Bone stress injury recovery and return to sport in runners: a multi-site prospective observational cohort study protocol across the USA and Canada

Abstract

Introduction

Bone stress injuries (BSIs) are common overuse injuries in athletes and are associated with prolonged recovery and high rates of reinjury. Despite their clinical importance, there is no consensus on how to define or assess BSI healing. Clinical, functional and imaging outcomes for BSI healing are used inconsistently across studies, limiting comparability and the development of evidence-based return-to-sport guidelines. This study aims to identify and characterise candidate outcomes for assessing BSI healing in runners, including their time course and variability across function, imaging and clinical domains.

Methods and analysis

This is a 1 year, multi-site prospective cohort study of male and female runners with a recent MRI-confirmed BSI of the tibia, metatarsals, femoral neck or sacrum. We will enrol participants across four clinical and academic research sites in the USA and Canada, within 3 weeks of diagnostic MRI and follow them longitudinally for 1 year through recovery, return to sport and full sport participation. Candidate outcomes for BSI healing include time to pain-free functional milestones (eg, hopping, jogging), completion of a return-to-run protocol, medical clearance for unrestricted activity, and changes in MRI grade and clinical severity scores. Secondary and exploratory measures include training patterns, wearable-derived activity and sleep metrics, biomechanical assessments, and serum proteomic profiles. We will use descriptive statistics to characterise the time course and variability of candidate outcomes. We will explore associations between outcomes and subsequent BSIs within 1 year using regression models.

Ethics and dissemination

This study has received institutional review board approval at all participating institutions. We will obtain written informed consent from all participants prior to any study procedures (and assent with parental consent for minors). We will disseminate results through peer-reviewed publications and scientific conferences.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Multi-site prospective cohort design across clinical and sport settings.

• Standardised return-to-run protocol to reduce variability in clinical management and rehabilitation.

• High-frequency, milestone-based outcome collection during recovery.

• Integration of clinical, imaging, behavioural and biological measures.

• Observational design limits causal inference.

Introduction

Bone stress injuries (BSIs) are common overuse injuries caused by repetitive loads applied to generally normal bone that exceed the adaptive capacity of the bone.¹ Female athletes and endurance runners are at a high risk of BSIs, especially in sports that expose the athlete’s lower extremities to repetitive impact, including cross country, gymnastics and track,^{2 3} are at especially high risk for BSIs. BSIs are injuries of insidious onset that can impact performance over weeks to months (due to the prolonged recovery period), and often lead to substantial financial and personal burden.^{1 4}

Current BSI management strategies include cessation of load-bearing activities, consideration for crutches and/or immobilisation, pain management, and a gradual return to activities until symptoms resolve.^{4 5} Premature return to sport can result in risk of complications and reinjury of the healing BSI.^{4 6–9} Notably, our prior work revealed that one-third of women with a tibial BSI will sustain at least one additional BSI within 1 year, suggesting that current return-to-sport guidelines are inadequate.⁴

MRI is generally recognised as the gold standard for diagnosing BSIs, which can be graded from least (grade 1) to most (grade 4 or 4b) severe.^10–12 Common injury locations include the tibia, metatarsals, lumbar spine, femoral neck and pelvis.² Femoral neck BSIs are typically associated with longer return to sport times and greater need for surgical intervention compared with other common BSI locations.²

Despite the clinical importance of optimising recovery, there are no gold standard or validated clinical, functional or imaging outcomes that are consistently used to assess BSI healing and guide return to sport. A variety of primary and secondary outcomes have been used in randomised controlled trials and prospective intervention studies, yet these outcomes are inconsistently defined. For example, the most commonly used outcome, time to pain-free single-leg hopping, varies substantially across studies, including differences in hopping protocol (eg, number of repetitions, hopping height and landing characteristics) and the reference timepoint used to define ‘day 0’ (eg, diagnosis, study enrolment, symptom onset or undefined).^13–16 The most commonly used objective assessment of BSI healing is change in MRI grade.^{17 18} However, changes in MRI grade have limited utility in determining readiness to return to sport.^{19 20} Together, this variability and lack of validity limit comparability across studies and impede the development of evidence-based rehabilitation and return-to-sport guidelines.

