Osteoporosis International

, Volume 23, Issue 1, pp 67–74

Excess body fat is associated with higher risk of vertebral deformities in older women but not in men: a cross-sectional study

Authors

    • Menzies Research Institute TasmaniaUniversity of Tasmania
  • S. J. Just nee Foley
    • Menzies Research Institute TasmaniaUniversity of Tasmania
  • S. J. Quinn
    • School of Medicine, Flinders Clinical EffectivenessFlinders University
  • T. M. Winzenberg
    • Menzies Research Institute TasmaniaUniversity of Tasmania
  • G. Jones
    • Menzies Research Institute TasmaniaUniversity of Tasmania
Original Article

DOI: 10.1007/s00198-011-1741-8

Cite this article as:
Laslett, L.L., Just nee Foley, S.J., Quinn, S.J. et al. Osteoporos Int (2012) 23: 67. doi:10.1007/s00198-011-1741-8

Abstract

Summary

Thinness is a risk factor for fractures, but the effect of obesity on fracture risk is less clear. We found an association between measures of obesity and prevalence and number of vertebral deformities in women but not in men, in a cross-sectional study of 1,011 participants aged 50–80 years.

Introduction

Low body weight is well recognised as a risk factor for fractures, but the association between overweight and fracture risk is less well described. This cross-sectional study describes the association between measures of obesity and vertebral deformities in 1,011 male and female participants in the Tasmanian Older Adult Cohort study.

Methods

Vertebral deformities (anterior wedging) of T4–L4 were determined by morphometric dual-emission X-ray absorptiometry. Body fat was assessed as weight, body mass index (BMI), waist–hip ratio (WHR), waist circumference and DXA measures of trunk fat (in percent) and total fat mass.

Results

The mean age of participants was 63 ± 7 years, and mean BMI was 28 ± 5. Prevalent thoracic vertebral deformities were associated with increasing weight [standardised β (Sβ) 0.29, p = 0.003], BMI (Sβ 0.33, p < 0.001), trunk fat (Sβ 0.20, p = 0.03), waist circumference (Sβ 0.19, p = 0.03) and fat mass (Sβ 0.23, p = 0.03), but not the WHR in women, and only with decreasing total fat mass in men. In addition, the number of vertebral deformities increased as weight, BMI or fat mass increased in women (all p < 0.05) but decreased with increasing total fat mass in men. Associations between fat mass and vertebral deformities were mainly linear, but there was some evidence of a threshold effect in women with a BMI ≥35.

Conclusions

There is a deleterious association between increasing amounts of body fat in women but not in men and the prevalence and number of vertebral deformities, which may reflect loading of the thoracic spine.

Keywords

ObesityOsteoporosis/epidemiologyRisk factorsVertebral deformityVertebral fracture

Introduction

Osteoporosis is a major public health problem [1]. Vertebral fractures are common osteoporotic fractures in older adults and can present clinically as severe back pain. However, most (60–70%) are clinically silent [2, 3] and can easily go undiagnosed. There is currently no agreed “gold standard” for the definition of osteoporotic vertebral fracture [4], and given the uncertainty around exactly which types of deformities that constitute fractures, we use the term “vertebral deformity” to describe “abnormal” anatomical shapes in vertebral bodies, which may or may not be associated with pain or other symptoms. Both mild [5] and severe [5, 6] vertebral deformities increase the risk of sustaining a subsequent fracture [6].

Slenderness has long been associated with osteoporosis and fractures [7], and low body weight has now been shown to be a risk factor for osteoporotic fractures especially of the hip [810]. The relationship between body mass index (BMI) and fracture risk is non-linear at low body weights, roughly linear at healthy body weights [8] and understudied at high body weights.

Lean mass is a predictor of bone mass through its mechanical load on the skeleton, with lean mass more strongly related to bone mineral density (BMD) than fat mass in cross-sectional studies in men [11] and postmenopausal women [1113]. As a person becomes increasingly obese, the proportion of body mass that is lean mass decreases in proportion to the total [13, 14]. When fat mass is adjusted for the effect of mechanical loading (total body weight), fat mass is negatively correlated with bone mass, suggesting that fat mass actually has a detrimental effect on bone [14, 15]. There are also many other possible associations between fat mass and bone mass on the molecular and cell level, including adipocyte hormones, estrogens, leptin, IL-6 and insulin [16, 17]. In addition to any effects relating to bone mass, obesity has been postulated to increase the risk of falling [18], as it has been shown to increase postural instability [19, 20] and decrease accuracy when performing rapid movements [19].

