Effect of daily physical activity on proximal femur

Clinical Biomechanics 21 (2006) 1–7 www.elsevier.com/locate/clinbiomech

Clinical Biomechanics Award 2004

Effect of daily physical activity on proximal femur Timo Ja¨msa¨

a,*

, Aki Vainionpa¨a¨ a,b,c, Raija Korpelainen Juhani Leppa¨luoto b

c,d

q

, Erkki Vihria¨la¨

a,e,1

,

a

d

Department of Medical Technology, University of Oulu, P.O. Box 5000, FI-90014 Oulu, Finland b Department of Physiology, University of Oulu, P.O. Box 5000, FI-90014 Oulu, Finland c Department of Sports Medicine, Deaconess Institute of Oulu, P.O. Box 5000, FI-90014 Oulu, Finland Department of Public Health and General Practice, University of Oulu, Oulu University Hospital, P.O. Box 5000, FI-90014 Oulu, Finland e Optoelectronics and Measurement Laboratory, University of Oulu, P.O. Box 5000, FI-90014 Oulu, Finland Received 28 September 2005; accepted 6 October 2005

Abstract Background. The incidence of osteoporotic fractures is increasing and has become one of the major health problems in developed countries. Physical exercise has been found to be effective in the prevention of osteoporosis. However, the optimal amount of exercise is not known. The aim of this study was to examine the association between the intensity of physical activity and bone mineral density at the proximal femur, using long-term quantification of daily physical activity. Methods. The study subjects were 64 women (age 35–40 years), who carried an accelerometer-based body movement recorder for 12 months for individual quantification of their daily physical activity. The average distribution of daily accelerations was defined using 33 acceleration levels. Findings. A significant relationship between physical activity data and proximal femur bone mineral density was found. Physical activity that induced acceleration levels exceeding 3.6 g correlated positively with the bone mineral density change at the proximal femur, the association being strongest at the femoral neck at 5.7 g (r = 0.416, P = 0.001). Interpretation. The association between physical activity and changes in proximal femur bone mineral density was dependent on the acceleration level of exercise. The quantity and quality of exercise can be monitored with the accelerometer-based physical activity monitor, and the method might be used for optimizing exercise for prevention of osteoporosis. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: Accelerometry; Body movement monitor; High-impact exercise; Osteoporosis; Physical activity; Prevention

1. Introduction Osteoporosis and osteoporotic fractures have become one of the major health problems in Western countries (Cummings and Melton, 2002; Kannus et al., 1995).

q The study was presented in part at the European Society of Biomechanics Conference in Ôs-Hertogenbosch, the Netherlands, in July 7, 2004. * Corresponding author. E-mail address: timo.jamsa@oulu.fi (T. Ja¨msa¨). 1 Current address: Newtest Ltd., Oulu, Finland.

0268-0033/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.clinbiomech.2005.10.003

The ageing of populations will further increase the health hazards and economic costs due to osteoporotic fractures. Although therapeutic effects of medical treatment have been shown (Bone et al., 2004; NIH, 2001), there is an urgent need for preventive strategies at the population level. Physical exercise has been found to be effective in the prevention of osteoporosis (NIH, 2001; Wallace and Cumming, 2000). It has beneficial effects on bone by increasing the peak bone mass in younger age groups, reducing age- and menopause-related bone loss in middle-aged populations, and decreasing the incidence of

