Shoulder pain and injury risk factors in competitive swimmers: A systematic review

To synthesize and assess the literature for shoulder pain and injury risk factors in competitive swimmers.


| INTRODUCTION
Swimming is popular as both a form of recreational exercise 1,2 and as a competitive sport, with over 22 000 swimmers competing in the National (American) Collegiate Athletic Association 2021-2022 season. 3As a form of recreation or exercise, swimming has numerous health benefits with relatively minimal stress/load on bones and joints, 4,5 making it a popular activity for all ages. 6However, competitive swimming involves consistently high training volumes and intensities, increasing the risk of injury. 7otwithstanding swimming's benefits, shoulder pain and injury prevalence and incidence is high in competitive swimmers.Injuries in competitive swimmers are typically ascribed to repetitive tissue strain. 8This is understandable given that these athletes perform thousands of shoulder revolutions each session for 10-12 sessions/week, 9 with some swimmers covering up to 110 km/week. 7While injuries are observed in the knee, lower back, and hip, the shoulder is undoubtedly the predominant location of pain and injury in competitive swimmers. 10 recent review of injury in competitive swimmers reported a total-body injury incidence of 2.6-3.0 injuries per 1000 h of swimming and a shoulder-specific injury prevalence of 23%-51% and 33%-41% in men and women respectively. 10Further, previous research shows self-reported shoulder pain in up to 97% of college-level swimmers (n = 30; 17M, 13F) during strenuous exercise 11 and 91% of elite junior competitors (n = 80; 42M, 38F) over a 1-month period. 7Despite advancements in sports science and rehabilitation, including improved understanding of intrinsic and extrinsic risk factors, shoulder pain and injury prevalence in competitive swimmers has not decreased suggesting missing or misunderstood risk factors.There may even be a rising trend for shoulder injuries given reports of a prevalence of 26% in 1993, 12 34% in 2002, 13 and 38% in 2012. 14These consistently high injury rates warrant further investigation into identifying risk factors to guide athlete screening/monitoring and injury prevention programs.
There is a growing body of evidence identifying pain and injury risk factors.Struyf et al. 15 reviewed the differences musculoskeletal dysfunctions in injured and non-injured swimmer's shoulders.They suggested four key areas for clinicians to be aware of when assessing and treating a swimmer's shoulder: (i) shoulder range of motion (ROM), (ii) shoulder laxity and instability, (iii) shoulder posture, and (iv) scapular dyskinesis.Hill et al. 16 previously conducted a review of shoulder pain and injury risk factors, identifying 18 risk factors, which were categorized into four groups: (i) shoulder joint anatomy and strength, (ii) activity history, (iii) demographics, and (iv) musculoskeletal determinants.This 2015 review included three prospective studies 14,17,18 ; however, since this review an additional six prospective studies (n = 457) [19][20][21][22][23][24] have been published and should be considered to further our understanding of shoulder injury risk factors in swimmers.There is a clear need to synthesize these data to aid practitioners in their athlete screening and injury prevention strategies.Therefore, this review aims to systematically synthesize the current literature and provide an update on shoulder pain and injury risk factors in competitive swimmers.

| METHODS
This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) 25 and registered on PROSPERO (CRD42021234093).

| Eligibility criteria
Original cohort, cross-sectional, and case-control studies, published in English were included in this review.To be included, studies must (i) include competitive swimmers aged from 10 to 40 years, this included any article that stated their sample was "competitive", (ii) provide a definition of pain/injury (use of a valid painscale tool or a positive orthopedic test were accepted), (iii) report an outcome measure including at least one potential shoulder pain or injury risk factor, (iv) use inferential statistics, and (v) in the event of multi-injury/ sport studies were only included if shoulder data specific to swimming was presented in isolation.Studies were excluded if they: (i) did not include competitive swimmers, (ii) if nonconservative/surgical interventions were applied, (iii) involved participants with medical movements.Importantly, shoulder strength-endurance may be the most clinically relevant modifiable intrinsic risk factor.