Both high body weight and fat have been shown to predict fracture in children [21, 22], with fat mass substantially inhibiting bone accrual in children with prior fracture [23]. However, the association between obesity and fractures in adults is less certain. A recent study in a fracture outpatient clinic at a tertiary hospital demonstrated a higher than expected prevalence of obesity in postmenopausal women with prior fractures [24], and obesity has been shown to be associated with vertebral [25, 26] and humerus fractures [27] in women, and hip, upper limb [28, 29] and lower limb fractures [28] in men. However, neither of the studies investigating vertebral fractures or deformities included male participants [25, 26], and one did not adjust for BMD [25], a major determinant of fracture risk [30]. The remaining study showed that the prevalence of vertebral fractures was associated with increasing BMI and weight, and the number of vertebral fractures was associated with BMI and height but not weight when BMI was substituted for BMI [26]. Neither used additional objective measures of fat mass such as DXA. Thus, the relationship between body mass and vertebral fracture risk when body weight is high is still unclear particularly in men. This is an important consideration given the increasing prevalence of overweight and obesity in the developed world [3133]. The aim of this cross-sectional study was, therefore, to describe the association between measures of obesity and vertebral deformities in randomly selected community-dwelling male and female participants aged 50–80 years.

Methods and procedures

Participants

The Tasmanian Older Adult Cohort (TasOAC) study is an ongoing, prospective, population-based study, which aims to identify factors associated with the development and progression of osteoarthritis and osteoporosis in older adults living in the community. Men and women aged 50–80 years in 2002 were selected from the electoral roll in southern Tasmania (population 229,000) using sex-stratified, simple random sampling without replacement (response rate 57%). Excluded participants were those living in an aged care facility or having contraindications to magnetic resonance imaging, as this was required to examine osteoarthritis progression. The Southern Tasmanian Health and Medical Human Research Ethics Committee approved the study, and we obtained written informed consent from all participants.

Anthropometrics

Weight was measured to the nearest 0.1 kg (with shoes, socks, bulky clothing and headwear removed) using a single pair of calibrated electronic scales (Seca Delta Model 707). Height was measured to the nearest 0.1 cm (with shoes and socks removed) using a stadiometer. BMI was calculated [weight (in kilograms)/height (in meters)2]. Waist and hip circumferences were measured to the nearest 0.1 cm using a constant tension tape (Figure Finder tape measure) directly over the skin.

Dual X-ray absorptiometry

Participants underwent a whole-body dual X-ray absorptiometry (DXA) scan using a Hologic Delphi densitometer (Hologic, Waltham, MI), from which body composition and bone density was determined. We excluded participants from the DXA scans if their weight exceeded 130 kg (n = 3). Spine DXA scans were read for the remaining participants. DXA measured the mass (in grams) of bone mineral, fat and lean mass of the whole body and compartments including the trunk as well as BMD (grams per square centimeter) at the hip (total hip) and spine (L1–L4). We calculated percentage trunk fat mass as the ratio of trunk fat mass divided by total trunk mass (i.e. the sum of fat mass, lean mass and bone mass).

Quantitative morphometry

One assessor (SJ) assessed the morphometric DXA (MXA) scans. One patient had a very poor-quality scan with no identifiable vertebrae and was omitted from the analyses. Using the lateral views, markers were placed on each of the four corners and in the centre of the superior and inferior surfaces of vertebrae T4–L4 in order to determine the dimensions. This was done visually using the “Markers” option in Hologic APEX software V2.2. Data were stored as co-ordinates and exported to Microsoft Access 2003 (Microsoft Corporation), then converted to length (in millimeters) using Pythagoras' theorem. We assessed anterior wedging—the most common type of vertebral deformity [34]. We defined a vertebral deformity as a ratio of the anterior to posterior heights of <0.8, representing ≥20% reduction in the height of the anterior portion of a vertebral body relative to the posterior height of that body. This was justified based on the criteria for subsidised access to authority prescriptions for anti-osteoporotic medication in Australia [35]. We assessed the presence or absence of deformities in the thoracic spine (T4–T12) and/or lumbar spine (L1–L4), and total number of wedge deformities in the whole spine (T4–L4).

Other measures

Self-reported estimates of current cigarette smoking prevalence were collected by questionnaire. Physical activity levels were determined using pedometers (Omron HJ-003 and HJ-102; Omron Healthcare, Kyoto, Japan), as described in Scott et al. [36].

