Prof. Jon Tobias's presentation from Osteoporosis 2016: Day-to-day levels of high impact physical activity are positively related to lower limb bone strength in older women: findings from a population based study using accelerometers to classify impact magnitude. Find out more at: https://nos.org.uk/conference

1. Professor Jon Tobias
Academic Rheumatology
Musculoskeletal Research Unit
University of Bristol
National Osteoporosis Society 2016
1
Day-to-day levels of higher impact physical activity are
positively related to lower limb bone strength in older women:
findings from a population based study using accelerometers to
classify impact magnitude

2. Background
2
• Exercise trials suggest that high impacts are beneficial for skeletal health in
postmenopausal women (1)
• Cross sectional studies using accelerometers suggest that day-to-day exposure to
high impacts is also positively related to hip bone mineral density (BMD) in
adolescents (2) and premenopausal women (3)
• Equivalent studies have not been performed in postmenopausal women, including
older individuals at risk of osteoporosis
1. M Martyn-St James, BJSM 2009, 43:898
2. K Deere et al JBMR 2012, 27:1887)
3. A Vainionpaa et al, OI 2006, 17:455

3. 1. To establish whether day-to-day exposure of older women to high vertical impacts,
assessed by seven day accelerometer recordings, is positively related to lower limb
bone strength
2. To determine whether any relationship with bone strength we find is explained by
relationships with BMD or bone size
3. To explore whether exposure to vertical impacts are related to bone turnover
markers, and whether these might also contribute to observed associations with
bone strength
3
Aims

4. 4
Bristol
Cohort for skeletal health in Bristol and Avon (COSHIBA): Unselected community-based study population
65-80 year old women recruited to
COSHIBA from GP practices (2007-09)
N= 3200
Invited to research clinic
(2015) N= 1064
Attended study clinic session
(2015) N= 463

5. 5
VIBE Clinic Assessments
•Accelerometry
•DXA (total body, lumbar spine & hip)
•pQCT (mid-tibia & distal radius)
•Bone formation (P1NP) and resorption (β-CTX) markers (from fasting blood samples)
•Short physical performance battery
•Grip strength
•Jumping mechanography

6. 6
VIBE Clinic Assessments
•Accelerometry
•DXA (total body, lumbar spine & hip)
•pQCT (mid-tibia & distal radius)
•Bone formation (P1NP) and resorption (β-CTX) markers (from fasting blood samples)
•Short physical performance battery
•Grip strength
•Jumping mechanography

7. 7
Seven day accelerometer recording
• Light weight (~55g) accelerometer worn in custom belt
for seven days, set to 50Hz sampling rate

8. 8
Seven day accelerometer recording
• Light weight (~55g) accelerometer worn in custom belt
for seven days, set to 50Hz sampling rate
• Seven day accelerometer trace was subsequently
down-loaded

9. 9
Seven day accelerometer recording
• Light weight (~55g) accelerometer worn in custom belt
for seven days, set to 50Hz sampling rate
• Seven day accelerometer trace was subsequently
down-loaded
• Y axis ‘peaks’ identified and categorised into low (0.5-
1.0g), medium (1.0-1.5g) and higher (>1.5g) impacts*
*K Deere et al, JAPA 2016 24:290

10. 10
Linear regression analysis
•Exposures
• Number of low, medium or higher impacts/week (log transformed)
•DXA outcomes
• Total hip BMD
• Femoral neck BMD
• Hip cross sectional moment of inertia (CSMI)
•Mid-tibial pQCT outcomes
• Periosteal circumference
• CSMI
•Adjustment
• Age, height, fat mass, lean mass, activity in other bands

11. 11
Characteristics of participants in final dataset (n=408)
Mean SD
Age (years) 76.8 3.0
Height (cm) 158.8 6.2
Weight (kg) 68.3 12.0
Accelerometer wear time (valid days) 5.4 1.5
Median 25th
Centile 75th
Centile
Low impacts (0.5-1.0g)/week 8809 4047 16882
Medium impacts (1.0-1.5)/week 345 99 764
Higher impacts (≥1.5g)/week 42 17 106

