Study selection
The search identified 6222 unique publications (Figure 1). Of the 215 screened full-texts, 184 were excluded (see Supplementary Material 3 for list of and reasons for excluding each publication from electronic bibliographic databases). Thirty-one publications were eligible for inclusion [10, 25, 27, 36-37, 41-66].
Developing a preliminary synthesis and exploring relationships between studies
Study characteristics
Reported review types include systematic reviews (n=14), literature reviews (n=6), reviews including meta-analysis (n=6), scoping reviews (n=3), and qualitative evidence syntheses (n=2). The primary research designs reported in the reviews were organised into 4 tiers (Figure 2), including Randomised control designs (number of reviews including this design=26), Non-randomised comparative designs (n=13), Single group observational studies (n=12), and Qualitative studies (n=5). Randomised controlled designs included RCTs (n=26), and CRTs (n=4; see Supplementary Material 4 for study datasets). Non-randomised comparative designs included non-randomised comparative studies (n=6), quasi-experimental design (n=3), cross-sectional studies (n=3), a survey with comparison group (n=1), and a case comparison study (n=1). Single group observational studies included pre-post studies (n=7), descriptive studies (n=6), and surveys without comparison group (n=2). Qualitative studies included unspecified qualitative research (n=3), interviews (n=2), focus groups (n=1), and participant observation (n=1). Some reviews reported primary research designs which could not be categorised or did not appear to be evaluative including feasibility or pilot studies (n=2), an experimental study (n=1), a case study (n=1), and a needs assessment (n=1).
The conditions reviewed included cancer (n=9), diabetes (n=7), cardiovascular disease (n=4), acquired brain injury, cerebral palsy, and spina bifida (n=1), asthma (n=1), kidney disease (n=1), HIV (n=1), and somatic illness (n=1). Six reviews included any chronic condition. The reviews included primary research across 26 countries or semi-autonomous regions (Figure 3). Reviews reported on primary research based in USA (n=24), Australia (n=14), Canada (n=12), the UK (n=11), Republic of Ireland (n=6), Netherlands (n=4), Argentina (n=3), China, Germany, Mozambique, South Korea, Sweden, Uganda, and Vietnam (n=2); Austria, Botswana, Denmark, Hong Kong, Iran, Israel, Jordan, Mali, Philippines, South Africa, Spain, and Taiwan (n=1). Reviews reported on primary research based in community settings (n=8), general hospitals (n=5), outpatient clinics (n=4), HIV/AIDS clinics (n=3), school or other educational settings (n=3), US Veterans Affairs centres (n=3), camps (n=2), primary care settings (n=2), church (n=1), haemodialysis centre (n=1), HIV/AIDS clinical trials unit (n=1), infectious disease hospital unit (n=1), physiotherapy department (n=1), public health clinic (n=1), seniors’ centre (n=1), stroke rehabilitation centre (n=1), University clinic (n=1), workplace (n=1).
Peer support intervention description
Reviews reported on peer support delivered in groups (n=21) and one-to-one (n=17), by telephone (n=21), face-to-face (n=16), and online (n=13). Descriptions of peer support intervention components most frequently comprised education (n=13), self-management techniques (n=9), discussion (n=7), reciprocal support (n=7), sharing personal experiences, and unspecified social support (n=6). Other specific forms of social support were numerous, wide ranging, and are listed in full in Table 1. Peers were consistently defined by their role in providing or exchanging support between people with similar experiences or circumstances, though only seven specified that this similarity extends to first-hand experience of living with a shared chronic condition, and three reviews did not give a definition of what constitutes peer support. The duration of peer support interventions ranged from a single session to 2 years. Most reviews reported on interventions with one contact per week, with a range of daily (for up to 1 week) to 3- and 12-monthly follow-ups.
Peer training content and methods were organised into 11 categories including delivery methods, counselling skills, communication skills, condition and treatment information and adherence, meta-competency and safety, social skills and story sharing, intervention and pedagogical theory, culture and religion, physical training techniques, intervention facilitation skills, and availability of community resources. Training most frequently involved teaching communication skills (n=7), condition and treatment information (n=5; see Supplementary Material 5 for training description table). Five reviews reported whether primary research peer support included the use of an intervention manual.
