A Prospective Longitudinal Study of Early Childhood Caries Onset in Initially Caries-Free Children

Abstract

Introduction:

Early childhood caries (ECC) is a complex oral disease that is prevalent in US children.

Objectives:

The purpose of this 2-y prospective cohort study was to examine baseline and time-dependent risk factors for ECC onset in initially caries-free preschool children.

Methods:

A cohort of 189 initially caries-free children aged 1 to 3 y was recruited. At each 6-mo study visit, children were examined using the ICDAS index; salivary samples were collected to assess mutans streptococci (MS), lactobacilli, Candida species, salivary cortisol (prior and after a stressor), and salivary IgA. Diet and oral health behavior were assessed from parent report. Child and family stress exposure was assessed from measures of psychological symptoms, stressful life event exposure, family organization and violence exposure, and social support. Sociodemographic factors were also considered. A Kaplan-Meier estimator of survival function of time to ECC and a Cox proportional hazards model were used to identify predictors of ECC onset.

Results:

Onset of ECC was associated with high salivary MS levels at baseline (log-rank test, P < 0.0001). Cox proportional hazards regression showed that the risk of dental caries significantly increased with salivary MS in log scale over the 6-mo period (hazard ratio, 1.08; P = 0.01). Other risk factors in the model did not reach statistical significance.

Conclusion:

Our results provide prospective evidence that an increase in salivary MS predicts ECC onset in young, initially caries-free children, confirming that a high salivary MS count likely plays a causal role in ECC onset, independent of covariates.

Knowledge Transfer Statement:

These results suggest that we must focus on reducing salivary MS counts in young children and preventing or delaying MS colonization in infants and young children determined to be at risk for ECC.

Introduction

Early childhood caries (ECC) is a complex multifactorial oral disease that is prevalent in US children <71 mo of age (Dye et al. 2007; Kopycka-Kedzierawski et al. 2008; Dye et al. 2017). ECC is challenging to treat successfully and has a disturbing propensity to recur following treatment. Several risk factors for ECC have been identified. Microbiological mechanisms have been identified, including intraoral bacteria (mainly mutans streptococci [MS] and lactobacilli [LB]) and a cariogenic diet (Klock and Krasse 1978; Plonka et al. 2012; Kirthiga et al. 2019; Thang Le et al. 2021); additionally, Candida species may exacerbate ECC onset and progression as suggested in recent studies (Raja et al. 2010; Xiao et al. 2016; Xiao et al. 2018; de Jesus et al. 2020). In addition, ECC risk is associated with a socioeconomic gradient (Dye et al. 2007; Dye et al. 2017), and many factors thought to index stress exposure have been suggested as risk factors for ECC (Seow et al. 2009; Fontana 2015; Kirthiga et al. 2019). Additionally, in preschool children, elevated salivary cortisol was significantly associated with dental caries: higher salivary cortisol reactivity was associated with thinner, softer enamel surfaces in exfoliated primary teeth (Boyce et al. 2010). Through its association with MS, LB, and Candida species in saliva, salivary cortisol may index stress-induced physiologic changes that directly or indirectly lead to ECC onset.

Despite considerable research in the area, basic questions persist about the risk mechanisms leading to ECC because of limits to study designs (e.g., reliance on cross-sectional analyses) and failure to consider competing risks within the same analysis. In addition, risk factors associated with dental caries in children and adults may or may not be directly comparable, as developmental factors concerning oral microflora and host defense mechanisms and dietary changes may pose particular or different sources of risk for ECC. For example, Morou-Bermudez and colleagues’ (2011) Caries Risk Pyramid has been proposed as a hierarchical conceptual model of an individual’s risk and needs to be validated. Similarly, a recent review of risk factors for ECC underscored the heterogeneity and low quality of research and stressed the need for longitudinal studies (Thang Le et al. 2021).

The purpose of our study was to provide a strong test of multiple alternative exposures that may predict ECC onset. We did this by prospectively following initially caries-free children for 2 y and assessing caries status, oral microbiology, oral health behavior, and multiple sources of stress at 6-mo intervals.

