Predictors of Physical Activity in Patients with Fibromyalgia: A Path Analysis

Abstract

Background/Objective:

To identify psychological factors that influence moderate-vigorous physical activity (MVPA) participation in patients with fibromyalgia.

Methods:

In this secondary data analysis, 170 patients received personalized exercise plans and completed baseline and follow-up assessments of self-reported physical activity at weeks 12, 24, and 36. Structural equation modeling was used to examine the predictive strengths of psychological factors [exercise self-efficacy, perceived barriers, and intention] on MVPA participation.

Results:

Using a threshold increase in MVPA of ≥10 metabolic equivalent hours per week (MET h/wk), three groups were defined based on subjects who: achieved a minimum increase of 10 MET h/wk that was sustained for at least 12 weeks (SUS-PA); achieved an increase of 10 MET h/wk that was not sustained for at least 12 weeks (UNSUS-PA); and did not achieve an increase of 10 MET h/wk (LO-PA). Increases in exercise self-efficacy and intention, and reductions in perceived barriers were associated with increased volume of PA, showing the greatest change in the SUS-PA, followed by UNSUS-PA. For the LO-PA group, there was no change in exercise self-efficacy, a decrease in intention, and an increase in barriers. Using path analysis, exercise self-efficacy and perceived barriers were associated with higher volumes of physical activity via greater intention to engage in MVPA.

Conclusions:

For patients with fibromyalgia, exercise self-efficacy, perceived barriers, and intention to exercise are important constructs for increasing physical activity. Our findings provide guidance for practitioners who seek to promote physical activity in fibromyalgia, and suggestions for researchers aiming to improve prediction models.

INTRODUCTION

Fibromyalgia is a chronic condition of unclear etiology characterized by widespread musculoskeletal pain and various degrees of fatigue, anxiety, cognitive difficulties, and non-refreshing sleep [1]. Due to the wide variability in patient presentation and associated symptoms, effective treatment remains a significant challenge for clinicians. While drug therapy can provide symptom relief for some patients, current treatment guidelines recommend a stepwise approach that prioritizes non-pharmacologic approaches (e.g. exercise) as first-line treatments for individuals with fibromyalgia [2].

Evidence compiled from multiple systematic reviews and meta-analyses supports the efficacy of several types of exercise-based interventions in the management of fibromyalgia [2-4]. Unfortunately, despite the well-documented benefits of exercise on patient symptoms and physical function, a high percentage of patients do not achieve recommended levels of moderate-vigorous physical activity (MVPA) and tend to remain highly sedentary [5]. The reasons for poor exercise adherence in fibromyalgia remain unclear, but are likely multifactorial. In healthy, pain free adults, a single bout of exercise typically leads to a period of reduced pain sensitivity, commonly referred to as exercise-induced hypoalgesia. In contrast, the pain response to exercise is variable in chronic pain populations, with some studies demonstrating reduced pain, while others show no change or even increased pain sensitivity following exercise [6].

Clinicians are generally aware that exercise can provide symptom relief for patients and are likely to encourage increased levels of physical activity during clinic visits. However, several factors can influence physical activity [7, 8] . For example, providing only education and advice about the benefits of exercise is not likely to motivate an individual that lacks confidence in his or her ability to become more active (exercise self-efficacy), does not plan to begin an exercise program (intention), and/or rationalizes that the perceived barriers to becoming more active (e.g. worsening symptoms) outweigh the potential benefits. In this regard, it is important for exercise-based interventions to be grounded on theoretical models that adequately explain and predict exercise behaviors.