Eleven reviews reported on the supervision of peer support. Three reviews simply indicated whether supervision was a component of the primary research intervention, and 2 reviews indicated whether supervision was professionally delivered. Six reviews described supervision in more detail including supervision of the first session only (n=1), weekly supervision from a psychologist, nurse, and community outreach coordinator (n=1), a study coordinator called participants to identify problems, including problems with the peer support relationship at week 3 of the intervention (n=1), an endocrinologist, an exercise physiologist and an exercise health psychologist were available to answer questions throughout the study period (n=1), psychologist supervision (n=1), nurse supervision (n=1), facilitators were present to ensure fidelity to the research protocol (n=1), children and young people overseen by adults (n=1).
Four reviews reported on screening practices to recruit peer support workers based on success (e.g. reduction in severity of condition or improved self-management) from past treatment (n=1), previous experience facilitating a group, ability to motivate, good listening and problem-solving skills (n=1), knowledge of diabetes and an interest in helping people and effective communication patterns (n=1), a competency test involving role play (n=1), severity of chronic condition (n=1); from the same community (n=1), demonstration of leadership qualities (n=1), ability to engage in conversation (n=1), give information clearly (n=1), share experiences and display appropriate listening skills (n=1). Twelve reviews reported on peer matching including matching by age (n=5), ethnicity (n=2), cultural similarities (n=2), type of cancer (n=2), life factors (n=1), gender (n=1), primary language (n=1), men who have sex with men (n=1), chronic pain (n=1), injection drug use (n=1), ex-smoker status (n=1).
Table 1
Intervention description table
Intervention components
|
Number of reviews
|
Education
|
13 (42%)
|
Self-management
|
9 (29%)
|
Discussion
|
7 (23%)
|
Reciprocally giving support to others
|
7 (23%)
|
Sharing personal experiences
|
6 (19%)
|
Unspecified social support
|
6 (19%)
|
Medications advice/ adherence
|
5 (16%)
|
Emotional support
|
4 (13%)
|
Activity scheduling/ planning leisure activities
|
3 (10%)
|
Addressing unspecified psychosocial issues
|
3 (10%)
|
Cognitive techniques
|
3 (10%)
|
Community outreach
|
3 (10%)
|
Encouragement
|
3 (10%)
|
Goal setting
|
3 (10%)
|
Mentoring
|
3 (10%)
|
Modelled recovery
|
3 (10%)
|
Addressing physical health concerns
|
2 (6%)
|
Counselling
|
2 (6%)
|
Exercise
|
2 (6%)
|
Monitoring condition/ symptoms
|
2 (6%)
|
Psychoeducation
|
2 (6%)
|
Strategic thinking
|
2 (6%)
|
Talking Circles (native American cultural discussion facilitation)
|
2 (6%)
|
Unspecified psychological support
|
2 (6%)
|
Answered patient questions
|
1 (3%)
|
Bullying support
|
1 (3%)
|
Coping skills
|
1 (3%)
|
Decision making
|
1 (3%)
|
Developing relationships
|
1 (3%)
|
Emphasise personal achievement
|
1 (3%)
|
Encouraging contact with clinicians
|
1 (3%)
|
Lifestyle change exercises
|
1 (3%)
|
Low-level advice
|
1 (3%)
|
Lead group activities
|
1 (3%)
|
Mindfulness
|
1 (3%)
|
Motivational support
|
1 (3%)
|
Needs assessment
|
1 (3%)
|
Outings for social integration and networking
|
1 (3%)
|
Problem-solving
|
1 (3%)
|
Recognising trauma
|
1 (3%)
|
Relaxation techniques
|
1 (3%)
|
Smoking cessation counselling
|
1 (3%)
|
Unspecified behavioural change
|
1 (3%)
|
Unspecified practical support
|
1 (3%)
|
Ways of taking action
|
1 (3%)
|
Outcome domains
Fifty-five outcome domains were identified and organised across 15 categories (Table 2). Mental health and psychosocial processes and adjustment were the most frequently reported outcome categories. Quality of life (n=14), self-efficacy (n=14), clinical surrogates (n=12), depression (n=11), distress (n=11), and health knowledge (n=9) were the most frequently analysed outcome domains in reviews. Psychosocial process and adjustment was the most variably measured category, spanning 21 outcome domains.
