Patient-friendly summary
If you read nothing else
Bottom line
Therapist-guided internet-based CBT is an effective and potentially safe option for chronic pain, producing small-to-medium improvements in psychological, physical, and daily-life outcomes versus passive care and performing comparably to or modestly better than active treatments.
Moderate evidencePublished
Evidence hierarchy
Study participants
Adults with chronic pain (mixed chronic pain syndromes, chronic back pain, fibromyalgia, arthritis, headache, IBS, and other musculoskeletal conditions)
Study Summary
This systematic review and meta-analysis pooled 33 randomized controlled trials (5,133 adults) testing therapist-guided internet-based cognitive-behavioral therapy (iCBT) for chronic pain conditions, including mixed chronic pain syndromes, chronic back pain, fibromyalgia, and arthritis. Compared with passive control conditions (waiting list or usual care), guided iCBT showed small-to-medium benefits across psychological outcomes (e.g., depression, anxiety, distress, catastrophizing, pain self-efficacy; effect sizes 0.34–0.47), physical outcomes (pain intensity and fatigue; 0.26–0.29), and impact-on-daily-life outcomes (pain interference and quality of life; 0.38–0.41). Against active control conditions (such as face-to-face CBT or other therapies), iCBT was about as effective for most outcomes and modestly better for distress, pain acceptance, and pain interference (after outlier removal). The authors conclude guided iCBT is an effective and potentially safe treatment for chronic pain, while noting that treatment safety and the contribution of individual treatment components are inconsistently reported.
Key Findings
| Finding | Detail | Impact |
|---|---|---|
| Guided iCBT outperformed passive controls across all measured domains | Versus waiting-list or usual-care controls, pooled standardized mean differences were 0.34–0.47 for psychological outcomes (anxiety, depression, pain self-efficacy, catastrophizing, pain acceptance, distress), 0.26–0.29 for physical outcomes (pain intensity, fatigue), and 0.38–0.41 for impact outcomes (pain interference, quality of life) — small to medium effects. | High |
| Against active treatments, iCBT was broadly comparable with a few small advantages | Pooled effects for depression, anxiety, catastrophizing, pain self-efficacy, and pain intensity were close to zero versus active controls. iCBT showed a small-to-medium benefit for distress (SMD 0.40), a small benefit for pain acceptance (SMD 0.15), and a small-to-medium benefit for pain interference (SMD 0.30) after removing one outlier. | High |
| Longer treatments were associated with larger effects for anxiety and quality of life | In meta-regressions (excluding two high-leverage long-duration studies, range 4–12 weeks), each additional treatment week was associated with a 0.10 increase in the effect on anxiety (p = .03) and a 0.10 increase in the effect on quality of life (p = .046) versus passive controls. No duration moderation was found for other outcomes. | Medium |
| Mode of therapist contact mattered mainly for pain self-efficacy | For most outcomes, whether contact was synchronous, asynchronous, or mixed did not change effect sizes. For pain self-efficacy, studies with mixed (SMD 0.58) or synchronous (SMD 0.67) contact had larger effects than those with purely asynchronous contact (SMD 0.19). A depression difference disappeared after outlier removal. | Medium |
| Treatment satisfaction was high and serious adverse events were not reported | Across 15 studies, participants were generally satisfied and would recommend the treatment. Among the 11 studies reporting adverse events, no serious adverse events were reported, though some studies listed hospitalization as a dropout reason. Symptom deterioration in intervention groups ranged from 0% to 13%, similar to or lower than controls. Dropout was related to lack of time, health issues, technical difficulties, and lack of computer skills (iCBT and active controls ~23%, passive controls ~16%). | Medium |
Versus waiting-list or usual-care controls, pooled standardized mean differences were 0.34–0.47 for psychological outcomes (anxiety, depression, pain self-efficacy, catastrophizing, pain acceptance, distress), 0.26–0.29 for physical outcomes (pain intensity, fatigue), and 0.38–0.41 for impact outcomes (pain interference, quality of life) — small to medium effects.