Given the strength of evidence supporting exercise as a therapeutic modality for fibromyalgia, identifying and overcoming barriers associated with the initiation and maintenance of exercise is critical to the successful management of patient symptoms. Despite strong theoretical links among the constructs of exercise self-efficacy, intention to change, barriers and benefits, these predictors of exercise behavior in people with fibromyalgia requires further elucidation. Therefore, the purpose of this study was to investigate how these theoretical factors influence physical activity participation in patients with fibromyalgia who: (1) increased and sustained participation in MVPA; (2) increased, but did not sustain MVPA levels; or (3) did not increase MVPA levels. The hypothetical structural equation model, as shown in Figure 1, addresses the relationship between the variables previously mentioned. We hypothesized that improvements in exercise self-efficacy, and a reduction in perceived barriers to exercise, would positively influence intention to exercise and, as a result, increased levels of MVPA.

Figure 1.

As shown in Figure 1, changes in perceived barriers and exercise self-efficacy influence intention to engage in PA, which in turn impacts changes in in the volume of PA (* p<0.05).

METHODS

Experimental Design.

Data were obtained from the Research to Encourage Exercise for Fibromyalgia (REEF) study, a randomized, attention-controlled clinical trial designed to investigate the effectiveness of motivational interviewing, a person-centered counseling approach designed to elicit a specific behavioral change [9], to encourage a more physically active lifestyle in adults with fibromyalgia. Details of the research design, measurement protocols, and exercise program for the REEF study have been described elsewhere [10]. Briefly, patients were randomized to either the motivational interviewing intervention group or an attention control group. Prior to the motivational interviewing counseling phase of the study, each participant received two supervised exercise sessions from a qualified fitness instructor and an individualized exercise prescription that included the initial exercise frequency, intensity, duration, as well as the progression of the exercise program over the duration of the study. After completing both exercise sessions, the motivational interviewing group received six exercise-based telephone calls over the ensuing 12 weeks. The attention control group received the same number of telephone contacts centered on fibromyalgia-related health education. Phone calls to both groups were discontinued at week 12 and subsequent outcome assessments were conducted at week 12 (immediate post-intervention), week 24 (3-month follow-up) and week 36 (6-month follow-up). The study protocol was carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki) and approved by the Indiana University Institutional Review Board (ClinicalTrials.gov Identifier NCT00573612).

Study Participants.

All study participants were referred from specialty or primary care clinics with an initial diagnosis of fibromyalgia, which was verified by the study physician (a rheumatologist). To be included, participants (aged 18 to 65 years) had to meet the following entry criteria: (a) American College of Rheumatology classification criteria for fibromyalgia [11]; (b) Brief Pain Inventory (BPI) pain score ≥4; (c) Fibromyalgia Impact Questionnaire–Physical Impairment (FIQ-PI) score ≥2; and (d) be on stable doses of fibromyalgia medications for at least 4 weeks. The exclusion criteria were known cardiovascular or pulmonary disease, any neurological or musculoskeletal disorders that would preclude moderate-intensity exercise, other inflammatory rheumatic conditions (e.g. rheumatoid arthritis, systemic lupus, or other connective tissue disease), previous participation (< 6 months) in MVPA on three or more days a week, and treatment with drugs affecting the chronotropic response to exercise (e.g. beta-blockers).

Clinical Outcome Measures.

The Fibromyalgia Impact Questionnaire (FIQ) was used to quantitate the overall impact of fibromyalgia over several dimensions of health status [12]. The FIQ contains 10 subscales (score range 0-10), which are summed to yield the total FIQ score (FIQ-Total). The FIQ-physical impairment (FIQ-PI) subscale assesses the patient’s ability to perform different types of physical activity. Higher scores on each of the subscales and the FIQ-Total indicate a greater severity of symptoms and/or a higher negative impact of fibromyalgia on the individual.

The Brief Pain Inventory (BPI) questionnaire was used to assess pain severity and the overall impact of pain on daily function. BPI pain intensity is the mean score of four items asking about the worst, least, and average pain in the last week, and the current level of pain [13]. The BPI has been proven reliable, valid and responsive to change among patients with chronic non-malignant pain [14].