Table 2
Outcome domains identified in reviews
Domains
|
Number of reviews
|
Mental health
|
18 (58%)
|
Depression
|
11 (35%)
|
Distress
|
11 (35%)
|
Anxiety
|
7 (23%)
|
Mental health
|
5 (16%)
|
Post-traumatic stress
|
2 (6%)
|
Wellbeing
|
2 (6%)
|
Suicidal ideation
|
1 (3%)
|
Psychosocial processes & adjustment
|
16 (52%)
|
Self-efficacy/ confidence
|
14 (45%)
|
Optimism/ Pessimism/ Hope
|
4 (13%)
|
Coping
|
3 (10%)
|
Empowerment
|
3 (10%)
|
Social coping
|
3 (10%)
|
Adjustment
|
2 (6%)
|
Psychosexual functioning
|
2 (6%)
|
Altruism
|
1 (3%)
|
Catharsis
|
1 (3%)
|
Comfort with clinician
|
1 (3%)
|
Illness uncertainty
|
1 (3%)
|
Motivation to volunteer
|
1 (3%)
|
Negative affect
|
1 (3%)
|
Perceived threat of condition
|
1 (3%)
|
Positive upward comparison
|
1 (3%)
|
Post-traumatic growth
|
1 (3%)
|
Self-understanding
|
1 (3%)
|
Sense of coherence
|
1 (3%)
|
Spirituality
|
1 (3%)
|
Stigma
|
1 (3%)
|
Suppression of affect
|
1 (3%)
|
Quality of life
|
14 (45%)
|
Clinical surrogate*
|
12 (39%)
|
Physical health & function
|
12 (39%)
|
Functional status
|
5 (16%)
|
Health status
|
4 (13%)
|
Physical health
|
4 (13%)
|
Adverse events
|
1 (3%)
|
Social integration & connectedness
|
12 (39%)
|
Social support
|
8 (26%)
|
Connectedness/ social network
|
3 (10%)
|
Social isolation
|
3 (10%)
|
Acculturation
|
1 (3%)
|
Community integration
|
1 (3%)
|
Interpersonal relationships
|
1 (3%)
|
Health knowledge
|
9 (29%)
|
Health care utilisation
|
9 (29%)
|
Health behaviour
|
8 (26%)
|
Self-care
|
8 (26%)
|
Treatment adherence
|
8 (26%)
|
Symptom severity
|
6 (19%)
|
Condition symptom severity
|
5 (16%)
|
Pain severity
|
2 (6%)
|
General level of activity
|
4 (13%)
|
Quality of communication with others
|
3 (10%)
|
Mortality
|
1 (3%)
|
*Clinical surrogates: Blood glucose (Fasting blood glucose); Blood pressure (Diastolic blood pressure & Systolic blood pressure); CD4 cell count; Estimated Glomerular Filtration Rate (eGFR); Glycated haemoglobin (HbA1c); Lipid Levels/ Lipid Profile (High-density lipoproteins & Low-density lipoproteins); Prostate-specific antigen; Resting heart rate; Triglyceride; Urinalysis (Glycosuria, Microhematuria, or Proteinuria); Viral Load; Weight (Body fat, Waist circumference, or Body Mass Index) |
Effectiveness
For the most frequently identified outcome domains, the majority of reviews reported mostly statistically non-significant effects of peer support (Table 3). Though most reviews reported findings favouring peer support, Table 3 shows that these were rarely significant. Few reviews found a statistically significant effect favouring peer support in a majority of the primary research measuring quality of life (n=1/12), self-efficacy (n=3/14), clinical surrogates (n=1/8), depression (n=1/10), distress (n=2/11), and health knowledge (n=4/9). Three meta-analyses pooled data on clinical surrogates, indicating small to medium statistically significant differences in favour of peer support effect sizes for HBA1c [55, 64-65]. Effect sizes were small though not statistically significant for quality of life and depression [53, 59], and negligible and not statistically significant for distress [54].
