WHY ARE THERE SO MANY, AND HOW SHOULD WE BEST PUT THEM TO USE IN THE CLINIC?
– Written by Paul Blazey, Bruce B. Forster, Clare L. Ardern, Michael Fredericson, Adam S. Tenforde, and Karim M. Khan
INTRODUCTION
The burden of bone stress injuries (BSIs) is often higher than other sport-related injuries1-5. Worse than the burden, athletes who suffer high-risk BSIs may never return to sport at their previous level of performance6. In short, BSIs are a big deal (or should we say a ‘bad deal’) for athletes, the clinicians who support them, and the teams they play for.
In the general sports or orthopaedic setting (i.e., with the public or recreational athletes), it may be acceptable—due to the cost of care—to take a cautious approach to managing BSIs. A ‘cautious approach’ meaning to keep non-professional athletes out of sport and exercise for longer than the longest suggested healing time for their respective BSI to avoid the risk of complications. However, in professional sport this is almost always undesirable to the athlete (detrimental effects of their career) and unacceptable to the team management (drop in team performance). Therefore, understanding the earliest possible opportunity that an athlete can successfully return to their sport, and the ideal treatment to get them there is important. Enter BSI classification systems.
BSI classification systems aim to grade the severity of an injury, and offer us (clinicians) the enticing prospect of accurately predicting a patient’s prognosis—most importantly, how long it will take an athlete to return to sport. However, a systematic review found 27 different classification systems related to bone stress injury8. This leaves many potential classification systems to choose from, and no gold standard system to apply to all athletes and all types of BSIs. Clinicians are often left to choose without guidance on which classification systems offer the best ability to detect and characterise injury (i.e., what imaging modality is best suited), or whether or not the systems have been validated for their ability to predict return to sport times.
We (the authors) as experts in bone stress injuries feel well placed to help clinicians working in professional sport pick through the decisions needed to choose a best imaging modality and classification system for their athletes. To be clinically useful a classification system should guide clinicians in two key areas: 1) help to form a prognosis for the athlete based on the severity of their injury and allow us to estimate a return to play timeframe; and/or 2) help direct which treatment options are appropriate for the athlete.
In this article, we review the most-used BSI classification systems. We provide an overview of why no universal BSI classification system has been adopted across sports medicine. In summarizing the literature, we hope readers working in professional sport will be aware of the nuances of classification systems and can work with athletes to make research informed choices about treatment and return to sport following a BSI.
THE EVOLUTION OF GENERIC AND SITE-SPECIFIC BSI CLASSIFICATION SYSTEMS
The first magnetic resonance imaging (MRI)-grounded BSI classification systems emerged in the mid-1990s. The MRI-based scales followed earlier attempts to classify bone stress using scintigraphy9. The scintigraphy and subsequent scales all follow the principle that bone loading leads to a continuum of effects dependent on the level of load received, and the ability of the bone to tolerate load. This continuum begins with normal remodeling (bone strain), followed by fatigue and bone exhaustion and, finally, cortical failure and fracture. MRI offered a way to better detect where a bone was along this continuum including more precise estimates of clinically relevant changes such as periosteal oedema, marrow oedema and cortical fracture. These clinical entities characterised by the changes seen on MRI and listed above are what we broadly call bone stress injuries, which depending on their severity, can be further broken down into ‘stress reactions’ or ‘stress fractures’ (see Figure 1).
The 1995 Fredericson BSI classification system proposed and validated a 0-4 scale for grading posteromedial tibia BSIs (4 being the worst and indicating the presence of a fracture line)10. Arendt et al., (1997) built on this by using the same scale and MRI approach but linking grade to prognosis across multiple BSI sites11; Nattiv et al. (2013) then correlated MRI grade, trabecular- vs cortical-rich bone, and female athlete risk factors with return-to-sport time12; Kaeding & Miller (2013) offered a ‘pragmatic’ classification that combined severity grade with anatomic risk (high- vs low-risk BSI) and a choice of which imaging modality to use13.
Site-specific BSI classification systems have also been proposed14,15. Classification systems for the navicular and femoral neck (both ‘high-risk’ BSI sites) were designed to guide clinical decision making. High-risk sites may require treatments that go beyond rest and a gradual return. A summary of the BSI classification systems outlined in this article is shown in Table 1.
