Self-help interventions for persistent physical symptoms: a systematic review of behaviour change components and their potential effects

ABSTRACT

Persistent physical symptoms (PPS) remain a challenge in the healthcare system due to time-constrained consultations, uncertainty and limited specialised care capacity. Self-help interventions may be a cost-effective way to widen the access to treatment. As a foundation for future interventions, we aimed to describe intervention components and their potential effects in self-help interventions for PPS. A systematic literature search was made in PubMed, EMBASE, PsycINFO and CENTRAL. Fifty-one randomised controlled trials were included. Interventions were coded for effect on outcomes (standardised mean difference ≥0.2) related to symptom burden, anxiety, depression, quality of life, healthcare utilisation and sickness absence. The Behaviour Change Technique (BCT) Taxonomy v1 was used to code intervention components. An index of potential was calculated for each BCT within an outcome category. Each BCT was assessed as ‘potentially effective’ or ‘not effective’ based on a two-sided test for binomial random variables. Sixteen BCTs showed potential effect as treatment components. These BCTs represented the themes: goals and planning, feedback and monitoring, shaping knowledge, natural consequences, comparison of behaviour, associations, repetition and substitution, regulation, antecedents and identity. The results suggest that specific BCTs should be included in new PPS self-help interventions aiming to improve the patients’ physical and mental health.

KEYWORDS:

• Somatoform disorders
• medically unexplained symptoms
• self-help
• behaviour change
• general practice
• systematic review

Introduction

Physical symptoms that are not attributable to any conventionally defined disease are highly prevalent in general practice (Rosendal et al., 2015; Steinbrecher et al., 2011). These symptoms represent a wide spectrum, ranging from self-limiting symptoms to persistent physical symptoms (PPS) including functional somatic symptoms and debilitating functional somatic syndromes and disorders (e.g., irritable bowel syndrome (IBS), fibromyalgia and chronic fatigue syndrome (CFS)) (Rosendal et al., 2017). PPS are seen in 17% of consultations in general practice and challenge patients, healthcare professionals and the social security system (Budtz-Lilly et al., 2015). PPS often remain ‘unexplained’ after extensive investigation, and they often cause reduced physical and mental health, considerable disability, prolonged sick leave and high healthcare costs (Barsky et al., 2005; Burton et al., 2012; Gjesdal et al., 2009; Henningsen et al., 2003; Lowe et al., 2008 Rask et al., 2015; Steinbrecher & Hiller, 2011;).

International guidelines for the management of PPS recommend a stepped care approach to address the entire severity spectrum of PPS. First, patients with self-limiting symptoms should be managed by general practitioners (GPs) in routine care. Second, patients with persistent symptoms should be offered specific treatment by GPs, if necessary in close collaboration with specialists. Third, patients with severe or chronic PPS should be offered treatment in specialised care (Rosendal, Christensen, et al., 2013; Schaefert et al., 2012; van der Feltz-Cornelis et al., 2012). Although this stepped care model is recommended, it lacks sufficient implementation (Kohlmann et al., 2018). Many GPs find the management of patients with PPS challenging and time-consuming (Rask et al., 2016), and many patients are left without a proper explanation for their symptoms and an adequate symptom management plan (Houwen et al., 2017; Olde Hartman et al., 2009). Furthermore, patients with PPS rarely receive specialised care due to limited resources and capacity (Brostrom, 2019; Kohlmann et al., 2018).

Evidence regarding functional somatic disorders points towards positive effects of cognitive behavioural therapy (CBT) and graded exercise therapy (GET) (Henningsen et al., 2018). Traditional CBT is a structured approach to the identification and modification of maladaptive thoughts and behaviours through cognitive restructuring and behavioural techniques (e.g., exposure). Third-wave CBTs such as Acceptance and Commitment Therapy (ACT) and mindfulness based therapy are more concerned with the individual's awareness of and relationship to inner processes (sensations, thoughts, emotions) than with their content (Frostholm & Rask, 2019). GET implies the stabilisation of physical activity accommodated to current capacity followed by planned gradual increases in physical activity or exercise with the aim to improve physical ability (Bavinton et al., 2004). While traditional CBT, ACT and GET are directed at behaviour change, mindfulness based therapy has no explicit focus on behavioural activation nor on modification (Frostholm & Rask, 2019).

General practice is the first point of contact in many healthcare systems. Hence, as specified by the stepped care approach, it is the front door to treatment for PPS. However, traditional CBT, third-wave therapies and GET rely on specially trained health care professionals, e.g., mental health specialists and physiotherapists. Moreover, such therapy is time consuming and resource demanding and is not available for mild or moderate cases of PPS in general practice. Studies of CBT- and CBT-associated treatments, such as the Extended Reattribution and Management Model, delivered by GPs have shown effects on GPs’ attitudes and behaviour (Rosendal et al., 2005). Yet, no documented effect has been shown on patient outcomes (Rosendal, Blankenstein, et al., 2013). Thus, there is clearly a need for new ways to optimise early treatment of PPS.

Self-help interventions with no or minimal support by healthcare professionals may be a viable solution to ensure improved early treatment. Technological advances allow for new delivery methods that diminish the financial and logistic barriers and support the provision of treatment services that are otherwise accessible only in specialised care settings (Bower et al., 2001; Cuijpers et al., 2010). A recent review of self-help interventions for PPS compared to usual care and waiting list controls found the use of self-help to reduce symptom severity and improve quality of life (van Gils et al., 2016). These findings have been supported by Vugts and colleagues, who reported small positive effects of eHealth interventions in patients with chronic pain and functional somatic syndromes (Vugts et al., 2018).

To guide the development of new self-help interventions for early PPS, it would be helpful to identify the ‘active ingredients’ or the ‘discrete components’ that should be considered in future interventions in order to achieve improved patient health, higher quality of life and lower societal costs (Davidoff et al., 2015; Michie et al., 2011). The Behaviour Change Techniques Taxonomy version1 (BCTTv1) was developed by Michie et al. as a means to facilitate accurate description of techniques applied in behaviour change interventions and to support the creation of new effective interventions. The BCTTv1 is a hierarchically structured coding scheme, including 93 behaviour change techniques (BCTs) that are observable, replicable and unreducible components of an intervention designed to alter behaviour (Michie et al., 2013). Vugts et al. applied the BCTTv1 on eHealth interventions for chronic pain and functional somatic syndromes and found indications of higher effects with regard to the use of exposure techniques only. However, BCTs are rarely applied as stand-alone techniques, and further descriptions of interventions according to their BCTs are needed to help better understand potentially active ingredients and determinants of the success or failure of behaviour change interventions in PPS.

The aim of this systematic review of self-help interventions for PPS was twofold. First, we aimed to identify and describe the behavioural change components of interventions including BCTs. Second, we aimed to explore the possible associations between the BCTs and the effects of the interventions on physical health, mental health and healthcare utilisation.

Methods

The review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Liberati et al., 2009). The Behavioural Change Wheel and the Behavioural Change Techniques were used as a framework for the deconstruction of the interventions and for the coding and analyses of their components (Michie et al., 2011). The review was registered in the International Prospective Register of Systematic Reviews (PROSPERO) database (registration number: CRD42019124462).

Selection criteria

We included randomised controlled trials targeting self-help interventions for PPS and meeting the following PICO criteria (Liberati et al., 2009): Participants were adults aged ≥18 years with PPS defined as non-acute symptoms of at least three months’ duration. PPS included functional symptoms, functional somatic syndromes and functional somatic disorders. Only participants with a non-specific primary diagnosis (except for functional syndromes and somatoform disorders) were eligible. Interventions should be self-help interventions delivered as text (in print or online), audio recordings, video recordings, or a combination of the former. They should be applicable as individual stand-alone treatments, i.e., without or with minimal assistance from a self-help group or a healthcare professional or therapist. We allowed for support by a healthcare professional or therapist at start-up and completion; throughout the intervention, any support or guidance should be of a standardised and non-therapeutic nature, e.g., email or telephone feedback following a pre-specified protocol. Furthermore, the interventions should target the individual (i.e., self-help group interventions were excluded) and be applicable independent of medications or any specialised equipment (e.g., wearable technology and spa). Comparators should be less profound active interventions compared with the primary intervention of interest (e.g., comparison condition could be without therapist guidance or without interactive parts offered in the primary intervention), treatment as usual (TAU) or waiting list control conditions. To be eligible, the studies should report on at least one of the following outcomes: symptom burden, mental disorder, quality of life, healthcare utilisation (contacts or costs) or sickness absence (Rief et al., 2017). Furthermore, the studies should be published in English as a full-body article in a peer-reviewed scientific journal.

Search strategy

The search strategy was developed in collaboration with a research librarian following the recommendations from the Cochrane Collaboration (Higgins & Green, 2011). The databases PubMed, EMBASE, PsycINFO and CENTRAL were searched from inception until 3 April 2019. Medical Subject Headings (MeSH), Emtree Terms and Psychological Index Terms were applied for searches in PubMed, PsycINFO, EMBASE and the Cochrane Central Register of Controlled Trials (CENTRAL) database and combined with free-text searches.

The search strategy included frequent symptoms in general practice for which an organic explanation is often lacking, including musculoskeletal problems, dyspepsia, headache, fatigue and dizziness (Rosendal et al., 2015). Key terms were identified and adapted from recent Cochrane reviews (Cramp et al., 2013; Fishman et al., 2011; Furlan et al., 2015; Pittayanon et al., 2018; Veys et al., 2016). Furthermore, the search strategy included key terms adapted from reviews on the management of functional somatic symptoms and disorders (Henningsen et al., 2018; Schaefert et al., 2012) and on self-help interventions for functional somatic symptoms and mental health disorders (Mayo-Wilson & Montgomery, 2013; van Gils et al., 2016). To identify randomised controlled trials, we used an adapted version of the ‘Cochrane Highly Sensitive Search Strategy for identifying randomised trials in MEDLINE: sensitivity- and precision-maximizing version’ (Higgins & Green, 2011). For a full description of the PubMed search strategy, see Appendix 1. Corresponding searches were conducted in PsycINFO, EMBASE and CENTRAL.

Screening procedure

All identified records were independently assessed for eligibility by two reviewers (MTR and MR or LF). First, title and abstract were screened. Records that were clearly ineligible were removed. Second, full text articles of the remaining records were retrieved and assessed, and a final decision was made on inclusion or exclusion of the study (MTR and TBG). Throughout the review process, discrepancy between reviewers was resolved by consensus. If consensus could not be reached in the abstract screening process, the record was passed on to full text assessment. Discrepancy between reviewers that could not be resolved by consensus in the assessment of the full text article was resolved by involving a third reviewer (LF or MR). The references were managed and the process tracked by the use of RefWorks and Covidence.

Data extraction

A data extraction form was developed, pilot-tested a priori and adapted to Covidence. Two reviewers independently extracted data (MTR and TBG or MWP). Subsequently, discrepancies were resolved through discussion (MR and MWP). We retrieved data on study description, including year of publication, country and setting (e.g., specialised care or general population). The methodological data included study design, number of included participants, number of groups, number of follow-up points, time to follow-up and number lost to follow-up. Data on sample characteristics included type of PPS (e.g., chronic pain, tinnitus, IBS or fibromyalgia), age, gender and symptom duration or time since diagnosis. Data on intervention characteristics included delivery method, duration, whether any assistance was provided by a healthcare professional or therapist, theoretical or therapeutic approach, main content of the intervention and the BCTs applied (further described below). For each included study, we defined one trial arm as the primary active intervention (also if the study included two or more arms). The primary intervention was defined as such by the authors. Alternatively, in case of juxtaposed intervention pairs, the most intensive one, or the one representing the most recent theoretical or therapeutic approach, was preferred, whereas the other was considered an active control group. Outcome data included mean and standard deviation on any of the defined outcomes at post-treatment or follow-up. Follow-up time points were defined as ≤3 month follow-up or >3 month follow-up to be comparable across studies. If more follow-up time points had been applied, the results of the follow-up closest to three months were extracted, i.e., if an outcome had been evaluated at both 6-month and 12-month follow-up, only the data for the 6-month follow-up were extracted.

