Women and men aged 18-70 years who met the American College of Rheumatology (ACR) 1990 criteria for CWP—pain persisting for more than 3 months, occurring in areas both above and below the waist, on both the right and left sides of the body, and including axial pain [1]—were included in the study. In Sweden, fibromyalgia is clinically diagnosed according to the ACR 1990 criteria.

Participants were excluded for the following reasons: serious physical or psychological disease or other conditions with restrictions on physical activity, not having a smartphone or computer, ongoing psychotherapy or physiotherapy, and insufficient knowledge of the Swedish language for completing the questionnaires and participating in the iCBT. Participants who had a score of >14 on the subscale for anxiety or depression in the Hospital Anxiety and Depression Scale (HADS) were contacted before randomization by a psychologist by telephone for an assessment about the severity of their psychological distress, and if needed, they could be referred to a relevant service within primary health care.

People with CWP were recruited via social media advertisements on Facebook and Instagram in western Sweden between November 2020 and October 2022. They responded to the advertisement on social media by sending an email to the research project inbox. Research assistant 1 sent written study information via email. Individuals who expressed interest after reading the information were screened for eligibility by research assistant 1 during a telephone call. If the person was eligible, the written consent form, along with a battery of paper-based baseline questionnaires, was sent home by mail to the participant, and the participant returned the forms by mail.

The control group received the same physiotherapy intervention as the intervention group, but as a stand-alone treatment. It consisted of 3 individual visits: at baseline, 4 weeks, and 12 weeks, during which a physical activity plan was created, revised, and progressively adjusted.

Fidelity was ensured by delivering the iCBT content identically to all participants through the digital platform, while the physical activity plan was implemented according to a standardized template. Adherence to the core components of the intervention was assessed based on the number of completed modules and attendance at all 3 physiotherapist visits.

The participants completed a battery of paper-based questionnaires at home at baseline and at 6-month follow-up. Follow-up at 6 months was chosen since the participants had access to the iCBT program for 6 months. The questionnaires were sent by mail, with reminders sent by email. The participants completed the same battery of questionnaires after 6 months. They received and returned the questionnaires by mail.

The additional outcomes were collected during the 12-week visit with the physiotherapist and thus not included as outcomes at the 6-month follow-up in the study protocol at ClinicalTrials.gov.

The power calculation was based on the primary outcome variable, pain intensity, which was assessed using a VAS (0-100 mm) [28]. Many factors impact what could be considered a clinically relevant difference in pain intensity between groups, and a suggestion is 20 mm on VAS (16) and a SD of 40-45, which is applied in the power calculation in this study. Statistical significance is P=.05 and power=80%. To be able to make analyses of subgroups based on the SCI-93 score, the number of patients was calculated to be 136. To be able to use nonparametric statistics, and to adjust for possible confounders as well as consider 20% dropout at the follow-up, the final sample size was calculated to be 200 participants.

The randomization was carried out by research assistant 2 and performed with block randomization with blocks of 4, with sealed, opaque, sequentially numbered envelopes prepared by a statistician. The randomization was stratified equally by sex and total score on the questionnaire SCI-93 (total score <50 or ≥50). The stratification by level of stress symptoms was conducted to create groups that were as comparable as possible with respect to stress symptom severity, as measured by the SCI-93. This was done in order to enable subgroup analyses of whether the effect of the intervention differed depending on the level of stress symptoms. Neither the research team nor any of the research assistants had access to the random allocation sequence. Participants and researchers were not blinded in this study. The physiotherapists who carried out the intervention in both groups did not know which group the participants were allocated to, if the participant did not mention it themselves.