2

T. Ja¨msa¨ et al. / Clinical Biomechanics 21 (2006) 1–7

falls in the elderly (Campbell et al., 1997; Greendale et al., 1995; Gregg et al., 2000; Heinonen et al., 1996; NIH, 2001). Meta-analyses have indicated that exercises including impact loading would be the most beneficial for bones, and high-impact activities most effective in improving femoral neck bone mineral density (BMD) (Wallace and Cumming, 2000; Wolff et al., 1999). High-impact exercise-induced benefits on BMD have also been shown to be maintained after the intervention (Kontulainen et al., 2004). Bone adaptation is a function of mechanical loading, and bone remodelling is influenced by the level and distribution of daily mechanical strain (Rubin and Lanyon, 1985; Wolff, 1892). The response of bone to mechanical stimuli is assumed to be threshold driven (Frost, 1987; Hsieh et al., 2001). Therefore, the loading level of exercise should be important. There are several studies on high-impact exercise interventions, but there are only a few studies that have estimated loadings (Bassey and Ramsdale, 1995; Bassey et al., 1998; Heinonen et al., 1996; Kemmler et al., 2004a). These estimations have been based on calculations of the ground reaction forces (GRF) of single jumps, or parts of regimens. The variation of loads in these jumping regimens has been equivalent to 2–6 times the body weight per jump. However, GRF measurements are limited to a fixed place and time. Accelerometer-based measurement of movement is an accepted method for monitoring physical activity with reasonable reproducibility (Eston et al., 1998; Janz, 1994; Menz et al., 2003; Servais et al., 1984). Accelerations during exercise have also been shown to relate with impact load forces (Janz et al., 2003; Servais et al., 1984). Thus, it might be supposed that high peak accelerations are associated with high peak strains. Continuous long-term measurements during exercise and normal life have not been performed so far, and there is no quantitative data to suggest the degree of intensity and amount of exercise needed for strengthening bone. Here we studied for the first time the association between the intensity of physical activity and proximal femur BMD, using long-term quantification of daily physical activity in a population-based intervention study.

2. Methods 2.1. Study design We performed a 12-month population-based randomized controlled intervention study, described previously in more detail (Vainionpa¨a¨ et al., 2005), to evaluate the effects of high-impact physical exercise on bone. In the previous article, we presented the BMD changes at different skeletal sites for the control group

and the exercise group, and the exercise effect was shown by the study group assignment. During the intervention study, the subjects wore an activity monitor to record their daily physical activity. Here we present the physical activity recording method and the association between the actual physical activity and changes in proximal femur BMD. 2.2. Study subjects The study was performed with 64 premenopausal women (age 35–40 years), who completed the original intervention study (Vainionpa¨a¨ et al., 2005), and in whom the BMD measurements and physical activity recordings were available. Half of the original 120 subjects were assigned to an exercise group, while the others continued their normal life. Thirty-four of the 64 women were included in the original training group, and 30 of the women in the control group. The training intervention consisted of a 12-month high-impact exercise program, which has been described in more detail previously (Vainionpa¨a¨ et al., 2005). Briefly, the training sessions were carried out three times a week for 12 months. Each workout lasted for 60 min and included step patterns, stamping, jumping, running and walking. Additionally, the participants were asked to train for 10 min daily at home. The subjects in the control group were asked to continue their normal daily lives. The study protocol was approved by the institutional ethical committee, and all participants gave an informed written consent. The procedure of the study was in accordance with the Declaration of Helsinki. 2.3. Measurement of physical activity Accelerometers have been widely used for recording physical activity. Accelerometric measurement can be considered as an indirect method for recording the intensity of impact loading. In this study, the accelerometer was worn close to the right iliac crest to have an estimate for the loading at the hip (Fig. 1). Previously, we tested the precision and accuracy of the accelerometer-based method using a three-dimensional prototype (Vihria¨la¨ et al., 2003). It was used for our preliminary measurements to estimate the acceleration levels at different exercise patterns and for validating the activity monitor used in the present intervention study. Briefly, the device consisted of three accelerometers (SCA-320, VTI Technologies, Vantaa, Finland), connected orthogonally, and a data logger (Tattletale Model 8v2 Data Logger, Onset Computer Corp., Bourne, Massachusetts, USA) with a sampling rate of 400 Hz. The data reduction method was adopted to maintain the principal characteristics of each acceleration peak. After data reduction, the device was able to record individual impact peaks continuously for a cou-

T. Ja¨msa¨ et al. / Clinical Biomechanics 21 (2006) 1–7

3

Fig. 2. The response of an accelerometer (upper) and a force plate (lower) during a jump. The first lower peak is for take-off and the second, the sharp peak for landing. Preliminary study, sampling frequency 400 Hz.

