Cost-Effectiveness of Silver Diamine Fluoride Depends on Caries Activity: A Decision Analytic Model

Globally, 48% of preschool children are affected by early childhood caries with 9% of children suffering from untreated dental caries in primary teeth [1]. Untreated caries can lead to pain, infections, poor masticatory function, and negative impacts on quality of life [2]. Locally, 50% of Singaporean children suffer from dental caries by age 6, and 90% remain untreated [3]. The disease distribution is skewed with 16% of children carrying 78% of lesions; children with low socio-economic status are disproportionally affected [3]. This disparity stemmed from reduced access to care and poorer treatment outcomes for children with lower socio-economic status [4], increasing the need to provide more cost-effective treatment options compared to traditional restorative options.

Silver diamine fluoride (SDF) has emerged as an alternative treatment choice for caries management in children. SDF is a minimally invasive, painless technique, which inhibits microbial activity and facilitates caries arrest [5]. Previous studies found that 38% SDF was effective in caries prevention and arrest in children compared to no intervention [6, 7]. A recent meta-analysis of five clinical trials found that the proportion of arrested dentinal caries was 51.62% ± 27.40% when SDF was applied annually or biannually [8]. Direct restorations are currently the standard restorative treatment of choice for carious teeth as they are able to restore function and aesthetics; however, they require cooperation from children to be performed chairside [9]. While stainless crowns (including Hall technique crowns) showed the greatest success rates, in many countries, direct restorations using composite resin and glass ionomer cement (GIC) were still routinely used. The success rates of these direct restorations were affected by the use of rubber dams, which typically require local anaesthesia and patient cooperation [10]. Similar to SDF, sodium fluoride (NaF) varnish can be used in less cooperative children but has lower effectiveness than SDF at managing cavitated lesions [11]. In the Singaporean healthcare system, the majority of carious lesions are managed with direct restorations in cooperative children, while NaF varnish is used in pre-cooperative and uncooperative children to arrest carious lesions until behaviour allowed for more comprehensive treatment. Despite its effectiveness in managing existing carious lesions, the cost-effectiveness of SDF relative to NaF and direct restorations is not well established.

Cost-effectiveness analysis (CEA) is a method to assess the resource use and health outcomes across competing options, enabling better resource allocation and economic efficiency [12]. Previous CEA studies using models have concluded that SDF was cost saving at managing caries in children [13‒15]. However, many of these studies assumed equivalent outcomes for SDF and its alternatives, without accounting for the relative effectiveness of SDF and follow-up treatment required in cases of failure. Davis et al. [13] conducted a retrospective chart review of 104 children who had SDF treatment matched with children who did not. They estimated that children who received SDF incurred USD 402 lower restorative costs and USD 292 lower overall treatment expenditures than non-SDF-treated children. Johhnson et al. [14] developed a model comparing the cost of direct restorations against SDF using Medicaid data, taking into account sedation costs in a population with an assumed uniform caries prevalence. The study estimated that SDF treatment avoided 2,049 to 60,542 restorative visits among Medicaid-enrolled children, with avoided per-restorative visit costs ranging from USD 100 to USD 350. Nguyen et al. [15] found that SDF in conjunction with standard care, direct restorations without general anaesthesia (GA), resulted in an average cost saving of AUD 171.01 (95% CI, 185.91–156.10) per child aged 2–10 years, with a mean effectiveness of 0.298 (95% CI, 0.296–0. 301) per avoided GA compared to standard care alone. So far, there is only one randomized control trial comparing the cost-effectiveness of 38% SDF treatment against GIC restorations with 5% NaF varnish. The cost of arresting caries in one tooth with SDF was 67.30 Indian Rupees (INR), and INR 225.50 using GIC restorations with NaF varnish. The incremental cost-effectiveness ratio (ICER) for SDF for the arrest of one carious tooth was −INR 89.90 (−USD 1.08), suggesting that SDF was more cost-saving than GIC restorations with NaF varnish [16]. However, the study had a short follow-up period of 6 months and included only a single SDF application. Further research with a longer review period is needed to assess the cost-effectiveness of SDF relative to other treatment modalities.

Currently, there are no studies considering the impact of factors that can influence cost-effectiveness such as caries activity at baseline and the relative cost of SDF. As multiple carious teeth can be treated at a fixed cost using SDF or NaF while each direct restoration results in incremental costs, the number of carious teeth at baseline will affect the cost-effectiveness of the different management options. The aim of this study was to assess the cost-effectiveness of SDF relative to NaF and direct restorations, taking into account patients’ caries activity.

