Outcome of a National Education Program on Supervised Daily Toothbrushing and Biannual Fluoride Varnish Application on Dental Caries in Chilean Preschool Children: An Ecological Cohort Study Open Access

Dental caries is estimated to be the most prevalent health condition worldwide and represents a significant challenge for public health, especially in childhood [1, 2]. The Global Burden of Disease Study estimated that 7.8% of children suffer from untreated caries in deciduous teeth [3], which significantly affects disadvantaged groups in industrialised and developing countries [1, 4]. Therefore, the complex interplay between personal, social, cultural, economic, and environmental factors can make developing effective oral health programs difficult [5].

Chile is facing up to this challenge by developing new public health programmes that aim to improve child oral health and reduce inequalities, with upstream and downstream interventions and strengthening of primary dental care, with actions developed jointly by health and educational teams and members of the community via a multidisciplinary approach [6], in addition to a national drinking water fluoridation programme that began in 1985, that currently covers approximately 80% of the urban population of Chile, with a recommended fluoride concentration ranging between 0.6 mg/L and 1.0 mf/L [7, 8]. These programs have been developed by Chilean oral health policymakers based on evidence from successful international examples, including the national child oral health improvement programme for Scotland – “Childsmile” [9, 10], leading to the creation of the “Sembrando Sonrisas” (Sowing Smiles).

Briefly, the “Sembrando Sonrisas” program is a complex intervention with a preventive approach, offering specific interventions focused on preschool-age children (from 2 to 5 years old) in state-funded nurseries and schools [11], including three main interventions: (1) oral health examination by a dental team (dentist and dental nurse) performed in the classrooms; (2) training of educators and implementation of daily supervised toothbrushing in nurseries (including the delivery of toothpaste with 1,000 ppm of fluoride and toothbrushes exclusively for nursery use); and (3) applications of fluoride varnish in nurseries twice a year by dentists. After the pilot test, these interventions were collectively implemented nationally in 2015 [11].

When it is challenging collecting primary data at the individual level due to high costs or complex logistics, using routine administrative healthcare service data are considered necessary in evaluating public health [12]. Such epidemiological research can help inform modifications and the future direction of public health programs and policies [13].

Chile has developed a protocol for collecting routine oral health data from primary care clinics within the public health system [14]. This protocol represents an opportunity to evaluate the population’s oral health condition over time and explore correlations with the characteristics of each municipality, such as exposure to community water fluoridation (CWF), which has been posited to affect population caries levels [15].

Despite having routine health data on dental examinations and the interventions of Sembrando Sonrisas at the municipality level of aggregation in Chile, to date there has not been any evaluation of whether this programme is associated with differences in the caries levels of children. Therefore, this study aimed to undertake an ecological analysis of the trends in caries experience and its association with the implementation of components of the Sembrando Sonrisas program among Chilean children, as well as the inequalities related to the programme’s delivery and outcomes.

This study follows the reporting guidelines of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [16]. An ecological cohort, with the municipality/year unit of analysis, was constructed using aggregated municipality-level data of routinely gathered anonymised electronic health records from the Chilean public health system from the Department of Health Statistics and Information, which collates routinely collected and anonymises electronic health records monthly, including all records of all dental examinations and interventions performed in the “Sembrando Sonrisas” program on nurseries. These datasets were linked via Unique Municipal Territorial Codes [17] to other non-health data sources: (1) Superintendence of Sanitary Services of the Ministry of Public Infrastructure of Chile for fluoride concentration and coverage of drinking water; (2) National Institute of Statistics of Chile for the Municipal Rurality Population Proportion [18]; and (3) the Public Health Department of the University of Chile for Municipal Socioeconomic Development Index (IDSE) [19]. All the data were obtained between 2015 and 2019.

The outcome measure of this study was dental caries experience in deciduous teeth, defined as the presence of decay (into dentine), missing (extracted) due to decay, or filled deciduous teeth, in two cohorts of Chilean children who were 5 years old in 2018 or 2019. These children attended state-funded nurseries and were beneficiaries of the “Sembrando Sonrisas” program. The first cohort was studied between 2015 and 2018, while the second cohort was studied between 2016 and 2019.

