Managing Stress Effectively: A Longitudinal Transtheoretical Comparison of Successful Changers, Relapsers, and Non-Changers

Background. Stress levels among Americans are considerable. This research examined Transtheoretical Model of Behavior Change (TTM) constructs for stress management in groups organized by longitudinal progress (dynatypes): Maintainers, Relapsers, and Stable Non-Changers. Methods. Secondary data analysis of a computer-tailored intervention group examined construct use over time across the three groups. Adults (n=427) meeting criteria for not engaging in stress management behaviors at baseline comprised the analytic sample. Participants received three TTM-tailored feedback interventions to help facilitate change at baseline, 3 and 6 months. Demographics, Stage of Change, 10 Processes of Change, Decisional Balance (Pros and Cons), and Self-Ecacy were assessed at baseline, 6, 12, and 18 months. Repeated measures MANOVA followed by ANOVAs, with Tukey follow-up tests assessed differences in use of TTM constructs longitudinally across dynatype groups. Results. Ten of the 13 TTM constructs differentiated between Successful Changers and Stable Non-Changers at baseline and over time. Relapsers were more similar to Successful Changers than to Stable Non-Changers in their use of all constructs, except Self-Ecacy. Conclusion. Findings suggest that baseline cognitive and behavioral constructs can improve prediction of different intervention outcomes 18 months later.


Introduction
Polls indicate that stress levels among the US population remain considerable (American Psychological Association, 2017a; 2017b; 2018; 2019; Gallup, 2019). An international poll found that more than half of Americans reported experiencing stress which was one of the highest rates surpassing the global average at 35%, placing the US at number seven out of 143 countries studied (Gallup, 2019). Furthermore, in the context of the unfolding COVID-19 pandemic, 45% of Americans reported that "worry and stress" related to COVID-19 "has had a negative impact on their mental health," a signi cant increase from 32% reported only one month earlier (Kaiser Family Foundation, 2020).
Stress is a process of negative appraisal of events and one's ability to cope with events (Lazarus & Folkman, 1984). Unmanaged daily stress can turn into chronic stress. Chronic stress has shown to negatively impact multiple organ systems leading to a range of preventable acute and chronic illnesses (Baum & Posluszny, 1999) as well as accelerated aging (Wikgren et al., 2012). A signi cant portion of Americans who suffer from chronic stress are not su ciently practicing effective stress management behaviors essential to reduction and prevention of chronic stress (APA, 2017a). Stress management is the practice of behaviors, such as engaging in regular physical activity and seeking and maintaining healthy social support, allowing one to cope effectively with cognitive and physical symptoms associated with stress (Murphy, 1996).
Given the prevalence of high stress levels in the US and health implications of poor stress management, wide-reaching, scalable interventions for stress management are imperative. Online interventions have advantages such as being e cient and effective for disseminating population-based research targeting public health issues including stress (Hester & Miller, 2006).
One meta-analysis of occupational stress management interventions (n=36) included workplace grouptraining, individual counseling, self-taught techniques, or a combination of diverse methods (Richardson & Rothstein, 2008), although no methods were disseminated online. Although the medium to large effect sizes indicated good effectiveness, the average intervention duration (7.4 weeks with a range of 3 days to 28 weeks) may be prohibitive for many people. In addition, many people lack access to workplace interventions.
A more recent meta-analysis of web-and computer-based stress management interventions (n=23) included a range of intervention durations between 2 and 12 weeks (Heber et al., 2017) and various treatment types including cognitive behavioral therapy, third-wave therapies, and alternative interventions. Stress results from 1-3-month follow-ups showed a small effect size and, for 4-6-month follow-ups, a medium effect size (Heber et al., 2017). Longer term maintenance or relapse after 6 months was not assessed.
Accordingly, one randomized trial recruited a large group of participants who were not practicing stress management. The Transtheoretical Model (TTM)-tailored computerized stress management feedback was delivered at three-time points during the rst 6-months and assessed every six months at follow-up until 18 months (Evers et al., 2006). Approximately 60% of treatment group participants began practicing effective stress management skills by the 6-month time point and maintained this behavior over 18 months, compared to 37% of the assessment-only control group (Evers et al., 2006). This intervention demonstrated signi cant positive effects on stress, depression, and stress management behaviors (Evers et al., 2006). These intervention effects were partially replicated in a separate multiple behavior trial (Prochaska et al., 2008). TTM is a model of intentional behavior change designed to reach individuals at all levels of readiness to change speci c behaviors. As such, it is widely used to develop, guide, and evaluate health behavior interventions, including smoking cessation (Prochaska et (Hall & Rossi, 2008;Prochaska et al., 1994). Self-E cacy refers to the con dence level one feels in one's ability to successfully change a behavior across a variety of challenging situations (Velicer et al.,1990). Processes of Change re ect the frequency with which an individual engages in ten experiential (cognitive/affective strategies) and behavioral activities or strategies shown to facilitate forward movement through the stages (Evers et al., 2006) (Table 1 de nes TTM constructs and shows sample items). Others praise my choice to use healthy strategies to manage stress.

