To examine the cost-effectiveness of fluoride varnish application by medical providers when implemented within a well-child periodicity schedule for Medicaid-enrolled children.
Cost-effectiveness was analyzed using published probabilities and costs. Input parameters included the effectiveness of fluoride varnish (35.4%) applied according to the well-child periodicity schedule up to 3 years of age at $16.00 per application, annual caries increment (14%), age-specific dental care usage rates (0.2% at 9 months to 19% at 42 months), and age-related nonhospital treatment costs ($292.00-$503.00) and hospital treatment costs ($2191.00-$2940.00). Sensitivity analysis was conducted to assess the effects for varying input parameters.
Well-child visits during primary care.
Children aged 9 to 42 months.
Application of universal fluoride varnish (fluoride varnish—all) at 9, 18, 24, and 36 months vs no intervention (fluoride varnish—none) was compared.
Main Outcome Measures
Cost per month without cavities and treatment averted during the first 42 months of life from a Medicaid payer's perspective.
Fluoride varnish improved clinical outcomes by 1.52 cavity-free months but at a cost of $7.18 for each cavity-free month gained per child and $203 for each treatment averted. Considerable uncertainty existed for some parameters. Fluoride varnish was cost saving when dental services and nonhospital treatment costs were 1.5 to 2 times greater, respectively, than our base case estimate.
Based on these assumptions, fluoride varnish use in the medical setting is effective in reducing early childhood caries in low-income populations but is not cost saving in the first 42 months of life. Potential total cost reductions with varying parameters suggest that evaluations using a longitudinal cohort are needed.
Disparities in children's oral health are large, and few improvements have been reported in oral health of high-risk populations.1 Results of the National Health and Nutrition Examination Survey suggest that the prevalence of dental caries and its treatment have become worse among children aged 2 to 4 years during the 1990s.2 By midchildhood, more than 50% of children have detectable caries, with disproportionate prevalence and severity of disease linked to lower socioeconomic and minority group status.3,4 These disparities arise from many complex determinants, including limited access to dental services by children at high risk.5 Few general dentists treat children younger than 4 years and only 7% report treating patients with Medicaid coverage often or very often.6 In 1996, the US Inspector General reported that fewer than 20% of Medicaid-enrolled children received any preventive dental care.7
Increasing difficulty in accessing dental services for young children at highest risk has led to innovative proposals to address this public health crisis.8 Medical settings provide an opportunity to increase use of preventive dental services for very young children during a developmental stage when oral health routines become established. Existing guidelines encourage primary care physicians to provide a broad range of anticipatory guidance to caregivers of preschool-aged children, including preventive dental care.9,10 The American Academy of Pediatrics and the American Academy of Pediatric Dentistry recommend that screening for early childhood caries should begin at the 1-year well-child visit.11 Almost a dozen states have expanded preventive dental services for children with public insurance to include topical fluoride application. However, the economic implications of physician-delivered interventions to prevent early childhood caries are not well understood.
Of the dental interventions available to nondental health care providers, fluoride demonstrates the strongest evidence for effectiveness in preventing dental caries.12 Physicians have prescribed fluoride supplements and recommended fluoridated toothpaste for many years, but topical application of fluoride in medical offices is a recent innovation. Fluoride varnish (FV) is one type of topical fluoride application by physicians that is increasingly being promoted for teeth in very young children because of its ease of application, effectiveness, and safety. When applied biannually for 2 years by dental professionals, it can reduce dental caries in the primary dentition by approximately one third.9,13,14
The purpose of this study was to examine the cost-effectiveness of FV application by physicians using a well-child periodicity schedule (WCPS) to prevent the occurrence and progression of early-stage dental caries in a high-risk population. Because of the early stages of adoption of this intervention among public insurance programs, this analysis does not use data from any current program; rather, the analysis is based on estimates and costs from the literature. Providing universal FV applications (FV–all [FVA]) to Medicaid-enrolled children was compared with providing no such intervention (FV–none [FVN]) to Medicaid-enrolled children during the first 42 months of a child's life.
