A systematic review and meta-analysis of the association between eating disorders and bone density.

Abstract

This meta-analysis investigates the effect of an eating disorder on bone mineral density in two eating disorder subtypes. Following conflicting findings in previous literature, this study finds that not only anorexia nervosa, but also bulimia nervosa has a detrimental effect on BMD. Key predictors of this relationship are discussed. This systematic review and meta-analysis investigates bone mineral density (BMD) in individuals with anorexia nervosa (AN) and bulimia nervosa (BN) in comparison to healthy controls (HCs). AN has been associated with low BMD and a risk of fractures and mixed results have been obtained for the relationship between BN and BMD. Deciphering the effect these two ED subtypes on BMD will determine the effect of low body weight (a characteristic of AN) versus the effects of periods of restrictive eating and malnutrition which are common to both AN and BN. We conducted a systematic search through the electronic databases MedLine, EMBASE and PsychInfo and the Cochrane Library to investigate and quantify this relationship. We screened 544 articles and included 27 studies in a random-effect meta-analysis and calculated the standardised mean difference (SMD) in BMD between women with a current diagnosis of AN (n = 785) vs HCs (n = 979) and a current diagnosis of BN (n = 187) vs HCs (n = 350). The outcome measures investigated were spinal, hip, femoral neck and whole body BMD measured by DXA or DPA scanning. A meta-regression investigated the effect of factors including age, duration since diagnosis, duration of amenorrhea and BMI on BMD. The mean BMI of participants was 16.65kg/m(2) (AN), 21.16kg/m(2) (BN) and 22.06kg/m(2) (HC). Spine BMD was lowest in AN subjects (SMD, -3.681; 95% CI, -4.738, -2.625; p < 0.0001), but also lower in BN subjects compared with HCs (SMD, -0.472; 95% CI, -0.688, -0.255; p < 0.0001). Hip, whole body and femoral neck BMD were reduced to a statistically significant level in AN but not BN groups. The meta-regression was limited by the number of included studies and did not find any significant predictors. This meta-analysis confirms the association between low BMD and AN and presents a strong argument for assessing BMD not only in patients with AN, but also in patients with BN.