K E Y W O R D S
athlete screening, injury prevention, rehabilitation, shoulder injury, shoulder pain, swimming conditions without reference to the shoulder, or (iv) involved cadavers and animal specimens.

| Literature search strategy
A librarian assisted with developing the electronic search strategy.Potentially relevant studies were identified through searching in CINHAL, SportDiscus, Scorpus, PubMed, and Embase databases between 1966 and April 2022 by two authors (S-A.L. and S.D.).The main search strings were "shoulder" AND "pain" OR "injur*" AND "competitive swim" (swim*) AND "risk factor*" (see online supplementary material S1 for full list).Synonyms, Medical Subject Headings (MeSH), Emtree terms, and major concepts derived from the main terms were used to refine our findings.Reference lists of all studies were reviewed for additional studies not found in the original search.

| Study selection
Articles from the initial database search results were imported into Endnote and duplicates removed.Remaining articles were imported into Rayyan, with two reviewers blinded before screening all titles and abstracts based on the eligibility criteria.In the event of disagreement, a third independent reviewer (S.D.) assisted to reach consensus.

| Study quality appraisal/risk of bias
Prior to quality assessment of all articles, six articles were chosen at random for independent evaluation to ensure all screening items were clear and understood by three reviewers (A.M., S-A.L., and S.D.).All disagreements were discussed and clarified as a group before quality assessment was conducted.Next, two reviewers (A.M. and S-A.L.) independently evaluated all articles with discrepancies mediated by a third researcher (S.D.).
Quality assessment of all articles was conducted using a modified Newcastle-Ottawa Scale. 268][29] The scale was modified to adjust to the specificity of the review question of interest (online supplementary material S2).Case-control and cohort studies were evaluated on a nineitem scale, with a maximum score of 12, whereas crosssectional studies were evaluated on a seven-item scale with a maximum score of 10.
Overall method quality was categorized using either high, acceptable, borderline, or unacceptable quality (Table 1).To be included in the analysis a minimum score of >33% of scale items was required, meaning that if a study did not satisfy at least 4 out of 12 (cohort and case control studies) or 4 out of 10 (cross-sectional studies) quality criteria, they were excluded from analysis.This method quality scoring was adopted and modified based on a previous systematic review on shoulder injury risk factors. 30

| Data extraction
Two authors (A.H. and S.D.) extracted half of the data into online supplementary material S1 which was then crosschecked (A.M.).Risk factors were categorized into one of four: (i) modifiable-intrinsic, (ii) modifiable-extrinsic, (iii) non-modifiable, and (iv) other/secondary risk factors and defined as follows: (i) Modifiable-intrinsic-any outcome measure with a demonstrated ability to change over time, that are measured on or within the individual (e.g., strength and range of motion (ROM)).(ii) Modifiable-extrinsic-any outcome measure with a demonstrated ability to change over time, that are measured external to the individual, or work performed by an individual (e.g., training load).(iii) Non-modifiable-any measure that has no demonstrated ability to change over time (e.g., injury history).(iv) Other/secondary-any measure that is potentially modifiable but lacks strong evidence regarding their modification (e.g., scapula position).

| Synthesis of results
This study used best evidence synthesis 31 as an alternative to meta-analysis due to the heterogeneity in outcome measures.This method has been used in the past to present the direction and strength of investigated risk factors. 30All risk factors were provided an evidence strength, and a direction of association.The strength of evidence was calculated using the criteria previously applied to suit risk factors in overhead sports, 30 Table 2 describes these criteria.The direction of association was indicated using either "for", meaning the evidence supports an association, or "against" meaning the evidence does not support an association.Several articles reported significant associations in a sub-sample (i.e., different age ranges, or male vs. female) of their study. 7,23,24,32The synthesized results reported in Table 4 only report associations found overall, with all significant associations found in subsamples reported in-text.Two articles only reported associations in subsamples of their data, with no analyses of the total data presented.When associations were not reported consistently across subgroups, that is, significant associations in one age group but not another, the authors of the original research were contacted and asked to provide their data to allow reanalysis following the methods as described originally (i.e., pooled data rather than in arbitrary age groups).

