Peer functioning

Peer status is the degree to which an individual is accepted or rejected by their peers [8]. Two systematic reviews focused more broadly on peer functioning, in which 14 studies collectively demonstrated children and youth with ADHD were more likely to be peer rejected than TD peers [4, 5, 9]. The implications of peer rejection for children and youth with ADHD were an increased risk of academic failure, school dropout, depression, anxiety, substance and/or alcohol misuse [10, 11]. Furthermore, children and adolescents with ADHD are more likely to self-report experiencing peer victimisation from their peers than TD peers [12–14]. As many as 57% of adolescents with ADHD reported that they experienced any form of peer victimization at least once per week [12]. However, parents and teachers also reported that children with ADHD were more likely to be bullies themselves or threatened other peers [13]. Adolescents who had lower satisfaction with family relationships and higher scores on the Behaviour Approach Scale were significantly associated with bullying perpetration [15], yet while female adolescents with ADHD were reported to be more likely to engage in bullying than their female TD peers, this was not a significant finding [14].

Friendship is distinctly different to peer status and is defined as a close relationship between two children that is mutual, reciprocal, and voluntary [8, 16]. Through friendship, children and youth learn how to cooperate, manage conflict, and express their emotions appropriately [17–21]. Friendship has been measured most commonly by peer nominations, parent, teacher and self-report questionnaires, and observation in structured and free play tasks [21–25]. Using these types of measures, friendship has been associated with successful adjustment during stressful periods for children and youth [16].

Conceptual models of friendship

With the existence of varying definitions of friendship and measures to operationalise these definitions, two prominent conceptual models of friendship have been presented within the literature [16, 21]. Fig 1 is a visual representation of social functioning that was developed by the authors to display how friendships has been operationalised in this current study and how it differs from other social functioning concepts. As friendship is multi-faceted, Hartup [16] proposed three domains of friendship: having friends, friendship quality and the identity of one’s friends which is represented by the light green boxes in Fig 1. Bagwell and Schmidt [21] conceptualised six domains (which is indicated in dark green in Fig 1) in their model of friendship experience, with three of their six domains (presence of friendship, friendship quality and the characteristics of friends) largely overlapping with Hartup [16]. The additional domains for Bagwell and Schmidt [21] were: interactions with friends, child characteristics (i.e., social skills, behaviours and competence, social cognition, emotional regulation), and the context of friendship (i.e., peer status). However, the inclusion of child characteristics and the context of friendship as domains of friendship contrasts with most existing literature [4, 5, 7, 9]. Multiple systematic reviews and one meta-analysis have conceptualised child characteristics with social functioning and context of friendship with peer status [4, 5, 7]. This distinction in the literature demonstrates recognition that child characteristics and peer status are separate constructs that rather contribute to a successful or unsuccessful friendship [4, 5, 7]. This distinction has been presented in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. A conceptualization of friendship based on existing literature.

Note. The conceptualisation of friendship has been adapted from both Hartup’s model of friendship and Bagwell & Schmidt’s model of friendship experience [16, 21].

https://doi.org/10.1371/journal.pone.0289539.g001

Therefore, we adopted a conceptualisation of friendship that reflected existing literature and included the following four domains: presence of friendship, friendship quality, characteristics of friends and friendship interactions which is indicated by the red boxes in Fig 1. The domain of presence of friends encompasses whether a child has a friend, the number of their reciprocated friends and the stability of their friendships (i.e. duration) [21]. Friendship quality is comprised of both positive features (i.e., companionship, support) and negative features (i.e., conflict) [26]. Under the domain of characteristics of friends, aspects may include understanding if friends share similar attitudes or interests and the social behaviours and competencies of friends [16, 21]. The domain of friendship interactions includes both the frequency of contact with friends and the quality of friendship interactions [21]. The quality of friendship interactions is distinct from friendship quality as it captures behaviour in games (e.g., compliance with rules), affect with friends, and how children make proposals [24, 27], rather than assessing the overall positive and negative features of a friendship [26]. Fig 1 provides a schematic representation of our conceptualisation of friendship based on existing literature.

