Validation Of Cariogram In Caries Prediction At Woman And Their Children 4 Years After Pregnancy – Longitudinal Randomized Control Study

Aim To evaluate Cariogram caries risk assessment during pregnancy with DMFT/dmft incidence at mother and children 4-years later.Background To validate baseline caries risk classifications according to Cariogram for pregnant women with the actual DMFT development after 4-year and the second aim was to validate assessed caries risk at pregnant women according to Cariogram with caries development at their children four years after. Methods The study population consisted of 96 pregnant women (average age 27, 4 years at baseline) that completed clinical baseline examination and salivary tests. A caries risk assessment was made using Cariogram model and according to it, participants were divided into five risk groups. The follow-up study was initiated 4 years later and the 80 pairs of mother and children (from that pregnancy) were re-examined using the same procedure as did at baseline. An individual caries risk profile and DMFT/dmft incidence were made for each woman and child. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for two cut-off were calculated to express the outcome. Results There was a strong association between the risk categories of pregnant women and their offspring as well as between caries development in children and the Cariogram risk categories of their mother in pregnancy. Sensitivity and PPV for new DMFT (ΔDMFT > 0) over 4 years for women was high (>80%) for those participants assessed with 0-60% “chance to avoid caries“, as well as diagnostic accuracy (74.00%). High specificity (91.00%), very high PPV (95.00%) and clinically useful values according to Youden’s index (0.53) were obtained for moderate and two lowest risk groups for dmft in children. Conclusions With the limits of this study, Cariogram was valid and highly predictive in caries risk assessment in children based on caries risk assessment of their mother in pregnancy. The Cariogram model can be useful tool for caries prediction in both groups of woman and their children based on caries risk assessment during pregnancy.

Conclusions With the limits of this study, Cariogram was valid and highly predictive in caries risk assessment in children based on caries risk assessment of their mother in pregnancy. The Cariogram model can be useful tool for caries prediction in both groups of woman and their children based on caries risk assessment during pregnancy.

Background
Caries is multifactorial disease and it is determined by the interaction of different microbial, genetic, immunological, behavioral, as well as environmental factors. The most common chronic disease in childhood is early childhood caries (EEC) that occurs in children below 71 months of age. It still remains huge problem both in developed and developing countries [1,2]. Many studies have demonstrated the association of MS (mutans streptococci) with ECC [3][4][5][6]. A systematic reviews confirmed that maternal factors influence bacterial acquisition, whereas colonization was mostly mediated by oral health and feeding habits [7,8] during the first year of life. Cross-sectional [9] and longitudinal [10] studies have corroborated the association between parental dental status and offspring caries.
Identifying children at greatest risk of caries in order to design appropriate preventive activity has been a goal of the dental profession for many years. It is expected that multiple factors, combined on an appropriate scale and accounted for possible interactions may improve the prediction of caries risk. The challenge is to develop really effective model for predicting caries risk. The Cariogram®, an algorithm based software, assesses an individual's caries risk profile and contains many "if "conditions -it can operate with 5 million combinations of caries related factors [11].The program offers recommendations for targeted preventive measures that should be implemented to avoid the formation of new caries lesions [12,13].
Cariogram has been used to assess the caries risk profile of schoolchildren [14,15], teenagers, young adults [16][17][18][19][20], orthodontic [21] and elderly patients [22]. Some studies used Cariogram to assess caries risk profile of preschool children [23][24][25][26][27][28]. In a previous study, we presented that Cariogram can be useful toll in caries risk assessment and identification of significant caries risk factors in a group of pregnant woman [29], but to our knowledge, there are no studies that used Cariogram of pregnant women in offspring risk profile prediction. Therefore, the aim of this study was to evaluate baseline caries risk assessment according to Cariogram during pregnancy with the DMFT/dmft development at mother and children after 4-years.
The null hypothesis was that children's caries risk profiles by Cariogram is associated with caries risk profiles (by Cariogram) of their mothers during pregnancy.