Fig. 1. Measurement of accelerations at the hip.

ple of days. The average peak amplitude error of the device was less than 2%, evaluated on a vibration tester by comparison with a calibrated industrial reference sensor (Vihria¨la¨ et al., 2003). In our preliminary study with 10 women (age 20–58 years, BMI 19.1–29.7 kg/m2), we estimated typical vertical acceleration levels attained in different exercise patterns, using the prototype described above. The acceleration of gravity (1 g) was subtracted, so that a value of zero corresponded to standing (=zero impact). Each exercise pattern was measured a total of 8 times for each subject during three measurement sessions. The peak values were predominantly obtained immediately after the heel contact (Fig. 2). The average peak accelerations (range of individual means) in the vertical direction were 0.8 g (0.7–1.2 g), 3.2 g (2.0–4.6 g) and 4.2 g (2.9–5.2 g) in walking at 5 km/h, running at 9 km/h and running at 13 km/h, respectively, measured on a treadmill. The average peak values for aerobic stepping, lateral jumping, counter-movement jumps, jumps without counter-movement, and drop jumps from 40 cm were 1.2 g (0.9–2.2 g), 2.0 g (1.1–3.3 g), 4.4 g (1.9– 6.5 g), 4.6 g (2.1–6.4 g) and 5.6 g (3.8–9.9 g), respectively, measured on a standard vinyl floor covering. All measurements were performed using light gymnastic shoes. The reproducibility error, given as the root-meansquare coefficient of variation (CVRMS), was 4.0%. The peak acceleration values had a high correlation (r = 0.989; n = 572 recordings) with the values obtained simultaneously using a standard optical motion analysis system (MacReflex, Qualisys Ab, Gothenburg, Sweden), operating at a rate of 100 frames/s (Vihria¨la¨ et al., 2004),

showing that the method accurately measures the local acceleration at the hip. The acceleration values also had a significant correlation with the GRF (r = 0.735 for the peak acceleration values, r = 0.937 for the area under the acceleration peaks; n = 462 recordings), measured with a force plate (Kistler 9287A with a Kistler 9865C charge amplifier, Kistler Instrumente AG, Winterthur, Switzerland), when the acceleration values were multiplied by body weight. In the present intervention study, we recorded the vertical acceleration peaks with a one-dimensional accelerometer-based human body movement monitor (Newtest Ltd.,Oulu, Finland). It was designed for long-term monitoring of physical activity, recording the histogram of daily number of impact peaks, classified according to peak acceleration, to describe the intensity of exercise. The continuous recording time was several weeks. The monitor was worn on a belt close to the right iliac crest. The physical activity monitor was validated against the prototype in simultaneous measurements during exercise training. For the individual quantification of their daily physical activity, all subjects of the training intervention study were asked to carry the monitor on their waist daily, during all waking hours, i.e. from morning to evening, for 12 months. The monitor gathered the data at the sampling rate of 400 samples per second, filtered, pre-analyzed and classified according to peak acceleration. The classified data were transferred into a server computer approximately every second week. The number of daily acceleration peaks (impacts) was analyzed at the 33 acceleration levels from 0.3 to 9.9 g as given by the activity monitor, 0 g corresponding to standing (acceleration of gravity subtracted), resulting in a 33-level histogram of impacts according to their peak acceleration value. The individual daily average number of impacts at each acceleration level was calculated for the analysis.