Methods

The Division of Pediatric Dentistry, Eastman Institute for Oral Health, recruited a cohort of 189 preschool children and their parents/primary caregivers. Children were aged 1 to 3 y, had confirmed caries-free status, and were eligible for Medicaid and Child Health Plus (i.e., at high risk for ECC). Children and their parents/primary caregivers aged ≥18 y were eligible to participate in the study irrespective of sex, ethnic origin, or race. Children with major medical problems and diagnosis of dental caries were not eligible for the study. Children taking antibiotics were not excluded, but any study visit was postponed for at least 30 d from the end of the antibiotic exposure. The study was conducted between February 2016 and February 2021. Participants were recruited from approximately 800 parents/primary caregivers who were approached to participate in the study over the course of a 3-y recruitment and enrollment period. A majority of individuals who opted not to participate in the study were not interested or had limited free time, work or family commitments, or language barriers. The study was approved by the University of Rochester Research Subject Review Board (No. 57726). Written informed consent was obtained from parents/primary caregivers of participating children prior to the study; participants were reimbursed for study participation.

Caries status was assessed with the International Caries Detection and Assessment System (ICDAS) (Ismail et al. 2007). At each visit, children were examined by a calibrated pediatric dentist; all children had an ICDAS score of 0 at enrollment.

Oral microbiology was assessed from a sample of approximately 2 mL of saliva, which was collected from children at baseline assessment and 6, 12, 18, and 24 mo afterward. Microbiological profiles included MS, LB, and Candida species; details related to the microbiological procedures are described elsewhere (Kopycka-Kedzierawski et al. 2021).

Child stress physiology, based on salivary cortisol, was measured at each visit from 3 salivary collections separate from those described for microbiology. SalivaBio’s Children’s Swab System (Salimetrics) was used to collect saliva samples prior to and 15 and 30 min following a previously described and validated laboratory stressor (Spangler and Grossmann 1993; O’Connor et al. 2013). Clock time, duration of each collection, and volume of collected saliva (assuming specific gravity = 1) were recorded. Following collection, saliva samples were immediately placed on ice and transported to the laboratory, where they were stored at −20 °C until processing.

Salivary cortisol and secretory IgA (sIgA) were measured in triplicate per the manufacturer’s instructions with the Expanded Range High Sensitivity Salivary Cortisol Enzyme Kit (1-3002) and the Salivary Secretory IgA Indirect Enzyme Immunoassay Kit (1-1602), respectively. Secretory IgA was measured in only 1 of the 3 saliva samples at each visit (typically the 30-min collection). All standard curves were fit to a 5-parameter logistic (GainData ELISA Calculator; Arigo Biolaboratories). Cortisol is reported as the concentration in saliva (mcg/dL), while sIgA is reported as the concentration in saliva (mcg/mL) and, after adjustment for flow rate, the sIgA secretion rate (mcg/min). The area under the curve for salivary cortisol was determined at 3 time points of collection: T0, from wake-up time to the first collection of cortisol, and 15 and 30 min following the stressor from T0. Log-transformed area under the curve was included in the Cox regression analysis.

We took considerable care in collecting saliva samples using established procedures to avoid confounds and contamination; for example, all saliva was obtained at the beginning of the visit to not be confounded by the subsequent dental examination, and no food or beverages were consumed within 30 min of saliva collection. We considered baseline and reactivity measures of cortisol, as well as the area under the curve, in analyses. Time of awakening and time of first sample collection were considered as covariates because of the strong diurnal influence on cortisol levels.

Child and family stress exposure was based on parent-reported measures of stress from multiple sources; each is widely used in studies of psychosocial stress. Specific measures included parental depression per the Center for Epidemiologic Studies Depression Scale (Radloff 1977). Anxiety and worry were based on the Penn State Worry Questionnaire (Meyer et al. 1990). Alcohol use was determined from the Alcohol Use Disorders Identification Test (Babor et al. 2001). Stressful life events were selected from a list of standard high-stress conditions (e.g., loss of income, health problems; Compas et al. 1989). Household disorganization and confusion were derived from the CHAOS tool (i.e., Confusion, Hubbub, and Order Scale; Matheny et al. 1995). Violence exposure was based on the Psychological Aggression and Physical Assault subscales of the Conflict Tactics Scale (Straus et al. 1996). Finally, caregiver social support was based on the Interpersonal Support and Evaluation List (Cohen and Hoberman 1983).

Diet, Oral Health Behaviors, and Sociodemographic Variables

Favorite snacks/drinks and oral health behaviors were identified from a parental questionnaire, which included questions related to the child’s eating and drinking habits, snacking choices, sippy cup use, and the type and amount of beverages and snacks consumed by the children; we also assessed breastfeeding and formula-feeding practices and the children’s oral hygiene regimens (Kressin et al. 2009). Sociodemographic variables included the children’s and parents/primary caregivers’ age, race, ethnicity, and gender; parent/primary caregiver educational attainment, occupation, and income were additionally assessed.