Coding of behavioural change techniques

The BCTTv1 was used to classify the applied behaviour change techniques. This hierarchically structured coding scheme includes 93 BCTs clustered in 16 overarching techniques that serve one or more intervention functions (education, persuasion, incentivisation, coercion, training, restriction, environmental restructuring, modelling or enablement) (Michie & Atkins, 2014). Intervention functions are broad categories of means by which an intervention may change behaviour. BCTs are defined as the hypothesised active component of an intervention designed to change behaviour, and it is the irreducible or discrete component that delivers the intervention functions (Michie et al., 2011; Michie & Atkins, 2014).

BCT coding assumptions

Prior to the BCT coding, two overall assumptions were made. First, we defined behaviour within the area of PPS to include both objectively observable activities and covert behaviour, i.e., behaviour that is apparent only to the person engaged in or experiencing it, including thoughts, feelings and physical sensations (Ramnerö & Törneke, 2008). Second, we assumed that the overall target behaviour across studies were self-management of symptoms, and we coded BCTs accordingly (no matter whether target behaviours were more specifically described as e.g., fatigue management, pain management, living values or coping).

We coded the BCTs in the primary intervention and control groups separately, and we extracted only the BCTs exclusively applied in the intervention group. Thus, only BCTs that were present in the intervention and absent in the control condition were considered for analysis (Samdal et al., 2017).

BCT coding procedure and manual development

The BCT coding was independently done by MTR and SHH, who both finished an online training course on the BCTTv1 (http://www.bct-taxonomy.com) prior to data extraction and coding. The coding of the BCTs was based on the intervention content as described in the retrieved full-text papers and (if accessible) in project protocols or other published materials. The coding was performed by the use of nVivo (https://www.qsrinternational.com/nvivo/nvivo-products).

The BCTTv1 gives detailed definitions of each of the 93 BCTs and provides examples of each technique. However, these guiding examples refer mainly to healthy lifestyle habits, whereas they are less compatible with behaviour change techniques reflecting psychological approaches and covert behaviour. Therefore, to aid the future refinement of BCTTv1, we developed supplemental coding rules and examples. Initially, five studies were selected for independent BCT coding, and the output by the two coders was compared. The two coders and LF, who is an experienced clinical psychologist and expert in the field of PPS, discussed all disagreements to reach consensus on the coding procedure. Extended descriptions, clarifications and specific examples were added to the BCT coding manual. Finally, all included studies were coded according to the BCT taxonomy. All discrepancies were resolved by consensus or by involving a third reviewer (LF), and further refinement of the coding manual was done.

Based on discussions between coders and the third reviewer during discrepancy resolution, we adapted the coding rules and added examples for 25 BCTs, including specifications concerning when to code and when not to code. Furthermore, we found it necessary to construct a new technique named ‘self-regulation of attention’ (11.5), which was added to the techniques describing regulation (see Appendix 2). As coding by the BCTTv1 is rather new to interventions within the field of PPS, we preferred consensus over formal agreement statistics (Presseau et al., 2015). Thus, the codebook was developed in several iterations; the coding of all interventions was revised until the two coders and the reviewer (SHH, MTR and LF) agreed on a final version, and all interventions were coded independently by the two coders following this final version and subsequently compared.

Risk of bias assessment

The quality assessment of the included studies was performed independently by MWP and PRJ. The Cochrane revised risk-of-bias tool for randomised trials (RoB 2) was used. This tool is structured into five domains, and it includes evaluation of bias arising from the randomisation process, bias due to deviations from the intended intervention, bias due to missing outcome data, bias in the measurement of the outcome, and bias in selection of the reported results (Sterne et al., 2019). Any disagreements between reviewers were resolved by a third reviewer (MR).

Analysis

Outcome categorisation

Each extracted outcome was categorised according to one of six mutually exclusive outcome clusters: (1) symptom burden, (2) anxiety, (3) depression, (4) quality of life including physical functioning and psychological well-being, (5) healthcare utilisation and (6) sickness absence (Appendix 3). The categorisation of outcomes was done by MTR and independently reviewed by MR. Disagreement was resolved by consensus.

Several of the included studies applied multiple outcome variables that could be categorised within the same outcome cluster. To include only one outcome within an outcome cluster for each study, we defined a hierarchy within each cluster.

Symptom burden: If provided, we extracted a symptom burden index generated across a number of outcome variables (e.g., headache index) or alternatively, an average outcome for symptom burden (e.g., average pain). Multidimensional measures (e.g., indexes) were preferred over single-item measures (e.g., pain severity measured with visual analogue scale (VAS)).

Anxiety and depression: Outcomes on anxiety and depression were extracted. Measures of general mental health that could not be defined as either anxiety or depression were excluded.

Quality of life: Measures of quality of life included both physical functioning and psychological well-being due to overlap in the domains encompassed and the questionnaire items applied. A summarised score was preferred. However, if this was not provided, outcome measures within single domains related to physical functioning were preferred over mental health. If two comparable quality-of-life measures were provided, symptom- or syndrome-specific measures were preferred over general quality of life measures.

Healthcare utilisation: If total healthcare utilisation (including total healthcare costs) was provided, we extracted this information. If only healthcare utilisation according to specific healthcare areas was available, we extracted information on hospital inpatient and outpatient visits and medication. We prioritised overall healthcare costs over healthcare costs related to specific symptoms or disorders (e.g., overall healthcare costs instead of healthcare costs related to low back pain).

Sickness absence: Sickness absence data on sickness days were extracted. We prioritised sickness absence in general over sickness absence related to specific symptoms or disorders (e.g., sickness absence days instead of sickness absence days due to low back pain).

Statistical analysis of intervention effect and index of potential

First, standardised mean differences (SMDs) were calculated for each study. Second, we investigated the direction of effect in the included studies. Third, we investigated the overall effect of the interventions. Finally, based on the effect of the interventions, we calculated an index of potential representing the potential effect for each BCT. The analyses were based on a vote counting method (Higgins & Green, 2011). Data were analysed separately for interventions compared with passive control groups and interventions compared with active control groups.

For each outcome within a study and for each time point (post-treatment, ≤3 month follow-up and >3 month follow-up), we calculated the SMD with corresponding 95% confidence intervals (CI) using Hedges g in order to adjust for small sample bias (Egger et al., 2001). The SMDs were graphically displayed using a forest plot.

To investigate the direction of effect in the included studies, a dichotomous variable for effect direction was created. The effect direction was deemed to favour the intervention if the calculated SMD was >0 and to favour the control group if the SMD was ≤0 without taking the 95% CI into consideration. Overall, evidence was considered to support self-help interventions for persistent physical symptoms when the proportion of studies favouring the intervention (SMD > 0) was statistically significantly different from 0.5.

To investigate the potential effect of the BCTs, a dichotomous variable based on the size of effect of the interventions was created. Following the general interpretation of SMD, interventions with SMD ≥0.2 representing small effect within an outcome cluster were defined as effective (Egger et al., 2001). If the SMD was <0.2, the intervention was defined as not effective. Overall, self-help interventions were considered effective when the proportion of effective interventions (SMD ≥ 0.2) was statistically significantly different from 0.5.

The influence of risk of bias on effect direction (proportion of studies favouring the intervention) and intervention effect (proportion of interventions with effect) with corresponding 95% CI were graphically illustrated according to risk of bias distribution.

In order to estimate the contribution of the BCTs to effectiveness, we calculated an index of potential for each technique. The index of potential represented the ratio of the number of times each BCT served as a component in an effective intervention (SMD ≥0.2) within an outcome cluster divided by the number of times the BCT served as a component in all trials within that outcome cluster (Gardner et al., 2017). A BCT was rated as having potential if the index of potential was statistically significantly different from 0.5, i.e., if the component was applied in more effective studies than in non-effective studies.

Estimates of proportions and index of potential were presented with 95% Bayesian credible interval (BCI) based on Jeffrey's interval for a binomial proportion, which is recommended for small samples (n ≤ 40) (Brown et al., 2001). We used an exact two-sided test for binomial random variables to test if (1) the proportion of studies favouring the intervention (SMD > 0), (2) the proportion of effective interventions (SMD ≥ 0.2) and (3) the index of potential for each BCT were equal to or different from 0.5. The alpha level was set at p < 0.05. Due to the different approaches used for the estimation of the 95% BCI and the p-value, these estimates did not always correspond. In these cases, we based our interpretation on the p-value. Analyses were done with Stata version 17.0.

Results

Literature search

We identified 8897 records in the four databases. Duplicates were removed, leaving 5127 records to be screened on title and abstract. Subsequently, 200 records were left for full-text screening. The final dataset comprised 54 papers (see Appendix 4) meeting the inclusion criteria and reporting on 51 trials (Figure 1).

Figure 1. PRISMA flow diagram illustrating the identification, screening and inclusion of studies.

Risk of bias assessment

Of the 51 included studies, only one was a high-quality study with low risk of bias across all evaluated domains (Figure 2). Thirty-four studies gave rise to some concern as to overall bias, whereas 16 studies were rated as having high risk of overall bias. Of the latter, 10 were rated as having high risk within one domain, four within two domains and two studies within three domains. The most frequent source of bias was bias in the measurement of the outcome. Across studies, subjective outcomes reported by the participants were used. A general concern was that outcomes could have been influenced by knowledge of the assigned intervention, apart from five studies of which four were single- or double-blinded.

Figure 2. Risk of bias assessment based on five domains and overall bias as defined by the revised Cochrane risk-of-bias tool for randomised trials (n = 51).

Study characteristics

The sample size mean was 137 in the included studies (n = 51). The sample size range was 24–855. The majority of the studies had been conducted in the US (n = 14, 27.5%) followed by Sweden (n = 9, 17.6%), the Netherlands (one including Belgium) (n = 6, 11.8%) and the UK (n = 4, 7.8%). Besides Europe, other nations represented were Canada, Australia, New Zealand, Taiwan and India (Table 1).

Table 1. Characteristics of included RCTs and intervention descriptions.