| Study selection
From the database search, a total of 1356 studies were identified.After removing duplicates and excluding studies not addressing the objective of our review, 55 studies remained and were screened based on inclusion criteria.Reference lists of the relevant papers were screened, and 5 additional studies were identified and assessed.A total of 24 papers were evaluated for quality, with 22 of these included in the best-evidence synthesis (Figure 1).

Strong evidence
Evidence provided by two or more high-quality studies and by generally consistent findings across these studies (≥75% of the studies reported consistent findings)

Moderate evidence
Evidence provided by one high-quality study and/or multiple studies of acceptable quality and by generally consistent findings (≥75% of the studies reported consistent findings)

Limited evidence
Evidence provided by one study of acceptable quality and/or one or more studies of borderline quality

Conflicting evidence
Inconsistent findings in multiple studies (<75% of the studies reported consistent findings) Note: Evidence criteria categories adopted from Asker et al. 30 T A B L E 2 Evidence criteria for best evidence synthesis.

| Synthesis of results
The best evidence synthesis is presented in Table 4.

| DISCUSSION
This review included 22 studies that investigated at least one shoulder pain or injury risk factor.No study was rated high quality and half of the included articles were borderline, using modified best evidence synthesis criteria. 30oth muscle recruitment (via surface electromyography [sEMG]) and the swimmer's competitive level had moderate evidence to support an association with shoulder pain or injury.However, there was no direction of association in the swimmers' competitive level, meaning that some studies indicated a higher injury incidence in elite swimmers whereas others indicated a higher incidence in sub-elite swimmers (Table 4).This is likely a result of different training practices of coaches.Similarly, various inconsistent findings were reported regarding muscle activity in injured swimmers compared to non-injured swimmers.Importantly, no study has investigated muscle activity using in-water sEMG, rather opting for nonecological tasks.Further, while studies consistently report differences in either sEMG amplitude or onset timing, the exact measures have not been replicated between studies.For example, during a standardized functional upper limb task Hidalgo-Lozano et al. 41 reported increased activity of scalene bilaterally in injured swimmers but no changes in upper trapezius.Conversely, Sabzehparvar et al. 43 reported increased upper trapezius, serratus anterior, and latissimus dorsi, but not of middle or lower trapezius (further discussion can be found in Section 4.4).This review also found a moderate level of evidence against the associations of several measures: external and internal rotational ROM, middle finger distance (backscratch) test, weight, training frequency (sessions/week), freestyle as a swimmers' specialty stroke, height, and sex.Meta-analysis of data was not possible due to the heterogeneity of studies including the lack of consistency in shoulder pain or injury definitions and limited risk factor reappraisal.Further, when risk factors were reappraised researchers often used alternative methods such as different speeds for the same isokinetic strength test.Overall, this review highlights the dearth of high-quality research investigating possible shoulder injury risk factors in competitive swimmers.We also highlight measures which may, and may not, be relevant in swimmer screening programs and future research.

| Modifiable-intrinsic risk factors
The literature has examined modifiable-intrinsic risk factors across four key areas, related to: ROM, isometric shoulder strength, isokinetic shoulder strength, and other (non-shoulder) strength and strength endurance.However, no modifiable-intrinsic risk factor had greater than limited evidence in support.
Improving shoulder ROM appears to be a common focus in competitive swimming training and injury prevention programs, evidenced by the 12 articles included in this review that investigated a ROM measure.While several individual studies report significant associations between ROM and shoulder pain or injury, 7,14,19,21,24,34,37 the best evidence synthesis suggests the literature does not support an association with any ROM measure.Of the 13 ROM measures investigated, 35 statistical tests of association were conducted, with only 8 reporting a significant association.There is conflicting evidence regarding the associations of pectoralis minor length, horizontal abduction ROM, glenohumeral laxity, and shoulder rotation T A B L E 3 Study scores for risk of bias and overall method quality.
T A B L E 4 Best evidence synthesis of shoulder injury risk factors in competitive swimmers.