Data Analysis

Sample Size Calculation

On the basis of our prior study, we assumed that the median caries-free time of survival for the children with high salivary MS levels at baseline was approximately 44.5 mo and that the hazard ratio of the caries-free survival among children with high versus low salivary MS status was 2.87 (Kopycka-Kedzierawski and Billings 2004). The method developed by Hsieh and Lavori (2000) based on the Cox regression model indicated that a sample size of 135 had 80% power to detect this difference (assuming a significance level of 5% and that the R² of MS on other covariates is 0.15).

Oral microbiological factors were dichotomized in the following manner (CFU/mL; Epstein et al. 1980; Zoitopoulos et al. 1996; Seow 1998): MS <10⁵ vs. ≥10⁵, LB = 0 vs. >0, and Candida = 0 vs. >0. Children were categorized as having low or high levels of MS in saliva. A high level of salivary MS was defined as ≥10⁵ CFU/mL, as at baseline, 18% of the children fell into the higher MS category. At baseline, approximately 84% of children had undetectable LB in saliva and 67% had undetectable Candida species. Given these observations, the cutoff point of 10⁵ CFU/mL of saliva between the high and low MS levels was chosen, as well as carriage or not of LB and Candida species.

Statistical Analyses

A Kaplan-Meier estimator of overall survival function of time to ECC onset was generated for the cohort of 164 initially caries-free children over the 24-mo study period (25 children were excluded from the analyses because their microbiological profiles were not available).

A Kaplan-Meier estimator of overall survival function of time to ECC onset stratified by baseline salivary MS in the cohort of children was generated taking into account high and low MS levels (cutoff, 10⁵) at baseline. The log-rank test was used to examine the difference between the groups of children in caries-free survival time. As the study follow-up period for each child was 2 y with follow-up visits at 6, 12, 18, and 24 mo, the time to ECC onset was interval censored. The sparse interval censoring made the estimation of the distribution of the time from baseline examination to ECC onset unstable. For this reason we made an assumption that if ECC was detected between study visits k and k + 1, the time to ECC onset was assumed to be uniformly distributed, and we took the middle point of the interval as the time of caries onset. If the child was caries-free at the last study visit, then the time was assumed to be right censored.

A Cox proportional hazards model was used to study the effects of several risk factors on the instantaneous rate (hazard function) of ECC onset. Race/ethnicity and gender were considered fixed in the model. Several risk factors were considered time-dependent, such as age, microbiological profiles, oral hygiene, salivary cortisol, salivary IgA, and psychosocial risk factors. For the time-dependent risk factors, we used the observed values at the visit right before ECC onset (for interval-censored dental caries time) or the values at the last visit (for right-censored dental caries time).

The significance level of each analysis was set at 0.05. All analyses were implemented with SAS 9.4 (SAS Institute Inc.).

Results

Descriptive Statistics

The demographic characteristics of the cohort of 189 caries-free children at baseline are presented elsewhere (Kopycka-Kedzierawski et al. 2021). The mean age of the children upon enrollment was 29.5 mo (SD, 9.1). Almost 75% of the families lived in an urban setting; 64.5% of the parents/primary caregivers’ educational attainment was less than high school, high school equivalent, or GED. The findings underscore that the study sample is at high sociodemographic risk, based on measures of education, place of residence, and income.

Table 1 depicts descriptive data of carious surfaces by study visit. The mean number of carious surfaces was 0.6 at 6 mo, and it increased to 2.3 at 24 mo. The range of carious surfaces was 0 to 60 across all study visits. Differences in carious surfaces by gender, race, ethnicity, and baseline age were not statistically significant (Kruskal-Wallis chi-square test, P > 0.05). As shown in Table 1, we started with 189 children at baseline; 130 children attended the last study visit. Despite the COVID-19 pandemic, the retention rate was approximately 70% over the study period.

Table 1.

Descriptive Statistics of the Carious Deciduous Surfaces by Study Visit.

	No. of Children		No. of Decayed Surfaces (d)
Study Visit	Examined	With Decayed Surfaces (d > 0)	Mean	SD	Range
Baseline	189	0	0	—	0
6 mo	152	20	0.6	2.0	0-13
12 mo	137	26	0.8	2.3	0-18
18 mo	122	39	1.4	2.8	0-17
24 mo	130	48	2.3	6.8	0-60

The Kaplan-Meyer estimators of the survival functions are presented in Figures 1 and 2.

