Targeted exercises against hip fragility
- First Online:
- Cite this article as:
- Nikander, R., Kannus, P., Dastidar, P. et al. Osteoporos Int (2009) 20: 1321. doi:10.1007/s00198-008-0785-x
- 365 Views
Compared to high-impact exercises, moderate-magnitude impacts from odd-loading directions have similar ability to thicken vulnerable cortical regions of the femoral neck. Since odd-impact exercises are mechanically less demanding to the body, this type of exercise can provide a reasonable basis for devising feasible, targeted bone training against hip fragility.
Regional cortical thinning at the femoral neck is associated with hip fragility. Here, we investigated whether exercises involving high-magnitude impacts, moderate-magnitude impacts from odd directions, high-magnitude muscle forces, low-magnitude impacts at high repetition rate, or non-impact muscle forces at high repetition rate were associated with thicker femoral neck cortex.
Using three-dimensional magnetic resonance imaging, we scanned the proximal femur of 91 female athletes, representing the above-mentioned five exercise-loadings, and 20 referents. Cortical thickness at the inferior, anterior, superior, and posterior regions of the femoral neck was evaluated. Between-group differences were analyzed with ANCOVA.
For the inferior cortical thickness, only the high-impact group differed significantly (~60%, p = 0.012) from the reference group, while for the anterior cortex, both the high-impact and odd-impact groups differed (~20%, p = 0.042 and p = 0.044, respectively). Also, the posterior cortex was ~20% thicker (p = 0.014 and p = 0.006, respectively) in these two groups.
Odd-impact exercise-loading was associated, similar to high-impact exercise-loading, with ~20% thicker cortex around the femoral neck. Since odd-impact exercises are mechanically less demanding to the body than high-impact exercises, it is argued that this type of bone training would offer a feasible basis for targeted exercise-based prevention of hip fragility.
KeywordsBone fragilityCortical boneExerciseFemoral neckMagnetic resonance imaging
The human skeleton is evolved for erect bipedal locomotion  and is particularly fit for long-distance running . Human bones are thus light to allow fast movement at low metabolic cost, stiff for proper leverage to locomotive muscle forces, and strong to bear habitual loads without breaking . As the bone tissue adapts, its structural design to accommodate prevailing mechanical loads [4, 5] the bone phenotype reflects both its structural rigidity and loading history. This being the case, studies of athletes provide useful ‘natural experiments’ to unravel the association of specific, long-term exercise-loading on bone structure and strength.
It was recently shown that thinning of the superolateral cortex at the femoral neck accounts for age-related hip fragility , and the anterior and antero-inferior cortices are particularly thin among hip fracture patients [7–10]. Regional age-related thinning of the femoral neck cortex has also been linked to reduced local mechanical stresses at old age in contrast to apparently higher bone-loading at young age [6, 9]. In their pivotal study , Mayhew et al. speculated whether the preservation of femoral neck cortical stability would require lifelong mechanical loading targeted to vulnerable cortical regions. Should such an effective and feasible targeted exercise against hip fragility be found, it would offer a cost-effective means to prevent hip fractures not only through bone-strengthening but also through improved muscle performance, balance, and, subsequently, reduced propensity to fall [11–13].
Very little is known, however, about the influence of specific, long-term exercise-loading on cross-sectional geometry of the femoral neck cortex . Virtually all such evidence [15, 16] has remained coarse and indicative because of the inherent inaccuracies and methodological limitations of the planar dual-energy X-ray absorptiometry (DXA) to describe three-dimensional bone geometry [17, 18].
In the present study, we assessed the proximal femur with three-dimensional magnetic resonance imaging (MRI), a method found recently useful for cortical bone evaluation [19–21]. In particular, we investigated whether the sports comprising mainly (1) high-magnitude vertical impacts, (2) moderate-magnitude impacts from rapidly varying, odd directions, (3) high-magnitude muscle forces, (4) low-magnitude impacts at high repetition rate, or (5) non-impact and non-weight-bearing muscle forces at high repetition rate were associated with thicker femoral neck cortex, and if so, to what extent and in which anatomic regions. Consequent to our tentative DXA-based hip structure analysis , we hypothesized that moderate-magnitude impacts from odd directions would be as beneficial in strengthening the femoral neck structure as the high-magnitude vertical impact-loading, known almost axiomatically osteogenic. We also hypothesized that the exercise-loading from odd directions is associated with thickened femoral neck cortex in anatomic regions considered crucial in terms of hip fragility [6–10].