4

T. Ja¨msa¨ et al. / Clinical Biomechanics 21 (2006) 1–7

2.4. Bone measurements

3. Results

Areal bone mineral density (BMD, g/cm2) was measured at the left proximal femur with dual-energy X-ray absorptiometry (DXA) (Hologic Delphi QDR, Bedford, Massachusetts, USA). The femoral neck, trochanter, and WardÕs triangle regions were analyzed separately. All scanning and analyses were done by the same operator. The scanner was calibrated daily by bone phantoms for quality assurance. All measurements were performed at the beginning and at the end of the intervention.

The physical activity monitor used in the current study was validated against the previously validated prototype. In simultaneous measurements during exercise training, the agreement between the physical activity monitor and the prototype was good (r = 0.971; n = 41 subjects). We recorded approximately a total of 150 million acceleration peaks during the intervention study. Physical activity data from 64 out of 80 subjects who completed the original intervention study were available, showing that 80% of the participants were compliant to use the monitor. The day-by-day compliance for the physical activity monitor was not controlled, but the daily average values of the days, when the monitor was in use, were analyzed. The individual average number of daily accelerations was calculated at different peak acceleration levels. Daily average distributions of the acceleration peaks at different acceleration levels are shown in Fig. 3, indicating that the subjects in the exercise group, on average, had a higher number of impacts at all acceleration levels. The individual distributions were normalized relative to the mean values of the control subjects. Fig. 4A shows an example of a very active person with high relative number of acceleration peaks at the levels of 4 g or more, while her number of accelerations at lower levels is normal. Fig. 4B represents a physically inactive person with a low relative number of accelerations at all levels. The results of the BMD measurements by study group assignment have been published elsewhere (Vainionpa¨a¨ et al., 2005) and summarized in Table 1. Here, the BMD change at the proximal femur appeared to correlate significantly with accelerations exceeding 3.6 g,

2.5. Statistical analysis The individual daily average numbers of acceleration peaks were defined at 33 acceleration levels. The average daily numbers were normalized relative to the corresponding mean values of the controls. PearsonÕs correlation coefficients were used to study the association between the relative number of acceleration peaks at each level and the percentage BMD changes. Thirtythree analyses were performed, separately for each of the histogram levels, having one BMD value (12-month change) and 33 numbers of impacts at 33 acceleration levels for each subject. This multiphase procedure was used to find out for the first time the effect of different activity levels on bone. Partial correlation was used to control for the influence of weight change on the correlation coefficients. Control and exercise groups were pooled for the correlation analysis. Values of P < 0.05 were considered statistically significant. The data were analyzed using the SPSS statistical package (SPSS 11.5 for Win).

10000 Exercise group

1000 Daily number

Control group

100

10

1

0.1

1

2

3

4

5

6

7

8

9 10

Acceleration level (g) Fig. 3. Distribution of the daily average number of impacts at different acceleration levels.

T. Ja¨msa¨ et al. / Clinical Biomechanics 21 (2006) 1–7 0.5

10

A P = 0.001

8

Correlation coefficient

Relative daily number

9

7 6 5 4 3 2 Mean of the controls

1 0

1

2

3

4

5

6

7

8

9 10

0.4 P = 0.01

0.3 P = 0.05

0.2

0.0

Ward's triangle Trochanter

1

2

3

4

5

6

7

8

9

10

Acceleration level (g)

Fig. 5. Correlation coefficient r between physical activity and BMD change at the femoral neck as a function of acceleration level. The P levels of the statistical significance are shown as dash lines.

