Prevalence and risk factors for bone loss in rheumatoid arthritis patients from South China: modeled by three methods

Table 3 Odd ratios of variables from selective models of 3 measurement sites

Detective sites	Osteopenia			Osteoporosis
Detective sites	Variables	Odd ratios (95%CI)	p	Variables	Odd ratios (95%CI)	P
Lumber spine^a	Age	1.15 (1.10, 1.21)	< 0.001	Age	1.25 (1.18, 1.33)	< 0.001
	BMI	0.88 (0.77, 0.99)	0.038	BMI	0.68 (0.57, 0.79)	< 0.001
	Serum 25(OH)D3 level	0.99 (0.98, 1.00)	0.030	Serum 25(OH)D3 level	0.97 (0.96, 0.98)	< 0.001
	TNFi usage in the last one year	0.27 (0.08, 0.84)	0.027	Serum 25(OH)D3 level
Femoral neck^b	Age	1.17 (1.12, 1.22)	< 0.001	Age	1.26 (1.18, 1.36)	< 0.001
	BMI	0.85 (0.77, 0.95)	0.003	BMI	0.62 (0.50, 0.75)	< 0.001
	Rheumatoid factor concentration	1.00 (0.99, 1.02)	0.183	Serum 25(OH)D3 level	0.98 (0.96, 0.99)	0.002
	Rheumatoid factor concentration	1.00 (0.99, 1.02)		sUA	1.00 (1.00, 1.01)	0.068
				Disease duration	1.00 (0.89, 1.13)	0.967
				Serum phosphate level	1.87 (0.48, 29.95)	0.631
Total hip^c	Age	1.11 (1.08, 1.15)	< 0.001	Age	1.15 (1.10, 1.21	< 0.001
	BMI	0.84 (0.77, 0.92)	< 0.001	BMI	0.68(0.58,0.78)	< 0.001
	TNFi usage in the last one year	0.37 (0.14, 0.93)	0.040	sUA	1.01(1.00, 1.01)	0.001

BMI Body mass index, 25(OH)D3 Calcifediol, TNFi Tumor necrosis factor-α inhibitor, sUA Serum uric acid level
^aPreselected logistic regression (models A) were optimal for both osteopenia and osteoporosis in lumber spine
^bLASSO (models B) were optimal for both osteopenia and osteoporosis in femoral neck (R)
^cModel A and B were optimal for osteopenia and osteoporosis respectively in total hip

ISSN: 1471-2474