Fragile External Phenotype of Modern Human Proximal Femur in Comparison with Medieval Bone


  • Harri Sievänen ScD,

    Corresponding author
    1. Bone Research Group, UKK Institute, Tampere, Finland
    • Bone Research Group, UKK Institute, PO Box 30, FI-33501 Tampere, Finland
    Search for more papers by this author
  • László Józsa,

    1. Department of Morphology, National Institute of Traumatology, Budapest, Hungary
    Search for more papers by this author
  • Markku Järvinen,

    1. Division of Orthopaedics and Traumatology, Department of Trauma, Musculoskeletal Surgery and Rehabilitation, Tampere University Hospital and Medical School, University of Tampere, Tampere, Finland
    Search for more papers by this author
  • Tero A Järvinen,

    1. Division of Orthopaedics and Traumatology, Department of Trauma, Musculoskeletal Surgery and Rehabilitation, Tampere University Hospital and Medical School, University of Tampere, Tampere, Finland
    Search for more papers by this author
  • Pekka Kannus,

    1. Bone Research Group, UKK Institute, Tampere, Finland
    2. Division of Orthopaedics and Traumatology, Department of Trauma, Musculoskeletal Surgery and Rehabilitation, Tampere University Hospital and Medical School, University of Tampere, Tampere, Finland
    Search for more papers by this author
  • Teppo L Järvinen

    1. Division of Orthopaedics and Traumatology, Department of Trauma, Musculoskeletal Surgery and Rehabilitation, Tampere University Hospital and Medical School, University of Tampere, Tampere, Finland
    Search for more papers by this author

  • The authors state that they have no conflicts of interest.


Proximal femur macroanatomy of 118 medieval and 67 contemporary adults, 84 contemporary elderly, and 48 contemporary hip fracture cases was evaluated. Within ∼1000 years, the femoral neck axis has become longer, and its cross-section has become proportionally smaller and more oval in shape. These changes in the present external phenotype alone account for ∼50% higher fall-induced stress compared with the medieval situation.

Introduction: Bones, as whole skeletal structures, adapt to mechanical stresses they customarily experience. Because the present, mechanized lifestyle apparently deprives our skeletons of vigorous, habitual physical exertion, we studied whether the proximal femur phenotype has evolved vulnerable to fragility fractures by time.

Materials and Methods: Proximal femur macroanatomy of 118 medieval and 67 contemporary adults, 84 contemporary elderly, and 48 contemporary hip fracture cases was evaluated. Using direct measurements of external bone dimensions and geometric properties, we estimated the fall-induced stress as an index of hip fragility.

Results: Within ∼1000 years, the femoral axis length has become substantially longer (analysis of covariance, body height adjusted, p < 0.001), whereas the neck circumference has not increased. The macroanatomy was found similar between the contemporary adult and elderly groups. In hip fracture cases, however, the femoral axis length was further lengthened (p < 0.001), but the circumference was somewhat smaller (p = 0.001). Consequently, the estimated fall-induced stress can be ∼1.5-fold today compared with the medieval times (p < 0.001), and the secular trend seemed to be worse in women (sex-time interaction, p = 0.001).

Conclusions: The modern, relatively slender phenotype of the proximal femur alone seems to increase the fall-induced stress considerably, and when this phenotype coincides the osteoporotic, internally deteriorated femoral neck structure, fracture risk is imminent. This mechanically compromised external phenotype underscores the importance of timely strengthening of the skeleton and its regular maintenance throughout life.


Hip fractures cause excess mortality and morbidity, whereas the aging populations will probably increase the problem in the near future.(1,2) The risk of hip fracture is attributable to aging as a physiological process and to concomitant complex interaction between osteoporosis, physical capability, various diseases, living habits, and environmental factors; falling on the hip is the major cause of hip fractures among the old population.(3,4)

Bones, as skeletal structures, adapt to mechanical stresses they customarily experience, and this adaptation occurs both during the lifetime of an individual and over evolutionary time.(5–7) The sedentary lifestyle of modern generations resting on ubiquitous automation and mechanization has apparently deprived our skeletons of versatile and vigorous, habitual physical exertion,(8) and possibly sculptured the phenotype of functionally vital proximal femur vulnerable to fall-induced fractures. In this respect, skeletal remains provide useful material to get insight into the skeletal adaptation under different living circumstances.(9)

Regarding the hip fragility, it is not simply the reduced bone mass or density but rather the external anatomic appearance of the proximal femur(10,11) and gradual deterioration of its internal cortical structure and trabecular architecture(12–17) that fundamentally count. On the other hand, the age-related bone loss affecting adversely the internal bone structure seems to be opposed by expansion of external bone dimensions so that the habitual load stresses are not essentially increased.(18) To study the secular trends in the external phenotype of the proximal femur, we assessed the macroanatomy of this bone region in medieval and present adult populations lived ∼1000 years apart and also in contemporary elderly people with and without hip fractures.