The dental examination data are collected in every municipality during visits to the nurseries performed by dentists following the WHO guidelines under natural light. Chile has 346 municipalities, and including both cohorts, a total of 692 municipality/year records were obtained as maximum possible record. Since caries experience in municipality/year units of analysis can vary depending on the population, especially in areas where small numbers of children have been examined, all analyses were weighted by the total number of dental examinations conducted in each municipality per year. Only those municipality/year records that did not have missing data for all the variables in the analyses were included. Caries experience represented a percentage for each municipality/year and age, but those percentages were considered a continuous variable for all analyses.

The “Sembrando Sonrisas” program includes two main activities: the first is nursery-based fluoride varnish applications (5% NaF, 22,600 ppm fluoride), and the second is the delivery of an oral health kit containing educational materials for supervised daily toothbrushing, including toothbrushes and 1,000 ppm fluoride toothpaste.

A rate between the sum of the total number of activities of each variable received by children before the age of five at the municipality level and the total dental examinations in children before the age of five at the municipality level was calculated for each explanatory variable. For the fluoride varnish rate a categorical presentation sub-grouped by fifths was assembled for further analyses (defined as “Sembrando FV” throughout). For the delivery of oral health kits, a categorical presentation was constructed by dividing the continuous variable into two categories: (1) “partial,” with a rate of less than 1, and (2) “complete,” with a rate greater equal to 1. This cut-off was selected to compare those municipality/years where all children received the kits with those where this was impossible (defined as “Sembrando kits” throughout).

The association between caries experience and the explanatory variables was analysed by univariate weighted (by municipality/year total dental examinations) least-squares linear regressions. Three variables were evaluated as potential confounders in the association between caries experience and the explanatory variables using forward selection models: (1) deprivation, measured through IDSE; (2) rurality, measured through RPP; and (3) time, measured in calendar years. Multivariate weighted linear regression (adjusted by all significant confounders) evaluated the association between caries experience and the explanatory variables.

A forward selection regression model including all of the explanatory variables that were associated with caries experience in the univariate analyses for each age and adjusted for all potential confounders and a categorical presentation for CWF exposure with two categories: (1) “No CWF,” including all municipality/year in the cohort unexposed to CWF; and (2) “CWF,” including those exposed to CWF (defined as “CFW exposure” throughout). The selection criteria for the models were the adjusted R-square, Akaike Information Criterion [20], and Mallow’s Cp (Cp) [21].

The independent associations of each explanatory variable were then assessed using a multivariate weighted least-squares linear regression model, which was adjusted for all of the other variables associated with caries experience according to forward selection. This multivariable approach means that all potential determinants of caries experience were included in the same model, producing results for each variable adjusted for the other determinants in the model. In addition, two-way interaction terms were added to the model to test whether any other variables modified the association with caries experience.

Socioeconomic inequalities in caries experience were assessed using the Slope Index of Inequality (SII), representing the absolute difference in caries prevalence across socioeconomic groups, accounting for population size and relative ranking; and the Relative Index of Inequality (RII), which represents the ratio of caries prevalence between the most and least advantaged groups, standardised by the mean prevalence in the population [22, 23]. To calculate SII and RII, municipalities were ranked into deciles based on IDSE, and weighted linear regression models were used, incorporating the number of dental examinations per municipality as weights. Univariate weighted least-squares linear regressions were used to assess inequalities in the distribution of Sembrando variables by IDSE tenths, using as the outcome a continuous presentation of Sembrando kits rate and Sembrando FV rate at the municipality level.

For all statistical analyses, data were imported into SAS version 9.4 (SAS Institute, NC, USA). All analyses were performed at the 5% significant level, and 95% confidence intervals were calculated. The regression results were expressed as differences in the least-squares mean.

Over the study period, dental examination data were available for 309,360 5-year-olds from the 637 municipality/years included in the analyses. Figure 1 shows the distribution of caries experience.

For the forward selection model (Table 1), the incorporation of “Year” and “Rurality Population Proportion” had no statistical significance in the model (p = 0.649 and 0.128 correspondingly), only IDSE has statistical significance (p < 0.0001). These results indicate that the most parsimonious model only includes “IDSE” (p < 0.0001), leaving aside the incorporation of “Rurality Population Proportion” and “Year.” Table 2 summarises the univariate and multivariate (adjusted by IDSE) weighted least-squares linear regression models for differences in caries experience across categories of the explanatory variables.