10

Relative daily number

Femoral neck

0.1

Acceleration level (g)

9

5

B

8 7

BMD in premenopausal women. The study showed that physical activity that included high accelerations was positively associated with the BMD changes, while the activity at lower intensity did not correlate with the BMD changes at the proximal femur. In the present study we used continuous long-term accelerometer-based monitoring of physical activity. There are several methods of analyzing accelerometric data; for example root-mean-square average, power spectrum integral and acceleration count have been previously used as measures of activity (Auvinet et al., 2002a; Janz et al., 2004; Kavanagh et al., 2004; Menz et al., 2003; Servais et al., 1984). At the vertical direction, the highest acceleration has been shown to be mainly related to the heel strike (Auvinet et al., 2002a,b; Bussmann et al., 2000; Derrick, 2004; Kavanagh et al., 2004; Menz et al., 2003), which was also found in our preliminary measurements. Here we used the daily number of vertical acceleration peaks at different acceleration levels to assess the intensity of exercise. The acceleration levels were also found to correlate with ground reaction forces during different exercise patterns. We used a multiphase procedure with 33 correlation

6 5 4 3 2 1 0

Mean of the controls

1

2

3

4

5

6

7

8

9 10

Acceleration level (g) Fig. 4. Distribution of the daily average number of accelerations normalized relative to the mean values of the control subjects. (A) An example of a very active person with high relative number of impacts at acceleration levels of 4 g or more, while her number of impacts at lower levels is normal. (B) A physically inactive person with low relative number of impacts at all acceleration levels.

the association being strongest at the femoral neck at the level of 5.7 g (r = 0.416, P = 0.001) (Fig. 5).

4. Discussion We studied here the relationship between the intensity of physical activity and change in proximal femur Table 1 Bone mineral density and body composition at baseline and at 12 months Control group (n = 30) Baseline 2

Femoral neck BMD; g/cm Trochanter BMD; g/cm2 WardÕs triangle BMD; g/cm2 Body weight, kg Fat mass, kg Lean body mass, kg

0.807 (0.109) 0.700 (0.089) 0.700 (0.114) 67.3 (12.9) 20.7 (8.6) 45.2 (4.8)

At 12 months

% Change

0.804 (0.106) 0.702 (0.085) 0.704 (0.110)

0.2 (1.9) 0.4 (2.0) 0.9 (4.0)

67.1 (13.2) 21.5 (8.4) 44.1 (6.7)

P#

Exercise group (n = 34) Baseline 0.789 (0.096) 0.696 (0.095) 0.685 (0.100) 67.6 (11.9) 20.6 (8.3) 46.0 (4.9)

At 12 months

% Change

0.798 (0.092) 0.702 (0.095) 0.702 (0.104)

1.2 (2.6)* 0.9 (2.2)* 2.5 (4.6)**

67.2 (12.1) 22.1 (9.2) 45.3 (4.6)

Values are mean (SD). NS = statistically non-significant, BMD = bone mineral density. * P < 0.05 annual change within the group. Paired samples t-test. ** P < 0.01 annual change within the group. Paired samples t-test. # P values for differences between the control group and the exercise group over the 12-month study period. Independent samples t-test.

