Making Interventions More Effective: Changes in Theoretical Constructs During Delivery of a Peer‐Led Physical Activity Program for Breast Cancer Survivors

ABSTRACT

Objective

Theory‐based physical activity (PA) interventions improve moderate to vigorous physical activity (MVPA) among breast cancer survivors, yet limited research has examined when psychosocial changes occur during intervention delivery. Understanding temporal changes in theoretical constructs during PA adoption may enhance intervention delivery and outcomes. The purpose of this study is to examine longitudinal changes in self‐efficacy, self‐regulation and affect (intrapersonal constructs from Social Cognitive Theory) and social support (interpersonal construct) during a 12‐week theory‐based PA program (webMFT) and their association with participants' active minutes (measured by Fitbit) over 12 weeks and intervention engagement (receipt of 12 weekly peer coaching calls).

Methods

In a randomized controlled trial with 61 participants, 30 were randomized to receive webMFT (mean age = 58.1 [SD = 8.55], 1.40 years post‐diagnosis [SD = 0.5], 80% Stage 0–1 cancer, 80% White) delivered by 10 peer coaches. Only the webMFT participants were assessed; the comparison group (MVPA Tracking) was not included in the analyses. webMFT participants received a Fitbit tracker and weekly PA coaching calls tailored to Fitbit PA data. At baseline, 4, 8 and 12 weeks, they completed questionnaires assessing multidimensional self‐efficacy (task, coping, scheduling), self‐regulation skills, affective response and social support for PA. Mixed effects regression models examined changes in constructs and their association with Fitbit active minutes/week and intervention engagement.

Results

Increases in total self‐efficacy, scheduling self‐efficacy, positive affect and social support from family and friends were significantly associated with participants' PA over 12 weeks. There were no significant associations between changes in self‐regulation skills, task or coping self‐efficacy with PA. Improvements in total self‐efficacy, scheduling self‐efficacy, self‐regulation, affective response and family social support were positively associated with intervention engagement.

Conclusions

Participants' self‐efficacy, specifically, for scheduling PA, positive affect and social support for PA adoption and intervention engagement merit future examination.

1. Background

Physical activity (PA) has shown to benefit cancer survivors' recovery and physical functioning leading to national recommendations for survivors to achieve at least 150 minutes/week of moderate‐to‐vigorous PA (MVPA) [1, 2]. Numerous randomized controlled trials (RCTs), some based on theories of behavior change to promote PA among cancer survivors, have been developed and tested [3]. Researchers have subsequently analyzed the effects of the interventions on the underlying theoretical constructs at postintervention and at follow up, with mediational analyses to identify the mechanisms of action [4, 5, 6, 7, 8, 9, 10, 11, 12]. However, the changes that occur during behavior change initiation (i.e., during intervention delivery) have received less attention with a few exceptions [13, 14]. It is important to examine the constructs during the intervention phase so that researchers can identify when changes occur and in which constructs so as to further improve the efficacy of interventions and guide intervention delivery.

We have conducted two theory‐based RCTs that demonstrated the efficacy of a peer‐delivered intervention in increasing breast cancer survivors' MVPA [15, 16]. Subsequently, we adapted the peer‐led intervention to a web platform creating webMFT to increase the scalability of the intervention and tested the intervention in a RCT [17]. The counseling component of our evidence‐based intervention is based on motivational readiness from the Transtheoretical Model (TTM) [18, 19] and Social Cognitive Theory (SCT) [20]. Key constructs from SCT (self‐efficacy, self‐regulation and affect) underlie the counseling that coaches/peers delivered to survivors [15, 16]. Self‐efficacy or the self‐confidence in achieving a specific goal is a central concept in TTM and SCT and is consistently identified as a psychosocial determinant of PA among cancer survivors [21, 22]. We assessed three types of self‐efficacy: task, coping and scheduling self‐efficacy [23] every 4 weeks during the 12‐week intervention phase to examine which type of efficacy changes and at which time‐point so that we could identify the type(s) of self‐efficacy that are associated with behavior change initiation. Task efficacy refers to confidence in performing the behavior/MVPA, coping efficacy refers to the confidence to perform the behavior in the face of challenges, and scheduling efficacy refers to the skill set relevant to scheduling exercise. The three domains are independently sensitive to contextual factors influencing exercise and can offer specific and distinctive routes for intervention strategies.