Depression severity was assessed with the Patient Health Questionnaire 8-item Depression Scale (PHQ-8), a brief self-administered questionnaire designed to evaluate major depressive disorder symptoms. The PHQ-8 allows a score (range: 0 to 24) based on the total number and severity of depressive symptoms noted over the previous two weeks [15].

Physical Activity Assessment.

The Community Health Activities Model Program for Seniors (CHAMPS) is a self-administered questionnaire that asks about the frequency, intensity, and duration of physical activity performed during a typical week. In older adults, CHAMPS has demonstrated internal consistency, content and construct validity, and responsiveness to change [16]. Due to its ability to capture a wide range of physical activities, CHAMPS was used to estimate the number of hours spent per week performing sedentary, light, and MVPA.

The term metabolic equivalents (METs) represents the oxygen requirements of various activities (e.g. sitting quietly = 1 MET = 3.5 ml O₂·kg⁻¹·min⁻¹) and multiples of this value provide a simple and practical classification scheme to quantity the intensity of various activities. Moderate-intensity physical activity is typically defined as 3 to 5.9 METs, whereas vigorous-intensity physical activities involves energy requirements ≥6 METs. MVPA, which corresponds to any activity ≥3 METs, has consistently been shown to reduce the health risks associated with numerous chronic health conditions [17]. Using the original scoring procedures from CHAMPS, each activity was assigned a corresponding MET value. Estimates of time spent in MVPA were calculated by multiplying the number of reported hours (h) spent performing each activity by its equivalent MET value and summed across all activities to determine the total number of MET hours per week (MET h/wk).

Physical Activity Group Classifications.

Current physical activity guidelines recommend that adults get at least 150 min/wk of moderate-intensity physical activity, 75 min/wk of vigorous-intensity physical activity, or an equivalent combination of MVPA [17]. When moderate- and vigorous-intensity activities are combined to meet physical activity recommendations, the minimum goal is to achieve approximately 10 MET h/wk above usual daily activities (range 7.5 to 12.5 MET h/wk) [18]. Changes in CHAMPS data were analyzed at each assessment period to identify the number of subjects who successfully increased (from baseline) and sustained an increase in MVPA of at least 10 MET h/wk for at least two 12-wk periods (baseline to week 12; week 12 to week 24; and week 24 to week 36). Under this standard, we defined three physical activity groups based on subjects who: (1) achieved a minimum increase of 10 MET h/wk that was subsequently sustained or increased for an additional 12 weeks (SUS-PA); (2) achieved a minimum increase of 10 MET h/wk that was followed by a decrease in physical activity for at least one 12-wk period (UNSUS-PA); or (3) did not achieve an increase of at least 10 MET h/wk from baseline (LO-PA) [19].

Model Variables.

The Exercise Self-Efficacy questionnaire is a 9-item scale designed to assess one’s belief and confidence about the ability to successfully increase exercise participation. Each item is rated on a 10-point scale ranging from 0 (not confident) to 10 (very confident) and summed to achieve a total score. This scale has good psychometric properties and is responsive to change [20].

Intention to exercise was assessed using a 4-item scale developed by Courneya, et al. [21]. All items are ranked on a 7-point Likert scale (1-7), with higher scores indicating greater intent. The first two items refer to the number of times an individual intends to exercise in the upcoming 4 weeks. The third item refers to the frequency of planned physical exercise; scores range from not at all to everyday. The final item asks respondents to indicate the likelihood of engaging in physical activity at least 12 times in the upcoming 4 weeks; scores range from definitely not to definitely. Scores for each item are transformed into z-scores which are then summed to create a total intention score. This scale has demonstrated good validity and reliability [21].

Perceived benefits and barriers to physical activity were assessed using a 16-item self-report questionnaire designed to assess a person’s decision-making in regard to exercise adoption [22]. The questionnaire has two components; 10 items compose positive perceptions of exercise (pros) scale, and six items represent avoidance of exercise (cons). Each item is ranked on a 5-point scale based on the importance of the item in making the decision to exercise (5 = very important, 1 = not important). This measure has shown good internal consistency and responsiveness to change [23, 24].

