The aims of this study were to estimate the risk of caries in the permanent teeth at 12 years of age and to describe the diagnostic accuracy of caries patterns in the primary dentition at age 4 years to predict caries at age 12 years. A prospective cohort study followed children from birth to age 12 years in the city of São Leopoldo, Brazil. Sociodemographic variables were collected at birth, and dental caries was measured at 4 and 12 years of age (n = 204). At 4 years, children were classified according to the presence of caries (cavitated and non-cavitated lesions), number of lesions, affected segment (anterior or posterior), and affected surface (occlusal, smooth, or proximal). Prediction of permanent dentition caries occurrence (DMFT ≥1) (primary outcome) involved Poisson regression with robust variance and standard diagnostic accuracy measures. The prevalences of caries at age 4 years (including non-cavitated lesions) and 12 years were 61.8% and 42.2%, respectively. All caries patterns in the primary dentition were associated with caries in the permanent dentition. In multivariable analysis, the strongest associations were carious lesions on the primary posterior teeth (RR 2.2; 95% CI 1.5–3.2) and occlusal surfaces (RR 2.1; 95% CI 1.4–3.0). Among patterns evaluated, the presence of any tooth with caries (cavitated or non-cavitated) had the highest sensitivity (73%), but any tooth with cavitated decay had the highest accuracy (67%). In conclusion, any dental caries experience in early childhood is strongly predictive of dental caries experience in early adolescence. Primary dentition carious lesions on the posterior teeth or occlusal surfaces and the presence of cavitated lesions were stronger predictors.

Dental caries can cause pain, functional limitation, and tooth loss, adversely impacting oral health-related quality of life [Feldens et al., 2016; Bernabe et al., 2020]. Caries can also contribute to absenteeism at school or work and place additional burdens on the healthcare system, impacting society as a whole [Righolt et al. 2018; Peres et al., 2019]. To date, caries continues to be the most prevalent chronic condition worldwide [Watt et al., 2019; Bernabe et al., 2020].

Dental caries is a biofilm-mediated, diet-modulated, multifactorial, noncommunicable, dynamic disease determined by biological, behavioral, psychosocial, and environmental factors [Machiulskiene et al., 2020]. An overarching strategy to address dental caries may feature a combination of population-based strategies focused on common risk factors and strategies that target high-risk groups [Watt 2005; Martignon et al., 2019; Watt et al., 2019]. Identifying high-risk individuals is an important step toward decision-making at the patient level, enabling dental care delivery that emphasizes prevention and funnels caries management resources toward those at highest risk, especially in countries where resources are scarce [Ismail et al., 2013; Twetman 2016; Martignon et al., 2019; Fee et al., 2020]. Many caries risk prediction models have been proposed, most involving protective factors as well as biological, behavioral, and socioeconomic risk factors. Although models are potentially more precise predictors than isolated variables [Ismail et al., 2013; Twetman 2016], model performance has been inconsistent [Young et al., 2017; Halasa-Rappel et al., 2019; Fontana et al., 2020; Twetman and Banerjee 2020]. Moreover, the need to collect data on multiple variables and the added complexity (and potentially added time and cost) could impede the use of more sophisticated caries risk models in public health programs [Twetman et al., 2013; Halasa-Rappel et al., 2019; Fontana et al., 2020].

Multiple investigations show previous caries experience to be the single best predictor of future caries [Mejàre et al., 2014; Twetman 2016; Hall-Scullin et al., 2017; Reyes et al., 2020]. However, many of these studies examine short prediction intervals (e.g., 1 year) and, by definition, rely on dental caries already having initiated to identify those who are most at risk [Twetman and Fontana 2009; Fontana et al., 2020]. Identifying children facing long-term elevated caries risk while still entirely in the primary dentition phase would help guide appropriate prevention prior to any damage to the permanent teeth. However, few studies have investigated whether and which early clinical patterns of dental caries can best predict later disease risk in the permanent dentition. Readily identified caries patterns in the primary dentition could potentially serve as simple long-term caries risk markers with potential for widespread application [Hall-Scullin et al., 2017].

Therefore, the objective of the present prospective cohort study was to estimate the risk of caries in the permanent teeth at age 12 years and to examine the diagnostic accuracy of selected patterns of caries in the primary dentition (age 4 years) to predict dental caries at age 12 years. Patterns of interest included any caries experience (including or excluding non-cavitated lesions), the number of lesions, as well as the segment (anterior and posterior) and surface (occlusal, proximal, and smooth) affected.

Study Design and Participants

The present study is a prospective cohort nested in a randomized clinical trial (ClinicalTrials.gov: NCT00629629) in which 500 children were recruited at birth from the maternity ward of the only public hospital in the city of São Leopoldo, Brazil. The aim of the intervention study was to investigate the effectiveness of nutritional counseling during the first year of life on different health outcomes, which has been published previously [Feldens et al., 2010a]. São Leopoldo has a population of approximately 213,000 residents, and all homes have access to the fluoridated public water supply (average 0.7 ppm F). Access to dental services in Brazil is universal and free, but utilization is low in the first years of life [Brasil Ministério da Saúde, 2012].

After childbirth, mothers with a singleton, full-term (≥37 weeks), normal weight (≥2,500 g) birth, free of congenital malformations and breastfeeding impediments (i.e., HIV/AIDS) were asked to participate in the study. After being informed of study procedures, 90% of the mothers agreed to participate. The cohort completed five study visits in the weeks and months after childbirth, with further follow-up at 12 months, as well as 4, 8, and 12 years of age. Dental examinations were completed only at 12-month, 4-year, and 12-year visits.

This cohort analysis includes dental data only from the 4-year and 12-year examinations (Fig. 1). Three hundred forty children were examined at age of 4 years. The number of participants recruited and retained at age 12 years (N = 204) determined the available sample for this study.

Fig. 1.

Cohort flowchart depicts the number of participants available for analysis at each time point. Only time points relevant to the current study are shown. Families were also contacted in the weeks and months after childbirth, at age 12 months and at age 8 years, but no dental examinations were conducted at those visits. N = 469 children originally entered the cohort at the first visit (near childbirth).