Decision analysis was used to assess the effects of dental disease and treatment costs in children under FVA and FVN strategies in a medical setting (Figure 1). We developed a Markov model of the natural history of dental caries, fluoride effectiveness, and associated health states (Figure 2). A Markov model is a recursive decision tree and is described by the ovals in the model representing the various health states and the arrows representing the transitions from one health state to another. Each arrow is associated with a probability that represents the likelihood that a child will transition from one health state to the next in a 3-month cycle. For example, a child may enter the model with no disease and in subsequent months either remains caries free or progresses to the caries state. Once the child has experienced dental disease, he or she can either cycle within this health state if no treatment is rendered or transition to receiving dental care. Once treatment is completed, the child returns to a caries-free health state and is at risk for subsequent cavitation. In less severe circumstances, treatment may be rendered in a dental office. When the disease is severe, multiple teeth are involved, or a child is medically compromised, treatment in a hospital with the patient under general anesthesia may be required. Either of these treatment scenarios results in a return to a caries-free state after treatment completion. A cycle spent in the no-disease health state is associated with a gain in effect of 3 cavity-free months, and transitions into dental care are associated with costs representing the level of treatment. Costs of FV treatment are incurred at cycles that correspond to selected visits in the WCPS.
Decision tree illustrates effects of dental disease and treatment costs after fluoride varnish (FV) application in children ages 6 to 36 months in a medical setting.
Markov model used to assess cost-effectiveness of fluoride varnish in the primary dentition.
We derived estimates from the literature for age-specific disease incidence, fluoride effectiveness, and treatment costs. The strategies were compared by determining the cost per additional cavity-free month during the first 42 months of a child's life. The intervention consisted of FV application by the pediatric primary care provider during regular well-child visits at 9, 18, 24, and 36 months of age during the WCPS. The first cycle of the model began at 9 months, the average age at which teeth first erupt into the mouth. Although many children have a 12-month medical visit, we chose 18 months as the second time point because the 12-month visit involves multiple activities, making FV difficult to add. The 1-year interval between the 24-month and 36-month visits reflects the American Academy of Pediatrics–recommended WCPS guidelines.15 Although the duration of treatment was the first 3 years of life, the cycles extended to 42 months to account for benefits incurred after the last FV application at the 36-month well-child visit. A half-cycle correction was applied to effectiveness. All analyses were performed using the TreeAge Pro Healthcare module (TreeAge Software Inc, Williamstown, Mass).16
The analysis was conducted from a Medicaid payer's perspective; thus, we assumed it would include those children enrolled in Medicaid and who accessed the medical system for well-child visits. Because children enrolled in Medicaid are from a lower socioeconomic group, we assumed they would have an overall constant high risk for the development of dental caries. Our analysis did not explicitly incorporate the relationship between probability of disease and use of the medical system. However, treatment cost estimates were based on severity of disease in children who received treatment. We further assumed that FV has negligible adverse effects and that effectiveness is the same when applied by dental and medical providers at any stage of development of the primary dentition.13 The probability of developing dental caries was constant, and dental utilization varied with child age. We assumed that once dental caries were detected, the child would be referred to a dental home for comprehensive oral health care and could continue to receive preventive dental care in a medical office.
The probability estimates of various health states were derived from published data for other lower socioeconomic groups similar to the one under investigation (Table 1).14,17- 24,26 Values were extracted by one of us (R.B.Q.) and were reviewed by all of us.
Because of the lack of longitudinal dental caries studies, the probability of cavitation was obtained from a cross-sectional, child-level caries prevalence survey of 5171 preschool-aged children recruited from mainly low-income populations.17 Annual estimates for dental caries were available for 4 age ranges: younger than 12 months (0.5%), 12 to 23 months (6.4%), 24 to 35 months (21.6%), and 36 to 42 months (35.1%). These data were used to determine quarterly probabilities by first choosing the midpoint for each age range and subsequently deriving incidence data by linear interpolation as average 3-month rates of cavitation. These calculations resulted in a 0.035 constant probability of a child having any new dental caries in each quarter.
The probability of a child receiving dental treatment once a carious lesion developed was determined from estimates of annual dental services use in children aged 1, 2, and 3 years enrolled in a state Medicaid program.18,19 The number of children receiving dental treatment was divided by the total number of children with dental disease. The latter figure was estimated by multiplying the prevalence of caries at ages 1, 2, and 3 years from Tang et al17 by the total cohort of children in the studies by Lee et al.18,19 Annual rates were estimated and then interpolated by the quarterly time points to obtain the number of children being treated, ranging from 0.03% to 19% in each 3-month cycle, depending on age.
To obtain the probability of treatment in the hospital, we used data from Kanellis et al20 to determine the percentage of full-time-equivalent children treated in the operating room at various ages (12-23 months [0.2%], 24-35 months [1.2%], and 36-47 months [0.9%]). We assumed a constant quarterly rate of the probability of receiving treatment in the hospital and used a step function in our analysis to reflect the variation across ages.
