Vertebral deformities as predictors of non-vertebral fractures

BMJ 1994; 309 doi: (Published 15 October 1994) Cite this as: BMJ 1994;309:991
  1. H Burger,
  2. P L A Van Daele,
  3. D Algra,
  4. A Hofman,
  5. D E Grobbee,
  6. H E Schutte,
  7. J C Birkenhager,
  8. H A Pols
  1. Department of Epidemiology and Biostatistics, Radiology, and Internal Medicine, Erasums University Medical School, PO Box 1738, 3000 DR Rotterdam, Netherlands
  1. Correspondence to: Dr Pols.
  • Accepted 12 July 1994

The estimated number of people fracturing a hip will increase from 1.7 million in 1990 to 6.3 million in 2050.1 It is therefore important to identify risk factors.2 Previous vertebral fractures have been shown to increase the risk of subsequent vertebral fractures.3 The degree of spinal deformity may, however, indicate the bone quality of the whole skeleton. We investigated the association between vertebral deformities and new non-vertebral fractures.

Subjects, methods, and results

This nested case-control analysis was carried out within a larger prospective follow up investigation of disease in elderly people.4 We studied 40 subjects who had a non-vertebral fracture on average 582 (range 287-1028) days after entering the larger study. Controls were subjects who had not had a non-vertebral facture during 648 (range 113- 1072) days of follow up and who were patients of the general practitioners of the patients with fracture.

On entering the study patients were asked about hip and wrist fractures in the past five years, history of hip fracture in parents or siblings, and frequent falling (more than once a month). Vertebral heights were measured from lateral radiographs and ratios were calculated according to Melton et al.5 A grade I or grade II deformity was recorded in a vertebra if one of the ratios was smaller than a reference value minus 2 SD or 3 SD, respectively. A mild spinal deformity was defined as one to three grade I vertebral deformities and at most one grade II deformity, and a severe deformity as more than three grade I or more than one grade II deformities. Radiographs of the spine were unavailable in one case and in two controls. We measured bone mineral density at the femoral neck by dual energy x ray absorptiometry. The associations of baseline variables with new non-vertebral fractures were expressed as relative risks adjusted for age and sex.

Ten patients had fractures in the hip, 17 in the radius or ulna, and 13 elsewhere. Mean age was 74.7 (SD 8.5) years in cases and 74.4 (SD 7.6) in controls; 90% of subjects were women. Subjects with a history of a wrist or hip fracture had a significantly higher risk (relative risk 3.1, 95% confidence interval 1.1 to 8.6) of fracture. A family history of hip fracture (2.0, 0.6 to 5.9) and frequent falling (4.7, 0.8 to 26.2) did not significantly increase the risk of fracture. The highest quartiles of bone mineral density (0.3, 0.1 to 1.1) and body mass index (0.8, 0.3 to 2.4) were associated with a non-significantly decreased risk.

The table shows the increasing relative risk of incident non-vertebral fractures with increasing spinal deformity. Adjustment for bone mineral density did not affect the risk estimates. The relative risks were also similar after other variables were adjusted for.

Relative risk of baseline spinal deformation for new non-vertebral fractures

View this table:


We found a strong relation between spinal deformity and subsequent non-vertebral fractures that was independent of bone mineral density. Our findings agree with the observations by Ross et al that previous vertebral fractures predict future vertebral fractures.3

There are at least two explanations of our findings. Firstly, an increased thoracic kyphosis caused by multiple vertebral deformities could shift the centre of gravity forward increasing the frequency of falling over. However, the relation of vertebral deformities with risk of fracture was independent of frequency of falling. Secondly, and we believe more likely, vertebral deformities may reflect impaired bone strength in the whole skeleton since the relation was independent of bone mineral density. We conclude that by assessing the degree of spinal deformity patients with a fourfold increased risk of future non-vertebral fractures can be identified.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
View Abstract