Type of PPS	Recruitment setting	Main inclusion criteria	Sample size	Females %	Age, y (SD)	Duration of symptoms/disorder, y (SD)	Intervention description	Methodological approach	Mode of delivery	Duration	Guidance	Control condition(s)	Follow-up points extracted
Amirova et al. (2017); UK
Fibromyalgia	Regional support groups	Diagnosis according to ACR criteria	IG: 67, ACG: 66, PCG: 58	93.7	48.1 (11.1)/50.5 (10.8)/49.0 (10.1)	11.6 (7.0)/14.1 (9.6)/11.0 (6.77)	The Mitchell Method Relaxation Technique including daily auditory guided imagery, muscular relaxation and breathing exercises guided by an audio recording (5 min 20 sec).	Mind-body theory	Online including written instructions and an guided audio recording	1 month	None	Attention control group, WL	Post treatment, 1 m FU
Andersson et al. (2002); Sweden
Tinnitus	General population and patient associations	≥6 mo tinnitus plus seen a general practitioner or ear, nose and throat physician	IG: 53, PCG: 64	46.0/48.0	48.5 (12.3)/47.2 (15.0)	6.2 (5.6)/6.4 (6.8)	Self-help programme including 10 components presented in six modules on a weekly basis, including homework and weeekly reports.	CBT	Website and e-mail	6 weeks	Participants received an encouraging e-mail when submittng a weekly report.	WL	Post treatment
Andersson et al. (2003); Sweden
Headache (migraine and tension type)	General population	Diagnosis according to IHS criteria	IG: 24, ACG: 20	81.8	40.3 (n.r.)	0.5–1 y: 4.5%/1–5 y: 31.8%/6–10 y: 11.4%/ > 10 y: 52.3%	Self-help programme with six weekly treatment modules, including information, applied relaxation and exercises. Weekly report cards for the exercises were used.	CBT	Website and e-mail	6 weeks	Weekly telephone support by therapist to forster adherence (lasting on average 10 min).	Self-help programme without telephone support	Post treatment
Berman et al. (2009); US
Chronic pain	Community settings	≥55 y plus ≥1 d/mo with functional impairment or average pain ≥3 (0–10)	IG: 52, PCG: 37	87.2	65.8 (range 55–91)	n.r.	Intervention including self-help content and exercises covering abdominal breathing, relaxation, expressive writing, creative visual expression and positive thinking delivered in six modules.	Mind-body theory	Website and e-mail	6 weeks	None	WL	Post treatment
Buhrman et al. (2004); Sweden
Chronic back pain	General population	≥3 mo pain plus been in contact with a physician	IG: 22, PCG: 29	62.7	43.5 (10.3)/45.0 (10.7)	8.8 (6.8)/10.5 (11.2)	Self-help intervention including psychological components, streching and physical exercises.	CBT	Website and telephone support	6 weeks	Weekly telephone support by CBT trained therapists (graduate students) to review homework, answer questions, provide reminders and maintain motivation. Supervision provided by clinical psychologist and physiotherapist.	WL	Post treatment
Buhrman et al. (2015); Sweden
Chronic pain with comorbid depression or anxiety	Specialised care and general population	Medical investigation(s), ≥ 3 mo pain plus depression >10 points (MADRS-S)	IG: 28, ACG: 24	85.0	50.7 (12.7)	3 mo–1 y: 12%/1–5 y: 21%/5–10 y: 13%/ > 10 y: 54%	Treatment programme including information, assignments and exercises. The first and the last part of the programme was the same for all participants, whereas parts of the programme focusing on behavioural activation and content about worry and insomnia were individualised. The full programme consisted of eight sections.	CBT	Website and e-mail	8 weeks	Weekly written feedback was given by CBT trained therapists (graduate students) and included positive reinforcement and replies to inquiries. Supervision provided by clinical psychologist.	Moderated online discussion forum	Post treatment
Carpenter et al. (2012); US
Chronic low back pain	General population	≥6 mo pain plus pain ≥4 (0–10)	IG: 70, PCG:71	83.0	42.5 (10.3)	8.6 (7.8)	Online ‘Wellness workbook’ consisting of a mind/body treatment rationale, pain education and CBT tecniques, including cognitive restructuring, stress management, relaxation training, mindfulness and values-based behavioural activation. Content was presented in six chapters, each taking 1–1.5 h to complete.	Enhanced CBT	Website	3 weeks	None	WL	Post treatment
Chalder et al. (1997); UK
Chronic fatigue	General practice	Fatigue ≥4 for >6 mo	IG: 70, PCG:80	74.7	36.0 (10.1)/35.0 (7.5)	n.r.	Self-help booklet, including information about fatigue, diary-keeping, cognitive and behavioural techniques and sleep management.	CBT	Booklet	n.r.	The self-help booklet was initiated by a nurse in general practice.	TAU	Post treatment
Chhabra et al. (2018); India
Chronic low back pain	Specialised care	Pain ≥5 (NPRS) for >12 weeks plus undergoing treatment	IG: 45, PCG: 48	n.r.	41.4 (14.2)/41.0 (14.2)	1.9 (1.8)/2.3 (2.1)	‘Snapcare’ app providing daily activity goals based on patient health status, activity of daily living and daily activity progress.	n.r.	App	12 weeks	None	TAU	Post treatment
Chiauzzi et al. (2010); US
Chronic back pain	General population, specialised care and patient associations	Pain ≥10 days/mo for ≥3 mo	IG: 104, ACG: 105	67.7	46.1 (12.0)	n.r.	Online self-management tailored through a recommendation engine that matched self-reported user chracteristics with lessons, interactive tools, personalised assessments and articles. Content presented in two weekly session (8 sessions in total), each taking at least 20 min to complete.	CBT	Website	Primary intervention: 4 weeks, booster period: 5 website visits per month during FU	None	Text-based back pain guide	Post treatment, 3 m FU, 6 m FU
Devineni and Blanchard (2005); US
Chronic headache (migraine and tension type)	General population	≥1 y pain plus self-reported medical diagnosis	IG: 39, PCG: 47	88.0/79.0	43.6 (12.0)/41.0 (11.8)	n.r.	Online behavioural regimen composed of progressive relaxation, limited biofeedback with autogenic training and stress management.	Mind-body theory	Website	4 weeks	None	WL	Post treatment
Dowd et al. (2015); Ireland
Chronic pain	Individuals with pain listed on a research database	≥6 mo pain	IG: 62, ACG: 62	90.3	44.5 (12.3)	10.8 (9.0)	Online ‘Mindfulness in action’, including a psychoeducation component, a mindfulness practice focus and a cognitive and behavioural change component. Content was presented in 12 sessions, each taking approx. 20 min to complete.	Mindful-ness	Online programme, audio recordings and e-mail	6 weeks	None	Written pain management psychoeducation delivered by e-mails	Post treatment, 6 m FU
Friedberg et al. (2016)/Meng and Friedberg (2017); US
Chronic fatigue syndrome	Specialised care and patient associations	≥6 mo fatigue plus 4 secondary symptoms	IG: 44, ACG: 45, PCG: 48	88.3	47.0 (10.8)/48.0 (12.4)/50.0 (11.3)	13.7 (7.0)/12.6 (8.5)/17.3 (10.7)	Fatigue self-management booklet (also available as audio CD) paired with relaxation audio CD, daily paper diary and step counter.	CBT	Booklet, audio CDs, paper diary and step counter	12 weeks	None	Fatigue self-management booklet (also available as audio CD), relaxation audio CD, daily online web diaries and actigraphs, TAU	Post treatment, 12 m FU
Friesen et al. (2017); Canada
Fibromyalgia	General population, general practice and patient associations	Prior diagnosed by physician, > 3 mo pain plus depression ≥5 (PHQ-9) or anxiety ≥5 (GAD-7)	IG: 30, PCG: 30	95.0	48.0 (11.0)	16.0 (10.0)	Online pain course consisting of lessons, homework assignments and additional resources.	CBT	Website and automated weekly e-mails	8 weeks	Weekly telephone contact (5–10 min) with therapist (doctorate-level graduate student trained in chronic pain) to summarise content, answer questions, reinforce progress and encourage practice. Supervision provided by registered psychologist.	WL	Post treatment
Gallagher et al. (2013); Australia
Chronic pain	Specialised care	>3 mo pain related impairment	IG: 40, ACG: 39	60.8	42.0 (11.0)/45.0 (11.0)	2.1 (1.6)/2.6 (1.7)	Booklet with metaphors and stories to help understand the biology of pain. Content presented in 80 pages divided in 11 sections.	Psycho-education	Booklet	3 weeks	None	Booklet with advice about managing pain	Post treatment, 3 m FU
Geraghty et al. (2018); UK
Low back pain	General practice	Current pain plus consulted general practitioner within 6 mo	IG: 29, ACG:30, PCG: 28	61.4	59.3 (10.4)/54.5 (13.7)/60.3 (16.3)	<1 y: 45.8%/1–5 y: 32.6%/6–10 y: 10.8%/ > 10 y: 10.8%	Online ‘SupportBack’ intervention focusing on self-regulatory processes, including goal setting, self-monitoring and tailored feedback to support physical activity. Content presented in 6 sessions.	n.r.	Website and e-mail	6 weeks	Three phone calls by senior physio-therapist (15–30 min, 1 h total) to support and encourage participants to use the SupportBack intervention.	Online SupportBack intervention without physio-therapist support, TAU	Post treatment
Hagatun et al. (2018); Norway
Insomnia	General population	Diagnosis according to DSM-IV criteria	IG: 95, ACG: 86	67.0	44.9 (13.0)	<1 y: 15.0%/1–5 y: 22.0%/6–10 y: 34.0%/ > 10 y: 28%	Online ‘SHUTi’ intervention, including sleep hygiene, sleep restriction, stimulus control and cognitive restructuring. Content presented in six modules and includes personalised treatment recommendations.	CBT	Website	9 weeks	None	Patient education website	Post treatement
Hesser et al. (2012); Sweden^a
Tinnitus	General population	>6 mo tinnitus, prior diagnosed by physician plus tinnitus distress ≥38 (THI)	IG: 35, ACG1: 32 ACG2: 32	43.4	48.5 (14.7)	0.8 (0.7)	Online treatment including text- and picture-based self-help material, exercises and homework. Content presented in eight modules.	ACT	Website	8 weeks	Asynchronous online communication with CBT/ACT trained therapist (psychologist or psychology students). Mean time spent per participant each week was 9 min, including response to messages and administration. Supervision was provided.	CBT with therapist feedback, online discussion forum	Post treatment
Hunt et al. (2015); US
Irritable bowel syndrome	Specialised care and patient associations	Prior diagnosed by physician plus meeting Rome III IBS criteria	IG: 32, PCG: 28	83.3	36.0 (15.7)/36.0 (12.6)	n.r.	Workbook for irritable bowel syndrome, including psychoeducation, relaxation training, cognitive model, avoidance behaviour and exposure. Content presented in 9 chapters.	CBT	Booklet or link to online version	6 weeks	None	WL	Post treatment
Irvine et al. (2015); US
Low back pain	General population	Pain within past 3 mo plus no impairment	IG: 199, ACG: 199, PCG: 199	60.0	n.r.	n.r.	Online FitBack intervention providing low back pain education and behavioural strategies to manage current pain and future pain episodes.	CBT	Website and e-mail	8 weeks	None	8 e-mails with links to 6 websites ith information about low back pain, TAU	Post treatment, 4 m FU
Janse et al. (2016); The Netherlands
Chronic fatigue	Specialised care	≥6 mo severe fatigue plus impairment 450–700 (SIP8) and/or <4 secondary symptoms	IG: 50, PCG: 50	68.0	median (range): 37.0 (18.0-63.0)/31.0 (18.0–62.0)	median (range): 5.5 (1.0–61.0)/4.0 (0.5-26.0)	Booklet with self-instructions based on the assumptions that fatigue-related cognitions and behaviour perpetuate fatigue. Content presented in 13 modules.	CBT	Booklet and e-mail	6 months	Cognitive behavioural therapists responded to fortnightly e-mails of the patients explaning certain aspects of the treatment relevant for the individual patient or suggesting how to proceed with the treatment. Supervision was provided.	WL	Post treatment
Jasper et al. (2014)/Conrad et al. (2015); Germany^b
Chronic tinnitus	General population, specialised care and patient associations	≥6 mo tinnitus plus tinnitus distress ≥18 (THI) or ≥8 (Mini-TQ)	IG: 41, ACG: 44	37.6	51.3 (9.8)/52.1 (9.0)	9.2 (7.9)/8.0 (7.1)	Online treatment combining CBT techniques with relaxation exercises. Content presented in 12 mandatory and 6 optional modules, each including general information, suggestions for exercising, worksheets and solutions for common problems.	CBT	Website	10 weeks	Weekly e-mail communication with therapist (clinical psychologist). Therapist spent max 10 min per week per patient. Supervision was provided.	Discussion forum	Post treatment
Jernelöv et al. (2012); Sweden
Insomnia	General population	Diagnosis according to AASM research criteria	IG: 44, ACG: 45, PCG: 44	82.0	47.9 (13.9)	11.8 (12.0)	Self-help intervention with two parts. The first part consists of psychoeducation about sleep and insomnia. The second part presents techniques, e.g., relaxation and visualisation techniques, sleep restriction and stimulus control, cognitive restructuring techniques, and sleep hygiene.	CBT	Booklet, CD, digital treatment guide and sleep diary	6 weeks	Weekly telephone contact with therapist (graduate student) following a structured guide (each call max. 15 min). Supervision provided experienced clinician.	Self-help booklet and CD without therapist support, WL	Post treatment, 3 m FU
Johnston et al. (2010); New Zealand
Chronic pain	Specialised care	Referred to a pain clinic	IG: 12, PCG: 12	62.5	Median (range): 43.0 (20.0–84.0)	n.r.	Self-help workbook ‘Living beyond your pain’ focusing on relationship with pain, values, thoughts, mindfulness and committed actions, including exercises. Content presented in eight chapters.	ACT	Booklet	6 weeks	None	WL	Post treatment
Kaldo et al. (2007); Sweden
Tinnitus	General population and specialised care	≥6 mo tinnitus, prior medical examination by specialist plus tinnitus distress ≥10 (TRQ)	IG: 34, PCG: 38	48.6	45.9 (13.0)/48.5 (15.7)	8.6 (8.4)/12.4 (11.7)	Self-help book, including introduction, treatment tools, evaluation and maintenance, with focus on tinnitus, noise sensitivity and hearing impairments.	CBT	Booklet	6 weeks	Brief weekly telephone calls by therapist (graduate student) focusing on treatment progress and advice on how to move on. Mean time spent per patient was 107 min. Supervision provided by clinical psychologists.	WL	Post treatment
Knoop et al. (2008); The Netherlands
Chronic fatigue syndrome	Specialised care	Diagnosis according to CDC criteria	IG: 85, PCG: 86	79.3	37.6 (10.0)/38.5 (10.6)	Median (range): 6.0 (1.0–35.0)/8.0 (1.0–35.0)	Self-instruction booklet containing information about chronic fatigue syndrome and weekly assignments.	CBT	Booklet	At least 16 weeks	Patients e-mailed or phoned a cognitive behavioural therapist fortnightly to report on progress and received a response.	WL	Post treatment
Lee et al. (2014); Taiwan
Bladder pain syndrome/interstitial cystitis	Specialised care	IC/BPS	IG: 40, PCG: 40	n.r.	46.5 (10.2)/49.5 (11.8)	n.r.	Online web service asking patients to check and follow diets and lifestyles suggested by physicians and providing a Q/A SMS application for management of symptom flares.	n.r.	Website and SMS application	8 weeks	None	TAU	Post treatment
Lee et al. (2018); Taiwan
Bladder pain syndrome/interstitial cystitis	Specialised care	Diagnosis of IC/BPS according to AUA	IG: 30, PCG: 30	100.0	42.9 (10.4)/46.3 (14.2)	n.r.	Online service providing 21 video clips for weekly health education and symptom flare management and weekly check for diets and lifestyles.	n.r.	App and reminders (via e-mail, SMS, voice call, or social network apps)	8 weeks	None	TAU	Post treatment