Table 3
Effect size (MD) of peer support interventions (or number of significant findings of primary studies)
|
Quality of life
|
Self-efficacy
|
Clinical surrogates
|
Depression
|
Distress
|
Health knowledge
|
Acquired brain injury, cerebral palsy, and spina bifida
|
Levy 2019
|
2/5
|
|
|
|
|
|
Asthma
|
Kew 2017
|
n=3, 0.40 (95% CI -0.02 to 0.81) favouring peer support
|
|
|
|
|
|
Cancer
|
Campbell 2004
|
0/5
|
|
|
|
|
0/1
|
Dunn 2003
|
0/1
|
0/1
|
|
|
0/1
|
* 0/2
|
Hoey 2008
|
0/3
|
1/3
|
|
1/3
|
0/3
|
0/8
|
Macvean 2008
|
0/1
|
0/1
|
|
1/3
|
|
|
McCaughan 2017
|
n=3, -0.11 (95% CI -0.47 to 0.24) favouring peer support
|
|
|
n=5, -0.37 (95% CI -0.75 to 0.00) favouring peer support
|
0/1
|
|
Meyer 2015
|
|
* 2/2
|
|
0/1
|
0/4
|
* 1/2
|
Hu 2019
|
* 2/3
|
* 2/3
|
|
1/6
|
* 2/4
|
* 2/3
|
Cardiovascular disease
|
Parry 2010
|
0/1
|
* 1/2
|
|
|
|
|
Small 2013
|
|
|
* HBA1c n=4, -0.26 (95% CI -0.41 to -0.11, I2 = 47.6% favouring peer support; Blood pressure n=1, −0.25 (95% CI −0.45 to −0.05) favouring peer support
|
|
|
|
Chronic disease
|
Enriquez 2016
|
|
0/2
|
* HBA1c 1/2, Virologic failure/viral load 2/3, CD4 cell count 1/2, Blood pressure 1/1
|
* 1/2
|
0/1
|
|
Fisher 2017
|
0/1
|
0/2
|
3/7
|
0/2
|
|
|
Dale 2008
|
0/2
|
0/1
|
|
|
|
|
Chronic kidney disease
|
Bennett 2018
|
|
|
0/2
|
0/1
|
|
|
Diabetes
|
Cici 2018
|
|
0/2
|
0/12
|
|
0/2
|
0/2
|
Dale 2012
|
|
2/5
|
HBA1c 4/14, Blood pressure 2/5, Cholesterol 1/6,Weight/BMI/Body Fat 5/9
|
|
|
* 2/3
|
Gatlin 2017
|
1/3
|
1/3
|
HBA1c 2/6, Blood pressure 2/5, Lipid profiles 0/4, Weight/BMI/Waist circumference 1/7
|
0/1
|
* 1/2
|
2/5
|
Kong 2020
|
|
|
|
|
n=10, -0.06 (95% CI -0.22 to 0.10) favouring peer support
|
|
Krishnamoorthy 2018
|
|
|
* HBA1c n=26, −0.28 (95% CI −0.45 to −0.11) favouring peer support
|
|
|
|
Qi 2015
|
|
|
* HBA1c n=13, −0.57 (95% CI −0.78 to −0.36) favouring peer support
|
|
|
|
Tang 2011
|
0/1
|
0/1
|
Glycaemic control 1/2, Lipid profile 0/1, Blood pressure 0/2, Weight/BMI 0/1
|
0/1
|
0/1
|
|
HIV
|
Boucher 2020
|
0/4
|
0/3
|
0/6
|
1/5
|
0/1
|
0/1
|
* Effect size (CI 95%) favours intervention, or at least half of primary research studies in reviews demonstrates significant difference |
Mechanisms
Twenty-three mechanisms or theories were identified across 9 reviews (Table 4). Most mechanisms or theories were identified in two reviews that focused on theory and treatment experience [10, 42]. Most reviews did not aim to address intervention theory. Appraisal, emotional and informational social support (n=5), social cognitive theory (n=3), and social comparison theory (n=3) were the most frequently cited mechanisms or theories of effect in peer support interventions.