The statistical program SPSS (version 28.0.1.1; IBM Corp) was used for data calculations. Descriptive data are presented as mean, median, SD, and range for continuous data and as number and percent for categorical data. The participants were analyzed in trial groups in a randomized trial. Comparisons between groups of change in outcomes were conducted using an independent t test. Within-group changes were calculated using a paired t test. The results for pain intensity are presented both on group level as change in pain intensity over time between the groups and on an individual level, divided into the proportion of participants in each group achieving moderate (≥30%) and major (≥50%) change in pain intensity [46], with differences between groups calculated using the Pearson chi-square test. Effect sizes representing between-group differences in change from baseline to 6-month follow-up were calculated using Cohen d [47]. The primary outcomes, pain intensity and pain locations, were analyzed using linear mixed effects models with repeated measures over time (baseline and 6-month follow-up). Fixed effects included group (intervention vs control), time, and the group-by-time interaction. A random intercept for participants was included to account for within-subject correlation. Potential confounders (sex, age, and pain duration) were first examined by adding each variable separately to model 1. Variables that were significantly associated with pain intensity/pain locations (P<.05) were subsequently included in the adjusted model. The final model (Model 2) included group, time, the group-by-time interaction, and significant confounders. Model assumptions were checked by inspection of residuals. Results are presented as P values, and statistical significance was set at P<.05. Additional analyses of between-group differences at the 12-week visit were conducted for achievement of individual goals using the Mann-Whitney U test, and adherence to WHO recommendations for physical activity and muscle-strengthening exercise was assessed using the Pearson chi-square test. All tests were 2-sided, and statistical significance was set at P<.05. For the subscales in the MFI-20 and SF-36, missing values were handled by replacing them with the mean of the completed items within that subscale, provided that at least half of the items in the subscale had been completed. This procedure allowed a score for the subscale to be calculated. For the SCI-93, the same procedure was applied; up to 10% missing values were allowed to be replaced with the mean of the completed items. For the FIQ total score, the total was calculated as the mean of the subscale scores. If any subscale score was missing, the mean of the remaining subscales was used. A maximum of 3 out of 10 subscales were allowed to be missing to compute the total FIQ score. Missing data were evaluated using the Little missing completely at random (MCAR) test, including baseline and 6-month follow-up outcome measures, treatment group, and selected demographic variables (age, sex, education, and duration of pain). No interim analyses and stopping guidelines were applied in this study.

The study was approved by the Swedish ethical review authority (registration number 2023-04031-0). Written informed consent was obtained from all participants on paper. The original informed consent covered the additional analyses, and no further consent was required. Participant privacy and confidentiality were ensured throughout the study. Data were deidentified prior to analysis, and all records were stored securely with access limited to authorized researchers. The study complied with relevant data protection regulations. The manuscript and supplementary materials do not contain any images in which individual participants can be identified. No financial or nonfinancial compensation was provided to participants for their participation in this study.

A total of 9 participants who scored >14 were contacted before randomization by a psychologist by telephone for a dialogue about the severity of their psychological distress. For 1 participant, the psychologist and the participant jointly decided that inclusion in the study was not appropriate, due to the participant’s mental health. Instead, the participant was referred to a relevant service within primary health care. The other 8 participants were included in the study. The recruitment process was delayed and obstructed by the COVID-19 pandemic, resulting in an interruption before the desired sample size of 200 participants could be reached. Data collection for baseline started in November 2020, and 6 months follow-up was completed in May 2023. A total of 129 participants were included and randomized in the study. A total of 82 (64%) participants completed the 6-month follow-up and were included in the analyses; 43 in the intervention group and 39 in the control group. The participants who were lost to follow-up did not return the questionnaires and did not give reasons for dropping out. The participant flow is described in Figure 2 [23].

There were no statistically significant baseline differences in age, number of pain locations, pain duration, FIQ pain intensity, FIQ fatigue, SCI-93, level of education, work status, sick leave, or disability pension between the 82 who participated in the 6-month follow-up and the 47 who were lost to the 6-month follow-up.

Little MCAR test was not statistically significant (χ²₉₅₀=1007.56; P=.10). Hence, the data could be MCAR, or the power was too low to detect a distinction from this (n=129). A statistically significant test would have indicated not MCAR.

Descriptive data of the participants are presented in Table 1. There were no statistically significant differences in descriptive data between the intervention group and the control group at baseline.