Figure 1.

Kaplan-Meier estimator of overall survival function of time to early childhood caries onset.

Figure 2.

Kaplan-Meier estimator of overall survival function of time to early childhood caries onset stratified by baseline salivary mutans streptococci (MS; cutoff, 10⁵).

As presented in Figure 1, the caries-free survival probability decreased with time for the study cohort. Caries-free survival probability was much lower for the initially caries-free children who harbored high salivary MS levels at baseline when compared with the children with low MS levels at baseline, as presented in Figure 2. The log-rank test indicates a significant difference between the groups in caries-free survival time distribution (P < 0.0001).

The results of the Cox proportional hazards model are presented in Table 2. The hazard ratio of 1.08 for the salivary MS (P = 0.01) suggests that for every unit of increase of salivary MS in log scale, the instantaneous rate of ECC onset increases by approximately 8%. None of the other risk factors in the model were independently significantly associated with ECC onset.

Table 2.

Results of the Cox Regression Analysis.

Variable	Parameter Estimate	SE	Hazard Ratio	Wald 95% CI		P Value
Age	0.0049	0.0167	1.005	0.9726	1.0383	0.7691
Sex: male vs female	−0.1484	0.2912	0.862	0.4872	1.5255	0.6103
Race
African American	0.2427	0.3583	1.275	0.6316	2.5727	0.4981
Caucasian	−0.0153	0.4806	0.985	0.3839	2.5260	0.9745
Ethnicity: Hispanic vs non-Hispanic	−0.2985	0.3878	0.742	0.3470	1.5866	0.4415
Log of
Mutans streptococci	0.0781	0.0329	1.081	1.0136	1.1532	0.0177
Lactobacilli	0.0560	0.0539	1.058	0.9515	1.1755	0.2992
Candida species	0.0671	0.0561	1.069	0.9581	1.1937	0.2311
Composite measure a
AUDIT	0.0755	0.0669	1.078	0.9459	1.2296	0.2590
CESD	0.0013	0.0178	1.001	0.9670	1.0369	0.9410
CHAOS	−0.0022	0.0561	0.998	0.8939	1.1139	0.9695
CTS	0.0046	0.0060	1.005	0.9930	1.0165	0.4362
EVENTS	−0.0310	0.0561	0.969	0.8685	1.0821	0.5801
PSWQ	0.0008	0.0103	1.001	0.9808	1.0212	0.9390
Time 0 b	−0.4904	1.8732	0.612	0.0156	24.0742	0.7935
Log area under the curve for cortisol	−0.2683	0.3325	0.765	0.3985	1.4672	0.4197
Secretory IgA
Concentration, mcg/mL	0.0042	0.0029	1.004	0.9984	1.0100	0.1564
Secretion rate, mcg/min	0.0003	0.0046	1.000	0.9913	1.0094	0.9486
Diet and feeding c	−0.0429	0.0359	0.958	0.8930	1.0278	0.2315
Oral hygiene d	−0.0026	0.0806	0.997	0.8517	1.1680	0.9740
Tooth monitoring e	−0.0184	0.2348	0.982	0.6197	1.5554	0.9376

AUDIT, Alcohol Use Disorders Identification Test; CESD, Center for Epidemiologic Studies Depression Scale; CHAOS, Confusion, Hubbub, and Order Scale (household disorganization and confusion); CTS, Conflict Tactics Scale; EVENTS, stressful life events; PSWQ, Penn State Worry Questionnaire.

^aHigher scores mean more symptoms or distress; see Methods for information on measures.

^bFrom wake-up time to first cortisol measurement.

^cIncluding sippy cup use, drinking liquids from the bottle, breastfeeding/taking formula at night, and having snacks during the day.

^dIncluding use of fluoridated toothpaste, frequency of toothbrushing, assisting with toothbrushing, and sharing a toothbrush with someone.

^eIncluding checking for cavities, child having a dentist and ever been to the dentist, taken antibiotics in the last 3 mo, and anyone in the household smoking.

Discussion

The results of this 2-y prospective cohort study advance prior work in demonstrating that salivary MS presence and high MS counts at baseline and longitudinally are significant predictors of ECC onset in initially caries-free preschool children, independent of psychosocial, sociodemographic, and oral health risks. These findings thus extend those of earlier investigators who demonstrated the relationship between caries in children and MS in cross-sectional studies (Klock and Krasse 1987; Alaluusua et al. 1989) and a recent umbrella review (Thang Le et al. 2021). Our results from the Cox proportional hazards model provide prospective evidence that each unit of increase in salivary MS level predicts an increased risk for ECC onset, consistent with previous reports that salivary MS presence and high MS counts are associated with ECC risk; this validates one aspect of the Caries Risk Pyramid conceptual model, mainly the biological risk factors that are most commonly incorporated into caries risk assessment models (Morou-Bermúdez et al. 2011).