Statistical Analyses.

Initial analyses were performed using ANCOVA models to determine the association between model variables and physical activity groups, adjusted for model variables’ baseline values (SAS v9.4). Controlling for REEF treatment group assignment, structural equation modeling (SEM) methods were used in the path analyses so that every path could be modeled at one time, accounting for the variance of each association. Paths were specified from both barrier and self-efficacy onto intent and intent onto physical activity. Baseline values were linked to each individual variable and analyses were adjusted by REEF treatment group assignment. Results are given as standardized path coefficients, for direct comparison to each other. The path analyses were performed using path analysis models in Mplus (Mplus v 7.31). All analytic assumptions were verified, with the theta parameterization being utilized due to the categorical nature of physical activity and treatment.

RESULTS

Baseline characteristics of study sample.

Table 1 shows the sociodemographic characteristics of the study participants. A total of 216 patients were enrolled in the REEF trial, of which 170 (78.7%) completed the outcome measures at baseline and each follow-up period and were included in these analyses. Of the 216, 15 subjects were excluded from the current study due to missing data, while an additional 31 subjects were excluded for not fitting into one of the three physical activity group definitions. Except for self-perceived physical impairment, which was slightly higher in those excluded (p=0.03), no differences were noted for any baseline demographic, clinical, or physical activity characteristics for those subjects excluded from the final analyses (n = 46, data not shown) compared to those included in the analysis (n = 170). Table 1 shows that most of study participants were female (94.7%) with a mean (SD) age of 46 (10.9) years. The majority of study participants had some education beyond high school (78.2%) and were employed (54.1%). Clinically, the study sample had a mean disease duration of 9.2 (7.1) years, was moderately depressed [PHQ-8=12.6 (5.1)] and reported moderate-to-severe physical impairment [(FIQ-PI=5.5 (1.5)] and pain severity [(BPI=6.0 (1.3)].

Table 1.

Baseline characteristics of 170 subjects with data at all 3 time periods ^‡

		SUS-PA (n=27)	UNSUS-PA (n=68)	LO-PA (n=75)	p	All Participants (n=170)
Demographics
Age (yr)		45.3 (11.7)	46.6 (10.5)	45.8 (11.0)	0.34	45.9 (10.9)
Gender (% female)		96.3%	92.7%	96.0%	0.62	94.7%
Ethnicity (% non-Hispanic)		96.3%	97.1%	100.0%	0.29	98.2%
Race (% white)		92.6%	85.3%	86.7%	0.63	87.1%
Education (% >high school)		81.5%	83.8%	72.0%	0.21	78.2%
Marital status (% married)		70.4%	60.3%	58.7%	0.55	61.2%
Employment (% employed)		44.4%	50.0%	61.3%	0.48	54.1%
Clinical Variables
BMI (kg/m2)		30.5 (6.9)	31.4 (6.6)	31.3 (7.9)	0.86	31.2 (7.2)
Duration of FM (yr)		8.2 (7.2)	9.3 (6.6)	10.2 (7.7)	0.54	9.2 (7.1)
PHQ-8 depression (range 0-24)†		13.0 (5.3)	12.2 (5.1)	12.3 (5.1)	0.36	12.6 (5.1)
BPI pain (range 0-10)†		6.0 (1.1)	6.1 (1.3)	6.0 (1.4)	0.58	6.0 (1.3)
FIQ-PI (range 0-10)†		5.2 (1.6)	5.7 (1.4)	5.5 (1.7)	0.30	5.5 (1.5)
FIQ-Total (range 0-100)†		66.6 (11.9)	68.4 (12.5)	66.4 (14.7)	0.08	67.4 (12.8)
Medications, % prescribed	Anticonvulsants	22.2%	30.9%	26.7%	0.67	27.7%
	Antidepressants	29.6%	63.2%	66.7%	<0.01*	59.4%
	Opiates	22.2%	27.9%	41.3%	0.10	32.9%
Study Variables
ESE		6.1 (1.7)	5.5 (2.0)	5.1 (2.1)	0.09	5.4 (2.0)
Intention		3.9 (1.4)	3.9 (1.3)	3.9 (1.3)	0.93	3.9 (1.3)
Benefits		42.5 (5.5)	41.3 (5.9)	42.1 (7.6)	0.69	41.8 (6.6)
Barriers		16.9 (5.8)	16.2 (4.7)	16.5 (4.2)	0.82	16.4 (4.7)
CHAMPS (mod-vigorous)		1.2 (2.6)	2.1 (5.1)	1.8 (3.3)	0.64	1.8 (4.0)

