Table 4.

Logistic regression Odds ratio (β-coefficient) and 95% CI for (A) a lower caries risk score in high risk subjects, or (B) maintain a low caries risk score in 2011 compared to at baseline (2005) and 2011; reference group in parenthesis

Variables retained in model1,2	β-coefficient	95% CI for β	P-value
(A) Odds ratio to have a lower caries risk score in 2011 if high risk in 2005 (n = 200)
Total number of visits to the dental office	0.92	0.89-0.96	<0.001
Counselling on tooth brushing with fluoridated toothpaste (lowest number odds = 1)	1.44	1.11-1.86	0.006
Counselling and training on tooth cleaning (lowest number odds = 1)	0.59	0.42-0.82	0.002
(B) Odds ratio to maintain a low/no caries risk score from 2005 to 2011 (n = 200)
Sex (women odds = 1)	0.34	0.17-0.68	0.002
Age group (youngest ten-year age group odds = 1)	1.05	1.02-1.09	0.005
Total number of visits to the dental office	0.86	0.80-0.91	<0.001
Counselling and training of tooth cleaning (lowest number odds = 1)	1.39	0.96-1.99	0.0802

¹⁾The basic model included sex, ten-year age group at 2005, living region, and total number of visits to the dental office, and numbers of visits with counselling on tooth brushing with fluoridated toothpaste, additional fluoride, dietary habits, and/or oral hygiene instruction. Variables not shown in the table did not meet the criteria of a probability <0.10 in the final step. The reference (odds ratio = 1) is to A) have no improvement and B) to not have a maintained low/no risk.

²⁾Models restricted to one counselling type at a time (covariates sex, ten-year age group at 2005, living region, and total number of visits to the dental office) confirmed the results from the basic model and reached statistical significance for counselling and training of tooth cleaning (p = 0.031) in section B.