NO ONE SYSTEM TO RULE THEM ALL - THE NEED FOR MULTIPLE SYSTEMS
We’ve highlighted that many BSI classifications exist and only discussed six of those most commonly used, but there are many more. Some of the classification systems aim to address prognosis, and others aim to direct treatment, most do not guide both prognosis and treatment. An international consensus statement called for consistent terminology and a single rating system through which to classify BSIs17 but the experts involved concluded that this is “easier said than done”. Four reasons why no universal classification system that guide both prognosis and treatment have currently been proposed include:
- a lack of consistent terminology;
- recommendations that include multiple potential imaging modalities;
- whether or not we need site-specific classifications; and
- a lack of agreement over the ability of classification systems ability to support prognosis or treatment decision-making.
We will consider in more detail each of these four barriers to effective implementation of BSI classification in clinical practice.
(1) Terminology
When does bone strain become a stress reaction; when does a stress reaction become a stress fracture? The term ‘bone strain’ used here refers to normal adaptation to load. In asymptomatic athletes it may be difficult to separate bone strains from stress reactions, especially as there is some debate over whether or not periosteal changes alone (seen on MRI) should be considered as “stress reactions” or as a normal response to bone loading. Non-symptomatic athletes compete with signs that could have been classed as “stress reactions”18-22. In clinical practice these terms sometimes get confused because of the lack of clearly delineated criteria that can support a move between each stage along the continuum. It is therefore unsurprising that athletes can find the discussion of their injury and its severity difficult to follow, especially when discussing less severe BSIs with limited clinical significance.
Our figure highlights that those who should be treated as having a ‘bone stress’ vs those who may be considered to be undergoing a normal reaction to load or ‘bone strain’ is based on whether or not they have clinical symptoms. Our belief is that any confusion in terminology can be resolved with better education for clinicians on the differentiation between bone ‘strain’ and ‘stress’ (see Box 1). Kaeding et al (2013) included clinical symptoms (such as pain) in their classification system but for reasons you will see in the next paragraph, theirs is not our favoured classification to use with professional athletes.
(2) Multiple imaging modalities
MRI has been established as the most sensitive imaging modality for diagnosing bone stress injuries17. Attempts to correlate radiography, bone scintigraphy, MRI and CT have proven unreliable23. Some authors of classification systems that rely on MRI have advocated for radiographic imaging prior to seeking an MRI. However, the rationale to reduce the costs associated with an unnecessary work-up are unlikely to apply in professional sport11,12. MRI is the gold standard for most BSIs and should be the imaging modality of choice when an athlete presents with pain consistent with a BSI. This makes the classification system proposed by Kaeding, et al (2013) less effective for use in the professional sport setting as grading relies on less accurate imaging methods13. What holds the field back from a blanket statement that ‘all BSIs should be characterised using MRI’ is that for high-risk BSIs, such as those sustained at the navicular, characterisation of a fracture line may be an important consideration that has additional treatment implications. This moves us to site-specific classifications.
(3) Site-specific classifications
Site-specific classifications have been put forward for ‘high-risk’ BSI. These include the navicular bone, which has notoriety for carrying a greater burden among BSI injury sites6,24,25. The Saxena classification recommends the use of CT for BSI injuries affecting the navicular15. MRI is best for detecting abnormal signal changes in bone as indicated above, it has excellent sensitivity. But CT is better at characterising the presence of a fracture line in the navicular (in this case specificity). This is backed by CT having a 100% positive correlation with correctly identifying a fracture line vs MRIs 71% correct correlation for navicular BSI26.
Recent data have suggested both compression- and tension-side femoral neck BSI should be considered ‘high-risk’27. The classification proposed by Rohena-Quinquilla argued for different gradings between patients with a macroscopic fracture (i.e. those with or without a clear fracture line), and then separated those with a fracture line covering >50% of femoral neck into the highest risk category (grade 4) requiring a surgical referral. However, even those with a fracture line that passes through >50% of the femoral neck do just as well with non-surgical management27. Therefore, this classification may be unnecessary, and the extra specificity which could be offered by CT scanning is not required.