Fig. 1.

Cohort flowchart depicts the number of participants available for analysis at each time point. Only time points relevant to the current study are shown. Families were also contacted in the weeks and months after childbirth, at age 12 months and at age 8 years, but no dental examinations were conducted at those visits. N = 469 children originally entered the cohort at the first visit (near childbirth).

Close modal

Data Collection

The following data were collected soon after childbirth: child’s sex and anthropometric data, family structure (child lives with both parents or nonnuclear), family income (Brazilian reais), and mother’s educational attainment (years of school completed). At age of 4 years, an experienced epidemiologist (dental surgeon specializing in pediatric dentistry) completed a clinical examination with the child seated on a chair. The teeth were brushed and dried with gauze. Each dental surface was examined with the aid of a mouth mirror and WHO probe under artificial light. The surfaces were categorized as sound, non-cavitated lesion, cavitated lesion, restored, or extracted. Early childhood caries (ECC) was defined as the presence of one or more carious surfaces, including non-cavitated lesions (ICDAS >0), or any missing or filled surfaces on any tooth (d1mfs ≥1) [Drury et al., 1999; Evans et al., 2018]. The children were subsequently categorized based on the segment (anterior and posterior) and tooth surface (smooth, occlusal, or proximal) affected by carious lesions. The children were also categorized based on caries experience counting only cavitated (ICDAS >2) lesions (i.e., one or more surfaces filled, missing, or with enamel breakdown, d2mfs ≥1) [Evans et al., 2018]. Intra-examiner reproducibility was previously tested with two examinations of 20 children aged 3–5 years (surfaces scored as sound, non-cavitated, or cavitated) with a 14-day interval between sessions (weighted kappa = 0.90).

The 12-year examination was performed with the adolescent seated by another dental surgeon, trained and calibrated by the dental examiner from the 4-year examination, who acted as a benchmark. The teeth were brushed and dried with gauze. Each tooth surface was examined with the aid of a mouth mirror and WHO probe under artificial light. The surfaces were examined for dental caries [WHO 1997], corresponding to ICDAS >3, and the number of decayed, missing, and filled teeth were recorded (DMFT index). The primary outcome of this study was caries experience in the permanent dentition (DMFT ≥1). Examiner calibration involved two examinations of 20 adolescents (surfaces scored as cavitated or not) with a 7-day interval between sessions. Unweighted kappa coefficients for inter-examiner and intra-examiner (12-year-old examiner) were 0.81 and 0.83, respectively.

Data Analysis

Data were analyzed using the SPSS program, version 20 (IBM Corp.). The χ2 test was used to test associations between caries patterns in the primary dentition and caries experience in the permanent dentition. Spearman’s rho quantified the correlation between the number of lesions in the primary and permanent dentitions. The Kruskal-Wallis test was used to analyze whether the number of affected permanent teeth differed in accordance with the caries patterns in the primary dentition. Poisson regression with robust variance was performed to estimate the risk of caries in the permanent dentition (caries experience as dependent variable, dichotomous) according to each caries pattern in the primary dentition (in separate models). Regression models were adjusted for child’s sex, mother’s age and years of education, family income, and allocation group in the nesting trial (i.e., children were originally recruited to test an intervention; allocation to the intervention could be a confounder). To describe diagnostic accuracy for predicting caries in the permanent dentition based on caries patterns in the primary dentition, sensitivity, specificity, accuracy, positive predictive values, negative predictive values (NPVs), and the proportions of false positives and false negatives were calculated.

The final sample comprised 204 participants who initiated the study at birth and were examined at 4 and 12 years of age. In post hoc calculations, this sample size provided 83% statistical power to detect a significant difference (p < 0.05) in the occurrence of caries on permanent teeth among children without caries (29.5%, n = 23/78) and those with caries (50.0%, n = 63/126) at age of 4 years. The early life characteristics of the sample visited at 4 years and the sample retained to 12 years of age were similar (Table 1), with no statistically significant differences for any of the variables. Mothers’ age at childbirth ranged from 14 to 45 years, with a mean (standard deviation [SD]) of 26.4 (6.7) years. Mothers’ years of education ranged from one to 13 years (mean: 7.2, SD: 2.7).

Table 1.

Characteristics of participants at 4 and 12 years of age

Cohort at 4 yearsaSample analyzedb
%Mean (SD)%Mean (SD)
Maternal variables and family environment 
 Mother’s age at childbirth, years  25.8 (6.7)  26.4 (6.7) 
 Mother’s schooling, years  6.9 (2.7)  7.2 (2.7) 
 Family structure: nuclear 71.3  71.2  
 Family income <2 × MMW 27.8  27.5  
Child’s variables 
 Male sex 56.7  59.3  
 Birth weight, g  3,357.6 (466.8)  3,378.7 (459.6) 
 Length at birth, cm  48.7 (2.0)  48.8 (2.0) 
 Consumed sugar prior to 6 months of age 86.4  83.9  
 ECC at 4 years 62.9  61.8  
 d1mft at 4 years  3.8 (4.4)  3.5 (4.3) 
Cohort at 4 yearsaSample analyzedb
%Mean (SD)%Mean (SD)
Maternal variables and family environment 
 Mother’s age at childbirth, years  25.8 (6.7)  26.4 (6.7) 
 Mother’s schooling, years  6.9 (2.7)  7.2 (2.7) 
 Family structure: nuclear 71.3  71.2  
 Family income <2 × MMW 27.8  27.5  
Child’s variables 
 Male sex 56.7  59.3  
 Birth weight, g  3,357.6 (466.8)  3,378.7 (459.6) 
 Length at birth, cm  48.7 (2.0)  48.8 (2.0) 
 Consumed sugar prior to 6 months of age 86.4  83.9  
 ECC at 4 years 62.9  61.8  
 d1mft at 4 years  3.8 (4.4)  3.5 (4.3) 

MMW, monthly minimum wage.

an = 340.

bn = 204.