SCT also posits that behavior is changed through self‐regulation strategies (e.g., goal setting, self‐monitoring, and reinforcement); the coach calls targeted the development of these skills among survivors. We also assessed changes in affective response to PA during the intervention phase because behaviors that enhance positive affect have a greater likelihood of being performed again. There is some evidence that affective experience predicted variations in exercise (within subject) among endometrial cancer survivors [14]. Finally, at the inter‐personal level, we assessed social support for PA from family and friends. Social support for PA has been shown to encourage adoption and maintenance of PA among those who have a similar cancer experience [24, 25]. Assessing the changes in these constructs during the intervention phase can guide interventionists about which constructs to focus their efforts and the timing of these efforts.

Our work extends the literature in several ways by: (a) conducting longitudinal analyses of changes in theoretical constructs (3 intrapersonal variables: self‐efficacy, self‐regulation and affect and one interpersonal variable: social support) during intervention delivery and their associations with behavior change initiation by collecting and analyzing weekly PA data, (b) examining the changes in theoretical constructs and associations with intervention engagement, defined as receipt of coaching calls, and (c) expanding the construct of self‐efficacy which has been previously shown to be a strong predictor of MVPA adoption to include three independent domains (task, coping and scheduling) that may change during the intervention phase and can inform intervention strategies.

The aims of this paper were to assess associations between participants' self‐efficacy, self‐regulation skills, affective response to PA, and social support for PA with weekly participation in PA over 12 weeks, obtained via a Fitbit activity tracker. Our hypothesis was that survivors who report increases in baseline self‐efficacy, increased use of self‐regulatory skills, increases in social support and positively‐valenced affective response to PA will demonstrate greater PA minutes/week than those who do not. Our second goal was to examine webMFT participants' self‐efficacy, self‐regulation skills, affective response to PA, and social support for PA, associations with intervention engagement (receipt of the 12 weekly calls from the peer coaches). Our hypothesis was that survivors who report increases in baseline self‐efficacy, greater use of self‐regulatory skills, and positively‐valenced affective responses to PA will have greater engagement with the intervention than those who do not.

2. Methods

2.1. Study Design

We conducted an RCT wherein 61 breast cancer survivors were randomly assigned to receive either webMFT (n = 30) or MVPA Tracking (n = 31) [26]. Briefly, peer coaches were trained to deliver webMFT to breast cancer survivors. Coaching calls were delivered weekly for 12 weeks using the web platform. MVPA Tracking participants were asked to self‐monitor their PA throughout the 12‐week intervention using the Fitbit. The study received Institutional Review Board approval at the University of South Carolina (Approval #Pro00117242). All study participants provided written informed consent.

2.2. Coaches

Coaches were recruited from four cancer care organizations across the United States. Participating cancer care organizations had a current peer mentoring program within their organization. Coaches were eligible to participate in the study if they were a current peer coach in their organization (at least 1 year) and were willing to participate in training and supervision throughout the study duration.

Overall, 10 coaches completed the online and virtual group training [17, 27] and were paired with a webMFT participant. Coaches followed the guidelines of the study (focusing on TTM and SCT constructs) and were supervised on a bi‐weekly basis by research staff. In addition to bi‐weekly supervision calls, the PI audited intervention calls and thus additional feedback was provided to the coaches to ensure fidelity in intervention delivery.

2.3. Survivors

Breast cancer survivors were recruited from partnering organizations (as described above) in addition to a local oncology practice and tumor registries. Participants were eligible if they were (1) ≥ 21 years old, (2) diagnosed with Stage 0–3 breast cancer (< 5 years), (3) sedentary (< 90 minutes/week of moderate intensity PA) in the past 6 months, (4) able to walk unassisted, and (5) owned a smartphone.