A CEA was conducted using a decision-tree model. The caries management option of interest was SDF treatment, and the comparators were NaF treatment, direct restorations performed chairside and under GA. Retrospective clinical data and data from literature were used in model construction. Ethics approval was granted by the National Healthcare Group Domain Specific Review Board (Singapore) (study reference 2021/00815). The study was reported according to the Consolidated Health Economic Evaluation Reporting Standards guidelines [17] (online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000544001).

This study adopted a healthcare system perspective in the Singapore public healthcare setting. The target population was children aged 1–6 years (preschool children) in order to examine caries management options of primary teeth. The total population was approximated to be 231,880 in 2019 [18].

An analytical decision-tree model was developed in R (https://github.com/tan-sharon/sdf-children-cea.git) (Fig. 1) to evaluate the relative cost-effectiveness of four caries management options for cavitated dentine lesions, International Caries Detection and Assessment System (ICDAS D4-D6), in primary teeth without pulpal involvement on a person level:

• (1)

  Twice yearly 38% SDF treatment

• (2)

  Twice yearly 5% NaF varnish treatment

• (3)

  Direct restorations performed chairside

• (4)

  Direct restorations performed under GA

A decision-tree model was chosen given the relatively short time frame prior to the exfoliation of primary teeth among preschool children in our study population. This study design is also in line with other prior studies assessing the cost-effectiveness of SDF among preschool children [14, 15], which enabled comparability across studies. The analytical time horizon was 12 months after first implementation of the caries management option. The reasons for adopting a 12-month time horizon were due to the lack of long-term data on the effectiveness of SDF and NaF in children [8, 11], as well as to align with the 12-month horizon used in previous CEA studies on SDF use in children [13, 15, 16].

Following treatment, participants would either become caries controlled or develop treatment failure (due to the failure to arrest primary caries, mechanical failure of restoration, secondary caries). Teeth with treatment failure were either restored or extracted. The probabilities of teeth becoming caries controlled (primary caries arrested or restoration intact) and developing treatment failure following the four caries management options were derived from the literature. The probabilities of restoration and extraction following treatment failure were derived from retrospective clinical data (online suppl. Table 2) from two public tertiary care centres, National Dental Centre Singapore (NDCS) and National University Centre for Oral Health Singapore (NUCOHS). As these centres also provide primary dental care for children, the clinical probabilities were likely representative of the broader Singaporean population. The baseline parameters of the model were reported in online supplementary Table 3.

The primary health outcomes were the number of carious teeth that became caries controlled and the number of tooth extractions avoided. ICERs, which summarize the ratio of differences in costs and outcomes between options, were tabulated as (1) incremental costs per carious tooth that became caries controlled and (2) incremental costs per extraction avoided.

Cost data were obtained from NDCS and NUCOHS for the year 2019. Costs were calculated in Singapore dollars (SGD 1.00 = USD 0.74). For the base-case scenario, fees from NDCS were used because of the larger patient pool (online suppl. Table 4). The breakdown of costs for each treatment option was reported in online supplementary Table 5.

For each caries management option, the entire study population was assumed to have received that option with the exception of direct restorations under GA since only a proportion of high-caries activity children underwent GA in the model. Modelling was carried out for the base-case scenario for children with high-caries activity and low-caries activity. The proportion of children with high-caries activity (15.6%) and low-caries activity (22.9%) was based on a previously published study among Singaporean preschool children [3]. In the base-case analysis, the number of teeth with caries at baseline was set at 1 and 7 for the low-caries activity and high-caries activity groups, respectively, based on mean estimates from the clinical datasets from NDCS and NUCOHS. For the high-caries activity group, the proportion of children requiring GA at the review visit following initial NaF and SDF treatments was set at 6.6%, based on clinical data. The study’s assumptions were (i) GA was never indicated for children in the low-caries activity group as the small number of lesions do not justify the medical risks of treatment under GA and (ii) replacement restorations for children that had restorations placed under GA initially were carried out chairside rather than undergoing a repeat GA (online suppl. Table 6). The study was performed from the healthcare system perspective in line with Singapore’s national guidelines for drug evaluation, without non-healthcare costs and indirect healthcare costs in the reference case analysis. As this study focuses on direct costs rather than indirect costs, the number of visits required to complete each management option was not modelled.

Deterministic sensitivity analysis was carried out for the number of carious teeth at baseline, 1 to 3 teeth as the low-caries activity group, and 4 to 10 teeth as the high-caries activity group [3]. One-way sensitivity analyses for base-case scenarios were also carried out for key parameters including the annual failure rate of restorations performed chairside, the rate of GA, the annual failure rate of SDF treatment, and the cost of procedures using NUCOHS prices.