The prevalence of ECC (including non-cavitated lesions) at 4 years of age was 61.8% (126/204), whereas the prevalence considering only cavitated lesions was 46.1% (94/204). The number of teeth with caries lesions (cavitated or non-cavitated) ranged from 0 to 20 (mean: 3.5, SD: 4.3, median: 2.0, Q1–Q3: 0–6.0). The prevalence of caries in the permanent dentition was 42.2% (86/204), and the number of affected teeth ranged from 0 to 7 (mean: 0.8, SD: 1.28). Untreated caries (D component) accounted for the largest portion of caries experience (58.8%), followed by the F (38.1%) and M (3.1%) components. The number of primary and permanent teeth with caries experience was positively correlated. The correlation coefficient was slightly higher when non-cavitated lesions in the primary dentition were included (r = 0.304; p < 0.001) compared to including only cavitated lesions (r = 0.284; p < 0.001).

Table 2 shows that the occurrence, number, and location of caries lesions at age of 4 years were significantly associated with the occurrence of caries and the DMFT index at age of 12 years. A gradual, constant increase in the experience of caries in the permanent dentition was found with each increase in the number of caries-affected primary teeth.

Table 2.

Caries experience at 12 years in permanent dentition according to caries patterns in primary dentition

Primary dentition (4 years)Permanent dentition (12 years)
variableN(%)DMFT ≥1pDMFTp
N(%)Mean(SD)
Prevalence of caries 
ECCa 
 No 78 (38.2) 23 (29.5) 0.004 0.54 (0.96) 0.004 
 Yes 126 (61.8) 63 (50.0) 1.04 (1.40) 
d2mfs≥1b 
 No 110 (53.9) 23 (36.6) 0.008 0.57 (0.94) 0.003 
 Yes 94 (46.1) 49 (52.1) 1.17 (1.52) 
Number of lesions (ECC) 
d1mft index 
 0 78 (38.2) 23 (29.5) <0.001 0.54 (0.96) <0.001 
 1–2 31 (15.2) 11 (35.5) 0.58 (0.92) 
 3–4 34 (16.7) 15 (44.1) 0.62 (0.85) 
 5–8 32 (15.7) 15 (46.9) 0.88 (1.24) 
 ≥9 29 (14.2) 22 (75.9) 2.21 (1.86) 
Affected segment 
Anterior 
 No 124 (60.8) 45 (36.3) 0.035 0.68 (1.05) 0.033 
 Yes 80 (39.2) 41 (51.2) 1.11 (1.53) 
Posterior 
 No 95 (46.6) 26 (27.4) <0.001 0.47 (0.90) <0.001 
 Yes 109 (53.4) 60 (55.0) 1.17 (1.46) 
Affected surface 
Smooth 
 No 123 (60.3) 43 (35.0) 0.010 0.60 (0.98) 0.003 
 Yes 81 (39.7) 43 (53.1) 1.22 (1.56) 
Occlusal 
 No 99 (48.5) 28 (28.3) <0.001 0.48 (0.88) <0.001 
 Yes 105 (51.5) 58 (55.2) 1.20 (1.48) 
Proximal 
 No 133 (65.2) 49 (36.8) 0.035 0.68 (1.05) 0.023 
 Yes 71 (34.8) 37 (52.1) 1.17 (1.57) 
Primary dentition (4 years)Permanent dentition (12 years)
variableN(%)DMFT ≥1pDMFTp
N(%)Mean(SD)
Prevalence of caries 
ECCa 
 No 78 (38.2) 23 (29.5) 0.004 0.54 (0.96) 0.004 
 Yes 126 (61.8) 63 (50.0) 1.04 (1.40) 
d2mfs≥1b 
 No 110 (53.9) 23 (36.6) 0.008 0.57 (0.94) 0.003 
 Yes 94 (46.1) 49 (52.1) 1.17 (1.52) 
Number of lesions (ECC) 
d1mft index 
 0 78 (38.2) 23 (29.5) <0.001 0.54 (0.96) <0.001 
 1–2 31 (15.2) 11 (35.5) 0.58 (0.92) 
 3–4 34 (16.7) 15 (44.1) 0.62 (0.85) 
 5–8 32 (15.7) 15 (46.9) 0.88 (1.24) 
 ≥9 29 (14.2) 22 (75.9) 2.21 (1.86) 
Affected segment 
Anterior 
 No 124 (60.8) 45 (36.3) 0.035 0.68 (1.05) 0.033 
 Yes 80 (39.2) 41 (51.2) 1.11 (1.53) 
Posterior 
 No 95 (46.6) 26 (27.4) <0.001 0.47 (0.90) <0.001 
 Yes 109 (53.4) 60 (55.0) 1.17 (1.46) 
Affected surface 
Smooth 
 No 123 (60.3) 43 (35.0) 0.010 0.60 (0.98) 0.003 
 Yes 81 (39.7) 43 (53.1) 1.22 (1.56) 
Occlusal 
 No 99 (48.5) 28 (28.3) <0.001 0.48 (0.88) <0.001 
 Yes 105 (51.5) 58 (55.2) 1.20 (1.48) 
Proximal 
 No 133 (65.2) 49 (36.8) 0.035 0.68 (1.05) 0.023 
 Yes 71 (34.8) 37 (52.1) 1.17 (1.57) 

a ECC, early childhood caries (including non-cavitated lesions).

bCaries experience in childhood considering only cavitated lesions.

In unadjusted models, the risk of caries in the permanent dentition was significantly higher in children who had caries in the primary dentition (cavitated or non-cavitated lesions), those with a greater number of lesions, those with lesions on anterior or posterior teeth, and those with lesions on smooth, occlusal, or proximal surfaces (Table 3). After adjustment for confounding factors, the relative increase in risk was greater when non-cavitated lesions were included in the diagnosis of primary dentition caries (RR: 1.70; 95% CI: 1.14–2.53) and when caries were found on occlusal surfaces compared to other surfaces (RR: 2.08; 95% CI: 1.44–3.00). The association was stronger in children with lesions on posterior teeth in the primary dentition (RR: 2.21; 95% CI: 1.51–3.23), whereas the association with lesions on anterior teeth lost statistical significance after adjustment for confounding factors.