Interested participants completed a phone screen with the Research Assistant; eligible participants signed an electronic informed consent. Upon completion of baseline assessments, participants in both groups received a Fitbit tracker to monitor their activity, a PA goal chart which gradually progressed their activity up to recommended 150 minutes/week of MVPA and PA tipsheets on topics such as safety and exercising in inclement weather.

2.4. Intervention Group (webMFT)

Participants randomized to webMFT were assigned a coach and received weekly calls for 12‐weeks. Participants were encouraged to wear their Fitbit and were sent weekly text reminders to sync their Fitbit data daily to ensure their activity was tracked and viewable to coaches. Coaches viewed participants Fitbit PA (active minutes as defined by Fitbit) prior to the calls and virtually shared (via the web platform) the report with the participant during the call. During the weekly calls coaches reviewed the participant's PA, problem solved barriers to activity, identified benefits and guided PA goals for the upcoming week. The counseling calls were aimed at increasing survivors' self‐efficacy by setting small, achievable MVPA goals each week, helping to problem solve barriers to PA, and providing feedback on their progress toward PA goals. Coaches helped participants to understand the weekly PA report and provided instruction on how to set achievable MVPA goals, how to plan for and exercise in varying weather conditions, and encouraged participants to self‐reward for achieving MVPA goals (i.e., training in self‐regulation skills). Coaches encouraged participants to find ways to make PA a more positive experience by exercising in pleasant surroundings, selecting activities that are enjoyable, such as dance, to enhance positive affect during PA. Finally, during the weekly calls, coaches specifically asked participants about social support in their network and encouraged them to seek support for PA in their home and work environment.

webMFT participants completed assessments (described below) at baseline, 4‐, 8‐, and 12‐weeks.

3. Measures

3.1. Background Variables

(1) Baseline participant demographics (age, marital status, race and ethnicity, education, etc.) were obtained via self‐report. (2) Patients' weight, height, disease and treatment variables (date of cancer diagnosis, cancer stage when applicable, treatments) were obtained from the participants' medical chart (after obtaining written consent) at baseline.

3.1.1. Psychosocial Constructs

The questionnaires assessing these constructs (as described below) were completed via RedCap via a secure link that research staff sent to each webMFT participant. Reminder emails were sent to prompt participants to complete the on‐line questionnaires. Participants received $25 for completion of each assessment (4‐ and 8‐week timepoint), totaling $50. At baseline, 4, 8 and 12 weeks, we administered the following measures:

1. Multi‐dimensional Self‐Efficacy for Exercise Scale [28]. This 9‐item scale assesses three domains of self‐efficacy (task, coping and scheduling); higher scores indicate greater confidence in exercising. This measure was used to evaluate changes in self‐efficacy domains during the 12‐week intervention.

2. Self‐regulation was assessed using the 12‐item Physical Activity Self‐Regulation Scale (PASR‐12) [29]. PASR‐12 contains 6 subscales: self‐monitoring, goal setting, eliciting social support, reinforcement, time management, and relapse prevention. Subscale scores are summed to provide an overall self‐regulation score. Higher overall scores indicate greater use of self‐regulation strategies to support PA.

3. Affective Response to PA. The Feeling Scale is a single‐item measure of the valence dimension of affect [30], used in PA studies and shown to be longitudinally associated with PA [31, 32]. Patients rate their feelings during PA on an 11‐point good/bad bipolar scale (+5 = very good, −5 = very bad). This scale was administered retrospectively to examine whether affective responses change during PA adoption.

4. Social Support. The Social Support for Exercise [33] measure assesses support from friends and family for PA. Participants report their support (family and friends were scored separately) on a 5‐point Likert scale with 1 = never and 5 = very often. Higher scores indicate more social support.

3.2. PA Adoption

We tracked weekly PA participation (defined as active minutes) using the Fitbit tracker. Coaches viewed their assigned participant's active minutes per week as shown on the web platform to track webMFT participants' progress in initiating behavior change (weekly active minutes).

