Introduction

Bone is a viscoelastic tissue with a unique ability to withstand applied loads.1 However, repetitive loading coupled with inadequate time for tissue recovery can result in overuse injuries to bone.2 The phenotypes of these injuries vary and have collectively been referred to as bone stress injuries.3

In recent decades, communal efforts have contributed to an improved understanding of bone stress injuries.2–7 Large epidemiological studies have shown that bone stress injuries, while infrequent in the general population (<1% of all injuries), are common (up to 20% of sports injuries) in athletes and military personnel who participate in activities involving repetitive impact loading, such as marching, running, jumping and dancing.6 8 The diagnosis of a bone stress injury is typically made by history and physical examination combined with imaging.4 9 10 Various risk factors have been described including those related to low energy availability,11 biomechanics7 12 13 and training load.14 15 The management of bone stress injuries is guided by multiple factors such as the anatomical site, injury severity and level of sports participation.2 4 5 Treatment approaches include activity modification, protected weight-bearing, immobilisation, physical therapy, nutritional counselling and, in some cases, surgical fixation.2 Even with optimal treatment, bone stress injuries disrupt participation in sports and other physical activities, and full recovery may take many months.4 5 7 Recognition that multiple risk factors may result in an elevated risk for future bone stress injury has led to interest in prevention strategies including diversifying sports participation during childhood and adolescence, training load management and aspects of nutrition including calcium, vitamin D intake and optimising energy availability.11 14 16–20

Prior terminology to describe bone stress injuries such as ‘stress reactions,’ ‘bone marrow oedema syndrome’ and ‘stress fractures’ lacks precision, and the need for universally accepted definitions result from inconsistent reporting measures across studies.4 5 In addition, barriers to clinical care and advancing science may result from a lack of agreement on diagnostic approaches, risk factors and management strategies.2 One solution to address these gaps in knowledge is the use of the Delphi method to establish consensus on clinically relevant topics.21 The sports medicine community has utilised the Delphi consensus technique on topics of interest, including but not limited to tendinopathies, muscle injuries and femoroacetabular impingement.22–24

To date, no international consensus has been developed on the topic of bone stress injuries in athletes. The aim of this Delphi study was to ascertain the level of agreement among panellists on three topic areas of bone stress injuries including (1) pathophysiology, diagnosis, terminology and classification systems; (2) risk factors, screening and prevention; and (3) management and return to sport.

MethodsStudy design

The study reporting adheres to the ACcurate COnsensus Reporting Document (ACCORD; for ACCORD checklist, see online supplemental file 1).25 26 Approval for the study was obtained from the Ethics Committee of the Medical Association of Hamburg (protocol number 2023–300375-WF) to perform a modified Delphi study on bone stress injuries. Other study registrations were not performed. An initial steering group was established to complete a literature review and to develop statements. The Delphi panel was assembled through a criteria-based identification and selection of international experts and stakeholders. A three-round electronic consultation was conducted between September 2023 and March 2024 using an institution-based software package (LimeSurvey, V.6.1.8, Hamburg, Germany).

Steering committee, search strategy and statement development

The Delphi study was initiated by a multidisciplinary steering committee (TH, KH, EK, KLP, MF, SJW, AST). Prior to the development of statements, a literature search was conducted in PubMed, Web of Science, Cochrane databases, SPORTdiscus and Google Scholar using the keywords ‘stress fracture*’, ‘stress reaction*’, ‘bone stress injur*’ and ‘stress response’. Google Scholar was used in incognito mode. Further studies were identified through a forward (‘cited by’) and backward (‘citation’) search. Additional relevant publications that were not identified through the literature search were added manually based on recommendations from the steering group. The articles were reviewed non-systemically and used to publish a primer on bone stress injuries covering the topics of epidemiology, pathophysiology, diagnosis, screening, prevention, management, quality of life and outlook.2 Based on this primer, the steering group of the current Delphi study identified three key areas requiring international consensus on bone stress injuries: (1) terminology, pathophysiology, diagnosis and classification; (2) risk factors, screening and prevention; and (3) management and return to sport.

Panel selection

An international expert and stakeholder panel was formed based on predefined criteria with the goal of recruiting at least 30 participants. The accuracy and reliability of Delphi results minimally improve with increasing sample size beyond 30 participants.27 Members of the steering committee were excluded from serving as panellists.28 29 Initially, panel experts were identified based on their scientific involvement in the study of bone stress injuries. No exceptions were made to this rule to avoid selection bias. Accordingly, a search on Medline (Pubmed) for peer-reviewed publications between April 2008 and March 2023 was performed using the following search string:

#1: bone stress injur*

#2: stress fracture

#3: stress reaction

#4: athlet*

#5: sport*

#1 OR ([#2 OR #3] AND ([#4 OR #5]).

