The aim of this analysis was to determine the influence of lifestyle, anthropometric and reproductive factors on the subsequent risk of incident vertebral fracture in men and women aged 50-79 years. Subjects were recruited from population registers from 28 centers across Europe. At baseline, they completed an interviewer-administered questionnaire and had lateral thoraco-lumbar spine radiographs performed. Repeat spinal radiographs were performed a mean of 3.8 years later. Incident vertebral fractures were defined morphometrically and also qualitatively by an experienced radiologist. Poisson regression was used to determine the influence of the baseline risk factor variables on the occurrence of incident vertebral fracture. A total of 3173 men (mean age 63.1 years) and 3402 women (mean age 62.2 years) contributed data to the analysis. In total there were 193 incident morphometric and 224 qualitative fractures. In women, an age at menarche 16 years or older was associated with an increased risk of vertebral fracture (RR = 1.80; 95%CI 1.24, 2.63), whilst use of hormonal replacement was protective (RR = 0.58; 95%CI 0.34, 0.99). None of the lifestyle factors studied including smoking, alcohol intake, physical activity or milk consumption showed any consistent associations with incident vertebral fracture. In men and women, increasing body weight and body mass index were associated with a reduced risk of vertebral fracture though, apart from body mass index in men, the confidence intervals embraced unity. For most variables the strengths of the associations observed were similar using the qualitative and morphometric approaches to fracture definition. In conclusion our data suggest that modification of other lifestyle risk factors is unlikely to have a major impact on the population occurrence of vertebral fractures. The important biological mechanisms underlying vertebral fracture risk need to be explored using new investigational strategies.

The aim of this population-based prospective study was to determine the incidence of limb fracture by site and gender in different regions of Europe. Men and women aged 50-79 years were recruited from population registers in 31 European centers. Subjects were invited to attend for an interviewer-administered questionnaire and lateral spinal radiographs. Subjects were subsequently followed up using an annual postal questionnaire which included questions concerning the occurrence of new fractures. Self-reported fractures were confirmed where possible by radiograph, attending physician or subject interview. There were 6451 men and 6936 women followed for a median of 3.0 years. During this time there were 140 incident limb fractures in men and 391 in women. The age-adjusted incidence of any limb fracture was 7.3/1000 person-years [pyrs] in men and 19 per 1000 pyrs in women, equivalent to a 2.5 times excess in women. Among women, the incidence of hip, humerus and distal forearm fracture, though not 'other' limb fracture, increased with age, while in men only the incidence of hip and humerus fracture increased with age. Among women, there was evidence of significant variation in the occurrence of hip, distal forearm and humerus fractures across Europe, with incidence rates higher in Scandinavia than in other European regions, though for distal forearm fracture the incidence in east Europe was similar to that observed in Scandinavia. Among men, there was no evidence of significant geographic variation in the occurrence of these fractures. This is the first large population-based study to characterize the incidence of limb fracture in men and women over 50 years of age across Europe. There are substantial differences in the descriptive epidemiology of limb fracture by region and gender.

In Europe there is a 3-fold variation, according to geographical center, in risk of vertebral deformity in men and women over the age of 50. We investigated the relationship between bone density, as assessed by dual-energy X-ray absorptiometry (DEXA) of the spine and hip and prevalent vertebral deformities in 13 of the 36 centers participating in the European Vertebral Osteoporosis Study (EVOS). Each center recruited an age-stratified sample of men and women aged 50 years and over, and of those who agreed to densitometry, 288/2088 women and 233/1908 men were found to have one or more deformities of the vertebrae between T4 and L4 as assessed by the McCloskey algorithm. DEXA was in each case performed on L2-L4, the proximal femur, or both. Bone densitometry results were cross-calibrated between centers using the European Spine Phantom prototype and results expressed as bone mineral density (BMD, g/cm 2 ). In both genders, subjects with deformities involving loss of anterior vertebral body height alone comprised over 20% of the total with deformities and these related poorly to BMD. Other classes of deformity were found by logistic regression to relate significantly to BMD in one or both genders, with odds ratios for the risk of any of these ranging from 1.67 to 2.11 for a 1 SD reduction in bone density at spine, femoral neck, or trochanter ( p < 0.001). Adjusting for anthropometric variables and BMD did not remove the effect of age on risk which rose 1.67-to 1.78-fold per decade according to gender. The greater unadjusted rate of increase in deformity risk with age in women was attributable to their faster rate of bone loss with age; after adjusting for age, body mass index (BMI), and BMD at the trochanter in grams per square centimeter, men had a 2-fold higher risk of deformity than women. Analysis of the relationship between mean bone density and the prevalence of deformity in each center demonstrated no significant differences between centers in either gender, after adjusting for BMD, age, and BMI together with an a posteriori statistical adjustment for imperfect cross-calibration of densitometers. It is concluded that BMD is an important determinant of deformity risk in both genders. Together with age, BMD explains much of the differences in risk both between the sexes and between individual geographical centers in