Studies reporting no association with risk
Best evidence synthesis (strength, association)

Flexion strength
Refs. width and injury.With regards to studies reporting associations, several interesting findings should be highlighted.
In a prospective cohort study Walker et al. 14 identified an increased risk of injury in swimmers with glenohumeral external rotational ROM <93° and ≥100°, 14 suggesting there is an optimal window of flexibility.Bansal et al. 34 also reported greater rotational ROM in injured swimmers in a cross-sectional study; however, eight other studies reported no association (Table 4).Further, when assessing males and females separately in a prospective study, Mise et al. 24 reported decreased shoulder rotation width in males, and increased width in females were associated with shoulder pain.Shoulder rotation width is a test whereby swimmer grasps a rod with both arms in full elbow extension, the minimum hand-to-hand distance the swimmer can move the rod from in front of their torso to behind their torso is the score they receive.A smaller distance indicates greater shoulder complex mobility.Additionally, Mise et al. 24 also found that decreased external rotation ROM was associated with pain in males' right but not left shoulders, and not in either female shoulder.
No statistical association was found in the combined data in the left and right shoulders of males and females, as presented in Table 4.The pragmatism of identifying different risk factors between left and right shoulder is questionable, particularly when these findings are not reproduced across sexes.It could be argued that breathing side may influence shoulder kinematics and thereby alter risk between shoulders; however, unilateral breathing has been assessed and found not to be associated with shoulder pain. 32Finally, while glenohumeral laxity was not associated with shoulder injury, 7,22 Sein et al. 7 did find joint laxity to be associated with impingement and extreme pain.However, these authors concluded that as joint laxity was not associated with training load, it is likely not a main contributor to shoulder pain.Future work should be directed toward investigating the ROM with conflicting evidence.
8][49] However, of the 27 different isometric and isokinetic shoulder strength measures investigated in six articles, only three significant associations were reported by two articles. 20,40The three significant associations were various external:internal rotational isokinetic strength ratios. 20,40In a cross-sectional study, Bak and Magnusson 40 found significantly greater eccentric external rotation:concentric internal rotation in injured swimmers compared to healthy swimmers but reported no difference in concentric external:concentric internal or eccentric external:eccentric internal rotations.Similarly, Drigny et al. 20 reported an eccentric external rotation:concentric internal rotation strength ratio <0.68 had a relative risk of 4.5, with eccentric external rotation:eccentric internal rotation strength ratio also demonstrating predictive ability for shoulder injury risk (<0.66) in their prospective study.It should be noted that this article found no association between concentric external rotation: concentric internal rotation ratio, or eccentric internal rotation ratio: concentric external rotation ratio and shoulder injury.Further investigation should be directed toward understanding the clinical significance of such ratios, especially regarding eccentric shoulder rotator contractions given their absence in swimming.While rotational strength imbalances appear to be associated with shoulder injury, maximal shoulder strength (relative to bodyweight or absolute) does not.However, while current evidence is limited given the lack of replication, it is plausible that shoulder strength-endurance may be more indicative of injury compared to maximal strength.
The highest rated prospective cohort study included in this review investigated the posterior shoulder endurance test and its relationship to shoulder pain. 21The test, first

