Geographical variation in DXA bone mineral density in young European men and women. Results from the Network in Europe on Male Osteoporosis (NEMO) study

We collected population-based young normal hip and spine BMD data from 17 centres across Europe to assess between centre differences and to compare reference values with the US NHANES-III data. There was strong evidence of between country heterogeneity, but not between centres within countries. Hip BMD mean values were lower in European women, but SD's differed little from the NHANES-III USA results in both sexes. It may be necessary to adjust NHANES-III based T-scores by adding/subtracting a country-specific adjustment factor. INTRODUCTION: It remains unclear whether young normal BMD reference values specific to an American population can be validly used for T-score calculation in Europeans. METHODS: We collected population based BMD data from 1163 men and 329 women aged 19-29 years from 17 centres across Europe to compare mean and SD values with the NHANES-III study USA results. BMD(g/cm2) was measured at the hip and spine using DXA densitometers cross-calibrated with the European Spine Phantom (ESP). The only exclusions were for technically inadequate scans. A linear regression model was used to derive reference values. To allow for direct comparison with published NHANES III study data, the cross-calibrated BMD values were converted using the ESP equations to Hologic QDR 1000 units. RESULTS: In men, the overall mean(SD) BMD values expressed in Hologic-QDR1000 units of measurement, were: femoral neck 0.912(0.132); trochanter 0.793(0.124); and L2-L4 spine 1.027(0.123). The respective estimates in women were: 0.826(0.115); 0.670(0.093); and 0.983(0.107). However the I2 statistic for heterogeneity indicated moderate to strong evidence of between-centre heterogeneity. There was, however, no significant heterogeneity observed between centres within countries, suggesting that this variation arose from national differences. Compared to the NHANES III population-based US data, the mean values in women were significantly lower at both sites due to some lower national European means. However, at all sites and in both sexes the SD's were very similar between the US and Europe. There was some evidence that recruiting volunteers resulted in biased values in women. CONCLUSION: Our T-score normal values for the lumbar spine (L2-L4) should be more reliable for spine-specific risk assessment than some non-representative normal ranges, and should be evaluated for that purpose in Europe. If T-scores are to be used to compare individual data with ranges seen in normal young subjects of the same nationality, it may be necessary to adjust femoral NHANES III-based T-scores by adding (or subtracting) a country-specific adjustment factor. In risk assessment it is probably sufficient to use NHANES III-based hip T-scores, as supplied for the hip by densitometer manufacturers, interpreting them in light of recent international meta-analysis data on the relationship between BMD and fracture risk.