready to face my day… I’m in charge of my own self, my life, my schedule, my health focus.” (Participant 4) Undeterred by ongoing fatigue, survivors of breast cancer with higher ESE described a sense of control through regular exercise: “I’m just getting to know [my fatigue] better and know how to take care of it better.” (Participant 2)

Despite regular exercise participation, survivors of breast cancer with lower ESE described less initiative to manage their fatigue through exercise as one woman described: “I know if I exercise, I’ll have energy … I toss that around in my head, but I don’t do anything about it.” (Participant 9) Another woman echoed, “I think a lot about getting back into the workout regimen, but it’s kind of like … when am I gonna have enough energy?” (Participant 7) Although women with lower ESE did perceive transient benefits of exercise, they expressed more uncertainty, stress, or even guilt related to sustaining a regular exercise routine: “…every day that I don’t exercise I’m sort of beating myself up like I should have done it, and I feel stressed that I didn’t do it.” (Participant 9) Women with lower ESE emphasized the impact of ongoing fatigue and perceived uncertainty of how fatigue may or may not change. One participant stated, “I don’t know if it’s just survivorship and this is, you know, what my life is like now. I don’t know.” (Participant 11)

Staying Motivated and Committed to Exercise

Women with higher ESE appeared to sustain their exercise routines through self-motivation and commitment, as Participant 4 remarked: “I have a lot to live for, and I wanna get every day that I can get. So, by hook or by crook … I’ll just keep on doing it, you know?” Motivation and commitment were bolstered among those with higher ESE by perceived effectiveness of self-selected exercise: “… it (high-intensity interval training) works for my body. It works for my brain.” (Participant 2)

Women with lower ESE expressed greater difficulty staying committed to regular exercise and relied on motivation from the external distractions that exercise provides. Participant 11 voiced the challenge in self-motivating: “To just get over that hump of like just doing it … you know, it’s just … easier to not.” Participant 8 described her reliance on the accountability and distraction provided by group exercise classes: “My body is tired. I don’t wanna do this, but I signed up for it … It’s just a way for me to get out of my normal day-to-day world.…”

Having Multiple Sources of Support

Finally, a difference emerged in how women with higher or lower ESE described their support systems in the experience of engaging in exercise. Women with higher ESE consistently described the importance of their support systems, detailing the influence of their family as well as their medical team. Participant 5 described that her oncologist was “a big promoter of exercise,” and Participant 3 stated that her oncology nurses “encouraged exercise … they basically said, ‘you’re gonna be fatigued, but if you can walk, it will make you feel better. It will reduce the fatigue.’” Participant 1 spoke about the insight provided by her physical therapist: “…You can take those type of things (body awareness, exercises) and incorporate them in your daily activities.”

Women with lower ESE, however, described feeling “untethered” after completion of cancer treatment, feeling like, “you fall off a cliff” and that “they kind of cut you loose.” Participant 9 described her attempt to engage in exercise despite the lack of support: “I thought I could just do it through my own willpower like I always used to.” Participant 10, who reported low ESE but initiated physical therapy before her interview, described her experience: “I just felt like there was so much that was left up to you after the fact, you know, and you’re kind of left on your own to work it all out … [Physical therapy] gave me things to do that make you feel like you’re in control again … that makes you feel better.”

Discussion

Our results build on prior evidence detailing the complex biobehavioral challenges related to exercise among survivors with CRF by quantifying the impact of CRF on ESE and highlighting the behavioral underpinnings of CRF as a barrier to exercise among survivors of breast cancer. Our mediation model situates ESE as a proximal factor to exercise participation, consistent with evidence that ESE is a predictor of initiating and adhering to exercise behaviors among survivors of breast cancer.^16,25–28 Identification of ESE as a mediator between CRF and exercise participation suggests that interventions to increase ESE may facilitate increased exercise participation among survivors with CRF. Our qualitative data support our quantitative results by illustrating how ESE influenced efforts made by survivors to sustain regular exercise while living with CRF and provide insight into how exercise recommendations may be tailored to promote sustainable exercise behaviors in this population.