Ninety-one adult female athletes competing actively at the national or international level and 20 non-athletic female referents participated in the study. The study protocol was approved by the Ethics Committee of The Pirkanmaa Hospital District, and each subject gave a written informed consent.
The athletes were recruited through national sports associations and local athletic clubs, and the referents were mainly students from the Pirkanmaa University of Applied Sciences. The athletes were nine triple-jumpers, ten high-jumpers, nine soccer players, ten squash players, 17 power-lifters, 18 endurance runners, and 18 swimmers. According to our previous protocol [15, 22], we classified these sports into five different exercise-loading types: high-impact (H-I); odd-impact (O-I); high-magnitude (H-M); repetitive, low-impact (L-I); and repetitive, non-impact (N-I) loading. The high-impact group comprised of the triple-jumpers and high-jumpers; the odd-impact group the soccer and squash players; the high-magnitude group the power-lifters; the repetitive, low-impact group the endurance runners; and the repetitive, non-impact group the swimmers.
Body height and weight were measured with standard methods in light indoor clothing without shoes. The body fat-% was assessed with DXA (GE Lunar Prodigy Advance, Madison, WI, USA). The training history was inquired via a recalled questionnaire including weekly sport-specific training hours and number of training sessions during at least the five preceding years. Medications, diseases, menstrual status, use of hormonal contraceptives, calcium intake, alcohol, tobacco, and coffee consumption, previous injuries and fractures were also inquired.
Muscle performance of lower extremities and functional agility were evaluated with standard methods [23, 24]. Maximal isometric leg extension strength was assessed at 90° knee flexion angle using a leg press dynamometer (Tamtron, Tampere, Finland). Dynamic maximal jumping power was assessed with a force-plate (Kistler Ergojump 1.04, Kistler Instrumente AG, Winterthur, Switzerland) using a counter-movement test. Functional agility was assessed via a figure-eight test by measuring the time for running two laps around two poles placed 10 m apart.
For clinical reference, conventional areal bone mineral density (BMD) was measured with DXA at the lumbar spine (vertebrae L1–L4) and femoral neck of the dominant side, while the primary bone analysis of this study was based on MRI of the femoral neck (see below). The dominant side denoted the lower limb primarily used for taking-off or kicking the ball, as appropriate. In our laboratory, the in vivo precision of the BMD measurement is ~1%.
For a mechanically less-loaded reference site, the non-dominant forearm was scanned with peripheral quantitative computed tomography (pQCT, Stratec XCT 3000, Stratec Medizintechnik GmbH, Pforzheim, Germany) . The non-dominant side denoted the hand not primarily used for throwing, holding a racquet, or writing, as appropriate. Trabecular density at the distal radius and cortical density at the radial diaphysis were evaluated as indices of trabecular architecture and cortical porosity, respectively. These traits are only marginally, if at all, associated with specific exercise-loading, even in the loaded forearms of squash players and weight-lifters [22, 26, 27]. In addition, the ratio of cortical to the total cross-sectional area was calculated as an index of cortical stability for both anatomic sites. The above pQCT traits were considered to provide reasonable reference data to assess the influence of potential genetic selection on bone structure of the present athletes. In our laboratory, the in vivo precision for the trabecular density is ~1%, for cortical density <1%, and for the cortical to total area ~4%.
The proximal femur cortical geometry was assessed with a 1.5-T MRI system (Siemens, Avanto Syngo MR B15, Erlangen Germany) in line with scanning principles found previously useful and precise . First, sagittal, axial, and coronal images of the pelvic region at the dominant side (the same side as with DXA) were obtained with two haste localization series, and these scout images were then used to specify the correct orientation of the imaging plane for the sequence taken for the femoral neck cortical evaluation. For this analysis, the plane of the femoral neck cross-sections was perpendicular to the oblique femoral neck axis. The used imaging sequence was a standardized axial T1-weighted gradient echo volumetric interpolated breath-hold (VIBE)-examination [FOV 35 × 26, TR 15.3 ms, TE 3.32 ms, slice thickness 1 mm without gaps, echo train length = 1, flip angle = 10°, matrix 384 × 288, the in-plane resolution (pixel size) 0.9 mm × 0.9 mm], covering the whole proximal femur from the femoral caput to the subtrochanteric level of the femoral diaphysis.