Many studies have shown a positive association between the severity of caries and the level of MS in saliva or plaque but were unable to draw conclusions about causal influence or direct effect. For example, a retrospective cohort study of US preschoolers concluded that young children with very high levels of salivary MS had a greater presence and extent of dental caries at the first dental visit and were 6 times more likely to experience additional cavities over time as compared with those without salivary MS present (Edelstein et al. 2016); however, no causal inference could be drawn. Our study design, which started with initially caries-free children and prospectively assessed MS for caries onset, provides substantial methodological leverage for drawing a causal link. The current results are in agreement with one other study of initially caries-free children that reported that caries onset was associated with high plaque MS (Fan et al. 2019).

In contrast with other studies of young children (Plonka et al. 2013a; Plonka et al. 2013b), we did not find statistically significant associations between toothbrushing, oral hygiene habits, cariogenic snack, or drink consumption and ECC onset, independent of sociodemographic and psychosocial risk and oral microbiology. Lack of adjustment for these alternative sources of risk in prior studies is one likely explanation for discrepant results. Likewise, we did not find that key established measures of child stress exposure—namely, parental depression, alcohol consumption, family conflict and disorganization, stressful life events, and low resources (or low socioeconomic status)—had an independent prediction of ECC onset. A lack of independent prediction was not attributable to restricted range or variability in these measures in this sample. Psychosocial and sociodemographic components of ECC risk have been widely suggested, and several reports indicated associations between psychosocial stress and ECC. However, prior studies had not considered these factors independently of alternative risks and oral microbiological risk in particular. The reliance on cross-sectional analyses or samples from children who were not initially caries-free constitutes another important limitation to prior reports. Our analytic approach, which included a comparatively comprehensive set of predictors, provided a very conservative test of a risk mechanism. Such an approach is warranted given the large number of proposed risks and the tendency in past studies to consider risk factors in isolation.

One limitation of our study is that we focused on salivary MS, LB, and Candida species as cariogenic organisms and not on microbiome analyses, which provide additional insight into cariogenic etiology. Additionally, we purposely targeted a cohort at high psychosocial risk; as such, our findings may or may not generalize to populations at lower risk. Finally, our period of surveillance, 2 y, was adequate for detecting ECC onset in a sizable minority of children. Yet, the findings may be particular to short-term ECC prediction in young children; further follow-up of the sample would be needed to assess if the findings extend to longer-term disease. Significant strengths of the study offset these limitations, including a prospective cohort design that involved young, initially caries-free children, which enabled us to assess them prior to caries onset. The longitudinal and repeated measures nature of the study provided significant leverage to test hypotheses about risk exposure and ECC onset; our assessment of psychosocial measurements was detailed and covered many aspects child stress exposure. Finally, ECC was assessed at 6-mo intervals by examiners calibrated to the gold standard, ICDAS criteria.

Our results provide some of the strongest evidence to date for a causal link between MS and ECC and indicate that high salivary MS count may be the most important risk factor in the causal pathway of ECC. The effect size is also clinically significant. ECC continues to affect young children, particularly those from poor and minority populations. Given these findings, preventive strategies may be best targeted to establishing a dental home at an early age and including assessment of salivary MS with modern and affordable chairside screening tests. Clinical work and research efforts to reduce salivary MS counts in young children may be the most promising manner of improving oral health and thus preventing ECC in young children.

Author Contributions

D.T. Kopycka-Kedzierawski, T.G. O’Connor, contributed to conception, design, data acquisition, analysis, and interpretation, drafted and critically revised the manuscript; R.J. Billings, C. Feng, contributed to conception, design, data analysis, and interpretation, drafted and critically revised the manuscript; P.G. Ragusa, K. Flint, C.L. Wong, S. Manning, contributed to conception, data acquisition, and interpretation, critically revised the manuscript; G.E. Watson, contributed to conception, design, data analysis, and interpretation, critically revised the manuscript; S.R. Gill, contributed to conception, design, data acquisition, and interpretation, critically revised the manuscript. All authors gave final approval and agree to be accountable for all aspects of the work.