Existing reviews on the friendships of children with ADHD

Two previous systematic reviews on peer functioning for children and youth with ADHD, included friendship as a variable [4, 5]. Although most of the 33 studies included in these reviews investigated peer status, only six of the studies examined friendship outcomes. Girls with ADHD were found to have fewer friends and friendships that were less stable than TD girls. The results of these studies of friendship included in these reviews did not show a significant correlation between externalising symptoms and comorbid oppositional behaviour or conduct disorder with friendship outcomes [4, 5]. Of the six studies already included in systematic reviews, the most recent was published in 2011 and numerous studies assessing friendship have since been published. Four literature reviews have investigated the friendships of children with ADHD [8, 17, 28, 29]. These literature reviews lacked the methodological rigour associated with systematic reviews and neither the systematic nor literature reviews assessed the methodological quality of their included studies. Collectively, this presents a substantial gap in the literature as the friendships of children with ADHD have not been synthesized or critiqued in over a decade.

Possible contributors to friendship challenges: Social skills, social cognition and emotional regulation

Children and youth with ADHD experience a breadth of friendship difficulties. The presence of friendships for children with ADHD were found to be fewer and shorter compared to TD children [30–32], with children self-reporting poorer friendship quality, with less positive features and more negative features [27, 30, 32–34]. Possible contributors to their poorer friendships may be difficulties with their social skills, social cognition, and emotional regulation abilities. While the distinction between social skills and friendships has been established, the development of these skills are crucial to enable successful peer relationships [7, 21]. Social skills include both verbal and non-verbal behaviours that support peer interactions such as appropriate facial expressions, sharing, helping others and turn taking [7]. The core symptoms of ADHD being inattentiveness and/or hyperactivity and impulsivity often impact the performance of these desired social skills within interactions such as interrupting others when they are talking, making irrelevant comments, increased conflict, [35–37].

Secondly, children and youth with ADHD have impaired social cognition which may also contribute to their poorer friendships. Social-cognitive skills include identifying and interpreting cues, perspective taking, forethought, cognitive biases, and self-perception of one’s performance [9, 35, 38]. Children’s difficulty with inter-personal empathy [39] could explain poorer quality interactions with friends, as cooperative play, games with rules, and successfully supporting another’s needs in play require perspective-taking. As children with ADHD have difficulty anticipating the impact of their behaviour on their friend’s emotional state and adjusting their actions accordingly, they may have fewer and shorter reciprocal friendships and develop friendships with children who have similar difficulties [35, 39, 40]. Children and youth with ADHD have reported impaired social informational processing which impacts their ability to respond to social situations appropriately [38]. Children and youth with ADHD may misinterpret social cues and can be more likely to respond to events with aggressive behaviour in neutral situations, known as a hostile attribution bias [41]. When children act defensively to perceived negative behaviour from peers, this can hinder the development of friendships [41]. Additionally, children and youth with ADHD may also hold a positive illusionary bias in which they perceive their own competence of their skills (i.e. social skills) better than their actual competence [42]. This presents a barrier to intervention if children are unable to identify areas of their own poorer social skills within their friendships.

A third explanation for their poorer friendships may be due to difficulties with emotional regulation [35, 43]. For children with ADHD, applying Barkley’s model of behaviour inhibition, highlights how it may be difficult for children to independently self-regulate [35]. Both children and youth with ADHD can struggle to self-regulate when they feel negative emotions (such as anger, frustration) and positive emotions (such as excitement) which this impacts their ability to demonstrate the desired behaviour in the social situation (i.e., staying calm after losing a game, using appropriate observable facial expressions and their tone of voice) [43, 44]. Their behaviour may be perceived as immature, bothersome or overly exuberant, in turn impacting their friendship interactions [45]. Further, difficulties with emotional regulation have been associated with more frequent incidents of verbal or physical aggression in children with ADHD, which also affect the way they interact with peers and respond to social situations [43]. In middle school students, emotional self-awareness and emotional control mediated the relationship between ADHD symptoms and poor social skills [43]. The aforementioned study was replicated by Cleminshaw and colleagues using an adolescent ADHD population where emotional dysregulation and presence of ADHD mediated parent-rated social skills of adolescents with ADHD [44]. Overall, difficulties with one or more of the above factors may be contributing to the both the reduced number of friends and poorer friendship stability, quality and interactions for children and youth with ADHD.