Changes in ESE can be influenced by clinicians through exercise interventions that include opportunities for exercise mastery, vicarious experience, verbal persuasion, and education focused on physiological state.^14,68–70 Exercise mastery, that is, perceived competence related to exercise, may be targeted by clinicians through focus on successful completion of exercises.^69,71 This may require breaking down or regressing exercise as recommended by current guidelines to allow for smaller but sooner success that can be built on. Vicarious experience enables a patient to observe someone else completing similar exercise, as through group exercise^69,72—a known facilitator of exercise among survivors with CRF.²⁰ Verbal persuasion occurs when a patient is encouraged or persuaded to complete an exercise; techniques include motivational interviewing, coaching, and education regarding the benefits of exercise.^69,72–74 Teaching fatigue self-management through the use of exercise may have a particularly important role considering how women with higher ESE in our sample more readily employed this strategy compared with participants with lower ESE. This approach could further enhance outcomes because increased self-efficacy for self-managing CRF has been shown to improve physical functional status among survivors with CRF.¹⁸

Finally, education regarding physiological state or the body’s responses to exercise has been shown to positively affect ESE.^69,75 This final technique can be addressed through various educational approaches, including teaching heart rate monitoring, rate of perceived exertion, and recognition of normal versus abnormal exercise responses. This education may be especially relevant for survivors of breast cancer with persistent CRF who may be deconditioned or living with comorbidities that may alter normal responses to exercise. Women with higher ESE in our sample described greater self-monitoring of exercise and its impact on their physical and mental well-being. Given the importance of sustaining an adequate dosage of exercise to reduce CRF,⁴⁹ the ability to self-monitor exercise may be critical for survivors of breast cancer with CRF. This framework to improve ESE through individualized exercise prescription, delivery, education, and support may facilitate more sustainable exercise participation among survivors with persistent CRF, which may lead to improved outcomes. Future research should assess how behaviorally informed interventions may promote greater exercise adherence and participation in this population. This is a necessary step to advance current guidelines and move toward personalized exercise prescription for survivors of cancer.

Identifying survivors of breast cancer with CRF and low ESE who may benefit from such approaches is critical. Future research should validate existing ESE measures or develop novel measures of ESE among survivors. Clinicians should screen for or discuss confidence as it relates to exercise and/or observe traits that may indicate low ESE. Patients with low ESE may shy away from exercise they perceive as a threat, lessen efforts toward exercise if difficulties arise, dwell on shortcomings and failures, or lack commitment to goals or personal desires related to exercise.⁷⁰ If a patient reports low ESE or demonstrates these behaviors, individualized adjustments to that patient’s exercise prescription can be made utilizing the strategies outlined above. A key difference in the experience of women with higher ESE was the role of a support system that often included members of their medical team, highlighting the important role of clinicians in providing support beyond provision of standardized exercise recommendations.

Limitations

The results of this study must be interpreted in the context of its limitations, including use of an unvalidated outcome measure (Self-Efficacy for Exercise scale), modest sample size, and cross-sectional study design. Although the Self-Efficacy for Exercise scale has not been validated among survivors with CRF, internal consistency was excellent in our sample. Although a clear inverse relationship was observed between CRF and ESE, clinical implications of our quantitative results are limited by the lack of MCID or established cutoff for this scale. Furthermore, our use of 70% as a delineator for higher versus lower ESE was chosen subjectively, which limits our ability to draw qualitative conclusions comparing these groups. Nevertheless, differences in experience between these groups emerged, and data saturation was achieved. Despite our modest sample size, observed effect sizes were large enough to achieve adequate power to draw statistical conclusions. Still, further investigation of these relationships is warranted, especially because the cross-sectional design of our study limits our ability to establish temporal cause-and-effect relationships between CRF, ESE, and exercise participation. Large longitudinal studies assessing CRF, ESE, and exercise participation before, during, and after treatment for cancer would greatly enhance our understanding of CRF’s effect on ESE and exercise behaviors among survivors of cancer.

Author Contributions

Concept/idea/research design: S. Wechsler, M.R. Fu, L. Wood Magee

Writing: S. Wechsler, M.R. Fu, K. Lyons, K.C. Wood, L. Wood Magee

Data collection: S. Wechsler, M.R. Fu, L. Wood Magee

Data analysis: S. Wechsler, M.R. Fu, K. Lyons, L. Wood Magee

Project Management: S. Wechsler

Fund Procurement: S. Wechsler

Consultation: S. Wechsler, M.R. Fu, K. Lyons, K.C. Wood, L.W. Magee

Funding

This work was supported by grant R21AG055149 from the National Institute on Aging and a Postdoctoral Research Grant from Boston College.

Ethics Approval

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Dana Farber Cancer Institute Office for Human Research Studies and risk level was deemed “not greater than minimal risk” (March 22, 2021; IRB Protocol #: 18–339).

Disclosures

The authors completed the ICMJE Form for Disclosure of Potential Conflicts of Interest and reported no conflicts of interest.