0.017 NS NS NS NS NS

6

T. Ja¨msa¨ et al. / Clinical Biomechanics 21 (2006) 1–7

analyses, separately for each of the 33 histogram levels, to find out the effect of different activity levels on bone. Combination algorithms for a more common view of the physical activity level will be under study in the future. In previous intervention studies, the assessment of exercise intensity has been based on the ground reaction forces of model performances. High-impact exercise resulted in positive exercise effects on femoral neck and lumbar spine BMDs in a study, where the peak forces of the exercise regimen varied between 2.1 and 5.6 times the body weight (Heinonen et al., 1996). A recent report from an exercise trial in early postmenopausal women with osteopenia also showed that supervised training twice a week and additional home training were enough to maintain, but not to improve lumbar spine and hip BMD. The average peak GRFs in different exercise regimens in that trial were between 1.8 and 3.5 times the body weight (Kemmler et al., 2004a,b). The intensity of skeletal exercise has been suggested to be defined by the loads applied to the bone (Turner and Robling, 2003). More specifically, Ôhigh intensityÕ has been defined as mechanical loading forces greater than four times body weight, Ômoderate intensityÕ as two to four times body weight and Ôlow intensityÕ as less than two times body weight (Witzke and Snow, 2000). Our findings are complementary to these previous results, giving information on the acceleration levels for exercise. Here, low- or moderate-intensity impact training below the acceleration level of 4 g was not equally effective for the proximal femur as high-impact exercise including accelerations of 4 g or more. More specifically, we reported net accelerations, where zero impact was adjusted to 0 g. Thus, the total loading affecting the bone is still 1 g higher when the acceleration of gravity is included. However, any comparison of the results of the previous studies with our data is difficult due to different method for assessing the intensity of exercise. We used an accelerometric sensor that was attached on the subject. The accelerations measured in a moving coordinate system differ from the values in the global one. Thus, the recordings can not be compared directly to ground reaction forces. However, the moving sensor may give even more accurate local loading estimates than GRF measurements do. Here the monitor was worn on a belt close to the right iliac crest to have an estimate for the loading at the hip. Recently, it was shown that impacts from atypical loading directions can be equally effective as highimpact loading in improving the strength of the femoral neck (Nikander et al., 2005). Here we measured only the vertical component of the acceleration. It remains to be seen if more accurate results might be found by using three-dimensional recording of accelerations. It has been shown that the osteogenic response saturates during prolonged exercise (Rubin and Lanyon,

1984; Umemura et al., 1997). Thus, load-induced bone formation is improved by rest periods, and osteogenic effectiveness is improved by adding more exercise sessions, rather than lengthening the duration of individual sessions (Turner and Robling, 2003). In the present study, we had up to three training sessions per week with additional daily home program. Here the physical activity monitor recorded the daily number of impacts, but it was not able to differentiate the rest periods. More studies are needed to find out the effect of splitting the exercise into multiple sessions. In conclusion, the changes in proximal femur BMD were associated with the acceleration level of physical exercise. The quantity and quality of exercise can be monitored with the accelerometer-based physical activity monitor, and the method might be used for optimizing exercise for prevention of osteoporosis.

Acknowledgements The authors would like to express their special thanks to Minna Tervo, the physiotherapist in our study team, for supervising the training and testing of the subjects. We thank Pentti Nieminen, Ph.D., for statistical advice, Mikko Ma¨a¨tta¨ for assistance with figures and the staff of the Department of Sports Medicine at Oulu Deaconess Institute for their assistance. The study was supported by the National Technology Agency of Finland; Newtest Ltd., Oulu, Finland; CCC Group, Oulunsalo, Finland; Fastrax Ltd., Vantaa, Finland; the Juho Vainio Foundation; the Research Foundation of the Institutes of Sports and the Finnish Foundation for Sports Research.

References Auvinet, B., Berrut, G., Touzard, C., Moutel, L., Collet, N., Chaleil, D., Barrey, E., 2002a. Reference data for normal subjects obtained with an accelerometric device. Gait Posture 16, 124–134. Auvinet, B., Gloria, E., Renault, G., Barrey, E., 2002b. RunnerÕs stride analysis: comparison of kinematic and kinetic analyses under field condition. Sci. Sports 17, 92–94. Bassey, E.J., Ramsdale, S.J., 1995. Weight-bearing exercise and ground reaction forces: a 12-month randomized controlled trial of effects on bone mineral density in healthy postmenopausal women. Bone 16, 469–476. Bassey, E.J., Rothwell, M.C., Littlewood, J.J., Pye, D.W., 1998. Preand postmenopausal women have different bone mineral density responses to the same high-impact exercise. J. Bone Miner. Res. 13, 1805–1813. Bone, H.G., Hosking, D., Devogelaer, J.P., Tucci, J.R., Emkey, R.D., Tonino, R.P., Rodriguez-Portales, J.A., Downs, R.W., Gupta, J., Santora, A.C., Liberman, U.A.for the Alendronate Phase III Osteoporosis Treatment Study Group, 2004. Ten yearsÕ experience with alendronate for osteoporosis in postmenopausal women. N. Engl. J. Med. 350, 1189–1199, Comment on N. Engl. J. Med. 351, 190–192.