In the univariate analysis, municipality-years with a partial kit delivery had a mean caries experience of 55.0% (95% CI: 53.3%, 56.7%), while those with complete delivery had 50.1% (95% CI: 48.6%, 51.7%). In the model adjusted for socioeconomic deprivation, caries experience in both groups was slightly higher overall, and the difference between partial versus complete delivery remained significant at 4.0% (95% CI: −6.3%, −1.8%, p = 0.0003) (higher in the partial group). For Sembrando FV, in the univariate analysis only the fourth quintile of fluoride varnish in nursery rate (IV: 1.72–1.90 per child) showed a significantly lower caries experience (49.7%, 95% CI: 47.2%, 52.3%) compared to the lowest quintile (I), with a difference of 4.8% (95% CI: −8.7%, −1.0%, p = 0.015); however, this difference was not significant after adjusting for deprivation.

Table 3 presents the results of the forward selection model, including both Sembrando interventions, “IDSE” and “CWF exposure.” In this model, “Sembrando FV” was not a significant explanatory variable (p = 0.535); indeed, adding this variable worsened the model fit (increasing Akaike Information Criterion and Cp and lowering adjusted R-square relative to models with the other variables.

Table 4 shows the final multivariable model including Sembrando kits and CWF exposure, adjusted by IDSE. Municipality-years with complete Sembrando kits had a caries experience 4.9% (95% CI: −7.1%, −2.6%, p < 0.0001) lower than those with partial delivery. Similarly, communities with CWF had a caries experience 7.5% (95% CI: −10.9%, −4.2%, p < 0.0001) lower than communities without CWF.

The model revealed a significant two-way interaction between “Sembrando kits” and “CWF exposure” (p < 0.0001). Therefore, a combined variable was tested in a model adjusted for IDSE. The category with “partial” delivery of “Sembrando kits” and “No CWF” had the highest caries experience of 65.1% (95% CI: 59.5%, 70.6%) and was used as the reference. No significant differences were observed between this group and the category with complete kit delivery and “No CWF” (p = 0.055).

The category with partial kit delivery and CWF exposure, as well as the category with complete delivery and CWF exposure, showed significantly lower caries experience. The latter group performed the best, with a caries experience of 51.7% (95% CI: 49.8%, 53.6%), which was 13.4% (95% CI: −19.3%, −7.5%, p < 0.0001) lower than the group with the worst results.

Using IDSE deciles as ranked socioeconomic categories, municipalities with lower deprivation had significantly lower caries experience. The SII was 14.7% (95% CI: 11.7%, 17.7%, p < 0.0001), indicating a 14.7% difference in caries prevalence between the most and least deprived municipalities. The RII was 0.283, reflecting that the most deprived municipalities had caries experience rates approximately 14.1% higher than the overall population mean, after accounting for population size and distribution.

Figure 2 shows the results of the socioeconomic inequality analysis of Sembrando interventions. Using the most socioeconomically deprived municipality/year group as reference (IDSE 1), no significant differences were observed in the rate of Sembrando kits and Sembrando FV with any category of IDSE.

This study used national municipality-level data from Chile to evaluate the association of a daily supervised toothbrushing and twice-per-year fluoride varnish application program and the caries experience of 5-year-old children who attend state-funded nurseries and schools. Unadjusted and deprivation-adjusted linear regressions revealed a significant association between municipalities achieving complete delivery of oral health kits for daily supervised toothbrushing and lower caries experience compared to those with partial delivery. In contrast, fluoride varnish application rates were not significantly associated with caries experience among the children covered by the program since its rollout. The distribution of both interventions appeared equitable across socioeconomic deprivation groups, but despite the equitable delivery of preventive interventions, we observed significant socioeconomic inequalities in caries experience.

The context in which daily supervised toothbrushing is delivered is essential for this intervention to succeed. According to a recent systematic review on the subject, the timing of the communication of the programmes, inadequate transfer of information among staff, frequent staffing turnover, lack of parental support, and staff feelings towards oral health were reported as barriers to daily supervised toothbrushing in nurseries and schools [24]. The findings showed that, in many municipalities, all children could not receive the kits with materials for daily supervised toothbrushing in nurseries and schools. Because rurality was not a significant informative variable in the model, and no differences were observed in the delivery of the kits between socioeconomic deprivation groups, we suggest that there is a need for studies that elucidate what conditions or barriers prevent universal access to materials to perform daily supervised toothbrushing in the Sembrando Sonrisas program. It is also necessary to explore under what conditions daily supervised toothbrushing is being carried out in nurseries and schools and whether teachers have the tools to support and supervise children in their oral healthcare.