Pooled effects for depression, anxiety, catastrophizing, pain self-efficacy, and pain intensity were close to zero versus active controls. iCBT showed a small-to-medium benefit for distress (SMD 0.40), a small benefit for pain acceptance (SMD 0.15), and a small-to-medium benefit for pain interference (SMD 0.30) after removing one outlier.
In meta-regressions (excluding two high-leverage long-duration studies, range 4–12 weeks), each additional treatment week was associated with a 0.10 increase in the effect on anxiety (p = .03) and a 0.10 increase in the effect on quality of life (p = .046) versus passive controls. No duration moderation was found for other outcomes.
For most outcomes, whether contact was synchronous, asynchronous, or mixed did not change effect sizes. For pain self-efficacy, studies with mixed (SMD 0.58) or synchronous (SMD 0.67) contact had larger effects than those with purely asynchronous contact (SMD 0.19). A depression difference disappeared after outlier removal.
Across 15 studies, participants were generally satisfied and would recommend the treatment. Among the 11 studies reporting adverse events, no serious adverse events were reported, though some studies listed hospitalization as a dropout reason. Symptom deterioration in intervention groups ranged from 0% to 13%, similar to or lower than controls. Dropout was related to lack of time, health issues, technical difficulties, and lack of computer skills (iCBT and active controls ~23%, passive controls ~16%).
Strengths
- Large, comprehensive evidence base: 33 RCTs and 5,133 participants drawn from six databases searched from inception to February 2022, with high inter-rater agreement (Cohen's kappa = 0.918).
- Followed PRISMA guidelines with a pre-registered protocol (PROSPERO CRD42017079422) and used random-effects models, sensitivity/outlier analyses, and retrospective power analyses.
- Analyzed iCBT against passive and active control conditions separately, giving more clinically actionable comparisons.
- Went beyond efficacy to report on under-studied IMMPACT domains: treatment satisfaction, global improvement, dropout reasons, adverse events, and symptom deterioration.
- Assessed publication bias (funnel plots, Egger's test, trim-and-fill) and explored treatment duration and therapist-contact mode as moderators.
Limitations
- Effect sizes were generally small to medium, and benefits against active comparison treatments were limited for most outcomes.
- Analyses comparing iCBT with active control conditions were underpowered (none reached the 0.80 power threshold), so some null findings may reflect limited power, though the authors note effects were close to zero.
- Moderate-to-high statistical heterogeneity was present for several outcomes (e.g., pain self-efficacy, distress, pain intensity, quality of life).
- Evidence of publication bias was found for the catastrophizing outcome (adjusted effect dropped from 0.43 to 0.31 after trim-and-fill).
- Risk-of-bias concerns: many studies had high or unclear risk for blinding of outcome assessment (e.g., unblinded statisticians) and selective reporting (42% inadequately registered).
- Follow-up results could not be pooled due to inconsistent reporting, so only post-intervention effects were quantified.
- Therapist involvement (total contact time) and the effectiveness of individual treatment components could not be analyzed; adverse events and safety were inconsistently reported across studies.
- Populations were predominantly female (79%) and the studies were largely from a small set of mostly European and Australian countries, which may limit generalizability.
Key Takeaways for Patients
What This Means for You
- 01Online cognitive-behavioral therapy guided by a therapist may help people with chronic pain feel less distressed, more accepting of pain, and less limited by it in daily life, with small-to-moderate benefits compared with simply waiting or receiving usual care.
- 02The benefits seen here are real but modest, and iCBT appears to work about as well as some in-person or other active therapies for many outcomes rather than dramatically better.
- 03Programs were generally completed at home over about 9 weeks (commonly 4 to 12), usually with a psychologist giving written or other personalized feedback — which can lower barriers like travel, cost, and stigma.
- 04Most participants were satisfied with the treatment, no serious harms were reported, and a small minority experienced worsening symptoms similar to control groups; this is a treatment to discuss with your clinician rather than a guaranteed cure.
- 05Common reasons people dropped out were lack of time, health problems, technical difficulties, and limited computer skills — practical factors worth considering before starting an online program.
Read the Full Paper
Access the complete peer-reviewed study from Internet Interventions
View Full Study