Abstract. The purpose of this study was to investigate variations in bone density between 16 European populations, 13 of which were participants in the European Vertebral Osteoporosis Study (EVOS). Men and women aged 50-80 years were recruited randomly from local population registers, stratified in 5-year age bands. The other three centres recruited similarly. Random samples of 20-100% of EVOS subjects were invited for dual-energy X-ray absorptiometry (DXA) densitometry of the lumbar spine and/or proximal femur using Hologic, Lunar or Norland pencil beam machines or, in one centre, a Sopha fan.-beam machine. Crosscalibration of the different machines was undertaken using the European Spine Phantom prototype (ESPp). Highly significant differences in mean bone density were demonstrated between centres, giving rise to betweencentre SDs in bone density that were about a quarter of a population SD. These differences persisted when centres using Hologic machines and centres using Lunar machines were considered separately. The centres were ranked differently according to whether male or female subjects were being considered and according to site of measurement (L2-4, femoral neck or femoral trochanter). As expected, bone mineral density (BMD) had a curvilinear relationship with age, and apparent rates of decrease slowed as age advanced past 50 years in both sexes. In the spine, not only did male BMD usually appear to increase with age, but there was a highly significant difference between centres in the age effect in both sexes, suggesting a variability in the impact of osteoarthritis between centres. Weight was consistently positively associated with BMD, but the effects of height and armspan were less consistent. Logmithmic transformation was needed to normalize the regressions of BMD on the independent variates, and after transformation, all sites except the femoral neck in females showed significant increases in SD with age. Interestingly, the effect of increasing weight was to decrease dispersion in proximal femur measurements in both sexes, further accentuating the tendency in women for low body mass index to be associated with osteoporosis as defined by densitometry. It is concluded that there are major differences between BMD values in European population samples which, with variations in anthro~ pometric variables, have the potential to contribute substantially to variations in rates of osteoporotic fracture risk in Europe.

The presence of a vertebral deformity increases the risk of subsequent spinal deformities. The aim of this analysis was to determine whether the presence of vertebral deformity predicts incident hip and other limb fractures. Six thousand three hundred and forty-four men and 6788 women aged 50 years and over were recruited from population registers in 31 European centers and followed prospectively for a median of 3 years. All subjects had radiographs performed at baseline and the presence of vertebral deformity was assessed using established morphometric methods. Incident limb fractures which occurred during the follow- up period were ascertained by annual postal questionnaire and confirmed by radiographs, review of medical records and personal interview. During a total of 40348 person-years of follow-up, 138 men and 391 women sustained a limb fracture. Amongst the women, after adjustment for age, prevalent vertebral deformity was a strong predictor of incident hip fracture, (rate ratio (RR) = 4.5; 95% CI 2.1-9.4) and a weak predictor of 'other' limb fractures (RR = 1.6; 95% CI 1.1-2.4), though not distal forearm fracture (RR = 1.0; 95% CI 0.6-1.6). The predictive risk increased with increasing number of prevalent deformities, particularly for subsequent hip fracture: for two or more deformities, RR = 7.2 (95% CI 3.0-17.3). Amongst men, vertebral deformity was not associated with an increased risk of incident limb fracture though there was a nonsignificant trend toward an increased risk of hip fracture with increasing number of deformities. In summary, prevalent radiographic vertebral deformities in women are a strong predictor of hip fracture, and to a lesser extent humerus and 'other' limb fractures; however, they do not predict distal forearm fractures.