Table 3.

Linear multivariable regression models predicting height loss over time intervals 1968–1980, 1980–1992 and 1992–2005

	Height loss (1968–1980) N=751		Height loss (1980–1992) N=557		Height loss (1992–2005) N=330
	Coefficient (SE)	P value	Coefficient (SE)	P value	Coefficient (SE)	P value
Severe erosion	0.20 (0.18)	0.27	0.48 (0.19)	0.011	0.43 (0.23)	0.050
Height at baseline	0.01 (0.01)	0.22	0.02 (0.01)	0.051	0.03 (0.02)	0.098
Birth year	−0.05 (0.01)	<0.001	−0.06 (0.01)	<0.001	−0.02 (0.03)	0.551
Physical activity	−0.11 (0.06)	0.07	−0.13 (0.10)	0.094	−0.019 (0.17)	0.907
Smoking	−0.02 (0.01)	0.27	0.06 (0.03)	0.031	0.120 (0.07)	0.072
Education	0.02 (0.07)	0.81	0.15 (0.12)	0.218	0.25 (0.23)	0.273
BMI	0.01 (0.01)	0.34	0.001 (0.02)	0.929	−0.015 (0.03)	0.609
Sparse trabeculation	0.16 (0.08)	0.038	0.24 (0.11)	0.023	0.45 (0.22)	0.040
Height at baseline	0.01 (0.01)	0.056	0.02 (0.01)	0.021	0.03 (0.02)	0.131
Birth year	−0.06 (0.01)	<0.001	−0.06 (0.01)	<0.001	−0.02 (0.03)	0.450
Physical activity	−0.14 (0.07)	0.019	−0.12 (0.07)	0.106	−0.10 (0.16)	0.528
Smoking	−0.02 (0.02)	0.320	0.018 (0.02)	0.439	0.13 (0.05)	0.011
Education	−0.02 (0.07)	0.773	0.14 (0.11)	0.190	0.20 (0.23)	0.388
BMI	0.012 (0.01)	0.239	0.02 (0.02)	0.165	−0.01 (0.03	0.717

Cortical erosion was dichotomised into severe erosion and normal (mild or moderate) erosion. The trabeculation was classified into sparse or normal (mixed or dense).

BMI, body mass index.