Objectives and research questions

Therefore, the aim of this systematic review was to systematically review and methodologically appraise the existing evidence of studies reporting on friendships of children and youth with ADHD. Our conceptualization of friendship included four domains: presence of friendship, friendship quality, characteristics of friends and friendship interactions. This study was guided by the following research questions:

1. What is the study design and reported aspects of friendship for children and youth with ADHD?
2. How do the friendships of children and youth with ADHD compare to TD children?
3. What associations are reported between friendship and social-emotional wellbeing and mental health in children and youth with ADHD?
4. What is the methodological quality of studies reporting on the friendships of children and youth with ADHD?

Protocol and registration

As per the PRISMA statement, the protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO registration ID CRD42021213718). The protocol may be accessed via the following URL: https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42021213718.

Eligibility criteria

Eligibility criteria were developed prior to the database search. Included studies were required to meet the following criteria: (1) participants needed to be aged 18 years or younger; (2) participants required a confirmed ADHD diagnosis by a qualified professional (e.g. psychiatrist or paediatrician) using recognised diagnostic tools such as the Diagnostic and Statistical Manual (DSM) of Mental Disorders 4^th or 5^th edition (DSM-III-R [48]; DSM-IV; [49] DSM-IV-TR; [50]; DSM-5; [1]), World Health Organisation (WHO) International Classification of Diseases 10^th Edition, (ICD-10; [51]) or Diagnostic Interview Schedule for Children Version IV (DISC-IV; [52]); (3) participants could have multiple diagnoses (e.g. ADHD and ODD) provided ADHD was the primary diagnosis; (4) measurement of friendship needed to be related to at least one of the adopted friendship domains (i.e., presence of friendship, friendship quality, characteristics of their friends, and friendship interactions); and (5) measurement could take the form of peer nominations, self- or parent-report friendship questionnaires, behavioural reports from parents and teachers, or clinician observations. Studies that used a quantitative study design, including exploratory designs, were eligible for inclusion.

Studies were excluded if they: (1) had participants described ‘at risk’ or used only parent and teacher reports of ADHD symptomology; (2) had participants where ADHD was a secondary condition (e.g., in addition to autism spectrum disorder); (3) reported on other aspects of social skills or peer functioning such as peer status, or peer problems without reference to friendship; (4) reported on outcomes after friendship intervention; and (5) were not published in a peer-reviewed journal or, were published in a language other than English.

Information sources and search strategy

The database selection and search strategy were developed in consultation with the first and third authors in collaboration with the Academic Liaison Librarian at the University of Sydney. A comprehensive search was completed across six databases: CINAHL, Eric, Embase, MEDLINE, PsycInfo and SCOPUS (see Table 1). The first author conducted the search on March 3, 2021, with no date limit applied. By identifying key words from similar reviews on friendship, this shaped the search strategy into three elements: (1) search terms related to ADHD; (2) search terms related to friendship and; (3) search terms related to both children and adolescence (see Table 1). A grey literature search using internet sources (Google Scholar) and manually searching reference lists was conducted by the first author to identify any additional studies.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Search strategy implemented across databases.

https://doi.org/10.1371/journal.pone.0289539.t001

Study selection

Study selection was performed by the first and seventh authors on the title and abstract of records. Prior to this, inter-rater agreement was established through a training session led by the first author on the eligibility criteria whereby both authors screened the same 10 title and abstracts independently using the eligibility criteria. A 100% inter-rater agreement was achieved. Thereafter, the first author independently screened all title and abstracts against the eligibility criteria and then, to ensure rating accuracy, 50% of the title and abstracts were randomly selected and screened by the seventh author. The Covidence software (https://www.covidence.org) was used for screening as it is the primary screening and data extraction tool recommended by the Cochrane Community [53]. Authors scored records as yes or no based on the eligibility criteria. Disagreements were discussed with the second and fifth authors to reach consensus. Weighted Kappa with linear weights was calculated to assess inter-rater agreement.