BSI healing outcomes provide limited insight into the underlying biological, behavioural and mechanical factors that may influence return to activity trajectories and risk of reinjury. This distinction is important, as individuals with similar clinical recovery timelines may differ in their risk of subsequent BSI.⁴ Established risk factors for BSI include reproductive hormone deficiency, energy deficiency, poor pre-existing bone health and increased mechanical loading/training volume beyond the capacity of the bone to adapt.¹ Emerging evidence suggests that further analysis of these risk factors, as well as serum biomarkers, may improve our understanding of BSI healing and subsequent BSI risk. For example, our prior work demonstrated that parathyroid hormone and serum protein profiles differ in women who sustained a single BSI versus multiple BSIs in 1 year.^{4 21} Collectively, these findings highlight the potential for biological and behavioural markers to complement clinical outcomes in understanding both recovery and reinjury risk.

The lack of standardised healing outcomes and the limited integration of biological and behavioural factors represent barriers to advancing both clinical care and interventional study design. A comprehensive approach that evaluates candidate healing outcomes alongside factors that may influence recovery is needed to move the field forward.

Accordingly, we aim to identify and characterise candidate outcomes for assessing BSI healing in runners, including their time course and variability. Secondary objectives include: (1) evaluating associations between candidate healing outcomes and subsequent BSIs within 1 year; (2) describing behavioural, biological and mechanical factors that may influence recovery trajectories and reinjury risk; (3) identifying standardised endpoints for future clinical intervention trials.

Methods and analysis

Study design

This is a 1-year, multi-site, prospective cohort study of runners with a recently diagnosed BSI. We will follow participants from MRI diagnosis through recovery and return to sport, with continued monitoring for subsequent injury for 12 months.

Setting

We will conduct in-person data collection at clinical research facilities in Bloomington, MN, Eugene, OR, Victoria, BC, and Stanford, CA, with additional remote data collection via electronic surveys and wearable devices.

Participants

We will enrol up to 105 male and female endurance runners with a current BSI. Enrolment began in August 2025 with data collection estimated to continue through June 2028.

Inclusion criteria

• MRI-confirmed BSI (grade ≥2 on the Fredericson scale¹⁰) of the tibia, metatarsals (2–4), femoral neck or sacrum.

• Age 16–35 years.

• Body Mass Index 17.5–30 kg/m².

• Participation in endurance exercise including running ≥24 kilometres per week prior to injury.

• Enrolment within 3 weeks of diagnostic MRI.

Exclusion criteria

• Endocrine or bone disorders affecting bone health (eg, osteogenesis imperfecta, osteopetrosis, rickets, diabetes or hyperparathyroidism).

• Use of bone-modifying medications (eg, prednisone, bisphosphonates).

• Current pregnancy or lactation.

• Contraindications to MRI.

• Injuries requiring surgical management.

• Anterior tibial BSI.

Recruitment and enrolment

We will identify participants through a combination of methods depending on study site location. Two participating sites are embedded within clinical environments and will use electronic medical record screening (eg, SlicerDicer) and clinician referral to identify individuals presenting with a recent BSI diagnosis. At these sites, patients may be approached during routine clinical care and consented at or shortly after their clinical visit. Additional participants may be identified through community outreach and direct contact with athletic organisations. The remaining two sites are academic and sports medicine institutes without direct electronic medical record access and will rely on clinical referrals, community outreach and direct contact with athletic organisations.

All interested individuals will complete a telephone screening to assess eligibility. We will invite eligible individuals for an in-person baseline visit, during which written informed consent (and assent for minors) will be obtained prior to any study procedures.

At the two non-clinic sites, we will use a dual-consent process for individuals with a suspected but not confirmed BSI. Participants first provide consent for a screening MRI to confirm diagnosis and injury grade. We will invite those who meet eligibility criteria based on MRI findings to provide assent/consent for full study participation.

Study procedures

All data collection and electronic surveys will be completed via Research Electronic Data Capture (REDCap).

Baseline visit

At enrolment, participants will undergo anthropometric assessment (height, weight) and a fasting blood draw. They will complete a baseline survey for medical history (including history of prior BSIs) and running/training history. We will obtain a baseline clinical severity score; the clinical severity score is a composite of seven sign and symptom assessments (pain during daily activities, night pain, pain during running and hopping, local tenderness, pain during percussion and local soft tissue swelling), each scored 0–3, for a maximum total score of 21.¹⁶ We will orient participants to electronic survey schedules and the paper-based return-to-run training diary. Those who opt in will authorise collection of Garmin wearable data including 60-day historical and 1-year prospective activity and daily health metrics (table 1).