Lin et al. (2017); Germany
Chronic pain	Specialised care	≥6 mo pain plus interference ≥2 (0–10)	IG: 100, ACG: 101, PCG: 101	84.1	41.7 (13.1)	9.5 (10.1)	Online ‘ACTonPain’ programme comprising modules focusing on information, assignments, relevant metaphors, and mindfulness exercises. Content presented in eight modules, each taking approx. an hour to complete.	ACT	Website and e-mail feedback	8 weeks	Personalised and standardised e-mail feedback from e-coaches (psychologists). Time spent on guiding per participant averaged 105 min. Supervision provided by psychological psychotherapist.	ACTonPain programme without e-coaching, WL	Post treatment, 6 m FU
Ljotsson et al. (2010); Sweden
Irritable bowel syndrome	Specialised care	Prior diagnosed by physician plus meeting Rome III IBS criteria	IG: 43, ACG: 43	84.7	34.6 (9.4)	6.3 (7.3)	Self-help manual based on exposure and mindfulness exercises with guidance. The manual was divided into five steps: (1) A rationale for the treatment and instructions on mindfulness, (2–4) Presentation of a psychological model of IBS and continued mindfulness exercises, (5) Exposure exercises and finally, relapse prevention.	Enhanced CBT	Website and closed online discussion forum	10 weeks	Asynchronous correspondence with CBT trained therapist (graduatestudent) providing weekly feedback. Time spent per participant averaged 165 min. Supervision provided by psychologist.	Online discussion forum and permission to contact student therapist for general support	Post treatment
Lorenz et al. (2019); Switzerland, Austria & Germany
Insomnia	General population and specialised care	Insomnia severity ≥8 (ISI)	IG: 29, PCG: 27	69.6	41.7 (17.3)/44.0 (20.1)	n.r.	Online treatment ‘mementor somnium’ focusing on psychoeducation, sleep restriction, relaxation, sleep hygiene, cognitive restructuring and changing sleep-related behaviours. Content presented in six sessions with guidance by an animated sleep coach.	CBT	Website	6 weeks	None	WL	Post treatment
Lorig et al. (2008); US^c
Arthritis and Fibromyalgia	General population	Diagnosed with osteoarthritis, rheumatoid arthritis or fibromyalgia	IG: 433, PCG: 422	89.8/90.5	52.2 (10.9)/52.5 (12.2)	n.r.	Online arthritis self-management programme consisting of interactive instructions, bulletin board discussions, tools such as exercise logs, medication diaries and tailored exercise programmes and arthritis helpbook containing all of the programme content. Participants were expected to spend 3–6 h each week on programme content.	CBT	Website, booklet and e-mail	6 weeks	Trained online peer moderaters reminding participants to log on, modelling action planning and problem solving, offering encouragement and posting to the bulletin boards.	TAU	12 m FU
Malouff et al. (2010); Australia
Tinnitus	General population, specialised care and patient associations	Tinnitus	IG: 84, PCG: 78	49.0/39.7	57.3 (13.7)/57.8 (13.3)	n.r.	Self-help book ‘Tinnitus: A self-Management Guide for the Ringing in Your Ears’ focusing on tinnitus education, cognitive reappraisal and restructuring, relaxation exercises, stress management tecniques, attention control techniques, use of self-instruction and lifestyle changes.	CBT	Booklet	2 months	None	WL	Post treatment
Menzies et al. (2006); US
Fibromyalgia	Specialised care	Diagnosis of fibromyalgia plus functional impairment >20 (FIQ)	IG: 24, PCG: 24	97.9	49.6 (10.5)	n.r.	Gudied imagery involving guided imagery audiotapes adressing familiarity with relaxation and imagery, the signal breath relaxation and imagery of a pleasant scene and reinforcement of the signal breath conditioning for relaxation. Tapes were practiced daily for two weeks.	Guided imagery	Audiotapes	Primary intervene-tion: 6 weeks; Follow-up period (4 weeks): Use of any of the tapes once a day	None	TAU	Post treatment, 10 weeks FU
Mulet et al. (2007); US
Masticatory myofascial pain and forward head posture	Specialised care and general population	Meet research diagnostic criteria for TMD, > 6 mo pain plus pain >4 (0–10)	IG: 20, ACG: 22	95.2	24.2 (10.0)	5.4 (3.9)	Self-care programme including counselling, patient education, reassurance about symptoms and encouragement to rest the masticatory muscles combined with Rocabado's 6 × 6 exercise programme, including 6 exercises to be performed 6 times a day repeated 6 times each.	n.r.	Written explanations and instructions	4 weeks	Treatment initiated by a physical therapist who conducted treatment review after 1 week.	Self-care programme without Rocabado's 6 × 6 exercises	Post treatment
Oerlemans et al. (2011); The Netherlands
Irritable bowel syndrome	General practice and patient associations	≥3 mo IBS according to ICPC or Rome III criteria	IG: 38, PCG: 38	91.9/76.9	35.9 (11.7)/40.6 (15.5)	n.r.	Digital diaries three times daily assessing quality of sleep, severity of complaints, intended activities for the day (mornings), severity of complaints, accomplished activities, cognitions and feelings (afternoons) and severity of complaints, satisfaction with activity level and achievements of the day (evenings).	CBT	Personal digital assistant (PDA)	4 weeks	Based on diaries and following a standardised protocol, situational feedback from psychologists was provided via SMS. Supervision provided by psychologist/senior researcher.	TAU	Post treatment, 3 m FU
Onieva Zafra et al. (2013); Spain
Fibromyalgia	Patient associations	≥3 y fibromyalgia diagnosis	IG: 30, PCG: 30	96.4	51.6 (6.3)	>3 y: 7.3%/>5 y: 29.1%/>10 y: 4.8%/>15 y: 21.8%	Music therapy (classical music mixed with salsa music) in two phases, each with initial face-to-face music therapy and with subsequent CDs for self-administration.	Music therapy	CD	4 weeks	Intervention phases initiated by 1 h face-to-face sessionwith researcher.	TAU	Post treatment
Peters et al. (2017); The Netherlands & Belgium
Chronic musculoskeletal pain	General population and patient associations	>3 mo pain	IG: 117, ACG: 116, PCG: 51	85.0	48.6 (12.0)	12.8 (10.1)	‘Happy despite pain’ intervention, including self-compassion exercises, ‘three good things’ exercises, savouring techniques, writing about and imagining the ‘best possible self’ and maintenance plan. Content presented in eight modules.	Positive psychology	Website and telephone and e-mail support	8 weeks	Forthnightly semi-structured telephone calls by graduate/newly graduated psychology students (mean duration of each call 15–20 min) and, forthnightly semistandardised e-mails to encourage to continue the programme and share any problems encountered.	iCBT including relaxation, stretching exercises, cognitive restructuring, coping techniques, body scan exercises and relapse prevention, WL	Post treatment
Rantonen et al. (2014)/Rantonen et al. (2016); Finland^d
Low back pain	Worksite	Pain intensity 10–34 (0–100) plus ≥2 weeks pain/y or ≥2 recurrent episodes/y	IG: 90, ACG: 92, PCG: 83	21.0/34.0/24.0	45.0 (8.0)/43.0 (7.0)/45.0 (8.0)	12.0 (9.0)/11.0 (7.0)/14.0 (9.0)	‘Back book’ based on a biopsychosocial model, focusing on LBP attitudes, inappropriate LBP behaviour, coping, avoiding LBP re-exacerbation and emphasing the importance of getting back to normal activities.	Psycho-education	Booklet	n.r.	Initial 20 min face-to-face book presentation and advice by occupational health nurse trained in LBP.	‘Back book’ without advice, natural course control group	Post treatment, 6 m FU, 12 m FU, 24 m FU
Riva et al. (2014); Switzerland
Chronic back pain	Specialised care	≥3 mo pain	IG: 27, ACG: 24	50.0/51.9	44.0 (13.6)/51.0 (14.1)	7.9 (7.2)/9.3 (8.7)	The ‘ONESELF website’ containing static features including a library, a first aid section and a frequently asked questions section and interactive features including a virtual gym, a testimonials and commentaries section, a weekly action plan and a quiz game.	n.r.	Website and SMS reminders	8 weeks	None	Static features of ‘ONESELF website’	Post treatment
Robinson et al. (2006); UK
Irritable bowel syndrome	General practice	Prior or current diagnosis of IBS and attending general practitioner	IG: 139, ACG: 141, PCG: 140	89.0	40.0 (14.4)	6.0 (7.2)	Self-help guide book containing information about lifestyle, diet and pharmacological and alternative therapies based on up-to-date evidence and patients’ anecdotal experiences.	n.r.	Booklet	n.r.	One group meeting (2 h) facilitated by trial coordinator.	Self-help guide book without group meeting, TAU	12 m FU
Sanders et al. (2007); US
Irritable bowel syndrome	Specialised care	Prior diagnosed by physician using Rome II IBS criteria	IG:17, PCG: 11	71.4/88.9	56.9 (14.6)/41.8 (10.3)	16.3 (15.0)/15.0 (12.7)	Self-help book ‘Breaking the bonds of irritable bowel syndrome’ containing review of the digestive system, diagnostic medical tests, diet, medical and psychological treatments and describing several exercises to assist identification of foods, situations, emotions and thoughts influencing IBS and providing CBT techniques to deal with difficult situations. Content presented in 5 modules.	CBT	Booklet and work sheets sent by e-mail	10 weeks	None	WL	Post treatment
Ström et al. (2000); Sweden
Headache	General population	≥6 mo pain plus headache frequency ≥1/week	IG: 20, PCG: 25	68.9	41.5 (range 23–62)/39.2 (23–61)	0.5–1 y: 6.7%/1–5 y: 35.6%/ > 5 y: 55.6%	Internet headache treatment based on applied relaxation, autogenic training and problem-solving training. Participants registered the frequency and duration of practising relaxation, and the results were reported once each week.	n.r.	Website and e-mail	6 weeks	None	WL	Post treatment
Townsend et al. (2001); US
Temporo-mandibular disorder	General population	Pain in the past year plus 1 of 5 secondary symptoms	IG: 14, ACG: 10	100.0	35.4 (9.5)/38.9 (8.2)	7.5 (0.8)/9.5 (2.5)	Self-help manual incorporating treatment rationale, awareness-training exercises, including deep breathing and progressive relaxation training, habit detection procedures, the use of competing responses to prevent or interrupt the target behaviour and relapse prevention. Content presented in 7 lessons.	Mind-body theory	Written materials and audiotape	App. 10 weeks	Weekly therapist (advanced graduate student) contact via e-mail or telephone (mean telephone contacts 9/mean length each 4 min) to answer questions and assist with problem solving.	Waiting list condition with weekly telephone contact to monitor pain levels and ensuring intervention enrolment	Post treatment
Trompetter et al. (2015); The Netherlands
Chronic pain	General population and online patient platforms	Pain ≥3 days/week for ≥6 mo plus pain intensity ≥4 (0–10)	IG: 82, ACG: 79, PCG: 77	76.8/75.9/75.3	52.9 (13.3)/52.3 (11.8)/53.2 (12.0)	>5 y: 58.5%/69.6%/61.0%	‘Living with pain’ intervention, including experiential exercises, metaphors and mindfulness exercises to be practised daily. Content presented in nine modules.	ACT	Website and e-mail	9–12 weeks	Structured e-mail counselling once a week provided by recently graduated psychology students to guide, encourage and support the process. Supervision by a registered cognitive behavioural therapist.	Expressive writing with structured e-mail correspondence once a week, WL	Post treatment, 6 m FU
Tummers et al. (2012); The Netherlands
Chronic fatigue syndrome	Community health centre	Diagnosis according to CDC criteria	IG: 62, PCG: 61	78.0	36.3 (12.1)/36.4 (13.6)	median (range): 4.0 (0.5–38.7)/5.0 (0.5–52.1)	Self-instruction material with information and assignments including establishment of treatment goals, precipitation and perpetuating factors, reducing the focus on fatigue, establishing a sleep routine, establishing an adequate activity pattern, illness beliefs, plan for work resumption, modifying expectations to social environment, increasing mental and social activities and relapse prevention. Content presented in 15 chapters.	CBT	Booklet	20 weeks	Forthnightly e-mail correspondance with psychiatric nurse trained in coaching. Supervision by a cognitive behavioural therapist.	WL	Post treatment
Vallejo et al. (2015); Spain^b
Fibromyalgia	Specialised care	Diagnosis according to ACR criteria	IG: 20, PCG: 20	100.0	51.6 (9.9)	8.7 (7.3)	Programme, including psychoeducation, relaxation training, emotional training, adjusting daily activities, insomnia and sexual dysfunctions, problem solving, cognitive restructuring attentional control, illness behaviours, cognitive processing and memory, and relapse prevention. Content presented in 10 sessions.	CBT	App	10 weeks	Digital messaging with therapist (junior therapist) to verify understanding of main topics, reinforce weekly schedule and resolve any problems. Supervision by a cognitive behavioural therapist.	WL	Post treatment
van Ittersum et al. (2014); Belgium
Fibromyalgia	Specialised care	Diagnosis according to ACR criteria	IG: 53, ACG: 52	93.0	46.5 (9.3)	median (range): 8.5 (1.0–30.0)/8.0 (0.1–25.0)	Pain neuroscience education using a cognitive approach to restructure thoughts and reconceptualise pain, including explanations and illustrations about pain physiology, mechanisms of central sensitisation, and tips and tricks on how to incorporate the information in daily life.	Psycho-education	Booklet	6 weeks	One telephone call from therapist (researcher) to provide opportunity for clarification, verify understanding and providing further information, patient allowed to ask questions twice by telephone or e-mail contact.	Written relaxation education, including exercises and instructions	Post treatment, 6 m FU
Vincent and Lewycky (2009); Canada
Chronic insomnia	Specialised care and general population	≥ 6 mo insomnia ≥4 nights/week plus daytime impairment	IG: 59, PCG: 59	66.9	n.r.	n.r.	Online treatment programme including psychoeducation, sleep hygiene, stimulus control, relaxation training, sleep restriction, hypnotic medication tapering and introduction to cognitive therapy. Content presented in five modules.	CBT	Website	5 weeks	None	WL	Post treatment, 1 m FU
Williams et al. (2010); US
Fibromyalgia	General practice and specialised care	Diagnosis according to ACR criteria	IG: 59, PCG: 59	95.0	50.5 (11.5)	9.4 (6.5)	Online ‘Living well with fibromyalgia’ behavioural self-management intervention, including educational lectures on fibromyalgia, educational, behavioural and cognitive skills to help with symptom management and behavioural and cognitive skills to facilitate adaptive life style changes. Content presented in 13 modules.	CBT	Website	6 months	None	TAU	Post treatment
Wilson et al. (2015); US
Persistent pain	General practice and general population	>3 mo pain plus current prescription(s) for opioids	IG: 57, PCG: 57	78.0	49.3 (11.6)		Online ‘Chronic pain management program’ consisting of learning modules related to cognitive, behavioural, social and emotional determinants and regulation and including exercises, relaxation and self-monitoring.	CBT	Online programme	8 weeks	None	TAU	Post treatment