Table Table 1.. Proximal Femur Macroanatomy Among 20- to 50-Year-Old Medieval and Contemporary Groups
original image


Medieval proximal femora were obtained from 10th to 13th century Hungarian cemeteries and were comprised of well-preserved, nonfractured bones from both sides. The age, sex, and body height of the medieval subjects were determined using the Acsádi and Nemeskéri scheme,(19) which uses the sex-specific stature estimation equations by Dupertuis and Hadden.(20) Contemporary nonfractured femora from adult and elderly people were obtained from autopsies of otherwise healthy Hungarian people who had died immediately in an accident or within 1 week thereafter. No musculoskeletal or endocrinological diseases (except osteoporosis among the elderly) were found in clinical and pathological examinations of these people. The body height was determined at autopsy.

The medieval material was comprised of 55 men and 63 women, 20–50 years of age, and the contemporary sample was made up of 41 men and 26 women, representing the same age range (Table 1). These two temporally separate samples of adult bones were considered to provide adequate information on the typical proximal femur phenotype among healthy 20- to 50-year-old, ethnically similar populations that lived in the same geographic region some 1000 years apart. The contemporary elderly material was comprised of 23 men and 61 women (Table 2).

The femora of the contemporary hip fracture cases (only cervical fractures were included) were obtained from autopsies of 15 men and 33 women (Table 2) who had died within 2 weeks after hip fracture. The material was comprised of the contralateral, nonfractured femora only.

For the macroanatomic description of the excised proximal femora, femoral axis length (FAL, from the tip of femoral head to the lateral aspect of femur along the neck axis), femoral neck length (FNL, the length of the anatomically distinct narrow neck region), superoinferior neck width (FNW) and circumference (FNC) at the narrowest section of the neck, and the neck-shaft angle between long axes of oblique femoral neck and femoral shaft (NSA) were measured directly with a caliper, tape measure, or angle-meter, as appropriate. The mean of left and right side results was used, and the hip fracture cases were excluded, for which the results from the nonfractured side were used. All measurements were done by one experienced observer.

Assuming a general cylindrical shape for the femoral neck cross-section and using the FNW and FNC data, the orthogonal, anteroposterior thickness (FNT) was estimated, and the total cross-sectional area (FNCSA) was calculated as πFNW·FNT/4. Furthermore, we evaluated the morphology of the femoral neck cross-section using the eccentricity as a measure of how much the given cylindrical cross-section deviated from being circular. The eccentricity was calculated by taking a square root of the following expression [1 – min(FNW; FNT)2/max(FNW; FNT)2]; a zero value denotes a perfect circle.

In biomechanical terms, the proximal femur can be approximated as a cantilevered beam.(21) Using the measured and calculated femoral dimensions in a standard engineering equation,(22) and (1) assuming that a typical worst-case impact force caused by a sideways fall onto the greater trochanter can be about eight times body weight (8 · BW),(23) (2) assuming conservatively that the body mass index (BMI) of all subjects was the same at 25 kg/m2 (it thus follows that BW = BMI · height2), and (3) the internal bone structure (i.e., cortical thickness and trabecular architecture) and bone material properties were similar in all groups, the estimate of fall-induced maximal net stress (Sfall) comprising both the bending and axial (compressive) stress components was calculated as follows:

equation image

where FNZmin is the minimum section modulus (index of bending strength, proportional to the third power of the narrowest diameter) of the given cylindrical cross-section.

Table Table 2.. Proximal Femur Macroanatomy Among Contemporary Elderly and Fracture Groups
original image

Mean and SD are given as descriptive statistics. Differences in the bone variables between the medieval, contemporary, and fracture groups were assessed by two-way analysis of covariance (ANCOVA); group and sex were used as factors. Because of substantial differences in mean body heights between different groups, the body height was used as a covariate in all analyses to keep the statistical treatment of data consistent.