3.3. Intervention Engagement

The web platform automatically tracked core metrics critical to intervention delivery (e.g., number of completed calls, duration of calls). For this paper, we used number of calls completed per participant as an index of engagement with the intervention.

4. Analyses

Baseline socio‐demographics were summarized using means (standard deviations) and frequencies (percentages) as appropriate. We present the data both across and within study arms for a complete understanding of the study sample (though the current paper is restricted to those randomized to the webMFT group).

As a preliminary step, within‐person changes in key psychosocial constructs (self‐efficacy, self‐regulatory skills, affective response to PA and social support) were assessed from baseline to 4 week, 8‐week and 12‐week follow‐ups using a series of generalized linear models (with time operationalized as factor‐level variables) to allow for simultaneously testing changes from baseline to each end point.

Next, using a series of GLMs, we examined associations between changes over time in self‐efficacy, self‐regulation skills, affective response to PA and social support with coaching call receipt (count outcome) and min/week of PA as measured by Fitbit (continuous outcome) at 12 weeks. Models controlled for confounders identified via correlation analysis and in the case of PA outcomes, also controlled for week 1 PA values. Count outcomes were modeled assuming a Poisson outcome distribution (or Negative Binomial should the outcome be over‐dispersed). PA outcomes assumed a Normal distribution. Of importance to note, these analyses were intended as hypothesis generating rather than strict statistical hypothesis testing. We present inferential statistics for the purpose of identifying a signal in the noise and for informing future studies. All analyses were run in R [34] with significance level set at $\alpha$ = 0.05 apriori.

5. Results

A full description of the study sample has been presented elsewhere [17]. In brief, participants randomized to the webMFT group (N = 30) were 58 years of age on average (SD = 8.55), 20% non‐white and 10% Hispanic/Latino. Table 1 describes the aggregate sample (N = 61) and baseline sociodemographic characteristics separately by study arm. Previously published work has shown that although not significant at a 5% level, between‐group differences were in the expected direction with greater Fitbit active minutes for webMFT compared to MVPA Tracking at 12 weeks (p = 0.22). Specifically, webMFT participants went from a mean of 150.83 min/week at week 1(SD = 136.56) to a mean of 174.43 min/week (SD = 120.98) at 12 weeks.

TABLE 1.

Baseline sociodemographic characteristics of study sample (overall and by arm), N = 61.

	Overall, N = 61	webMFT, N = 30	MVPA tracking, N = 31
Age, years	57.57 (11.73)	58.10 (8.55)	57.06 (14.29)
Race
White/caucasian	50 (82%)	24 (80%)	26 (83.9%)
Black/african american	9 (14.8%)	4 (13.3%)	5 (16.1%)
Other	2 (3.3%)	2 (6.7%)	0
Ethnicity
Not hispanic or latino	58 (95.1%)	27 (90%)	31 (100%)
Hispanic or latino	3 (4.9%)	3 (10%)	0
Education
At least some college (%)	55 (90.2%)	27 (90%)	28 (90.3%)
Employment
Full time/part time	34 (55.7%)	20 (66.7%)	14 (45.2%)
Retired	20 (32.8%)	8 (26.7%)	12 (38.7%)
Homemaker	5 (8.2%)	2 (6.7%)	3 (9.7%)
Unemployed	2 (3.3%)	0	2 (6.5%)
Income
$20–29,999	2 (3.3%)	0	2 (6.7%)
$40–49,999	5 (8.3%)	3 (10%)	2 (6.7%)
$50–59,999	6 (10%)	3 (10%)	3 (10%)
$60–69,999	4 (6.7%)	2 (6.7%)	2 (6.7%)
$70–79,999	6 (10%)	4 (13.3%)	2 (6.7%)
$80–89,999	5 (8.3%)	4 (13.3%)	1 (3.3%)
$90,000+	29 (48.3%)	13 (43.3%)	16 (53.3%)
Don't know	3 (5%)	1 (3.3%)	2 (6.7%)
Surgery
Lumpectomy	38 (64.4%)	17 (58.6%)	21 (70%)
Mastectomy	23 (38.3%)	14 (46.7%)	9 (30%)
Treatment
Chemotherapy	31 (50.8%)	15 (50%)	16 (51.6%)
Radiation	47 (77%)	23 (76.7%)	24 (77.4%)
Hormone therapy	30 (49.2%)	16 (53.3%)	14 (45.2%)

Note: N (%) for categorical variables; mean (standard deviation) for continuous variables.

