We conducted an electronic search in each of the following three databases: PsycINFO, PubMed, and Web of Science from their origins until 15 August 2017. PsycINFO and PubMed are two of the main databases used in the fields of Psychology and Medicine (Baker Pistrang, & Elliott, 2015). Additionally, we considered to conduct a search with the multidisciplinary academic database Web of Science, as friendships topic might be achieved by other disciplines.

We used the following combinations of terms: “pain” and “child* / infant / adolescent* / juvenile / teen* / young* / school age” and “peer* / friend* / interpersonal / social*” in the title and/or abstract of the articles. Some of these terms have been previously used in searches from salient reviews in the field of pediatric chronic illnesses, including pain (Forgeron et al., 2010; Kohut, Stinson, Giosa, Luca, & van Wyk, 2014; Lewandoski, Palermo, Stinson, Handley, & Chambers, 2010; Tong, Jones, Craig, & Grewal, 2012). The full search strategy is described in detail in Table S1, S2 and S3.

Two of the authors independently reviewed the titles and abstracts of all the citations. They were blind to authors and institutions, and all the articles that met the criteria were full-text reviewed. In addition, reviewers examined reference lists of all the articles included in order to identify any additional articles that might have been missed by the search strategy. Articles were selected using the following inclusion criteria:

(C1) Primary sources (e.g., not meta-analysis, reviews, letters, or commentaries)

(C2) Published in English or Spanish.

(C3) Targeted school-aged children (6-12 years old) and/or adolescents (13-18 years old).

(C4) Children or adolescents presenting one of the following chronic pain conditions: lower limb pain, neck and shoulder pain, back pain, abdominal pain, headache and migraines, fibromyalgia, and complex regional pain syndrome. Therefore, we did not include pain due to medical conditions, medical procedures, or disease.

(C5) Included a specific subgroup analysis or specific attention to peer and friend relationships.

Any disagreement was discussed until agreement was reached. Inter-rater agreement for full-text screening was 93% (Cohen, 1968).

The heterogeneity of study designs in the articles selected for the review (descriptive, case-control, and cohort) and the nature of studies (no intervention studies or randomized controlled trials were found) made it not possible to follow the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines (Moher, Liberati, Tetzlaff, Altman, & PRISMA Group, 2009) and assess the quality of the methodology with a unique structured quality scale. To overcome this problem and do a proper assessment according to the nature of studies, we followed the solution adopted by Jarde (Jarde, Losilla, & Vives, 2012) by using a different checklist for each study design. Therefore, the methodological quality of quantitative studies was reviewed using Downs and Black’s (1998) checklist, and for qualitative studies, the Critical Appraisal Skills Programme (2013) was used. The quality appraisal for each study is described in Table S4 and S5. No study was rejected due to low methodological quality.

Predefined data were extracted by one of the authors: authors and year of publication, participants and chronic pain condition, study design, primary focus of the study, and instruments and measures used to assess peer and friend relationship characteristics (Table 1). Detailed data about the mechanisms involved in peer and friend relationships were extracted (Table 2), as well as whether a theory was used to explain these relationships. A second reviewer checked the information selected through each process, and a consensus was reached if there were any discrepancies.

Table 1

Table 1

Table 1

Table 1

Table 1

Table 1

Table 1

Table 1

Table 1

Table 1

Table 1

Table 1

Table 1

Table 1

Table 1

Note. CP = chronic pain; RP = recurrent pain; CRPS = complex regional pain syndrome; RAP = recurrent abdominal pain; JPFS = juvenile primary fibromyalgia syndrome; CSC = cross-sectional cohort; PC = prospective cohort; CSD = cross sectional descriptive; CC = case-control; LD = longitudinal descriptive; Q = quantitative; QL = qualitative.

Table 2

During the data extraction process, we contacted one author in order to obtain the full-text version of one of the articles selected. However, when some data were missing (i.e., mean age or standard deviation, or specific items of questionnaires used), we did not attempt to contact authors for further information or search for extra data.

After reviewing the articles, and to fulfill the second objective involving the identification of mechanisms, we followed this four-step process: first, to detect the behaviors, cognitions, affects, or even strategies that are cited or described in some way in the study; second, to check whether strategies were attributed to the child or adolescent in pain or to his peers and friends; third, to collect and organize data in different categories according to their similarity; and fourth and finally, to assign a label to each category in order to define and differentiate them as mechanisms.

Table 1 describes and summarizes the characteristics of the studies reviewed.