First, the medieval and contemporary adult groups were compared to reveal secular (∼1000 years) trends in traits describing the proximal femur macroanatomy. Then, the contemporary adult and elderly groups were compared as to whether the proximal femur macroanatomy was similar between these temporally near (∼40 years) groups. Finally, the contemporary elderly and fracture groups were compared as to whether some macroanatomic traits were particularly augmented among the fracture group.


Descriptive data of the proximal femur macroanatomy in different study groups are given in Tables 1 and 2. As expected, the dimensions in the male proximal femur were significantly (height-adjusted p < 0.001) larger compared with female bones in general, whereas the femoral neck shape (eccentricity) and NSA values were similar between sexes.

Within the studied span of ∼1000 years, FAL has become considerably longer in both sexes: ∼13% in men and ∼19% in women (height-adjusted p < 0.001; Fig. 1A). Of note, the secular trend seemed to be more pronounced in women (height-adjusted, sex-group interaction, p = 0.054). In the Middle Ages, the narrow neck region (FNL/FAL) took up about one quarter of the femoral axis length, but today it seemed to consistently occupy almost one third of this region (height-adjusted p < 0.001). In contrast, neither FNC nor FNCSA has increased since the Middle Ages (Fig. 1B; Table 1). However, the morphology of the femoral neck cross-section seemed to have undergone significant secular shaping (height-adjusted p < 0.001 for eccentricity) similarly in both sexes. The contemporary cross-sectional shape is more oval, as can be judged from significant trends in the FNW and FNT data (height-adjusted p = 0.006 and p = 0.008, respectively). The NSA showed a consistent ∼4° secular increase among both sexes (height-adjusted p < 0.001).

Compared with their ∼40 year younger counterparts, the contemporary elderly people had a similar proximal femur macroanatomy (Figs. 1A and 1B; Tables 1 and 2), with the cross-sectional shape excluded (height-adjusted p = 0.008 for eccentricity; Table 1). Accordingly, the mean femoral neck shape among the elderly appeared to be even more oval than in the adult group.

Compared with the contemporary elderly group, the hip fracture cases had ∼10% longer FAL (height-adjusted p < 0.001; Fig. 1A; Table 2) and FNL (height-adjusted p < 0.001) but somewhat smaller FNC (height-adjusted p = 0.001; Fig. 1B) and FNCSA (height-adjusted p = 0.050). Of note, the smaller femoral neck size seemed to concern men in particular (height-adjusted sex-time interaction, p = 0.014 for FNC). The FNL/FAL ratio and the femoral neck morphology among the hip fracture cases were not statistically different from the data observed in the contemporary elderly group. In contrast, NSA among the hip fracture cases seemed to be significantly higher (height-adjusted p = 0.003), and again, the increased angle concerned men in particular (height-adjusted sex-group interaction, p = 0.020).

Figure Figure 1.

Differences in (A) femoral axis length (FAL) and (B) circumference of the femoral neck (FNC) in medieval and contemporary groups. Bars represent mean and whiskers represent SD. The statistical significances of between-group differences are given in Tables 1 and 2.

In concert with what could be expected from the observed secular trends in the proximal femur macroanatomy, fall-induced maximal net stress (Sfall) among contemporary adult people was 1.3- to 1.5-fold in comparison with their medieval peers in a similar standing height fall onto the hip (height-adjusted p < 0.001; Fig. 2), the secular increase being higher in women than in men (height-adjusted sex-group interaction, p = 0.001). Between the contemporary elderly and adult groups, there was also a borderline (height-adjusted p = 0.06; Table 1) difference in Sfall, apparently caused by elderly women's slightly longer FAL compared with younger women's FAL (height-adjusted sex-group interaction, p = 0.06).The Sfall among the contemporary hip fracture cases did not differ significantly (p = 0.15) from the elderly group (Fig. 2; Table 2). The difference in Sfall was only ∼10% for women but was ∼40% for men, being in line with the above described anatomic differences between elderly and fractured men (Table 2).

Figure Figure 2.

Estimated fall-induced maximal stress (Sfall) affecting the femoral neck in medieval and contemporary groups normalized to the average level of medieval men. Bars represent mean and whiskers represent SD. The statistical significances of between-groups differences are given in Tables 1 and 2.


