Article Text
PDF
Abstract
Background Previous injury is a strong risk factor for recurrent lower limb injury in athletic populations, yet the association between previous injury and a subsequent injury different in nature or location is rarely considered.
Objective To systematically review data on the risk of sustaining a subsequent lower limb injury different in nature or location following a previous injury.
Methods Eight medical databases were searched. Studies were eligible if they reported lower limb injury occurrence following any injury of a different anatomical site and/or of a different nature, assessed injury risk, contained athletic human participants and were written in English. Two reviewers independently applied the eligibility criteria and performed the risk of bias assessment. Meta-analysis was conducted using a random effects model.
Results Twelve studies satisfied the eligibility criteria. Previous history of an ACL injury was associated with an increased risk of subsequent hamstring injury (three studies, RR=2.25, 95% CI 1.34 to 3.76), but a history of chronic groin injury was not associated with subsequent hamstring injury (three studies, RR=1.14, 95% CI 0.29 to 4.51). Previous lower limb muscular injury was associated with an increased risk of sustaining a lower limb muscular injury at a different site. A history of concussion and a variety of joint injuries were associated with an increased subsequent lower limb injury risk.
Conclusions The fact that previous injury of any type may increase the risk for a range of lower limb subsequent injuries must be considered in the development of future tertiary prevention programmes.
Systematic review registration number CRD42016039904 (PROSPERO).
- subsequent injury
- sports injury
- athletic injury
- injury risk
- epidemiology
Statistics from Altmetric.com
Introduction
Lower limb injuries often occur following previous injury in athletic populations.1–3 Subsequent injury risk increases as the number of previous injuries rises.4 5 The first, or index injury, can be minor, whereas subsequent injuries may be more severe. This is evidenced by greater sports incapacity (time-loss from sport) and a larger burden on sports medical resources for subsequent injuries compared with index injuries.1 6 7 Furthermore, sustaining more than one injury within a season reduces an athlete’s chances of achieving their performance goals by up to 68%.8
Tertiary prevention aims to reduce health complications following an index injury. To achieve this, the risks for sustaining subsequent injury must be understood so that rehabilitation programmes, return to sport preparation and athlete load management can be tailored to mitigate this risk.9–11 Implementation of targeted preventative measures to diminish the risk of subsequent injury occurrence is possible where injury aetiology and mechanisms are known.11 12
A recurrent injury is a repeat of the same injury at the same location, whereas a subsequent injury may differ in nature or location.13 14 A previous history of a lower limb injury is a strong risk factor for recurrence of that same injury.15–18 In contrast, the relationship between previous injury and a subsequent lower limb injury that differs in nature or location is rarely considered.13 19
This review aims to synthesise the existing evidence to answer the clinical question: is previous injury associated with subsequent lower limb injury at a different anatomical site and/or of a different nature in athletic populations?
Methods
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines were followed.20 The review protocol was registered on the PROSPERO international prospective register for systematic reviews website (http://www.crd.york.ac.uk/PROSPERO) (registration number: CRD42016039904). This study considered the BJSM systematic review methodological elements recommendations21 and used the Assessing the Methodological Quality of Systematic Reviews tool.22
Literature search
A systematic search of eight electronic databases (MEDLINE, Embase, CINAHL, SPORTDiscus, Web of Science, Scopus, AMED and AUSPORT) was conducted from their origin to the 7 August 2016. Medical subject headings and free-text search terms were derived from the research question.23 The search strategy used for MEDLINE is provided in online supplementary file 1 .
To ensure that all of the relevant studies were located, the reference list of each of the included studies was screened (backward citation tracking), forward citation tracking of included articles was performed using the Web of Science (Thomson Reuters, New York, New York, USA) and the ‘similar articles’ function within the PubMed database was used.24
Study selection
Two independent reviewers (LAT and MKD) screened the records by title and abstract using predetermined eligibility criteria. The full texts of potentially eligible studies identified by either reviewer were retrieved and independently screened by the same two reviewers. Disagreement was resolved by consensus, if not a third independent reviewer (JEG) was consulted to resolve the discrepancy. Agreement between the two reviewers for title and abstract screening and full-test screening was calculated for absolute agreement and using the Prevalence and Bias Adjusted Kappa (PABAK)25 using the epiR package in the R computer programming environment.26 27
Inclusion criteria
Studies were eligible for inclusion if they: (1) reported occurrence of lower limb subsequent injury, which was of a different anatomical site and/or of a different nature to preceding injury, (2) provided some form of risk assessment or contained sufficient data for risk calculations to be performed, and (3) were written in English language.