^‡Values are means (standard deviation) for continuous variables and percentage for categorical variables.

^*p < 0.05

^†Higher score indicates a worse state of health. Antidepressants: Non-tricyclic antidepressants

Abbreviations: SUS-PA: ≥10 MET h/wk increment in MVPA and sustained; UNSUS-PA, ≥10 MET h/wk increment in MVPA, followed by decline; LO-PA, did not achieve ≥10 MET h/wk increment in MVPA; CHAMPS, Community Health Activities Model Program for Seniors; PHQ-8, Patient Health Questionnaire-8; BPI, Brief Pain Inventory; FIQ-PI, Fibromyalgia Impact Questionnaire-Physical Impairment; ESE, Exercise Self-Efficacy.

At baseline, the study sample represented a mostly low active fibromyalgia population, reporting average MVPA levels below the recommended 10 MET h/wk (mean = 6.6 MET h/wk). After 36 weeks, 27 subjects (15.9%) reported a sustained increase of at least 10 MET h/wk for ≥12 weeks (SUS-PA). Sixty-eight subjects (40.0%) reported an increase in MVPA that was not sustained (UNSUS-PA). The remaining subjects (n=75, 44.1%) did not achieve an increase in MVPA (LO-PA). Baseline demographic and clinical characteristics were mostly similar among the three subgroups; however, fewer participants in the SUS-PA group were receiving antidepressant medications compared to the UNSUS-PA and LO-PA groups (p<0.01). There were no differences among the groups in baseline model variables (exercise self-efficacy, intention to exercise) perceived benefits or perceived barriers.

Associations Between Model Variables and Physical Activity Group.

Table 2 shows the magnitude of change for each model variable from baseline to week 36 and the bivariate associations between changes in model variables and the three physical activity groups. From baseline to week 36, the SUS-PA and UNSUS-PA groups reported greater intention to engage in MVPA, whereas the LO-PA group reported less intention (p<0.0001). Compared to the UNSUS-PA and LO-PA groups, the SUS-PA group reported the largest improvement in exercise self-efficacy (p=0.0009). The SUS-PA and UNSUS-PA groups both reported reductions in perceived barriers to exercise, while the LO-PA group reported a higher number of perceived barriers (p=0.0022). No differences in the perceived benefits of exercise were found among groups (p=0.2654).

Table 2.

Bivariate associations between changes in cognitive variables and physical activity groups

	SUS-PA (n=27)	UNSUS-PA (n=68)	LO-PA (n=75)	p
Baseline to week 36 changes
Change in intention	−1.06 (0.26)ab	−0.40 (0.16)ac	0.52 (0.16)bc	<0.0001
Change in exercise self-efficacy	−0.79 (0.34)ab	−0.06 (0.21)ac	0.68 (0.21)bc	0.0009
Change in perceived barriers	0.91 (0.80)ab	1.65 (0.51)ac	−0.81 (0.49)bc	0.0022
Change in perceived benefits	0.27 (1.44)b	0.40 (0.91)c	2.66 (0.89)bc	0.2654

Values are means (standard deviation).