The external phenotype of the proximal femur among contemporary healthy adult people seems to have become longer but not respectively wider in all directions, manifesting as a more oval cross-sectional shape. The present phenotype is distinct from the more robust medieval proximal femur. Although some lengthening of the femoral axis has been observed in modern female populations only four decades apart,(24,25) the disproportionate increase of the narrow neck region accompanied by a change in cross-sectional shape among both sexes, as indicated by our data, has not been shown before. Compared with a mere lengthening of the femoral axis, the disproportionately longer narrow part of an already longer lever arm declines the mechanical capacity of the proximal femur to cope with unusual external impact loading caused by a fall.(5) This is particularly so if the cross-sectional dimensions of the narrow neck region had not proportionally increased and thus reduced the incident stress within the bone structure. Indeed, this seemed to be the case (Figs. 1B and 2.).

From the locomotive aspect(21)—and in line with bone's established ability to adapt to prevalent loading milieu(6,7)—the contemporary phenotype of proximal femur is fit for stresses occurring mainly in the superoinferior direction (corresponding to FNW), which is those taking place during common daily locomotion, such as walking and occasional running, but not necessarily for multidirectional and vigorous physical exertion that could have been regularly encountered during heavy physical work and other activities in the Middle Ages. Accordingly, the structural design of the contemporary proximal femur, given the oval cross-section and long narrow neck region, is not appropriate for impact loading from an unusual direction (e.g., the one occurring during a sideways fall on the posterolateral side of the greater trochanter).(26,27) A more circular cross-section with proportionally wider diameters, the predominant phenotype of the proximal femur in the medieval population, is not so distinctly adapted to any specific type of loading or direction but basically represents an efficient geometry to cope with any loading.

A long femoral axis is known to predispose to hip fractures.(10,11) It is recalled that the external macroanatomy of the proximal femur, the modest increases (∼0.5 mm/decade) in cross-sectional diameters excluded,(28) does not change with age after cessation of longitudinal growth. Nevertheless, our contemporary people with cervical hip fractures not only had a longer femoral axis but also a smaller femoral neck cross-section than their contemporary elderly counterparts. This finding underpins the obvious reason for the increased fragility of the contemporary proximal femur—a relatively long and thin femoral neck region within an initially long bone structure. In fact, a lower risk of hip fracture among different contemporary populations is associated with a shorter femoral neck.(29,30)

What is behind the relatively slender phenotype of the proximal femur among contemporary people compared with people who lived in the Middle Ages? Obviously, definite answers cannot be given. However, a period of ∼1000 years is too short for systematic evolutionary changes in genetics. There is evidence from recent prehistory and the last 2000 years that the adult height in many groups has been periodically equal to modern humans of the same region.(31) Thus, the observed differences in the stature and proximal femur macroanatomy may be taken as reflections of prevalent conditions during growth and function in general, those pertaining to secular improvements in nutrition, general health, and well being, and the progressive reduction in overall physical activity in modern developed societies.(8)

The contemporary people extensively use automation and other labor-saving devices in industry, transport, agriculture, and at home. Undoubtedly, the intensity, amount, and frequency of hard physical work have dramatically reduced by time. Whereas the good nutrition and health care have created optimal conditions for growth—unleashing the “hidden” genetic potential for axial growth of the narrow neck region at the proximal femur, the prevalent physical exertion has not provided sufficiently large and versatile osteogenic stimuli so that all dimensions of the femoral neck cross-section would have enlarged in proportion to the lengthened functional lever arm. The 4° increase in the NSA observed in this study may be regarded as a compensatory adaptation to reduce the functional lever arm and thus the body weight–induced bending moment at erect walking position during locomotion.(21) In line with this observation, higher NSAs have been generally linked to sedentary existence and mechanization among recent human populations.(32) In this respect, these findings of increased NSA and smaller femoral neck cross-section among the men with cervical fractures are noteworthy.

The proximal femur macroanatomy, the cross-sectional dimensions excluded, is fundamentally shaped to its final form during adolescence, and this process is largely under genetic control.(33,34) Nevertheless, there is still sufficiently room for exercise-induced modulation of the femoral neck structure,(35) a potential that may turn out to be crucial in reducing hip fragility in later life. Bone accrual can be best enhanced during the growing years with specific exercises,(36,37) and this opportune time to lay a strong foundation for mechanically competent bone structures for later life should not be missed. Evolution has basically equipped us with an elegant and efficient locomotive, musculoskeletal apparatus,(21,38) and we should use it properly.

