Article Text
PDF
Abstract
Background Little is known about the value of clinical examination in relation to diagnostic imaging for acute groin injuries in athletes. Primary aim: to investigate whether clinical examination tests predict a positive or negative MRI result (MRI±). Secondary aim: to assess accuracy of clinical tests to localise injury in MRI+ cases.
Methods We consecutively included 81 male athletes with acute groin injuries. Standardised clinical examination (palpation, resistance and stretch tests) and MRI were performed within 7 days of injury. Diagnostic statistics including positive and negative predictive values (PPV/NPV) were calculated.
Results 85 acute injuries were found on MRI in 64 (79%) athletes with 17 (21%) athletes having MRI− injuries. Palpation had the highest NPV (91–96%, (95% CI 69% to 99%)). 3 specific adductor examination tests (resisted outer range adduction, squeeze test with hip neutral and long lever, and passive adductor stretch) showed 80–81% (95% CI 63% to 91%) probability of an MRI+ adductor lesion when positive, all with high accuracy of a correct MRI location (PPV 93–97% (95% CI 76% to 100%)). Hip flexor tests showed poor ability to predict MRI+ lesions (PPV 34–63% (95% CI 20% to 84%)) and low accuracy (PPV 17–71% (95% CI 7% to 85%)).
Conclusions 21% of athletes had negative imaging and the absence of palpation pain was best at predicting an MRI− result. Specific adductor examination tests accurately predicted MRI+ adductor injuries. Hip flexor clinical tests were poor at predicating and localising MRI+ injuries in the hip flexors. Clinical examination appears sufficient to diagnose acute adductor injuries, whereas MRI could assist in accurately locating acute hip flexor injuries.
- Groin
- Diagnosis
- Radiology
- Examination
- Sport
Statistics from Altmetric.com
Introduction
Acute groin injuries are common in many sports, but reporting of groin injuries in epidemiological studies often lack description of pain onset (acute/gradual),1–3 which limits information on diagnostic variations, injury incidence and treatment related to this injury type. Diagnosing groin pain in athletes can be challenging due to the complex regional anatomy,4 multiple differential diagnoses5 and heterogeneous terminology.6 Diagnostic terminology in acute groin injuries, however, appears less diverse than for long-standing groin pain, and is usually related to specific muscles. Although this could simplify diagnosis of acute groin injuries, discrepancies between clinical diagnoses and radiological findings have been found.7 The reasons for this discrepancy are still unknown, but could be related to the pain presentation of the athlete during the clinical examination. More data on the association between clinical examination and imaging findings in acute groin injuries are needed to understand the diagnostic value of commonly performed clinical examination tests and imaging.
Specific clinical examination tests for athletes with long-standing groin pain have good intra-tester and inter-tester reproducibility.8 Paired combinations of clinical examination tests are used to categorise groin pain into several defined clinical entities,9 ,10 and there is international agreement on diagnostic taxonomy for athletes with long-standing groin pain.5 This taxonomy was also suggested to be usable for acute groin injuries, but data to inform decision-making on the best clinical examination tests are lacking.5
MRI is currently considered the best available investigation to confirm the location of acute muscle injuries, due to its supreme soft tissue contrast and high resolution,11 although considerable prevalence of imaging negative results in acute muscle pain has been reported.7 ,12 ,13
Since MRI is not readily available for many clinicians, further investigation of the association between clinical examination tests and the presence of MRI lesions and their location is required.
The primary aim of this study was to investigate the ability of specific clinical examination tests to predict a positive or negative MRI injury in the tested muscles group in athletes with acute groin injuries. The secondary aim was to assess the accuracy of the clinical examination tests to predict the location of the injury in MRI-positive cases.
Methods
Participants
Athletes with an acute groin injury were consecutively included prospectively over two sports seasons (August 2013 to May 2015) from an outpatient department at a specialised sports medicine hospital in Qatar. Inclusion criteria were: male athletes, 18–40 years old, participating in competitive individual or team sports, attending the hospital clinic within 7 days of acute-onset groin pain sustained during sport, and with MRI performed within 7 days of injury. Exclusion criteria were: gradual onset or exacerbation of ongoing groin pain, clinical signs or symptoms of prostatitis or urinary tract infection, or other known coexisting chronic diseases, such as significant hip osteoarthritis. Ethical approval was obtained by the Shafallah Medical Genetics Center and the Anti-Doping Lab Qatar Institutional Review Boards (IRB project number 2012-013 and EXT2014000004). Written informed consent was obtained from all athletes at inclusion.
