The Impact of Multidirectional Sports on Bone Health in Female Runners and Bone Stress Injury Risk

Abstract

Purpose:

This study investigates the connection between playing multidirectional sports (MDS) in youth and bone stress injuries (BSIs) in female runners. BSIs, encompassing stress reactions and fractures, are prevalent among female runners. We explored whether engaging in sports like soccer or basketball during younger years could contribute to building stronger bones, thereby reducing the risk of BSIs in these athletes.

Methods:

We recruited female collegiate cross-country runners and divided them into two groups: 1) RUN: athletes with a background solely in cross-country training, recreational running, swimming, and/or cycling, and 2) RUN+MDS: athletes with additional experience in soccer or basketball training and competition. High-resolution peripheral quantitative computed tomography (HR-pQCT) was employed to assess bone properties at various sites: the distal tibia, common BSI locations (tibia, fibula, and 2nd metatarsal diaphysis), and high-risk BSI locations (base of the 2nd metatarsal, navicular, and proximal diaphysis of the 5th metatarsal). Radius scans served as control sites for comparison.

Results:

The RUN+MDS group (n=18) exhibited significantly enhanced bone characteristics at the distal tibia compared to the RUN group (n=14). Specifically, RUN+MDS showed greater cortical area (+17.1%) and thickness (+15.8%), as well as increased trabecular bone volume fraction (+14.6%) and thickness (+8.3%) (all p<0.005). Failure load, an indicator of bone strength, was 19.5% higher in the RUN+MDS group (p<0.001). In the fibula diaphysis, the RUN+MDS group had 11.6% greater total area and 11.1% greater failure load (all p≤0.03). At the 2nd metatarsal diaphysis, total area was 10.4% larger in RUN+MDS, accompanied by greater cortical area and thickness, and an 18.6% higher failure load (all p<0.05). Furthermore, RUN+MDS athletes presented greater trabecular thickness at the base of the 2nd metatarsal and navicular, and enhanced cortical area and thickness at the proximal diaphysis of the 5th metatarsal (all p≤0.02). No significant differences were found at the tibial diaphysis or radius between the groups.

Conclusion:

Our findings suggest that encouraging young athletes to participate in multidirectional sports and delaying specialization in running can lead to a more robust skeletal structure, potentially reducing the incidence of BSIs. This study highlights the long-term benefits of diverse athletic engagement during youth for bone health and injury prevention in female runners.

Keywords: EXERCISE, PHYSICAL ACTIVITY, RELATIVE ENERGY DEFICIENCY IN SPORT, RUNNING, STRESS FRACTURE, STRESS REACTION

INTRODUCTION

Bone stress injuries (BSIs), encompassing stress reactions and stress fractures, are understood to arise from mechanical fatigue. This occurs when repetitive loading, even below the bone’s yield threshold, leads to the accumulation of microdamage (1). Microdamage is a natural biological process that triggers skeletal renewal through targeted remodeling (2). However, when workload is not optimally managed, such as with rapid increases in loading intensity, microdamage can accumulate excessively, coalesce, or extend, ultimately leading to the development of BSIs (3).

The formation of microdamage is influenced by the interplay between the number of loading cycles a bone endures and the magnitude and rate of tissue stresses and strains generated within the bone. This relationship is often described by an inverse power law, indicating that even small reductions in stress and strain can significantly increase the number of loading cycles a bone can withstand before fatigue failure. For instance, in the context of running loads, it’s estimated that a mere 10% decrease in tissue stress and strain can double the number of loading cycles to failure (4).

One effective strategy to reduce tissue stresses and strains under a given load is to develop stronger bones. Mechanically-induced bone adaptation is a key mechanism for achieving this. Our previous research using professional baseball players as a controlled model demonstrated that the humeral diaphysis adapts to throwing-related loading by nearly doubling in strength. This adaptation effectively maintains tissue strains below injury thresholds (5). In an animal model, we observed that even moderate (<10%) gains in bone mass induced by controlled loading resulted in an exponential (>100-fold) increase in bone fatigue resistance. This dramatic improvement is attributed to the reduced tissue-level strain experienced during each loading cycle (6).

Participation in multidirectional sports (MDS) during childhood and adolescence may be a crucial factor in building a more resilient skeleton and, consequently, reducing the risk of BSIs. Studies across both athletic and military populations have indicated that individuals with a history of playing ball sports exhibit a lower incidence of BSIs (7–9). The underlying hypothesis is that ball sports contribute to the development of a stronger skeletal structure (10). Ball sports such as basketball and soccer involve multidirectional movements and loading patterns, which introduce high-magnitude strains and rapid strain rates across various bone regions (11). Research has shown that basketball and soccer players possess enhanced skeletal properties (12, 13) and experience lower bone strains during physical activity compared to runners and sedentary controls (14, 15).

