Potential errors inherent in quantitative densitometric analysis of orthopaedic radiographs. A study after total hip arthroplasty.

We sought to determine whether x-ray densitometry could be used to provide quantitative information about the growth and resorption of bone in a retrospective analysis of the radiographs of patients who had a non-cemented hip prosthesis. We examined variables that were likely to be uncontrolled in routine orthopaedic radiographs for their effect on radiographic (film) density. Changes in film lots, exposure, target distance, field variability (heel effect), and femoral orientation were independently analyzed for radiographs of precisely oriented femora of cadavera and a radiology step-wedge. Despite attempts to normalize these variables, we found that retrospective comparison of serial radiographs of patients for densitometric analysis of change in the bone was prone to significant error. In uncontrolled radiographs, non-uniformity of exposure, intensity of the radiation field, target distance, and deviation in femoral rotation will be likely to result in substantial variations in the apparent thickness or density of the femoral cortex and make quantitative comparisons of serial radiographs unreliable.