To investigate whether simple measurements made on conventional pelvis plain film could predict hip fractures independently of bone mineral density.

The authors analyzed the simple radiographs and dual x-ray absorptiometry scan of women over the age of 60 years who had sustained a neck fracture (n=40), trochanteric fracture (n=40) and non-fracture volunteers (n=40).

In an age-adjusted univariate logistic regression model, the most reliable risk factor of hip fracture in bone mass was bone mineral density of the femoral neck. In the measurements of bone architecture, the most important risk factor was Singh index and in bone geometric measurements, it was hip axis length. There were no statistical differences in the measurements between neck fractures and trochanteric fractures. In a multivariate model, these three variables were shown to be significant improvements in the assessment of hip fracture risks.

The authors concluded that simple measurements of proximal femoral geometry made on pelvic radiographs could predict hip fracture as well as bone density of the hip.