Female cross-country athletes are known to have the highest rates of BSIs among sports populations (16, 17). Compromised bone health may be a significant contributing factor. Studies have revealed that up to 40% of female adolescent cross-country runners have a z score below −1 for spine areal bone mineral density (aBMD) (18). Relative Energy Deficiency in Sport (RED-S), often resulting from insufficient caloric intake to meet the demands of training, is a known contributor to poor bone health in some athletes. However, an early specialization in running, often at the expense of participation in other sports, could also play a critical role. A study on female high school distance runners found that those who specialized early and intensely in distance running (running >9 months per year without participating in other sports) had lower spine aBMD compared to their counterparts who did not specialize to the same degree (19).

To further investigate the potential of ball sports in enhancing bone properties in female cross-country runners, our study compared collegiate-level runners who had participated in MDS during puberty with those who primarily engaged in running (along with swimming or cycling). The primary tool for assessment was high-resolution peripheral quantitative computed tomography (HR-pQCT), used to evaluate bone properties at the distal tibia and common BSI sites (diaphysis of the tibia, fibula, and 2nd metatarsal [MT]). We also assessed bone properties at specific high-risk BSI sites, including the base of the 2nd MT, navicular, and proximal diaphysis of the 5th MT. Injuries at these high-risk sites are particularly concerning due to their propensity for delayed or nonunion and the elevated risk of progressing to complete fractures (20).

METHODS

Participants

A convenience sample of non-pregnant female participants, aged 18 years and older, was recruited for this study. Participants were required to be active members of a Division I or II cross-country team within the National Collegiate Athletic Association (NCAA), and could also compete in track events of 1,500 meters or longer. We excluded athletes participating in shorter events as these may alter lower extremity loading distribution. Power athletes (sprinters, jumpers, hurdlers) and endurance athletes (middle and distance runners) exhibit different patterns of BSI incidence, with the former being more prone to foot BSIs and the latter to leg BSIs (21).

The study protocol was reviewed and approved by the Institutional Review Board of Indiana University, and all participants provided written informed consent prior to participation. Potential participants completed a screening questionnaire detailing their historical involvement in various sports. For each sport they had trained and/or competed in more than twice per month for at least 4 months per year, they were asked to specify the age(s) of participation, frequency (times per month), and duration (months per year).

Participants were categorized into the running group (RUN group) if their reported sports history included only cross-country, recreational running/jogging, swimming, and/or cycling. The running and MDS group (RUN+MDS group) included participants who, in addition to the activities listed for the RUN group, had a history of training and/or competing in either soccer or basketball at least twice per week for a minimum of 6 months per year for 5 or more years, starting before the age of 10. These criteria were designed to ensure: 1) MDS participation occurred during adolescence, a critical period for bone mechanoadaptation, and 2) participants had sufficient exposure to MDS in terms of frequency and duration.

Exclusion criteria for both groups were: 1) a history of participating in gymnastics more than twice per month for at least 4 months per year for more than 2 years. Gymnastics was excluded because gymnasts often have bilaterally enhanced radial bone health, which could confound our use of the non-dominant radius as a control site (22); 2) a history of BSI or fracture on both sides of the body in any of the bones to be imaged; 3) lower extremity surgery or immobilization for more than 2 weeks within the past 2 years; and 4) a known metabolic bone disease or systemic condition known to influence bone health.

Participant characteristics

We collected anthropometric data including height and weight, and gathered self-reported information on BSI history (number and locations), age of menarche, and personal best 5 km cross-country time. The Low Energy Availability in Females Questionnaire (LEAF-Q) was administered to assess for physiological symptoms indicative of low energy availability, such as gastrointestinal and menstrual function issues. A LEAF-Q score of 8 or higher was considered indicative of elevated risk (23). Body composition measures, including appendicular skeletal muscle mass relative to height (ASM/height2; kg/m2) and whole-body aBMD, were obtained using whole-body dual-energy x-ray absorptiometry (DXA) (Norland Elite; Norland at Swissray, Fort Atkinson, WI). Regional DXA scans were also performed to assess total hip and spine aBMD.

High-resolution peripheral quantitative computed tomography (HR-pQCT)

HR-pQCT scans were conducted using an XtremeCT II scanner (Scanco Medical, Bruttisellen, Switzerland) operating at 68 kVp and 1.47 mA with a voxel size of 60.7 μm. We imaged the radius, tibia/fibula, metatarsals (MTs), and navicular (Fig. 1). Scans were performed on the participants’ non-dominant arm and ipsilateral leg. In cases of a history of BSI or fracture in the bone to be imaged, the contralateral side was scanned instead. Participants were positioned supine, and the limb to be scanned was immobilized using a padded carbon fiber cast provided by the scanner manufacturer, anatomically shaped for the arm and leg. For MT and navicular scans, the foot was plantarflexed onto the leg cast, with a wedge placed under the forefoot to align the 2nd MT parallel to the scanner’s z-axis. The hip and knee were flexed to reduce stress on the anterior ankle while allowing the scanner gantry door to be lowered. Elastic straps around the lower leg and foot and bolsters under the thigh and leg provided additional stabilization and support.

Fig. 1.