Sembrando Sonrisas also includes twice-per-year fluoride varnish applications in nurseries and schools. Only in the unadjusted model was a significantly lower caries experience of 5-year-olds observed in those municipalities with higher FV delivery rates, compared to a group of municipalities with lower rates. However, this result did not remain in the socioeconomic deprivation-adjusted model, which was the most parsimonious according to the forward selection. A recent systematic review by de Sousa et al. [25] in 2019 concluded that fluoride varnish application had a modest and uncertain anticaries effect in preschool children, even in the context of preventive programmes. Oliveira et al. [26] suggested that the professional application of fluoride varnish twice a year for 2 years did not significantly reduce the incidence of caries in preschoolers from areas with fluoridated water in Brazil.

An RCT by Muñoz-Millán et al. [27] assessed a twice-per-year fluoride application intervention versus placebo in rural children of Chile without access to water fluoridation who were already receiving supervised toothbrushing. They concluded that biannual fluoride varnish application was not effective in preschool children from rural non-fluoridated communities at a high risk of caries in Chile [27]. Kidd et al. [28], in their evaluation of the Scottish Childsmile programme, determined that the component of fluoride varnish application in nurseries was not independently associated with the caries experience of children; instead, nursery toothbrushing and regular dental visits were associated strongly with a reduction in the caries levels of the children participating in the programme. McMahon et al. [29] showed in a high-quality RCT that fluoride varnish provided a modest and non-significant reduction in dental caries experience at a relatively high cost when delivered in a nursery setting twice a year from 3 years of age, over and above the multiple-component treatment as usual preventive interventions delivered in the Childsmile programme.

This is especially relevant for the Chilean Sembrando Sonrisas program, where this intervention represents the most significant part of the programme’s budget since its application is carried out exclusively by dentists in addition to the costs involved in purchasing the fluoride varnish [11]. However, variations in application techniques, frequency, and baseline caries risk could also influence the effectiveness of fluoride varnish [30], suggesting the need for further clinical research to clarify its role in such contexts.

An interesting finding in this study was the observed association between lower caries levels and the combined presence of supervised daily toothbrushing kit delivery and CWF in certain municipalities. Although supervised toothbrushing programs have been implemented in nurseries and schools in various countries, including some with concurrent access to fluoridated water [31], relatively few have undergone comprehensive evaluation. Among those, the Scottish Childsmile program [10, 28] is one of the most extensively studied. However, Scotland does not have water fluoridation, and, to our knowledge, there are no evaluations of supervised toothbrushing programs that have examined their potential combined association with CWF at the population level. The present study provides data on this overlap in implementation and its association with caries, contributing preliminary insight into an area that has not been well documented to date.

Sembrando Sonrisas proposes universal coverage in state-funded nurseries and schools; that is, it benefits all children ages two to five who attend these establishments [11], so knowing there is no unequal distribution of its interventions between municipalities despite their socioeconomic level is an important finding. Nevertheless, if evaluated from a national point of view, by only benefiting state-funded nurseries and schools, Sembrando Sonrisas corresponds to an intervention with proportionate universalism since the inclusion of this programme has not yet been considered in private nurseries [11]. Despite being a free right, not all children in Chile attend preschool education. According to OECD data, only 45% of 3-year-old children attend nurseries. Of this, 42% attend state-funded nurseries, and 57.4% attend private nurseries, which are not included in Sembrando Sonrisas. The rest of the children do not attend any preschool education [32, 33].

Despite the equitable distribution of the program’s interventions, significant socioeconomic inequalities in caries experience were observed. This finding underscores the persistent impact of broader social determinants on oral health and is consistent with previous studies highlighting the challenges of reducing oral health inequities, particularly in vulnerable populations [34, 35]. Thus, it is unsurprising to observe a similar association in Chilean children. Further research is needed to explore the long-term trends in caries experience and inequalities in Chilean children, and to monitor how the implementation of programs may relate to changes in oral health outcomes across different age groups over time.