T. Ja¨msa¨ et al. / Clinical Biomechanics 21 (2006) 1–7 Bussmann, J.B., Damen, L., Stam, H.J., 2000. Analysis and decomposition of signals obtained by thigh-fixed uni-axial accelerometry during normal walking. Med. Biol. Eng. Comput. 38, 632–638. Campbell, A.J., Robertson, M.C., Gardner, M.M., Norton, R.N., Tilyard, M.W., Buchner, D.M., 1997. Randomised controlled trial of a general practice programme of home based exercise to prevent falls in elderly women. Br. Med. J. 315, 1065–1069. Cummings, S.R., Melton, L.J., 2002. Epidemiology and outcomes of osteoporotic fractures. Lancet 359, 1761–1767. Derrick, T.R., 2004. The effects of knee contact angle on impact forces and accelerations. Med. Sci. Sports Exerc. 36, 832–837. Eston, R.G., Rowlands, A.V., Ingledew, D.K., 1998. Validity of heart rate, pedometry, and accelerometry for predicting the energy cost of childrenÕs activities. J. Appl. Physiol. 84, 362–371. Frost, H.M., 1987. Bone ‘‘mass’’ and the ‘‘mechanostat’’: a proposal. Anat. Rec. 219, 1–9. Greendale, G.A., Barrett-Connor, E., Edelstein, S., Ingles, S., Haile, R., 1995. Lifetime leisure exercise and osteoporosis. The Rancho Bernardo study. Am. J. Epidemiol. 141, 951–959. Gregg, E.W., Pereira, M.A., Caspersen, C.J., 2000. Physical activity, falls, and fractures among older adults: a review of the epidemiologic evidence. J. Am. Geriatr. Soc. 48, 883–893. Heinonen, A., Kannus, P., Sieva¨nen, H., Oja, P., Pasanen, M., Rinne, M., Uusi-Rasi, K., Vuori, I., 1996. Randomised controlled trial of effect of high-impact exercise on selected risk factors for osteoporotic fractures. Lancet 348, 1343–1347. Hsieh, Y.F., Robling, A.G., Ambrosius, W.T., Burr, D.B., Turner, C.H., 2001. Mechanical loading of diaphyseal bone in vivo: the strain threshold for an osteogenic response varies with location. J. Bone Miner. Res. 16, 2291–2297. Janz, K.F., 1994. Validation of the CSA accelerometer for assessing childrenÕs physical activity. Med. Sci. Sports Exerc. 26, 369–375. Janz, K.F., Rao, S., Baumann, H.J., Schultz, J.L., 2003. Measuring childrenÕs vertical ground reaction forces with accelerometry during walking, running, and jumping: The Iowa Bone Development Study. Pediatr. Exerc. Sci. 15, 34–43. Janz, K.F., Burns, T.L., Levy, S.M., Torner, J.C., Willing, M.C., Beck, T.J., Gilmore, J.M., Marshall, T.A., 2004. Everyday activity predicts bone geometry in children: the Iowa bone development study. Med. Sci. Sports Exerc. 36, 1124–1131. Kannus, P., Parkkari, J., Niemi, S., 1995. Age-adjusted incidence of hip fractures. Lancet 346 (8966), 50–51. Kavanagh, J.J., Barrett, R.S., Morrison, S., 2004. Upper body accelerations during walking in healthy young and elderly men. Gait Posture 20, 291–298. Kemmler, W., Lauber, D., Weineck, J., Hensen, J., Kalender, W., Engelke, K., 2004a. Benefits of 2 years of intense exercise on bone density, physical fitness, and blood lipids in early postmenopausal osteopenic women: results of the Erlangen Fitness Osteoporosis Prevention Study (EFOPS). Arch. Int. Med. 164, 1084–1091. Kemmler, W., von Stengel, S., Beeskow, C., Pintag, R., Lauber, D., Weineck, J., Hensen, J., Kalender, W., Engelke, K., 2004b. Umsetzung moderner trainingswissenschaftlicher Erkenntnisse in