Skeletal sites of interest included the fibular (A) and tibial diaphysis (B), distal tibia (C), base of the 2nd MT (D), navicular (E), proximal diaphysis of the 5th MT (F), and 2nd MT diaphysis (G). In each panel, the left image displays a representative HR-pQCT slice at the site of interest. The right image in each panel shows a representative 3D reconstruction of the analyzed slices. Panel E illustrates the slices used to identify the floor of the navicular facet (*) and ridges on the distal navicular at the navicular-cuneiform complex (arrows), which served as landmarks for defining the volume of interest. In panels D and E, the region of interest (ROI) manually traced within the cortical shell is indicated on the HR-pQCT slice.

For radial and tibial sites, scans comprised 168 slices (10.2 mm of bone length) and were acquired following established protocols (24). Reference lines were placed at the medial edge of the distal radius articular surface and the center of the tibia joint surface. Scan stacks were centered at 4% (distal radius) and 30% (radial diaphysis) of bone length proximal to the radius reference line, and 7.3% (distal tibia) and 30% (tibial diaphysis) of bone length proximal to the tibia reference line. Bone length measurements were taken in triplicate before scanning using a segmometer (Realmet Flexible Segmometer, NutriActiva, Minneapolis, MN). Fibula diaphysis outcomes were derived from the tibial diaphysis scans, as both bones are imaged simultaneously at this location.

Multi-stack scans were utilized for imaging the MTs and navicular. Up to nine stacks (each 168 slices or 10.2 mm bone length) were used to cover all 5 MTs. A reference line was positioned distal to the head of the 2nd MT (the most distal MT region), and stacks were imaged proximally to extend beyond the styloid process of the 5th MT (the most proximal MT region). For the navicular, three scan stacks were acquired proximally from a reference line at the navicular-cuneiform joint. All scans were evaluated for motion artifacts using a standard scale from 1 (no motion) to 5 (significant blurring and discontinuities). Only scans with a score of 3 or less were included in the analysis.

Reconstructed images of the distal radius and tibia, and diaphysis of the radius, tibia, fibula, 2nd MT, and proximal 5th MT were analyzed using the manufacturer’s standard protocol. The outer periosteal surface was semi-automatically contoured, and the inner endosteal surface was identified using the manufacturer’s automatically generated endocortical contour. Images were filtered with a low-pass Gaussian filter (sigma 0.8, support 1.0 voxel), and fixed thresholds were applied to segment trabecular and cortical bone (320 and 450 mgHA/cm3, respectively).

For the 2nd MT diaphysis, we analyzed 168 slices (10.2 mm bone length) spanning the midshaft. Bone length was defined as the distance between the most proximal and distal slices containing the 2nd MT base and head, respectively. For the proximal 5th MT, bone length was measured from the most distal and proximal slices containing the MT head and styloid process. One hundred slices (6.1 mm bone length) were analyzed, starting one-third of the 5th MT bone length distal to the tip of the styloid process (Fig. 1F). This region is located distal to the articulation between the 4th and 5th MTs and corresponds to Zone III of the 5th MT, a common site for BSIs (25).

At the distal radius and tibia (trabecular-rich sites with a cortical shell), we measured: total volumetric BMD (Tt.vBMD, mgHA/cm3) and area (Tt.Ar, mm2); cortical vBMD (Ct.vBMD, mgHA/cm3), area (Ct.Ar, mm2), and thickness (Ct.Th, mm); and trabecular vBMD (Tb.vBMD, mgHA/cm3), area (Tb.Ar, mm2), bone volume/total volume (Tb.BV/TV, %), thickness (Tb.Th, mm), number (Tb.N, 1/mm), and separation (mm).

At the radial, tibial, fibular, 2nd MT, and proximal 5th MT diaphysis (cortical-bone predominant sites), we measured: Ct.vBMD (mgHA/cm3), Tt.Ar (mm2), Ct.Ar (mm2), Ct.Th (mm), cortical porosity (Ct.Po, %), and minimum (IMIN, cm4), maximum (IMAX, cm4), and polar (IP, cm4) moment of inertias. Diaphyseal robustness was calculated as Tt.Ar divided by bone length (26), using tibia length as a surrogate for fibula length as fibula length was not directly measured.

Micro-finite element analysis (Scanco Medical FE software version 1.13) was utilized to estimate failure load (kN) at the radial and tibial sites, and the fibula and 2nd MT diaphyses. Each voxel within the segmented images was assigned a modulus of 10 GPa and Poisson’s ratio of 0.3. Axial compression was simulated, and failure load was defined as the load at which 5% of elements exceeded 1% strain (27). In addition to raw values, cortical, trabecular, and failure load outcomes at radial and tibial sites were converted to z scores using reference data obtained from the same HR-pQCT scanner (28).

Analyses at the base of the 2nd MT and navicular were limited to trabecular outcomes due to the minimal cortical bone at these locations. For the base of the 2nd MT, 100 slices (6.1 mm bone length) were analyzed, starting 5% of bone length from the most proximal slice containing the 2nd MT. Navicular outcomes were obtained from 100 slices (6.1 mm bone length) centered midway between the floor of the facet articulating with the talus head and the most distal navicular portion at the navicular-cuneiform complex. At both sites, a region of interest was manually traced inside the outer bone boundary every 10 slices (Fig. 1D, E). The region was then morphed on intervening slices and visually inspected. Images were filtered (sigma 0.8, support 1.0 voxel), and trabecular bone was segmented (threshold = 320 mgHA/cm3) to measure Tb.BV/TV, Tb.Th, Th.N, and Tb.Sp.

