Endocrine Abstracts

Many risk factors predict fractures overall, but other factors may be associated with the incidence of vertebral fractures in particular. We carried out a 10-year prospective study of a population-based group of women, ages 50 to 85 years. At baseline we measured bone mineral density (BMD) and collected fasting blood samples for biochemical measurements including serum calcium, alkaline phosphatase, parathyroid hormone, creatinine, phosphate, albumin and thyroid hormones. Medical and lifestyle data were obtained by questionnaire. Incident vertebral fractures were determined by a single radiologist from spinal radiographs at 0, 2, 5, 7 and 10 years, and non-vertebral fractures were confirmed by radiologist reports.

Survival time to first fracture was used in a Cox regression model to identify risk factors for vertebral and for non-vertebral fractures. Risk factors that predicted both types of fracture included age, BMD at the lumbar spine, hip or total body, years of estrogen exposure and prevalent vertebral fracture.

However, a number of factors predicted vertebral fractures only. The predictors that remained significant when adjusted for age included low serum T3 (triiodothyronine), low serum albumin and low body fat. Neither TSH nor T4 predicted fracture. Adjusted for age, the relative risks (RR) per SD decrease and the 95% confidence intervals were: serum T3, RR=1.71 (1.25,2.33), p<0.001; serum albumin, RR=1.74 (1.12,2.70), p<0.05; body fat (g), RR=1.55 (1.01,2.38), p<0.05. Both T3 and albumin also predicted vertebral fracture independently of spine or hip BMD. We have more recently confirmed the association between low baseline albumin levels and incident vertebral, but not non-vertebral, fractures in a much larger cohort of 2720 women.

Low levels of T3, as found in 'sick euthyroid syndrome', and of albumin are each associated with sickness and chronic ill health. We conclude that general poor health may predict vertebral fractures but does not predict non-vertebral fractures.