Article Text

Original research
Investigating correlates of athletic identity and sport-related injury outcomes: a scoping review
  1. Tian Renton1,2,
  2. Brian Petersen3,
  3. Sidney Kennedy1,2,4
  1. 1Rehabilitation Sciences Institute, University of Toronto Faculty of Medicine, Toronto, Ontario, Canada
  2. 2Centre for Depression and Suicide Studies, St Michael's Hospital, Toronto, Ontario, Canada
  3. 3Faculty of Kinesiology and Physical Education, University of Toronto, Toronto, Ontario, Canada
  4. 4Department of Psychiatry, University of Toronto Faculty of Medicine, Toronto, Ontario, Canada
  1. Correspondence to Tian Renton; tian.renton{at}


Objectives To conduct a scoping review that (1) describes what is known about the relationship between athletic identity and sport-related injury outcomes and (2) describes the relationship that an injury (as an exposure) has on athletic identity (as an outcome) in athletes.

Design Scoping review.

Participants A total of n=1852 athletes from various sport backgrounds and levels of competition.

Primary and secondary outcome measures The primary measure used within the studies identified was the Athletic Identity Measurement Scale. Secondary outcome measures assessed demographic, psychosocial, behavioural, physical function and pain-related constructs.

Results Twenty-two studies were identified for inclusion. Samples were dominated by male, Caucasian athletes. The majority of studies captured musculoskeletal injuries, while only three studies included sport-related concussion. Athletic identity was significantly and positively associated with depressive symptom severity, sport performance traits (eg, ego-orientation and mastery-orientation), social network size, physical self-worth, motivation, rehabilitation overadherence, mental toughness and playing through pain, as well as injury severity and functional recovery outcomes. Findings pertaining to the association that an injury (as an exposure) had on athletic identity (as an outcome) were inconsistent and limited.

Conclusions Athletic identity was most frequently associated with psychosocial, behavioural and injury-specific outcomes. Future research should seek to include diverse athlete samples (eg, women, athletes of different races, para-athletes) and should continue to reference theoretical injury models to inform study methodologies and to specify variables of interest for further exploration.

  • sports medicine
  • orthopaedic sports trauma
  • rehabilitation medicine

Data availability statement

All data relevant to the study are included within the article or have been uploaded within supplemental files. All data extracted and summarised within this scoping review were obtained from published peer-reviewed journal articles. Please refer to articles referenced.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Strengths and limitations of this study

  • The search strategy was constructed in consultation with a University of Toronto librarian.

  • Citation management (EndNote) and systematic review citation screening software (Covidence) were used to allow reviewers to independently screen citations and extract data.

  • Data extraction variables thoroughly described the study sample, injuries sustained, theoretical models referenced, athletic identity scores and timeline of administration, significant key findings as well as study strengths and limitations.

  • A quality assessment was not conducted, and level of evidence ratings were not assigned to studies.


Participation in sport, be it in a formal (eg, registered league) or informal (eg, pick-up, drop-in) setting, is a popular pastime for individuals the world over. Positive benefits associated with sport participation include increased mental toughness,1 perseverance1 2 and positive self-esteem,2–4 as well as the development of fine and gross motor skills, team work and problem-solving abilities.5 These benefits are aside from the countless physical (eg, maintenance of a healthy body weight6), mental (eg, reduction in depression7 and anxiety symptoms8) and cognitive benefits (eg, improved academic performance9 and memory recall10) associated with physical activity in general. Despite these benefits, negative outcomes should also be considered, namely risk of injury. However, not all athletes are created equal, nor are their respective risks of sport injury. This is illustrated by several large-scale epidemiological studies describing marked differences in injury incidence when stratified by sport.11–16 Internal risk factors, such as an athlete’s biological and physical characteristics (eg, age, sex, anthropometry, skill level and physical fitness) as well as their psychological predisposition (eg, personality, history of stressors and availability of coping resources) are also posited to modify injury risk.17–19 External factors, such as level of competition and playing surface, have also been implicated.18 19