Short-term precision assessments were conducted on duplicate tibial (including fibula) and radial scans with repositioning in 15 individuals. Root mean square coefficients of variation (RMS-CV) were <0.6% for bone density (Tt.vBMD, Ct.vBMD, Tb.vBMD) and <0.8% for bone size (Tt.Ar, Ct.Ar, Tb.Ar) at both distal and diaphyseal sites, and 1.8–2.9% and <1% for estimated failure load at distal and diaphyseal sites, respectively. RMS-CVs for trabecular microarchitecture outcomes (Tb.BV/TV, Tb.N, Tb.Th, Tb.Sp) at distal sites ranged from 0.7–3.2%. Precision data for MT and navicular site outcomes are not currently available.

Statistical analyses

Statistical analyses were performed using two-tailed tests with a significance level of α=0.05 in IBM SPSS Statistics (v28; IBM Corporation, Armonk, NY). Participant demographics were compared between the RUN and RUN+MDS groups using unpaired t-tests for ratio data and Chi-squared tests for nominal data. Univariate general linear model analyses were used to compare DXA and HR-pQCT outcomes between groups. LEAF-Q score and whole-body lean mass were included as covariates in group comparisons for DXA and trabecular HR-pQCT outcomes (Tb.BV/TV, Tb.N, Tb.Th, Tb.Sp). In addition to LEAF-Q score and whole-body lean mass, bone length was included as a covariate in analyses of HR-pQCT measures of bone size (Tt.Ar, Ct.Ar, Tb.Ar, Ct.Th) and strength (IMIN, IMAX, IP, failure load).

RESULTS

Participant characteristics

We recruited 14 participants for the RUN group and 18 for the RUN+MDS group. The two groups were statistically similar in terms of age, age of menarche, LEAF-Q score, current and past menstrual function (assessed via LEAF-Q), age at which they started cross-country, total years competing in cross-country, and best 5 km cross-country performance (all p=0.38–0.99) (Table 1). The frequency of participants at risk of low energy availability (LEAF-Q score ≥8) was also similar in both groups (p=0.53). There was no significant difference in BSI history between the groups (p=0.26) or in the number of BSIs per athlete (p=0.14). Within the RUN+MDS group, participants played soccer (n=7), basketball (n=6), or both (n=5) for an average of 10.3±3.2 years (range = 5–14 years), starting at an average age of 7.7±2.4 years prior to menarche (range = 5–11 years) and continuing until age 16.1±2.5 years (range = 12–19 years).

Table 1.

Participant characteristicsa

Open in a new tab

aBMD = areal bone mineral density; BSI = bone stress injury; BV/TV = bone volume per tissue volume; HRpQCT = high-resolution peripheral quantitative computed tomography; LEAF-Q = Low Energy Availability in Females Questionnaire; MDS = multidirectional sport

aData are mean ± SD, except for BSI history and LEAF-Q score ≥8 [frequencies] and z scores [mean (95% confidence interval)]

bValues adjusted for LEAF-Q score and whole-body lean mass

cSelect outcomes at the distal radius. See Supplementary figures 1 & 2 for full data

dValues adjusted for LEAF-Q score, whole-body lean mass, and bone length

eSelect outcomes at the radial diaphysis. See Supplementary figures 3 & 4 for full data

fNormative reference data for this outcome is not available

*Significant difference between groups (p<0.05)

†Confidence interval does not cross zero

The RUN and RUN+MDS groups showed no significant differences in height, weight, BMI, whole-body lean mass, and percent fat mass (all p=0.09–0.84) (Table 1). However, DXA measurements revealed that whole-body and lumbar spine aBMD were significantly higher in the RUN+MDS group compared to the RUN group, by 5.8% and 8.0% respectively (all p≤0.05). No group differences were observed for hip areal BMD (p=0.31). Notably, the RUN group exhibited below-reference lumbar spine aBMD (z score < 0), whereas both groups had above-reference hip aBMD (z score > 0).

HR-pQCT outcomes at the distal radius and radial diaphysis showed no significant differences between the RUN and RUN+MDS groups (see Supplemental Figs. 1 and 2) (all p=0.10 to 0.98). Distal radius and radial diaphysis outcomes in both groups were generally comparable to reference values, with a few exceptions: lower Ct.Ar and failure load in the RUN group, and higher Ct.vBMD in the RUN+MDS group at the radial diaphysis (see Supplemental Fig. 3A, B).