Data analysis

Students' t-tests and chi square tests were used to compare differences in means and proportions, respectively, between participants with and without deformities. GEE binomial models were used for the binary data (presence or absence of vertebral deformity), clustering on each individual participant with a logit link to account for correlated readings within individuals. Associations between the predictor variables and outcomes were checked for linearity using the Stata command “fracpoly”. Logistic regression was used to give odds ratios in men as the GEE model did not converge. Negative binomial regression was used to analyse count data (number of vertebral deformities) and correct for overdispersion. Data were standardised in order to present sex-specific standardised coefficients, which enable meaningful comparison of variables with different units of measurement. Analyses were adjusted for age, hip and spine BMD. There were statistically significant interactions between vertebral deformities (either prevalence or number) and sex for trunk fat (in percent) and total fat mass (adjusted for lean mass). Therefore, data for females and males were analysed separately. Statistical significance was defined as a p value ≤0.05 (two tailed). We used Stata 10.0 (StataCorp LP) for all statistical analyses.

Results

Participants

A total of 1,011 people (48.4% men) between 50 and 80 years of age (mean, 62.6 years) were included in this study. The study population was 98% white or of Caucasian ancestry, consistent with the Tasmanian community. The majority of the women (98.4%) were either menopausal (19%) or postmenopausal (77%). Table 1 shows characteristics of the study population stratified by presence or absence of deformity. Women who had at least one vertebral deformity were older and shorter with higher BMI and had greater percentage body fat (total and trunk fat) than women without vertebral deformities. Men with at least one vertebral deformity were older than men without vertebral deformities and had lower fat mass [trunk fat (in percent), fat mass (in percent)] and higher lean mass (in percent).
Table 1

Demographics of study participants with and without one or more vertebral deformity (≥20% anterior wedging) by sex

 

Females

Males

No deformities (n = 328)

Deformities (n = 194)

p value

No deformities (n = 245)

Deformities (n = 244)

p value

Age

61.3 ± 6.8

64.1 ± 7.7

<0.001

62.3 ± 7.3

63.7 ± 7.5

0.046

Current smoking (%)

12%

8%

0.165

12%

16%

0.144

Weight (kg)

70.6 ± 12.3

72.9 ± 15.7

0.070

84.2 ± 13.6

83.6 ± 12.7

0.613

Height (cm)

161.0 ± 6

159.8 ± 6

0.025

173.8 ± 6.2

174.2 ± 6.1

0.415

BMI

27.3 ± 4.6

28.6 ± 6.1

0.005

27.8 ± 3.9

27.6 ± 3.9

0.415

Trunk fat (%)

37.2 ± 7.0

38.5 ± 7.1

0.045

29.4 ± 6.3

28.2 ± 6.8

0.038

Waist circumference

87.9 ± 12.1

89.8 ± 13.9

0.089

99.2 ± 11

98.7 ± 10.5

0.583

Waist–hip ratio

0.86 ± 0.07

0.86 ± 0.07

0.625

0.98 ± 0.06

0.98 ± 0.06

0.874

Fat mass (%)

39.4 ± 5.4

40.6 ± 5.4

0.031

28.0 ± 4.8

27.1 ± 5.2

0.044

Lean mass (%)

57.8 ± 5.1

56.8 ± 5.1

0.018

68.9 ± 4.5

69.8 ± 4.9

0.049

Spine BMD (g/cm2)

0.97 ± 0.15

0.96 ± 0.17

0.471

1.07 ± 0.18

1.05 ± 0.17

0.235

Hip BMD (g/cm2)

0.91 ± 0.14

0.90 ± 0.15

0.457

1.04 ± 0.14

1.02 ± 0.15

0.297

Mean ± standard deviation

Prevalent vertebral deformities

The prevalence of at least one vertebral deformity was 37% in women (n = 194) and 50% in men (n = 244). The mean number of deformities was 0.6 in women (range 0–5) and 1.0 in men (range 0–8). The majority of vertebral deformities (96%) were in the thoracic region (T4–T12) with a prevalence of 36% in women and 47% in men (Table 2).
Table 2