Despite individual athlete (eg, physicality, disposition) and sport-specific differences (eg, type, level, frequency of involvement, injury risk), all athletes are thought to embody an ‘athletic identity’ (AI). Initially defined by Brewer et al in 1993, AI is defined as ‘the exclusivity and strength with which an individual identifies with the athlete role, and looks to others for confirmation of that role’.20 To some extent, an athlete’s self-perception of their AI can provide an important measure of their longevity in sport.21 Stronger AIs have been associated with positive health outcomes, increased sport engagement, enhanced athletic performance, improved global self-esteem and confidence, as well as improved social relationships.20 22–25 Conversely, following a sport-related injury, stronger AIs have been associated with depressive symptoms.26 It has also been suggested that athletes who hold a stronger AI may neglect other identities and role responsibilities to maintain the athlete role.20 Therefore, a strong AI may be helpful in some cases and harmful in others, especially within a sport injury context.

Athletes will continue to sustain injuries so long as sport exists, thus illustrating the need to understand factors associated with recovery. To inform stakeholders’ (eg, clinicians, coaches, athletes) understanding and expectations, many theoretical injury recovery models have been developed, several of which are presented here: The Biopsychosocial Model27 28 ; Biopsychosocial Model of Stress and Athletic Injury29 ; Integrated Model of Psychological Response to the Sport Injury and Rehabilitation Process30 ; and Cognitive Appraisal Model of Psychological Adjustment to Athletic Injury.31 Although not specific to sport, some models have been developed to explain and predict outcomes associated with a specific injury, such as concussion (Neurobiopsychosocial Model of Concussion32). Others have been adapted from existing models (Transactional Stress Model33) to suit a sport injury context (Injury Response Model34 35). For a more comprehensive review of select models, please see the following article.36 Despite variation in the labelling used within the models cited above, constructs can be categorised as modifiable (ie, flexible, subject to intervention) or non-modifiable (ie, fixed, unchanging). With respect to addressing recovery outcomes, attention is best focused on modifiable factors because they are subject to intervention. Prior to implementing an intervention however, efforts should focus on describing recovery outcomes observed for a given factor. To our knowledge, AI (a modifiable factor) has not been summarised in detail with respect to its association with sport injury recovery outcomes.

To address this knowledge gap and to provide a comprehensive summary of what is known about AI in relation to sport-related injury outcomes, authors conducted a scoping review. To guide this review, the following questions were established a priori:

  1. Is there an association between athlete self-reported AI and response to a sport-related injury? If so, what is known? Response to injury is operationally defined as any outcome observed following injury (eg, psychosocial, behavioural, functional, cognitive or performance).

  2. Is there an association between a sport-related injury (as an exposure) and athlete self-reported AI (as an outcome)? If so, what is known?


Search strategy and study identification

Search strategies and terms were developed in consultation with a University of Toronto health science librarian (EN; 20 January 2020). The following databases were searched in March and April 2020 by one reviewer (TR): MEDLINE, EMBASE, SPORTDiscus, CINAHL, APA PsycInfo, and Sport Medicine & Education Index (Proquest). The number of citations identified were recorded in table 1.

Table 1

Search strategies by database

Search results were exported to EndNote37 and duplicates were discarded (n=334). Thereafter, article titles and abstracts (n=1122) were exported to Covidence.38 Covidence collates each reviewer’s decision to accept or reject a citation and identifies screening conflicts for resolution. The programme also populates a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart to reflect the number of citations included or excluded at each screening stage (see online supplemental appendix 1). Reasons for exclusion were cited at the full-text screening stage only. Studies identified for inclusion at full-text screening also had their reference lists reviewed for additional studies. was also searched using the following terms: “athlete”, “identity”, “injury” and “sport”, but did not identify any additional studies. TR and BP independently performed each stage of the screening process (titles, abstracts and full-text screening) as well as full-text data extraction. After completing each stage, reviewers met virtually (via Zoom) to discuss and resolve conflicts. Progression to the next screening stage occurred only after 100% agreement was achieved. The same process was applied throughout the data extraction phase. For quality assurance, this scoping review was structured according to the PRISMA extension for scoping reviews checklist (see online supplemental appendix 2).