To characterize the simultaneous effect of parameter uncertainty, probabilistic sensitivity analysis was carried out using Monte Carlo simulations with 10,000 runs. Gamma distributions were selected for cost parameters, beta distributions were selected for probabilities, and a uniform distribution was selected for the number of teeth with treatment failure. For each run, parameter inputs were randomly sampled from the distribution of each variable, and the option with the highest NMB was identified. The cost-effectiveness acceptability curve was plotted to quantify the probability of SDF being cost-effective (out of the 10,000 runs) at different WTP thresholds [21]. Statistical analyses were carried out in R version 4.3.2.

For both low- and high-caries activity groups, SDF incurred a lower total cost and lower effectiveness in terms of total caries controlled and extractions avoided compared to direct restorations chairside and under GA (Table 1).

For the low-caries activity group (1 carious tooth at baseline), the number of caries-controlled teeth was 27,410 for SDF, 42,480 for direct restorations chairside, and 11,098 for NaF. The number of extractions avoided were 49,504 for SDF, 51,614 for direct restorations chairside, and 47,220 for NaF. At baseline, the total cost of SDF treatment, direct restorations chairside, and NaF treatment were SGD 13,178,738, SGD 13,321,858, and SGD 15,208,010, respectively, for the population (Table 1). The NaF option was dominated, meaning NaF was more expensive and less effective than SDF and direct restorations chairside. The ICER for SDF compared to direct restorations was SGD 9 per caries-controlled tooth and SGD 68 per extraction avoided (Fig. 2). At a WTP threshold of SGD 30, comparing SDF with direct restorations chairside, the NMB was negative (favours direct restorations) at −SGD 6 per child for caries control. The NMB was positive (favours SDF) at SGD 2 per child for the avoidance of extraction.

For the high-caries activity group (7 carious teeth at baseline), the number of caries-controlled teeth were 130,709 for SDF, 202,570 for direct restorations chairside, 227,892 for direct restorations via GA, and 52,922 for NaF. The number of extractions avoided were 236,062 for SDF, 246,123 for direct restorations chairside, 249,668 for direct restorations GA, and 225,172 for NaF. At baseline, the total cost of SDF treatment, direct restorations chairside, direct restorations under GA, and NaF treatment were SGD 23,488,088, SGD 42,690,258, SGD 83,979,163, and SGD 34,084,209, respectively, for the population (Table 1). The NaF option was dominated, meaning NaF was more expensive and less effective than SDF. The ICER for SDF relative to direct restorations chairside was SGD 267 per caries-controlled tooth, while the ICER for SDF relative to direct restorations under GA was SGD 622 per caries-controlled tooth. Comparing SDF with direct restorations chairside and direct restorations under GA, the NMBs were positive (favours SDF) at SGD 471 and SGD 1,592 per child, respectively, for caries control. The ICER for SDF relative to direct restorations chairside was SGD 1,909 per extraction avoided, while that for SDF relative to direct restorations under GA was SGD 4,446 per extraction avoided (Fig. 2). The NMBs were positive (favours SDF) at SGD 522 and SGD 1,661 per child, respectively, for the avoidance of extraction.

For the low-caries activity group, NMBs for caries control and avoidance of extraction were positive (favours SDF) at WTP thresholds below SGD 7 and SGD 56, respectively. For the high-caries activity group, NMBs for caries control and avoidance of extraction were positive (favours SDF) at WTP thresholds below SGD 267 and SGD 1,907, respectively (online suppl. Table 7).

Sensitivity analyses were conducted to compare between SDF and direct restorations chairside as the NaF option was dominated. At a WTP threshold of SGD 30, NMB was always positive (SDF preferred) for the avoidance of extraction, regardless of the number of carious teeth at baseline. The NMB was positive (SDF preferred) for caries control only if the number of carious teeth at baseline was two or more and increased proportionally with the number of lesions (Fig. 3).

Increasing the annual failure rate of direct restorations chairside to 28% resulted in a higher ICER of SGD 84 per caries-controlled tooth for the low-caries activity group and SGD 424 per caries-controlled tooth for the high-caries activity group. The ICER per extraction avoided was SGD 597 and SGD 3,029 for low- and high-caries activity groups, respectively. NMBs were positive for all groups, favouring SDF. Decreasing the annual failure rate of direct restorations chairside to 2% resulted in a lower ICER per caries-controlled tooth of −SGD 64 (low-caries activity) and SGD 112 (high-caries activity) and lower ICER per extraction avoided of −SGD 456 (low-caries activity) and SGD 801 (high-caries activity). At a 2% annual failure rate of direct restorations chairside, the NMBs for both caries control and extraction avoidance for the low-caries activity group were negative favouring direct restorations chairside (Table 2).

