Caries is the disease process that causes cavities (Figure 1). The natural history of ECC begins with a newly erupted tooth that is not yet colonized by cariogenic bacteria. It progresses through the establishment of a biofilm that may be visible as plaque, followed by demineralization of the enamel resulting in a precavity lesion, or "white spot." With continued progression, the enamel breaks down, resulting in a cavity. In our modeling, all phases of this dynamic chain from initial biofilm formation to cavity are termed "caries," and the physical breakdown of the tooth structure (cavitation) is termed "cavities." A cavity becomes symptomatic when it progresses to affect dentin and pulp. Various preventive and disease management interventions may affect this progression at any point along the way. At each point in the simulation, children move in at least 2 directions as they age and as ECC stages progress. With intervention, children can move in an additional direction if the intervention is sufficient to change risk status.

We set up the model to continuously reflect the initial distribution of children among age, risk, and illness groups in the absence of any interventions. Interventions change rates of flow from one status to another and, over time, yield different patterns of ECC prevalence. The model calculates a number of variables including overall fraction of children with cavities and cumulative costs of restorative care so that program costs, cost savings for restorative care, and net costs can be attributed to different interventions and combinations of interventions.

We used multiple data sources to quantify the model and the effects of simulated interventions. To determine ECC prevalence by age and income and confirm the previously reported relationship between oral health and income, we obtained state-specific information from the Colorado Child Health Survey (9) (Rickey Tolliver, State of Colorado, written communication, November 5, 2009). We used a positive response to a question querying parents about children's prior dental pain, cavities, broken or missing fillings, or teeth pulled because of cavities as a surrogate for a finding of ECC on clinical examination. The National Health and Nutrition Examination Survey (NHANES) stratifies pediatric cavity experience by income levels of less than 100%, 100% to 199%, and 200% or more of the federal poverty level (FPL) (10), but Colorado data supported using income breaks of up to 200% of the FPL (18.6% ECC prevalence) for "high risk," 201% to 300% of the FPL (15.0% ECC prevalence) for "moderate risk," and 300% or more of the FPL (8.4% ECC prevalence) for "low risk." We categorized children for whom income data were not available (6.0% of respondents) as high-risk because they had similar ECC experience. Of the 431,070 children in Colorado's population of 0- to 5-year-olds, 31.4% were classified as high-risk, 17.8% were moderate-risk, and 50.8% were low-risk; 8.2% were aged birth to 6 months, 24.7% were 7 to 24 months, and 67.1% were 25 to 72 months. To quantify the fraction of children with untreated and treated cavities at baseline and to compensate for lack of parental knowledge about their children's nonsymptomatic cavity status, we adjusted Colorado findings to reflect treated and untreated cavity prevalence reported for the 1999-2002 NHANES survey after adjusting for Colorado's income distribution. A prior secondary analysis of these NHANES data (11) was used to calculate the selected Colorado income bands.

We derived the fraction of children with symptomatic cavities from the Government Accountability Office, which conducted a secondary analysis of NHANES data to determine the number of children considered to be "in urgent need of treatment" (12). To estimate precavity caries fractions, we backward-extrapolated the rate of change in cavities between age groups (between the youngest and middle age groups from the Colorado survey and between middle and older age groups from NHANES). Results were consistent with a study that measured both precavity lesions and cavities among children (13). We used the model to more finely calibrate rates of children moving between stages of tooth decay and finally adjusted the model by using data on rates of childhood dental fillings from the Medical Panel Expenditure Survey (MEPS) (14). We revised recurrence rates (new cavities among children with prior repair) upward to maintain the model's ability to reflect initial prevalence patterns. The model produces restorative visit rates that fall within the range suggested by MEPS data.

In addition to the prevalence of ECC, the model seeks to anticipate the extent of primary-tooth cavity experience, measured in average numbers of decayed and filled teeth. Using NHANES data (10), we adjusted for differences between national and Colorado's income distribution and rates of cavity experience for 6- to 24-month-olds and scaled downward from the rates for children younger than 6 years.

To calculate cost savings attributable to avoided restorative care from various interventions, we obtained data for care delivered in both dental offices and ambulatory or hospital sites under general anesthesia. For office-based care, MEPS reported average restorative dental costs of $216 in 2004 for all children, adjusted for inflation at 5% per annum to $276 in 2009 dollars. We attributed an average cost of $7,204 based on Medicaid data adjusted to reflect market-level charges for the 12% of children younger than 6 years requiring care under anesthesia reported by Colorado Medicaid (M. Sajovitz, Colorado Department of Health Care Policy and Financing, written communication, September 17, 2010) and the National Survey of Ambulatory Surgery (15) (M. Hall, Centers for Disease Control and Prevention [CDC] National Center for Health Statistics, written communication, August 4, 2010).