Clinical examination
All athletes were initially assessed by a sports medicine physician, with a research physiotherapist present, to determine eligibility. On athlete consent, a more extensive standardised clinical examination was performed by the physiotherapist, who was blinded to the MRI results. Three types of clinical pain provocation tests were used: palpation, muscle resistance and stretch tests.8 ,9 The clinical examination tests were grouped into three categories: (1) hip adductor, (2) hip flexor and (3) abdominal tests (table 1). Full test descriptions are included in online supplementary appendix 1. All clinical examination tests included bilateral comparison, and were deemed positive if the athlete recognised the specific acute injury pain for the tested area. If the player reported pain or discomfort during a test unrelated to the acute injury pain, the test was considered negative. The modified Thomas test is normally used as a hip flexor test, but in this study it was also considered as an abdominal test and hence considered positive if an athlete reported abdominal pain during the test either in addition to or isolated from any other groin pain.
Categorisation of the specific clinical examination tests (full description of tests can be found in online supplementary appendix 1)
supplementary appendix
MRI assessment
MRI was performed on a 1.5 T system (Magnetom Espree, Siemens) using a body matrix coil and eight standardised sequences as described in detail in a previous publication.7 In brief, the protocol included two coronal T1-weighted and short T1 inversion recovery (STIR), one sagittal fat-suppressed proton density-weighted, three axial T1-, fat-suppressed proton density-, and fat-suppressed T2-weighted, and two axial oblique fat-suppressed proton density- and T2-weighted sequences. The MRI scoring was performed by an experienced musculoskeletal radiologist not involved in the clinical practice and blinded to any clinical information. Each potential muscle location was scored individually as MRI-positive (findings of oedema with or without structural disruption, as well as complete tears or avulsions) or MRI-negative (no acute lesion). This dichotomisation has shown almost perfect intra-rater and inter-rater reproducibility both on an athlete level (k=1.00 and 0.89, respectively (95% CI 0.76 to 1.00)) and for the number of acute lesions per athlete (k=0.94 and 0.91, respectively (95% CI 0.77 to 1.00)).14 All MRI lesions were categorised into (1) ‘adductor’ (adductor longus, adductor brevis, adductor magnus, pectineus, gracilis and obturator externus), (2) ‘hip flexor’ (rectus femoris, iliacus, psoas (or iliopsoas if both or no differentiation could be made) and sartorius), (3) ‘abdominal’ (rectus abdominis, transverse abdominis, internal and external obliques, and conjoint tendon) or (4) ‘other’.
Statistical methods
Predicting a positive or negative MRI
We used the individual clinical examination tests relevant for the specific areas as described in table 1. The 2×2 tables were created using a clinical examination test result (positive/negative) compared with the MRI result (positive/negative) in the location relevant for the test, for example, a ‘true positive’ could be a positive adductor stretch test and a positive MRI lesion in the adductors. A ‘false positive’ could be a positive adductor stretch test and a negative MRI for an adductor lesion. We then calculated prevalence of the positive examination tests, sensitivity (Sen), specificity (Spe), positive and negative likelihood ratios (LR+ and LR−), and predictive values (positive predictive value (PPV) and negative predictive value (NPV) in per cent) for all tests including continuity corrected CIs. We used a receiver operating characteristic (ROC) curve including area under the curve (AUC) to determine an optimal cut-off point for the number of tests discriminating between a positive and negative MRI in the tested area. The AUC is also reported for all tests individually.
Predicting accurate injury location in MRI-positive cases
Since there is no definite way of determining which structure is truly injured in MRI-negative cases, we therefore excluded all MRI-negative cases for use in the further analysis. For this part of the analysis, we thereby consider MRI as the reference standard. This will in turn provide a clearer overview of the accuracy of the individual tests, which can then be used to guide diagnosis of players with a negative MRI or players without MRI performed. Different 2×2 tables were then created using the specific test results, now in comparison to a positive/negative MRI in a different location, for example, a ‘true positive’ being a positive adductor stretch test and an MRI lesion in the adductors. A ‘false positive’ could be a positive adductor stretch test, but an MRI lesion in the iliopsoas and not the adductors. Sen, Spe, LR+, LR−, PPV and NPV, as well as AUC, were calculated again for each test. Values were also calculated for cases where all the examination tests were positive or all were negative for both analyses. Statistical analysis was performed using SPSS software (V.21; IBM Corp) and an online statistical calculator (vassarstats.net/clin1.html).