AASM = American Academy of Sleep Medicine, ACG = active control group, ACR = American College of Rheumatology, ACT = acceptance and commitment therapy, AUA = American Urological Association, BPS = Bladder Pain Syndrome, CBT = cognitive behavioural therapy, CDC = Centers for Disease Control and Prevention, CFS =Chronic Fatigue Syndrome, DSM-IV = Diagnostic and Statistical Manual for Mental Disorders, fourth edition, FIQ = Fibromyalgia Impact Questionnaire, FU = follow-up, GAD = Generalised Anxiety Disorder, HIS = International Headache Society, IBS = Irritable Bowel Syndrome, IC = Interstitial Cystitis, ICPC = International Classification of Primary Care, IG = intervention group, ISI = Insomnia Severity Index, LBP = low back pain, n.r. = not reported, MADRS-S = Montgomery Åsberg Depression Rating Scale, Mini-TQ = Mini-Tinnitus Questionnaire, NPRS = Numeric Pain Rating Scale, PCG = passive control group, PHQ = Patient Health Questionnaire, SIP = Sickness Impact Profile, TAU = treatment as usual, THI = Tinnitus Handicap Inventory, TMD = Temporomandibular Disorder, TRQ = Tinnitus Reaction Questionnaire, WL = waitlist.

^a Two active interventions were compared with a less intensive ACG. Only data for the primary intervention compared with the less intensive ACG were retrieved.

^b One study arm assigned to group CBT was excluded from the review.

^c Only data on patients with fibromyalgia was retrieved.

^d Outcome measures based on registerdata were retrieved at 12 months FU or 24 month FU.

The studies included individuals recruited from general practice (n = 6, 11.8%), specialised care (n = 24, 47.1%), general practice and specialised care (n = 1, 2%), the general population (n = 12, 23.5%), community settings (including a community health centre) (n = 2, 3.9%), patient associations and support groups (n = 4, 7.8%), a work site (n = 1, 2%) and a research database (n = 1, 2%). Several studies recruiting patients from healthcare settings combined this recruitment method with online and newspaper advertisements.

The included studies represented a broad array of PPS. Pain (e.g., musculoskeletal pain and headache) and pain disorders, including fibromyalgia, were the most frequently studied (n = 28, 54.9%). These were followed by IBS (n = 5, 9.8%), tinnitus (n = 5, 9.8), chronic fatigue/CFS (n = 5, 9.8%) and insomnia (n = 4, 7.8%). Less frequently, bladder pain syndrome (n = 2, 3.9%) and temporomandibular disorder (n = 2, 3.9%) were studied.

Women were overrepresented in the majority of studies. However, six studies (11.8%) included more men than women; five studies focused on tinnitus and one concerned forestry workers with low back pain.

In 40 of the included studies (78.4%), the design comprised two arms. Of these, the primary intervention was compared to TAU or a waiting list control group (passive control condition) in 28 cases. Thus, in 12 of the studies with two arms, the intervention was compared to an active control condition. Of the 11 studies reporting a three-armed trial design (21.6%), 10 compared the primary intervention with both an active and a passive control condition, and one investigated two active interventions and compared these with a less intensive active control conditions.

Characteristics of primary interventions

The majority of primary interventions were delivered in an online format, e.g., a website or an app (n = 32, 62.7%) (Table 1). In 15 interventions (29.4%), the main content was delivered in a booklet format. In two cases (3.9%), a booklet was combined with online material. In two interventions (3.9%), the main content was delivered by audiotapes or CDs. The duration of the interventions ranged from 3 weeks up to 6 months. Almost half of the studies (n = 25, 49.0%) reported an intervention period of six or eight weeks.

More than half of the primary interventions were based on CBT (n = 28, 54.9%). Two of these interventions applied enhanced CBT, including mindfulness exercises and values-based behavioural activation adopted from acceptance and commitment therapy (ACT). Other approaches were ACT (n = 4, 7.8%), mind–body theory, including relaxation and breathing exercises (n = 4, 7.8%), psychoeducation (n = 3, 5.9%), mindfulness (n = 1, 2.0%), guided imagery (n = 1, 2.0%), positive psychology (n = 1, 2.0%) and music therapy (n = 1, 2.0%). Eight primary interventions (15.7%) lacked description of the applied theoretical or methodological approach to behaviour change and could not be classified.

Of the primary interventions evaluated, 27 (52.9%) were guided by a healthcare professional or therapist, whereas 24 (47.1%) were pure self-help. Guidance was delivered by graduate students (n = 7), psychologists/licensed therapists (n = 6), graduate students or psychologists (n = 2), nurses (n = 3), physiotherapists/physical therapists (n = 2) and peers (n = 1). In six studies, guidance was done by therapists or researchers without further specifications. Six of the primary guided interventions were compared with a similar non-guided intervention, i.e., differences between intervention groups and active controls could be ascribed to social support.

Overall intervention effects according to the selected outcomes

Direction of effect

We investigated the direction of effect of the primary interventions according to the outcome categories. For symptom burden, anxiety, depression and quality of life at post-treatment, the direction of the effect favoured the interventions more frequently than it favoured the TAU or waiting list conditions (p < 0.001–0.002). Compared with active control conditions, only for quality of life at post-treatment a positive effect direction was reported by more studies than not (p = 0.008) (see Appendix 5, Table a). For all other time points and for the outcome clusters of health care utilisation and sickness absence, studies were few and no differences as to number of studies favouring the interventions and studies favouring the control groups could be detected.

