Skip to main content
. 2020 May 19;20:147. doi: 10.1186/s12903-020-01125-3

Table 4.

Selected variables using training data shown to be associated with gingival bleeding by multi-level logistic regression analyses (n = 3394)

Variables Categories Null model Model with individual and contextual variables
OR (95% CI) p
Fixed effect
Individual level
hukou 1
Rural 0.61 (0.52, 0.72) < 0.01
Family size Urban 1
One child 1.41 (1.19, 1.68) < 0.01
Father’s educational level More than one child 1
Middle school or lower a 0.53 (0.45, 0.63) < 0.01
Mother’s educational level More than middle schoolb 1
0.71 (0.59, 0.86) < 0.01
Brushing frequency (times/day) 1
2.03 (1.46, 2.82) < 0.01
0.93 (0.70, 1.23) 0. 60
0.35 (0.26, 0.48) < 0.01
Dental floss use 1
0.58 (0.41, 0.83) < 0.01
Sugar-containing soft drink/Pop/ milk/yogurt/tea/coffee/water 1
2.11 (1.80, 2.49) < 0.01
Dental visit (previous year) 1
1.44 (1.19, 1.74) < 0.01
Random effect
Contextual level variance (SEc) 0.20 (0.12) 0.13 (0.08) 0.10
AICd 4562.8 4051.4
Average intercept −0.173 −0.02246
VPCe 0.0469 0.031252

Abbreviations: aMiddle school or lower, ≤9 years; bMore than middle school, > 9 years; cSE standard error, dAIC Akaike Information Criteria, eVPC variance partitioning coefficient attributable to the second level (regions)