Chronic pain conditions addressed most in studies were: abdominal pain (Bandell-Hoekstra et al., 2002; Caes et al., 2015; Carter et al., 2002; Castarlenas et al., 2015; Eccleston et al., 2008; Forgeron et al., 2013; Forgeron et al., 2011; Greco et al., 2007; Larsson & Sund, 2007; Konijnenberg et al., 2005; van Tilburg et al., 2015), headache or migraines (Caes et al., 2015; Carter et al., 2002; Castarlenas et al., 2015; Donovan et al., 2013; Eccleston et al., 2008; Guite et al., 2007; Konijnenberg et al., 2005; Langeveld et al., 1997; Larsson & Sund, 2007; Meldrum et al., 2009; Metsähonkala et al., 1998; Vannatta et al., 2008), back pain (Caes et al., 2015; Carter et al., 2002; Eccleston et al., 2008; Forgeron et al., 2011; Forgeron & McGrath, 2008; Larsson & Sund, 2007), limb pain (Caes et al., 2015; Castarlenas et al., 2015; Forgeron & McGrath, 2008; Larsson & Sund, 2007), and musculoskeletal pain (Forgeron et al., 2013; Guite et al., 2007; Konijnenberg et al, 2005).

Samples sizes ranged from 5 to 2360. Some of the studies included additional sources of information besides children and adolescents, such as clinicians (Donovan et al., 2013), caregivers (Donovan et al., 2013; Guite et al., 2007), and teachers (Greco et al., 2007; Kashikar-Zuck et al., 2007; Vannatta et al., 2008). Nine of the 20 studies (Forgeron et el., 2013; Forgeron et al., 2011; Greco et al., 2007; Karwautz et al., 1999; Kashikar-Zuck et al., 2007; Konijnenberg et al., 2005; Metsähonkala et al., 1998; Vannatta et al., 2008) included a comparison group of children and/or adolescents without chronic pain.

Children and adolescents were recruited from clinical services in the majority of the studies, but in some cases, they were also recruited from the community (Caes et al., 2015; Donovan et al., 2013; Metsähonkala et al., 1998) or school (Bandell-Hoekstra et al., 2002; Castarlenas et al., 2015; Greco et al., 2007; Langeveld, Koot, Loonen, Hazebroek-Kampscheur, Passchier, 1996; Larsson & Sund, 2007).

Considering the use of standardized instruments, only a few studies used the same instruments: Bath Adolescent Pain Questionnaire (BAPQ) (Eccleston et al., 2005) to assess self-perception development (Caes et al., 2015; Eccleston et al., 2008), Revised Class Play (RCP) (Masten, Morison, & Pellegrini, 1985) in combination with Three Best Friends (TBF)/Best Friends Nomination (BFN) (Bukowski & Hoza, 1989) to describe patterns of social behavior and reciprocated friendships (Kashikar-Zuck et al., 2007; Vannatta et al., 2008) and the Quality of Life Headache in Youth questionnaire (QLH-Y) (Langeveld et al., 1996) to assess social interaction with peers (Bandell-Hoekstra et al., 2002; Langeveld et al., 1996). In addition, three studies used specific pain instruments (Caes et al., 2015; Castarlenas et al., 2015; Eccleston et al., 2008), and one of them used instruments oriented toward health and illness in a broad sense (Konijnenberg et al, 2005). The rest of the instruments were related to the social area (classmate’s pain experience (Castarlenas et al., 2015), peer relationships (Bandell-Hoekstra et al., 2002; Langeveld et al., 1996; Price, Spence, Sheffield, & Donovan, 2002), perceived social support-friend (Procidano & Heller, 1983), social acceptance (Asher, Singleton, Tinsley, & Hymel, 1979; Harter, 1988; Vorst, 1990), social experiences (Crick & Bigbee, 1998; social skills (Gresham & Elliot, 1990), role and social limitation due to physical problems (Landgraf, Abetz, & Ware, 1996), illness behavior (Walker & Zeman, 1992), and pain response (Walker, Smith, Garber, & van Slyke, 1997). In parallel, ten studies used their own techniques designed for the study (focus group, workshop, qualitative and in depth interviews, or paper-and-pencil/internet surveys) (Carter et al, 2002; Donovan et al., 2013; Forgeron et al., 2013; Forgeron & McGrath, 2008; Karwautz et al., 1999; Larsson & Sund, 2007; Meldrum et al., 2009; Metsähonkala et al., 1998).

