Overall, 6,825 children (52.76% females) were examined, of which 882 (12.92%) were of non-European background. The distribution of caries figures for both European and non-European children across the five different Italian areas (Figure 1) showed that caries prevalence was statistically significant higher in children with non-European background compared to European children (72.59% vs 41.62% respectively χ2= 297.29 p<0.01).
A not equal distribution by sex across both European and non-European groups was recorded; a higher caries prevalence in females in the non-European group (80.79%) was found, while in European background sample a higher prevalence in males (61.12%) was observed (data not in table).
Descriptive and bivariate analysis (by European and non-European background) is displayed in Table 1.
Table 1. Caries experience in European and Non-European children across Geographical areas, Household, Parents’ occupation and educational level.
|
|
European
|
Non-European background
|
|
|
Caries-free
|
Caries-Ex
|
OR (95%CI)
|
Caries-free
|
Caries-Ex
|
OR (95%CI)
|
|
|
n (%)
|
n (%)
|
|
n (%)
|
n (%)
|
|
Geographical areas*
|
North-West
|
1037 (65.18)
|
554 (34.82)
|
reference
|
88 (28.85)
|
217 (71.15)
|
reference
|
North-East
|
662 (55.26)
|
536 (44.74)
|
1.52 (1.29-1.77)
|
4 (4.35)
|
88 (95.65)
|
8.92 (3.07-25.90)
|
Central
|
614 (53.67)
|
530 (46.33)
|
1.62 (1.38-1.89)
|
42 (28.97)
|
103 (71.03)
|
0.99 (0.64-1.54)
|
South
|
685 (50.15)
|
681 (49.85)
|
1.86 (1.60-2.16)
|
79 (34.96)
|
147 (65.04)
|
0.75 (0.52-1.09)
|
Islands
|
471 (73.25)
|
172 (26.75)
|
0.68 (0.55-0.84)
|
29 (25.22)
|
86 (74.78)
|
1.20 (0.74-1.96)
|
|
Mantel Haenszel trend of odds χ2=3.94 p= 0.03
|
Mantel Haenszel trend of odds χ2=1.55 p= 0.22
|
Household
|
1st quartiles
|
511 (42.13)
|
702 (57.87)
|
8.68 (6.66-11-31)
|
106 (24.65)
|
324 (75.35)
|
22.41 (8.50-59.14)
|
2nd quartiles
|
646 (36.85)
|
1107 (63.15)
|
10.83 (8.35-14.03)
|
92 (22.83)
|
311 (77.17)
|
24.79 (9.22-66.63)
|
3rd quartiles
|
1674 (74.83)
|
563 (25.17)
|
2.12 (1.68-2.68)
|
44 (88.00)
|
6 (12.00)
|
reference
|
4th quartiles
|
638 (86.33)
|
101 (13.67)
|
reference
|
--
|
--
|
--
|
|
Mantel Haenszel trend of odds χ2=735.80 p<0.01
|
Mantel Haenszel trend of odds χ2=28.18 p<0.01
|
Parents’ occupation
|
Un-employment/unskilled
|
559 (48.31)
|
598 (51.69)
|
3.28 (2.83-3.79)
|
235 (29.71)
|
556 (70.29)
|
0.16 (0.04-0.67)
|
Skilled jobs/qualified
|
429 (39.94)
|
645 (60.06)
|
4.60 (3.94-5.37)
|
5 (8.33)
|
55 (91.67)
|
0.73 (01.13-4.05)
|
White-collar
|
2319 (75.37)
|
758 (24.63)
|
reference
|
2 (6.25)
|
30 (93.75)
|
reference
|
|
Mantel Haenszel trend of odds χ2=382.30 p<0.01
|
Mantel Haenszel trend of odds χ2=18.37 p<0.01
|
Parents’ education
|
Low
|
187 (41.93)
|
259 (58.07)
|
2.12 (1.73-2.59)
|
209 (28.32)
|
529 (71.68)
|
1.04 (0.64-1.67)
|
Intermediate
|
1181 (58.41)
|
841 (41.59)
|
1.09 (0.97-1.22)
|
6 (11.54)
|
46 (88.46)
|
3.13 (1.17-8.40)
|
High
|
2101 (60.48)
|
841 (41.59)
|
reference
|
27 (29.03)
|
66 (70.97)
|
reference
|
|
Mantel Haenszel trend of odds χ2=37.01 p<0.01
|
Mantel Haenszel trend of odds χ2=0.38 p=0.54
|
Odds Ratio (OR) and 95%Confidence Interval (95%CI). The Mantel Haenszel trend of odds was calculated. *Geographical areas were ordered following the mean Gross National Product (GNP) per capita for each Italian area.