Scapula dyskinesis
Refs. [22,32]   N = 201 (201F) Note: Risk factors are presented with studies that found a significant association, studies that found no significant association, and the best evidence synthesis.
Strong evidence, moderate evidence, limited evidence, conflicting evidence, against an association, in favor of an association, unknown association, an * indicates this study found a significant difference in a subsample of their studied cohort.
T A B L E 4 (Continued)  described in baseball players, 50 involves an athlete laying prone on a plinth with their test arm hanging vertically off the side.The participant then raises their arm to 90° horizontal abduction with a dumbbell weighing approximately 2% of bodyweight, the participant repeats this until volitional exhaustion. 21,50This exercise has been shown to elicit high levels of activation in trapezius, supraspinatus, and infraspinatus. 50In their sample of competitive swimmers, Feijen et al. 21found for every one additional repetition the swimmer performed, their risk of shoulder pain decreased by 5%.Tate et al. 11 modified the posterior shoulder endurance test and found a significant increase toward the end of the season.Unfortunately, this investigation did not statistically analyze the test's relationship with shoulder pain.Beach et al. 35 further support the association of strength-endurance in abduction and external rotation and pain.This cross-sectional study found mean peak torque (after 50 repetitions at 240°/sec) in both external rotation and abduction accounted for 54% and 60% of the variation in shoulder pain rating in left and right arms, respectively.In addition to abduction and external rotation endurance, one case-control study found multiple significant findings relating to various measures of shoulder and core strength-endurance. 46While being careful not to overstate the clinical relevance of shoulder strength-endurance given its limited evidence, is perhaps the modifiable-intrinsic risk factor with the greatest supporting evidence.When combined with its capacity to be modified with appropriate shoulder injury prevention programs, strength-endurance has potential to be a highly clinically relevant measure.Due to the impact of shoulder fatigue during swimming, it is possible that increased endurance of the posterior rotator cuff can enhance the stability of the humeral head in the glenoid.In turn, this may help mitigate the shoulders tendency for functional impingement, particularly during longer swimming sets. 51Future studies should reappraise shoulder strengthendurance, particularly the posterior shoulder test, to confirm its association with shoulder pain or injury, and whether any such association is causal or consequent.

| Modifiable-extrinsic risk factors
Despite 81% of Olympic swimming events taking less than 140 s, training volume is relatively high compared to other cyclical sports. 52Barry et al. 53 surveyed 31 swim coaches and support staff (78% of whom had >10 years experience in competitive swimming) and found that 96% monitored training distance and 92% measured session rating of perceived exertion (RPE).Of these experts, 38% perceived their training volume monitoring practices to be either "very" or "extremely effective" in preventing injury.
However, the current evidence does not support these expert beliefs.This review found moderate evidence against the association of training frequency (sessions/week) and injury, limited evidence against training time, and conflicting evidence supporting an association of training distance and shoulder injury.No included study investigated session RPE.
Although perhaps logical to consider absolute training load (i.e., training time and/or distance) as injury risk factors, only two included articles reported a significant association between injury and training time 7 or distance. 7,24Mise et al. 24 found injured swimmers swam significantly further than non-injured swimmers when males and females were pooled together.However, when separated by sex, there was a significant difference in females only.Further, Sein et al. 7 found all who swam >20 h/week, or >60 km/week exhibited supraspinatus tendinopathy on MRI, and swimming >15 h/week doubled the chance of tendinopathy while those swimming >35 km/week had four times the rate of supraspinatus tendinopathy.While these absolute training load measures have some individual support as shoulder injury risk factors, the overall evidence is not conclusive.An earlier systematic review of swimming training load and pain, injury, and illness 54 also reported no clear evidence of association between training volume and pain but did present one article 55 that found an association between retrospectively reported yearly training distance and injury.However, this article was excluded from the current study as shoulder injury data were not presented separately from other injuries.Another study 13 did report a significant association between weekly training time and pain but was also excluded from the current study due to unacceptable method quality (see online supplementary material S3).
A possible reason why no clear association is seen between training load and injury is that these absolute measures fail to account for the rate of load application.For example, this monitoring fails to differentiate between swimming 1 km in 50 m maximal efforts, versus 1 km in recovery.An example of a swimmer's relative training load is in their acute:chronic workload ratio.In fact, this ratio was monitored by 31% of coaches in Barry et al. 53 sample of expert practitioners.In swimming, acute:chronic workload ratio has been investigated by one research team who found every 1-unit increase in this ratio lead to a shoulder injury odds ratio of 4.3. 21However, the use of this metric has been scrutinized for its inaccuracies and statistical artifacts and cannot currently be recommended for use in reducing injury risk. 56With the advent of water-proof inertial measurement unit's, more specific shoulder measures may become available to monitor including number and velocity of shoulder revolutions, potentially increasing the validity of relative training loads.Such research is currently being conducted in water polo, 57 and should be repeated in swimming populations.Future studies should also consider monitoring internal load including RPE, sleep, and psychological inventories as suggested by Soligard et al. 58 Further to training load, the swimmer's competitive level had moderate evidence supporting its association with shoulder injury.Two cross-sectional and one cohort study have analyzed competitive level, and all found a significant association. 7,21,38However, the direction of this association is not clear.Sein et al. 7 found that swimmers who competed at a higher level had greater rates of supraspinatus tendinopathy.Feijen et al. 21found regional swimmers had a lower injury risk than club swimmers, but no significant differences were present for national versus club, or international versus club swimmers.Matsuura et al. 38 found that elite swimmers who made the Japanese Olympic team had less injury prevalence than the elite swimmers who did not qualify for the Olympics.Matsuura et al.'s study is likely confounded as injured swimmers would likely be underperforming, therefore negatively impacting team selection.While there is moderate evidence suggesting different competitive levels have different injury rates, the relationship between these is not clear and is potentially an artifact of different training volumes or intensities, both in and out of the water.