Study group
This longitudinal study was performed as follow up of randomized control study previously presented by Dolic and al [29].The study population consisted of 96 pregnant women who were between 20 and 42 years old (average age 27, 4 years) at the start of the study. They lived in different socio-economic areas of the city of Banja Luka, Bosnia and Herzegovina. They were recruited at Pregnancy counseling center at Public Hospital Banja Luka. Among the 213 pregnant women who were initially eligible, 96 fulfilled the inclusion criteria and were included in the study. The study included only pregnant women a) who were in last trimester of a normal, single fetus pregnancy, b) who didn't have high-risk pregnancies, C) who didn't have previous medical conditions, pregnancy complications, or pregnancyrelated issues that required hospitalization, d) who did not have any chronic disease, e) who had not taken antibiotics or other drugs during pregnancy, f) who gave their written consent to participate in research.
The follow-up study was initiated 4 years later and the mother and their children (from that pregnancy) were re-examined by the same examiner, using the same procedure as that followed at baseline. The follow-up sample included 80 pairs of mother-child (83, 33 % recall rate), with 16 participants dropout-10 did not want to participate in the follow up study and 6 had moved from the area.
In Banja Luka, pregnant women and children up to the age of 15 have access to free and regular dental care at public dental clinics. At baseline and follow-up the women were given detailed written and verbal information about the outline of the study, and they signed an informed consent for themselves and their children. During the study period, the participants were not informed about any result and they received regular dental care. All the participants were residents in areas with low natural fluoride content in the drinking water supply (<0.30 mg/l).
The research has been conducted in full accordance with the 1964 Helsinki declaration and its later amendments or comparable ethical standards and ethical approval for the study was obtained from The Research Committee of Faculty of Medicine, University of Banja Luka.

5
The study consisted of five steps: a questionnaire, interview, clinical examination, saliva sampling, and assess of caries risk using the Cariogram. The questionnaire contained questions about: general health and medication, diet and oral hygiene habits including tooth brushing frequency and the use of fluoride.
All examinations were performed by one trained and calibrated examiner in the morning, between the meals and at least one hour after a meal and after teeth brushing. A plane mirror, WHO-CPI probe and air syringe were used in the clinical visual-tactile examination following WHO criteria [26]. Clearly visible lesions with cavities on tooth surfaces were classified as dental caries (i.e.d3-level cavities), whereas changes in transparency, initial enamel demineralization with intact surfaces and no cavitation were noted as intact teeth. The teeth were not professionally cleaned and no radiographs were taken. Caries prevalence was registered at the Decayed-Missed-Filled-Teeth (DMFT) level. The oral hygiene and oral biofilm amount were estimated using a mirror and periodontal probe in accordance with the Sillness and Löe plaque index.
Saliva sampling consisted of: measurement of the saliva secretion rate, measurements of saliva buffer capacity using a Dentobuff® Strip test, measurements of mutans streptococci in the saliva using a Dentocult® SM Strip mutans test and measurements of lactobacilli in the saliva using a Dentocult® LB test according to the manual of the manufacturer. Paraffin-stimulated whole saliva was collected for 5 minutes and the secretion rate was expressed as ml/min. Buffer capacity was categorized as high, medium or low. The density of colonies (Streptococcus mutans and Lactobacillus) was compared with a chart provided by the manufacturer (Orion Diagnostica, Finland).
Data for the computerized caries risk assessment were entered into the Cariogram to obtain an individual caries risk profile. The following five Cariogram categories were used: "very high risk" = 0-20% chance to avoid caries; "high risk" = 21-40% chance to avoid caries; "moderate risk" = 41-60% chance to avoid caries; "low risk" = 61-80% chance to avoid caries; and "very low risk" =81-100% chance to avoid caries.
The patient and the patient's ordinary dental team were not informed about calculated Cariogram risk category. The patient's regular dental team was responsible for preventive and restorative dental 6 care during the entire study period.

Statistical methods
All data were processed with the SPSS software (version 16.0, Chicago Ill., USA). χ2 test of contingency was used to compare the difference between groups. Parametric ANOVA and Student's test for independent samples (if the difference variance observed characteristics were not statistically significant) and nonparametric Mann-Whitney test (if the difference in the variance of the observed characteristics statistically significant) were used to compare the mean values of the characteristics.
Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were calculated from two-by-two tables. P values less than 0.05 were considered as statistically significant.