In the high-caries activity group, decreasing the rate of GA for the management of treatment failures following SDF treatment resulted in a higher NMB per child of SGD 505 for caries control and SGD 557 for extractions avoided. Increasing the rate of GA resulted in a lower NMB per child of SGD 460 for caries control and SGD 511 for extractions avoided compared to the base-case scenario analysis (Table 2).

At a higher annual failure rate of 56.1% for SDF treatment, the NMB of SDF relative to direct restorations in the high-caries activity group was lower than the base-case scenario at SGD 381 for caries control and SGD 447 for extractions avoided. Contrarily, in the low-caries activity group, the SDF option was dominated by the direct restorations chairside option. The NMB per child for direct restorations was −SGD 18 for caries control and −SGD 8 for avoidance of extraction. With a lower annual failure rate of SDF of 19%, the direct restorations chairside option was dominated by the SDF option in both low- and high-caries activity groups (Table 2).

For the low-caries activity group, the SDF option was dominated by the direct restorations option with higher incremental cost, lower number of caries-controlled teeth, and fewer extractions avoided. For the high-caries activity group, the ICER comparing SDF to direct restorations chairside was SGD 186 per caries-controlled tooth and SGD 1,331 per extraction avoided, and the NMBs were SGD 311 and SGD 362, respectively (Table 2).

At a WTP threshold of SGD 30, SDF had a 99.8% probability of being cost-effective for caries control and 100% probability of being cost-effective for avoidance of extraction for the high-caries activity group. This was 41.5% and 49.2%, respectively, for the low-caries activity group (online suppl. Fig. 1).

SDF was the most cost-effective option in most scenarios, based on the ICERs and NMBs, which concurs with past studies [13‒15]. However, this study found that cost-effectiveness of SDF depended on caries activity, with direct restorations chairside being more cost-effective for caries control in children with 1 carious tooth. The clinical advantages of SDF such as lack of need for specialized equipment and minimal cooperation for application meant it is a viable alternative to direct restorations, particularly in high-caries children [22]. Similar to the present study, a previous study examining root caries in high-caries risk adults found that twice yearly SDF treatment was more effective and less costly than NaF [23]. While there are other caries arrest options such as nanosilver fluoride that do not result in dark staining, these agents are not available in Singapore and many parts of the world [24]. Even though the NaF option was dominated in all scenarios by SDF in this study, NaF’s use is still appropriate in certain clinical situations, such as when the dark staining of SDF is an aesthetic concern [25].

The exception to SDF being the most cost-effective was in the scenario of a child with low-caries activity (i.e., 1 carious tooth at baseline), with an ICER less than the WTP threshold when compared to direct restorations chairside for caries control. This differed from a previous model that found that SDF resulted in lower costs when compared to direct restorations in all of their scenarios [14]. However, that study did not consider variable caries activity in its study population. A possible reason may be that a fixed cost is incurred for SDF regardless of its use to treat single or multiple lesions. Additionally, SDF treatments may require multiple visits [26], thus making it less cost-effective for treating a single carious tooth. In Singapore, the cost-effectiveness of SDF may have been affected due to the cost incurred for a special licence to import SDF in Singapore, resulting in a higher cost of SDF treatment compared to that reported elsewhere [15].

Cost-effectiveness of SDF was affected by its clinical effectiveness against dental caries. The failure rate of SDF in arresting caries varies greatly in the literature, from 19.0% [27] to 56.1% [6]. Assuming the lowest failure rate of 19%, SDF was found to be cost-effective in all scenarios. However, the probability of 51.62% [8] used in this study is likely a more accurate reflection of the real-world effectiveness of SDF as the lower failure rate (19.0%) was derived from older studies [27]. Nonetheless, actual failure rate may vary between populations and each country should conduct its own modelling. Another factor that could affect cost-effectiveness of SDF was the failure rate of direct restorations. This study assumed a 20% failure rate for direct restorations in primary teeth [10], which is higher than the failure rates of 1.6–2.9% reported in permanent teeth [28]. The higher failure rates in primary teeth may be attributed to the lack of cooperation in young children and technical challenges in restoring primary teeth (e.g., size and tooth morphology), which are not issues associated with SDF treatment. The high failure rates of direct restorations have also led to the increased use of alternative restorative techniques such as Hall technique crowns, which have shown superior success rates [29]. Even after reducing the failure rate of direct restorations to 2%, SDF remains cost-effective in high-caries activity children.

