Caries Preventive Interventions and Oral Health Inequalities: A Scoping Review

Scoping reviews are used to identify and provide an overview of the available evidence. The main consideration to perform a scoping review is when it is not aimed to produce a synthesized answer to the research question (Munn et al. 2018). Therefore, to answer the aims of this review, a scoping review was conducted. The review reports according to the PRISMA extension for conducting Scoping Reviews (PRISMA-ScR) (Tricco et al. 2018). The review protocol was published on PROSPERO (CRD42020156635).

A literature search was conducted in EMBASE, Medline (Ovid), Web of Science core collection, and Cochrane Central Register of Controlled Trials in November 2019, and a search update was performed in March 2021. The full search strategy was built with the support of the librarian of the Erasmus Medical Centre and is available in the supplementary file. The search was initially designed for EMBASE and adapted for all other databases. Search terms for dental caries were combined with terms for preventive medicine, intervention studies, social groups, and children. No date and language restrictions were applied.

For the selection of gray literature and unpublished results, to limit the risk of reporting bias, we performed an electronic search in WorldCat, Scientific Electronic Library Online (scielo.org), opengrey.eu, Open Access Theses and Dissertations (oatd.org), and Google Scholar, based on the following keywords: intervention, oral health, caries prevention, public health, and socially disadvantaged children.

Titles and abstracts were independently screened by 2 researchers (AM-L and YY) using EndNote to make the initial selection. All potentially relevant studies were read full-text by 2 researchers (AM-L and YY). In case of discrepancy, a third reviewer (LK) was consulted. Thereafter, reference lists of systematic reviews identified in the literature search and of all included studies were screened for additional relevant studies.

From the final set of relevant studies, the following data were extracted: study characteristics (first author, publication year, country, study design, study population size, age, duration, setting), type of intervention, control group, outcome(s), outcome assessment, the measures of social disadvantage (e.g., indicator of SES, ethnic background, or of area deprivation score), and effect size measures used.

The effect on dental caries of each intervention was evaluated for the lowest and highest group of social disadvantage and assessed by mean differences (MDs) or risk ratios (RRs) with corresponding confidence intervals (CIs) calculated by Student’s t test or the Mantel–Haenszel procedure with a significance level of P < 0.05. To identify the potential direction of each intervention on oral health inequalities, the (significance of) effect estimates were compared. Due to the differences in study design and outcome measures and statistical heterogeneity, no statistical synthesis of results was performed, and no pooled summary measures were presented. We categorized the differential effects of each intervention per social group as follows: 1) the intervention is indicative of a potential reduction in oral health inequalities: the intervention preferentially reduced dental caries in the most disadvantaged group; 2) the intervention is indicative of a potential widening of oral health inequalities: the intervention preferentially reduced dental caries in the least disadvantaged group; and 3) the intervention is indicative of no preferential impact on oral health inequalities: the intervention equally reduced dental caries in both social groups or did not show an overall effect. Accordingly, we reported determinants of the interventions, such as type, length, frequency, setting, or the level of influence. The level of influence on health behavior was categorized as individual (knowledge, attitudes, and beliefs), interpersonal (family, friends, peers, or interactions), organizational (institutionalized rules or regulations), community (social networks and norms), and policy level (local, regional, or national policies and laws) (McLeroy et al. 1988; Rimer and Glanz 2005).

This is a scoping review, and generally all available evidence regardless of the methodological quality is included. In order to examine how previous research was conducted and which elements contribute to the quality of the evidence, we additionally present a risk of bias analysis. These results will give guidance and recommendation for future research. To assess the risk of bias in the individual studies, the Quality Assessment Tool for Quantitative Studies from the Effective Public Health Practice Project was used (Effective Public Healthcare Panacea Project 1998; Jackson and Waters 2005). This is a validated assessment method to rate the methodological quality of both randomized and nonrandomized intervention studies recommended for evaluating public health programs (Jackson and Waters 2005; Armijo-Olivo et al. 2012). The tool includes 6 components, including selection bias, design, confounders, blinding, data collection methods, and withdrawals and dropouts. Study components were rated as strong, moderate, or weak, and the study was rated strong if none of the components were rated weak, moderate if 1 component was rated weak, or weak if 2 or more components were rated weak.