Studies defined as having effect

Figures 3 and 4 depict SMDs with regard to symptom burden and quality of life. At post-treatment, 71.9% (n = 23) of the studies comparing the intervention with a passive control condition had an SMD ≥0.2 for symptom burden, which was statistically significantly more studies than studies having an SMD <0.2 (p = 0.02). The same was evident for quality of life at post-treatment where 78.8% (n = 26) of the studies were found to be effective (p = 0.001). No other statistically significant differences as to the criterion of SMD ≥0.2 were found (see Appendix 5, Table b).

Figure 3. Standardised mean difference (95% CI) of the effect of the primary interventions compared with passive control groups on symptom burden at post-treatment (n = 32).^a

Figure 4. Standardised mean difference (95% CI) of the effect of the primary interventions compared with passive control groups on quality of life at post-treatment (n = 33).

Risk of bias influence

We investigated whether risk of bias may have influenced the results. As can be seen from Figure 5, studies with high risk of bias may have led to overestimation of both the proportion of studies favouring the interventions and the proportion of interventions with effect (SMD ≥0.2) with regard to symptom burden when intervention groups were compared with active controls. However, no consistent pattern was seen across outcome categories, time points (post-treatment, ≤3 month follow-up, >3 month follow-up) and comparison groups (see Appendix 6).

Figure 5. Proportion of studies (95% CI) favouring the intervention (SMD > 0) and proportion of interventions (95% CI) with effect (SMD ≥ 0.2) at post-treatment, ≤3 month follow-up and >3 month follow-up according to risk of bias distribution. SMD = standardised mean difference.

Potential effect of behaviour change techniques

Fifty-five BCTs were applied in at least one intervention (Figure 6). The most frequently applied BCT in interventions compared with a passive control condition was instruction on how to perform the behaviour. In interventions compared with an active control condition, the most frequently used BCT was behavioural practice or rehearsal. For distribution of BCTs according to theoretical approach, see Appendix 7.

Figure 6. Frequency of behaviour change techniques applied in studies comparing the primary intervention with an active control group (n = 23) and/or a passive control group (n = 38).

The index of potential with corresponding 95% BCI for each BCT within an outcome cluster was calculated for studies with passive controls and active controls, respectively. The results for BCTs applied in at least five studies within an outcome cluster are shown in Tables 2 and 3. The number of studies measuring health care utilisation and sickness absence was too low to be included in the tables.

Table 2. Effect (SMD> = 0.2) of interventions and index of potential of BCTs (95% BCI) applied in each symptom cluster in studies with passive control groups (n = 38)^a,b,c.

	Symptom burden					Anxiety					Depression					Quality of life
	Effect, n	All, n	Index of potential	95% BCI	p	Effect, n	All, n	Index of potential	95% BCI	p	Effect, n	All, n	Index of potential	95% BCI	p	Effect, n	All, n	Index of potential	95% BCI	p
Post treatment
Problem solving	11	14	0.79	(0.53;0.94)	0.057	5	9	0.56	(0.25;0.83)	1.000	8	11	0.73	(0.43;0.92)	0.227	12	16	0.75	(0.51;0.91)	0.077
Goal setting (outcome)	10	12	0.83	(0.56;0.96)	0.04	5	7	0.71	(0.35;0.94)	0.453	5	7	0.71	(0.35;0.94)	0.453	13	13	1.00	(0.82;1.00)	<0.001
Action planning	9	14	0.64	(0.38;0.85)	0.424	4	8	0.50	(0.20;0.80)	1.000	6	9	0.67	(0.35;0.90)	0.508	13	16	0.81	(0.58;0.94)	0.021
Feedback on behaviour	6	9	0.67	(0.34;0.90)	0.508						3	5	0.60	(0.21;0.91)	1.000	7	10	0.7	(0.39;0.91)	0.344
Self-monitoring of behaviour	18	24	0.75	(0.56;0.89)	0.023	5	10	0.50	(0.22;0.78)	1.000	7	13	0.54	(0.28;0.78)	1.000	19	24	0.79	(0.60;0.92)	0.006
Self-monitoring of outcome(s) of behaviour	6	8	0.75	(0.41;0.94)	0.289						4	5	0.80	(0.37;0.98)	0.375	7	9	0.78	(0.46;0.95)	0.180
Feedback on outcome(s) of behaviour	3	5	0.60	(0.21;0.91)	1.000											3	5	0.60	(0.21;0.91)	1.000
Social support (unspecified)	10	17	0.59	(0.36;0.79)	0.629	5	7	0.71	(0.35;0.94)	0.453	6	10	0.60	(0.31;0.85)	0.754	13	18	0.72	(0.49;0.89)	0.097
Instruction on how to perform behaviour	23	31	0.74	(0.57;0.87)	0.011	7	14	0.50	(0.26;0.74)	1.000	12	18	0.67	(0.44;0.85)	0.238	25	32	0.78	(0.62;0.90)	0.002
Information about health consequences	20	24	0.83	(0.65;0.94)	0.002	5	10	0.5	(0.22;0.78)	1.000	8	12	0.67	(0.39;0.88)	0.389	20	26	0.77	(0.58;0.90)	0.009
Information about social consequences	14	17	0.82	(0.60;0.95)	0.013	4	7	0.57	(0.23;0.86)	1.000	6	9	0.67	(0.35;0.90)	0.508	16	19	0.84	(0.64;0.95)	0.004
Information about emotional consequences	16	19	0.84	(0.64;0.95)	0.004	5	8	0.63	(0.29;0.88)	0.727	6	10	0.60	(0.31;0.85)	0.754	17	20	0.85	(0.65;0.96)	0.003
Demonstration of the behaviour	7	9	0.78	(0.46;0.95)	0.180	3	5	0.6	(0.21;0.91)	1.000	4	5	0.8	(0.37;0.98)	0.375	8	9	0.89	(0.59;0.99)	0.039
Social comparison																5	5	1.00	(0.62;1.00)	0.063
Prompts/cues	10	16	0.63	(0.38;0.83)	0.454	3	5	0.60	(0.21;0.91)	1.000	4	7	0.57	(0.23;0.86)	1.000	10	16	0.63	(0.38;0.83)	0.454
Associative learning	5	6	0.83	(0.44;0.98)	0.219											6	6	1.00	(0.67;1.00)	0.031
Behavioural practice/rehearsal	16	25	0.64	(0.44;0.80)	0.230	7	13	0.54	(0.28;0.78)	1.00	10	16	0.63	(0.38;0.83)	0.454	21	26	0.81	(0.63;0.92)	0.002
Graded tasks	3	7	0.43	(0.14;0.77)	1.000											6	8	0.75	(0.41;0.94)	0.289
Credible source	13	19	0.68	(0.46;0.86)	0.167	5	8	0.63	(0.29;0.88)	0.727	7	11	0.64	(0.35;0.86)	0.549	14	19	0.74	(0.52;0.89)	0.064
Reduce negative emotions	8	9	0.89	(0.59;0.99)	0.039											8	10	0.80	(0.50;0.96)	0.109
Self-regulation of attention	11	18	0.61	(0.38;0.81)	0.481	5	8	0.63	(0.29;0.88)	0.727	6	9	0.67	(0.35;0.90)	0.508	15	18	0.83	(0.62;0.95)	0.008
Distraction	4	5	0.8	(0.37;0.98)	0.375											6	7	0.86	(0.50;0.98)	0.125
Adding objects to the environment	8	10	0.80	(0.50;0.96)	0.109	0	6	0	(0;0.33)	0.031	4	8	0.50	(0.20;0.80)	1.000	9	10	0.90	(0.62;0.99)	0.021
Body changes	16	22	0.73	(0.52;0.88)	0.052	4	11	0.36	(0.14;0.65)	0.549	8	14	0.57	(0.32;0.80)	0.791	18	24	0.75	(0.56;0.89)	0.023
Framing/reframing	16	21	0.76	(0.55;0.90)	0.027	7	13	0.54	(0.28;0.78)	1.00	10	15	0.67	(0.42;0.86)	0.302	20	23	0.87	(0.69;0.96)	<0.001
Valued self-identity																5	5	1.00	(0.62;1.00)	0.063
> 3-month follow-up
Self-monitoring of behaviour																4	5	0.8	(0.37;0.98)	0.375
Instruction on how to perform behaviour	3	5	0.6	(0.21;0.91)	1.000											4	5	0.8	(0.37;0.98)	0.375
Information about health consequences	3	5	0.6	(0.21;0.91)	1.000											4	5	0.8	(0.37;0.98)	0.375

BCT = behavioural change technique, BCI = Bayesian credible interval.

^a Only results for BCTs applied in at least five studies within an outcome cluster are included in the table.

^b A two-sided test for binomial random variables was used (alpha level p < 0.05).

^c Note that the 95% BCI and the p-value from the exact test do not always correspond. In the interpretation, the p-value was prioritised.

Table 3. Effect (SMD> = 0.2) of interventions and index of potential of BCTs (95% BCI) applied in each symptom cluster in studies with active control groups (n = 21)^a,b,c,d.

	Symptom burden					Anxiety					Depression					Quality of life
	Effect, n	All, n	Index of potential	(95% BCI)	p	Effect, n	All, n	Index of potential	(95% BCI)	p	Effect, n	All, n	Index of potential	(95% BCI)	p	Effect, n	All, n	Index of potential	(95% BCI)	p
Post treatment
Problem solving	5	5	1.00	(0.62;1.00)	0.063						5	5	1.00	(0.62;1.00)	0.063	6	6	1.00	(0.67;1.00)	0.031
Goal setting (outcome)																4	5	0.8	(0.37;0.98)	0.375
Action planning	3	5	0.60	(0.21;0.91)	1.000						5	5	1.00	(0.62;1.00)	0.063
Feedback on behaviour	4	5	0.80	(0.37;0.98)	0.375						4	5	0.80	(0.37;0.98)	0.375	5	5	1.00	(0.62;1.00)	0.063
Self-monitoring of behaviour	4	7	0.57	(0.26;0.86)	1.000											4	6	0.67	(0.28;0.92)	0.688
Social support (unspecified)	3	5	0.60	(0.21;0.91)	1.000											5	7	0.71	(0.35;0.94)	0.453
Instruction on how to perform behaviour	3	5	0.60	(0.21;0.91)	1.000						4	5	0.80	(0.37;0.98)	0.375	5	7	0.71	(0.35;0.94)	0.453
Information about health consequences																5	5	1.00	(0.62;1.00)	0.063
Information about emotional consequences																3	5	0.60	(0.21;0.91)	1.000
Demonstration of the behaviour	2	6	0.33	(0.08;0.71)	0.688
Prompts/cues	3	6	0.50	(0.17;0.83)	1.000											4	5	0.80	(0.37;0.98)	0.375
Behavioural practice/rehearsal	5	10	0.50	(0.22;0.78)	1.000	4	5	0.80	0.37;0.98	0.375	5	6	0.83	(0.44;0.98)	0.219	7	9	0.78	(0.46;0.95)	0.180
Credible source	5	7	0.71	(0.35;0.94)	0.453						4	5	0.80	(0.37;0.98)	0.375	6	8	0.75	(0.41;0.94)	0.289
Reduce negative emotions																4	5	0.80	(0.37;0.98)	0.375
Self-regulation of attention	3	6	0.50	(0.17;0.83)	1.000						4	5	0.80	(0.37;0.98)	0.375	6	8	0.75	(0.41;0.94)	0.289
Body changes	3	6	0.50	(0.17;0.83)	1.000											3	5	0.60	(0.21;0.91)	1.000
Framing/reframing	4	5	0.80	(0.37;0.98)	0.375						4	5	0.80	(0.37;0.98)	0.375	6	7	0.86	(0.50;0.98)	0.125

BCT = behavioural change technique.

^a Only results for BCTs applied in at least five studies within an outcome cluster are included in the table.

