Table 3 Effect of allostatic load on the social gradients in the extent of periodontal diseases variables in the American population 1988–1994 (NHANES III)

	Education groups			Poverty : income ratio
	>12 years	12 years	<12 years	Poverty : income ratio
Extent of gingival bleeding (N = 4295)†
Model 1 regression coefficient (95% CI)	0	2.71*** (1.70 to 3.73)	5.40*** (3.27 to 7.51)	−0.86*** (−1.26 to −0.47)
Model 2 regression coefficient (95% CI)	0	2.13*** (1.03 to 3.24)	4.58*** (2.53 to 6.63)	−0.84*** (1.24 to −0.44)
Extent of loss of periodontal attachment (N = 4295)†
Model 1 regression coefficient (95% CI)	0	4.36*** (2.33 to 6.40)	10.58*** (7.78 to 13.38)	−0.89*** (−1.46 to −0.32)
Model 2 regression coefficient (95% CI)	0	4.14*** (2.07 to 6.21)	10.27*** (7.45 to 13.09)	−0.86** (−1.42 to −0.30)
Extent of periodontal pocket depth (N = 4295)†
Model 1 regression coefficient (95% CI)	0	0.95* (0.15 to 1.75)	2.43*** (1.18 to 3.68)	−0.37*** (−0.54 to −0.20)
Model 2 regression coefficient (95% CI)	0	0.81* (0.01 to 1.61)	2.24*** (1.09 to 3.39)	−0.36*** (−0.54 to −0.19)

*p<0.05, **p<0.01, ***p<0.001.
†Extents of gingival bleeding, loss of attachment and pocket depth, all continuous variables. Model 1 controls for education, poverty : income ratio, medical insurance, sex, ethnicity, age and smoking. Model 2 additionally controls for the aggregated allostasis variable.