Unadjusted descriptives of self‐efficacy, self‐regulation, social support and affective response are presented over time for webMFT participants in Table 2, including subscale scores for self‐efficacy. Although total self‐efficacy scores did not significantly change from baseline to 4‐weeks, there were significant increases among webMFT participants from baseline to 8‐weeks (p = 0.03) in the expected direction. However, from baseline to 12‐weeks, scores significantly decreased (p = 0.02). When considering individual subscales, there was a significant increase in Scheduling score from 4 to 8 weeks (p = 0.01) but significant decreases in Scheduling and Coping scores from 8 to 12 weeks (p's = 0.01). Self‐regulation scores significantly increased baseline to 4‐weeks (p < 0.001) and trended toward significant improvements from baseline to 8‐week and 12‐week follow‐ups (p's < 0.10). Participants reported early increases in positive affective response (p < 0.001 from baseline to 4‐weeks) with no further changes at 8 or 12‐week follow‐ups). Finally, there were significant increases in social support (both friends and family participation from baseline to 4‐weeks, p's < 0.02) with no further changes between 4 and 12 weeks.

TABLE 2.

Unadjusted changes in psychosocial constructs from baseline through 12‐week follow‐up.

	Baseline	4‐weeks	8‐weeks	12‐weeks	p‐value (for within‐person change)
Self‐efficacy
Task	22.31 (6.05)	23.46 (4.93)	23.08 (4.91)	22.89 (4.96)	0.18 (baseline to 4w)
					0.21 (4 to 8w)
					0.38 (8 to 12w)
Coping	17.08 (6.99)	17.58 (6.21)	19.63 (4.59)	16.96 (7.71)	0.33 (baseline to 4w)
					0.08 (4 to 8w)
					0.01 (8 to 12w)
Scheduling	22.88 (5.87)	21.85 (5.37)	24.08 (4.19)	21.48 (7.15)	0.12 (baseline to 4w)
					0.01 (4 to 8w)*
					0.01 (8 to 12w)*
Self‐efficacy (total score)	21.06 (5.58)	20.56 (4.92)	22.23 (4.37)	18.60 (6.09)	0.47 (baseline to 4w)
					0.03 (4 to 8w)*
					0.02 (8 to 12 w)*
Self‐regulation (PASR)	29.17 (10.50)	42.71 (5.63)	44.67 (7.02)	38.30 (9.64)	< 0.001 (baseline to 4w)*
					0.09 (4 to 8w)
					0.08 (8 to 12w)
Affective response	2.08 (1.84)	3.00 (1.23)	3.07 (1.36)	2.87 (1.75)	< 0.001 (baseline to 4w)*
					0.42 (4 to 8w)
					0.37 (8 to 12w)
Social support (family participation)	15.64 (5.29)	22.64 (10.08)	24.00 (9.85)	23.70 (2.22)	< 0.001 (baseline to 4w*)
					0.13 (4 to 8w)
					0.50 (8 to 12w)
Social support (friend participation)	13.30 (3.69)	17.11 (8.19)	17.29 (9.86)	19.63 (10.43)	0.02 (baseline to 4w*)
					0.48 (4 to 8w)
					0.09 (8 to 12w)

Note: Mean (SD) reported at each timepoint. Within‐person change in constructs from baseline to 12 weeks was assessed (p‐value reported to compare changes from baseline to 4w, 4w to 8w and 8w to 12w).

^* p < 0.05.