All authors with four or more first or senior authorships were considered for invitation to serve as panellists. This criterion was based on preliminary searches, taking into account a substantial number of candidates not willing to participate.22 Additional experts and stakeholders were then added to the panel to meet the strategic goal of promoting diversity among panel members.29 30 The selection of additional panellists is outlined below in the sections Equity, diversity and inclusion statement and Patient, public and clinical expert involvement. The invited panel candidates were approached via email. All experts and stakeholders who initially agreed to serve as panellists were allowed to participate in each of the three Delphi rounds. All panellists were asked to assess only those statements related to their field of expertise.

Equity, diversity and inclusion statement

Additional scientific experts with fewer than four senior or first authorships but in the order of the search results were invited to serve as panellists based on predefined diversity criteria: (1) at least three experts per continent (except Antarctica) and (2) at least 40% female and male experts.

Patient, public and clinical expert involvement

To include perspectives beyond scientific knowledge, each identified expert was asked to nominate a non-scientific stakeholder (eg, clinician, coach, athlete) that should be included in the Delphi process. The recommended clinical experts and stakeholders were subsequently invited to participate in the Delphi study. Athletes and coaches were only counted as representatives if they have a currently active role in their sports without being a clinician or scientist (eg, clinicians and scientists who used to be athletes/coaches do not count as representatives). In addition, the involvement of athletes, coaches and public is planned in the dissemination of the study findings.

Modified Delphi process

A modified Delphi method was utilised. Specifically, in the first round, the steering group formulated preselected statements. Piloting of the survey instruments was not performed. For ordinal statements, a 5-point Likert scale was used (strongly agree, agree, neutral, disagree and strongly disagree). For each statement, an optional free-text box was provided with encouragement for panellists to add comments to support their decision or suggest modifications. Additionally, an open text box was implemented in the first round to gather information, from the panellists’ point of view, on the lack of evidence and what additional categories would benefit from international consensus. All statements were sent to the panellists via email, with a reminder function integrated to reduce the attrition rate. No virtual or in-person meeting was held to ensure anonymity among panellists during the consensus process. All statements with agreement or disagreement of ≥80% were omitted for the subsequent rounds. All remaining statements were subjected to modification based on the panellists’ feedback and were admitted to the next round. Anonymity was applied at all stages (blinding the identity of panellists to each other and votes). The second round consisted of all statements that did not achieve consensus in the first round. For each statement, the voting results from the previous round were provided (anonymous group voting results and the panellist’s own vote). An example of the controlled feedback technique applied is provided in online supplemental file 2. This feature allowed every panel member to reflect on their own response from the previous round. The panellists were then asked to reconsider their previous ratings. Statements with ≥80% agreement or disagreement were excluded from the subsequent round. All other statements were subjected to modification based on the experts’ feedback and admitted to the next round. Similarly, a third round containing the remaining statements was then sent to the experts for final voting.

Statistical analysis and reporting

Levels of agreement to the Delphi statements are presented as percentage (%). Responses for ‘strongly agree’ and ‘agree’ were categorised as agreement, ‘neutral’ as undecided and ‘disagree’ or ‘strongly disagree’ as disagreement. Descriptive statistics were used to report demographic data of the participants. For each round, median and IQR voting results are presented. A χ2 analysis was performed to investigate responses by demographics (number of cases treated per year). Statistical significance was defined as p<0.05. All statistical analyses were performed by using SPSS V.29.0 (IBM, Armonk, New York). Based on the previous reports,11 28 consensus was defined a priori as agreement or disagreement of ≥80% for each statement. Three levels of agreement have been defined:

Full consensus: ≥80% of panellists agreed on the Delphi statement; no panellists disagreed.

Consensus with one or more disagreements: ≥80% of panellists agreed on the Delphi statement; one or more panellists disagreed.

Failure of consensus: <80% of panellists agreed on the Delphi statement.

Discussion

The purpose of this modified Delphi process was to ascertain the level of agreement among experts and stakeholders on key domains of bone stress injuries including (1) pathophysiology, diagnosis, terminology and classification systems; (2) risk factors, screening and prevention; and (3) management and return to sport. Agreement was reached on 41 out of 58 statements. Certain statements failed to reach international consensus; these statements may reflect topics that require future research.