Table 1. Schedule of study procedures.

Study procedure	Baseline (week 0)	Weeks 1–4 (weeks 1–8*)	Visit 2 (week 4; week 8*)	Weeks 4–8 (weeks 8–12*)	Visit 3 (week 8; week 12*)	Post-RTR (ongoing)	12-month exit
In-person assessments
Informed consent	●
Height & weight	●
Fasting blood draw	●
Validated surveys†	■				■		■
Baseline questionnaire	■
Clinical severity score	●		●		●
MRI—injured site (Fredericson grade)			●		●
Single-leg hop analysis (OpenCap)					●
Adverse event assessment			●		●
Ongoing monitoring
Functional outcomes survey‡	■ 2×/week	■ 2×/week	■	■ 2×/week	■
Subsequent injury survey						■ Monthly	■
Daily training diary (paper+photo upload)	▬	▬	▬	▬	▬
Wearable device data—Garmin§	◊	◊	◊	◊	◊	◊	◊

● In-person assessment; ■ Electronic survey (REDCap); ▬ Daily; ◊ Optional.

^*Femoral neck BSIs: visit 2 occurs at week 8 (not week 4); visit 3 occurs at week 12 (not week 8); monitoring periods adjusted accordingly.

^†Validated surveys: EDE-Q: Eating Disorder Examination Questionnaire; LEAF-Q: Low Energy Availability in Females Questionnaire (modified; females only); LEAM-Q: Low Energy Availability in Males Questionnaire (males only); PSQI: Pittsburgh Sleep Quality Index; PHQ-8: Patient Health Questionnaire-8; GAD-7: Generalized Anxiety Disorder Scale; BAS-2: Body Appreciation Scale-2.

^‡Functional outcomes survey frequency: 2×/week from enrolment until return-to-run protocol complete.

^§Wearable device data collection is optional (participant opt-in).

BSI, bone stress injury; RTR, return to run.

Participants will also complete a battery of validated surveys at time of enrolment. Female participants will complete the modified Low Energy Availability in Females Questionnaire (LEAF-Q), a 22-item tool assessing low energy availability risk across injury, gastrointestinal and reproductive/menstrual subscales²²; male participants will complete the 42-item Low Energy Availability in Males Questionnaire (LEAM-Q; partially validated).²³ All participants will complete the Eating Disorder Evaluation Questionnaire (EDE-Q), a 28-item measure of eating disorder symptoms and related psychopathology, comprising four subscales: restraint, eating concern, shape concern and weight concern.²⁴ Sleep quality will be assessed using the Pittsburgh Sleep Quality Index (PSQI), a 19-item instrument evaluating sleep quality across seven domains, including subjective quality, latency, duration, efficiency, disturbances, medication use and daytime dysfunction.²⁵ Depressive symptoms will be assessed using the 9-item Patient Health Questionnaire-8 (PHQ-8)²⁶ and anxiety severity using the 7-item Generalized Anxiety Disorder Scale (GAD-7).²⁷ Body image will be evaluated using the 10-item Body Appreciation Scale-2 (BAS-2).²⁸ The full battery is estimated to require approximately 30 to 45 min to complete and will be administered via REDCap at baseline, one follow-up visit and at 12 months. Item-level completion rates will be monitored and reported.

Specimen collection and processing

We will perform a baseline blood draw (approximately 30 mL) using phlebotomy-trained study staff following an overnight (8 hours) fast. We will separate samples for serum and plasma and store them at −80°C until analysis using the SomaScan platform (SomaLogic, Inc., Boulder, CO), which simultaneously quantifies approximately 7000 human proteins, including markers relevant to bone metabolism, inflammation and metabolic regulation. We will archive additional serum for future analyses.

Overview of return-to-run progression

We will provide participants with a standardised phased return-to-run (RTR) protocol developed in collaboration with experts in BSIs, including metabolic bone researchers, sports medicine physicians, endocrinologists, biomechanists, exercise physiologists and physical therapists. This phase-based standardised RTR protocol will reduce variability in rehabilitation and enable consistent longitudinal assessment of recovery. The RTR protocol consists of progressive phases that systematically increase loading through walking, jogging and running intervals (table 2). Progression is guided by symptom response, with participants required to remain pain-free at each stage before advancing to the next phase.