Finally, we examined whether changes in these key psychosocial constructs were associated with changes in active min/week (as collected via Fitbit) among webMFT participants. Change in total self‐efficacy score over time was significantly associated with PA at 12‐weeks (b = 5.25, SE = 2.07, 95% CI: 1.19–9.31, p = 0.01) such that greater increases were associated with greater active min/week. A one‐unit increased change in self‐efficacy was associated with a 5 min/week difference in PA when controlling for Week 1 PA. There were no significant associations between changes in coping or task scores and changes in active min/week (p's > 0.25). There was, however, a significant positive association between changes in scheduling over time and changes in active min/week (b = 0.60, SE = 0.20, 95% CI: 0.24–0.99, p = 0.01) such that greater increases in self‐efficacy for scheduling was associated with greater improvements in PA over 12 weeks.

Results did not indicate an association between self‐regulation and active min/week (b = 0.09, SE = 0.15, 95% CI: −0.20 to –0.38, p = 0.59). However, change in affective valence over time was significantly associated with PA at 12‐weeks (b = 1.12, SE = 0.08, 95% CI: 0.96–1.28, p = 0.01) such that greater increases were associated with greater min/week PA. Finally, changes in Social Support (friends) were significantly associated with changes in active min/week, b = 0.38, SE = 0.18, 95% CI: 0.03–0.73, p = 0.04 such that greater improvements were associated with more PA over time. An even stronger effect was seen for Family participation, b = 0.41, SE = 0.11, 95% CI: 0.19–0.63, p = 0.01. See Figure 1a‐d.

FIGURE 1.

(a–d) Theoretical constructs (baseline, 4, 8 and 12 weeks) and fitbit active minutes/week (weeks 1–12). Figures show changes over time in the theoretical constructs (lines) alongside activity minutes (from the Fitbit) as bars.

When considering whether changes in theoretical constructs were associated with intervention engagement, we found significant associations between change in total self‐efficacy scores, scheduling scores, self‐regulation, social support (family) and positive affect with intervention engagement in the expected direction (greater improvements associated with higher engagement), Table 3.

TABLE 3.

Association between changes in theoretical constructs and intervention engagement.

	Association with weekly call receipt
Self‐efficacy
Task	0.001 (0.01), 95% CI: −0.02 to 0.02, p = 0.78
Coping	0.01 (0.002), 95% CI: 0.006–0.02, p = 0.01*
Scheduling	0.01 (0.003), 95% CI: 0.004–0.02, p = 0.01*
Self‐efficacy (total score)	0.01 (0.004), 95% CI: 0.002–0.02, p = 0.02*
Self‐regulation (PASR)	0.003 (0.002), 95% CI: 0.001–0.01, p = 0.049*
Affective response	0.02 (0.01), 95% CI: 0.0004–0.03, p = 0.04*
Social support (family)	0.02 (0.01), 95% CI: 0.0004–0.024, p = 0.02*
Social support (friends)	0.001 (0.001), 95% CI: −0.01 to 0.002, p = 0.91

Note: Regression coefficient (b), Standard deviation (SD), 95% Confidence Interval (CI) and p‐value (p) reported.

^* p < 0.05.

6. Discussion

Examining psychosocial constructs and their effects of PA behavior and engagement with the intervention is integral for the development and refinement of PA interventions to improve PA adoption. Our results showed promising trends in theoretical constructs and the outcomes (PA and intervention engagement) with self‐efficacy, especially scheduling, and affective response. Additionally, social support from family and friends was strongly associated with active min/week. Intervention engagement was significantly associated with improvements in total self‐efficacy, scheduling self‐efficacy, self‐regulation, affective response (intra‐personal variables), and social support (interpersonal variable).

