MEASUREMENT OF PERIPROSTHETIC BONE MINERAL DENSITY USING WINDOWS VERSUS DOS BASED ANALYSIS SOFTWARE.

Abstract

Dual energy X-ray absorptiometry (DXA) is a precise tool for measuring bone mineral density (BMD) around total joint prostheses. The Hologic ‘metal-removal hip’ analysis package (Hologic Inc, Waltham, Massachusetts) is a DOS-based analysis platform that has been previously validated for measurement of pelvic and proximal BMD after total hip arthroplasty (THA). This software has undergone a change in the operating platform to a Windows-based system that has also incorporated changes to DXA image manipulation on-screen. These changes may affect the magnitude of random error (precision) and systematic error (bias) when compared with measurements made using the previously validated DOS-based system. These factors could influence interpretation of longitudinal studies commenced using the DOS system and later completed using the Windows system. The aims of this study were to compare the precision and bias of pelvic and femoral periprosthetic BMD measurements made using the Windows versus the DOS analysis platform of the Hologic ‘metal-removal hip’ software. A total of 29 subjects (17 men and 12 women) with a mean age of 51years (SD±10), who had undergone hybrid THA using a cemented stem and uncemented cup. Subjects underwent duplicate DXA scans of the hemipelvis and proximal femur taken on the same day after a period for repositioning.. Scans were obtained with the patient lying supine in the scanner with the legs in extension and the foot in a neutral position. Scans were carried out using the same Hologic QDR 4500-A fan-beam densitometer in ‘metal-removal hip’ scanning mode. The DXA scan acquisitions were analysed using both the DOS and the Windows versions of the analysis software. The same observer made all analyses (NRS). Pelvic scans were analysed using a four region of interest model and femoral scans were analysed using a seven region of interest model. Precision was expressed as coefficient of variation (CV%) and compared between methods using the F-test. Systematic bias was examined using the Bland and Altman method and paired t-test. The CV% for the pelvic regions of interest (n=4) varied from 3.92 to 8.54 and from 2.36 to 5.96 for the Windows and DOS systems, respectively. The CV% for the net pelvic region was 3.04 and 2.36 for Windows versus DOS, respectively (F- test, p> 0.05). The CV% for the femoral regions of interest (n=7) varied from 1.58 to 4.14 and from 1.84 to 4.65 for the Windows and DOS systems, respectively. The CV% for the net femoral region was 1.75 and 1.51 for Windows versus DOS, respectively (F- test, p> 0.05). Absolute BMD values for the net pelvic region were similar (Bland-Altman, Windows minus DOS value mean = -1.0%, 95% CI −7.5 to 5.6; t-test p.0.05). Absolute BMD values for the net femoral region were also similar (Bland-Altman, Windows minus DOS value mean = 1.3%, 95% CI −8.3 to 10.8; t-test p.0.05). In summary precision of the measurements using the 2 operating systems was similar and there was no systematic bias between methods. These data suggest that scans analysed using each platform may be used interchangeably within the same study subjects, without the need of a calibration correction.