Distal tibia

At the distal tibia, the RUN+MDS group demonstrated a 12.4% higher Tt.vBMD compared to the RUN group (p<0.01) (Fig. 2A), with a 15.9% increase in Tb.vBMD (p<0.01) (Fig. 2C). Cortical vBMD did not differ significantly between groups (p=0.74) (Fig. 2B). While overall size (Tt.Ar) of the distal tibia was similar between groups (p=0.41) (Fig. 2D), the RUN+MDS group had significantly greater cortical area (Ct.Ar, +17.1%, p=0.002) (Fig. 2E) and cortical thickness (Ct.Th, +15.8%, p=0.015) (Fig. 2G). Trabecular bone volume/total volume (Tb.BV/TV) was 14.6% greater (p=0.005) (Fig. 2H), and trabecular thickness (Tb.Th) was 8.3% greater (p=0.005) (Fig. 2I) in the RUN+MDS group. Trabecular number and spacing showed no significant differences (all p=0.53–0.63) (Fig. 2J, K). Critically, failure load at the distal tibia was 19.5% higher in the RUN+MDS group compared to the RUN group (p<0.001) (Fig. 2L).

Fig. 2.

Distal tibia properties in RUN and RUN+MDS. A) Total volumetric bone mineral density [Tt.vBMD]; B) cortical vBMD [Ct.vBMD]; C) trabecular vBMD [Tb.vBMD]; D) total area [Tt.Ar]; E) cortical area [Ct.Ar]; F) trabecular area [Tb.Ar]; G) cortical thickness [Ct.Th]; H) trabecular bone volume/tissue volume [Tb.BV/TV]; I) trabecular thickness [Tb.Th]; J) trabecular number [Tb.N]; K) trabecular spacing [Tb.Sp], and; L) failure load. Mean ± 95% confidence interval and individual participant data (open circles) are shown. *p<0.05 between groups.

HR-pQCT outcomes at the distal tibia in the RUN group were largely consistent with reference values, except for Tt.Ar (z score = −0.52; 95% confidence interval [CI] = −1.01 to −0.04) (see Supplemental Fig. 3C). In contrast, the RUN+MDS group showed elevated distal tibia HR-pQCT outcomes compared to reference values for Tt.vBMD, Tb.vBMD, Ct.Ar, Ct.Th, Tb.BV/TV, and Tb.Th. The z score for failure load at the distal tibia in the RUN+MDS group was +0.80 (95% CI = +0.43 to +1.16), indicating significantly enhanced bone strength.

Tibial diaphysis

No significant differences were found between the RUN and RUN+MDS groups in any HR-pQCT measures at the tibial diaphysis (all p=0.42–0.94) (see Supplemental Fig. 4). HR-pQCT outcomes at the tibial diaphysis were comparable to reference values across all measures, with the exception of Ct.Th, which was elevated in both the RUN (z score = +0.91; 95% CI = +0.51 to +1.31) and RUN+MDS (z score = +0.89; 95% CI = +0.45 to +1.32) groups (see Supplemental Fig. 3D).

Fibula diaphysis

The fibula diaphysis in the RUN+MDS group was 11.6% larger in total area (Tt.Ar) and had 12.0% greater cortical area (Ct.Ar) compared to the RUN group (all p<0.03) (Fig. 3B, C). The minimum moment of inertia (IMIN) at the fibula diaphysis was 28.5% greater in the RUN+MDS group (p=0.02) (Fig. 3F), while maximum moment of inertia (IMAX) and polar moment of inertia (IP) did not reach statistical significance (all p=0.06–0.13) (Fig. 3G, H). Cortical vBMD, Ct.Th, and Ct.Po at the fibula diaphysis showed no differences between groups (all p=0.22–0.95; Fig. 3A, D, E). The fibula diaphysis in the RUN+MDS group exhibited 8.4% greater robustness and 11.1% greater strength compared to the RUN group (all p≤0.03) (Fig. 3I, J).

Fig. 3.

Fibula diaphysis properties in RUN and RUN+MDS. A) Cortical volumetric bone mineral density [Ct.vBMD]; B) total area [Tt.Ar]; C) cortical area [Ct.Ar]; D) cortical thickness [Ct.Th]; E) cortical porosity [Ct.Po]; F) minimum second moment of area [IMIN]; G) maximum second moment of area [IMAX]; H) polar moment of inertia [IP]; I) robustness, and; J) failure load. Mean ± 95% confidence interval and individual participant data (open circles) are shown. *p<0.05 between groups.

Second metatarsal (2nd MT) diaphysis

The 2nd MT diaphysis in the RUN+MDS group had a 10.4% larger total area (Tt.Ar) and significantly greater cortical area (Ct.Ar, +18.7%) and cortical thickness (Ct.Th, +19.9%) compared to the RUN group (all p<0.05) (Fig. 4A, C, D). Maximum moment of inertia (IMAX) and polar moment of inertia (IP) at the 2nd MT diaphysis were 32.5% and 23.3% greater, respectively, in the RUN+MDS group (all p<0.04) (Fig. 4H & I). Cortical vBMD, Ct.Po, and IMIN at the 2nd MT diaphysis showed no differences between groups (all p=0.40–0.86) (Fig. 4A, E, F, G). The 2nd MT diaphysis in the RUN+MDS group was 9.2% more robust and 18.6% stronger than in the RUN group (all p<0.04) (Fig. 4I, J).

Fig. 4.