The caries prevalence was statistically associated (Mantel Haenszel trend of odds χ2=3.94 p=0.03) to the area of living ordered by GNP among European children: the caries prevalence increases as the GNP decreases. In children with a non-European background, caries prevalence was not associated with the area of living. The caries prevalence (CariesEx) was statistically significant different across the different household income quartiles, the parents’ occupation both in children with European and non-European background (p<0.01). Caries figure was also statistically significant associated (p<0.01) to behavioral factors: Breastfeeding, Pacifier at night, Brushing frequency, Cariogenic diet, Smoking habit of parents [see Additional file 1].
The social levels, by equivalized household income, for caries experience and cavitated lesions prevalence for both European and non-European children are clearly demonstrated in Figure 2: children in the highest household income group had lower levels of caries (either caries experience and cavitated lesions prevalence) than their counterparts in the lowest income group.
The coefficients from the multivariate analysis with negative values indicate that the social disparity in caries favors the more socially advantaged children (Table 2).
Table 2. Multivariate regression coefficients of caries prevalence.
European background
Number of observations = 5,942
Caries observations (Non-zero observations)=2,473
Caries-free observations (Zero observations)=3,469
Log likelihood = -3602.97 Likelihood Ratio χ2(5)=311.48 p<0.01
Covariate
|
IRR±Std.Err. (p-value)
|
95%CI
|
Parents’ occupation level (White-collar)
|
0.80±0.03 (<0.01)
|
0.74 / 0.86
|
Parents’ educational level (High)
|
1.08±0.05 (0.06)
|
1.01 / 1.17
|
Number of kids in the family (One child)
|
0.95±0.11(0.03)
|
0.72 / 0.98
|
Area of living (North-West)
|
0.86±0.01 (<0.01)
|
0.84 / 0.90
|
Household income (4th quartile)
|
0.73±0.15 (<0.01)
|
0.69 / 0.78
|
constant
|
0.22±0.12 (0.07)
|
-0.02 / 0.46
|
Inflate
Gender
|
23.05
|
|
Gender (Females) Constant
|
-23.35±0.11 (<0.01)
|
-23.57 / -23.13
|
Natural log of alpha
|
-147.17±112.96 (0.21)
|
-362.57 / 80.21
|
IRR=Coefficients of the Zero-inflated negative binomial logistic model transformed to incidence-rate ratios
constant= The zero negative binomial regression estimate;
alpha= The estimate of the dispersion parameter.
Non-European background
Number of observations = 883
Log likelihood = -468.65 Likelihood Ratio χ2(5)=99.80 p<0.01
Covariate
|
Odds Ratio ±Std.Dev (p-value)
|
95%CI
|
Parents’ occupation level (White-collar)
|
1.59±1.43 (<0.01)
|
1.08 / 2.33
|
Parents’ educational level (High)
|
1.32±0.18 (0.04)
|
1.02 / 1.73
|
Number of kids in the family (One child)
|
1.49±0.15 (<0.01)
|
1.22 / 1.81
|
Area of living (North-West)
|
0.85±0.04 (<0.01)
|
0.74 / 0.94
|
Household income (3rd quartile)
|
0.35±0.06 (<0.01)
|
0.26 / 0.48
|
Gender (Female)
|
2.19±0.40 (<0.01)
|
1.54 / 3.13
|
constant
|
1.27±0.66 (0.651)
|
0.45 / 3.49
|
Taking into consideration of the different caries prevalence in the two population (42.61 % in European children and 72.59% in children with non-European background), it was decided to run a Zero-inflated negative binomial logistic model with European background children and a logistic model in non-European children.