Bone stress injury arises from cumulative microdamage, occurs on an injury continuum and may result in a complete bone fracture

The presumed underlying pathophysiology of bone stress injuries is an imbalance in bone metabolism favouring microdamage accumulation over its replacement via targeted bone remodelling.2 31 The Delphi panel affirmed that with continued stress and strain, injuries may progress along a pathology continuum. It was collectively indicated by the panellists that local bone tissue weakening may result in a partial or complete bone fracture. The concept of overuse injuries to bone that exists on a continuum has been proposed by others3 32 33 and is important to developing disease models that can advance treatment and prevention of these injuries.

The extent to which microdamage occurs is influenced by the magnitude, frequency, duration and direction of loading.33 It is plausible that a generally low training workload may exceed the bone’s tissue tolerance to stress and strain in athletes with impaired bone metabolism, thus considering these injuries as being multifactorial.11 However, the exact mechanisms are not fully understood.2 Greater understanding of the underlying pathophysiology is essential to inform prevention and treatment of bone stress injuries, for example, training interventions or pharmacological approaches.

The essentials of consistent terminology

The panel collectively accepted that, at present, bone stress injury is the preferred umbrella term. Until now, the clinical entity of activity-related, non-traumatic bone injuries has seen multiple terms, including stress reactions, stress fractures, fatigue fractures, hairline cracks, bone marrow oedemas or bone marrow lesions. To account for the broad spectrum of bone stress injuries, several classification systems have been described.9 10 34 Two commonly applied classification systems are those by Arendt et al 9 and Fredericson et al,10 and both use distinct imaging findings to assess the severity of bone stress injuries (eg, presence of a fracture line). The classification system described by Kaeding and Miller34 is of interest for surgical decision-making as displaced fractures and non-unions are classified separately. A universally accepted bone stress injury classification system is yet to be developed, though recommended by the Delphi panel. The need for further international collaboration in this field stems from the potential benefits of grading bone stress injuries in informing decision-making and prognosis, particularly in high-level athletes.4 5

Improving recognition and accurate diagnosis of bone stress injuries through comprehensive case history, physical examination and MRI

In sports medicine, case history taking and clinical examination are key skills that can be taught and learnt.35 The Delphi panel identified a number of features that may raise high suspicion of the presence of a bone stress injury, for example, recent changes in training volume or intensity. A comprehensive physical examination, including the observation of morphological variations (eg, pes cavus, bowlegs, pelvic tilt), is the basis for further clinical investigation.2 Superficial bone stress injuries such as the tibial shaft allow for palpation of the injury site.36 Depending on the injury location, specific tests have been described.2 36 In a study of 80 adolescents with tibial bone stress injuries, the vertical single leg hop test demonstrated highest sensitivity as compared with tap/percussion test, vibration test, weight bearing lunge test and fulcrum test.37 Notably, the authors highlighted that no individual test demonstrated high sensitivity or specificity. However, combining multiple tests resulted in higher likelihood for accurate diagnosis or exclusion of a bone stress injury. The limited physical exam features and characteristics to detect bone stress injuries remain a challenge for clinical evaluation.

Although patient history and clinical exam are critical components, imaging plays a crucial role in the diagnostic workup.4 38 39 Plain radiography has the benefit of being widely accessible and is an agreed on initial imaging modality, however, MRI is considered the imaging modality of choice by most Delphi panellists. Other imaging modalities are of limited utility but may be used in selected cases.2 4 5 Ultrasound was rarely considered useful by the panel in identifying bone stress injuries. Scintigraphy is less commonly used due to radiation exposure and its clear disadvantages as compared with MRI.5 For detailed bone imaging, CT is a standard imaging technique worldwide although rarely used for assessing bone stress injuries. However, its use can be beneficial for differential diagnosis workup, aiding decision-making in high-risk bone stress injuries and for assessing fracture consolidation.39–41 Newer imaging modalities are constantly evolving, and the use of volumetric interpolated breath-hold examination (VIBE) MRI, photon CT or dual-energy CT may offer new opportunities in the detection of bone stress injuries in the future.39 42 However, increased accessibility to advanced imaging technology is not without disadvantage, that is, the misinterpretation/overdiagnosis of asymptomatic imaging findings.43 44

Risk factors for bone stress injuries are multifactorial and result from factors related to skeletal loading, underlying bone health and other health behaviours

The improved understanding of the underlying pathophysiology provides a clear explanation for the occurrence of bone stress injuries in athletes experiencing rapid transitions in training load, for example, increases in training volume or intensity.2 A suboptimal bone workload is present when the magnitude of loading cycles constantly surpasses the tissue’s ability to withstand repetitive loads.45 While in vivo microdamage accumulation is the driving force in the development of bone stress injuries,46 the Delphi panellists recognise a clear association with risk factors related to bone health.