Once we identified the mechanisms, following the process described above (methods section), we classified them in two main categories called isolation or involvement mechanisms, depending on their consequences: a tendency to isolate a child or adolescent in pain from his/her peers or friends, or the opposite, a tendency to involve these children with their peers or friends.

Table 2 presents a detailed list and description of the mechanisms involved in peer and friend relationships.

As mentioned above, the mechanisms arose mainly from qualitative data, whereas only two mechanisms were based on quantitative data: catastrophizing and rewarding pain behaviors (Bandell-Hoekstra et al., 2002; Caes et al., 2015; Castarlenas et al., 2015; van Tilburg et al., 2015). The majority of the mechanisms were captured from the experiences of children or adolescents with chronic pain with their peers or friends. Two mechanisms, Looking for alternative friends and Rewarding pain behaviours (both from peers/friends’ perspective), were exclusively taken from studies that examined the impact of the chronic pain on friendships, including healthy adolescents’ point of view (Castarlenas et al., 2015; Forgeron et al., 2013). We identify the mechanisms described in 13 articles. The other six studies (Karwautz et al, 1999; Kashikar-Zuck et al., 2007; Konijnenberg et al., 2005; Langeveld et al., 1996; Larsson & Sund, 2007; Metsähonkala et al., 1998; Vannatta et al., 2008) provided complementary information about social functioning.

In relation to our third objective, from all the studies selected, we identified 3 of the 5 theories cited in the introduction section to describe and explain the relationships between being in pain and friendships. Additionally, three other theories were identified. And only three of the studies were clearly theory-based, with a theory guiding the study (Bandell-Hoekstra et al., 2002; Caes et al., 2015; Forgeron et al., 2011).

The Social Learning theory (Bandura, 1977) was cited in one study (Guite et al., 2007) as a general reference in the background section, and it was related to withdrawal from regular activities or social contact. Due to their activity restriction, children and adolescents in pain had fewer opportunities to develop social skills and salient behavioral models to learn and imitate; therefore, they would show delayed social development. As variables, the authors analyzed functional disability and self-perceived competence (including social acceptance), and the latter was an important aspect to understand the relationship between pain and functional disability, as the theory postulates.

Lazarus’ (1966) Theory about coping with stress appeared in two articles (Bandell-Hoekstra et al., 2002; Forgeron & McGrath, 2008). One of the studies (Bandell-Hoekstra et al., 2002) describes the pain coping strategies (such as seeking social support, internalizing and externalizing behaviours and so on) to manage stressful situations (pain) and analyzes them in terms of pain severity. Authors based their study in this theory and concluded that pain coping strategies have an important impact on pain severity. The other article (Forgeron & McGrath, 2008) discusses whether adolescents who are not interested in seeking peer support could appraise sharing their pain as somehow shameful, and thus feel incompetent in their ability to discuss their pain with others and consequently do not use this pain coping strategy (and lose the opportunity to use social support as a resource to cope with stress). The study only assessed a few variables related to the theory (such as perceived social support, social anxiety, self-perception, coping, and disability). Direct relationships among these variables remain unclear, and authors explicitly cited the theory only in Discussion sections (Forgeron & McGrath, 2008).

The FAM (Vlaeyen & Linton, 2000) was used as the basis for one study (Caes et al., 2015) that tested some of the variables in this model. Authors showed that high pain-related anxiety is associated with greater disability in adolescents (in general), and with a self-perception of greater impairment in social functioning particularly in girls. Therefore, authors concluded that social functioning should be explored as an integral part of a Fear Avoidance Model.

Although it was not previewed according to previous literature, we identified three other theories that might also be used to explain friendships. These are the Theory of Interdependence (Kelley & Thibaut, 1978) and the Theory of Equity (Walster, Walster, & Berscheid, 1978), both based on a cost-benefit ratio, and the Social Information Processing (SIP) (Crick & Dodge, 1994) model to interpret social situations.

The Theory of Interdependence (Kelley & Thibaut, 1978) and the Theory of Equity (Walster et al., 1978) were used in one study (Forgeron et al., 2013) to partially explain the results, specifically the decrease in the time spent with friends based on a cost-benefit ratio. In other words, children and adolescents in pain spend less time with close friends because they need more time to fulfil their needs related to the pain condition, and, consequently, their friends look for alternative partners.

The SIP (Crick & Dodge, 1994) model was used in one study (Forgeron et al., 2011) to lead the study and to assess, through narrative vignettes, whether the interpretation of friendship interactions was supportive or non-supportive. Results reflected a tendency among adolescents with chronic pain to interpret non-supportive social situations with close friends as more distressing.