Table 3.

Associations between primary dentition caries patterns and permanent dentition caries

Primary dentitionUnadjusted modelAdjusted model a
RR(95% CI)pRR(95% CI)p
Prevalence of caries 
ECC 
 No 1.00  0.007 1.00  0.008 
 Yes 1.70 (1.15–2.49) 1.70 (1.14–2.53) 
d2mfs ≥1 b 
 No 1.00  0.008 1.00  0.012 
 Yes 1.55 (1.12–2.15) 1.53 (1.10–2.14) 
Number of lesions (ECC) 
d1mft index 
 0 1.00  <0.001 1.00  <0.001 
 1–2 1.20 (0.67–2.16) 1.10 (0.60–2.03) 
 3–4 1.50 (0.90–2.49) 1.52 (0.90–2.55) 
 5–8 1.59 (0.96–2.63) 1.76 (1.06–2.95) 
 ≥9 2.57 (1.72–3.84) 2.51 (1.66–3.81) 
Affected segment 
Anterior lesion 
 No 1.00  0.032 1.00  0.104 
 Yes 1.41 (1.03–1.94) 1.32 (0.95–1.84) 
Posterior lesion 
 No 1.00  <0.001 1.00  <0.001 
 Yes 2.01 (1.39–2.91) 2.21 (1.51–3.23) 
Affected surface 
Lesion – smooth surface 
 No 1.00  0.010 1.00  0.009 
 Yes 1.52 (1.11–2.08) 1.54 (1.11–2.14) 
Lesion – occlusal surface 
 No 1.00  <0.001 1.00  <0.001 
 Yes 1.95 (1.37–2.79) 2.08 (1.44–3.00) 
Lesion – proximal surface 
 No 1.00  0.031 1.00  0.039 
 Yes 1.41 (1.03–1.94) 1.42 (1.02–1.97) 
Primary dentitionUnadjusted modelAdjusted model a
RR(95% CI)pRR(95% CI)p
Prevalence of caries 
ECC 
 No 1.00  0.007 1.00  0.008 
 Yes 1.70 (1.15–2.49) 1.70 (1.14–2.53) 
d2mfs ≥1 b 
 No 1.00  0.008 1.00  0.012 
 Yes 1.55 (1.12–2.15) 1.53 (1.10–2.14) 
Number of lesions (ECC) 
d1mft index 
 0 1.00  <0.001 1.00  <0.001 
 1–2 1.20 (0.67–2.16) 1.10 (0.60–2.03) 
 3–4 1.50 (0.90–2.49) 1.52 (0.90–2.55) 
 5–8 1.59 (0.96–2.63) 1.76 (1.06–2.95) 
 ≥9 2.57 (1.72–3.84) 2.51 (1.66–3.81) 
Affected segment 
Anterior lesion 
 No 1.00  0.032 1.00  0.104 
 Yes 1.41 (1.03–1.94) 1.32 (0.95–1.84) 
Posterior lesion 
 No 1.00  <0.001 1.00  <0.001 
 Yes 2.01 (1.39–2.91) 2.21 (1.51–3.23) 
Affected surface 
Lesion – smooth surface 
 No 1.00  0.010 1.00  0.009 
 Yes 1.52 (1.11–2.08) 1.54 (1.11–2.14) 
Lesion – occlusal surface 
 No 1.00  <0.001 1.00  <0.001 
 Yes 1.95 (1.37–2.79) 2.08 (1.44–3.00) 
Lesion – proximal surface 
 No 1.00  0.031 1.00  0.039 
 Yes 1.41 (1.03–1.94) 1.42 (1.02–1.97) 

RR, relative risk; CI, confidence interval; ECC, early childhood caries.

aAdjusted for child’s sex, mother’s age and years of education, family income, and allocation group in the nesting trial.

bCaries experience in childhood considering only cavitated lesions.

Table 4 shows that the inclusion of non-cavitated lesions (cut point: d1mft ≥1) had greater sensitivity to predict permanent dentition caries experience (73.3%) compared to the criterion that only considered cavitated lesions (68.1%), corresponding to a smaller proportion of false negatives (26.7% vs. 31.9%, respectively). However, due to a lower percentage of false positives, overall accuracy was higher when considering only cavitated lesions (66.7%) versus including non-cavitated lesions (57.8%). Each increase in the criterion cut point of the number of teeth affected by ECC reduced sensitivity and increased specificity. The greatest accuracy was found when the cutoff point was d1mft ≥7.

Table 4.

Diagnostic performance of primary dentition caries patterns to predict occurrence of any permanent dentition caries