Table 2. Summary of return-to-run protocol.

Phase	Suggested activities	Criteria to complete phase
1: Protected weightbearing	Activities of daily living, swimming, cycling	Pain-free walking without an assistive device for 3 continuous days
2: Light loading	Walking, swimming, deep-water running, elliptical, stair climber, strength training	Pain-free dedicated walking for 30 min Pain-free single-leg hop × 10
3: Progressive loading	Progressive jogging programme, plus activities from prior phases to maintain cardiovascular fitness	Pain-free jogging for 30 min
4: Return-to-run preparation	Progressive running programme	Pain-free run progression. Ready to return to unrestricted running

Full protocol includes day-by-day instructions with modification suggestions based on individual symptoms.

Longitudinal monitoring during return-to-run progression

Timing of advancement through the RTR protocol will be captured through a series of twice-weekly, electronically administered functional outcome surveys. The survey series is composed of five individual surveys that query participants for milestone achievement according to the phases of the RTR protocol, with participants advancing through the series based on self-reported pain. The surveys progress as follows: (1) pain-free walking without an assistive device for 3 consecutive days; (2) 10 pain-free single-leg hops on injured leg; (3) pain-free 1-mile run; (4) pain-free 30-min run; (5) completion of RTR progression (table 2). Each survey is repeated twice weekly until the participant reports pain-free activity. Once pain-free activity is reported, the participant will advance to the next survey in the series at the next timepoint until they have successfully achieved all milestones and returned to unrestricted running.

Participants will also record physical activity using a paper-based daily training diary documenting exercise duration, distance, pace and pain with activity. The diary captures completion dates for key candidate outcomes including first pain-free 1-mile jog, first pain-free 30-min continuous jog and completion of the final RTR phase. These diary-captured dates serve as a pre-specified secondary source for time-to-event outcomes in case electronic survey timestamps are unavailable. Additionally, diary documentation allows for analysis of adherence to the provided RTR progression. Diary pages will be scanned by study staff at each in-person visit. On completion of the RTR protocol, participants will be instructed to photograph and upload completed pages for final data collection. Participants who consent to optional wearable device data collection will provide objective activity data, including stride metrics, cadence, ground contact time, heart rate and sleep duration.

Follow-up visits

Participants will return for in-person assessments at approximately 4 and 8 weeks post-MRI diagnosis (or 8 and 12 weeks for femoral neck BSIs). Each follow-up visit includes clinical severity scoring, MRI of the injured site, adverse event assessment and repeat surveys at the third visit. To determine change in MRI grade, de-identified digital imaging and communication files in medicine files for diagnostic and study MRIs will be graded by a single, board-certified, fellowship-trained musculoskeletal radiologist using the Fredericson scale¹⁰: Grade 1, periosteal oedema on T2 with normal marrow on T1 and T2; Grade 2, periosteal oedema with marrow oedema on T2; Grade 3, periosteal and marrow oedema on both T1 and T2 without a fracture line; Grade 4, periosteal and marrow oedema on T1 and T2 with a visible fracture line.

At the third in-person visit, participants who have completed the pain-free, single-leg hopping progression will undergo a biomechanical assessment using a markerless motion capture system (OpenCap). During this assessment, three mobile phone cameras will capture 10 single-leg hops to quantify lower extremity joint kinematics and kinetics. Video recordings, which include participants’ faces, will be stored in protected health information-compliant secure storage. Participants will be asked to provide consent/assent to sharing de-identified processed data with the OpenCap development team for algorithm development.²⁹

Long-term follow-up

Following self-guided RTR protocol completion, participants will complete monthly electronic surveys to report subsequent BSI diagnoses. Additionally, participants will record date of medical clearance for unrestricted activity and any treatment they received beyond standard of care, including nutritional guidance with a registered dietitian, physical therapy, sports psychology services, bone stimulation and shock wave therapy. A final assessment at 12 months will include a final subsequent BSI survey, as well as the validated questionnaires described previously.