Study inclusion criteria

  1. AI was assessed using a self-report quantitative measure.

  2. Study sample consisted of at least one group with a sport-related injury which prevented them from engaging in sport.

  3. Injuries were real or hypothetical (ie, imaginary).

  4. Studies captured athletes of any age and playing status (eg, amateur or professional, retired or active). Studies that included athletes with disabilities (eg, para-athletes) were permissible, however, the injury must have been secondary to the existing disability (ie, study must pertain to a sport-related injury).

  5. An objective measure was used to assess the injury or post-rehabilitation status or post-injury AI.

Study exclusion criteria

  1. Article not available in the English language.

  2. Full-text article could not be located following direct request to author(s) (if not available online).

  3. Injury was not specified or assessed for severity.

  4. AI was not self-reported (ie, was reported by a coach, teammate or parent).

  5. Conference proceedings or abstracts.

  6. Qualitative studies.

  7. Systematic, scoping or narrative reviews.

  8. Theses or dissertations.

  9. Consensus statements.

Data extraction

The following data were extracted from each of the included studies and logged independently by reviewers into a blank, preformatted table (see table 2 for template).

Table 2

Article data extraction

  1. Description of sample: country of origin, sample size, sex, race, age, recruitment source, sport background, level of sport and history of sport involvement (eg, frequency and years of participation).

  2. Injury descriptors: definition of injury used (if any), type and severity of injury, time removed from sport, rehabilitation protocol administered and surgical details (if any).

  3. Study methodology: study design, primary and secondary objectives.

  4. Theoretical support: author and model or theory used.

  5. Outcome measures: AI measured used, timeline of administration, AI score and additional outcome measures used.

  6. Key findings: findings related to AI and other measured variables.

  7. Study strengths and limitations.

Findings are presented as a narrative summary, and where possible, presented as a tally (ie, number of studies that reported on a given finding) to denote trends in the literature. In keeping with the purpose of scoping review methodology which is ‘…to identify knowledge gaps, scope a body of literature, clarify concepts or to investigate research conduct’39 as well as ‘… to identify strengths [and] weaknesses … in the research’,40 studies will not undergo quality review (ie, assessment of bias) or be assigned a Level of Evidence rating.

Patient and public involvement

No patient(s) involved.


The search strategy identified 1456 records for consideration (see table 1 for databases searched, search terms used and number of records identified). Two additional articles were identified via hand searching of the included article reference lists. One additional article was previously known to others, but not identified in the searches. Two articles contained multiple studies. A total of 20 publications reporting on 22 studies were eligible for inclusion. Studies used cross-section observational (n=8), prospective longitudinal (n=13) and mixed-methods (n=1) designs.

Sample descriptors

Studies originated from Australia (n=1), Canada (n=1), Israel (n=1), Slovenia (n=1) and the USA (n=18). Most studies included both sex groups, except for three studies which included all-male samples41–43 and one which included an all-female sample.44 A total of n=1852 athletes were included; individual study samples ranged from a minimum of n=6 (45) to a maximum n=316.43 Participants were a minimum of 1343 to a maximum of 70 years old.35 Participants were recruited from several clinical and non-clinical settings, with one study failing to specify a recruitment source45 (see table 2, column 2).

Athletes were involved in a range of team and individual sports; however, several studies did not specify sport background.26 35 46–51 Furthermore, two studies included a small proportion (3%48 and 4%51) of self-defined ‘non-athletes’. Authors of this review chose to include these studies due to the small number of non-athletes (n=7 total) included in analyses. Samples consisted of recreational (eg, house league) and competitive athletes (eg, elite, National Collegiate Athletics Association). Several studies did not report on this metric.26 35 49 52–54 Sport involvement (eg, frequency of and years involved in sport) was heterogeneous and reported within six studies.35 43 45 55–57 Sport participation ranged from 30 min35 to 14.19 (SD=9.40) hours per week55 and years of sport involvement ranged from 6.64 years (SD=3.98)57 to 11.17 years (SD=4.31)45 (see table 2, column 2).

