Table 3

 Multiple linear regression analysis of BMD (g/cm2) at different sites of measurement (dependent variable), demographic and disease variables (independent variables)

Femoral neckSpine L1–4
βSE95% CIβSE95% CI
BMD, bone mineral density; SE, standard error; CI, confidence interval; BMI, body mass index; R2, percentage of total variance explained in the model. Numbers needed for regression equation: original study: 1 = remission cohort, 2 = RAPIT trial; sex: 0 = women, 1 = men; corticosteroid use: 0 = never, 1 = ever. §P<0.05; ‡P<0.01.
Original study−0.0610.017−0.095 to −0.027‡−0.0070.021−0.049 to 0.035
Sex−0.180.084−0.34 to −0.016§−0.410.10−0.61 to −0.21‡
Age (years)−0.0050.001−0.006 to −0.004‡−0.0060.001−0.007 to −0.004‡
Interaction (age × sex)0.0040.0010.001 to 0.007‡0.0080.0020.005 to 0.012‡
BMI (kg/m2)0.0060.0010.004 to 0.009‡0.0050.0020.002 to 0.009‡
Corticosteroid use−0.0280.023−0.062 to 0.007−0.0380.021−0.080 to 0.005
Larsen score: hands and feet−0.0004<0.001−0.001 to 0.000‡−0.0002<0.001−0.001 to 0.000
R22217