12. 12
Impacts vs Hip DXA
CSMI = cross sectional moment of inertia (mm4
)
Beta shows SD change in DXA outcome per doubling in number of impacts per week, adjusted for
age, height, fat mass, lean mass, activity in other bands
Low impacts Medium impacts Higher impacts
Beta lower CI upper CI p Beta lower CI upper CI p Beta lower CI upper CI p
Total hip BMD 0.087 -0.030 0.203 0.145 -0.055 -0.158 0.047 0.287 0.030 -0.042 0.101 0.417
Fem neck BMD 0.072 -0.048 0.192 0.237 -0.072 -0.177 0.033 0.180 0.033 -0.041 0.107 0.380
Hip CSMI 0.008 -0.099 0.115 0.882 -0.108 -0.202 -0.014 0.025 0.067 0.001 0.133 0.045

13. 13
Impacts vs tibia pQCT
CSMI = cross sectional moment of inertia (mm4
)
Beta shows SD change in pQCT outcome per doubling in number of impacts per week, adjusted for
age, height, fat mass, lean mass, activity in other bands
Low impacts Medium impacts Higher impacts
Beta lower CI upper CI p Beta lower CI upper CI p Beta lower CI upper CI p
Peri circ -0.002 -0.021 0.018 0.866 -0.007 -0.024 0.010 0.436 0.015 0.003 0.027 0.017
CSMI 0.016 -0.033 0.064 0.524 -0.031 -0.073 0.012 0.159 0.042 0.012 0.072 0.006

14. 14
Impacts vs Bone turnover markers
Low Impacts Medium Impacts Higher Impacts
Beta lower CI upper CI p Beta lower CI upper CI p Beta lower CI upper CI p
β-CTX ng/l -0.048 -0.206 0.111 0.556 0.027 -0.117 0.171 0.713 0.085 -0.017 0.187 0.103
P1NP ng/l -0.021 -0.172 0.129 0.780 -0.039 -0.174 0.097 0.577 0.127 0.031 0.224 0.010
Beta shows SD change in bone marker per doubling in number of impacts per week, adjusted for
age, height, fat mass, lean mass, activity in other bands

15. Discussion
• Day-to-day exposure of older women to higher vertical impacts is positively associated
with bone strength of the hip and tibia.
• This relationship appears to be explained by alterations in bone size as opposed to
BMD.
• An equivalent positive association was not observed with low or medium impacts.
• Increased rates of bone formation relative to resorption may contribute to these
relationships between higher impacts and bone strength
15

16. Limitations
• Difficulty in inferring causal inferences from cross sectional studies
16

17. 17
Impacts vs tibia pQCT: additional confounder adjustment
Higher Impacts
Model Beta lower CI upper CI p
Periosteal
circumference Age, height, fat mass, other bands 0.012 0.000 0.025 0.056
+ Co-morbidities 0.012 0.000 0.025 0.055
+ Social class 0.012 0.000 0.025 0.055
+ Bone active medication 0.014 0.002 0.027 0.028
Cross sectional
moment of inertia Age, height, fat mass, other bands 0.037 0.006 0.069 0.020
+ Co-morbidities 0.038 0.006 0.069 0.020
+ Social class 0.037 0.006 0.069 0.021
+ Bone active medication 0.040 0.008 0.072 0.015

18. Limitations
• Difficulty in making causal inferences from cross sectional studies
• Seven day accelerometer recordings may not fully capture long term patterns of
physical activity
• Since effect sizes were relatively small, the clinical significance of relationships
between higher impacts and bone strength which we observed is unclear
• Participants are likely to represent a relatively healthy sub-population, limiting
generalisability of our observations
18

19. Conclusions
• Habitual levels of higher, but not medium or low, vertical impacts in older women are
positively associated with lower limb bone size and strength.
• Equivalent relationships are not observed for low or medium impacts.
• The benefit of day-to-day physical activity in older women for their skeletal health
appears to be explained by exposure to higher impacts, despite their rarity and the
relatively low g levels used to define these.
19

20. Acknowledgements
Kimberly Hannam1
Kevin Deere1
April Hartley1
Usama Al-Sari1
Emma Clark1
William Fraser2
1
Musculoskeletal Research Unit, University of Bristol School of Clinical Sciences
2
University of East Anglia
http://www.bristol.ac.uk/clinical-sciences/research/musculoskeletal/rheumatology/research/vibe/
20