Other site-specific considerations have included a suggested difference between healing times in stress fractures that affected trabecular-rich vs cortical-rich bone. Nattiv et al. (2013) found longer healing times for trabecular BSIs such as those sustained at the femur, sacrum or pubic bones12. However, this finding was based on only nine such cases in these bones. Given the longer healing times associated with navicular or anterior tibial BSI (both ‘cortical bone’), a dichotomy in classifications between trabecular and cortical bone does not appear to be a clear differentiator in terms of prognosis. It would therefore appear to be inappropriate—without further evidence—to make a classification scale that adapts to cortical- or trabecular-rich bones.
(4) Prognostic or treatment-guiding capacity
In our introduction we highlighted that classification systems need to perform at least one of two functions, they must guide: i) prognosis and/or ii) treatment. A 2022 meta-analysis of pooled MRI grading criteria (majority used Fredericson or Arendt classifications, or modifications thereof), demonstrated that there is a good (not great) correlation between MRI grading and time to return to sport (r=0.55)28. In addition, the Kijowski validation study demonstrated that the Fredericson classification system could be collapsed into 3 categories for tibial BSIs: grade 1 or “mild” BSI (limited to periosteal reactions only); grades 2 and 3 or “moderate” BSI (periosteal and marrow oedema); and grade 4 “severe” BSI (evidence of a cortical fracture line, plus severe marrow oedema)16. These three categories offered the most accurate predictions for when an athlete would be able to return to sport. As a sidebar, adolescent athletes may require their own classification system, but this goes beyond the scope of our article29. The classification systems proposed by Arendt and Nattiv, or Fredericson appear to offer good prognostic value, but their precision is likely impacted by individual athlete factors.
None of the classification systems above were used to guide treatment. This may be because the treatment/management of low-risk BSI cases is fairly settled, involving: relative rest; an assessment and management of modifiable risk factors; and a gradual (individualised) return to sport. The only differing factor may be the time to return. Treatment guidance derived from a classification system may therefore be unnecessary in low-risk BSIs.
HOW WOULD WE RECOMMEND HEALTHCARE PRACTITIONERS, COACHES AND ATHLETES THINK ABOUT BSI CLASSIFICATION SYSTEMS IN PROFESSIONAL SPORT?
Based on our (the authors) experience of the professional sport environment and the available evidence, we cannot recommend a single classification system for BSIs. However, we do have some suggestions that may help to guide the athlete, coach and/or healthcare professional in charge of identifying and managing athletes who present with a potential bone stress injury.
1) For the athlete – Our clinical experience suggests that athletes with a BSI often present (report pain for the first time) after the onset of a cortical break (higher-grade BSI). Make sure to report “niggles” however minor they may appear. Although a niggle (such as repeated cramping) may be nothing to worry about, early detection can prevent significant time away from sport30. And remember most overuse related bone stress injuries are easily managed if detected early and still in grades 1-3.
2) For the coach – For sports that carry a high-risk of navicular BSI, clinical experience suggests that a regular monitoring and awareness program with athletes should be encouraged. A study of 207 Olympic track and field athletes backs up the assertion that athletes with navicular BSI most often present post stress fracture (very few stress reactions—grades 1-3—in their data). There was also a high-risk of significant time loss associated BSIs, with it taking on average 199 days to return to full training following a stress fracture at any site24. If you work with athletes who are more likely to experience BSI, create a ‘brave space’ for athletes to come forward with pain. Make it clear that in the long-term dealing with the injury swiftly and early is likely to lead to more training or playing time and therefore will likely have long-term performance benefits.
3) For the healthcare professional – In elite athletes, systems such as the one proposed by Kaeding and Miller (2013) are of little use because the classification settles for using low precision imaging methods. Lack of precision limits the ability to support an accurate estimate of the time to return to sport. Notably MRI grading systems by Fredericson, Arendt and Nattiv demonstrate sound correlations between severity and return to sport times, perhaps unsurprising given their shared heritage. Even though Fredericson’s original classification was validated for tibial BSI we suggest there is limited benefits of one system over the other, especially as the differences are marginal (such as whether to limit changes to periosteal edema to classify grade 1). All three classification systems are summarised in Table 2. We suggest that you choose one of these classifications to use with all BSIs, and that the grading scale is likely to provide reasonable support in estimating prognosis (time to return to sport).