For the direction-specific analysis of cortical thickness, the femoral neck cross-section was divided into inferior, anterior, superior, and posterior regions (Fig. 1) according to principles described previously [7, 11]. The regional cortical thickness was calculated similarly to the mean cortical thickness by averaging ~30 to 45 thickness values in each sector.
All above-described segmentations of the MRI images were done blinded to the exercise-loading classification, and the measurements and calculations were done in Matlab environment (Math Works Inc., Natick, MA). The mean data from the two adjacent MRI slices were used in the statistical analysis.
Statistical analyses were done with SPSS 15.0 (SPSS Inc., Chicago, IL.). Means and standard deviation (SD) are given as descriptive statistics. Primary comparisons were made between the exercise-loading groups and the reference group. The between-group differences in bone traits were estimated by analysis of covariance (ANCOVA) using age, body height and weight as covariates. Sidak correction was used in the post-hoc tests to adjust for multiple comparisons. A p value of less than 0.05 was considered statistically significant.
Descriptive characteristics of exercise-loading and reference groups
Sports-specific training hours /week
Competing career (years)
Isometric leg press (kg)
Counter movement jump (W/ kg)
Figure-8 running (s)
At the non-dominant radius, trabecular and cortical densities in the exercise-loading groups did not differ significantly from the reference group (data not shown), the mean inter-group differences ranging from −3.9% (repetitive, low-impact group) to 7.2% (high-impact group) for trabecular density, and from −1.3% (high-magnitude group) to 0% (high-impact group) for cortical density. Further, the mean ratio of cortical to the total cross-sectional area was equal in the pooled exercise-loading group and the reference group: 0.32 at the distal radius in both groups, and 0.76 at the radial diaphysis in both groups.
Descriptive characteristics of the femoral neck cross-sectional geometry in the exercise-loading and reference groups
Total area, (mm2)
Cortical area, (mm2)
Cortical area/Total area
Antero-posterior diameter (mm)
Infero-superior diameter (mm)
Section modulus (mm3)
Descriptive characteristics of cortical thickness at the inferior, anterior, superior, and posterior regions of the femoral neck in the exercise-loading and reference groups
Cortical thickness (mm)
We demonstrated here that a specific long-term exercise-loading, particularly the one producing high-magnitude impacts or moderate-magnitude odd impacts to the hip, has considerable ability to make the femoral neck cortex thicker—not only at the primary weight-bearing inferior region but also at the other cortical regions considered critical in terms of hip fragility [6–10]. In contrast, high-magnitude exercise-loading was not at all associated with thicker femoral neck cortex at any region. This somewhat surprising finding may be explained by the fact that all movements during a typical power-lifting performance (e.g., a squat), while involving maximal muscle forces, are very slow by nature, keeping also the rate of loading low. During impact-exercises, the rate of loading is apparently much higher. Impact-loading is known to be associated with thickened cortical bone at peripheral bones, such as radius, humerus, and tibia [16, 22, 26], and in the present study, this was found to be the case for the femoral neck as well. Thus, it seems the impact-loading is the very type of exercise with potential to thicken the cortical bone in particular.
Given the influence of regional cortical thinning at the femoral neck on hip fragility, our observations have direct clinical relevance. Fall-induced hip fractures among elderly people have become a serious public health challenge with keen interest in finding feasible preventive strategies [11, 13]. Of all the modifiable risk factors for fragility fractures, regular physical activity is unique because it can strengthen both bones and muscles, improve balance and gait, and subsequently prevent falling , the predominant cause of hip fracture . Further pinpointing the potential of physical activity in preventing fractures, two recent large epidemiological cohort studies have estimated that if the aging population exercised briskly 3 to 4 h a week, the number of hip fractures could diminish by one third [29, 30]. Whether any specific type of exercise-loading would be particularly effective in reducing the incidence of fractures is yet hardly known. So far, only one cohort study of elderly former soccer and ice hockey players (N.B., odd-impact type of exercise-loading), who had stopped their active playing more than 30 years earlier, suggested that the incidence of fragility fractures was halved among athletes compared to their age-matched peers without athletic background .