Total self‐efficacy demonstrated significant improvements by 8‐weeks compared to baseline, with scheduling self‐efficacy showing notable improvements at weeks 4 and 8. These findings are consistent with SCT and TTM, which position self‐efficacy as influential during the early adoption phase of PA [9, 21, 22], and align with evidence linking higher total self‐efficacy to greater exercise adherence with the use of strategies such as goal‐setting, self‐monitoring, and social support [35]. However, total self‐efficacy, as well as scheduling and coping self‐efficacy declined from week 8 to week 12, with total self‐efficacy falling below baseline levels by the end of the intervention. This pattern suggests that early gains in confidence were not maintained as the intervention progressed. One potential explanation is the increasing behavioral demands of the program (e.g., progression toward 150 min/week), which may have introduced greater time‐management and coping challenges reflected in the decline in scheduling and coping self‐efficacy scores from Week 8 to the end of the program. Prior literature indicates that self‐efficacy can fluctuate when task complexity increases (e.g., as weekly PA goals increased over time) or when initial novelty and support fade [36, 37]. Future training of coaches should focus on the importance of assisting survivors to find time to be physically active in their daily schedule and helping them to cope with barriers to PA so that they build confidence in these domains. These results may also reflect variations in the coaches' implementation of behavior change techniques over the 12 weekly calls, which will be examined in subsequent analyses.

Fitbit active minutes/week in Week 1, as reported previously [17], in the webMFT group were relatively high, averaging 150.83 minutes/week. (Note that accelerometer measured PA was used to screen out participants who were too active at baseline and not PA as measured by the Fitbit.) We were surprised to see these high levels of active min/week in the first weeks of the program, since the Actigraph measured PA showed that participants were sedentary at baseline. Unfortunately, we did not measure PA using the Actigraph during the 12‐week program, the units were used only at baseline and at post‐intervention. We acknowledge that the high active min/week at baseline may have contributed to a ceiling effect for reaching the outlined study goals, thus limiting the potential for influencing psychosocial changes and large behavioral changes.

The coach calls incorporated behavioral strategies such as self‐monitoring, goal setting, positive reinforcement, and barrier problem solving. In addition to the coach calls, the use of the Fitbit tracker, weekly reminders to sync the Fitbit, graded weekly PA goals to increase active minutes up to the national recommendations, and PA tip sheets, aimed to enhance self‐efficacy throughout the intervention. Further coding and analysis of the frequency, timing, and delivery of behavior change techniques implemented throughout the intervention calls could provide deeper insight into understanding the dynamic changes in psychosocial constructs observed across the 12‐week intervention [38].

Self‐regulation demonstrated early improvements and trended positively over time; however, it was not significantly associated with active min/week over 12 weeks. This differs with prior research, which often reports at least a moderate association between self‐regulatory skills—such as goal‐setting and self‐monitoring—and PA behavior change [39, 40]. While literature frequently highlights early gains in self‐regulation as a precursor to later behavioral change [41, 42], our findings diverge from this expectation. One plausible explanation is the cohort's high level of Fitbit active minutes at Week 1, thereby limiting the scope for meaningful change. In addition, the use of the Fitbit based self‐monitoring may have influenced these relationships. For example, in a similar intervention among cancer survivors, the inclusion of Fitbit and self‐monitoring platform resulted in significant increases in MVPA at 12 weeks (+53.6 min/week at 12 weeks), although these gains attenuated at follow‐up (+24.6 min/week at 24 weeks), with minimal added benefit from supplemental behavioral supports [43]. These findings suggest that self‐monitoring tools may be effective for short‐term PA increases, particularly in more active individuals, but may be less effective in sustaining behavior over time. In our present study, the absence of an association between self‐regulation and PA coupled with the declines in self‐efficacy during the later stages of the intervention, suggests that while external monitoring may have supported early engagement, it may have not been sufficient to foster sustained internal regulatory skills or confidence as intervention demands increased and novelty diminished.

Although participants began the program with relatively high PA at Week 1 (measured by the Fitbit), affective response showed meaningful improvements, increasing significantly by week 4 and then stabilizing for the remainder of the intervention. Prior work suggests that more positive affective reactions are associated with greater long term participation in PA [31]. In addition, research on affective memory indicates that both in‐task and recalled affect influence future exercise behavior, with lower‐intensity exercise generally evaluated more positively [44]. While affect was assessed at baseline and 4, 8, and 12 weeks (i.e., not during PA itself), the observed improvements in reported affect in our sample were associated with PA over 12 weeks. This pattern aligns with broader literature that positive emotional experiences during PA may help support continued participation [45, 46], even when changes in self‐regulatory skills were not sustained.