Exclusion criteria
Studies were excluded if they: (1) only considered index injuries or recurrent injuries sustained at the same anatomical site and of the same nature on either the ipsilateral or contralateral side, (2) were not related to athletic human populations, (3) reported on surgical methods or attempted to reduce the risk of injury through interventions, (4) only included subsequent degenerative joint conditions, chronic instability, skin injuries, non-injury-related training soreness or delayed onset of muscle soreness or (5) were reviews, case series, case studies, opinion articles, abstracts or conference proceedings.
Data extraction and risk of bias assessment
Data extraction was completed by one of the authors (LAT) using a structured form. A second author (MKD) then independently verified the extracted data. Data extracted included study design, surveillance period, participant characteristics (age, gender, height, weight, body mass index, sport type and level of competition), injury and subsequent injury definitions (mapped to the Injury Definitions Concept Framework (IDCF) classifications28), injury characteristics (number, type, location, severity and mechanism) and any associated injury risk data (ORs, relative risks (RRs), incidence rate ratios and so on). The authors of three studies were contacted to provide clarification and raw injury data.29–31
Where summary risk estimates were not provided in the original papers, univariate RRs (cohort studies) or ORs (case–control) with 95% CI were calculated from the available raw data (Stata/IC V.14.1). The injury data from the three studies that used a χ2 test32 33 or Fisher’s exact test30 was used to compute a univariate RR. Data were obtained from the authors of two studies in order to segregate ACL injury as an index injury from a severe knee injury allowing inclusion of data into the meta-analysis,30 and removal of the data from 3 of the 310 participants who had either sustained a recurrent injury or a degenerative joint condition31 allowing for univariate RR calculation. Effect sizes for the point estimates were defined as follows: small effect point estimate=1.0 to <1.25, medium effect point estimate=1.25 to 2.0 and large effect point estimate >2.0.34
Risk of bias was assessed using the Newcastle-Ottawa Quality Assessment Scale (NOS) that was created for the assessment of cohort and case–control studies and also allows for customisation to reflect the review question of interest.35 36 A tailored version of the NOS, incorporating explicit decision rules, was created in a checklist format specific for sports injury research (see online supplementary file 2). Each study was assessed in three discrete domains in accordance to their study design: selection of the study participants, comparability of the different study groups and outcome/exposure result. Maximum scores of 4, 2 and 3 were awarded for these respective domains, which were considered individually and not combined for an overall total as recommended by the PRSIMA guidelines.20
Two independent reviewers (LAT and MKD) completed the risk of bias assessment for all studies. Discrepancy was resolved through discussion or a third independent assessor (JEG) if consensus was not possible. Intraclass correlation coefficients (ICC3,1) were calculated to measure the agreement between the two assessors.37
Where possible, risk of publication bias was assessed using funnel plots (RevMan V.5.3, The Cochrane Collaboration, Copenhagen, Denmark).38 A fixed effects model generated 95% CI.39 The fixed effects model was only used during the generation of graphs.
Data synthesis
Quantitative synthesis: meta-analysis
Where three or more studies examined the risk of the same subsequent injury following the same index injury using an equivalent summary statistic, a meta-analysis was considered.20 Due to the heterogeneous nature of the included studies, a random effects model with Mantel-Haenszel weighting was used.20 Heterogeneity was assessed using the I2 statistic. Heterogeneity was considered high when I2 was greater than 50%.40
Qualitative synthesis
In circumstances where it was not appropriate to pool the results due to clinical heterogeneity (varying nature of index injury, subsequent injury and statistical method), the results were ordered systematically via a forest plot without summary estimates as recommended in the PRIMSA statement.20 The results were grouped according to the statistical method used to determine the risk of subsequent injury and the type of index injury that preceded a subsequent injury.