VariableSEN (%)SPEC (%)PPV (%)NPV (%)ACCU (%)FN (%)FP (%)
ECC 
 Yes (dmft ≥1) 73.3 46.6 50.0 70.5 57.8 26.7 53.4 
d2mfs ≥1 a 
 Yes 68.1 65.9 52.1 79.1 66.7 31.9 34.1 
ECC – cutoff points 
 dmft ≥2 67.4 50.8 50.0 68.2 57.8 32.6 49.2 
 dmft ≥3 60.5 63.6 54.7 68.8 62.3 39.5 36.4 
 dmft ≥ 4 52.3 70.3 56.2 66.9 62.7 47.7 29.7 
 dmft ≥5 43.0 79.7 60.7 65.7 64.2 57.0 20.3 
 dmft ≥6 39.5 83.9 64.2 65.6 65.2 60.5 16.1 
 dmft ≥7 33.7 89.0 69.0 64.8 65.7 66.3 11.0 
 dmft ≥8 26.7 93.2 74.2 63.6 65.2 73.3 6.8 
 dmft ≥9 25.5 94.1 75.9 63.4 65.2 74.5 5.9 
 dmft ≥10 19.8 94.9 73.9 61.9 63.2 80.2 5.1 
Location 
Segment 
 Anterior lesion 47.7 66.9 51.2 63.7 58.8 52.3 33.1 
 Posterior lesion 69.8 58.5 55.0 72.6 63.2 30.2 41.5 
Surface 
 Occlusal 67.4 60.2 55.2 71.7 63.2 32.6 39.8 
 Proximal 43.0 71.2 52.1 63.2 59.3 57.0 28.8 
 Smooth 50.0 67.8 53.1 65.0 60.3 50.0 32.2 
VariableSEN (%)SPEC (%)PPV (%)NPV (%)ACCU (%)FN (%)FP (%)
ECC 
 Yes (dmft ≥1) 73.3 46.6 50.0 70.5 57.8 26.7 53.4 
d2mfs ≥1 a 
 Yes 68.1 65.9 52.1 79.1 66.7 31.9 34.1 
ECC – cutoff points 
 dmft ≥2 67.4 50.8 50.0 68.2 57.8 32.6 49.2 
 dmft ≥3 60.5 63.6 54.7 68.8 62.3 39.5 36.4 
 dmft ≥ 4 52.3 70.3 56.2 66.9 62.7 47.7 29.7 
 dmft ≥5 43.0 79.7 60.7 65.7 64.2 57.0 20.3 
 dmft ≥6 39.5 83.9 64.2 65.6 65.2 60.5 16.1 
 dmft ≥7 33.7 89.0 69.0 64.8 65.7 66.3 11.0 
 dmft ≥8 26.7 93.2 74.2 63.6 65.2 73.3 6.8 
 dmft ≥9 25.5 94.1 75.9 63.4 65.2 74.5 5.9 
 dmft ≥10 19.8 94.9 73.9 61.9 63.2 80.2 5.1 
Location 
Segment 
 Anterior lesion 47.7 66.9 51.2 63.7 58.8 52.3 33.1 
 Posterior lesion 69.8 58.5 55.0 72.6 63.2 30.2 41.5 
Surface 
 Occlusal 67.4 60.2 55.2 71.7 63.2 32.6 39.8 
 Proximal 43.0 71.2 52.1 63.2 59.3 57.0 28.8 
 Smooth 50.0 67.8 53.1 65.0 60.3 50.0 32.2 

SEN, sensitivity; SPEC, specificity; PPV, positive predictive value; NPV, negative predictive value; ACCU, accuracy; FN, false negative; FP, false positive.

a Caries experience in childhood considering only cavitated lesions.

Regarding location, the highest sensitivity (69.8%) and NPV (72.6%) values were associated with the occurrence of caries in the posterior segment and on occlusal surfaces (sensitivity: 67.4%; NPV: 71.7%). The occurrence of proximal lesions and lesions on the anterior teeth had lower sensitivity (43.0% and 47.7%, respectively) and slightly lower accuracy (59.3% and 60.3%, respectively) than posterior or occlusal lesions, each of which had an accuracy of 63.2%.

The present birth cohort study estimated the strength of association between caries patterns in the primary dentition and caries occurrence in the permanent dentition and investigated which caries patterns in the primary dentition best predicted this outcome. Based on the study findings, primary dentition caries is a strong indicator of caries occurrence in early adolescence. All primary dentition caries patterns examined produced reasonable diagnostic accuracy, ranging from 58% to 67%. Carious lesions on the posterior teeth or occlusal surfaces and the presence of cavitated lesions were relatively more accurate predictors, but inclusion of non-cavitated lesions resulted in the greatest sensitivity.

The association between the occurrence of caries in both dentitions has previously been described, but many of the cohort studies to report these associations included children in the mixed dentition phase [Motohashi et al., 2006; Skeie et al., 2006; Tagliaferro et al. 2006; Zhang and van Palenstein Helderman 2006; Hall-Scullin et al., 2017]. With a baseline mixed dentition sample, the permanent dentition may already be affected by caries lesions, and previous lesions in the primary teeth may be lost to exfoliation. In contrast, a baseline primary dentition sample followed to the permanent dentition phase offers a clearer temporal ordering with less information loss. Previous longitudinal studies that followed children from the exclusive primary dentition to the exclusive permanent dentition reported relative associations similar in magnitude to the present study, with the incidence of permanent dentition caries from 1.5 to more than 2 times greater among children with primary dentition caries than those caries free [Hause 1997; Peres et al., 2009; Lee et al., 2015; Du et al., 2017].

Few studies have described different diagnostic accuracy measures for caries in the permanent dentition based on disease experience in the exclusive primary dentition [Li and Wang 2002; Ekbäck et al., 2012; Saethre-Sundli et al., 2020]. In general, these studies found that the metrics used for classifying primary dentition caries presented trade-offs between sensitivity and specificity [Zhang and van Palenstein Helderman 2006; Peres et al., 2009; Ekbäck et al., 2012]. Given this trade-off, the most useful diagnostic accuracy measure in any context depends in part on the potential consequences of false negatives and false positives [Bonita et al., 2006]. When the consequences of not detecting a disease outweigh the harms of unnecessary treatment, more sensitive tests are valued [Fletcher et al., 2012]. In the context of dental caries, the goal of identifying high-risk children is to direct the provision of prevention strategies, such as nutritional counseling and the application of dental sealants and topical fluoride [Martignon et al., 2019]. Such interventions are noninvasive, inexpensive, and have few adverse effects, particularly compared to the treatment costs and sequelae of dental caries. Thus, there is an argument for prioritizing sensitivity and NPV in caries risk prediction.

In the present study, the inclusion of non-cavitated lesions in the diagnosis of ECC, as proposed by the World Health Organization and the International Association of Paediatric Dentistry [Phantumvanit et al., 2018; Tinanoff et al., 2019], yielded greater sensitivity in predicting permanent dentition caries. The criterion of any cavitated or non-cavitated lesion captured nearly three quarters of the children in this study to later experience permanent dentition caries. A cohort study involving preschool children in Brazil found that initial, moderate, or extensive lesions were all associated with an increase in caries after 2 years (still in the primary dentition), although children with only initial lesions had less severe outcomes [Guedes et al., 2018]. In the present study, the higher sensitivity achieved by including non-cavitated lesions also increased the proportion of false positives. In choosing risk indicators, over-prioritization of sensitivity has the obvious downside of losing the personally tailored, and potentially cost-saving, approach of reserving the most intensive caries prevention for the highest risk children. Whether caries prevention programs should be risk stratified or designed to provide preventive care universally is an active area of clinical and policy research. The effectiveness of a risk-based strategy depends at least partly on the accuracy of risk prediction.