Social support from friends and family improved significantly between baseline and 4 weeks (p's < 0.02), with no further change observed from weeks 4–12. These early gains likely reflect the specific encouragement from peer coaches during weekly calls to engage social networks for support in their PA goals. Importantly, changes in social support from friends and family were positively associated with changes in active min/week. Social support from family was also associated with intervention engagement. Notably, our coaches were physically active breast cancer survivors, which aligns with recent evidence that leveraging social support between survivors can improve MVPA even in the absence of expert facilitation [47]. These findings underscore the potential for interventions that prioritize supportive peer relationships to enhance PA among breast cancer survivors.

In this study, changes in key theoretical constructs, namely total self‐efficacy, scheduling self‐efficacy, self‐regulation, affective response and social support were positively associated with receipt of intervention calls. These findings suggest that improvements in these constructs may play a role in supporting engagement in the intervention, an important goal for behavior change, while also reflecting the dynamic interplay between psychosocial factors and sustained participation. Previous evidence demonstrates that greater behavioral engagement predicts increases in self‐efficacy over time, suggesting that intervention engagement may contribute to changes in key psychosocial constructs [42]. Interpretation of these changes should consider potential bidirectional relationships between intervention engagement, psychosocial constructs, and PA behavior. Call completion, while indicative of feasibility, also reflects variability in participant exposure to the intervention. As previously reported, on average, 90% of calls were completed, meeting our prespecified feasibility criteria. Although the webMFT program also included the use of the Fitbit activity tracker, and educational tipsheets, the weekly calls from the coaches were central to the program.

7. Strengths

This study captured longitudinal changes in key theoretical constructs, offering insight into how psychosocial variables‐intrapersonal and interpersonal‐change across a 12‐week peer led intervention for cancer survivors. It also linked these changes with PA behavior outcomes and intervention engagement, providing a more nuanced understanding of intervention implementation.

7.1. Clinical and Research Implications

These results help inform intervention delivery and guide interventionists on which constructs they need to focus their efforts to enhance PA adoption among cancer survivors. By assessing theoretical constructs during the intervention phase and not only at the end of the intervention, the results also provide a more granular understanding of when the constructs change during the intervention phase and their impact (or absence of an impact) on PA adoption and engagement with the intervention.

8. Limitations

While the primary study effect did not show a significant between‐group difference in MVPA as measured by accelerometers [17], psychosocial constructs were not assessed in the MVPA Tracking group (comparison) during the 12‐week intervention, limiting interpretation of results. Future research should include assessment of psychosocial constructs in both the intervention and comparison group to make conclusive statements about causal relationships of psychosocial constructs with behavioral outcomes and intervention engagement. Additionally, mediators were not formally tested, which could help to further clarify the mechanism of action for change in behavioral outcomes. When interpreting results, we acknowledge that the study participants were a relatively small and select group of breast cancer survivors: chiefly white (80%) well educated, with higher incomes.

9. Conclusions

In conclusion, SCT‐related constructs improved during the first 4‐weeks of the intervention phase. Early changes in self‐efficacy, especially scheduling efficacy, reported affect, and social support may play a role in supporting PA initiation among breast cancer survivors. Targeting these constructs during the initial phase of intervention delivery may help facilitate PA adoption. Further work is needed to examine the dynamic interplay of theoretical constructs during delivery of theory‐based interventions, as well as the influence of external supports (e.g., self‐monitoring technologies) and internal self‐regulatory processes. Improved understanding may help clarify how peer‐coaching strategies influence engagement and PA outcomes to inform more effective interventions.

Author Contributions

Dr. Pinto conceptualized the study and its design. Dr. Dunsiger conducted the statistical analyses. All authors contributed to the first draft of the manuscript and have commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.