WHAT ABOUT THAT PESKY HIGH-RISK BSI THAT DOESN'T FOLLOW THE EXPECTED PATHWAY?
An international consensus agreed that there are at least four ‘high-risk’ BSI locations17. These are the superior cortex 'tension-side’ of the femoral neck, the anterior cortex of the tibial diaphysis, the navicular and the base of the fifth metatarsal. We highlighted earlier that research suggests differentiating between tension- and compression-side femoral neck BSIs may be unnecessary with both carrying a risk of progressing to full fracture, we therefore suggest considering both types of femoral neck BSI as high-risk27.
High-risk BSIs lead to longer return to sport times. Alongside an MRI, those with high risk BSI may benefit from characterising the severity of a fracture line and its exact location. Therefore, individual classifications may be required for high-risk BSI sites25.
We suggest a two-step process:
- Clinicians follow the MRI standard for low-risk BSIs. Consider grading using the 0-4 scale.
- For high-risk BSIs of the anterior tibia, navicular and if needed the base of the fifth metatarsal, CT is added to the MRI-standard to assess for the presence, extent, and specific location of a cortical fracture.
The reason we do not suggest CT for femoral neck BSI is that MRI alone is often sufficient. This is especially true if MRI is taken using newer scanning techniques such as the Volumetric Interpolated Breath-hold Examination (VIBE) T1-weighted images that produce CT-like appearances31. Avoiding CT in younger patients, such as athletes, who are still of reproductive age reduces ionizing radiation.
By including step 2, you can ensure that the sensitivity of the MRI, and in equivocal cases the specificity of the CT can be taken into account when making treatment decisions at high-risk BSI sites. While the significance of the degree of cortical fracture is unknown (i.e., treatment differences between a fracture that propagates to 30%, or 50% or 75% of the depth of a bone), having a standardised process for imaging and classifying these injuries will better allow us to track the outcomes and look for associations in sport-related outcomes. The Saxena classification (Table 1) is an example of a CT-derived bone specific classification scale and can be combined with an MRI graded image to support both prognosis and treatment decision-making. A summary of our suggestions for each of the high-risk BSI sites is found in Table 3.
Any new classifications at high-risk sites would need to be validated for their ability to improve clinical outcomes. Given the difficulty of conducting such work in athletes, prospective registry of all athletes who are managed using surgical or non-surgical treatments and correlated against a uniform BSI grading on MRI +/- CT may support greater decision-making power when choosing between surgical or non-surgical approaches.
Future developments in BSI classification should not just account for severity, location and symptoms but should account for individual athlete risk profiles that modify prognosis or treatment decisions. Risk profiles may include the need to perform additional investigations (e.g., DXA scans to establish BMD scores), menstrual status and blood tests to assess hormone function, and screening tools that assess lifestyle factors such nutrition and eating behaviours. Readers are directed to the International Olympic Committee's (IOC) consensus statement on Relative Energy Deficiency in Sport (REDs) for a list of potential overlapping risk factors that may impact an athlete's bone health37.
CONCLUSIONS
This article outlines a perspective on best practice decision-making using what we currently know about BSI classification systems with athletes. Although much of what we recommend may also apply to the general public, there are often additional barriers outside of professional sport, such as diagnostic delays (related to imaging accessibility).
We suggest two main takeaway points for clinicians in professional sport (or anyone with the resources to follow these suggestions):
- That a unified classification system based upon those proposed by Fredericson/Arendt/Nattiv be used for all BSIs (low- and high-risk), to establish a severity score (between 0-4). This classification system can be used to provide a guideline on the time needed to return to sport, although factors such as the bone injured and individual athlete factors (e.g., low total bone mineral density) may impact the accuracy of time predictions.
- In addition to the first recommendation, BSIs that occur at high-risk sites (except the femoral neck) may benefit from a CT scan if imaging is non-diagnostic or for navicular BSI. The CT scan being used specifically to measure the extent of any fracture line.
Follow-up studies are needed to assess the significance of different degrees of cortical fracture, and whether the extent of a fracture line can be used to predict prognosis for high-risk BSI sites. We also expect classification systems to evolve as: (i) imaging techniques improve and (ii) stress fracture management advances.