After 6 months, 16 (37%) participants of the 43 in the intervention group had completed all 14 modules in the iCBT program, 19 (42%) participants had completed between 7 and 13 modules, and 8 (19%) participants had completed 0-6 modules. A total of 88 participants (46 in the intervention group and 42 in the control group) completed all 3 physiotherapy visits (baseline, 8 weeks, and 12 weeks).

There were no differences found between the intervention and the control groups for change in any of the secondary outcomes after 6 months. Both groups showed statistically significant within-group improvements in SCI-93, MFI general fatigue, HADS anxiety, and SF-36 pain (Table 2). There was no statistically significant difference between the groups in the participants’ ratings on the PGIC after 6 months (intervention group: median 3.0, IQR 3.0-4.0; control group: median 3.5, IQR 3.0-4.0; P=.59).

Achievement of individual main goals and intermediate goals in the physical activity plan and achievement of WHO recommendations for physical activity and strengthening exercises are presented in Table 3. The participants in the intervention group were more likely to have reached their main goals and their intermediate goals at the 12-week visit than the participants in the control group. No difference was found between the groups at the 12-week visit in achievement of the WHO recommendations of aerobic physical activity and strengthening exercise (Table 3).

This study hypothesized that the combined intervention of therapist-guided iCBT plus a face-to-face physiotherapist-guided person-centered physical activity plan would result in superior improvements in pain, stress-related symptoms, psychological distress, and physical activity among individuals with CWP than the stand-alone physical activity plan. The findings did not support this hypothesis, as no significant differences in improvement were observed between the groups for either the primary outcomes of pain intensity and pain locations or the secondary outcomes of associated symptoms and amount of physical activity. However, both groups demonstrated modest improvements across several outcomes.

These findings are partly in line with a systematic review investigating the effect of CBT on patients with combined pain and stress-related disorders, in which CBT was found to have no effect on pain, but rather on psychological symptoms [20]. The improvements observed in this study are also supported by a systematic review and meta-analysis of CBT combined with exercise in adults with chronic pain, which reported small to moderate improvements in pain intensity, functional disability, and psychological outcomes such as anxiety and self-efficacy [21]. In this study, the absence of additional effects of iCBT beyond a structured physical activity intervention may suggest that physical activity alone contributes substantially to improvements in both physical and psychological outcomes, potentially limiting the detectable added benefit of iCBT in this context. However, as this study may have been underpowered, a larger sample size might have revealed a between-group difference.

There was a statistically significant difference between the groups in achievement of individual goals at the 12-week visit with the physiotherapist, favoring the combined intervention. Similar findings have been reported in a previous study showing that a CBT-based pain management program may contribute to the attainment of personally meaningful goals among individuals with chronic pain [48]. In the intervention group, the iCBT program started about 8 weeks after the participant’s first physiotherapist visit. The iCBT started later because it was considered too demanding to initiate both new physical activity habits and behavioral therapy at the same time. The participants got access to 2 new modules in the iCBT program per week. By the 12-week physiotherapist visit, the participants had access to approximately 8 modules of the total of 12 modules. Thus, the iCBT program could have contributed to improvements or behavioral changes among the participants in the intervention group that enabled them to achieve their individual goals. Approximately half of the participants in each group reached the WHO recommendations for physical activity [29] at the third visit. The physiotherapist-guided person-centered physical activity plan may have contributed to the observed improvements, aligning with previous research on patients with CWP [35]. However, it remains difficult to determine with certainty whether the improvements were attributable to the activity plan, since cohorts with chronic pain have previously shown small symptom improvements over time, particularly in studies with long-term follow-up [12,49].

The main limitations of this RCT are that it was underpowered due to difficulties with participant recruitment during the COVID-19 pandemic and that the loss to follow-up was high, which affects the validity of the study results. The subgroup analyses based on the SCI-93 score were not performed since statistical power was not reached. One possible reason for the high loss to follow-up is that the questionnaires were completed on paper and had to be returned by mail, which may have required additional effort from the participants. Due to limited statistical power, the study may not have been able to detect potential differences in effects between 2 groups receiving active treatments. This was also seen in a systematic review that found iCBT to be promising in ameliorating chronic pain and its associated symptoms when compared to passive control treatments, while it may not yield superior effects when compared to active treatments [18].