Self-liberation
Choosing and commitment to act or belief in ability to change I promise myself that I will take active steps to manage my stress. This study used dynatype comparison analyses to examine TTM stress management constructs over 18 months. These analyses will examine whether patterns of TTM stress-related construct use over time were comparable to or different from those found for other health behaviors: dietary fat reduction, physical activity, smoking cessation, and sun protection. Based on previous ndings from other behavioral areas, Relapsers and Maintainers were predicted to be similar at baseline on TTM variables but were expected to be doing signi cantly better on these variables than Stable Non-Changers. At 18 months, Relapsers' performance on all TTM processes were expected to fall in between that of Maintainers and Stable Non-Changers.

Participants
This secondary data analysis included treatment group data from a TTM-tailored stress management randomized trial (Evers et al., 2006). All participants provided informed consent and all study methods were approved by an Institutional Review Board. In the parent dataset (n=1085), two directories from market research companies were purchased. Next, individuals were contacted via mail and telephone. Eligibility criteria required that individuals were over 18 years old, English-speaking, experiencing some degree of stress at baseline, and not currently practicing effective stress management. A recruitment rate of more than 70% was reported, including individuals from across 48 states. Additional details and outcomes are described elsewhere (Evers et al., 2006).
Participants were excluded from this study if they had completed only the baseline intervention or if they did not have all TTM constructs available for analysis. For participants who were missing some follow-up time points, the last observation (pre-Action or Action/Maintenance) from the previous time point was carried forward prior to determine their dynatype group. After data cleaning, the analytic sample included N=427 adults who were in the treatment group and followed over time.

Intervention
The TTM-tailored intervention targeting stress management produced three computerized feedback reports at baseline, 3 and 6 months on Stage of Change for stress management and each of three TTM stress-related constructs: Decisional Balance, Self-E cacy, and Processes of Change (Evers et al., 2006).

Measures
Demographics Gender, age, race, and marital status were assessed at baseline ( Table 2). Note. Bold indicates p-value <.05 Stage of Change Participants were presented with the de nition for effective stress management: "Stress management includes regular relaxation and physical activity, talking with others, and/or making time for social activities" and were then asked, "Do you effectively practice stress management in your daily life" . Five response categories allowed participants to self-classify into one of ve stages of change for stress management: Precontemplation (not intending to begin in the next 6 months), Contemplation (intending to begin in the next 6 months), Preparation (intending to begin in the next 30 days), Action (currently practicing stress management behavior, but for less than 6 months) or Maintenance (currently practicing stress management behavior for at least 6 months) ( Self-E cacy On a 10-item scale, participants rated their con dence in their ability to manage stress effectively during challenging situations (e.g., when I am feeling frustrated) on a ve-point Likert scale ranging from 1 = Not at All Con dent to 5= Very Con dent (Evers et al., 2006). Cronbach's alpha for the 10-item con dence scale was also good (α = .874) ( Table 1).
Processes of Change The 10 processes are experiential (cognitive) or behavioral (overt) activities that individuals use to modify problem behaviors. Participants rated 30 items re ecting the 10 Processes of Change for effective stress management on a ve-point Likert scale re ecting their frequency of use in the past month (1 = Never; 3 = Occasionally; 5 = Repeatedly) (Evers et al., 2006). Overall internal consistency for Processes of Change was also good (α = .887) ( Table 1).

Dynatype Group
Three de ned groups re ecting participant change patterns over time were determined based on participants' progress from pre-Action stages at baseline to the Action or Maintenance stages at 6, 12, and 18 months post-baseline. Successful Changers included participants who were in a pre-Action stage at baseline and subsequently moved to Action or Maintenance at any follow-up time point (6, 12, or 18 months) and remained in Action/Maintenance through 18 months. Relapsers included those who were in a pre-Action stage at baseline and moved to Action/Maintenance at one follow-up timepoint (6 or 12 months) but moved back into a pre-Action stage at 18 months. Finally, Stable Non-Changers included those who remained in pre-Action stages at all three follow-up time points (6, 12, and 18 months).

