We read with great interest the 2010 American College of Rheumatology (ACR) guidelines for the prevention and treatment of glucocorticoid (GC)–induced osteoporosis published in Arthritis Care & Research. In our rheumatology clinic, we started a project to improve our clinic's adherence to guidelines for GC-induced osteoporosis. The ACR guidelines state that men ages >50 years and postmenopausal women should receive a bisphosphonate if they are taking ≥7.5 mg of prednisone for 3 months, even if their risk of fracture is low (FRAX score predicts <10% risk of a 10-year major osteoporotic fracture) ([1]). Because this was rated as level A evidence, we expected to find excellent data demonstrating that GCs increase the risk of fracture and bisphosphonates prevent fractures in these low-risk patients. However, this was not the case.

There is clear evidence that GCs decrease bone mineral density (BMD) ([2]) and cause fractures ([3]), but it is not clear if those receiving GCs with T scores greater than −2.5 have an increased risk of fracture. In a 2003 study, van Staa et al stated that fracture risk was increased at all BMDs in those receiving GCs, but they did not state whether the differences were statistically significant at higher BMDs; on the basis of a figure in this article, it appears that there was close to 0% risk of fracture in patients with the highest BMDs ([4]). It is also unclear whether bisphosphonates prevent fractures in those receiving GCs with nonosteoporotic BMDs. Several studies have demonstrated that GCs increase the risk of fracture through mechanisms independent of BMD ([4, 5]), yet improvements in BMD were used as the primary outcome measure in the majority of clinical trials. The small number of randomized controlled trials (RCTs) in those receiving GCs reporting clinically relevant outcomes such as a reduction in fractures did not distinguish whether there was an effect in nonosteoporotic patients. A pooled analysis of 2 RCTs examined fracture reduction among those receiving GCs without preexisting vertebral fractures and found fewer fractures in the bisphosphonate-treated group ([6]). However, this result was not statistically significant and the study did not comment on the bone density of the patients included in this outcome, so some patients might have been osteoporotic at baseline. There are no studies of bisphosphonates demonstrating a reduction in fractures in patients with osteopenic or normal BMDs.

Based on the available evidence, we should treat older individuals receiving GCs who are already osteoporotic and high-risk osteopenic patients such as those with an estimated 10-year major osteoporotic fracture risk >20%. But should we treat low-risk individuals receiving GCs including those with a normal BMD? The answer to this question is probably not. According to the Third National Health and Nutrition Examination Survey data set, 41% of women and 68% of men ages >50 years have normal bone density ([7]). In this population, the benefits of bisphosphonates are unclear, so we may be causing more harm than good because of the side effects of bisphosphonates, including the small risk of atypical femur fractures ([8]). The 2007 European League Against Rheumatism guidelines are less specific and recommend individualizing decisions about bisphosphonate use based on risk factors and BMD ([9]). Kanis et al recently published a simple calculation for incorporating the GC dose into the FRAX tool, which can help us distinguish which of these nonosteoporotic patients would benefit from treatment with bisphosphonates ([10]). This approach seems much more reasonable, unless further data supporting bisphosphonate use for GC users with higher BMDs become available.