Data collection and extraction

In alignment with the Cochrane Handbook for Systematic Reviews of Interventions, two data extraction tables were created, trialled on studies, and refined based on feedback from the first, second, third and fifth authors [54]. One table pertained to participant and study characteristics, including: study design, country of study, sources of participants, participant age, gender and comorbidities, eligibility criteria, friendship outcome and measures, source(s) of data, and methodological quality. The second table included: the friendship findings of each study mapped against our conceptualisation of friendship, the associations between friendship and social-emotional wellbeing and mental health (if reported), and the main findings reported in the included studies, with a focus on comparing friendships to TD children. The first author extracted all data. Authors from one of the included studies were contacted via email for a subset of data as the age range of the sample exceeded 18 years old, but the mean was under 18 years [55].

Methodological quality

To assess the methodological quality of included studies, the Standard Quality Assessment Criteria for Evaluating Primary Research Papers from a Variety of Fields (QualSyst checklist) was administered as it can be used for all quantitative studies [56]. The Appraisal tool for Cross-Sectional Studies (AXIS) was also applied, as most included studies used a cross-sectional design [57].

The first author independently rated studies against the QualSyst checklist and AXIS. The seventh author also rated a randomly selected 50% overlap on the included studies using the QualSyst checklist. If disagreements arose, consensus was reached through discussion with the first, second and seventh authors. The fourth author also rated a randomly selected 50% overlap on the included studies using the AXIS checklist. Weighted kappa with linear weights was calculated to determine inter-rater agreement. To achieve a quality score using the QualSyst checklist, fourteen items were rated as yes = 2, partial = 1, no = 0 or not applicable to achieve a maximum score of 28, which is reported as a percentage. Items marked not applicable were excluded from the calculation. Studies were scored as strong (>80%), good (60–79%), adequate (50–59%) or poor quality (<50%) [56]. The AXIS has 20 items which are rated yes, no or don’t know/comment and no quality score is calculated [57]. The proportion of the 20 items adequately addressed (i.e., scored yes) was calculated.

Study selection

In total, 5252 abstracts were retrieved from six databases. Duplicate abstracts (n = 2630) and other publication types (i.e., conference abstracts, reports; n = 116) were excluded prior to screening, leaving 2506 records to be imported into Covidence (https://www.covidence.org) for title and abstract screening. The first author independently screened all 2506 title and abstracts with 1253 (50%) randomly selected records screened independently by the seventh author. There was strong inter-rater agreement kw = 0.81 (95% CI = 0.76–0.86). 99 studies were identified for full text review, with 23 of the 99 studies meeting the inclusion criteria along with one additional study [58] identified during handsearching. Due to the broad search strategy, a significant number of studies were excluded as irrelevant abstracts such as pharmacological and neurological imaging abstracts were captured in the database search. A total of twenty-four studies were included for data extraction. Fig 2 shows a PRISMA flow diagram detailing the study selection process.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. PRISMA flow diagram.

https://doi.org/10.1371/journal.pone.0289539.g002

Study characteristics

Of the twenty-four studies, twenty-three studies were cross-sectional and one was longitudinal [24]. Most studies were conducted in Canada (n = 11), with the remaining conducted in the United States, Australia, Bahrain, China, Denmark, Greece and Israel (see Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Summary of participant and study characteristics.