Among interventions of interest are those that delay transmission of cariogenic bacteria to children. To provide baseline values, the model was quantified with fractions of children by age and risk with detectable levels of cariogenic bacteria, specifically Streptococcus mutans. On the basis of studies of S. mutans prevalence among mixed-income (16) and low-income groups (17) and a putative "window of infection" of 19 to 31 months (18), we spread acquisition of S. mutans across the model's age groups. Simulations with the model were run over 10 years to determine changes in fractions of children younger than 6 years with cavities, untreated cavities, and symptomatic cavities; numbers of decayed and filled teeth; and costs of restorative care. We weighted results against estimated program costs to estimate costs, benefits, and net savings for various interventions. Interventions were applied to entire populations or to age or risk subgroups.

Interventions considered in the analysis included 1) educational programs that reduce consumption of sugary drinks, nocturnal bottle use, and other harmful behaviors; 2) efforts to reduce S. mutans transmission from parents and other caregivers to children using xylitol gum, chlorhexidine, or behavioral interventions; and 3) use of xylitol products directly with older children; 4) aggressive screening for and treatment of caries activity to reduce progression to cavities; 5) expanded use of fluoride varnish; 6) focused preventive care and education for children who already have cavities to reduce recurrence; 7) expansion of community water fluoridation to the entire population; and 8) motivational interviewing with strong educational and behavioral components. Motivational interviewing is a brief interactive approach to counseling and educating parents that focuses on skills that move patients to action. Interventions were clustered into 6 categories: fluoride exposure, transmission reduction, xylitol administration, clinical treatment, motivational interviewing, and combinations of these.

Simulation 1.1 assumes that community water fluoridation is expanded to 24.6% of Colorado's population not currently served; that fluoridation reduces cavity prevalence by 25.4%, half the average reduction reported by CDC (19) to compensate for other sources of pediatric fluoride exposure (William Maas, Pew Center on the States, written communication, September 28, 2010); and that the cost of fluoridation is $0.50 per person for the expanded population, including adults (20). The next 3 simulations consider variants of fluoride varnish application: to all children older than 6 months twice annually (1.2), to high-risk children older than 6 months 3 times annually (1.3), and to all children older than 24 months twice annually (1.4). All assume that fluoride varnish reduces decay of primary teeth by one-third (21,22) at a cost of $16 per child per application (William Maas, Pew Center on the States, written communication, September 28, 2010).

Interventions aimed at reducing transmission of cariogenic oral bacteria to children, simulations 2.1 and 2.2, are modeled on assumptions of 88% S. mutans colonization reduction among 7- to 24-month-olds and 64% reduction among 25- to 72-month-olds when mothers are provided with education and treated to suppress their oral bacteria (23,24). Associated caries reductions of 73% are anticipated (25,26), at a cost of $100 per mother.

The effect of treating children with xylitol is reported only for children aged 24 months or older (27). Because research on xylitol interventions has reported both low-impact (44%) and high-impact (73%) reductions in cavity experience (28), simulations are modeled separately for each of these levels. Simulations 3.1 and 3.2 address all children aged 24 months or older, and simulations 3.3 and 3.4 address only those of this age group who are at high risk. Simulation 3.5 posits xylitol intervention beginning at age 7 months for children of all risk levels. The assumed cost of xylitol interventions is $100 per child.

Simulations 4.1 and 4.2 assume white-spot lesions are identified and treated before they become cavities for children older than 6 months. Simulation 4.1 assumes a fraction of untreated caries that is treated per month equal to the model's rates for treating cavities, and 4.2 assumes a more aggressive program of screening and treatment. In simulations 4.3 and 4.4, follow-up care is intensified for children who have had prior restorative care to limit recurrence by 50% or 75%. The 2009-adjusted cost for caries management of $242 is the median between MEPS-reported costs for preventive and restorative care (14).

Motivational interviewing, with appropriate follow-up, can reduce cavity prevalence by 63% (29), at an estimated per-child cost of $100. Simulations 5.1 and 5.2 consider motivational interviewing for all children and for high-risk children.

Simulation 6.1 combines fluoride varnish for all children older than 6 months (1.2) with clinical treatment of active caries (4.1). Simulation 6.2 adds clinical care of cavities to reduce recurrence (4.3). Simulation 6.3 adds motivational interviewing (5.1). Combining interventions can have cumulative and complementary effects. Combining several interventions can produce a smaller fraction of children with cavities than can any of the single interventions.