Injury descriptors

Musculoskeletal (MSK) injuries were the most common injuries cited. Nine studies reported exclusively on anterior cruciate ligament (ACL) surgical outcomes, while two42 54 exclusively examined concussion. The remaining 11 studies captured various MSK injuries. Of these 11 studies, 1 did not specify an exact injury but indicated injury to lower or upper extremities,41 1 captured both MSK injuries and concussion43 and 2 studies did not define the injuries sustained.26 53 Of these two, one indicated injury severity on a scale ranging from 1 (mild) to 3 (severe)26 while the other stated that the majority of injuries were ACL tears, but did not specify the exact proportion.53 Time away from sport due to injury varied, ranging from 24 hours41 43 to 63 weeks.55 Ten studies did not specify a length of absence. Three studies41 43 57 reported on athletes who sustained multiple injuries during the data collection period while the remaining 19 captured a first (ie, initial) injury only (see table 2, column 3).

Definitions and theoretical models

Operational definitions of injury were specified in each study except one.26 Those that captured ACL and concussions exclusively, indicated a diagnosed ACL tear or diagnosed or self-reported concussion in lieu of an operational definition. Eleven studies referenced injury models as a means of justification for study methodologies used. The most frequently cited model was the Integrated Model of Response to Sport Injury.30 Several other theories unrelated to sport injury were also referenced (see table 2, column 5).

Wiese-Bjornstal et al’s injury model30 (see online supplemental appendix 3) suggests an athlete’s cognitive appraisal (eg, rate of perceived recovery, cognitive coping, etc.) of the injury is a primary driver of outcome (ie, physical, behavioural and emotional). Seven studies explicitly measured cognitive appraisal via subjective rehabilitation progress,48 coping skills and strategies used,35 44 45 53 psychological response to injury,44 readiness to return to sport56 and rehabilitation beliefs.58 Most outcome measures sought to typify athlete personal factors. A small proportion of studies (n=6) used measures that isolated situational factors (eg, sport, social and environmental),35 41 46 51 54 58 but only assessed social support (eg, availability, quality and source).

Measuring AI

The Athletic Identity Measurement Scale (AIMS),59 7-item or 10-item version, was used exclusively to quantify the strength of AI (see table 2, column 6). The AIMS consists of three subscales: social identity (ie, the extent to which the individual views themselves as occupying the athlete role), exclusivity (ie, the extent to which the individual defines their self-worth based on the athlete role), and negative affectivity (ie, the extent to which the individual experiences negative emotions from undesired outcomes associated with the athlete role).59 The findings summarised below are specific to AI. Analyses that did not consider AI were excluded from the summary. Findings were grouped into the following categories: demographic, psychosocial, behavioural, injury-specific and pain. Several studies also investigated the association between injury (as an exposure) and AI (as an outcome). These findings are presented at the end of this section.


Findings pertaining to AI and sex were presented in two studies but were inconsistent. One study found that sex significantly predicted AIMS subscale scores, with males having significantly higher scores on each subscale (eg, social, exclusivity and negative affect) than females.57 Padaki et al also compared AIMS scores by sex (M=56.6 vs 53.4 for females and males, respectively), but this difference was not significant (p=0.092). They also examined AIMS scores by sport involvement (single vs multisport athletes) and was the only study to have done so. Interestingly, single-sport athletes reported a significantly stronger AI (M=57.7) compared with multisport athletes (M=52.8, p=0.043). Two studies investigated AI and age,42 49 with both identifying a negative non-significant association (as age increased, AI decreased) (see table 2, column 7).


Depressive symptoms were measured in six studies, but only five presented findings in relation to AIMS scores. Correlational analyses were conducted in two of the studies35 52 while regression models were constructed in the other three.26 47 53 Correlational analysis identified a large positive significant association between AI and depression scores,52 while findings from the other study identified a small negative but non-significant association.35 Beta coefficients generated from regression models illustrated a similar positive relationship between AI and depressive symptom severity, while also adjusting for several covariates. Two studies included AIMS scores in their models as an interaction term, one with injury severity26 and one with number of days since surgery.47 Although both models indicated that interaction terms explained a greater variance in depression scores compared to when AIMS scores were entered alone, only one interaction coefficient was significant.47 Despite evidence suggesting that athletes with stronger AIs were more likely to experience depressive symptoms following a sport-related injury, findings also indicated that they experienced greater improvements in their mood throughout the post-surgical follow-up period.47 Four studies assessed anxiety, but only one study compared anxiety symptoms (eg, sport-related performance, somatic, concentration disruption and worry) to AI.54 Despite anxiety symptoms being positively, although weakly, correlated to AI (r=0.14; 0.13; 0.21; 0.05, respectively, for the type of anxiety symptoms noted in the previous sentence), findings were not significant. Another study assessed athletes for symptoms of post-traumatic stress disorder (PTSD; eg, hyperarousal, avoidance and intrusive thoughts)49 and compared PTSD scores between ‘high’ and ‘low’ AI groups prior to ACL reconstructive surgery. Group differences were not significant.