| Non-modifiable risk factors
There was moderate evidence against an association between height, sex, and age with shoulder pain.While there was only limited evidence supporting shoulder injury history and future injury, articles reporting an association exhibited dramatic effect sizes. 14,22,33A history of shoulder injury increased risk of future injury, by 4.7, 33 7, 22 and 4.1 times. 14Further, Walker et al. 14 showed those with a shoulder injury history were 11.3 times more likely to sustain a future shoulder injury lasting >2 weeks.These large effects highlight the need for primary injury prevention in adolescent swimmers.However, this may be somewhat inflated due to articles potentially failing to differentiate between an index injury, exacerbation, reinjury, and new injury.These terms have previously been defined in a consensus statement on methodologies for injury surveillance in aquatic sports. 59All future research should consider adopting these terms to not only avoid the inflation of injury history as a risk factor, but also to improve inter-study consistency allowing for future metanalyses.

| Other/secondary risk factors
Injured swimmers appear to exhibit different muscle activity profiles when performing functional upper extremity tasks, 41,43 and scapula elevation. 45This has led to thus far unsupported speculation as being a possible causal mechanism for injury recurrence and suggestions that motor control exercises be used in injury rehabilitation. 41,43,45imilarly, it has been suggested that swimmers reporting lower pain-pressure thresholds 42 and fatigue-related changes in scapula orientation 44 are at higher risk of injury.However, caution is recommended when concluding these are shoulder pain or injury risk factors.These articles tested swimmers with and without shoulder pain in a single session potentially experiencing some pain inhibition requiring compensatory increases in activity of other muscles.Due to the study designs, it is not possible to conclude whether these other/secondary risk factors were present prior to the shoulder pain or occurred with the presence of pain.This is particularly relevant for Hidalgo-Lozano et al. 41 who claim the presence of trigger points leads to the development of shoulder pain based on the finding that swimmers with painful shoulders presented with active muscle trigger points.

| Findings from prior reviews
The current study expands upon a 2015 systematic review by Hill et al. 16 who reported joint instability, internal and external rotation ROM, pain history, and the swimmer's competitive level as the shoulder injury risk factors with the highest level of confidence.Our updated evaluation incorporating recent research, challenges the perceived association between rotational ROM and shoulder injury, and downgrades joint laxity to a conflicting evidence.Nevertheless, the updated evidence continues to support the swimmer's competitive level (moderate evidence) and pain history (limited evidence) as variables associated with shoulder injury.
Hill's review also reported several variables associated with shoulder injury but supported with low confidence, the majority of these now contravened by the most recent evidence, including rotational strength, sex, and breathing side.In fact, training volume is the only variable that is still supported by the current evidence.In addition, years of swimming experience and use of hand paddles currently displaying conflicting evidence, adding a layer of complexity to their roles as potential contributors to shoulder injury.