Proportion of participants with anterior wedging (≥20%) by vertebra

Vertebra

Women

Men

Number

Percent

Number

Percent

T4

14/468

3.0

12/368

3.3

T5

30/502

6.0

29/449

6.5

T6

57/514

11.1

60/473

12.7

T7

74/519

14.3

74/485

15.3

T8

63/518

12.2

87/488

17.8

T9

18/518

3.5

57/485

11.8

T1

8/517

1.5

27/484

5.6

T1

24/515

4.7

59/484

12.2

T1

21/517

4.1

42/485

8.7

L1

7/519

1.3

23/486

4.7

L2

0/520

0.0

3/488

0.6

L3

0/519

0.0

2/487

0.4

L4

0/504

0.0

0/452

0.0

Prevalence of at least one deformity

194/522

37.2

244/489

49.9

There was a positive association between prevalent vertebral deformities and weight, BMI, trunk fat (in percent), waist circumference and total fat mass (adjusted for lean mass) in women in the thoracic spine, but not in men, while there was a negative relationship between prevalent vertebral deformities and total fat mass only in men (Table 3). The association between prevalent vertebral deformities and BMI remained statistically significant after further adjustment for height (data not shown). In a sensitivity analysis, we limited the vertebral deformities to those in participants with back pain (n = 122, 23.3% in women; n = 140, 28.6% in men) as a proxy marker for clinical fracture and compared these with the participants with no deformities and no back pain (Table 3). This increased the strength of the associations for weight, BMI, trunk fat (in percent), waist circumference and total fat mass in women. The association between vertebral deformity and fat mass in men reduced in strength and was no longer statistically significant.
Table 3

Association between presence or absence of at least one vertebral deformity (≥20%) and measures of body fat in the thoracic spine stratified by sex

 

At least one vertebral deformity (≥20%) (yes or no)

At least one vertebral deformity (≥20%) PLUS back pain vs no deformitya

Females, standardised β (95% CI)

Males, standardised β (95% CI)

Females, standardised β (95% CI)

Males, standardised β (95% CI)

Weight

0.29 (0.11 to 0.47)

0.07 (−0.09 to 0.23)

0.40 (0.16–0.64)

0.16 (−0.05 to 0.37)

BMI

0.33 (0.17 to 0.50)

0.05 (−0.10 to 0.20)

0.47 (0.27–0.68)

0.13 (−0.08 to 0.33)

Trunk fat (%)

0.20 (0.02 to 0.37)

−0.09 (−0.24 to 0.06)

0.30 (0.07–0.53)

−0.02 (−0.22 to 0.18)

Waist circumference

0.19 (0.02 to 0.37)

0.05 (−0.11 to 0.20)

0.25 (0.02–0.48)

0.15 (−0.05 to 0.35)

Waist–hip ratio

−0.005 (−0.16 to 0.15)

0.06 (−0.08 to 0.21)

−0.03 (−0.23 to 0.18)

0.17 (−0.02 to 0.36)

Total fat massb

0.23 (0.03–0.43)

−0.19 (−0.36 to −0.02)

0.30 (0.04–0.56)

−0.17 (−0.40 to 0.05)

All analyses are adjusted for age, hip and spine BMD. Bold denotes statistical significance

aParticipants with vertebral deformity and no back pain are omitted from this analysis

bFurther adjusted for lean mass

There were no statistically significant associations between body mass and vertebral deformities in the lumbar region in either men or women (data not shown). Model diagnostics for Table 3 showed that no transformations of the data from linear improved the fit of the model (data not shown). However, Fig. 1 suggests that the relationship between prevalence of vertebral deformities is linear in women until a BMI of 35, with a marked increase in prevalence of vertebral deformities in women with BMI ≥35. Compared to women of average age, hip and spine BMD and a BMI of 20–24.9, BMI of 35–39.9 had a 2.4-fold increased risk (odds ratio) of vertebral deformity (95% CI 1.3–4.6), and women with BMI of ≥40 had an odds ratio of 3.6 (95% CI 1.8–7.3). Women in the low, healthy and mildly obese BMI categories did not show increased risk of vertebral deformities [BMI <20—OR 1.3 (95% CI 0.4–3.7), p = 0.66; BMI 25–29.9—OR 1.0 (0.7–1.4), p = 0.86; and BMI 30–34.9—OR 1.4 (0.9–2.1), p = 0.15]. In men, none of the subcategories demonstrated increased or decreased risk [BMI <20—OR 1.0 (0.2–4.0); BMI 25–29.9—OR 0.9 (0.6–1.3); BMI 30–34.9—OR 1.1 (0.7–1.8); BMI 35–29.9—OR 0.8 (0.3–1.8); BMI ≥40 no estimates (n = 2)].
https://static-content.springer.com/image/art%3A10.1007%2Fs00198-011-1741-8/MediaObjects/198_2011_1741_Fig1_HTML.gif
Fig. 1