The present results suggest that lesion location may affect caries prediction accuracy. Greater sensitivity and NPV were observed for lesions on the posterior teeth than the anterior teeth and on occlusal surfaces compared to smooth or proximal surfaces. The lower prevalence of proximal lesions may partially explain the lower sensitivity. Notably, the characteristic lesions of ECC on anterior teeth had low sensitivity, similar to results reported elsewhere [Li and Wang 2002]. These findings may be explained, at least in part, by the fact that eating practices associated with anterior tooth lesions, such as bottle feeding with sweetened liquids [Feldens et al., 2010b], are uncommon after eruption of the permanent teeth. However, the strong association between caries occurrence in both dentitions demonstrates disease persistence from childhood to adolescence in most patients, suggesting the continuing influence of caries risk factors, such as socioeconomic disadvantage and sugar consumption, previously confirmed as caries risk factors in this sample [Feldens et al., 2010b].

Some implications for clinical practice should be addressed. In Brazil, only one-third of dentists report using a caries risk assessment tool and only 13% record their assessment in the clinical chart [Tagliaferro et al. 2020]. Using the clinical presentation of caries in the primary dentition to indicate caries risk in the permanent dentition represents a simple and practical clinical parameter that requires no additional technology or expense; the dentist needs only identify non-cavitated lesions, which is standard care in pediatric dental practice. For caries risk assessment outside clinical dental practice, such as in school-based programs or medical examinations, visually obvious cavitation in the posterior teeth or occlusal surfaces may be the more practical risk indicator. Primary dentition caries alone is far from perfect predictor of future caries occurrence. However, the patterns of presentation identified in the present study provide insight into clinical conditions to be tested in more complete and multifactorial caries risk models.

Identifying children at greater risk of permanent dentition caries can contribute to the planning of specific interventions, such as professional fluoride application and sealant use or shorter recall appointment intervals [Martignon et al., 2019]. Moreover, actions should be based on the multifactorial nature of the disease and target engagement of patients to improve health behaviors, especially regarding sugar intake. The effectiveness of this measure depends on the intensity and frequency of counseling [Feldens et al., 2010a; Martignon et al., 2019]. As sugar intake is a common risk factor for other noncommunicable diseases, such actions should be integrated with upstream actions directed at the entire population, especially actions that do not depend on a change in behavior and can reduce inequalities. This includes reducing sugar availability at schools, advertising regulation, warning labels on sweetened foods and beverages, and excise taxes on sugary products [Watt et al., 2019].

The present study has limitations, among them the number of dropouts from the cohort over time. However, the dropout rate is similar to that found in other birth cohorts [Peres et al., 2020], and no significant differences were found between the children analyzed and those of the baseline cohort, minimizing the possibility of selection bias. Second, there may have been measurement error in the diagnosis of caries, especially in the primary dentition, in which non-cavitated lesions were included. Single examiners were used at both examination visits but different examiners between visits. Examiner reproducibility was adequate but should be interpreted cautiously. Third, although the use of different caries patterns in the primary dentition to predict caries in the permanent dentition has advantages from a clinical point of view due to its simplicity, it is limited as a multifactorial model of caries occurrence, which was not the objective of the present study. Lastly, the present study involved a population predominantly of low socioeconomic status, and the pattern and frequency of caries in the primary and permanent dentition was similar to that found in most low- and middle-income countries [Phantumvanit et al., 2018; Bernabe et al., 2020]. As some diagnostic accuracy measures, such as NPV and positive predictive value, are affected by the prevalence of the disease, the findings of this study cannot necessarily be extrapolated to other geographic and population contexts, such as low caries prevalence.

In conclusion, dental caries experience in the primary dentition was a strong predictor of dental caries in the permanent teeth 8 years later. Given that caries experience can be assessed visually, it serves as a simple and convenient marker of long-term caries risk status that can be implemented in a variety of settings to identify children in need of more intensive caries prevention. In this study, the presence of any caries experience, including non-cavitated lesions, had the most favorable sensitivity and NPV of all primary dentition specifications considered and thus served as a viable risk indicator when assessing early lesions when maximizing prevention is the goal. By comparison, as predictors, carious lesions on the posterior teeth or occlusal surfaces and the presence of cavitated lesions offered more balance between sensitivity and specificity. Overall, these findings suggest that simple, objective characteristics assessed during the exclusive primary dentition phase can help identify children who require individualized preventive care beginning well before the eruption of the first permanent teeth.

The Federal University of Health Sciences of Porto Alegre Nutrition Research Group (NUPEN) contributed to participant recruitment, data collection, and data management.

Research Ethics Committees of the Universidade do Vale do Rio dos Sinos approved the study protocol (No. 13-116). After research staff explained the study and all phases of data collection, parents provided written free and informed consent for themselves and on behalf of their children.

The authors have no conflicts of interest to declare.

Grant support was from the Rio Grande do Sul Research Support Foundation, the National Council for Scientific and Technological Development, and the Coordination for the Improvement of Higher Education Personnel (CAPES). The information presented is solely the responsibility of the authors and does not necessarily represent the official views of the sponsoring organizations.

Carlos Alberto Feldens, Vanessa Simas Braga, and Márcia Regina Vítolo conceived study design; Carlos Alberto Feldens and Priscila Humbert Rodrigues contributed to data collection; Carlos Alberto Feldens and Benjamin W. Chaffee analyzed data and drafted the initial manuscript; and Carlos Alberto Feldens, Vanessa Simas Braga, Paulo Floriani Kramer, Márcia Regina Vítolo, Priscila Humbert Rodrigues, Elisa Maria Rosa de Barros Coelho, and Benjamin W. Chaffee contributed to data interpretation and provided critical manuscript comments and revisions.