7

ein knocheanaboles Training fu¨r fru¨h-postmenopausale Frauen. Die Erlanger Fitness Osteoporose Pra¨ventions Studie (EFOPS). Osteologie 13, 1–13. Kontulainen, S., Heinonen, A., Kannus, P., Pasanen, M., Sieva¨nen, H., Vuori, I., 2004. Former exercisers of an 18-month intervention display residual aBMD benefits compared with control women 3.5 years post-intervention: a follow-up of a randomized controlled high-impact trial. Osteoporos. Int. 15, 248–251. Menz, H.B., Lord, S.R., Fitzpatrick, R.C., 2003. Acceleration patterns of the head and pelvis when walking on level and irregular surfaces. Gait Posture 18, 35–46. NIH Consensus Development Panel on Osteoporosis Prevention Diagnosis and Therapy. Osteoporosis prevention, diagnosis, and therapy, 2001. JAMA 285, 785–795. Nikander, R., Sieva¨nen, H., Heinonen, A., Kannus, P., 2005. Femoral neck structure in adult female athletes subjected to different loading modalities. J. Bone Miner. Res. 20, 520–528. Rubin, C.T., Lanyon, L.E., 1984. Regulation of bone formation by applied dynamic loads. J. Bone Joint Surg. 66A, 397–402. Rubin, C.T., Lanyon, L.E., 1985. Regulation of bone mass by mechanical strain magnitude. Calcif. Tissue Int. 37, 411–417. Servais, S.B., Webster, J.G., Montoye, H.G., 1984. Estimating human energy expenditure using an accelerometer device. J. Clin. Eng. 9, 159–171. Turner, C.H., Robling, A.G., 2003. Designing exercise regimens to increase bone strength. Exerc. Sport Sci. Rev. 31, 45–50. Umemura, Y., Ishiko, T., Yamauchi, T., Kurono, M., Mashiko, S., 1997. Five jumps per day increase bone mass and breaking force in rats. J. Bone Miner. Res. 12, 1480–1485. Vainionpa¨a¨, A., Korpelainen, R., Leppa¨luoto, J., Ja¨msa¨, T., 2005. Effects of high-impact exercise on bone mineral density: a randomised controlled trial in premenopausal women. Osteoporos. Int. 16, 191–197. Vihria¨la¨, E., Saarimaa, R., Myllyla¨, R., Ja¨msa¨, T., 2003. A device for long term monitoring of impact loading on the hip. Mol. Quant. Acoust. 24, 211–224. Vihria¨la¨, E., Oksa, J., Karkulehto, J., Korpelainen, R., Myllyla¨, R., Ja¨msa¨, T., 2004. Reliability of an accelerometry in the assesment of body movements. Technol. Health Care 12, 122–124. Wallace, B.A., Cumming, R.G., 2000. Systematic review of randomized trials of the effect of exercise on bone mass in pre- and postmenopausal women. Calcif. Tissue Int. 67, 10–18. Witzke, K.A., Snow, C.M., 2000. Effects of plyometric jump training on bone mass in adolescent girls. Med. Sci. Sports Exerc. 32, 1051– 1057. Wolff, J., 1892. Das Gesetz der Transformation der Knochen. Translation by Maquet, P., Furlong, R., 1986: The Law of Bone Remodelling. Springer, Berlin. Wolff, I., van Croonenborg, J.J., Kemper, H.C., Kostense, P.J., Twisk, J.W., 1999. The effect of exercise training programs on bone mass: a meta-analysis of published controlled trials in pre- and postmenopausal women. Osteoporos. Int. 9, 1–12.