• coeliac disease
• osteoporosis
• bone fracture

Metabolic bone disease is a well established complication of coeliac disease. Early reports focused mainly on the association with vitamin D deficiency and osteomalacia¹ but more recently attention has turned to osteoporosis.² A number of studies have demonstrated reduced bone mineral density in individuals with coeliac disease, particularly in untreated cases and in those who fail to respond fully to a gluten free diet.³ Furthermore, the need to consider coeliac disease as a pathogenetic factor in individuals presenting with osteoporosis has been emphasised; this particularly applies to those with clinical features of the disease and those who fail to respond to treatment for their osteoporosis.

The clinical significance of the observed reduction in bone mineral density associated with coeliac disease is unclear. While a small proportion of patients undoubtedly develop severe osteoporosis with multiple fragility fractures, the question of whether fracture risk is significantly increased across the clinical spectrum of coeliac disease has only recently been explored. Studies in small numbers of selected patients, drawn from hospital clinics, have suggested that fracture risk is increased^4,5; however, such studies are likely to overestimate risk and cannot be generalised to the coeliac population as a whole.

A more representative approach is to study fracture rates in population based cohorts of coeliac disease patients and controls. In this issue of Gut, Thomason and colleagues⁶ report the results of one such study in 274 patients and 224 control subjects [see page 518]; no significance difference in fracture risk was demonstrated between these two groups, the overall age and sex adjusted odds ratio being 1.05 (95% CI 0.66–2.25). For forearm or wrist fractures and for low trauma fractures in men, there was a trend towards increased risk with odds ratios of 1.21 and 1.28, respectively, but these were not statistically significant. In a larger population based case control study of 1021 patients with coeliac disease, Vestergaard and Mosekilde⁷ also found no significant increase in fracture risk either before or after diagnosis, with incidence rate ratios of 1.15 (1.00–1.32) and 1.19 (1.06–1.33), respectively.

These two studies thus provide no definitive evidence for increased fracture rate in association with coeliac disease. However, the slight (but non-significant) increase in risk observed in both studies might suggest that the failure to demonstrate statistical significance was due to limitations of the study design rather than to the true absence of an association. Important considerations in this respect are the size of the study and the robustness of the methods used for ascertainment of coeliac disease and fracture. Power calculations should take into account all of these factors as poor ascertainment of disease status and/or fracture affects the sample size required. Furthermore, the validity of the power calculations depends critically on the assumed fracture rate in the control population, which in reality varies considerably between studies. The study of Thomason and colleagues⁶ was powered to demonstrate a twofold or greater increase in fracture risk in the coeliac population but did not have sufficient power to demonstrate smaller increases in risk. In contrast, the study of Vestergaard and Mosekilde⁷ was powered to show an increase in incidence risk ratio of 1.2–1.5; however, the relatively low validity (78%) of the diagnostic criteria used for coeliac disease is likely to have reduced the power of the study and may have led to underestimation of risk ratios.

While an increase in fracture risk cannot be excluded on the basis of these studies, it can be concluded that the magnitude of any increase, if present, is small and thus for the majority of individuals with coeliac disease absolute risk of fracture is low. Furthermore, prospective studies have shown significant improvement in bone mineral density after introduction of a gluten free diet,^3,8 indicating that bone densitometry at diagnosis may considerably overestimate short term fracture probability. This has implications for clinical practice as bone densitometry for all patients with coeliac disease, as advocated in some guidelines,⁹ will not be cost effective and may result in unnecessary treatment with bone preserving agents. A more rational approach is to restrict bone densitometry to the minority of individuals in whom short term (5–10 years) fracture probability is high; risk factors for fracture have not been specifically identified in coeliac disease but are likely to include non- compliance with or failure to respond fully to a gluten free diet, glucocorticoid therapy, untreated hypogonadism, age, low body mass index, and previous fragility fracture.

Effective treatment strategies for osteoporosis in individuals with coeliac disease have not been defined. However, it seems reasonable to advocate similar lifestyle measures to those recommended in postmenopausal osteoporosis¹⁰; these include vitamin D repletion where required and maintenance of adequate calcium intake, using calcium supplements if necessary. In osteoporotic patients with malabsorption, intravenous therapy with pamidronate or zoledronate should be considered; in others, the therapeutic options should be based on evidence from studies in postmenopausal women and men with osteoporosis.

At the present time therefore available evidence indicates that the risk of fragility fractures is, at most, only slightly increased in coeliac disease and that the absolute risk of fracture in the majority of these individuals is low. Bone densitometry should therefore be reserved for the minority with a high fracture probability, selected on the basis of risk factors. Screening of all patients with coeliac disease, as currently recommended in the British Society of Gastroenterology guidelines, cannot be justified and represents an inappropriate use of resources. The existing guidelines should therefore be revised to accommodate these considerations, which also apply to patients with inflammatory bowel disease.¹¹