The data that support the findings of this study are not publicly available to maintain the confidentiality of research participants. Data may be made available on reasonable request. Further inquiries can be directed to the corresponding author.

1.
GBD 2017 Oral Disorders Collaborators
Bernabe
E
,
Marcenes
W
,
Hernandez
CR
,
Bailey
J
,
Abreu
LG
.
Global, regional, and national levels and trends in burden of oral conditions from 1990 to 2017: a systematic analysis for the global burden of disease 2017 study
.
J Dent Res
.
2020
;
99
(
4
):
362
73
.
2.
Bonita
R
,
Beaglehole
R
,
Kjellström
T
Basic Epidemiology. 2
Geneva
World Health Organization
2006
. p.
212
.
3.
Brasil Ministério da Saúde
. SB Brasil 2010: Pesquisa Nacional de Saúde Bucal: Resultados Principais.
Brasília
Ministério da Saúde
2012
. Available from: https://bvsms.saude.gov.br/bvs/publicacoes/pesquisa_nacional_saude_bucal.pdf.
4.
Drury
TF
,
Horowitz
AM
,
Ismail
AI
,
Maertens
MP
,
Rozier
RG
,
Selwitz
RH
.
Diagnosing and reporting early childhood caries for research purposes. A report of a workshop sponsored by the national institute of dental and craniofacial research, the health resources and services administration, and the health care financing administration
.
J Public Health Dent
.
1999
;
59
(
3
):
192
7
.
5.
Du
Q
,
Yu
M
,
Li
Y
,
Du
H
,
Gao
W
,
Mei
H
.
Permanent caries experience is associated with primary caries experience: a 7-year longitudinal study in China
.
Community Dent Oral Epidemiol
.
2017
;
45
(
1
):
43
8
.
6.
Ekbäck
G
,
Ordell
S
,
Unell
L
.
Can caries in the primary dentition be used to predict caries in the permanent dentition? An analysis of longitudinal individual data from 3–19 years of age in Sweden
.
Eur Arch Paediatr Dent
.
2012
;
13
(
6
):
308
11
.
7.
Evans
RW
,
Feldens
CA
,
Phantunvanit
P
.
A protocol for early childhood caries diagnosis and risk assessment
.
Community Dent Oral Epidemiol
.
2018
;
46
(
5
):
518
25
.
8.
Fee
PA
,
Riley
P
,
Worthington
HV
,
Clarkson
JE
,
Boyers
D
,
Beirne
PV
.
Recall intervals for oral health in primary care patients
.
Cochrane Database Syst Rev
.
2020
10
CD004346
.
9.
Feldens
CA
,
Ardenghi
TM
,
Dos Santos Dullius
AI
,
Vargas-Ferreira
F
,
Hernandez
PAG
,
Kramer
PF
.
Clarifying the impact of untreated and treated dental caries on oral health-related quality of life among adolescents
.
Caries Res
.
2016
;
50
(
4
):
414
21
.
10.
Feldens
CA
,
Giugliani
ERJ
,
Duncan
BB
,
Drachler
Md L
,
Vítolo
MR
.
Long-term effectiveness of a nutritional program in reducing early childhood caries: a randomized trial
.
Community Dent Oral Epidemiol
.
2010a
38
4
324
32
.
11.
Feldens
CA
,
Giugliani
ERJ
,
Vigo
Á
,
Vítolo
MR
.
Early feeding practices and severe early childhood caries in four-year-old children from southern Brazil: a birth cohort study
.
Caries Res
.
2010b
44
5
445
52
.
12.
Fletcher
RH
,
Fletcher
SW
,
Fletcher
GS
. Clinical Epidemiology: The Essentials. 5th ed.
Philadelphia (PA)
Lippincott Williams & Wilkins
2012
. p.
275
.
13.
Fontana
M
,
Carrasco-Labra
A
,
Spallek
H
,
Eckert
G
,
Katz
B
.
Improving caries risk prediction modeling: a call for action
.
J Dent Res
.
2020
;
99
(
11
):
1215
20
.
14.
Guedes
RS
,
Piovesan
C
,
Ardenghi
TM
,
Emmanuelli
B
,
Braga
MM
,
Mendes
FM
.
Presence of initial caries lesions as a risk factor for caries in preschool children: a cohort study
.
Caries Res
.
2018
52
1-2
32
41
.
15.
Halasa-Rappel
YA
,
Ng
MW
,
Gaumer
G
,
Banks
DA
.
How useful are current caries risk assessment tools in informing the oral health care decision-making process
.
J Am Dent Assoc
.
2019
;
150
(
2
):
91
102.e2
.
16.
Hall-Scullin
E
,
Whitehead
H
,
Milsom
K
,
Tickle
M
,
Su
T-L
,
Walsh
T
.
Longitudinal study of caries development from childhood to adolescence
.
J Dent Res
.
2017
;
96
(
7
):
762
7
.
17.
Hausen
H
.
Caries prediction–state of the art
.
Community Dent Oral Epidemiol
.
1997
;
25
(
1
):
87
96
.
18.
Ismail
AI
,
Tellez
M
,
Pitts
NB
,
Ekstrand
KR
,
Ricketts
D
,
Longbottom
C
.
Caries management pathways preserve dental tissues and promote oral health
.
Community Dent Oral Epidemiol
.
2013
;
41
(
1
):
e12
40
.
19.
Lee
H-J
,
Kim
J-B
,
Jin
B-H
,
Paik
D-I
,
Bae
K-H
.
Risk factors for dental caries in childhood: a five-year survival analysis
.
Community Dent Oral Epidemiol
.
2015
;
43
(
2
):
163
71
.
20.
Li
Y
,
Wang
W
.
Predicting caries in permanent teeth from caries in primary teeth: an eight-year cohort study
.
J Dent Res
.
2002
;
81
(
8
):
561
6
.
21.
Martignon
S
,
Pitts
NB
,
Goffin
G
