Background Low bone mineral density (BMD) is a strong determinant of fracture which occurrence is related to increased mortality, mainly in men1. Several clinical decision rules have been developed to identify men with higher probability of low BMD, trying to create an efficient process of referring patients to dual-energy Xx-ray absorptiometry (DXA) evaluation2. However, clinicians tend to avoid these rules because the lack of consensus regarding their accuracy among men.
Objectives To identify and assess the accuracy of published clinical screening tools designed to identify men with low BMD.
Methods We systematically searched Medline through August 2013 for any study describing the validation of instruments for low bone mass prediction in men and we hand searched reference lists of included articles. Data was extracted on participants' characteristics, DXA features and tools properties in terms of risk factors considered in the predictive models and their discriminatory performance (area under the ROC curve, sensitivity, specificity). Methodological quality of studies was assessed using a modified QUADAS checklist.
Results A total of 1484 citations were initially identified and 22 articles were included in our analysis. Fourteen different screening tools have been identified: 5 guidelines proposed by health entities, 7 newly instruments developed specifically for men and 2 tools initially created for women and then validated in men. Only 3 of the 14 identified tools were validated more than once – OST, OSTA and MORES. When predicting a femoral neck T-score<-2.5, with different tools cut-offs tested, OST had an AUC ranging between 0.740-0.990, sensitivity 6-100% and specificity 51-94%; AUC of OSTA ranged between 0.848-0.850, sensitivity and specificity between 83-91% and 66-67%, respectively; and MORES presented similar psychometric characteristics (ranges: AUC, 0.820-0.842; sensitivity, 80-95%; specificity, 61-70%). None of the cut-offs tested for each tool score was evidently more accurate than the alternatives. Moreover, there was high heterogeneity across studies regarding source populations, age distribution, ethnic background, low BMD diagnostic criteria and DXA equipment contributing to the different ability of tools in correctly identify low BDM. A global moderate quality of reports was observed with a mean of 10.8 items (range: 8-15) in 19 possible modified QUADAS items.
Conclusions Only OST, OSTA and MORES were validated in more than one sample. The tools had acceptable predictive capacity and performed similarly in terms of overall accuracy but no clear cut-offs to predict low BDM, for any score, emerged from our analysis. OST and MORES, for their simplicity, accuracy and replication, seem to be more adequate for routine clinical practice use.
Kannegaard, P.N., et al., Excess mortality in men compared with women following a hip fracture. National analysis of comedications, comorbidity and survival. Age Ageing, 2010. 39(2): p.203-9.
Schwartz, E.N. and D.M. Steinberg, Prescreening tools to determine who needs DXA. Curr Osteoporos Rep, 2006. 4(4): p.148-52.
Disclosure of Interest None declared