Analytic Plan
Participants use of TTM constructs was assessed on four occasions (0,6,12, and 18 months) over an 18month period and all scores were standardized to baseline scores (T-score: M = 50, SD = 10) to permit comparisons between constructs over time Velicer, et al., 1998). Means across time for each group were plotted for each of 13 TTM constructs. Repeated measures 3 (dynatype group) X 4 (time points) MANOVAs with follow-up ANOVAs, and Tukey post-hoc tests were conducted on all TTM constructs to examine dynatype group differences. Estimates of population effect sizes were calculated using omega-squared with 90% con dence intervals (due to the nature of the F distribution, which serves as the basis for calculating con dence intervals for omega-squared, a 90% con dence interval provides 95% coverage, equivalent to an alpha level of .05) (Steiger, 2004). IBM SPSS Statistics 23 was utilized for all analyses.

Demographic Characteristics
Of the total analytic sample (n=427), 59.0% (n=252) were Successful Changers, 17 Main Analyses Findings Table 3 shows MANOVA TTM construct differences with effect sizes between dynatype groups across baseline, 6, 12 and 18-months. Table 4 shows follow-up pairwise comparison results based on Tukey tests. Signi cant dynatype group differences were also observed for the Behavioral Processes at baseline, 6, 12, and 18-months, with medium to large effect sizes at most occasions. Post-hoc Tukey tests conducted with summary scores for combined Behavioral processes indicated that Successful Changers scored signi cantly higher on Behavioral processes than Stable Non-Changers at baseline and 6-months. At 12months, Successful Changers scored signi cantly higher than Relapsers and Relapsers scored signi cantly higher than Stable Non-Changers on combined Behavioral Processes of Change. For both the Experiential and Behavioral Processes, effect sizes tended to be larger at 12 and 18 months than at baseline and 6 months.

Discussion
All TTM stress management constructs, except the Cons, signi cantly differentiated between Successful Changers and Stable Non-Changers, with Relapsers falling in between these two groups. These results replicated TTM dynatype group patterns of change found for smoking cessation ( , that also found Relapsers to be a distinct group. Moreover, these ndings support TTM's view that experiencing a relapse can often teach individuals important skills that can help further progress towards subsequent attempts to change behavior. More than Stable Non-Changers at all time points, Successful Changers engaged in cognitive activities such as raising awareness of the negative effects of stress or assessing how their environment was affected by not managing stress effectively. At baseline, Successful Changers used nine out of 10 cognitive and behavioral processes signi cantly more often than Stable Non-Changers. This supports the validity of TTM, in that using these Processes facilitated forward transitions and maintenance of stress management. In terms of Behavioral Processes of Change, Successful Changers used Counterconditioning (substitution) and Self-Liberation (choosing and commitment to act) signi cantly more than Non-Changers and Relapsers at all time points. Among the Processes of Change, Relapsers most consistently used Helping Relationships (seeking social support). Relapsers also increased their use of Stimulus Control (managing cues) and Counterconditioning (substitution) at 6 months, which then plateaued at 12 and 18 months. Over time, Relapsers' process use was most similar to that of Stable Non-Changers'. Stable Non-Changers' process use was signi cantly different than Successful Changers' process use, except for Dramatic Relief. Ultimately, Stable Non-Changers did not use behavioral processes as much as the other two dynatype groups and Relapsers did not su ciently utilize cognitive and affective experiential processes.
Effect sizes for Processes of Change between dynatypes over time were large, especially at later assessments, building upon previous ndings in other behavioral areas that Processes were salient predictors of stage movement (Levesque et al., 2011;Sun et al., 2007). Among Experiential Processes of Change, Consciousness Raising had medium effect sizes at baseline and 6 months, with particularly large effect sizes at 12 and 18 months. These ndings indicate that attending to information related to the importance of stress management is particularly important in both the initiation and especially, maintenance of effective stress management. Effect sizes for Environmental Reevaluation and Self-Reevaluation at 12 and 18 months were also large, supporting their importance during stress management maintenance. Environmental Reevaluation re ects awareness of the impact that one's own effective stress management behaviors can have on those around them. Self-Reevaluation re ects one's self image as someone who manages stress well, which these results suggest could be a powerful motivator for initiating and maintaining engagement in stress management behaviors.
Decisional Balance results were that the evaluation of the bene ts of effective stress management (Pros) differentiated between dynatypes, whereas evaluation of drawbacks (Cons) did not. Pros showed a small effect size at baseline, 6 and 12 months, and a medium effect size at 18 months. At the baseline and 18month time point, Successful Changers rated Pros of stress management more highly than Stable Non-Changers and Relapsers did. Over time, Successful Changers were stable in their appraisal of Pros. This nding reveals that individuals who continue to remind themselves of the bene ts of stress management can sustain change over time. At 6 and 18 months, Successful Changers rated Pros as signi cantly more important than Relapsers. Thus, the 6 to 18-month time period may be a critical time for Relapsers, during which intervention efforts may improve e cacy by reemphasizing the bene ts of stress management.
Self-E cacy did not differentiate between dynatypes at baseline, but did show a small effect size at 6, 12 and 18 months. Successful Changers' con dence improved over time and was signi cantly different from Relapsers at 12 months. The effect size between dynatypes from baseline to 18 months was particularly large, re ecting a threefold increase. This nding reveals that building con dence over time is especially important for both initiating and sustaining effective stress management.
Finally, there were more individuals 65 and older in the Non-Changer group compared to the other two groups. This may indicate the di culty of changing behavior later in life or it may re ect a cohort effect in this group of older adults. More randomized research will be needed to better understand these gender and age group differences.