Prior work has characterised aspects of nutrition as contributing to bone stress injuries, including low energy availability.11 47 48 Low energy availability is the cause of Relative Energy Deficiency in Sport (REDs) and is associated with higher rates of bone stress injuries.48–51 Low energy availability is more common in female endurance athletes,52 and female athletes have been suggested to have elevated risk for bone stress injuries.53 Our panel identified similar risk factors for bone stress injury, and those related to low-energy availability include insufficient caloric intake to meet energy demands, history of eating disorder, low body mass index, low bone mass, menstrual dysfunction in women and low testosterone in men. Importantly, bone stress injuries may serve as the impetus for amateur athletes to consult a medical professional for the first time. Therefore, clinicians must be knowledgeable about the frameworks used for identifying REDs.54

Besides impaired bone health, probability of fracture is greater in the presence of increased stress and strain through abnormalities or variations in biomechanics and movement patterns.33 Panellists agreed that change of habitual footwear represents a modifiable risk factor in athletes whose sporting activity includes high volumes of running (eg, distance runners, triathletes, orienteers). Introducing new footwear has been previously proposed as a risk factor for bone stress injury. For example, use of ‘advanced footwear technology’ containing carbon fibre plates and responsive midsole foam has been described in a small population of injured endurance athletes with bone stress injuries of the tarsal navicular7 and resulted in calls to study this form of footwear.55 Earlier work has shown that minimalist footwear (shoes with flexible midsole, minimal stack height and no foot arch support) may contribute to metatarsal injuries.56 However, the Delphi panel did not agree that minimal shoes or advanced footwear technology increase the risk for bone stress injuries, such that specific footwear cannot currently be recommended to modify injury risk. Nevertheless, findings from the present consensus statement combined with earlier work suggest the need for athletes to use caution regarding changes in footwear.

Besides footwear, the panel was in agreement that intrinsic biomechanical patterns may contribute to risk of bone stress injury in running athletes. This is consistent with prior work published on clinical and musculoskeletal modelling studies, indicating that low step rates were associated with increased injury risk.13 57 Accordingly, gait retraining interventions, such as adopting a higher step rate, have been previously suggested.13 However, in line with the preferred movement pathway paradigm,58 gait retraining in athletes is not routinely recommended by the Delphi panel. The panel though acknowledged that gait retraining may be a treatment intervention in athletes with a history of recurrent bone stress injuries.59

Clinical assessment of risk factors includes bone mass scanning and laboratory testing

Notably, a history of bone stress injury is seen to be among the strongest risk factors for developing a subsequent bone stress injury.53 60 Thus, it is crucial to take steps to mitigate risk factors through sufficient energy intake, appropriate nutrition or training load monitoring.11 36 59 Consensus was not reached for routinely obtaining a dual-energy X-ray absorptiometry study to measure bone mineral density (BMD) values in at-risk sports such as running if athletes are injury free. However, the threshold for defining low bone mass in athletes was agreed as BMD Z-score ≤−1.0. A similar threshold has been previously proposed for male athletes61 and for female athletes participating in weight-bearing sports by the American College of Sports Medicine.62 However, the use of traditional bone health screening principles in sports medicine has been challenged, and new approaches such as sport-specific BMD Z-scores or bone microarchitecture assessment via high-resolution peripheral quantitative CT have been proposed.47 63

The panel reached consensus that laboratory testing is recommended to evaluate patients with bone stress injury with particular recommendations for screening for low serum vitamin D status. Further laboratory testing may be indicated and are similar to those proposed for assessing REDs.11 The potential advantage of assessing bone biomarkers in response to acute or long-term exercise remains ambiguous.64 The panel concluded that athletes with a bone stress injury should be provided nutrition counselling and supplemental vitamin D if serum levels are insufficient. Furthermore, the panel acknowledged that lifestyle factors including sleep and psychological stress may be important to address consistent with prior work suggesting these mechanisms being associated with rapid bone loss.65

Other instrumental assessments to identify risk factors may potentially comprise movement analyses in order to identify biomechanical abnormalities associated with an increased risk of bone stress injury.12 66 Movement patterns in sports, such as running gait analysis, may be assessed clinically without advanced technology, but the use of camera-based systems or inertial measurement units provides higher accuracy, especially in high-speed motion analysis.67

Management of bone stress injuries depends on anatomical location and is mostly non-surgical

The anatomical location of bone stress injury should guide management. While most anatomical sites of bone stress injury are low risk, panel consensus on high-risk locations was achieved for the superior cortex of the femoral neck, anterior cortex of tibial diaphysis, navicular and base of fifth metatarsal. These anatomical locations of injury are fewer in number than prior definition for high-risk locations68 but are consistent with literature describing the need to consider surgical management for injuries to the fifth metatarsal,69 navicular,41 femoral neck70 and anterior tibial cortex.71

Standard fracture care may include periods of immobilisation and partial weight-bearing for lower extremity bone stress injuries. Elite athletes may consider surgery, particularly with the goal of facilitating early return to sport.72 When considering surgical treatment, panellists may support the decision based on features including delayed union (no signs of union>3 months), non-union (no signs of union>6 months), recurrent injury, fracture displacement and injury site being prone to treatment complications.