Three-month FV effectiveness estimates were calculated from a study that examined the caries-preventive effects of semiannual FV application in children beginning at 3 years of age.14,21 This study was chosen because it was a randomized, controlled trial of FV effectiveness conducted in preschool-aged children and had the youngest subjects at enrollment of any FV study. Results indicated a preventive fraction of 35.4% per 6-month application of FV, although ranges in the literature vary from 30% to 63%.9 Reasoning that the effectiveness of FV would undergo a constant proportional decline, we calculated effectiveness estimates for the intervals under study based on an average effectiveness of 35.4% for 6 months.14,21 The resulting average effectiveness in the first 3 months was 57.1% and the average effectiveness in the second 3 months was 27.1%. Applying this same rate of decline, the average effectiveness of FV declined to 14.1% between 6 and 9 months and to 7.4% between 9 and 12 months from the time of FV application.
A $16.00 FV application fee was assigned at the beginning of the 9-month, 18-month, 24-month, and 36-month cycles. This figure accounts for direct costs only and was the midpoint of the range for 2004 fees ($13.00-$19.00) that some Medicaid programs reimburse nondental providers for FV applications.22- 24 Because our analysis was conducted from a Medicaid payer's perspective, the calculated average reimbursement cost of the FV to physicians was an appropriate number to use, with no additional payment for FVA considered. Costs related to dental treatment in the hospital setting were obtained from the experience of preschool-aged children enrolled in the Iowa Medicaid Program.20 Medicaid reimbursements per child ranged from $2100.00 to $2940.00 once adjusted for inflation using the Medical Care Component of the Consumer Price Index to reflect the 2003 US $ value.25 Nonhospital (office-based) treatment-related costs were obtained from a retrospective study by Ramos-Gomez et al.26 We used the 20th percentile of nonhospital costs for children experiencing decayed, extracted or missing, or filled teeth (DMFT), an epidemiologic index to describe children's oral health status (2-5 DMFT = $170-$854 and 6-10 DMFT = $394-$1373). The costs associated with 2 to 5 DMFT were used for children younger than 24 months and costs for 6 to 10 DMFT were used for children older than 36 months. Interpolation of these 2 numbers was used to estimate nonhospital costs for children aged 24 to 35 months. The costs were inflated to reflect the 2003 US $ value and then deflated by 30% to reflect average Medicaid reimbursement fees as a proportion of estimated charges (Table 1).
The primary outcomes were the average overall cost of each strategy, the number of months without cavities per child, and the incremental cost per cavity-free month gained moving from FVN to FVA. The incremental cost per cavity-free month, or the incremental cost-effectiveness ratio (ICER), was calculated as the ratio of the difference in cost between alternatives to the difference in effectiveness between the alternatives. A secondary outcome included cost per treatment averted. The FVA strategy was cost-effective if it improved outcomes and increased total costs (FV plus dental treatment), but the cost per additional unit of outcome was deemed to be worthwhile to the purchaser. The FVA strategy was cost saving if outcomes improved and total costs were less than what dental treatment costs would be in the absence of the intervention. We conducted sensitivity analyses on the effect of varying the following parameters on our primary outcomes: frequency and effectiveness of FV application, probabilities of cavitation, receiving any restorative treatment, or receiving treatment rendered in the hospital, and costs (hospital, nonhospital, and discount rate). Because some of the parameters varied with age, multipliers ranging from 0.5 to 4 were used to assess the effects of changing these parameters on the outcomes associated with the 2 strategies. Costs (dollars) and effect (cavity-free months) were discounted using an annual rate of 3%, as recommended by the panel on Cost-effectiveness in Health and Medicine.27
The results of the base case cost-effectiveness analysis (discounted) indicated that FVA was more effective than FVN, providing an additional 1.52 cavity-free months per child by 42 months of age (Table 2). Restorative treatment costs were reduced by $52.00 in the FVA group. However, this reduction does not offset the $64.00 cost of 4 applications. When including the direct costs associated with the FV application, the increase in cavity-free months was at an additional cost of $10.93 during the entire period, or an ICER of $7.18 per cavity-free month gained. The ICER improved as children aged, because of the increase in disease prevalence (Figure 3). Our analysis also showed that using FVA would cost Medicaid $203.00 (beyond treatment and intervention costs) for 1 treatment averted (hospital or nonhospital) over the 42-month simulation period (Table 3).
Results illustrate changing incremental cost-effectiveness ratio over simulation time by quarter discounted.