Association between body mass index and prevalence of vertebral deformities by sex. Number of participants in individual groups listed above each bar

Number of vertebral deformities

The number of vertebral deformities increased with increasing BMI in women but not in men (Table 4, Fig. 2). Table 3 shows that in addition to BMI, increasing number of vertebral deformities is also associated with increasing weight and total fat mass (adjusted for lean mass) in women. The association between number of vertebral deformities and BMI remained statistically significant after further adjustment for height (data not shown). In men, there was no relationship between number of vertebral deformities and weight or BMI but the number of vertebral deformities was negatively associated with fat mass, which persisted after further adjustment for putative confounders, such as smoking and steps per day. Model diagnostics for the analyses shown in Table 4 showed that no transformations of the data from linear improved the fit of the model (data not shown).
Table 4

Association between number of vertebral deformities (≥20%) and measures of body fat

 

Females, standardised β (95% CI)

Males, standardised β (95% CI)

Weight

0.24 (0.08 to 0.41)

0.03 (−0.10 to 0.17)

Body mass index

0.29 (0.13 to 0.44)

0.03 (−0.11 to 0.16)

Trunk fat (%)

0.12 (−0.03 to 0.28)

−0.12 (−0.25 to 0.005)

Waist circumference

0.01 (−0.001 to 0.02)

0.001 (−0.011 to 0.01)

Waist–hip ratio

−0.02 (−0.18 to 0.14)

0.04 (−0.08 to 0.17)

Total fat massa

0.19 (0.01 to 0.37)

−0.17 (−0.32 to −0.02)

All analyses are adjusted for age, hip and spine BMD. Bold denotes statistical significance

aFurther adjusted for lean mass

https://static-content.springer.com/image/art%3A10.1007%2Fs00198-011-1741-8/MediaObjects/198_2011_1741_Fig2_HTML.gif
Fig. 2

Association between body mass index and number of vertebral deformities (≥20%) stratified by sex

To investigate whether the choice of definition of vertebral deformity was important, we conducted a sensitivity analysis using a threshold-free approach, with degree of anterior wedging as a continuous variable as the outcome. The results were largely similar to the results in Table 3 with regard to statistical significance and direction of association with the exception of anterior wedging and total fat mass in men, which was not statistically significant [standardised β (Sβ) −0.003 (−0.007 to 0.001); p = 0.15].

Discussion

This is the first study to investigate the association between excess body fat assessed objectively and vertebral deformities in a population-based setting and the first study to investigate the association in men. It demonstrates that there is a dose–response relationship between both prevalence and number of vertebral deformities and multiple measures of body fat in women but not consistently in men. The association was independent of age and bone density, and appears linear in nature, with some evidence of a threshold effect at a BMI of 35.

There was no relationship between prevalence or number of vertebral deformities and body fat in men, except for a negative relationship with total fat mass (after adjustment for lean mass). Men with higher lean mass may have had occupations involving more physical work than men with lower lean mass and thus have had a higher risk of sustaining vertebral deformities from occupational trauma. The higher prevalence of deformities in men observed in this study is consistent with previous Australian data on vertebral deformities [37].

The reasons for the male–female differences are uncertain. As men and women have different distributions of trunk fat, related to the presence of breasts in women, the relationship between trunk fat mass and vertebral deformities in women could be related to breast size and therefore could explain the sex differences. Unfortunately, no direct measures of breast mass were collected in TasOAC so we are unable to investigate this further. In addition, there were relatively few men with a BMI greater than 35, meaning we had much less power in men (despite a similar sample size), and the fat mass (in percent) is lower in men compared to women, for a similar BMI. These could account for the differences observed.

The association in women appears driven by overall fat mass, as weight, BMI, waist circumference and total fat mass were all associated with prevalence and number of vertebral deformities in women. Waist circumference is more strongly correlated with abdominal fat content (as measured by computed tomography or DXA) than the waist–hip ratio [38, 39], and a better predictor of atherogenic metabolic disturbances associated with abdominal obesity than waist hip ratio [39], as it measures visceral rather than subcutaneous fat. This may explain why waist circumference was associated with vertebral deformity in women, but waist hip–ratio was not. We attribute the strengthened association between obesity and vertebral deformities seen in participants with vertebral deformities and back pain to positive associations between obesity and both pain and prevalent vertebral deformities in women. There is no association between back pain and vertebral deformity, therefore associations between obesity and vertebral deformity in our sample are not confounded by pain.