Table 4
Mechanisms identified in reviews
|
Social cognitive theory
|
IMB Theory
|
Uncertainty management theory
|
Appraisal, emotional, informational support
|
Illness or Social identity theory
|
Readiness stage of wellness motivation
|
Theory of planned behaviour
|
Empowerment
|
Peer leadership or advocacy
|
Sense of connection
|
Experiential knowledge
|
Finding meaning
|
Isolation
|
Sharing
|
Helping
|
Reciprocity
|
Role satisfaction
|
Emotional entanglement
|
Self-determination & autonomy motivation
|
Social comparison theory
|
Orem’s self-care theory
|
Rodger’s person-centred approach
|
Collective voice & mobilisation
|
|
Cancer
|
Dunn 2003
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
✓
|
|
|
|
Lee 2018
|
✓
|
|
|
|
|
|
|
|
✓
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Meyer 2015
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
✓
|
|
|
|
Walshe 2018
|
|
|
|
✓
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
✓
|
|
|
|
|
Chronic disease
|
Embuldeniya ‘13
|
|
|
|
|
|
|
|
|
|
✓
|
✓
|
✓
|
✓
|
✓
|
✓
|
✓
|
✓
|
✓
|
|
|
|
|
|
Enriquez 2016
|
|
✓
|
|
✓
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
✓
|
|
✓
|
✓
|
|
|
Diabetes
|
Qi 2015
|
✓
|
|
|
✓
|
|
|
|
✓
|
|
|
|
|
|
|
|
|
|
|
✓
|
|
|
|
|
|
HIV
|
Boucher 2020
|
✓
|
✓
|
✓
|
✓
|
✓
|
✓
|
✓
|
✓
|
✓
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Somatic illness
|
Kingod 2017
|
|
|
|
✓
|
✓
|
|
|
|
|
|
✓
|
|
|
|
|
|
|
|
|
|
|
|
✓
|
IMB Theory (Information, motivation, and behavioural skills health behaviour model); Biopsychosocial ICF theoretical framework (Biopsychosocial International Classification of Functioning, Disability and Health theoretical framework) |
Assessing the robustness of the synthesis
Overall confidence in reviews were rated high in two Cochrane reviews [46, 53], low (n=5) and critically low (n=24). Most reviews were rated ‘critically low’ due to 2 recurring critical weaknesses: absence of explicit reference to a protocol established prior to the conduct of the review (n=16), and the absence of reference to a list of excluded studies and justification for exclusions (n=18; Table 5). However, each of these reviews did not necessarily set out to be read as a comprehensive systematic review. Of the reviews that assessed risk of bias, most used the Cochrane risk of bias tool (n=13), and a further three referred to the EPOC-specific Cochrane risk of bias tool [67]. Reviews’ risk of bias findings were rated favourably with low risk of bias and high confidence in the findings (n=5); equivocal or unclear (n=6); and rated unfavourably with high risk of bias and low confidence in the findings (n=8).
Best evidence synthesis
Due to a lack of ‘high quality’ comprehensive systematic reviews, this best evidence synthesis presents a descriptive summary of Dale and colleagues (2008) and Kew and colleagues (2017) only [46, 53]. Dale and colleagues reviewed across chronic conditions, though included some primary research involving peers without direct experience of a shared chronic conditions (i.e., people with shared spiritual beliefs). We abstracted only data concerning people with direct experience of a shared chronic condition (post myocardial infarction or with angina). Peer support included educational telephone-based interventions based in Australia and USA. Dale and colleagues measured health status, mental health, quality of life, self-efficacy, and behaviour change (i.e., diet). No differences were found between any physical health outcomes, mental health, quality of life or self-efficacy. Peer support telephone calls were associated with dietary change in one primary research study only.
Kew and colleagues (2017) reviewed peer support for asthma [53]. Peer support comprised education, strategic thinking, discussion, and encouragement in the context of the Triple A programme that teaches older participants to educate and empower their peers based in Australia, Jordan, and USA. Kew and colleagues measured quality of life, severity of asthma (via exacerbations requiring a course of oral steroids, and asthma control), health care use, health behaviours (i.e., smoking), and adverse events. The asthma-related quality of life random-effects model was imprecise and showed no differences (MD 0.40, 95% confidence interval -0.02 to 0.81). Most other outcomes did not show an effect favouring peer support. One study found a reduction in adherence in peer support and comparator. Asthma control and nicotine dependence favoured peer support, though this finding was not statistically significant.
Meta‐analysis was prevented by heterogeneity between studies, weaknesses in blinding and incomplete reporting. Both reviews shared the conclusion that their findings should be interpreted with caution due to weaknesses in the underlying primary research literature, and that more high-quality clinical trials are indicated.