https://doi.org/10.1371/journal.pone.0289539.t002

Participants

Across the twenty-four studies, there were 1509 participants with ADHD with a mean age of 11.4 years (range 5–18 years; SD = 3.0). A total of 1197 TD participants with a mean age of 11.8 years (range 6.8–18 years; SD = 3.1) were included across nineteen studies that used a TD comparison group. Participants with ADHD were mostly male, 68.4% and 63.7% of TD participants were male. The IQ of participants varied between 70–120 on the Wechsler Abbreviated Scale of Intelligence [95, 96], though five studies did not report IQ eligibility criteria [59, 66, 68, 73]. In five studies, participants’ diagnosis was confirmed using the DSM-5, seven studies used the DSM-IV, two studies used the DSM-IV-TR, DISC-IV or the KSADS [85], and one study used the ICD-10 F90 Hyperkinetic disorder criteria [51] or DSM-III-R [51]. Seven studies did not report the diagnosis confirmation tool (see Table 2). Twenty studies used rating scales to confirm ADHD symptomology, with the Conners rating scale-revised or Conners-3 [79, 80] most commonly used across twelve studies (see Table 2).

Friendship outcomes

Regarding the first research question, nine primary friendship outcomes were identified across the twenty four included studies: friendship quantity (n = 16), friendship stability (n = 6), difficulty making friends (n = 1), perceived social support (PSS) from friends (n = 4), friendship quality (n = 9), self-perception of competence in friendship (n = 1), the quality of interactions with friends (n = 5), frequency of contact with friends (n = 4) and characteristics of friends (n = 5; see Table 2). These reported outcomes were then mapped against our four friendship domains (see Table 3). The most investigated domain was the presence of friendships (n = 17), which included reported outcomes of friendship quantity, friendship stability and difficulty making friends. This was followed by the second domain, friendship quality (n = 12), which included reported outcomes on friendship quality and PSS from friends. The third domain, friendship interactions (n = 10), included the reported outcomes of quality of interactions, perception of friendship competence and frequency of contact. The fourth domain, characteristics of friends (n = 5) included reported outcomes on the characteristics of friends (see Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Summary of friendship outcomes.

https://doi.org/10.1371/journal.pone.0289539.t003

Friendships of children with ADHD

In addressing research question two, the main findings on the friendships of children and youth with ADHD are reported under each friendship domain, with a focus on comparing their friendships to TD children.

Friendship, social-emotional wellbeing, and mental health

Regarding research question three, we identified thirteen studies that assessed friendship as a correlating (n = 10), mediating (n = 1) or moderating (n = 2) factor in children’s social-emotional wellbeing and mental health (see Table 3). Aspects of social-emotional wellbeing and mental health measured across the studies were loneliness (n = 3), peer victimization (n = 4), externalising symptoms/behaviour (n = 4), depressive or anxiety symptoms (n = 2), affect (n = 1), hope (n = 1), perceived competence (n = 1), self-worth (n = 1), sibling relationships (n = 1).

Higher quality friendships were reported to be associated with a reduced degree of loneliness for boys with ADHD [64] and youth with ADHD and comorbid learning difficulties [59]. Smit and colleagues [75] found that children who reported negative friendship quality also reported to be more lonely although, no significant association between loneliness was found for children with comorbid externalising disorders. Two studies investigated the associations between friendship quality and depressive and anxiety symptoms [66, 69]. Only one study found increased friendship quality to mediate depressive and anxiety symptoms in adolescents with ADHD [66]. Children with ADHD who had a co-morbid externalising disorder were predicted to have more negative and fewer positive features of friendship quality, however, positive or negative friendship quality was not predicted by a comorbid internalising disorder [34].

In children with ADHD, two studies found that the higher friendship quality was correlated with increased overall sense of hope and competence [68] and self-worth [53]. One study reported that increased friendship quality was associated with increased positive affect but found no reduction in negative affect [59]. One study investigated the role of friendship quality on peer victimization and bully status and found a correlation between victimized youth reporting less PSS from friends [58].