Apparently, a fragile proximal femur can cope with normal living of an aged person , e.g., the mechanical demands caused by slow walking and daily chores. On the other hand, a sideways fall on the hip can impose a large stress on the femoral neck from such a direction the cortical bone is not typically adapted to, rendering the bone highly susceptible to fail [32, 33]. In the present study, we found that the odd-impact exercise-loading was associated with consistently ~20% thicker cortical bone around the femoral neck (Fig. 3). One may then speculate that in the long-term, these initially thicker cortices obtained in young adulthood may provide a vital reservoir of cortical bone to be lost with aging, and thus turn out be crucial in maintaining femoral neck stability at the vulnerable cortical regions [6–10]. Obviously, this calls for further evidence.
As to the strength of present evidence, the cross-sectional design is compromised by inability to show causal effect and possibility for selection bias. Basically, individuals with genetically strong musculature and skeleton can be physically more active and more prone to start an athletic career in their youth. On the other hand, the large mean differences in bone traits (~20% or more) observed in this study are difficult to explain by selection bias alone. Further, the mean trabecular density, cortical density, and the ratio of cortical bone to the total cross-sectional area in the non-dominant and less-loaded radius were similar in all exercise groups (including the referents). This implies that there were no substantial inborn group-differences between the studied bone traits, and the present findings can be taken as reflections of actual influence of specific exercise-loading on the femoral neck cortical geometry and thickness. It is also received that the strongest scientific evidence to verify the causal associations between the distinct exercise-loading types and femoral neck cortex (i.e., a randomized controlled exercise intervention trial) would be quite impossible to obtain. Such a study would have problems with sufficient sample size to provide adequate statistical power; with subject compliance and drop-outs; with several co-existing confounding factors; with precision of measurements vs. small expected treatment effects in cortical bone traits; and with long-enough intervention to see any treatment effects with confidence.
According to our findings, high-impact and odd-impact exercise-loadings appear to have a similar ability to thicken the femoral neck cortex—not only at the primary weight-bearing inferior region but also at the other regions, which are considered critical in terms of hip fragility [6–10]. As odd-impact exercises involve typically reaction forces of no more than approximately five times body weight [34, 35], they are not mechanically as demanding to the body as vertical high-impact loadings, which can produce peak reaction forces from approximately ten up to 20 times of body weight [16, 36]. In essence, the softer, mechanically less demanding odd-impact exercises may suit much better for many common people than vigorous high-impact exercises, and thus, feasible exercise regimens may be build on the principles of odd-impact loading. In such programs, the duration, frequency, and intensity of training could be customized to different age-groups with varying background in physical ability and interests [34, 37–42].
In conclusion, our study suggests that exercise regimens comprising moderate-magnitude impacts from varying, odd directions can provide a feasible basis for targeted exercise against hip fragility. Ball games, dancing, modified gymnastics, and aerobic exercises involving rapid turns and movements are practical examples of such odd-impact exercises.
We thank all the participants for their contribution and time, including Saturdays, for this study. The collaboration with the involved sports associations and sports clubs as well as with their coaches is greatly appreciated. We thank Ms Taru Helenius for managing the measurement schedule, Ms Ulla Hakala and Ms Ulla Honkanen for DXA and pQCT measurements at the UKK Institute, and Ms Arja Hilander and Ms Anu Kuhanen for the MRI measurements at the Tampere University Hospital. The help of Chief Physicist Pertti Ryymin from the Tampere University Hospital in defining the appropriate MRI sequence for this study is gratefully acknowledged. The financial support of the Medical Fund of the Pirkanmaa Hospital District, Finnish Ministry of Education, National Graduate School for Musculoskeletal Disorders and Biomaterials, and the Päivikki and Sakari Sohlberg Foundation is greatly appreciated.
Conflicts of interest