^b In two studies (Andersson et al., 2003 & Friedberg et al., 2016), the applied BCTs were used in both the intervention and the active control condition, i.e., these studies did not contribute to the analyses of potentially effective BCTs.

^c A two-sided test for binomial random variables was used (alpha level p < 0.05).

^d Note that the 95% BCI and the p-value from the exact test do not always correspond. In the interpretation, the p-value was prioritised.

A total of 15 BCTs showed potential as treatment components with positive effect on outcomes at post-treatment in studies comparing the interventions with passive control conditions (Table 2). Of these BCTs, eight showed potential as treatment components to reduce symptom burden (index of potential range 0.74–0.89, p = 0.004–0.04) and 14 to increase quality of life (index of potential range 0.75–1.00, p ≤ 0.001–0.039). Seven BCTs, i.e., goal setting (outcome), self-monitoring of behaviour, instruction on how to perform behaviour, information about health, social and emotional consequences and framing/reframing, showed potential as treatment components to positively influence outcomes of both symptom burden and quality of life (see explanation of BCTs below). For the outcomes of anxiety, depression, health care utilisation and sickness absence, no BCTs were found potentially effective as treatment components. Rather, one BCT, adding objects to the environment, was applied in six studies measuring anxiety without being able to show any effect, and thus this BCT seemed to be related to less effect (index of potential = 0, 95% BCI: 0;0.33, p = 0.031).

In trials comparing the primary intervention with an active control group, only one BCT showed potential effect as treatment component at post-treatment (Table 3); problem solving had been applied in six studies that consistently showed positive effect across studies on quality of life (index of potential = 1.00, 95% BCI: 0.67;1.00, p = 0.031).

No BCT showed potential as treatment component with regard to any of the outcome categories at the two follow-up time points (i.e., ≤3 months and >3 months), neither in studies with passive controls, nor in studies with active controls.

Potentially effective behavioural change techniques

Table 4 displays the BCTTv1 (Michie & Atkins, 2014) definition of and the context-specific examples of the 16 BCTs showing index of potential statistically significantly greater than 0.5 in either primary interventions compared with passive control conditions or compared with active control conditions. These BCTs were related to the overarching themes of BCTs: goals and planning, feedback and monitoring, shaping knowledge, natural consequences, comparison of behaviour, associations, repetition and substitution, regulation, antecedents and identity.

Table 4. Definitions (Michie & Atkins, 2014) and examples of behaviour change techniques with potential effect.

BCT	BCT number	BCT cluster	Definition	Examples extracted from included studies	Index of potential according to PCG or ACG
Problem solving	1.2	Goals and planning	Analyse, or prompt the person to analyse, factors influencing the behaviour and generate or select strategies that include overcoming barriers and/or increasing facilitators (includes ‘Relapse Prevention’ and ‘Coping Planning’).	Included was a problem-solving section in which subjects were presented with different methods aimed to be useful in the identification of problems, coping with problems in general, and coping with headache-related problems (Andersson 2003). In the last module, participants were encouraged to develop individualised plans with the purpose of maintaining the skills acquired during treatment (Hesser et al., 2012).	ACG
Goal setting (outcome)	1.3	Goals and planning	Set or agree on a goal defined in terms of a positive outcome of wanted behaviour.	Patients started with formulating their personal goals for therapy. Achievement of these goals should, when attained imply that the patient was recovered (i.e., no longer severely fatigued and no longer disabled by the fatigue) (Janse et al., 2016).	PCG
Action planning	1.4	Goals and planning	Prompt detailed planning of performance of the behaviour (must include at least one of context, frequency, duration and intensity). Context may be environmental (physical or social) or internal (physical, emotional or cognitive) (includes ‘Implementation Intentions’).	Sleep restriction instructions include setting a sleep window to number of hours slept during the previous 4–7 nights based on sleep diary recordings (no lowest limit given), fixating the wake-up time, and increasing the scheduled time in bed with 15 min contingent on sleep efficiency exceeding 85% during the last 4–7 days (Jernelöv et al., 2012).	PCG
Self-monitoring of behaviour	2.3	Feedback and monitoring	Establish a method for the person to monitor and record their behaviour(s) as part of a behaviour change strategy.	Participants are asked to write down everything they have done to reduce/manage their pain and record their behaviours and sensations during increased pain situations (functional analysis) for a week (Lin et al., 2017).	PCG
Instruction on how to perform the behaviour	4.1	Shaping knowledge	Advise or agree on how to perform the behaviour (includes ‘Skills training’).	Each session included a prerecorded presentation designed to build skills associated with mindfulness and instructions on how to cultivate and sustain positive emotional experiences, particularly within social relationships (Dowd et al., 2015).	PCG
Information about health consequences	5.1	Natural consequences	Provide information (e.g., written, verbal, visual) about health consequences of performing the behaviour.	Information regarding headache and the role of psychological factors was provided (Andersson et al., 2003).	PCG
Information about social and environmental consequences	5.3	Natural consequences	Provide information (e.g., written, verbal, visual) about social and environmental consequences of performing the behaviour.	Explain that sometimes people unnecessarily avoid or refrain from all kinds of activities because of their complaints. This so-called avoidance behaviour can aggravate the complaints by causing stress and a vicious circle develops (Oerlemans et al., 2011).	PCG
Information about emotional consequences	5.6	Natural consequences	Provide information (e.g., written, verbal, visual) about emotional consequences of performing the behaviour.	Information about tinnitus, the distress it can lead to, and the aims and methods of CBT for tinnitus (Kaldo et al., 2007).	PCG
Demonstration of the behaviour	6.1	Comparison of behaviour	Provide an observable sample of the performance of the behaviour, directly in person or indirectly e.g., via film, pictures, for the person to aspire to or imitate (includes ‘Modelling’).	Videos used gain-framed messaging delivered by an animated whiteboard-style coach and behaviourally focused live-action instructional videos on ergonomics and exercises (Irvine et al., 2015).	PCG
Associative learning	7.8	Associations	Present a neutral stimulus jointly with a stimulus that already elicits the behaviour repeatedly until the neutral stimulus elicits that behaviour (includes ‘Classical/Pavlovian Conditioning’).	The goal of this technique is to enable you to achieve relaxation in response to a self-produced cue (e.g., saying the word relax) (Malouff et al., 2010).	PCG
Behavioural practice/rehearsal	8.1	Repetition and substitution	Prompt practice or rehearsal of the performance of the behaviour one or more times in a context or at a time when the performance may not be necessary, in order to increase habit and skill.	It encourages readers to practice each of the three different approaches to relaxation and to pick the one (or two) that work best and use them daily (Hunt et al., 2015).	PCG
Reduce negative emotions	11.2	Regulation	Advise on ways of reducing negative emotions to facilitate performance of the behaviour (includes ‘Stress Management’).	The Learning Center content includes (…) methods for managing negative emotions such as anger, fear and depression (Lorig et al., 2008).	PCG
Self-regulation of attention	11.5^a	Regulation	Attention is brought to a direct experience of events in the mind and body by fostering a non-elaborative awareness of thoughts, feelings and sensations as they arise, i.e., rather than getting caught up in ruminative, elaborative thought streams. Includes mental imagery in which neural perceptions are employed to foster change in emotional or physical experience or to transform the imagery content. Includes mindfulness, imagery and visualisation.	The mindfulness instructions included exercises to be practiced daily, aimed at bringing the participant into immediate awareness of current GI symptoms, thoughts, feelings and behavioural impulses (Ljotsson et al., 2010).	PCG
Adding objects to the environment	12.5	Antecedents	Add objects to the environment in order to facilitate performance of the behaviour.	Module 3 presented relaxation training and provided MP3 files for paced breathing, progressive muscle relaxation, imagery-induced relaxation and selfhypnosis (Vincent and Lewycky, 2009).	PCG
Body changes	12.6	Antecedents	Alter body structure, functioning or support directly to facilitate behaviour change.	Applied relaxation: Tension and relaxation of different muscle groups; relaxation through controlled breathing; establishing a relaxation routine in everyday life (Jasper et al., 2014).	PCG
Framing/reframing	13.2	Identity	Suggest the deliberate adoption of a perspective or new perspective on behaviour (e.g., its purpose) in order to change cognitions or emotions about performing the behaviour (includes ‘Cognitive structuring’).	In an adaptive dialogue, different typical thoughts that insomnia patients often have are discussed. The aim is to identify dysfunctional sleep-related thoughts and replace them with sleep-compatible cognitions (Lorenz et al., 2019).	PCG

ACG = active control group, PCG = passive control group.

^a The technique ‘Self-regulation of attention’ was defined by the authors.

Goals and planning

The BCT problem solving (1.2) was used for the coding of intervention content that involved identification of and coping with problems and the development of individualised plans for maintaining the skills learned throughout the intervention. Goal setting (outcome) (1.3) was used for the formulation of personal goals related to treatment. The action planning (1.4) technique involved a broad array of actions, ranging from detailed planning of sleep restriction following a specified plan to behavioural activation plans, the scheduling of pleasant events and specification of committed actions in interventions based on ACT.

Feedback and monitoring

Self-monitoring of behaviour (2.3) involved a method for the patients to journal or record their behaviour and experience, e.g., in the form of an electronic or paper-based diary.

Shaping knowledge, comparison of behaviour and repetition and substitution

Instruction on how to perform the behaviour (4.1) and behavioural practice/rehearsal (8.1) were the most frequently applied BCTs (Figure 6). These two BCTs were often used jointly with demonstration of the behaviour (6.1) involving both patient narratives modelling different aspects of behaviour and visual illustration of exercises (e.g., animations and videos).

Natural consequences

Information about health consequences (5.1) typically involved information about the symptoms or syndromes in scope, e.g., tinnitus, fibromyalgia or IBS. This technique was often delivered as one of the elements in a psycho-educative approach, including information about social and environmental consequences (5.3) and information about emotional consequences (5.6).

Associations

Associative learning (7.8), i.e., the process in which a new response becomes associated with a particular stimulus, included sound enrichment and relaxation in response to a self-produced cue in patients with tinnitus and sleep hygiene in patients suffering from insomnia.

Regulation

Regulation techniques included the BCT reduce negative emotions (11.2) and the new technique described by the authors, self-regulation of attention (11.5). The first primarily represented content supporting the identification of causes of stress and methods directed at reducing stress. The latter was used for mindfulness, imagery and visualisation.

Antecedents

The BCT adding objects to the environment (12.5) was used for interventions that supplied participants with some sort of physical object but was also used for interventions that gave participants access to downloadable mp3 files etc. in order to place online programmes on an equal footing with paper-based programmes supplemented by exercises on CDs or videotapes. The technique body changes (12.6) was used for intervention components directly targeting the body and its function and included breathing exercises, relaxation exercises, sleep restrictions and strength and stretching exercises.

Identity

The technique of framing/reframing (13.2) was applied across different methodological approaches including CBT, ACT, mind–body theory and psychoeducation. This technique was directed at helping the patient to identify maladaptive thinking patterns and distorted core beliefs, to challenge them and to create more helpful alternatives.

Discussion

To our knowledge, this review is the first to address behaviour change components and their potential effect as treatment components in self-help interventions for PPS independently of the delivery method (e.g., text- or internet-based). We included 51 randomised controlled trials, which primarily applied psychological methods across a broad array of PPS, such as chronic pain, fibromyalgia, chronic fatigue/CFS, IBS, insomnia and tinnitus. At post-treatment, interventions overall showed effect on symptom burden and quality of life; however, only in comparison with passive control groups. Sixteen BCTs were found to meet the applied criteria for potential effect. These BCTs were related to goals and planning, feedback and monitoring, shaping knowledge, natural consequences, comparison of behaviour, associations, repetition and substitution, regulation, antecedents and identity.