Paul Blazey BPhysio(Hons), MSc1,4
Bruce B Forster MSc, MD, FRCPC2
Clare L. Ardern BPhysio(Hons), PhD3,4,5
Michael Fredericson MD6
Adam S Tenforde MD7
Karim M. Khan MD, PhD, MBA1,4,8
1. School of Kinesiology, Faculty of Education, University of British Columbia, Vancouver, British Columbia, Canada
2. Department of Radiology, University of British Columbia, Vancouver, British Columbia, Canada
3. Department of Physical Therapy, The University of British Columbia Faculty of Medicine, Vancouver, British Columbia, Canada
4. Centre for Aging SMART, The University of British Columbia, Vancouver, British Columbia, Canada
5. La Trobe Sport and Exercise Medicine Centre, La Trobe University, Melbourne, Victoria, Australia
6. Department of Orthopaedic Surgery, Division of Physical Medicine & Rehabilitation, Stanford University, Stanford, California, United States of America
7. Department of Physical Medicine and Rehabilitation, Harvard Medical School, Spaulding Rehabilitation Hospital, Charlestown, Massachusetts, United States of America
8. Department of Family Practice, The University of British Columbia, Vancouver, British Columbia, Canada
Contact: paul.blazey@ubc.ca
References
1. Crunkhorn M, Toohey L, Etxebarria N, et al. Injury epidemiology in elite triathletes: a 4-year prospective study. J Sci Med Sport. 2022;25(Suppl 1):S14.
2. Ekegren CL, Quested R, Brodrick A. Injuries in pre-professional ballet dancers: incidence, characteristics and consequences. J Sci Med Sport. 2014;17(3):271–5.
3. Maquirriain J, Ghisi JP. The incidence and distribution of stress fractures in elite tennis players. Br J Sports Med. 2006;40(5):454–9.
4. Wasserstein DN, Dahm F. The epidemiology of stress fractures in sport. In: Robertson GAJ, Maffulli N, editors. Fractures in sport. Cham: Springer International Publishing; 2021. p. 29–33.
5. Wayner RA, Brown Crowell CN, Bovbjerg V, et al. Epidemiology of bone-stress injuries and health care use in Pac-12 cross-country athletes. J Athl Train. 2024;59(6):641–8.
6. Burne SG, Mahoney CM, Forster BB, et al. Tarsal navicular stress injury: long-term outcome and clinicoradiological correlation using both computed tomography and magnetic resonance imaging. Am J Sports Med. 2005;33(12):1875–81.
7. Popp KL, Ackerman KE, Rudolph SE, et al. Changes in volumetric bone mineral density over 12 months after a tibial bone stress injury diagnosis: implications for return to sports and military duty. Am J Sports Med. 2021;49(1):226–35.
8. Miller T, Kaeding CC, Flanigan D. The classification systems of stress fractures: a systematic review. Phys Sportsmed. 2011;39(1):93–100.
9. Roub LW, Gumerman LW, Hanley EN Jr, et al. Bone stress: a radionuclide imaging perspective. Radiology. 1979;132(2):431–8.
10. Fredericson M, Bergman AG, Hoffman KL, et al. Tibial stress reaction in runners: correlation of clinical symptoms and scintigraphy with a new magnetic resonance imaging grading system. Am J Sports Med. 1995;23(4):472–81.
11. Arendt EA, Griffiths HJ. The use of MR imaging in the assessment and clinical management of stress reactions of bone in high-performance athletes. Clin Sports Med. 1997;16(2):291–306.
12. Nattiv A, Kennedy G, Barrack MT, et al. Correlation of MRI grading of bone stress injuries with clinical risk factors and return to play: a 5-year prospective study in collegiate track and field athletes. Am J Sports Med. 2013;41(8):1930–41.
13. Kaeding CC, Miller T. The comprehensive description of stress fractures: a new classification system. J Bone Joint Surg Am. 2013;95(13):1214–20.
14. Rohena-Quinquilla IR, Rohena-Quinquilla FJ, Scully WF, et al. Femoral neck stress injuries: analysis of 156 cases in a U.S. military population and proposal of a new MRI classification system. AJR Am J Roentgenol. 2018;210(3):601–7.
15. Saxena A, Fullem B, Hannaford D. Results of treatment of 22 navicular stress fractures and a new proposed radiographic classification system. J Foot Ankle Surg. 2000;39(2):96–103.