With regard to controlling active caries and preventing tooth extractions, SDF was less effective compared to direct restorations chairside and under GA. The NMB of SDF increased proportionally with caries activity, implying that the decision to utilize SDF should consider the caries activity of the patient. As the study demonstrated that direct restorations were more clinically effective than SDF, the implication of this finding in clinical practices is the consideration of this option over SDF in children with low-caries activity, especially in urban areas with good access to dental care [30]. Moreover, considering the side effect of staining in SDF, the WTP threshold may be higher than the SGD 30 for direct restorations adopted in this study [31].

The strength of this study is that it is the first CEA study on SDF accounting for varying numbers of carious teeth and likelihood of treatment under GA. The study also utilized clinical data from multiple centres to improve the clinical relevance of the model in the context of the Singaporean healthcare system. The findings provide insights for clinicians and decision-makers to optimize the allocation of resources by delivery of the most appropriate and cost-effective options, especially in the development of clinical guidelines and public health policies. SDF being cost-effective for high-caries activity children in all scenarios tested suggests that it should be considered a viable alternative to conventional restorative treatment for these patients. A limitation of the study is the assumption of constant annual failure rates and extraction rate across all tooth types and surfaces, lesion size, and depth. Given that it has been suggested that SDF has a higher arrest rate in anterior teeth compared to mandibular first molars [32], and that the caries arrest rates of SDF in our study are based on studies that primarily focused on molar teeth, SDF may be more cost-effective if there is a higher proportion of anterior teeth with caries in the population. To mitigate this limitation, a sensitivity analysis was carried out varying the annual failure rate and number of teeth with treatment failures to improve the robustness of the model. Related to this is the relatively short time horizon of 12 months used in the model. The results may not be generalizable over a longer period given the wide variation in estimates of restorative failure reported in the literature [10]; the cost-effectiveness of SDF relative to direct restorations in the long run will depend on major assumptions regarding the relative failure rates of SDF and direct restorations over time. Another limitation is the use of an older source [3] for caries prevalence and distribution in the model; however, a range of carious teeth at baseline for low-caries activity (1 to 3 teeth) and high-caries activity (4 to 10 teeth) was used during deterministic sensitivity analysis to account for any change in prevalence over time. Lastly, despite the robustness of the model, the findings may be limited to the healthcare system in Singapore as the model assumed specific costs and conditions. Previous studies on the cost-effectiveness of SDF also suffer from this issue with methodology [13‒15] as they typically do not account for the varying characteristics of populations, healthcare systems, and local economic factors, limiting the generalizability of their findings. Therefore, parties should conduct its own CEA with input of local factors in order to obtain a more accurate picture of the cost-effectiveness of SDF.

In conclusion, SDF is cost-effective at a WTP threshold of SGD 30 in most situations. However, this study found that direct restorations were more effective for caries control and prevention of extractions. In view of both cost-effectiveness and effectiveness for caries control, the direct restorations option in low-caries activity children should be considered in setting policy for caries management. Local factors, which influence the cost and effectiveness of SDF treatments, should be taken into account when assessing the cost-effectiveness of SDF compared to direct restorations.

The authors would like to thank Professor Finbarr Allen for his mentorship and guidance on this study.

This study protocol was reviewed and approved by the National Healthcare Group Domain Specific Review Board (Singapore) (NHG DSRB study reference 2021/00815). The need for written informed consent was waived by the National Healthcare Group Domain Specific Review Board (Singapore) (NHG DSRB study reference 2021/00815).

The authors have no conflict of interest to declare. All authors have made substantive contributions to this study and/or manuscript, and all have reviewed the paper prior to its submission.

This study was supported by the National Medical Research Council (Grant MOH-000741): Health Services Research New Investigator (Grant HNIG20nov-0006).

S.H.X.T. is considered the first author and S.H. is the senior author. S.H., X.G., H.-L.W., B.W.P.L., and C.H.L.H. conceived the idea and were responsible for funding acquisition. S.H., C.W.M.L., and B.W.P.L. participated in data collection. S.H., Y.W., and S.H.X.T. conducted the data analysis. S.H., S.H.X.T., and C.H.L.H. led the interpretation and writing. All authors gave their final approval and agree to be accountable for all aspects of the work.

The data that support the findings of this study are not publicly available due to host institution data policies but are available from the corresponding author (S.H.) upon reasonable request.