Properties of the 2nd MT diaphysis in RUN and RUN+MDS. A) Cortical volumetric bone mineral density [Ct.vBMD]; B) total area [Tt.Ar]; C) cortical area [Ct.Ar]; D) cortical thickness [Ct.Th]; E) cortical porosity [Ct.Po]; F) minimum second moment of area [IMIN]; G) maximum second moment of area [IMAX]; H) polar moment of inertia [IP]; I) robustness, and; J) failure load. Mean ± 95% confidence interval and individual participant data (open circles) are shown. *p<0.05 between groups.

High risk BSI sites

At high-risk BSI sites, the RUN+MDS group exhibited greater trabecular thickness (Tb.Th) compared to the RUN group at both the base of the 2nd MT (+5.3%) and the navicular (+10.1%) (all p≤0.02) (Table 2). The increased Tb.Th at the navicular contributed to an 8.9% higher trabecular bone volume/total volume (Tb.BV/TV) in the RUN+MDS group (p=0.01). The proximal diaphysis of the 5th MT in the RUN+MDS group showed significantly greater cortical area (Ct.Ar, +10.9%) and cortical thickness (Ct.Th, +11.6%) compared to the RUN group (all p≤0.02). Moments of inertia (IMIN, IMAX, IP) at the proximal diaphysis of the 5th MT did not differ significantly between groups (all p=0.06–0.09).

Table 2.

HRpQCT properties at selected high risk BSI sites in RUN and RUN+MDSa

Open in a new tab

aData are mean ± SD

bValues adjusted for LEAF-Q score and whole-body lean mass

cValues adjusted for LEAF-Q score, whole-body lean mass, and bone length

DISCUSSION

Our findings provide compelling evidence supporting the recommendation that runners should engage in ball sports during their youth to foster skeletal development and potentially mitigate the risk of BSIs (3, 10, 29). Female collegiate cross-country runners with a history of participating in MDS (soccer, basketball, or both) before and throughout puberty exhibited enhanced bone microarchitecture and greater strength at the distal tibia, as well as improved bone size and strength at the diaphysis of the fibula and 2nd MT, compared to runners who primarily focused on running (and low-impact sports like swimming or cycling). The observed strength enhancements ranged from an average of 11.1% at the fibula diaphysis to a substantial 19.5% at the distal tibia. Furthermore, those who played MDS also showed improvements in bone microarchitecture at high-risk BSI sites, specifically the base of the 2nd MT, navicular, and proximal diaphysis of the 5th MT. Notably, no benefits of MDS were observed on the properties of the tibial diaphysis.

Previous research in both military and athletic populations has consistently demonstrated that prior participation in ball sports is associated with a reduced risk of BSIs (7–9). In a study involving over 1,000 military recruits, Milgrom and colleagues (8) found that playing MDS (primarily basketball) for at least 2 years before infantry basic training more than halved the risk of developing a BSI. Similarly, Fredericson and colleagues (7) reported that participation in soccer or basketball reduced BSI risk by nearly half in a cohort of 274 elite runners. In both studies, the hypothesized mechanism was that ball sports enhanced skeletal health, although this was not directly tested. More recently, Rudolph et al. (30) reported that women who spent a greater proportion of their youth playing MDS had a lower risk of experiencing multiple BSIs. While our study did not find a statistically significant difference in BSI history between the groups, likely due to our limited sample size and statistical power for this outcome, our data strongly support the hypothesis that playing MDS during youth indeed enhances skeletal health—in our case, in females who later become collegiate-level cross-country runners.

The RUN+MDS group demonstrated enhanced whole-body and lumbar spine aBMD compared to the RUN group. Notably, the RUN group exhibited lower lumbar spine aBMD compared to reference data (z score < 0), potentially reflecting the systemic demands of distance running on energy availability and hormone levels. In our RUN group, 21% of participants (3 out of 14) presented with low bone mass (z score < −1), consistent with the 19% reported by Tenforde et al. (31) in a similar cohort. In contrast, none of the participants in the RUN+MDS group had low bone mass. Despite a similar proportion of participants in both groups being at risk of low energy availability (LEAF-Q score ≥ 8), the normalized lumbar spine aBMD to reference values in the RUN+MDS group, along with their greater whole-body aBMD compared to the RUN group, suggests that MDS participation during youth contributes to improved general bone health. These findings align with Rauh et al. (19), who reported that high school distance runners specializing intensively in running (distance running only for ≥9 months per year) were five times more likely to have low aBMD compared to low sport specializers (≥1 non-running sport and distance-running sport/s for ≤8 months per year). Lower whole-body bone mass and lumbar spine aBMD have been prospectively identified as predictors of BSI in female athletes (32).

Our HR-pQCT data from the distal tibia provide further support for the beneficial effects of MDS on bone health, particularly at trabecular-rich sites. The distal tibia in the RUN+MDS group showed a greater amount of trabecular bone (higher Tb.vBMD and Tb.BV/TV) with thicker trabeculae (greater Tb.Th) compared to the RUN group. Additionally, the RUN+MDS group exhibited enhanced cortical bone microarchitecture (greater Ct.Ar and Ct.Th). These microstructural advantages culminated in a 19.5% greater bone strength (failure load) at the distal tibia in the RUN+MDS group compared to the RUN group. The magnitude of strength enhancement observed in the MDS-RUN comparison is comparable to the 18.7% difference in distal radius bone strength between the racquet and non-racquet arms of collegiate tennis players, a within-subject controlled model that minimizes the influence of genetic and systemic factors (24). Further underscoring the benefits of MDS during youth, the RUN+MDS group had significantly higher Tt.vBMD, Tb.vBMD, Ct.Ar, Ct.Th, Tb.BV/TV, Tb.Th, and failure load at the distal tibia compared to reference data (z score > 0) (28), while no such differences were observed in the RUN group.