Outcomes

Candidate outcomes were selected based on prior literature demonstrating substantial variability in definitions and implementation of BSI healing endpoints. Outcomes were intentionally grouped across three areas: functional, imaging and clinical. These outcomes capture complementary aspects of recovery and are intended to facilitate identification of outcomes most suitable for future intervention trials.

Primary candidate outcomes (BSI healing)

The primary objective of this study is descriptive and focuses on the time course and variability of seven candidate outcomes:

Time-to-event outcomes (days from diagnostic MRI): (1) 10 pain-free single-leg hops on the injured side (both sides for sacrum) without shoes and on a hard surface; (2) pain-free 1-mile continuous jogging; (3) pain-free 30-min continuous jogging; (4) completion of RTR protocol; (5) receipt of medical clearance for unrestricted activity (table 3).

Table 3. Summary of outcomes.

Outcome	Domain	Data source	Variable type
Primary candidate outcomes
Time-to-event outcomes (days from diagnostic MRI)
10 pain-free single-leg hops on injured limb (bilateral for sacral BSIs), without shoes on a hard surface	Functional	Phase-specific survey (REDCap)	Time-to-event (days)
Pain-free 1-mile continuous jogging	Functional	Phase-specific survey (REDCap); training diary	Time-to-event (days)
Pain-free 30-min continuous jogging	Functional	Phase-specific survey (REDCap); training diary	Time-to-event (days)
Completion of return-to-run (RTR) protocol	Functional	Phase-specific survey (REDCap); training diary	Time-to-event (days)
Receipt of medical clearance for unrestricted activity	Clinical	Participant report (REDCap); medical records review	Time-to-event (days)
Change-based outcomes (at in-person visits: weeks 4 &and 8 for tibial/metatarsal/sacral BSIs; weeks 8 and 12 for femoral neck BSIs)
Change in MRI grade from diagnostic MRI (Fredericson scale)	Imaging	MRI	Ordinal change score
Change in clinical severity score from baseline study visit (composite of 7 signs/symptoms; 0–21)	Clinical	In-person assessment	Continuous change score
Secondary outcomes
Incidence of subsequent clinically diagnosed BSI within 12 months	Clinical	Monthly survey (REDCap); medical records review	Binary (yes/no)
Compliance with RTR protocol (% of prescribed sessions completed)	Behavioural	Training diary	Continuous (%)
Time to completion of individual RTR protocol phases	Functional	Training diary; phase-specific survey (REDCap)	Time-to-event (days)
Exploratory outcomes
Behavioural and training metrics: daily exercise duration, distance, pace and training load patterns	Behavioural	Training diary; wearable device	Continuous
Wearable-derived running and activity metrics: stride count, cadence, stride length, ground contact time, vertical oscillation, heart rate, step count, sleep duration (optional; Garmin participants only)	Behavioural	Garmin wearable device	Continuous
Lower extremity joint kinematics (ankle, knee, hip) and estimated joint kinetics during single-leg hopping	Biomechanical	OpenCap markerless motion capture	Continuous
Serum proteomic profiles	Biological	Aptamer-based proteomics (fasting blood draw)	Continuous
Patient-reported psychosocial and behavioural outcomes (EDE-Q, LEAF-Q/LEAM-Q, PSQI, PHQ-8, GAD-7, BAS-2)	Psychological	Validated questionnaires (Research Electronic Data Capture; REDCap)	Continuous (scale score)

BAS-2, Body Appreciation Scale-2; BSI, bone stress injury; EDE-Q, Eating Disorder Examination Questionnaire; GAD-7, Generalised Anxiety Disorder Scale; LEAF-Q, Low Energy Availability in Females Questionnaire; LEAM-Q, Low Energy Availability in Males Questionnaire; PHQ-8, Patient Health Questionnaire-8; PSQI, Pittsburgh Sleep Quality Index; REDCap, Research Electronic Data Capture; RTR, return to run.

Change-based outcomes (site-specific follow-up visits: 4 and 8 weeks for tibial, metatarsal and sacral BSIs; 8 and 12 weeks for femoral neck BSI): (6) change in MRI grade from diagnosis; (7) change in clinical severity score from baseline visit (table 3).

These outcomes were selected to inform the identification of clinically meaningful and reproducible endpoints for future randomised controlled trials targeting BSI recovery. Outcomes will be captured using phase-specific surveys, training diaries, clinical assessments and imaging.