The study selection is summarized in the Figure. Electronic searches retrieved a total of 3,639 unique references. Titles and abstracts were screened, and 174 studies were selected for full-text screening. Finally, a selection of 14 articles was included in the present study. The main reasons for exclusion were non-European countries, no stratification into social groups, no evaluation of an intervention program, or studies without a control group.

A summary of the included studies is presented in Table 1. The date of publication ranged from 1989 to 2020. Of the 14 included studies, 3 were randomized controlled trials (Ellwoodet al. 2004; Qadri et al. 2018; Winter et al. 2018), 7 were quasi-experimental or nonrandomized studies (Winter et al. 1989; Evans et al. 1996; Freeman et al. 2001; Levin et al. 2009; Drosen et al. 2010; Wagner and Heinrich-Weltzien 2017; Kidd et al. 2020), and 4 were historical comparison studies, including repeated cross-sectional surveys (Dargent-Paré et al. 1999; Heinrich-Weltzien et al. 2007; McMahon et al. 2011; MacPherson et al. 2013). Of all studies, 5 were conducted in Germany (Winter et al. 1989; Heinrich-Weltzien et al. 2007; Drosen et al. 2010; Wagner and Heinrich-Weltzien 2017; Qadri et al. 2018), 7 in the United Kingdom (Evans et al. 1996; Ellwood et al. 2004; Levin et al. 2009; McMahon et al. 2011; MacPherson et al. 2013; Winter et al. 2018; Kidd et al. 2020), 1 in France (Dargent-Paré et al. 1999), and 1 in Ireland (Freeman et al. 2001). Three studies evaluated 1 or more components of the “Childsmile” oral health program (McMahon et al. 2011; MacPherson et al. 2013; Kidd et al. 2020). Most of the included studies were performed at nursery and primary schools (Dargent-Paré et al. 1999; Freeman et al. 2001; Heinrich-Weltzien et al. 2007; Levin et al. 2009; Drosen et al. 2010; MacPherson et al. 2013; Qadri et al. 2018; Winter et al. 2018; Kidd et al. 2020), 2 were performed at home (Winter et al. 1989; Ellwood et al. 2004), 1 was performed in a community setting (Evans et al. 1996), and 2 had a mixed setting at home and at a dental clinic (McMahon et al. 2011; Wagner and Heinrich-Weltzien 2017). Seven studies assessed caries in the primary dentition (Ellwood et al. 2004; Evans et al. 1996; Kidd et al. 2020; MacPherson et al. 2013; McMahon et al. 2011; Wagner and Heinrich-Weltzien 2017; Winter et al. 1989), 5 assessed caries in the permanent dentition (Dargent-Paré et al. 1999; Freeman et al. 2001; Heinrich-Weltzien et al. 2007; Levin et al. 2009; Qadri et al. 2018), and 2 studies assessed both (Drosen et al. 2010; Winter et al. 2018). The following indicators were used to measure social disadvantage of the children: parental employment status (Winter et al. 1989; Evans et al. 1996; Dargent-Paré et al. 1999), poverty (Freeman et al. 2001), area deprivation (Ellwood et al. 2004; Levin et al. 2009; McMahon et al. 2011; MacPherson et al. 2013; Kidd et al. 2020), ethnic background (Dargent-Paré et al. 1999; Heinrich-Weltzien et al. 2007; Drosen et al. 2010), or a summary score of 2 or more socioeconomic indicators (Wagner and Heinrich-Weltzien 2017; Qadri et al. 2018; Winter et al. 2018).

In Table 1, the global rating of the methodological quality of the included studies is presented, and an extended table including the rating of all 6 components is presented in Appendix Table 1. Of the included studies, 3 were rated as strong (Winter et al. 1989; Qadri et al. 2018; Kidd et al. 2020), 5 were rated as moderate (Evans et al. 1996; Freeman et al. 2001; Levin et al. 2009; Wagner and Heinrich-Weltzien 2017; Winter et al. 2018), and 6 were rated as weak in terms of study quality (Dargent-Paré et al. 1999; Ellwood et al. 2004; Heinrich-Weltzien et al. 2007; Drosen et al. 2010; McMahon et al. 2011; MacPherson et al. 2013).