The follow-up assessment was conducted 6 months after baseline. This time frame was selected because behavioral change and the establishment of regular physical activity routines were expected to require time. However, it would also have been valuable to collect additional follow-up data at 16 weeks in order to assess outcomes immediately after the completion of the intervention. Another limitation is that the questions used to determine whether participants fulfilled the WHO recommendations for physical activity were not validated. In addition, these questions were only collected at the third physiotherapy visit. Consequently, adherence to the WHO recommendations at baseline could not be determined. While the physical activity plan, including the goal setting, has been used and evaluated in terms of content and patient experiences [34,36], the specific rating of goal achievement used in this study has not previously been applied, and its validity has not yet been established, which may affect the level of confidence in the findings.

Although all participants in this RCT fulfilled the inclusion criteria for CWP, they constituted a heterogeneous population. Pain duration varied from 4 months to 45 years, and pain spread ranged from 3 to 18 locations. The variance in symptom severity across participants could have affected the outcomes, as those with more severe symptoms and longer pain duration may have required greater support than what was provided in this intervention. There was also a large range in the study sample regarding the extent of stress-related symptoms as rated on the SCI-93. This was part of the study plan, since the research question concerned whether the participants could benefit from an iCBT program for stress regardless of their stress level. The variability within this CWP population is consistent with previous findings suggesting the existence of subgroups with differing symptom severities in fibromyalgia, a subgroup of CWP [50], thereby supporting the applicability of this study’s results. A possible bias could be that the participants were aware that iCBT constituted the intervention of interest rather than the comparator intervention. However, as the intervention did not demonstrate superior effects, this potential expectancy bias may have had a limited impact on the outcomes.

Despite these limitations, this study has several strengths. The randomized controlled design reduces the risk of bias. The study was conducted in a real-world clinical setting in a patient population with complex symptoms, enhancing external validity. In addition, mostly validated outcome measures were used, and appropriate statistical methods were applied to account for repeated measures and potential confounders. The intervention in this study was designed to affect both psychological and physiological aspects of pain using iCBT for stress in combination with a person-centered plan for physical activity. After 6 months, 37% of the participants had completed all 12 modules in the iCBT program, which is in line with previous data from the same region in Sweden showing that 34% of patients participating in iCBT in primary health care completed all modules in an iCBT program [51], suggesting that adherence in this study reflects real-world implementation of iCBT in similar clinical settings. The CBT was chosen to be internet-based in this study since the participants were recruited from a large geographical area in western Sweden, involving many primary health care centers. iCBT could easily be provided to all participants regardless of location. Both the person-centered health plan and the iCBT are provided in Swedish primary health care and require minimal health care resources, offering a potentially cost-effective approach.

This study contributes a novel perspective by examining whether a stress-focused intervention influences pain-related outcomes, addressing an area that has received limited attention in previous research. No differences in pain intensity, number of pain locations, or associated symptoms were observed between the intervention group receiving therapist-guided iCBT focused on stress combined with person-centered physical activity and the control group receiving physical activity alone. However, both groups demonstrated improvements over time across several outcomes. Participants in the intervention group achieved their individually set goals to a greater extent than those in the control group, suggesting a potential added value of the intervention in supporting goal attainment. In a clinical context, these findings support person-centered, collaboratively developed physical activity plans as a central component in the management of CWP, while stress-management interventions may be considered as complementary support to facilitate behavioral change and goal attainment. Although iCBT did not provide additional improvements in pain or other standard outcomes, it may still be beneficial for selected patients, particularly those with higher levels of perceived stress, which warrants further investigation. The relatively high rate of loss to follow-up warrants caution when interpreting the findings, as it may affect both the internal validity and the generalizability of the results.