Values with the same superscript indicate statistically significant pairwise differences using the Tukey adjustment (e.g. SUS-PA and UNSUS-PA have significantly different intention scores, as they both share the superscript ‘a’).

Change variables were calculated as baseline minus follow-up; positive change scores indicate follow-up was smaller than baseline and negative change scores indicate follow-up was larger than baseline (e.g. self-efficacy’s group 1 is in the appropriate direction).

Path Model Analysis.

Standardized regression coefficients for our hypothesized paths are shown in Figure 1. The association between exercise self-efficacy and intention to exercise was significant, (path coefficient = 0.237; p<0.05), suggesting that an improvement in exercise self-efficacy expectations had a direct significant effect on intention to exercise. Perceived barriers and intention to exercise were negatively correlated (path coefficient = −0.169; p<0.05), suggesting that a reduction in perceived barriers to exercise had a direct significant effect on intention to exercise. The association between intention and MVPA participation also was significant, with a path coefficient of 0.515 (p<0.05). No bidirectional relationship was observed between exercise self-efficacy and perceived barriers.

DISCUSSION

In the current study, we investigated factors leading to change in physical activity behavior over a 36-week intervention study period in patients with fibromyalgia. We grouped patients into three categories, regardless of intervention group assignment. We compared subjects who over the 36-week period increased and sustained MVPA (n = 27) to those who increased but did not sustain (n = 68) and those who never increased (n = 75). We examined the role of self-efficacy for exercise, perceived barriers and benefits on intention to exercise, and the role of intention on MVPA.

Over the 36-week period, participants who never increased their exercise behavior experienced decreases in exercise self-efficacy, in perceived benefits and in intention, and increases in perceived barriers. Several factors may explain these decreases in factors important for exercise participation, including comorbid depression, which has been linked to low self-efficacy and higher levels of fatigue among chronic pain patients [1], as well as fears about the potential for worsening (or new) pain symptoms. In the current study, a higher number of participants in the LO-PA group were unemployed and taking anti-depressant medications. Although exercise self-efficacy and depression scores were not different between groups at study entry, we have previously reported that depression is a partial mediator between pain intensity and physical functioning in patients with fibromyalgia [25]. Finally, while these participants may have come into an exercise study ready to engage, somehow they were unable to gain in factors that increase exercise behavior. Thus, it is possible that their final scores on each of these model variables represent a return to what might be their typical scores, and that their higher baseline score represents a hopeful approach to the research study. The two groups that increased their exercise behavior, on the other hand, increased in exercise self-efficacy and intention and decreased in perceived barriers, but experienced no change in perceived benefits.

The results of the structural equation model indicate that intention to exercise directly predicts increased and sustained participation in MVPA in patients with fibromyalgia. Intention is common to several health behavior models that informed our study, including the Theory of Planned Behavior [26], Transthoretical Model [27], and Social Cognitive Theory [28]. Previous research demonstrates that intention is associated with the initiation of exercise [29] . Furthermore, individuals who report greater intention to be physically active have greater success in increasing levels of MVPA [30]. However, in a meta-analysis of prospective correlational studies, intention explained only 33% of the variability in physical activity behavior [31], a finding confirmed recently in one study of walking as exercise for fibromyalgia [7]. Our data showed that intention was the strongest predictor in the model, with a direct, significant effect on MVPA, explaining 44% of the variance.

Despite the predictive value of intention on physical activity, there are clearly other factors at play that keep intention from automatically resulting in behavior change. Termed the “intention-behavior gap,” [32] this phenomenon presents a significant challenge for researchers and clinicians and underscores the need to examine not only variables that can predict behavior change, but additional variables that may explain why these well-studied predictors do not result in change for all participants, all the time. For intention, some of these additional variables might be intrapersonal such as behavioral control [7], past behavior [33], and enjoyment of exercise [34], or broader, such as the built environment [35]. It is also important to note that intention is measured on a continuum. It is possible that intention needs to reach a certain level on the continuum to elicit the behavior change. Future work should determine if this is the case, where that level stands, and how to move individuals along the continuum toward the higher levels of intention. Findings from such studies would contribute to further development of the theories that have intention as a main variable by reducing the intention-behaivor gap and improving its predictive value.