The decision to employ an ecological cohort design over a pre-post design for the evaluation of the Sembrando Sonrisas program was informed by several key considerations. The ecological cohort approach allowed for the analysis of naturally occurring variations in program implementation across different municipalities, facilitating the assessment of associations between intervention coverage and caries experience at the population level [36]. This design may be particularly pertinent when randomization is impractical and when interventions are implemented at the community or policy level as it enables the examination of exposure-outcome relationships across diverse settings [37]. In contrast, a pre-post design, which compares outcomes before and after an intervention within the same population, may be susceptible to confounding by temporal trends and other concurrent changes and differences at the population level [13]. Therefore, the ecological cohort design was deemed more appropriate for assessing the program across various municipalities. However, given the inherent limitations of ecological studies, causality cannot be inferred [38]. Therefore, findings must be interpreted as associations at the group level and not as individual-level effects [37] and do not provide direct evidence to support the implementation of supervised toothbrushing or fluoride varnish applications as standalone policies for caries prevention in children.

Several important unmeasured confounders could have influenced the associations observed in this study. For instance, differences in home-based access to toothbrushing with fluoride toothpaste and parental involvement in children’s oral hygiene practices, nursery-level infrastructure, fidelity of intervention implementation, dietary behaviours and sugar consumption, sex, ethnicity, or broader municipal health initiatives were not available at the municipal level. These factors may vary significantly between municipalities and could have confounded the association between intervention coverage and caries levels. Additionally, the use of an ecological methodology introduces the possibility of misclassification bias, as area-level indicators – such as municipal socioeconomic deprivation – may not accurately reflect individual socioeconomic status, and measures of CWF may not correspond precisely to individual water consumption or total fluoride exposure from all sources. Using the calendar year variable to account for the timing of the interventions makes it impossible to accurately assess when the interventions were performed, which could also affect the caries outcomes. Information collected in nurseries and schools was not primarily intended for research, along with the fact that dentists are not calibrated in their diagnoses and treatments. Hence, it is possible to assume an inherent variability due to the complexity of the aggregated records from routine health information.

In conclusion, this study provides one of the first national-level assessments of the Sembrando Sonrisas programme in Chile using routine administrative and contextual data. The findings reveal significant associations between supervised daily toothbrushing kit delivery and lower caries experience in municipalities, particularly where this intervention co-occurred with CWF. No association was observed between fluoride varnish application rates and caries levels. Despite equitable distribution of interventions, substantial socioeconomic inequalities in caries experience of 5-year-olds remain. These results offer descriptive insight into how program coverage patterns relate to oral health indicators at the municipal level. While limited by the ecological design, the absence of individual-level data, and the possibility of unmeasured confounding, this study contributes preliminary evidence that may inform future research. Continued monitoring and more detailed evaluations will be important to understand long-term trends and to explore how different components of oral health policy relate to changes in child caries levels and equity.

This study was conducted ethically in accordance with the World Medical Association Declaration of Helsinki. By the provisions of Chilean law on Data Protection and Private Life, the processing of secondary health data carried out by the Ministry of Health ensures the confidentiality of personal data to users who register as part of the Public Health System. In addition, patients’ personal data are subject to data dissociation procedures, so the information obtained is anonymised and cannot be associated with a specific or determinable person. The law of access to public information establishes that aggregated pre-anonymised routine health data and the non-health information used in the assembly of this dataset can be publicly accessed under the principle of transparency of the Chilean Government’s public function. Therefore, no specific ethical approvals were required for the use of these data, and no consent to participate was necessary. The study protocol was approved by the Ethics Committee of the Faculty of Medicine, Veterinary and Life Sciences of the University of Glasgow, UK (Project No.: 20021’122).

The authors have no conflicts of interest to declare.

This work was funded by the “Development of Advanced Human Capital for Chile” programme of Chile’s National Research and Development Agency. However, the funding organisation was not involved in the study design, data collection, analysis, interpretation, manuscript writing, or decision to submit the manuscript for publication.

Andres Celis: conceptualization, methodology, formal analysis, writing – original draft, and funding acquisition. Jorge Celis-Dooner: writing – review and editing. Alex D. McMahon: conceptualization, methodology, formal analysis, writing – review and editing, and supervision. David I. Conway and Lorna M.D. Macpherson: conceptualization, writing – review and editing, and supervision.

Researchers who require access to the database can download it from the University of Glasgow’s “Enlighten: Research Data” repository, DOI: 10.5525/gla.researchdata.1331.