| Limitations and recommendations
Before swimming practitioners apply any of the current review's findings, it is important to highlight that it is not known whether the presented risk factors are causal or noncausal.This synthesizes known associations between various measures and shoulder pain and injury, some of these measures may occur concurrently with pain or injury without being an underlying cause.Currently, it is not known if altering any risk factor will decrease injury incidence.Preliminary evidence suggests exercise-based programs in swimmers could reduce shoulder injury incidence. 48,51,60,61Nevertheless, before meaningful conclusions can be drawn, subsequent investigations should include the following methodological improvements, larger samples, refined pain definitions, and longer monitoring periods.For swimming practitioners looking to use the presented evidence for their athlete monitoring or injury prevention programs, we highlight the following areas for consideration: (i) A swimmer's ROM is not likely a predictor of injury.
Monitoring changes throughout a season/career may produce meaningful outcomes; however, it is not clear if "improving" ROM is beneficial.(ii) Maximal shoulder strength is likely not associated with shoulder injury.While studies have associated various shoulder rotation strength ratios with pain, due to the lack of maximal eccentric contraction during swimming, we do not believe these ratio's to be practically relevant injury risk factors.Improving shoulder strength-endurance may be of more benefit.(iii) Monitoring absolute training load as a means of shoulder injury risk is not supported by the literature.Relative training loads and new technologies may be worth consideration.(iv) Clinicians and coaches should be aware that previously injured swimmers are at high risk of re-injury.
Greater care and monitoring for swimmers with shoulder injury history is needed during their rehabilitation and return-to-swim programming.Efforts in primary injury prevention may be exponentially beneficial.
To improve on the current research, there is a need for highly controlled prospective studies in competitive swimmers.Risk factors presented in this review should be reappraised using the same methods in different populations to validate the measure and cutoff scores. 62We suggest the need for risk factor reappraisal is greater than the need for additional exploratory investigations given the current body of knowledge.Excluding training load monitoring, potential risk factors are typically monitored at a single time point while the associated injury may occur several months later.Future prospective studies should monitor modifiable-intrinsic risk factors at regular intervals throughout the collection period (e.g., every 1-3 months), reflecting their modifiable nature.Tate et al. 11 have shown many shoulder metrics change throughout the season, including posterior shoulder endurance, horizontal adduction ROM, and internal and external ROM, highlighting the need to regularly monitor.However, this article failed to statistically associate these variables with pain.As more prospective studies are conducted, it is critical for subsequent systematic reviews to consider their approach.These reviews should exclusively include only prospective designs or weight these studies greater than crosssectional studies due to their inherently greater quality.Future research should adopt the injury definitions defined by Mountjoy et al. 59 , this is particularly relevant to the definitions of pain and injury, index injury, exacerbation, reinjury, and new injury.Adopting these definitions will allow future reviews to pool data between studies improving inferences that can be drawn.New technologies, such as inertial measurement units, should be investigated particularly as they relate to relative training loads.Finally, internal training loads such as RPE, sleep quality, and other psychological metrics should be investigated for their possible role in performance and injury.

| PERSPECTIVES TO SPORTS MEDICINE
A thorough understanding of shoulder injury risk factors is critical for the development of athlete screening and injury prevention protocols in competitive swimmers.As evidenced in this review, there is contention regarding risk factors, with many researchers exploring new potential risk factors as opposed to confirming existing risk factors in new populations.While it is not known whether the identified risk factors in this review are causal or noncausal, clinicians may use these risk factors to develop screening protocols and injury prevention programs in their athletes.