Discussion of results

We applied an empirically derived taxonomy, the BCTTv1 for behaviour change technique, which has primarily been applied within the field of lifestyle risk behaviour, e.g., smoking, obesity, alcohol consumption and physical inactivity. However, the BCTTv1 is gaining ground in clinical psychological interventions (Dugdale et al., 2016; Pearson et al., 2020). Many of the BCT techniques (e.g., information about health consequences, instruction on how to perform a behaviour, framing/reframing, goal setting and exposure) are integral parts of both traditional CBT and ACT. Yet, in line with a review on stress management apps and an explorative study of techniques used in a treatment programme for substance use disorders (Christmann et al., 2017; Dugdale et al., 2016), we found some adaptations necessary for the coding of techniques applied in the psychological interventions used in the included studies. Therefore, we developed a codebook with supplemental coding rules and guiding examples, including a new category of regulation labelled ‘self-regulation of attention’ to be used for techniques aimed at bringing attention to a direct experience of events in the mind and body by fostering non-elaborative awareness of thoughts, feelings and sensations. Furthermore, we specified how varying therapeutic approaches should be coded. The codebook may lay the ground for future mappings of intervention contents and development of new ways to address PPS.

The majority of studies involved the presentation of didactic material with the aim to support a bio-psycho-social illness understanding and supported patients in practising new skills through instruction on how to perform the wanted behaviour, behavioural practice or rehearsal and by the provision of methods for the patients to self-monitor their behaviour. These are all techniques used within psychoeducation, which may also encompass techniques such as problem solving, goal setting (outcome) and action planning that were less often used, yet was found to be potentially effective as treatment components also (Lukens & McFarlane, 2004). Psychoeducation has been found to reduce symptoms of psychological distress and depression and may alter illness behaviour in patients with persistent physical symptoms (Donker et al., 2009; Frølund Pedersen et al., 2019). However, psychoeducation is primarily one component in a multicomponent treatment; only in three of the included studies, psychoeducation constituted the entire intervention.

The most commonly used overarching therapeutic approaches were traditional CBT and third wave therapies such as mindfulness and ACT. These therapeutic approaches are characterised by certain techniques; thus, the application of specific BCTs will to some extend have been driven by the therapeutic approach chosen. The BCTs framing/reframing, reduce negative emotions, self-regulation of attention and valued self-identity are not exclusively related to distinct therapeutic approaches. Yet, they are at the core of CBT, mindfulness and ACT. While valued self-identity was used in too few studies to investigate the potential effect of this technique with sufficient certainty, the other three techniques could be related to effect on symptom burden and/or quality of life. Traditional CBT and third wave behaviour therapies have shown to be effective in PPS (Henningsen et al., 2018; Ost, 2008; O’Connor et al., 2018). As described, most interventions included a wide range of techniques, and there may be a synergistic effect of these combinations. Thus, the potential effect (or lack of effect) of one BCT may not be explained by that specific component but rather by the pattern of components in which it is embedded (Goodwin et al., 2016; Peters et al., 2015). Therefore, rather than supporting specific therapeutic approaches, our findings point to the importance of addressing cognitions and regulation of emotions and consciousness as well as introducing techniques of information and skill building which are common techniques applicable across a wide range of psychological approaches.

Attrition and non-compliance are two major obstacles to self-guided psychological treatment (Karyotaki et al., 2015). Qualitative studies have suggested that integration of feedback or support in internet-based treatment programmes may be crucial for adherence, especially among extrinsically motivated patients (Heinrich et al., 2016; Wilhelmsen et al., 2013). Several of the included studies provided e-mail or telephone support by a therapist or researcher, but more innovative approaches were also used, such as automatic and standardised text messages (SMS) containing reminders, repetition or motivational support (Lin et al., 2015), or an animated coach asking questions and delivering personalised feedback based on the patient's response (Lorenz et al., 2019). While such strategies could be expected to promote treatment adherence and thereby the effect of the intervention, we did not find the techniques social support (unspecified) and credible source to meet criteria for index of potential on any of the outcomes. Investigations of the effect of therapist guidance and support in interventions for PPS have shown diverging results. One review found that therapist contact did not significantly influence the effect of interventions (van Gils et al., 2016), whereas another review found small effects in interventions without guidance and larger effects in guided interventions, depending on the experience of the therapist (Vugts et al., 2018). Based on our study, we cannot conclude on the causal effect of techniques. Yet, our results did not support the need for therapist guidance in self-help interventions for PPS.

Interventions using associative learning were found to be effective in increasing quality of life in six out of six cases when compared with passive control groups. Apart from one study, associative learning was used solely in studies of tinnitus and insomnia. Despite the fact that associative learning could not be related to reduced symptom burden, we cannot rule out that tinnitus and insomnia may respond better to treatment than symptom disorders with a more extensive bodily expression, e.g., pain conditions or conditions characterised by excessive fatigue and exhaustion, such as CFS. Thus, whether associative learning contributes to or diagnosis primarily account for the positive effect on quality of life is uncertain. In the review by Vugts et al. (2018), the effect sizes of computer-based interventions varied with specific functional syndromes. However, the number of studies addressing specific syndromes was too small to detect any statistically significant difference between the specific diagnostic categories. Nevertheless, in the development of new interventions, the selection of BCTs should be based on both the index of potential and the evidenced effect on the specific condition in question.

Strengths and limitations

We performed an extensive systematic literature search in four databases, building on previous reviews within the field of PPS. This approach was taken to ensure that all relevant studies were identified. Study screening and selection, risk of bias evaluation, data extraction and BCT coding were done independently by two reviewers, and consensus was reached through discussion or involvement of a third reviewer. Furthermore, we adapted and specified coding rules and provided examples for BCTs in a supplemental codebook. We also defined a new technique of self-regulation of attention to allow for a more detailed and specific coding of therapeutic approaches that are frequently taken to patients with PPS. Thus, the adapted codebook ensured more valid coding of the potentially active ingredients in the interventions applied. As the observed effect of an intervention is susceptible to placebo effects and the control group used for comparison, we calculated the index of potential for the interventions by comparing with passive and active control groups separately. Furthermore, we evaluated the index of potential for BCTs at post-treatment and at two follow-up time points in order to be able to discern potential immediate effects from potential long-term effects of the BCTs applied.

Despite these strengths, a number of limitations must be addressed. First, no firm definition of self-help exists. We allowed for support at the beginning or completion of a treatment programme, while all support throughout the treatment period should be of a standardised non-therapeutic nature. However, based on the description of the interventions, it was sometimes difficult to conclude whether the guidance was standardised and non-therapeutic. All inclusion criteria were applied to each study by two independent reviewers, and discrepancies were resolved by consensus or a third reviewer. Furthermore, we based the evaluations of self-help on the time devoted to guidance, whether guidance followed a pre-specified protocol, and whether guidance was face-to-face or based on e-mails or telephone contacts. However, we cannot rule out that others may have reached another conclusion as to whether guidance was therapeutic or not and whether trials should be included or excluded. For example, in a previous review of self-help interventions for PPS, no firm definition of the maximum duration of a contact was applied, and studies using face-to-face contacts for up to four hours were included (van Gils et al., 2016).

Second, the BCT coding of complex psychological interventions such as CBT and ACT may be challenged by the fact that specific features and non-specific factors such as quality of relational interactions may neither be captured by the taxonomy, nor described in the published literature (Michie et al., 2018). We based the coding on published peer-reviewed articles and study protocols, which may have described the interventions applied in limited detail (Dombrowski et al., 2007). We retrieved supplemental material, including protocols referred to, publication of secondary analyses and other publicly available materials (e.g., published manuals), but we did not contact authors for missing data or more detailed information. Instead, we chose a pragmatic approach to the BCT coding; this is a departure from the conservative use prompted by the BCTTv1 training course, but it considers the limited space allowed in scientific papers to describe interventions. On the one hand, this approach may have led to higher numbers of coded BCTs. On the other hand, it may have provided a more detailed picture of how the interventions were conducted in real life.

Third, the studies included in the review were of a varying quality. All studies except one represented high risk, or at least some concern, of overall bias, which reflects lack of high-quality studies. The most frequent source of bias was lack of blinding of patients and therapists with regard to intervention condition. This could limit the validity of the results, as subjective outcomes were applied; participants were thus both the target for treatment and the assessor of the outcome of the treatment. In effect studies of psychological treatments, blinding is often not applicable. A recent meta-analysis found no difference in treatment effect when comparing trials with and without blinding of patients, health care professionals or outcome assessors. Thus, blinding may not be as important as previously believed (Moustgaard et al., 2020). Yet, the effect of the studies included in this review, and thus the results of this review, must be interpreted with reasonably caution.

Fourth, the designs of the included studies did not allow for estimation of the effect of the individual BCTs. Meta-analyses and meta-regressions have been used for estimating the effect of specific BCTs on health risk behaviours (Michie et al., 2018). However, studies addressing PPS are heterogeneous in many ways; the symptoms or functional somatic syndromes investigated, intervention theory and content, intensity, delivery and outcomes evaluated, and this may question the used method (Peters et al., 2015). Furthermore, from a clinical point of view, the contribution of meta-regressions may be limited, especially in consideration of the aim to support the development of future interventions addressing PPS. This method relies on a high number of high-quality studies and in an extensive meta-analysis of 46 studies, Vugts et al. found exposure to be the only BCT associated with effect sizes (Vugts et al., 2018). Instead, we applied a vote counting method used by Gardner et al., (2016) and Martin et al. (2013). We defined effective interventions as interventions in which the SMD was equal to or greater than 0.2 in one of the defined outcome clusters and investigated, whether the specific BCT had been applied in more effective than ineffective studies. This method may be subject to increased probability of the identification of potentially effective BCTs among techniques that are frequently applied, while being unable to differentiate between less effective and less used BCTs. Furthermore, this method does not allow for isolation of the effect of one BCT from the effect of the cluster of BCTs in which it is embedded, nor for determination of the effect of a certain pattern of BCTs (Michie et al., 2018). However, the choice of method allowed for a detailed description of BCTs applied, and the results point to BCTs that are more likely than others to contribute to effects and should be considered for future treatment manuals.

Finally, due to the amount of published research within the field of PPS and the broad approach applied (i.e., including functional somatic symptoms and not limiting the review to a specific functional somatic syndrome or disorder), we chose not to include grey literature. Thus, we cannot rule out the existence of other material that could contribute to the results of the present review.

Conclusion

In this review, we found evidence for the potential positive effect of 16 BCTs as treatment components in self-help interventions for PPS with regard to outcome of physical health including quality of life. Our description of the practical application of BCTs in specific contexts, targeting varying groups of PPS and using a variety of psychological approaches, may inform everyday clinical practice and future management. However, the methods applied allow us neither to draw any firm conclusion on the causal relationships between specific BCTs and effects, nor conclusion on the effects in patients with early PPS in specific. Rather, our results endorse that certain BCTs may contribute to changes in symptom burden and quality of life in patients with PPS. Thus, the findings point to BCTs that should definitely be taken into account, yet without excluding the consideration of others, in the development of new interventions in this field.

Author contributions

MTR, MR and LF initiated and designed the study. MTR did the literature search, study screening, study inclusion, data extraction and BCT coding in collaboration with MR, MWP and SHH. The risk of bias assessment was done by MWP and MR. TBG and PRJ contributed to the screening, data extraction and risk of bias assessment process (see acknowledgements). EØ analysed the data in collaboration with MTR. MTR drafted the manuscript and MR, LF, MWP, SHH and EØ provided critical revision. All authors approved the final version of the manuscript.

Acknowledgements

We are grateful to psychologists Tine Bennedsen Gehrt (TBG), PhD and Pernille Rahbek Jensen (PRJ) for their competent assistance with article screening, data extraction and risk of bias assessment.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

Innovation Fund Denmark (grant number 8056-00040B) and the Tryg Foundation (grant number 7549-07).