Comparison of spine and femur reference data in native Chinese women from different regions of China

The aim of this study is to explore the differences of BMD reference curves at various skeletal sites among Chinese women from different regions of China and to investigate the feasibility of establishing a unified national BMD reference database for Chinese women. We measured BMD at the posteroanterior (PA) lumbar spine, femoral neck, trochanter and Ward's triangle by dual-energy X-ray absorptiometry bone densitometer in 3,422 Changsha women of South Central China, aged 20-84 years. The documented BMDs of reference populations of women in all other areas included Shanghai ( n =2,111) and Nanjing ( n =3,174) in the East, Shenyang ( n =1,213) in the Northeast, Kunming ( n =523) in the Southwest, Chongqing ( n =811) in the Midwest and Xian ( n =1,320) in the Northwest. We adopted the cubic regression as the fitting model for reference curves of BMD that varied with age, conducted conversions of BMD measured by various bone densitometers from different manufacturers and compared the differences between standardized BMD (sBMD) reference curves and combined ones for women from different areas. Our results revealed that by comparing variances in women from different areas, the average variances of non-standard BMD were 0.8-30.8% at the PA spine, 0.7-24.5% at the femoral neck, 0.6-29.9% at the trochanter and 1.1-54.7% at Ward's triangle, while average variances of sBMD either significantly decreased or disappeared (0.8-3.9% at the PA spine, 0.7-8.6% at the femoral neck, 0.6-8.3% at the trochanter and 1.1-29.9% at Ward's triangle). The sBMD reference curves were highly positive-dependent with combined ones ( r =0.913-0.999, P =0.000). At the PA spine and trochanter, the effect of combined sBMD curves presented well in women from different areas, except for those from Shanghai at the PA spine and Shenyang at the trochanter, indicating that sBMD curves were significantly different from pooled ones; at the femoral neck and Ward's triangle, the effect of combined sBMD reference curves was poor, indicating that sBMD curves demonstrated significant differences from pooled ones in women from a majority of these areas. We conclude that, in high density population areas, sBMD reference curves showed no significant geographic differences in women from various regions. In women from different areas, sBMD reference curves present good pooled results at the PA spine and trochanter. The less ideal combining effect of the sBMD curves at both femoral neck and Ward's triangle might be caused by the intrinsic differences from the different measuring instruments.