AI was significantly associated with several other, although more abstract, psychosocial constructs including sport performance traits, physical self-worth, motivation and social network size. Traits associated with sport performance such as ego-orientation (example scale item: ‘The most important thing is to be the best athlete’) and mastery-orientation (example scale item: ‘My goal is to learn new skills and get as good as possible’) were significantly associated with AI as represented by the moderate effect sizes observed.54 One study correlated physical self-worth (ie, perceived sport competence, perceived muscular and physical strength and conditioning) to AI and identified a positive moderate and significant association among athletes shortly after they began a rehabilitation programme.35 One study also identified a small significant association between AI and generalised motivation.51 Similarly, a moderate positive significant association was also identified between motivational climate in sport (as facilitated by parental figures) and AI. Athletes with stronger AIs also maintained greater intrinsic and extrinsic motivation towards participation in sport.54 Although social support was assessed in seven studies, only two presented findings in relation to AI. Findings indicated that the maintenance of larger social networks was moderately positively and significantly associated with AI.54 Petrie et al also examined the relationship between AI and social support but with respect to family, friends and significant others. Small positive but non-significant associations were identified between support provided by family, friends and AI but a negative association for significant others (see table 2, column 7).


Several studies investigated the relationship between AI and rehabilitation overadherence, motivation, completion of exercises and accompanying treatments (eg, cryotherapy). One study identified a small significant positive association between AI and beliefs pertaining to rehabilitation overadherence55 and another found that stronger AIs significantly and independently predicted overadherence (ie, ignoring practitioner recommendations and attempting to expedite the rehabilitation process).56 Contrariwise, one study found that athletes with AIs >75th percentile were less likely attempt to return to sport prior to medical clearance.43

Exercise completion was assessed in three studies.46 50 51 Findings were inconsistent. In one study, correlational analyses identified a small positive but non-significant association between AI and exercise completion.51 Authors also entered AI as an interaction term in regression models. When entered with subjective stress,50 a small positive significant interaction was found. However, when entered with age in a different study, a negative significant association was identified.46 Researchers also found that younger athletes were significantly more likely to complete their exercises and cryotherapy treatments compared with older athletes. Interestingly, the opposite relationship was observed in an earlier study but findings were not significant.51

In alignment with the findings discussed above, athletes with stronger AIs were significantly more likely to place a greater value on and maintain greater motivation towards the rehabilitation process.58 Similarly, beliefs and attitudes regarding rehabilitation were also examined.57 Authors allocated athletes into subgroups based on their AIMS score (‘low’=<25th percentile; ‘moderate’=between 25th and 75th percentile; ‘high’=>75th percentile). Athletes in the ‘high’ subgroup reported significantly greater positive attitudes and tendencies to play through pain and injury than athletes in the ‘low’ and ‘moderate’ groups. When entered into a hierarchical regression model, AIMS exclusivity and negative affect subscales significantly predicted attitudes pertaining to toughness (ie, regarding risk, pain and injury in sport), social role choice (ie, willingness to accept risk, pain and injury in sport as a part of the athlete role) and ‘pressed’ (ie, the perception of pressure felt from others to play with pain and injury) across each subgroup. However, only the AIMS negative affect subscale was found to be a significant independent predictor of perceived injury behaviours (ie, intention to play through injury).57 A similar finding was identified by Kroshus et al in their investigation of concussion reporting behaviours. They found that athletes with stronger AIs were slightly and significantly more likely to engage in non-reporting behaviours than athletes with weaker AIs.42 Additional variance was explained when perceived concussion reporting norms were added to their model (see table 2, column 7).