Results
Search results
The electronic search yielded 1297 potentially relevant articles. Screening of titles and abstracts left 45 articles to be assessed by full text. There was an absolute rate of selection agreement between the two reviewers of 98% and a PABAK of 0.96 (95% CI 0.94 to 0.97). Screening full texts left 16 potentially eligible articles. There was an absolute rate of screening agreement of 80% and PABAK of 0.60 (95% CI 0.31 to 0.81). A preliminary discrepancy for study eligibility existed between the reviewers for nine of the articles.3 4 13 41–46 All nine of the studies were excluded when consensus was reached following discussion between reviewers. A further five papers were identified for inclusion during backward citation tracking, forward citation tracking and PubMed ‘similar articles’ search.47–51 At the end of this process, 12 studies were eligible for the review (figure 1).29–33 47–53
PRISMA study selection flow chart. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.
Characteristics of the included studies
Ten of the 12 included studies had a cohort design,29–32 47 49–53 while two used case–control designs (table 1).33 48 The type of index and subsequent injuries examined varied; hamstring injury was the most common subsequent injury and was evaluated in seven studies.29 30 32 47 50 51 53 The definition of index injury and subsequent injury was mixed across the studies with ‘sports incapacity’ (time-loss) being the most prevalent definition used compared with the ‘clinical examination’ and ‘athlete self-report’ definitions.28 The two most common sports investigated were Australian football29 30 32 50 and soccer.31 49 53
Summary of study characteristics
Risk of bias assessment
The ICC for absolute inter-rater agreement was 0.65 (95% CI 0.52 to 0.75). The included cohort studies had a lower risk of bias than the included case–control studies (see online supplementary file 3). The cohort studies performed best in the outcome (average score 3/3; 100%) and selection domains (3.5/4; 87.5%), while receiving lower scores for the comparability domain (1.2/2; 60%). The two case–control studies performed best in the selection domain (2.5/4; 62.5%) but were found to have lower scores in the exposure (1.5/3; 50%) and the comparability (0/2; 0%) domains. Neither of the case–control studies and only 5 of the 10 cohort studies29 30 49 50 53 controlled for both age and previous injury in multivariate or multiple variable analysis. Funnel plots were created for the assessment of publication bias; however, due to the small number of studies included in each, meaningful interpretations from the funnel plots could not be determined.
Subsequent injury risk
Quantitative synthesis: meta-analysis
Two subsequent injury types were assessed with meta-analysis. The risk of subsequent hamstring injury following an ACL injury compared against athletes without a previous ACL history was explored (three studies, figure 2), with minimal heterogeneity (I2=0%) verified. A history of ACL injury was associated with a significantly higher risk of subsequent hamstring injury (RR=2.25, 95% CI 1.34 to 3.76, p=0.002).
Meta-analysis forest plot showing RR and 95% CI of subsequent hamstring risk in athletes with a previous history of ACL injury (cases 1.1.1) or chronic groin pain (cases 1.1.2) compared against athletes without a previous history (controls).
The risk of subsequent hamstring injury for athletes with and without a previous history of chronic groin pain was also evaluated (three studies, figure 2). High heterogeneity (I2=62%) and no significant change in associated risk of subsequent hamstring injury were detected (RR=1.14, 95% CI 0.29 to 4.51, p=0.850).
Qualitative synthesis
Subsequent injury risk was summarised in forest plots grouped in accordance to the statistical analysis used in the study that provided the data. Studies that controlled for potential confounders were plotted together (figure 3), while the univariate results were presented together in the same forest plots (figures 4 and 5). It is important to note that the point estimates are calculated using different statistical methods across the studies, which is highlighted in the forest plots through the use of different shaped point estimates.
Risk of subsequent injury following a specific index injury calculated through multivariate or multiple regression models. Note: different statistical analyses of subsequent injury risk are used between studies, which are represented with different shaped point estimates: circle, HR; rectangle, RR; diamond, OR. a≤8 weeks following hamstring injury; b>8 weeks following quadriceps injury; c>8 weeks following calf injury. RR, relative risk.
Risk of subsequent injury following a specific index injury calculated through univariate models or simple regression models. Note: different statistical analyses of subsequent injury risk are used between studies, which are represented with different shaped point estimates: circle, HR; rectangle, RR; diamond, OR. aRisk calculated from raw injury data available; bRisk calculated from raw injury data sought from study authors. RR, relative risk.