Secondary outcomes

Secondary outcomes include: (1) incidence of subsequent clinically diagnosed BSI within 12 months; (2) compliance with the RTR protocol (percentage of prescribed sessions completed) and (3) time to completion of individual protocol phases (table 3).

Exploratory outcomes

Exploratory outcomes span five areas: (1) behavioural and training metrics (daily exercise duration, distance, pace, progression and training load patterns derived from diary and wearable data); (2) wearable-derived running and activity measures in the subset of participants using a Garmin device (stride count, cadence, stride length, ground contact time, vertical oscillation, heart rate, step count and sleep duration); (3) biomechanical outcomes from OpenCap assessment (lower extremity joint kinematics at the ankle, knee and hip, and estimated joint kinetics during single-leg hopping); (4) biological outcomes (aptamer-based proteomic profiles) (5) patient-reported outcomes from the validated instruments described above (EDE-Q, LEAF-Q/LEAM-Q, PSQI, PHQ-8, GAD-7, BAS-2). Where multiple data sources are available for a given outcome, pre-specified rules will prioritise time-stamped electronic survey data, with diary or wearable data used to supplement missing values and confirm milestone timing (table 3).

Statistical analysis

We will use descriptive statistics to characterise the distribution and variability of candidate outcomes. We will conduct analyses for the overall cohort and stratify by injury location and sex to characterise potential differences in outcome trajectories across subgroups. We will explore associations between candidate outcomes and subsequent BSI using logistic regression models, adjusting for relevant covariates such as age, BMI, training history and self-reported menstrual status. We will use sensitivity analyses to assess the influence of protocol adherence by excluding participants who complete fewer than 80% of prescribed RTR sessions.

Sample size

This study will enrol a convenience sample with a goal of 90 completers across all participating sites including approximate targets of 30 tibial BSIs, 30 metatarsal BSIs, 15 sacral BSIs and 15 femoral neck BSIs. To account for anticipated dropout and loss to follow-up, we plan to enrol up to 105 participants over an 18-month enrolment period.

Because the primary aim is descriptive, we did not perform formal power analysis. The proposed sample sizes are expected to yield reliable estimates of outcome variance across injury locations and sexes. These variance estimates will directly support sample size calculations for future randomised controlled trials targeting BSI recovery.

Data management

We will collect and manage study data using REDCap electronic data capture tools. Additional data sources include wearable-derived activity and sleep metrics obtained through a secure Stanford-based platform for Garmin data integration, and biomechanical data collected using OpenCap.

We will store all study data on secure, password-protected servers with access restricted to authorised study personnel. We will store identifiable information separately from study data and replace with unique study identifiers for analysis. We will train all study personnel in data privacy and confidentiality procedures, and data handling will comply with institutional policies and applicable regulatory standards. Data quality will be monitored through routine range checks and internal review procedures.

Patient and public involvement

Patients and members of the public were not involved in the design, conduct, reporting or dissemination plans of this research.

Ethics and dissemination

This study has received institutional review board approval at all participating institutions: HealthPartners Institutional Review Board (IRB), Bloomington, MN (approval #A25-047); Stanford University IRB, Stanford, CA (approval #76887); Canadian Sport Institute IRB, Victoria, BC, Canada (approval #H24-02905) and the University of Oregon IRB, Eugene, OR (approval #00001449). We will obtain written informed consent from all participants prior to any study procedures (and assent with parental consent for minors).

Participant safety will be monitored throughout the study. Adverse events will be assessed at each in-person visit and documented in REDCap. The principal investigator will provide oversight and review of all adverse events. Protocol deviations will be monitored on a routine basis by each site principal investigator and the overall principal investigator. Clinically significant incidental findings on study MRIs will be communicated to participants by the study physician. Participants will be referred to appropriate clinical care as indicated.

De-identified study data will be made available to qualified researchers on reasonable request to the corresponding author following publication of primary results.

We designed this study to address a critical gap in BSI management by systematically evaluating outcomes used to assess healing and return to sport. Findings will inform future prospective and randomised controlled trials and support more consistent clinical decision-making.

We will disseminate results through peer-reviewed publications, scientific conferences, clinical education and social media.

Data availability statement

Data are available upon reasonable request.