Because of considerable uncertainty about some of the parameters, the base case used conservative assumptions. Therefore, sizable increases in some of these parameters may be a reasonable possibility. Because some of the parameters varied with the age of a child, the sensitivity analyses were conducted using a multiplier; for example, a multiplier of 2 for nonhospital treatment costs means that costs indicated in Table 1 are doubled. The base case values provided for all parameters were multiplied by the multiplier values indicated for each of the sensitivity analyses. Sensitivity analyses are reported using a 3% discount rate because all results were insensitive to reasonable changes for the discount rate.
The analysis demonstrated that FVA resulted in lower total costs (intervention and treatment) when FV effectiveness was 1.25 times greater than our base case assumption of 35.4% average caries reduction for 6 months. A sensitivity analysis was conducted to reflect the biannual application regimen reflected in the dental literature starting at the 9-month WCPS. Using this protocol, the total number of FV applications increased from 4 to 6 times. Results indicated effectiveness only slightly greater than the base case schedule at 1.87 cavity-free months, but with an ICER of $15.59 per cavity-free month gained, an additional $8.41 when compared with the base case results.
The results were sensitive to the overall probability of receiving treatment, receiving this treatment in a hospital, and rate of cavitation. The FVA strategy became cost-saving when the probability of receiving restorative treatment was 1.5 times greater than our baseline estimates of 0.2% to 19%, depending on the child's age; the likelihood of treatment occurring in the hospital was 1.5 times greater than the baseline case, ranging from 14% to 31%; and cavitation rates were nearly doubled.
Results were sensitive to costs of hospital and nonhospital treatment, with FVA becoming cost saving when nonhospital costs were doubled and hospital costs increased 1.5 times greater than for the base case.
Two-way sensitivity analysis was conducted to illustrate the joint effect of changing the probabilities of receiving treatment and nonhospital costs. Results indicated that FVA was cost saving when greater numbers of children received dental treatment and nonhospital costs doubled from the base case costs. In a second analysis, varying both the fluoride effectiveness and probability of children with caries seeking dental care, we found no evidence of cost savings at lower fluoride effectiveness levels. When fluoride effectiveness falls below two thirds of our base case effectiveness, FVA does not save money even with increased use of dental treatment (Figure 4).
Results illustrate 2-way sensitivity analysis between fluoride varnish effectiveness and dental utilization.
To improve access to oral health preventive services for children at risk, several Medicaid programs reimburse for FV application in the medical setting. Our analysis, from the Medicaid payer's perspective, found that FV application in a low-income population showed modest improvement in outcome of 1.52 additional months in a cavity-free state between 9 to 42 months of age but would cost Medicaid $7.18 per cavity-free month gained or $203 per restorative treatment averted.
Sensitivity analysis showed that assumptions of sometimes substantial increases in dental services utilization rates, nonhospital costs, FV effectiveness, and disease rates were required to demonstrate cost savings, with multipliers ranging from 1.25 to 2 times base case estimates. Nevertheless, the thresholds at which FVA became cost-savings fell within the range of possible values for most parameters, primarily because the absolute numbers used in our base case estimates were conservative, representing low dental services utilization and reimbursement rates among children enrolled in Medicaid reported in the literature. Since publication of these estimates, dental services in some states have increased among Medicaid populations because of the policy emphasis on access issues, often through increasing reimbursement fees to dental providers.28 Other changes that could result in dental service utilization rates that are higher than our base case include the addition of screening and referral to any preventive oral health program in the medical home, the recent policy by the American Academy of Pediatrics11 recommending dental referral of children at high risk at about 1 year of age rather than the traditional 3 years, and the increase in the number of pediatric dentists during the last decade.29
Increasing the probability of nonhospital treatment influenced the cost-effectiveness of our model and is notable because our base case estimates were selected at the 20th percentile from the only study in the literature reporting nonhospital costs in this age cohort.26 Doubling these conservative estimates results in costs that fall slightly above the midpoint of reported ranges. Independently, hospital treatment costs reflected the highest dollar amount in the model and, thus, have an important role, given the proportion of children requiring hospitalization.
Doubling caries rates that shifted FVA to a dominant strategy in the sensitivity analysis results in a disease rate that is also plausible among certain populations that use Medicaid services, including those of Hispanic and Native American origin.4,30 Similarly, the multiplier of 1.5 used for effectiveness of FV in our sensitivity analysis is within the range of preventive fractions observed in clinical studies.9 Our estimates may be optimistic, because overall effectiveness levels may be attenuated when applying FV in a younger population that may be more behaviorally challenging than older children.