Results
Risk categories for children in regard to risk categories of their mother during pregnancy are shown in Figure 1. The results displayed that 85.00% (n = 68) of the children were in the same risk category as their mother during pregnancy (at baseline). A higher-risk category then their mother during pregnancy had 10% (n=8) of children. Decreased risk than their mother during pregnancy had only 5% (n=4) of children (P < 0.001). The distribution of women with new DMFT in relation to the baseline Cariogram risk category, mean 7 DMFT at baseline and 4 years follow up, as well as mean DMFT increment are shown in Table 1. The positive relationship was statistically significant and the women in very high and high risk categories had more DMFT than those in very low risk group (p<0.05). The difference in mean DMFT at the baseline and follow up for very low group was not statistically significant.
Sensitivity, specificity and predictive values for new DMFT (ΔDMFT > 0) over 4 years for women are displayed in Table 2. Sensitivity (Se) and specificity (Sp) of the Cariogram were compared at two cutoff points. The moderate risk group was combined with the low risk and very low group and it was taken as cut off 1. As cut off 2, it was combined the moderate, the high and very high risk groups.
Sensitivity and PPV was high (>80%) for cut-off 2, for those assessed with 0-60% "chance to avoid caries", as well as diagnostic accuracy (74.00%). When the moderate risk combined with the two lowest risk groups sensitivity, diagnostic accuracy and NPV were lower than 60%.  dJ = sensitivity + specificity -1.

LR-Likelihood ratio
The distribution of children with dmft in relation to the Cariogram risk category of their mother in pregnancy and mean dmft of children are shown in Table 3. The children of high and very high risk mothers in pregnancy had more DMFT than those in very low risk group (p<0.05). Sensitivity, specificity and predictive values for dmft (ΔDMFT > 0) in children are displayed in Table 4.
Sensitivity (Se) and specificity (Sp) of the baseline Cariogram (pregnant women) were compared at two cut-off points on the same way like for women. The moderate risk group was combined with the low risk and very low group and it was taken as cut off 1. As cut off 2, it was combined the moderate, the high and very high risk groups. High specificity (91.00%), very high PPV (95.00%) and clinically useful values according to Youden's index (0.53) were obtained for cut-off 1 (moderate and two lowest risk groups). When the moderate risk combined with the two lowest risk groups sensitivity and PPV were higher than 82%.  d J = sensitivity + specificity -1.

Discussion
Oral health has been an important aspect of adult, as well as, children well-being and quality of life.
Untreated dental caries associated with discomfort/pain can affect weight gain, normal growth and 9 child's development, therefore preventive and therapeutic measures must be based on the most current scientific and clinical knowledge available [2]. Despite the fact that the Health Insurance Fund of Bosnia and Herzegovina covers full costs of dental care for children younger than 15-years, the ECC prevalence is extremely high and the most of the lesions are untreated [2]. In countries like Bosnia and Herzegovina, with high caries prevalence, caries risk assessment is very important, but there is only data about caries risk assessment in the 12-years-old children [18] and pregnant women [29].
In the present study, the majority of participants (women and children) were at high risk of caries sensitivity, and specificity [33]. Often, it may be very impractical to be achieved simultaneously simplicity and accuracy. But in more recent years, sensitivity/specificity, predictive powers and Youden's index have been applied for validation in Cariogram studies [34][35][36].
The principle of sensitivity / specificity is to use a specific cut-off value for the factor under investigation and to define a specific outcome of the test. In present study, the relatively high years-old preschool children resulted in high sensitivity for future caries 5 years later, but the method lacked precision and accuracy [24].
The predictive ability of Cariogram for pregnant women at both cut-off points had specificity lower than 70%, which may categorized in high-risk group some individuals with actual low caries risk and unnecessary preventive measures may be taken. But, on the other hand, according to PPV (>78%) in both cut-offs, it can be seen that a higher percentage of patients will actually develop caries. PPV is even higher (95%-cut off 1, 85%-cut off 2) for a possibility of caries developing in offspring if pregnant women is in the higher risk of caries according Cariogram.
In recent studies from Hong Kong, the results showed that the Cariogram for preschool children generally exhibited a higher accuracy [26,37]. In the present study, for validation of caries experience in children using Cariogram model of their mothers during pregnancy, accuracy was higher than 70% for both cut-offs. Also, validation used in this study obtained clinically useful values according to Youden's index, when moderate and two lowest risk groups have combined.

Conclusions
With the limits of this study, Cariogram was valid and highly predictive in caries risk assessment in children based on caries risk assessment of their mother in pregnancy. The Cariogram model can be useful tool for caries prediction in both groups of woman and their children based on caries risk 11 assessment during pregnancy. Figure 1 Risk categories for children in regard to risk categories of their mother during pregnancy