Results
During the study period, 100 athletes presented with an acute groin injury eligible for inclusion. Of these, 13 athletes did not wish to participate in the study, and 6 athletes initially consented, but MRI was performed more than 7 days after injury. A total of 81 athletes could therefore be included in the study analysis (mean age: 25.8 years (SD 4.4, range 18–37), mean height: 179.7 cm (SD 9.1, range 162–210), mean weight: 77.5 kg (SD 14.1, range 47–125)). All athletes were registered with national sports federations and playing at the highest or second highest national level. The athletes were mainly football players (47 soccer (57%) and 16 futsal (20%)). The other sports were basketball, handball, volleyball, table tennis and shot putt, from which seven, five, four, one and one athlete were included, respectively.
In 17 athletes (21%), no acute lesion was found on MRI. In the remaining 64 patients with MRI-positive findings, 85 different acute lesions were found (table 2). The adductor longus was the most frequent MRI lesion location (52% of all athletes). Only one MRI lesion (1.2%) was found in the abdominals. There were four MRI-positive lesions in muscles not initially considered in the groin area. Three vastus medialis injuries were reported in conjunction with an adductor injury, as well as one isolated tensor fascia latae injury. There were two cases where an acute MRI lesion was reported in more than one classification grouping: one combined adductor and abdominal (adductor longus and rectus abdominis), and one combined adductor and hip flexor (adductor longus, gracillis and sartorius). No adverse effects were reported from performing any of the included tests.
Overview of all MRI lesions in the 81 athletes subdivided in the four main categories
Predicting a positive or negative MRI
Results for the predictive ability of an MRI lesion in the specific muscles examined with the individual examination tests presented in table 3 for the adductors and table 4 for the hip flexors.
Overview of clinical examination tests for the hip adductor muscles in relation to a positive or negative MRI finding in the adductors
Overview of clinical examination tests for the hip flexor muscles
The adductor tests with the highest PPV of a positive MRI in the adductors were resisted adduction in outer range, squeeze-0° and passive adductor stretch (PPV 80–81% (95% CI 63% to 91%), table 3A). Positive hip flexor tests had poor ability to predict a positive MRI in the hip flexors (PPV 34–63% (95% CI 20% to 84%), table 4A). Since there was only one MRI lesion in the abdominals, analyses of the individual clinical examination tests for the abdominals were not performed. Instead, a simple overview of the frequency of positive findings in the tests relevant for the abdominals can be found in online supplementary appendix 2, where it is noticeable that clinical examination tests were often positive without MRI findings. Owing to severe pain, two athletes could not be positioned in the Thomas tests position, and an additional four athletes could not perform the second part of this test involving knee movement, and were therefore excluded from the specific test analyses (four of these six athletes were diagnosed with an adductor longus avulsion). The ROC curve analysis showed that the cut-off number of positive tests for the adductors and hip flexors was 3 (both p<0.001).
supplementary appendix
Predicting injury location in MRI-positive cases
Results of the accuracy of the clinical examination tests are shown in tables 3B and 4B,C for the adductors, iliopsoas and rectus femoris, respectively. In general, when adductor tests were positive, there was a high probability of finding an adductor injury on MRI (PPV 89–100% (95% CI 60% to 100%), table 3B). Specific palpation of the gracilis and pectineus was, however, less accurate with PPVs of 33% (95% CI 2% to 87%) and 38% (95% CI 10% to 74%), respectively, both with a relatively low prevalence resulting in large CIs. Positive tests for iliopsoas and rectus femoris injuries showed poor predictive ability in accurately locating injuries (PPV 17–71% (95% CI 7% to 85%), table 4B, C).
Discussion
Clinicians can be confident in locating or ruling out adductor injuries using simple clinical tests. In contrast, the low PPVs of accurately locating hip flexor injuries suggest that MRI could improve the accuracy in the diagnosis of these injuries. The discussion focuses on the PPV and NPV, as these are more easily interpreted by clinicians and patients. Owing to the lower number of MRI hip flexor lesions, and resulting statistical uncertainty, these specific results should be interpreted as a foundation for further research.
Predicting a negative MRI
Around one in five MRIs were negative. If all clinical tests were negative, we found very low probability of finding an acute lesion on MRI: 0% (95% CI 0% to 24%) for adductors and 7% (95% CI 1% to 24%) for hip flexors. There were only two cases where all hip flexor tests were negative and a positive hip flexor lesion was reported on MRI, both in the sartorius. Therefore, it is highly unlikely to find an acute lesion on MRI, when all standardised clinical tests used in the current study are negative.
Predicting a negative MRI might be harder when some tests are positive and others negative. The results of individual tests should therefore be considered. Of these, palpation showed the highest probability of a negative MRI with an NPV of 91% (95% CI 69% to 98%) for adductor palpation and 96% (95% CI 85% to 99%) for hip flexor palpation. There were only two cases with false-negative adductor palpation tests. For clinical purposes, no pain on palpation can be considered a key test to predict a negative MRI.