The distal tibia location assessed in our study is not typically considered a high-risk site for BSI, although high-risk BSIs of the medial malleolus do occur in the vicinity. However, when examining more BSI-prone sites, we observed significant benefits of MDS at the diaphysis of both the 2nd MT and fibula. Metatarsal BSIs account for nearly a quarter of all BSIs in female collegiate cross-country runners and over a third of all BSIs in collegiate athletes (17). Approximately 80% of MT BSIs occur in the 2nd and 3rd MTs (33), and biomechanical analyses suggest that bending strain during running is greatest in the 2nd MT compared to other MTs (34, 35). During the stance phase of running, loading of the 2nd MT primarily occurs in the dorsal-plantar plane around a medial-lateral axis located proximally in the bone. The proximal axis location is due to the relative fixation of the proximal 2nd MT within a mortise formed by the three cuneiforms and the 1st and 3rd MTs. This proximal fixation creates a cantilever effect during stance, resulting in compressive and tensile stresses on the dorsal and plantar surfaces of the diaphysis, respectively (36–39).

The loading mechanics of the 2nd MT during stance help explain the pattern of bone adaptation observed at the mid-diaphysis in the RUN+MDS group. The 2nd MT diaphysis in the RUN+MDS group was 10.4% larger (greater Tt.Ar), 19.9% thicker (greater Ct.Th), and exhibited 23.3% greater torsional resistance (greater IP) compared to the RUN group. Further analysis of the orthogonal axes of IMAX and IMIN revealed that the RUN+MDS group had a 32.5% greater IMAX at the 2nd MT diaphysis, but no significant group differences in IMIN. The IMAX plane, representing the plane of greatest bending resistance, within the 2nd MT diaphysis closely aligns with the plantar-dorsal plane (40). Thus, the predominant bone adaptation in the RUN+MDS group appeared to occur in the plane of maximal loading during stance. We also measured an 18.6% greater failure load at the 2nd MT diaphysis in the RUN+MDS group, although this was assessed under longitudinal compressive loading, not bending.

Similar to the findings at the 2nd MT, the fibula in the RUN+MDS group exhibited an 11.6% greater size (Tt.Ar) compared to the RUN group. The fibula is the third most common site for BSI in collegiate athletes, accounting for 9.7% of all BSIs (17). While the fibula bears less than 20% of the axial load on the leg (41), it serves as a crucial site for muscle attachments, and muscle forces are a primary driver of bone loading (42, 43). Studies suggest that runners exhibit limited adaptation at the fibula compared to controls, whereas soccer players show enhanced fibular bone mass and geometric properties (44). Our data support the notion that MDS participation may lead to a more robust fibula, better able to withstand loading and reduce the risk of subsequent BSIs.

The greater total area (Tt.Ar) observed at both the 2nd MT and fibula diaphyses in the RUN+MDS group is particularly noteworthy. Bone strength is proportional to the fourth power of the distance of bone material from the neutral axis. This principle implies that even small increases in bone size can result in disproportionately large gains in bone strength. Increased bone strength, in turn, reduces tissue-level stress and strain during each loading cycle, leading to an exponential increase in bone fatigue resistance (6). In the context of BSIs, bone geometry has been recognized as a critical determinant of MT stress during loading (45, with enhanced geometric properties of the 2nd MT helping to normalize stresses in response to increased external loads (38, 45, 46.

There is a broad consensus that the majority of gains in bone size associated with mechanical loading occur during the rapid bone modeling and periosteal bone apposition characteristic of skeletal immaturity (3), although bone size can still be influenced in skeletally mature individuals (47–50). The skeletal size adaptations induced during youth have the potential to persist long-term (5, highlighting the importance of encouraging MDS participation and delaying running specialization during growth.

Interestingly, we found no significant differences between the groups at the tibial diaphysis, the most frequent site for BSI in female cross-country runners (17). This finding was unexpected, as soccer and basketball players have been reported to have enhanced bone properties (12, 13) and reduced tibial strain per given load (14, 15). Our previous work has also demonstrated that multidirectional basketball activities load different regions of the tibial diaphysis (11, and HR-pQCT studies have revealed dominant-to-nondominant leg differences at the tibial diaphysis in collegiate tennis players (24). In the current study, both the RUN and RUN+MDS groups exhibited elevated cortical thickness (Ct.Th) at the tibial diaphysis compared to reference data (z score > 0), but this was the only measure showing differences. The reasons for the lack of group differences at the tibial diaphysis remain unclear. However, this absence of difference at the tibial diaphysis may strengthen the validity of the differences observed at other skeletal locations, suggesting that these are true effects of MDS participation and not simply due to general between-group variations in skeletal properties driven by genetics or systemic factors.