Injury-specific outcomes

Injury severity, risk and functional outcomes were examined in several studies. Significant small effect sizes were identified between AI and physician-rated injury severity.26 Similarly, another study indicated that stronger AIs were moderately positively and significantly associated with concussion symptom severities at follow-up time points (~14–21 days and ~21–28 days post-concussion). When entered into a hierarchical regression model, AI significantly predicted post-concussion symptom severities ~21–28 days following injury.54 With respect to injury risk, one study found that athletes with AIMS scores <25th percentile faced a greater risk compared with those >25th percentile, but this difference was not significant.43 Notably, athletes with AIMS scores >75th percentile were significantly more likely to have incurred a subsequent injury during the data collection period.

Only one study assessed functional recovery outcomes. Measured 6 months following ACL reconstructive surgery, AI was moderately positively and significantly associated with improved joint stability (ie, less anterior and posterior laxity in the knee joint, improved single leg hopping scores and improved subjective knee function (ie, limping, locking, instability, support, swelling, stair climbing and squatting)).51 Findings were replicated in regression models which indicated that AI was a significant and positive independent predictor of joint stability. Psychological distress was identified as a significant negative independent predictor (see table 2, column 7).


Measures assessing subjective ratings of pain were administered in six studies, however only two analysed pain ratings in relation to AIMS scores.47 52 Both studies identified small negative non-significant associations between AI and post-surgical pain ratings (see table 2, column 7).

The relationship between injury as an exposure and AI as an outcome

Of the three studies that assessed AI at multiple time points,44 45 48 only one44 assessed AI prior to and following injury. One study found that AIMS scores decreased significantly over time (pre-surgery compared with 6, 12 and 24 months post-surgery) after adjusting for age, sex and rehabilitation progress.48 Scores did not change significantly between pre-op and 6 months nor between 12-month and 24-month follow-up, but all other comparisons were significant. Madrigal and Gill also assessed AIMS at two time points: pre-season and return to sport.44 Small decrements in AI were observed but were non-significant. The final study did not conduct tests of statistical significance45 (see table 2, columns 6 and 7).

Study strengths and limitations

The studies captured within this review have several strengths and limitations for the reader to consider. First, the body of literature spans a 25-year period (1993–2018). This artefact implies that any trend or change with respect to athletes’ conceptualisation of AI that may have occurred as a result of cultural progression (ie, a shift over time in group norms, the importance of the athlete role, and cultural values and ideals as they pertain to sport) is represented within the data itself. Most studies either defined a specific injury (eg, ACL tear) or provided an operational definition of sport injury, thus ensuring that inclusion criteria were applied consistently. Due to exclusive use of the AIMS, AI was conceptualised and assessed equivocally across all studies. This allows for a direct comparison of AIMS scores from one study to another. Finally, almost half of the studies included athletes from a variety of sport backgrounds, increasing the external validity of these respective studies’ findings.

One of the most important limitations for readers to consider is that AI was not the primary construct of interest within the majority of the studies identified; only seven studies26 41 43 48 55 57 explicitly stated that AI was a primary variable of interest within objective statements, and therefore the main variable of interest within statistical tests. Therefore, it is possible that significant relationships between AI and the assessed injury outcomes were present but went unidentified. Being that a self-report measure was used to quantify the strength of AI, reports may have been skewed by a social desirability bias; athletes may have reported a stronger AI than their actual AI because this would be seen as desirable to other members (eg, teammates, coaches) of their social group. Another limitation with respect to the AIMS was timing and frequency of administration; 17 of 22 studies administered the AIMS following an injury and 19 studies administered the AIMS at one time point. Therefore, the existing body of literature cannot speak definitively to (1) any change over time with respect to the relationships observed between AI and the various injury outcomes observed, and (2) the relationship (if any) that exists between an injury (as an exposure) and AI (as an outcome).

Being that most studies were conducted in the USA, findings represent athletes who embody Western cultural values and attitudes towards sports and athletics. Females and athletes who identify as having a disability (eg, para-athletes) are under-represented in the literature, thus limiting the applicability of findings to these athlete populations. Studies captured a variety of MSK injuries, but few investigated AI in athletes who had sustained a sport-related concussion. Findings may not be generalisable to this population. The majority of studies had small samples sizes (n<100: n=15; n>100: n=7). This may have limited the type (eg, correlation vs regression modelling) and the extent (eg, number of predictor variables included in regression models) of statistical tests performed by authors. Overall, sport involvement (eg, frequency and years of involvement) as well as injury severity was poorly described within most studies. This oversight makes it difficult to gauge the dose–response relationship that exists between sport involvement and AI, and how this then relates to the injury outcomes observed (see table 2, column 8).