Limitations
This study has some limitations including sample homogeneity. Future studies should include more diverse, heterogeneous samples to replicate and extend these ndings. This is particularly important given the role of culture in the perceptions of stress and coping goals, abilities and strategies (Wong, Since these ndings were based on participants from one study in one TTM-tailored treatment group, these ndings may have limited generalizability. Future studies should examine other treatments longitudinally to determine whether these ndings can generalize beyond this treatment. In addition, Successful Changers were the largest group in this sample (59%), in contrast to the other two dynatype groups (Relapsers and Stable Non-Changers) which had much smaller sample sizes. Thus, the small sample sizes in the Relapsers and Stable Non-Changers groups limited statistical power to detect some differences. Finally, sociodemographic covariates such as income, education, and urban/rural status were not included here, and hence unmeasured confounders may have in uenced our ndings.

Conclusions
Despite these limitations, four important ndings may inform future replication studies and tailored interventions for effective stress management: 1. TTM construct use facilitated stress management change, as Successful Changers used these constructs more frequently than the other groups.
2. Interventions should include reminders for participants of all the bene ts of stress management uniquely for them, particularly for those who may be struggling or those with previous relapses.
3. Since Relapsers engaged in behavioral strategies more than those who did not change but did not engage in su cient cognitive and affective strategies, interventions for individuals who have a relapse history may be more effective if they focus on cognitive strategies. Such strategies are important to utilize, given that they can cultivate intrinsic awareness and motivation to engage in and maintain behavior change.
4. Signi cant baseline differences between those who ultimately changed and those who did not suggest some predictions. Aside from the Cons, Self-E cacy, and Reinforcement Management, the remaining 10 dynamic TTM process variables differentiated between Successful Changers and Stable Non-Changers at baseline (see Table 4). Future research could build upon these ndings by tailoring interventions even more using these TTM constructs. This information could enable interventionists to detect Stable Non-Changers earlier, even at baseline, and then, provide enhancements that could assist them to make more progress.
In addition, based on these ndings, healthcare systems and health care providers are encouraged to be more proactive and screen for or assess patients' levels of stress routinely and discuss the negative implications of chronic stress along with the range of mental health and health bene ts that can be attained through effective stress management. Such an approach can also encourage timely and appropriate referrals which can both improve patient well-being and potentially reduce other healthcare costs. For one example, the National Comprehensive Cancer Network, the overseeing organization of Comprehensive Cancer Care designated hospitals, advocates routine use of a "distress thermometer" to assess current distress as part of the clinical assessment of a patient's vital signs that include weight, temperature, pulse, blood pressure, respiration, and pain. Such excellent clinical care should not be limited to cancer patients, but applied equally well across many settings, including primary care. Finally, encouraging the establishment of strong intrinsic rewards early in treatment is also recommended which can help an individual both adopt and maintain the practice of effective stress management longer-term.

Declarations
Ethics approval and consent to participate. All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2000. Informed consent for participation was obtained from all participants. All protocols were approved by the University of Rhode Island Institutional Review Board.
Consent to publication. Informed consent for publication was obtained from all participants.
Availability of data and materials. The data that support the ndings of this study are available from Pro-Change Behaviors Systems, Inc. but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from Kerry Evers (kevers@prochange.com) upon reasonable request and with permission of Pro-Change Behavior Systems, Inc.