The evolving role of shockwave, orthobiologics, shoe technology and antigravity treadmills in rehabilitation and fracture healing

The field of bone stress injury treatment has seen innovations and advancements in recent decades. Research has been conducted to identify strategies to promote faster bone healing with variable results, and similarly our panel did not reach consensus on adjuvant strategies to promote healing. No agreement was reached on prescription of low-intensity pulsed ultrasound, extracorporeal shockwave or electromagnetic stimulation. Minimally invasive interventions such as bone marrow aspirate injection or autologous platelet-rich technologies have been explored to expedite fracture healing73; our panel did not reach consensus on use of these treatments. Similarly, no conclusion on the potential role of (shock-absorbing) footwear or insoles can be made at present.74 In rehabilitation, an increasing number of studies are currently investigating the role of modern treadmill technology (eg, antigravity or underwater treadmills).75–77 While modern innovations certainly play an evolving role in clinical care of bone stress injuries, further evidence on their use or retention is warranted.

There is an unmet need for return to sports guidelines for athletes with bone stress injuries

Rehabilitation protocols should be provided through structured physical therapy.78 A number of clinical criteria identified by the panel may guide return to sport. These include both the anatomical location and severity of injury and level of sport participation. Additionally, stepwise return to activity may be guided by pain-free status with sports and daily activities, with bone loading tests and with palpation over the injury site.

Multidisciplinary collaboration was recognised as a key aspect of managing bone stress injuries. In countries without a fully recognised sports medicine specialty, athletes may seek consultation with family doctors, orthopaedists, physiatrists and other healthcare professionals for assistance with sports injuries. As with any overuse injuries in athletes, the necessity for interdisciplinary collaboration becomes apparent. For example, orthopaedists may lack deep knowledge of concepts of low energy availability, while endocrinologists may lack knowledge of fracture care.

Future research areas

Some Delphi statements failed to achieve consensus, highlighting areas with a need for future research. Injury prevention requires the definitions and correction of both training errors and lifestyle risk factors. This includes a detailed understanding of how low-energy availability impacts bone health.79 80 Implementing innovative strategies in the treatment and rehabilitation of bone stress injury is another field with a need for research.2 48 73 77 This may comprise the use of shockwave or antigravity treadmills as well as surgical approaches.2 The enthusiasm for new technological interventions is evident in the sports medicine community; however, evidence-based practices are paramount to ensure that the health and safety of our athletes remain a top priority. Nevertheless, future studies can benefit from new research methods such as finite element analysis, big data analyses and wearable technology.2 Translational research may contribute to clinical care by investigating types of bony microdamage, mechanotransduction signalling pathways and fracture resistance mechanisms (eg, sacrificial bonding and dilatational band formation).81–83

Strength and limitations

The steering group invited a diverse group of experts and key stakeholders. To minimise risk of bias in voting members, experts were selected using objective criteria documented by peer-reviewed publications from multiple countries. The panel was supplemented by inviting stakeholders including athletes and coaches using pre-defined invitation criteria. However, a balanced view on the topic should not be taken for granted and the panel group may not necessarily reflect opinions of others. Also, differences in healthcare systems (eg, unequal access to modern imaging modalities) are a major challenge when defining universally accepted recommendations. To consider this limitation, terms like ‘when accessible’ or ‘if feasible’ were added to some consensus statements, acknowledging the varying healthcare resources around the globe. We used predefined criteria to improve diversity but did have more limited participation from certain geographic locations (eg, Africa and South America), coaches/athletes and from certain healthcare professions; in part, this reflects limited resources to publish on the topic. Furthermore, some topics such as the pathophysiology of bone stress injuries would benefit from in vivo studies, which represent higher evidence level as compared to expert consensus. Another limitation is the attrition in panellist participation, which may have been due to the length/comprehensiveness of the survey. This (modified) Delphi study differs from traditional approaches in that all 41 experts and stakeholders who agreed to participate in the consensus process were invited to vote in every round, regardless of their participation in previous rounds.