Literature describing the relationship between AI and sport-related injury outcomes has grown steadily over the past 25 years. Importantly, 18 of 22 studies identified for inclusion in this review originated from the USA. This is important to consider when interpreting the findings presented herein given the cultural importance that different societies place on specific sports and the athlete role.60–62 The athletes described were representative of many different sports and varying levels of competition, thus increasing the external validity of this review’s findings to the general athlete population. Importantly, half of the identified studies referenced a theoretical model to inform study design and methodology. However, most investigators did not discuss or interpret their findings within the context of the models originally used to position their work. The integration of novel findings as they relate to the theoretical injury outcome models referenced is necessary to progress towards predictive modelling.

Injury outcomes associated with AI were grouped into five categories. Psychosocial, behavioural and injury-related outcomes dominated the literature, with relatively few studies reporting results within demographic and pain-related categories. Several studies identified moderate to strong positive relationships between AI and depressive symptoms following injury. This aligns with cognitive diathesis–stress models of depression63–68 as well as previous research that has identified sport injury as a risk factor for depression in athletes.69–72 When an athlete is unable to engage in sport, as is the case when an athlete sustains an injury, depressive symptoms may occur due to ego dissonance (ie, an incongruence between who an individual believes themselves to be and their ability to fulfil their role responsibilities). As per cognitive diathesis–stress models,67 athletes low in self-complexity (ie, a self-schemata consisting of a limited number of identities or significant identity overlap) are subject to a greater risk of experiencing depression following an identity disruption (eg, a sport injury) than athletes who maintain a multifaceted self-schemata (ie, maintenance of multiple identities and roles). However, this explanation fails to account for if and how the strength and importance of a given identity (eg, AI) moderates depression risk. Alternatively, depressive symptoms may manifest due to the fact that the athlete is no longer receiving the reciprocal benefits associated with role engagement. For example, studies captured in this review identified a significant positive relationship between AI and physical self-worth35 and general motivation.51

Behaviourally, evidence suggested that athletes with stronger identities were more likely to overadhere to prescribed rehabilitative protocols.55 56 This could be due to an athlete’s attempt to remain in an ego syntonic state. The athlete seeks congruence between who they think they are (an athlete) and their associated role responsibilities (engaging in competition, training with teammates), so they engage in behaviours that will expedite their recovery. This behaviour may be useful, as evidence suggested that stronger AIs were associated with improved functional outcomes.51

Interestingly, pain appears to be negatively associated (although non-significantly) with AI. This might suggest that an element of mental toughness or grit accompanies stronger AIs (ie, the ability to play through and downplay pain); both of the above traits having been previously associated with sport involvement.1 2 It may also be the case that athletes with stronger AIs develop better coping skills to deal with injury pain and are better equipped to “push through”. An alternative explanation: athletes with stronger identities opt to “push through” minor injuries and ignore minor indicators of injury (ie, pain) up to a certain threshold, which is supported by study findings.47 52 Additional support for this explanation is provided by studies that identified positive significant associations between AI and injury severity.26 54

As stated previously, only three studies44 45 48 assessed AI at multiple time points, with only one of these three having assessed AI prior to and following injury.44 Based on the available literature, there is insufficient evidence to define the relationship that exists (if any) between an injury (as an exposure) and AI (as an outcome).

Strengths and limitations

Readers should consider the following strengths and limitations of the methodology used in this review. The search strategy used to identify studies was co-constructed with the help of a University of Toronto librarian. This collaboration ensured that (1) the relevant databases for the review topic were searched, (2) the search strategy notation was applied correctly for each database, and (3) that the search terms (eg, key words, subject headings) were exhaustive and appropriate to capture studies relevant to the review topic. To prevent bias, Covidence was used to blind reviewers’ decisions to accept or reject articles throughout all screening stages. Use of Covidence also ensured that all studies identified within the search were reviewed (ie, records were not missed). Finally, data extraction was conducted independently by both reviewers. This reduced the probability that study findings were transcribed erroneously within the data table and summarised incorrectly.