The exclusive erect, bipedal locomotion, originating at more than three millions years ago, accounts for the well-known, heterogeneous cortical bone distribution within the femoral neck,(20,39,40) naturally evident at present.(12,14,17) This anatomic fact can be deleterious at old age when one falls onto the hip, and the proximal femur becomes impacted from an unfavorable direction.(17,26) More likely so, the external phenotype of the proximal femur represents the contemporary slender appearance, a structural design that can inherently augment fall-induced stress. Our finding that the secular augmentation seemed to concern more women than men may be explained in part by different hip anatomy, but apparently the influence of sex-specific variation in physical activity and work over time probably plays a role also.(41)

Besides the robust external phenotype of medieval proximal femur, there is evidence that trabecular bone architecture in medieval women and men has been better conserved until old age among both sexes, probably because of a higher level of physical activity, being in contrast to prevalent osteoporosis and bone fragility among the present elderly population.(42) In our study, we could not take age-related osteoporosis, usually manifesting as deteriorated trabecular architecture (thinning of trabeculae and loss of connectivity), locally thinned cortical wall, and increased local cortical porosity,(12–17,43) into account. Nevertheless, it is obvious that when the modern, slender external phenotype coincides the internally fragile, osteoporotic proximal femur, the risk of a fall-induced hip fracture must be multiple compared with the medieval, robust external phenotype.

This study has limitations, and prudence should be exercised when interpreting the results. Despite the relatively large total sample size representing the same ethnic origin, the subgroups remained relatively small and may not be representative of the target populations generally. We did not know the actual stature of the medieval people. Although the bone dimensions are known to scale with each other(44) and with body height, there can be a substantial amount of individual variation in skeletal traits at any given height, and the relationships are not necessarily linear. Further complicating matters, the relative proportions of body segment lengths (e.g., trunk versus legs) may have undergone secular changes,(45) rendering the stature estimates subject to bias. In this study, all macroanatomic traits were statistically adjusted for body height to assure consistent treatment of data from groups with clearly different body heights. Given the adjustment, however, the inaccuracy of height estimation may be crucial and contribute to false conclusions. The reconstruction equations of stature are known to be compromised by several inconsistencies (typical accuracy is ±3 cm), and those by Dupertuis and Hadden(19) are argued to overestimate the body height.(46) Thus, regarding our medieval proximal femora, the mean estimated fall-induced stress, the summary measure of the loaded bone structure, may have been overestimated. Also, we assumed a similar BMI in the medieval and contemporary samples to get a compatible estimate of body mass for the stress estimation in each group. This assumption contradicts the prevailing notion that the medieval people were leaner and lighter than the contemporary people. Altogether, it seems that the fall-induced stress was overstated among the medieval people, and the magnitude of secular trend may have been even more substantial. On the other hand, it is stressed that the actual in vivo mechanical loading caused by a fall is a much more complex issue than could be grasped with the present simple approach. The incident load imposed on the femoral neck, possibly breaking the bone, is influenced by many nonskeletal factors, including the individual falling mechanics (e.g., direction, rate, protective movements before impact)(23,47,48) and landing surface characteristics.

In this study, both the medieval and contemporary data were statistically analyzed in a similar and conservative fashion using the appropriate 2 >2 factorial design, and quite consistent findings, indicating a clear secular mechanical compromise in the external phenotype of proximal femur, were obtained. Naturally, given several assumptions and simplifications, these findings are regarded as suggestive, and further complementary data are needed.

In conclusion, the external phenotype of the modern proximal femur seemed to be relatively slender compared with the medieval phenotype. It is particularly noteworthy that the present external phenotype alone may engender a considerably increased fall-induced stress, and when it coincides with the osteoporotic, internally deteriorated femoral neck structure, fracture risk is imminent. This mechanically compromised, external phenotype underscores the importance of timely strengthening of the skeleton and its regular maintenance throughout life.


This study was supported in part by grants from the Medical Research Fund of Tampere University Hospital, the Research Council for Physical Education and Sports, Ministry of Education, and the AO Research Fund, Switzerland.