Predicting a positive MRI
When all grouped clinical examination tests were positive, we found the highest probability of a positive MRI, with a PPV of 83% (95% CI 58% to 96%) for all adductor tests, and 86% (95% CI 42% to 99%) for all hip flexor tests positive. All tests were positive in only 31% of cases, so while this combination has fairly good predictive ability, athletes often present with fewer positive tests.
We examined whether a minimum number of positive tests exists to predict a positive MRI. We found three positive tests to be the optimum cut-off, yet the predictive ability of this three test cut-off for adductor and hip flexor tests was worse than the majority of the individual tests (PPV 76% (95% CI 62% to 87%), and 54% (95% CI 34% to 73%)), respectively. Clinicians are therefore better off relying on their interpretation of individual tests rather than a specific cut-off number of positive tests.
The best individual adductor examination tests to predict a positive MRI in the adductors were resisted outer range adduction, squeeze test with neutral hip and long lever arm (squeeze-0°),9 ,10 and passive adductor stretch test. The predictive values of these tests were similar to having all tests positive, and as these tests are more often positive (43–59%) than all adductor tests are positive (22%), these individual tests are likely to be of most benefit for clinicians attempting to predict a positive MRI in the adductors.
The best individual hip flexor test to discriminate between a positive and negative MRI in the hip flexors was palpation. However, the PPV of an MRI hip flexor lesion was similar to chance. All individual hip flexor tests had limited capacity to predict a positive MRI, indicating that hip flexor tests were often positive without positive MRI findings in the hip flexors.
Predicting injury location in MRI-positive cases
The highest accuracy in predicting the MRI lesion location was when all tests for either the adductors or hip flexors were positive. When all adductor tests were positive, the PPV of an MRI injury in the adductors was 100%. This suggests that when clinicians are presented with an athlete with all adductor tests positive, there is no need for an MRI if the sole purpose is to confirm the injury location. For the iliopsoas and rectus femoris test, this was considerably lower with a PPV of only 71% (95% CI (0.30% to 0.95%) and 43% (95% CI 0.12% to 0.80%)) for an accurate location, respectively. Clinicians can therefore not be certain of an accurate location even when all tests are positive and can consider imaging if a higher level of confidence regarding injury location is required.
The very high PPVs of most of the individual adductor tests indicate that MRI is not needed to confirm the injury location, and that the diagnosis of an acute adductor injury can be comfortably made using clinical examination alone. In our study, squeeze-0° had the highest PPV with a 97% (95% CI 85% to 100%) probability of an accurate injury location. This squeeze-0° test has also been shown to be a superior test to other angles (45° and 90°) for pain provocation induced in the adductor region,15 and seems to be the best resistance test to include in the assessment of acute and long-standing adductor-related groin pain. The clinical entity adductor-related groin pain has been defined as ‘adductor tenderness AND pain on resisted adduction testing’,5 which is in line with our results, where adductor palpation also had a high probability of an accurate location, PPV 92% (95% CI 79% to 97%). For palpation of the adductors, we specified three different adductor muscles, as these were thought to have the highest palpation reproducibility. We found that palpation pain of the pectineus and gracilis was not very prevalent, which is most likely reflected in the poor PPVs compared with the adductor longus. Clinicians should therefore be cautious in specifying clinical diagnosis to these structures.
Palpation of the iliopsoas and rectus femoris were the individual tests with the highest probability of an accurate injury location as reported on MRI. Both had a 100% probability of a negative MRI when the test was negative; however, the probability of an accurate location was not much different than chance when the test was positive. Accurately predicting the injury location in the initial clinical examination of acute hip flexor injuries therefore appears challenging. The poor PPVs of these tests could explain the previously reported discrepancies between the clinical diagnoses and imaging findings for acute iliopsoas and rectus femoris injuries.7 This could also have an influence in relation to the general taxonomy of groin pain, where the clinical entity ‘iliopsoas-related groin pain’ is defined as iliopsoas tenderness, and is deemed more likely if there is pain on resisted hip flexion and/or pain on stretching.5 Our study shows that using these clinical criteria is likely to result in inaccurate classification of acute injuries in this area. Further research with a larger number of hip flexor injuries is required to provide more specific recommendations in this regard.