Risk of subsequent injury following a specific index injury calculated through univariate analysis or simple regression models. Note: different statistical analyses of subsequent injury risk are used between studies, which are represented with different shaped point estimates: rectangle, RR; triangle, injury rate ratio.aRisk calculated from raw injury data available, # x-axis truncated at 0.1. RR, relative risk; IRR, injury rate ratio.
Significant associations and trends indicating an increased susceptibility for subsequent injury were found to be common for most of the index injury types, including concussion, hamstring, quadriceps, adductor, calf, back, knee and ankle injuries. The exception to these findings were the associations identified for the odds of an ankle sprain following a knee injury (figure 4) and subsequent groin strain injuries and knee cartilage injuries following a lumbar stress fracture (figure 5), which were not found to be associated with an increased risk.
Discussion
This review provides evidence of an association between subsequent injury risk following an index injury of a different type consistent with the premises of the subsequent injury categorisation (SIC) model.13 A recent paper examining injuries in elite Australian football players demonstrated that subsequent injuries are commonly related to the occurrence of previous injuries of different types, and reporting of only recurrent injuries would result in lower injury rates.54 Therefore, any injury is critical in an athlete’s injury profile when assessing their risk for subsequent injury. Previous injury history has also been suggested to modify the complex interaction between other injury factors, which is an important consideration in identifying causal factors.55 The identification of the relationship between index and subsequent injuries provided in this review forms a foundation for tertiary prevention programmes to be developed when an understanding of the injury aetiology and mechanisms have been further established.11 12
History of ACL injury was associated with an increased risk of subsequent hamstring injury
Athletes who had a history of an ACL injury at any stage throughout their career were found to be at more than twice the risk for sustaining a subsequent hamstring injury compared with those without a history of ACL injury (RR=2.25, 95% CI 1.34 to 3.76, I2=0%, figure 2). An additional study,30 not included in the meta-analysis due to hamstring re-injury risk only being evaluated, found that a second hamstring injury was associated with an almost fourfold increase (RR=3.89, 95% CI 1.55 to 9.78) of occurrence if an athlete also had a history of ACL injury. Two of the studies defined a previous ACL injury as an ACL reconstruction,30 32 while the other two studies did not specify the severity or management of the injury required to be a recordable injury event.47 50 The type of ACL graft used in the ACL reconstruction (hamstring or patellar autograph) was specified in two of the studies30 32 but was not significantly associated with subsequent hamstring injury rates. This finding may indicate that the graft type used during ACL reconstruction is not associated with subsequent hamstring injury, which is contrary to previous hypotheses. Hypotheses for the mechanisms that contribute to this associated increase in subsequent hamstring injury include altered biomechanics following ACL injury,30 32 the type of graft used,56 the harvesting procedure undertaken in the use of hamstring autographs57 and residual hamstring weakness following hamstring graft.56
Tertiary prevention for hamstring injury should be considered in ACL rehabilitation programmes and ongoing preventative programmes throughout an athlete’s career to mitigate the associated increased risk of hamstring injury.11 Investigation into the mechanisms that contribute to this increase in associated risk is warranted to strengthen the likelihood of success of tertiary prevention programmes.