Our study has several notable strengths, including the use of advanced HR-pQCT technology to obtain detailed measures of bone microarchitecture and micro-finite element estimated strength. We also recruited relatively well-matched RUN and RUN+MDS groups in terms of BMI and potential for low energy availability, and included the radius as a control site. The absence of group differences at both the distal radius and radial diaphysis further supports that the lower extremity differences observed are not due to generalized skeletal enhancements. However, our study also has limitations, including its cross-sectional design, focus on female athletes only, relatively small sample size, micro-finite element strength modeling limited to axial compressive loading, and reliance on physical activity recall for group allocation.

CONCLUSIONS

In conclusion, our study demonstrates that female cross-country runners who participated in MDS during their younger years have enhanced lower extremity bone size, microarchitecture, and strength compared to runners who primarily engaged in running and low-impact sports. These group differences were evident at the distal tibia, common BSI sites (fibula and 2nd MT diaphyses), and select high-risk BSI sites (base of the 2nd MT, navicular, and proximal diaphysis of the 5th MT). However, no differences were observed at the tibial diaphysis. These findings strongly suggest that coaches, parents, and healthcare providers should encourage young athletes to delay specialization in running and promote participation in MDS, particularly during childhood and adolescence, to foster the development of a more robust skeleton and potentially prevent BSIs. While our study focused on basketball and soccer, it is plausible that other MDS, such as gymnastics, volleyball, and field hockey, could offer similar bone health benefits.

Supplementary Material

Supplemental Data File (.doc, .tif, pdf, etc.)_1SDC 1: Suppl_fig_1_distal_radius_graphs.pdf – Figure S1 – Distal radius properties in RUN and RUN+MDS

NIHMS1826153-supplement-Supplemental_Data_File___doc___tif__pdf__etc___1.pdf (129KB, pdf)

Supplemental Data File (.doc, .tif, pdf, etc.)_2SDC 2: Suppl_fig_2_radial_shaft_graphs.pdf – Figure S2 – HRpQCT properties of the radial diaphysis in RUN and RUN+MDS

NIHMS1826153-supplement-Supplemental_Data_File___doc___tif__pdf__etc___2.pdf (86.1KB, pdf)

Supplemental Data File (.doc, .tif, pdf, etc.)_3SDC 3: Suppl_fig_3_HRpQCT_z_graphs.pdf – Figure S3 – Z scores for HRpQCT properties of the distal radius, radial diaphysis, distal tibia, and tibial diaphysis in RUN and RUN+MDS

NIHMS1826153-supplement-Supplemental_Data_File___doc___tif__pdf__etc___3.pdf (479.5KB, pdf)

Supplemental Data File (.doc, .tif, pdf, etc.)_4SDC 4: Suppl_fig_4_tibial_shaft_graphs.pdf – Figure S4 – HRpQCT properties of the tibial diaphysis in RUN and RUN+MDS

NIHMS1826153-supplement-Supplemental_Data_File___doc___tif__pdf__etc___4.pdf (86.4KB, pdf)

Acknowledgements

This work was supported by grants from the National Institutes of Health (NIH/NIAMS P30 AR072581) and the Indiana Clinical Translational Science Award/Institute (NCATS UL1TR002529-01). The authors declare no financial conflicts of interest related to the content of this article. The findings of this study do not represent an endorsement by the American College of Sports Medicine. The research was conducted and reported with integrity, honesty, and adherence to ethical standards of data handling and presentation.

REFERENCES

[References from the original article – to be included here in the final output]

Associated Data

Supplementary Materials

Supplemental Data File (.doc, .tif, pdf, etc.)_1SDC 1: Suppl_fig_1_distal_radius_graphs.pdf – Figure S1 – Distal radius properties in RUN and RUN+MDS

NIHMS1826153-supplement-Supplemental_Data_File___doc___tif__pdf__etc___1.pdf (129KB, pdf)

Supplemental Data File (.doc, .tif, pdf, etc.)_2SDC 2: Suppl_fig_2_radial_shaft_graphs.pdf – Figure S2 – HRpQCT properties of the radial diaphysis in RUN and RUN+MDS

NIHMS1826153-supplement-Supplemental_Data_File___doc___tif__pdf__etc___2.pdf (86.1KB, pdf)

Supplemental Data File (.doc, .tif, pdf, etc.)_3SDC 3: Suppl_fig_3_HRpQCT_z_graphs.pdf – Figure S3 – Z scores for HRpQCT properties of the distal radius, radial diaphysis, distal tibia, and tibial diaphysis in RUN and RUN+MDS

NIHMS1826153-supplement-Supplemental_Data_File___doc___tif__pdf__etc___3.pdf (479.5KB, pdf)

Supplemental Data File (.doc, .tif, pdf, etc.)_4SDC 4: Suppl_fig_4_tibial_shaft_graphs.pdf – Figure S4 – HRpQCT properties of the tibial diaphysis in RUN and RUN+MDS

NIHMS1826153-supplement-Supplemental_Data_File___doc___tif__pdf__etc___4.pdf (86.4KB, pdf)