In summary, under these study assumptions, FV application will require that Medicaid provide additional financial resources to prevent disease in children younger than 36 months. However, assuming continued FV applications, this intervention may result in cost savings beyond 36 months of age as illustrated by the decrease in cost per cavity-free month (as measured by the ICER) over time, declining from $578.00 initially to $7.18 at 42 months. Cost savings may result from the effects of increasing onset of dental caries with age and the more frequent applications observed in dental offices, reflecting similar periods as those used in clinical trials of FV effectiveness. Additional benefits not examined in this study may accrue to society through improvements in the oral health status of the nation's most vulnerable children. These benefits include an increase in children's ability to learn, decreased use of dental emergency services, and increased caregiver productivity by postponing or diminishing time away from work.30,31
Many issues related to the cost-effectiveness of an FV Medicaid program are important when considering its implementation in a medical setting. First, research is needed to assess children's transition of care after the intervention in a medical setting to a dental setting to determine the long-term economic effect of the intervention. Second, better targeting of children at highest risk for dental caries could help improve the economic attributes of these programs for very young children. For example, holding all other assumptions constant, achieving cost savings would require the ability to identify a subset of the population with cavitation risk 2 times higher than our base case estimate. Improved caries risk assessments by primary care medical providers that go beyond identifying children of certain socioeconomic background are needed.
Third, timing and compliance issues with WCPS are important in the ability to target this high-risk population, particularly in the latter part of our simulation period. A change to annual WCPS after age 2 years creates a gap in the opportunity to apply FV at the desired 6-month interval to achieve maximum effectiveness. In addition, FV application depends on well-child visit compliance. Evidence suggests that in high-risk populations adherence to the recommended WCPS is not ideal, and some states report that only 36% of Medicaid-eligible children younger than 2 years receive all recommended visits in managed care programs.32 Such patterns may decrease the opportunity of adequate frequency and timely application of FV. We conservatively accounted for this underuse by using only 4 of the well-child visits in our simulations. However, some of the FV application points occurred at times when the frequency of well-child care visits is decreasing and risk for caries is increasing.
The results of this study must be interpreted in the context of our study assumptions. Overall, limited data were available to derive our probabilities and costs. For example, even though the assumption on effectiveness of FV was made for the primary dentition, the data available for this parameter included only children older than 2 years. This limitation points to the need for longitudinal, controlled studies of the effectiveness of FV in the Medicaid population starting in the first year of life. More up-to-date information and studies of better quality and quantity could help improve some of the uncertainties around the base case estimates and refine the accuracy of our economic evaluation.
Our findings examined intermediate measures of health and were limited to the first 42 months of life; thus, effect on quality of life beyond this time frame and spending for dental services after early intervention cannot be reasonably inferred and data are not available on which to model such an effect. The denominator in the cost-effectiveness ratios was expressed as caries-free child-months rather than a standard health outcome such as quality-adjusted life-years. Efforts should be made to standardize measures that can help facilitate comparison of such programs with other medical and dental interventions. Finally, the outcomes used in our study were dichotomous, namely, the presence or absence of disease. Incremental changes in severity of caries measured by the number of teeth or surfaces affected after an FV program could produce additional benefits that were not captured in our analysis.
Based on probabilities and costs derived from the literature and other secondary sources, our findings suggest that during the first 3 years of life FVA may not save money compared with FVN. However, the ICER decreased with time, approaching cost savings by 48 months of age. The sensitivity analyses also found cost savings for the intervention under higher but reasonable dental service utilization rates and nonhospital costs, which were estimated conservatively in our base case because of the traditionally poor access to dental care for children enrolled in Medicaid. The estimates in this study are based on data that are incomplete, and further research is needed to determine the benefits and costs of a dental FV program based in a primary medical care setting and how it interfaces with care in the dental home.
Correspondence: Rocio B. Quiñonez, DMD, MS, MPH, Department of Pediatric Dentistry, Campus Box 7450, School of Dentistry, University of North Carolina, Chapel Hill, NC 27599 (firstname.lastname@example.org).
Accepted for Publication: July 8, 2005.
Funding/Support: This study was supported by grant R01 DE 013949 from the National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, Md.
Quiñonez RB, Stearns SC, Talekar BS, Rozier RG, Downs SM. Simulating Cost-effectiveness of Fluoride Varnish During Well-Child Visits for Medicaid-Enrolled Children. Arch Pediatr Adolesc Med. 2006;160(2):164-170. doi:10.1001/archpedi.160.2.164