History of chronic groin pain was not associated with subsequent hamstring injury
A history of chronic groin pain in athletes was not found to be significantly associated with sustaining a subsequent hamstring injury (RR=1.14, 95% CI 0.29 to 4.51, I2=62%, figure 2). The definition of chronic groin injury was not specified in the two studies with non-significant relationships with subsequent hamstring strain.47 50 These two studies only considered a previous history of injury within the 12 months prior to the study’s commencement. The third study in the meta-analysis required bone marrow oedema on MRI at the time of injury, and injury history was not limited the preceding 12 months. This review suggests that chronic groin pain is not associated with future hamstring strains; however, new research that categorises groin pain in accordance with the recent Doha agreement on terminology and definitions for groin pain in athletes may change this conclusion.58
A previous lower limb muscle injury was associated with an increased risk for a subsequent muscle injury a different site
Index injuries to the hamstrings, quadriceps, adductors and calves were associated with a subsequent muscle injury at a different site. Two of the four studies controlled for additional variables,29 53 including age and previous injury that can both act as confounders for future injury risk. Index hamstring injuries were associated with a significant increase in risk of future calf29 53 and quadriceps injuries.29 Index quadriceps injuries were significantly associated with an increased risk of calf injuries,29 but associations with subsequent hamstring injury was found in only one of the three studies that investigated this association.53 Adductor muscle injuries were associated with increased subsequent quadriceps and calf injury regardless of whether simple or multiple analysis was performed.53 Previous calf injury was found to be associated with subsequent quadriceps injury risk,53 and one study found an association for subsequent hamstring injury following a calf injury through multiple regression analysis,29 despite two other studies finding no univariate association.47 50
The associations with future lower limb muscular injury could be linked to a change in running biomechanics29 30 and inadequate compensatory movement and motor control strategies following injury.53 Adequate rehabilitation, including training load monitoring, after injury determines if an athlete is reconditioned to the specific demands of the sport to mitigate the risk of injury when returning to play.9 Athletes commonly become deconditioned following an injury that causes a reduction in their chronic workload level. Returning to high workload levels without appropriate training load prescription, the athlete will be susceptible to injury recurrence and exposed to a higher risks for other injuries at a different anatomical site and/or of a different nature.9 59 60
Initial back injury was associated with increased risk for lower limb subsequent injury
A back injury was associated with subsequent hamstring injury.30 51 A previous history of lumbar stress fracture in cricket fast bowlers was associated with a 1.5 times increase in hamstring injury risk,51 while any severe back injury sustained by an elite or subelite Australian football player was associated with subsequent hamstring injury by more than a twofold.30 Cricket fast bowlers with a previous history of lumbar stress fracture were also found to have a fourfold increased risk of sustaining a calf injury and a twofold risk of sustaining a subsequent quadriceps injury.51 The same study did not find any significant association between lumbar stress fractures and groin or ankle sprains and joint injuries.
Lumbar spine pathology has been hypothesised to potentially contribute to hamstring muscle weakness, particularly with increasing athlete age as greater lumbar spine pathology is commonly observed. This reduced level of muscular strength has been proposed to increase the risk of muscular strain injuries.51 61
History of lower limb joint injury was associated with other types of subsequent injury
An index injury at the knee joint was found to be associated with more than a 3.5 times increased risk of sustaining a leg or foot injury in elite female soccer players.49 Associations between sustaining a new knee injury (non-ACL) in elite male soccer players and a previous history of an ACL injury was found to be twice as high in one study.31 Possible explanations for these associated subsequent injuries may include access to medical treatment and rehabilitation,49 inadequate levels of rehabilitation,31 premature return to sport31 and altered kinematics and diminished proprioception following knee injury.31
Concussion was associated with an increase in odds and rate of lower limb injury
Concussion was found to be significantly associated with sustaining a subsequent lower limb musculoskeletal injury.48 52 A 2.5-fold increase in the odds of sustaining a lower limb musculosketal injury was found for Division I NCAA athletes in the initial 90 days following return to play when compared with non-concussed athletes. The increased association was maintained at 12 months postconcussion, where concussed athletes were 64% more likely to experience a lower limb musculoskeletal injury.48
The mechanisms for subsequent injury following concussion remain unknown. Hypotheses for increased subsequent injury following concussion include deficits in gait, dynamic balance and neuromuscular control,48 52 62 63 reduced reaction times and cognitive processing speeds,48 52 disrupted cortical pathways,48 63 and playing positions or aggressive styles of play that expose players to an increase risk of injury.64 65 Altered athlete workloads that occur following a graduated period of physical and cognitive rest following concussion, as currently recommended by the most recent consensus statement guidelines for the management of concussion in sport,66 could also be associated with an increased risk of subsequent injury.
Limitations
Despite the extensive search strategy conducted in this review, the inclusion of studies only written in English was a limitation. The meta-analysis included only a small number of studies. This meant that definitive conclusions could not be drawn through funnel plot analysis. Therefore, it was not possible to either confirm or eliminate the possibility of publication bias. Heterogeneity of study methodology within the included studies (varying nature of index injury, subsequent injury and statistical method used) must also be considered when interpreting the overall findings of this review.