There are some differences between our sample and those from the literature, such as in the meta-analysis of BMI and fracture risk by De Laet and colleagues [8]. Our sample is heavier, with an average BMI of 27.2 g/cm2. This is likely to reflect secular changes in obesity trends as our data were collected at a later date. De Laet et al. [8] noted that the relationship between BMI and fracture risk when unadjusted for bone density was strongest in the people with BMI <20. This effect was not statistically significant in our cohort, probably due to the low prevalence of BMI <20 (1.5% in our sample, compared to 8.5% in De Laet et al. [8]). Whilst the statistical modelling suggests a linear association for both prevalence and increasing number of vertebral deformities in women, there was some evidence (Fig. 1) of a threshold for increased prevalence and number of vertebral deformities in women with BMI ≥35 and also less than 20. This supports the findings of Premaor and colleagues [24], who found a high prevalence of obesity amongst female attendees of a fracture outpatient clinic.

The effect of obesity on vertebral deformities may vary by fracture type and by sex. Two studies demonstrated increased risk of lower limb fractures among obese women, but no increased risk with fractures of other types [13, 27]. In men, Nielson reported that obesity increased the risk of hip fractures [28], and Ensrud et al. [9] showed no association with obesity after adjustment for age and BMD [8].

Ferrar et al. argue that some types of mild deformities are “non-fracture variants” rather than vertebral deformities. These include short vertebral height (SVH), which they report as being more common in older, heavier women, unrelated to osteoporosis [40, 41], equally prevalent in pre- and postmenopausal women [41], and therefore unimportant. However, the limited evidence on the natural history of “mild” vertebral deformities suggests that they do predict subsequent fracture [5] and may even represent an epidemiologically distinct subtype of vertebral “fractures” [42]. This may also explain the different demographic pattern of women with SVH between pre- and postmenopausal women in Ferrar and colleagues' comparison of these groups [41]. Pollintine et al. describe a compelling aetiological mechanism as to how constant mechanical loading could cause microcracks leading to progressive vertebral deformities without necessarily resulting in actual fracture [43]: this is particularly relevant in the context of obesity. Since the association between vertebral deformity and measures of obesity we observed in our study was strengthened by limiting the participants with vertebral deformity only to those with back pain (a proxy measure of clinical fracture), this supports our thesis that this is a real association and not an artefact of the definition of vertebral deformity.

Unlike the findings reported by Ferrar and colleagues [40, 41], we did find that BMD was associated with deformities. Hip BMD had a significant negative association with wedge deformities in both women and men after further adjustment for age, in the associations in Table 3. Spine BMD was not significant, which is also consistent with the results from Dubbo [37]. The final model has both BMD measures in it, and neither is significant after adjusting for the other.

This study has some limitations. We were unable to confirm the relationships we observed with clinically defined vertebral fractures as the self-reported prevalence of vertebral fractures was low (3%). We also omitted rarer types of vertebral deformities such as compression and biconcave deformities [34], but the misclassification rate of participants who had compression or wedge deformities but not anterior wedging is very low (n = 27, 4.7%), suggesting that the omission of this data is unlikely to have affected our results. There are a number of ways to define a vertebral deformity, but since the results of the analyses using anterior wedging as a continuous measure are similar to the results in Table 3, this suggests that our results are unaffected by varying the definition of vertebral deformities. Due to the cross-sectional nature of this analysis, we were unable to assess antecedent factors relating to the vertebral deformity or time elapsed since the deformity and were unable to make any inferences about cause and effect relationships. Longitudinal studies are required to address causality. In conclusion, there is a deleterious association between the prevalence and number of vertebral deformities and increasing amounts of body fat in women but not in men, which may reflect loading of the thoracic spine.

Acknowledgements

We especially thank the participants who made this study possible, and we gratefully acknowledge the role of TasOAC staff and volunteers in collecting the data, particularly research nurses Catrina Boon and Pip Boon. Tisha Carter and Rose Ford conducted the densitometry. We thank the Health IT team, particularly Tim Albion and Alistair Chilcott for their expertise in converting the MXA data into a useable format.

Conflicts of interest

None.

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2011