One study investigated the associations between internalising and externalising behaviours and the presence of friendship [31]. Only children with comorbid anxiety were reported to have fewer dyadic friends, as this variable was not significant for children with comorbid oppositional defiant disorder-conduct disorder [31]. Three studies produced mixed findings on the associations between presence of friendship as a protective function against peer victimization [63, 67, 73]. A negative correlation was reported between numbers of friends and being a victim of bullying [67]; Ma, however, reported that this correlation pertained to boys [73]. Conversely, in girls with ADHD, the presence of a friend was reported to reduce peer victimisation as having at least one friend moderated the association between behavioural risk factors (internalising, externalising behaviours and social competence) and victimisation [63]. An association was reported where children with ADHD who had externalizing disorders were also more likely to be bullies themselves if they had more friends [67]. The presence of friendship was found to be predicted by quality of sibling relationship in children with ADHD whereas, for TD children, the quality of their sibling relationships could predict both presence and quality of friendship [32].

Methodological quality and risk of bias

For the fourth research question, twenty-two studies had strong methodological quality and two studies had good methodological quality [33, 56] (see Table 2). Inter-rater agreement for QualSyst checklist represented strong agreement kw = 0.81 (95% CI = 0.68–0.94), with an average of 14.39 (SD = 1.5) out of 20 items rated as yes on the AXIS. The most common areas where studies needed to report more explicitly to improve their quality score was using a cross-sectional design [10, 17, 27, 30, 32, 33, 58, 59, 63, 64, 67, 69, 70, 73, 74]. There was a risk of selection bias as 14 studies used sampling across one or two locations from referrals and/or self-selection processes. Only three studies justified the sample size using a power analysis [34, 67, 75], with remaining studies reporting small sample sizes. Effect sizes were not completely reported for seven studies.

Limitations

This study was strengthened by adhering to the PRISMA protocol [47] completing a rigorous search across six databases and implementing two critical appraisal tools to assess methodological quality [56, 57]. Due to the exploratory nature of this systematic review, eighteen included studies used a cross-sectional design. Consequently, our ability to report on the evolving nature of friendship and casual relationships across social and emotional wellbeing outcomes is limited. Further, a secondary reviewer was unable to assess 50% overlap using AXIS due to time constraints. No meta-analysis was conducted due to the clinical heterogeneity seen in variable friendship measures, and small sample sizes.

Implications for research and clinical practice

We propose five areas for further investigation. Firstly, there is a need for longitudinal studies using the real-life friendships of children and youth with ADHD. This will enable greater insights into how the friendship domains may evolve over time in conjunction with social, emotional and cognitive development. Secondly, future studies should assess the mediating and moderating roles of the four friendship domains in the prevention of social, emotional and mental health difficulties. We propose that further studies should also investigate how underlying capacities, including social-cognitive skills and emotional regulation skills, contribute to the poorer friendship experience across the friendship domains of children and adolescents with ADHD. By examining these relationships, this could inform existing interventions, such as Children’s Friendship Training [106], Program for Evaluation and Enrichment of Relational Skills (PEERS) for adolescents [107] or classroom based interventions such as the Making Social Accepting Inclusive Classrooms (MOSAIC), and novel friendship interventions to target the underlying capacity (i.e. emotional regulation) in a naturalistic setting (i.e. with friends or classmates). We also identified a range of self-reporting, parent reporting and clinical observation measures to assess our friendship domains. We recommend that future studies assess the psychometric properties of these tools to compare their usefulness for future studies. Finally, as fourteen of the studies were conducted in English speaking countries, the results in the review reflect a limited cultural perspective of friendships. Therefore, cross-cultural comparisons of friendships should be considered in future studies to understand how cultural values may impact children and youth’s perspective and experience of friendship.

Clinically, by synthesising current literature in our four domains, clinicians can use these domains to explore a client’s strengths and weaknesses in their friendships more effectively. Thus, children and youth with ADHD may experience greater success in their friendships and prevent the negative impacts on their development and well-being.