The majority of the included studies performed well in the risk of bias assessment; however, the results from a number of studies should be considered with some caution as there was no adjustment for the confounders of age and previous injury history. It is also important to note that there may be other biases inherent in the results of the included studies that were not identified by the NOS. Unmeasured variables the such as an individual’s aggressive style of play or playing position, nutrition, match congestion and psychological aspects could also potentially impact on the risk of injury and may modify the association between previous and subsequent injury.
Recommendations for future research
Previous research on recurrent injury has often not accounted for an athlete’s full injury history. Consequently, the impact that an injury of another type has in the contribution to an athlete’s overall injury risk profile is likely to have been underestimated.13 14 Future studies should consider all types of sports injuries and make appropriate adjustments in their statistical analysis to control for potential confounders such as age, previous injury history and player position. Categorisation of subsequent injury data, using frameworks such as the SIC model13 14 or the multistate framework for the analysis of subsequent injury in sport67 should also be performed in future sport injury studies to aid subsequent injury analysis. Presentation of absolute risk estimates should be included in all future work. Absolute risk estimates provide clinicians, coaches and athletes with a more clinically applicable and patient-specific risk measure when compared with RR measures. Utilisation of absolute risk estimates may improve the mutual understanding and better inform all personnel involved in the shared decision-making process when considering an athlete’s risk of injury.68 Moreover, the statistical challenges in addressing multiple and subsequent injuries needs to be adequately addressed, as recommended elsewhere.19 69 70
This review did not investigate the aetiology or mechanisms for subsequent injury. Understanding these factors provides a strong foundation for the development and implementation of successful tertiary prevention programmes.12 Future research is warranted to establish these mechanisms that will guide the development of specific prevention programmes.
Conclusion
Previous injury history is associated with an increased susceptibility to different subsequent lower limb injuries. A previous history of ACL injury was associated with subsequent hamstring injury by more than twofold. A range of other lower limb injuries were found to be associated with future lower limb injuries at a different site and/or of a different nature. Concussion and back injuries were also associated with an increase in subsequent lower limb musculoskeletal injury.
What is already known?
A previous history of a lower limb injury is a strong risk factor for recurrence of that same injury in athletic populations.
Previous injury has been suggested to modify the complex interaction between other injury determinants, which is an important consideration in identifying causal factors and in the development tertiary prevention programmes.
The association between previous injury and a subsequent injury that differs in nature or location is rarely considered.
What are the findings?
A history of ACL injury was found to be associated with more than a twofold increase in subsequent hamstring injury risk compared with athletes without a history of ACL injury, while chronic groin pain was not associated with subsequent hamstring injury.
A history of lower limb muscular or joint injuries was associated with a variety of lower limb subsequent injuries that are of a different type.
Concussion was found to be associated with increased odds and rate of subsequent lower limb musculoskeletal injury in athletic populations.
How might it impact on clinical practice in the future?
Future injury can be associated with previous injury of different types, and this should be accounted for in the design and implementation of injury prevention programmes.
Tertiary injury intervention programmes for hamstring injuries should be implemented following an ACL injury within athletic populations.
Athletes who sustain a lower limb injury should routinely be identified as being at an increased risk of future injury and have their training programmes adapted accordingly to mitigate the risk of subsequent injury.
References
Footnotes
Contributors LAT, CFF, JLC, MKD and JEG contributed to the original concept. LAT and JEG designed the search strategy that LAT executed. LAT and MKD undertook the process of inclusion/exclusion and independently assessed methodological criteria. LAT extracted all data from the included studies and MKD confirmed the accuracy of this. All authors contributed to the drafting and final approval of the manuscript. This work was undertaken by LAT as a component of his PhD under the supervision of authors JLC, CFF, MKD and JEG.
Funding This work was supported by an Australian Government Research Training Program Scholarship awarded to author LAT for support during his PhD. JLC was supported by an NHMRC Practitioner Fellowship (ID058493). The Australian Centre for Research into Injury in Sport and its Prevention (ACRISP) is one of the International Research Centres for Prevention of Injury and Protection of Athlete Health supported by the International Olympic Committee (IOC).
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.