16. Kijowski R, Choi J, Shinki K, et al. Validation of MRI classification system for tibial stress injuries. AJR Am J Roentgenol. 2012;198(4):878–84.
17. Hoenig T, Hollander K, Popp KL, et al. International Delphi consensus on bone stress injuries in athletes. Br J Sports Med. 2025;59(2):78–90.
18. Kornaat PR, de Jonge MC, Maas M. Bone marrow edema-like signal in the athlete. Eur J Radiol. 2008;67(1):49–53.
19. Bergman AG, Fredericson M, Ho C, et al. Asymptomatic tibial stress reactions: MRI detection and clinical follow-up in distance runners. AJR Am J Roentgenol. 2004;183(3):635–8.
20. Kornaat PR, van de Velde SK. Bone marrow edema lesions in the professional runner. Am J Sports Med. 2014;42(5):1242–6.
21. Major NM, Helms CA. MR imaging of the knee: findings in asymptomatic collegiate basketball players. AJR Am J Roentgenol. 2002;179(3):641–4.
22. Paajanen H, Hermunen H, Ristolainen L, et al. Long-standing groin pain in contact sports: a prospective case-control and MRI study. BMJ Open Sport Exerc Med. 2019;5(1):e000507.
23. Beck BR, Bergman AG, Miner M, et al. Tibial stress injury: relationship of radiographic, nuclear medicine bone scanning, MR imaging, and CT severity grades to clinical severity and time to healing. Radiology. 2012;263(3):811–8.
24. Kelly S, Waring A, Stone B, et al. Epidemiology of bone injuries in elite athletics: a prospective 9-year cohort study. Phys Ther Sport. 2024;66:67–75.
25. Hoenig T, Eissele J, Strahl A, et al. Return to sport following low-risk and high-risk bone stress injuries: a systematic review and meta-analysis. Br J Sports Med. 2023;57(7):427–32.
26. Saxena A, Behan SA, Valerio DL, et al. Navicular stress fracture outcomes in athletes: analysis of 62 injuries. J Foot Ankle Surg. 2017;56(5):943–8.
27. Steele CE, Cochran G, Renninger C, et al. Femoral neck stress fractures: MRI risk factors for progression. J Bone Joint Surg Am. 2018;100(17):1496–502.
28. Hoenig T, Tenforde AS, Strahl A, et al. Does magnetic resonance imaging grading correlate with return to sports after bone stress injuries? a systematic review and meta-analysis. Am J Sports Med. 2022;50(3):834–44.
29. Ditmars FS, Ruess L, Young CM, et al. MRI of tibial stress fractures: relationship between Fredericson classification and time to recovery in pediatric athletes. Pediatr Radiol. 2020;50(12):1735–41.
30. Whalan M, Lovell R, Sampson JA. Do niggles matter? increased injury risk following physical complaints in football (soccer). Sci Med Footb. 2020;4(3):216–24.
31. Koh E, Walton ER, Watson P. VIBE MRI: an alternative to CT in the imaging of sports-related osseous pathology? Br J Radiol. 2018;91(1088):20170894.
32. Hunt KJ, Anderson RB. Treatment of Jones fracture nonunions and refractures in the elite athlete: outcomes of intramedullary screw fixation with bone grafting. Am J Sports Med. 2011;39(9):1948–54.
33. Mologne TS, Lundeen JM, Clapper MF, et al. Early screw fixation versus casting in the treatment of acute Jones fractures. Am J Sports Med. 2005;33(7):970–5.
34. Goodloe JB, Cregar WM, Caughman A, et al. Surgical management of proximal fifth metatarsal fractures in elite athletes: a systematic review. Orthop J Sports Med. 2021;9(9):23259671211037647.
35. Herterich V, Baumbach SF, Kaiser A, et al. Fifth metatarsal fracture – a systematic review of the treatment of fractures of the base of the fifth metatarsal bone. Dtsch Arztebl Int. 2021;118(35–36):587–94.
36. Bušková K, Bartoníček J, Rammelt S. Fractures of the base of the fifth metatarsal bone: a critical analysis review. JBJS Rev. 2021;9(10):e21.00010.
Header Image by Doha Stadium Plus Qatar (Cropped)