,
Mazevet
M
,
Douglas
GVA
,
Newton
JT
.
CariesCare practice guide: consensus on evidence into practice
.
Br Dent J
.
2019
;
227
(
5
):
353
62
.
22.
Machiulskiene
V
,
Campus
G
,
Carvalho
JC
,
Dige
I
,
Ekstrand
KR
,
Jablonski-Momeni
A
.
Terminology of dental caries and dental caries management: consensus report of a workshop organized by ORCA and cariology research group of IADR
.
Caries Res
.
2020
;
54
(
1
):
7
14
.
23.
Mejàre
I
,
Axelsson
S
,
Dahlén
G
,
Espelid
I
,
Norlund
A
,
Tranæus
S
.
Caries risk assessment. A systematic review
.
Acta Odontol Scand
.
2014
;
72
(
2
):
81
91
.
24.
Motohashi
M
,
Yamada
H
,
Genkai
F
,
Kato
H
,
Imai
T
,
Sato
S
.
Employing dmft score as a risk predictor for caries development in the permanent teeth in Japanese primary school girls
.
J Oral Sci
.
2006
;
48
(
4
):
233
7
.
25.
Peres
KG
,
Thomson
WM
,
Chaffee
BW
,
Peres
MA
,
Birungi
N
,
Do
LG
.
Oral health birth cohort studies: achievements, challenges, and potential
.
J Dent Res
.
2020
;
99
(
12
):
1321
31
.
26.
Peres
MA
,
Barros
AJ
,
Peres
KG
,
Araújo
CLP
,
Menezes
AMB
.
Life course dental caries determinants and predictors in children aged 12 years: a population-based birth cohort
.
Community Dent Oral Epidemiol
.
2009
;
37
(
2
):
123
33
.
27.
Peres
MA
,
Macpherson
LMD
,
Weyant
RJ
,
Daly
B
,
Venturelli
R
,
Mathur
MR
.
Oral diseases: a global public health challenge
.
Lancet
.
2019
;
394
(
10194
):
249
60
.
28.
Phantumvanit
P
,
Makino
Y
,
Ogawa
H
,
Rugg‐Gunn
A
,
Moynihan
P
,
Petersen
PE
.
WHO global consultation on public health intervention against early childhood caries
.
Community Dent Oral Epidemiol
.
2018
;
46
(
3
):
280
7
.
29.
Righolt
A
,
Jevdjevic
M
,
Marcenes
W
,
Listl
S
.
Global-Regional-and Country-Level Economic Impacts of Dental Diseases in 2015
.
Int J Paediatr DentJ Dent Res
.
2018
;
97
(
5
):
501
7
.
30.
Saethre-Sundli
HB
,
Wang
NJ
,
Wigen
TI
.
Do enamel and dentine caries at 5 years of age predict caries development in newly erupted teeth? A prospective longitudinal study
.
Acta Odontol Scand
.
2020
;
78
(
7
):
509
14
.
31.
Skeie
MS
,
Raadal
M
,
Strand
GV
,
Espelid
I
.
The relationship between caries in the primary dentition at 5 years of age and permanent dentition at 10 years of age - a longitudinal study
.
Int J Paediatr Dent
.
2006
;
16
(
3
):
152
60
.
32.
Tagliaferro
EPd S
,
Pereira
AC
,
Meneghim
Md C
,
Ambrosano
GMB
.
Assessment of dental caries predictors in a seven-year longitudinal study
.
J Public Health Dent
.
2006
;
66
(
3
):
169
73
.
33.
Tagliaferro
EPd S
,
Silva
SRCda
,
Rosell
FL
,
Valsecki Junior
A
,
Riley Iii
JL
,
Gilbert
GH
.
Caries risk assessment in dental practices by dentists from a Brazilian community
.
Braz Oral Res
.
2020
;
35
:
e017
.
34.
Tinanoff
N
,
Baez
RJ
,
Diaz Guillory
C
,
Donly
KJ
,
Feldens
CA
,
McGrath
C
.
Early childhood caries epidemiology, aetiology, risk assessment, societal burden, management, education, and policy: global perspective
.
Int J Paediatr Dent
.
2019
;
29
(
3
):
238
48
.
35.
Twetman
S
.
Caries risk assessment in children: how accurate are we
.
Eur Arch Paediatr Dent
.
2016
;
17
(
1
):
27
32
.
36.
Twetman
S
,
Banerjee
A
.
Caries risk assessment
. In:
Chapple
ILC
,
Papapanou
PN
,
Organizadores
, editors. Risk Assessment in Oral Health: A Concise Guide for Clinical Application.
Cham
Springer International Publishing
2020
. p.
89
100
.
37.
Twetman
S
,
Fontana
M
.
Patient caries risk assessment
. In:
Pitts
N
, editor. Detection, Assessment, Diagnosis and Monitoring of Caries.
Basel
KARGER
2009
. p.
91
101
.
38.
Twetman
S
,
Fontana
M
,
Featherstone
JDB
.
Risk assessment – can we achieve consensus
.
Community Dent Oral Epidemiol
.
2013
;
41
(
1
):
e64
70
.
39.
Watt
RG
.
Strategies and approaches in oral disease prevention and health promotion
.
Bull World Health Organ
.
2005
;
83
(
9
):
711
8
.
40.
Watt
RG
,
Daly
B
,
Allison
P
,
Macpherson
LMD
,
Venturelli
R
,
Listl
S
.
Ending the neglect of global oral health: time for radical action
.
Lancet
.
2019
;
394
(
10194
):
261
72
.
41.
World Health Organization
. Oral Health Surveys – Basic Methods. 5th ed.
Geneva
World Health Organization
2013
.
42.
Young
DA
,
Alvear Fa
B
,
Rogers
N
,
Rechmann
P
.
The effect of calibration on caries risk assessment performance by students and clinical faculty
.
J Dent Educ
.
2017
;
81
(
6
):
667
74
.
43.
Zhang
Q
,
van Palenstein Helderman
WH
.
Caries experience variables as indicators in caries risk assessment in 6–7-year-old Chinese children
.
J Dent
.
2006
;
34
(
9
):
676
81
.