Table 3. The influence of the periodontal status on incident type 2 diabetes in modified Poisson regression analysis.

	Adjusted Relative Risk for incident diabetes (95% confidence interval)
	Model 1		Model 2
Independent Variables	A	B	A	B
Prediabetes	‒	5.89 (4.06 ‒ 8.56)‡	‒	5.94 (4.08 ‒ 8.63)‡
(reference: no prediabetes)
Gingival hemorrhage	1.32 (0.95 ‒ 1.85)	1.23 (0.90 ‒ 1.70)	‒	‒
(reference: no gingival hemorrhage)
Tooth loosening	‒	‒	1.73 (1.14 ‒ 2.64)*	1.73 (1.18 ‒ 2.53)†
(reference: no tooth loosening)

*: P<0.05

†: P<0.01

‡: P<0.001.

BMI: body mass index. Model 1 used gingival hemorrhage as an oral status indicator. Model 2 used tooth loosening as an oral status indicator. A: not including prediabetes as a covariate. B: including prediabetes as a covariate. The other covariate consisted of age, current smoking habits, BMI, family history of diabetes, hypertension, alcohol heavy consumption (≥ 40 g/day), and exercise habits (> 30 minutes, ≥ 2 days/week).