Limitations
The relatively low prevalence of some injuries, especially hip flexor injuries, is reflected in the fairly large CIs, which should be considered in the interpretation of our results. Therefore, we cannot draw firm conclusions from the study for hip flexor or abdominal injuries. However, on the basis of our results, clinicians dealing with injured athletes on an individual basis can be comfortable in diagnosing the location of an adductor injury based on clinical examination alone, whereas MRI might be needed to diagnose injury location in an athlete with an acute hip flexor injury. We also found that the MRI result was often negative, reflecting limitations in using MRI as a diagnostic gold standard for injury location in athletes with acute groin injuries. Further research is required to elucidate whether current MRI methods are not sensitive enough to detect minor injuries, or whether there could be other explanations for the acute pain that potentially cannot be visualised at all. In the future, more advanced MRI such as compositional and other techniques including T2 mapping, diffusion-weighted imaging or spectroscopy might help in elucidating these cases further. A previous study using MRI in the assessment of long-standing groin pain16 showed that in asymptomatic soccer and non-soccer players, adductor longus musculotendinous lesions were reported in 2 out of 34 (6%) and 40 (5%) cases, respectively. A clear distinction between whether these reported lesions could potentially be related to an older injury was, however, not included in that study. All positive MRI lesions in our study were divided into acute and non-acute, with only the acute lesions included in the analysis. We found high intra-rater and inter-rater reproducibility for this differentiation; intra-rater k=1.00 (95% CI 1.00 to 1.00) absolute agreement 100%, and inter-rater k=0.86 (95% CI 0.71 to 1.00) absolute agreement 96.7% (Serner et al14). We believe this limits the potential for false-positive MRI findings, which could have a minor influence on the values provided in this study. Additionally, it is beyond the scope of this study to conclude on the clinical implications in relation to potential treatment and prognostic variations in response to the diagnostic considerations, such as injury severity.
Conclusion
Specific adductor examination tests (resisted outer range adduction, adductor stretch and the squeeze test in hip neutral position) individually provided ∼80% probability of predicting a positive MRI in the adductors. These adductor examination tests also provided very high probability of predicting an accurate injury location. In contrast, individual hip flexor pain provocation tests had poor ability to predict a positive MRI, and poor accuracy. The absence of palpation pain was the best test result to predict a negative MRI in athletes with acute groin injuries.
What are the findings?
A negative MRI was found in one in five athletes (21%) with acute groin injury. The absence of palpation pain in the adductors and hip flexors had the highest predictive values for ruling out acute injury on MRI.
Positive individual adductor tests provide ∼80% probability of a positive MRI in the adductors, with the highest positive predictive values for resisted outer range adduction, squeeze test with hip neutral and long lever, and passive adductor stretch.
These specific adductor tests also show high accuracy (92–97% probability) of confirming that the injury is located in the adductors on MRI.
Individual hip flexor tests are poor at predicting a positive MRI and locating which hip flexor muscle is injured.
How might it impact on clinical practice in the future?
Clinicians can use clinical examination in acute adductor injuries to predict whether an MRI will show an acute injury in the adductors or not. Additionally, since the clinical adductor examination tests showed high accuracy in MRI-positive cases, the clinical examination appears sufficient to confirm the location of acute adductor injuries. In contrast, when clinicians assess acute hip flexor injuries, they can be less confident about predicting the injury location, as seen on MRI, using clinical tests. This knowledge may influence the decision whether or not imaging with MRI is required.
Acknowledgments
High appreciation goes to the multidisciplinary diagnostic and treatment approach provided in collaboration between the various departments within Aspetar Sports Medicine Hospital, as well with the club medical staff under the National Sports Medicine Programme. Additionally, the authors would like to thank Ivan Stankovic for his assistance with the anatomical drawings for the clinical examination test pictures, and Arnlaug Wangensteen and Arnhild Bakken for their assistance with patient inclusion.
References
Footnotes
Twitter Follow Andreas Serner at @aserner
Contributors AS, AW, JLT, KT and PH contributed to the conception and design of the study. All authors contributed to the analysis and interpretation of the data. AS drafted the article, and all authors revised it critically and approved the final article.
Funding The study was funded by Aspetar Orthopaedic and Sports Medicine Hospital.
Competing interests Outside this work, AG has received consultancies, speaking fees and/or honoraria from Sanofi-Aventis, Merck Serono and TissuGene, and is President and shareholder of Boston Imaging Core Lab (BICL), LLC, a company providing image assessment services. FR is Chief Medical Officer and shareholder of BICL, LLC.
Ethics approval Shafallah Medical Genetics Center and the Anti-Doping Laboratory Qatar Institutional Review Boards (IRB project no. 2012-013 and EXT2014000004).
Provenance and peer review Not commissioned; externally peer reviewed.