Exercise self-efficacy had a direct, significant effect on intention to exercise, predicting 24% of its variance. This finding is consistent with a meta-analysis that found support for the causal effect of self-efficacy on intention and behavior [36]. Importantly, a recent meta-analysis showed that exercise self-efficacy was consistently associated with physical activity [8], although it may only have a short-term impact [37]. Given its important role in prediction of exercise behavior, and the fact that fibromyalgia patients must initially exercise through pain and discomfort to reach clinical improvements, interventions that increase self-efficacy may be especially useful in this population.

Perceptions of both positive and negative consequences of performing a specific behavior, i.e. perceived barriers and benefits, are thought to be important for behavior change [38, 39]. In the current study, the awareness of the potential benefits of physical activity did not predict increased MVPA participation. This finding matches a meta-analysis examining Social Cognitive Theory constructs and physical activity, which found that these perceptions did not predict physical activity [8]. It is possible that for individuals with fibromyalgia, the suggestion that exercise may help to attenuate pain symptoms may seem paradoxical, given that the pain associated with increased activity often prevents many patients from attempting to follow through with recommendations to become more active [40] . Furthermore, an active lifestyle is unlikely to be initiated unless one believes he or she is capable of sustaining higher levels of physical activity [7].

On the other hand, our findings support the importance of reducing perceived barriers, as they had a significant effect on intention to becoming more active. The model shows that the perception of barriers predicted intention to exercise, accounting for 28% of the variance. Our instrument assessed time commitment and exhaustion after exercise as two barriers to exercise. The expectation of exhaustion after exercise, although potentially accurate, is unfortunate, as a recent systematic review and meta-analysis of 17 RCTs of exercise for FM showed a that exercise in fact reduces overall fatigue [41], and results from the FAST study show that low levels of physical activity are associated with less, not more fatigue [42]. Given the consensus that physical activity provides symptom relief and other health benefits for patients with fibromyalgia, identifying and reducing perceived barriers to becoming more active should be part of every patient encounter.

Limitations of this study include a lack of demographic diversity in this predominately Caucasian, female sample with fibromyalgia. Thus, we are unable to generalize these findings to individuals with different demographic characteristics of our study sample, or to other chronic pain conditions. We relied on self-reported data; therefore, it is possible that some subjects may have been misclassified when assigned to physical activity groups.

Given the complexity of physical activity behavior, and its value in increasing fibromyalgia quality of life [43], it is imperative that the best combination of predictors of physical activity be identified and targeted for interventions. Compared to most research on fibromyalgia and exercise, the results of this study are based on a large sample size and longer study duration. Our findings suggest that future model construction and testing about exercise among patients with fibromyalgia should incorporate exercise self-efficacy, perceived barriers, and intention to exercise as important constructs. However, these variables alone will not lead to broad success in interventions to increase physical activity. Other variables that may predict the model variables included here, should also be explored and included in future studies.

Our findings suggest that a significant proportion of fibromyalgia patients are not able to garner exercise self-efficacy, or change their perceptions of benefits and barriers, which results in lack of intention to exercise and consequently sedentary behavior. Although further study is needed to identify best practices for evaluating and addressing these constructs with patients, it is important for clinicians to identify the clinical and/or lifestyle factors that impact their patients so that they too can be guided toward a more physically active lifestyle. In our data, there were significant differences between the group that increased and sustained MVPA and the other two groups in antidepressant medication. Thus, it is possible that factors such as depression might play a role, and may explain the problems initiating and sustaining MVPA. Other factors likely play a role as well, and merit future research.