With respect to methodological limitations, authors did not conduct a quality and bias assessment of the identified studies. This is required and necessary prior to delineating implications for clinical care or conducting an intervention that seeks to alter AI in an attempt to improve injury outcomes. However, authors wish to remind readers that this is not the purpose of a scoping review73 and is instead better suited to a systematic review. Researchers who wish to update this review with newly published literature should consider the use of a rigorous and widely accepted method of qualitative evaluation (eg, Downs and Black’s Checklist for Quality Assessment74). The exclusion of qualitative studies, theses/dissertations and non-English articles may have resulted in the exclusion of relevant data. Finally, the search strategy used herein primarily used databases (eg, PubMed) to identify relevant studies. The incorrect labelling (eg, MeSH subject headings) of studies or studies published within journals not indexed within the databases searched were therefore missed (if any).


Findings from this review highlighted several significant and positive associations between AI and psychosocial (eg, depressive symptoms, performance traits, physical self-worth, motivation), behavioural (eg, rehabilitation overadherence, playing through pain and suspected injury) and injury-related (eg, function and injury severity) outcomes. Assessing AI prior to the start of a rehabilitation protocol may give both the athlete and treating clinician a road map of what to expect with respect to mindset, behaviours and recovery outcomes. Importantly, readers should consider the floor and ceiling effects of AI with respect to the relationships identified. A somewhat limited variability in mean AIMS scores does not allow for a complete representation of the AI as it relates to injury outcomes. Future studies should aim to capture athletes with a wider range of AIMS scores (ie, AI of varying strengths) as well as non-athletes who have also experienced an injury. Readers should also consider the over-representation of Caucasians, males, able-bodied athletes and MSK injuries identified in this review. Homogeneity in these domains limits the external validity of findings to other racial groups, females and populations with sport-related concussion. Subsequent studies should include para-athletes as no study included in this review considered this population. Importantly, limitations associated with study design and methodology within this body of literature preclude any causal inferences from being made (ie, AI as a cause of the injury outcomes observed).

This review also highlights a large gap in knowledge with respect to the association (if any) that exists between injury (as an exposure) and AI (as an outcome). Studies must adopt prospective longitudinal designs that assess AI prior to and following the occurrence of injury in order to speak to this relationship. Additional consideration should be given to the inclusion of multiple long-term follow-up observations. As per the Wiese-Bjornstal et al injury model,30 an athlete’s cognitive appraisal of the injury event is a central tenant to the outcomes observed. Despite its importance, few studies directly assessed an athlete’s cognitive appraisal of their injury. Researchers may wish to inform the development of their study protocols while referencing a theoretical model. This will facilitate a more holistic understanding of the outcomes observed.

Data availability statement

All data relevant to the study are included within the article or have been uploaded within supplemental files. All data extracted and summarised within this scoping review were obtained from published peer-reviewed journal articles. Please refer to articles referenced.


The authors would like to thank Mrs Erica Nekolaichuk (University of Toronto librarian) for helping construct the search strategy used within this review. The authors would also like to thank Dr Nick Reed, Dr Sakina Rizvi, and Dr John Cairney for their critical review of and feedback on the original scoping review protocol. The authors also wish to thank the peer reviewers for their comments; their feedback has substantially improved the quality of this written work.


Supplementary materials


  • Contributors TR was responsible for establishing the research questions, developing and conducting the literature search, performing the title, abstract and article screening process, extracting data from eligible articles, drafting and submitting the manuscript for publication, as well as responding to peer-reviewer feedback and completing the required revisions. BP was responsible for performing the title, abstract and article screening process, extracting data from eligible articles and contributing to results and discussion sections of the manuscript draft. SK was responsible for helping establish the research questions, advising on data extraction elements and editing/revising the manuscript draft prior to submission for publication.

  • Funding This work was supported by a Canadian Institutes of Health Research (CIHR) Frederick Banting and Charles Best Canada Graduate Scholarship Doctoral Award (CGS-D) (fund # 505508).

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.