In European children, the estimated coefficients (Parents’ occupation level, Number of kids in the family, Area of living and Household income) of the Zero-inflated negative binomial logistic model transformed to rate ratios, were statistically significant associated (p<0.01) to caries prevalence. White-collar parents, family living in North-West and highest household income (4th quartile) played a protective effect on caries prevalence. For the non-European children, all the social background factors were statistically significant associated to caries prevalence (p<0.01, except for Parents’ educational level, p=0.04) with an increment of risk in children with white-collar parents and parents with a high education level, while a protective effect was observed for highest quartile of household (3rd quartile) and highest GDP-related area (North-West area). All behavioral and life-style habits considered in the survey were statically significant associated (p<0.01) to caries prevalence in children with European background, while in non-European children the use of pacifier at night was not statistically significant associated to caries prevalence [see Additional file 2].
Overall the social gradient is clearly able to demonstrate how the worst social factors expose children to a greater risk for each caries level (except cavitated) and the overall caries prevalence (data not in table). The social gradient was statistically significant associated (p<0.01) to the different caries levels and experience in children with European background (Table 3).
Table 3. Association between Social Gradient (from best to worst) and Caries levels (Enamel lesions, Pre-cavitated lesions, Cavitated lesions) and total Caries prevalence adjusted by areas of living.
European background
Social Gradient
|
Enamel Lesions
|
Pre-cavitated Lesions
|
Cavitated Lesions
|
Overall Caries prevalence
|
|
OR (95%CI)
|
OR (95%CI)
|
OR (95%CI)
|
OR (95%CI)
|
Best
|
reference
|
reference
|
reference
|
reference
|
Good
|
1.34 (1.17-1.54)
|
3.22 (2.35-4.40)
|
4.30 (3.61-5.13)
|
3.24 (2.88-3.66)
|
Bad
|
1.31 (1.04-1.64)
|
5.81 (3.74-9.03)
|
2.87 (2.27-3.64)
|
2.78 (2.31-3.34)
|
Worst
|
3.97 (1.14-13.83)
|
6.61 (0.90-48.60)
|
6.42 (1.99-20.75)
|
16.77 (3.56-83.74)
|
Mantel Haenszel trend of odds
|
χ2=19.02 p< 0.01
|
χ2=60.28 p< 0.01
|
χ2=203.23 p< 0.01
|
χ2=325.97 p< 0.01
|
Non-European background
Social Gradient
|
Enamel Lesions
|
Pre-cavitated Lesions
|
Cavitated Lesions
|
Overall Caries prevalence
|
|
OR (95%CI)
|
OR (95%CI)
|
OR (95%CI)
|
OR (95%CI)
|
Best
|
reference
|
reference
|
reference
|
reference
|
Good
|
--
|
--
|
3.62 (0.10-124.40)
|
19.87 (0.04->1000)
|
Bad
|
--
|
--
|
0.23 (0.01-7.43)
|
--
|
Worst
|
--
|
--
|
0.52 (0.11-2.35)
|
7.53 (0.33-170.47)
|
Mantel Haenszel trend of odds
|
--
|
--
|
χ2=0.37 p= 0.54
|
χ2=23.32 p< 0.01
|
Odds Ratio (OR) and 95% Confidence Interval (95%CI) and the Mantel Haenszel trend of odds were calculated.
For enamel lesions and pre-cavitated lesions very few observations were present so the association between Social Gradient was not run.
Subjects with the worst gradient had usually the highest Odds-ratios for caries prevalence. As caries experience was so high in children with non-European background, no social gradient was detected.