Of the 6 studies that were indicative of a potential reduction in oral health inequalities (Evans et al. 1996; McMahon et al. 2011; MacPherson et al. 2013; Wagner and Heinrich-Weltzien 2017; Winter et al. 2018; Kidd et al. 2020), 3 evaluated (a part) of the “Childsmile” intervention (McMahon et al. 2011; MacPherson et al. 2013; Kidd et al. 2020). Five interventions included a daily exposure to fluoride via water or toothbrushing (Evans et al. 1996; McMahon et al. 2011; MacPherson et al. 2013; Winter et al. 2018; Kidd et al. 2020), 3 were performed from birth onward (Evans et al. 1996; McMahon et al. 2011; Wagner and Heinrich-Weltzien 2017), duration varied from 2 to 6 y, and the setting differed per study. All studies evaluated dental caries in the deciduous dentition and 1 additionally in permanent teeth (Winter et al. 2018). Furthermore, all programs incorporated the wider organizational or public policy level by implementing a setting where teeth are brushed at school regularly or having fluoridated water.

Of the 4 studies that were indicative of a potential widening in oral health inequalities (Dargent-Paré et al. 1999; Heinrich-Weltzien et al. 2007; Levin et al. 2009; Qadri et al. 2018), 3 had an infrequent exposure to the intervention program (Heinrich-Weltzien et al. 2007; Levin et al. 2009; Qadri et al. 2018), and for 1 study, frequency was unknown (Dargent-Paré et al. 1999). All studies were performed in primary schools and evaluated dental caries in the permanent dentition (Dargent-Paré et al. 1999; Heinrich-Weltzien et al. 2007; Levin et al. 2009; Qadri et al. 2018). Regarding the level of influence, 1 study focused at the individual and interpersonal level (Qadri et al. 2018) (i.e., having oral health education in class), and 3 others included a combination of the individual (i.e., provision of fluoride) or interpersonal (i.e., oral health education in class) and organizational level (i.e., implementing a fixed moment of mouth rinsing or toothbrushing in class) (Dargent-Paré et al. 1999; Heinrich-Weltzien et al. 2007; Levin et al. 2009).

Of the 5 studies that showed results that did not indicate a differential impact on oral health inequalities (Winter et al. 1989; Freeman et al. 2001; Ellwood et al. 2004; Drosen et al. 2010; Winter et al. 2018), 3 were performed at primary schools (Freeman et al. 2001; Drosen et al. 2010; Winter et al. 2018) and 2 at a home setting (Winter et al. 1989; Ellwood et al. 2004). The program frequency was mainly once or a few times a year (Winter et al. 1989; Ellwood et al. 2004; Drosen et al. 2010; Winter et al. 2018). The age when the intervention started and its duration varied. Two studies evaluated caries in mixed dentition (Drosen et al. 2010; Winter et al. 2018), 2 in deciduous dentition (Winter et al. 1989; Ellwood et al. 2004), and 1 in permanent dentition (Freeman et al. 2001). Also, the level of influence varied per study, with 2 studies solely focusing on the individual level (i.e., provision of toothpaste or toothbrush) (Winter et al. 1989; Ellwood et al. 2004), 1 merely on the organizational level (i.e., nutrition guidelines for school break) (Freeman et al. 2001), and 2 incorporated a mix of individual, interpersonal, and organizational level (i.e., oral health education, provision of fluoride or toothbrush, supervised toothbrushing in class, or fluoride varnish application) (Drosen et al. 2010; Winter et al. 2018) (Table 3).

The present scoping review identified (components of) interventions that have the potential to reduce oral health inequalities among children. Of the included studies, 6 presented intervention strategies with a potential reduction in oral health inequalities, 4 presented intervention strategies that were indicative of a widening in oral health inequalities, and 5 presented intervention strategies that were indicative of no preferential impact on oral health inequalities. Common characteristics of interventions that may reduce oral health inequalities were a high frequency of the intervention program (daily), an early onset (from birth onward), targeting at the deciduous dentition, and incorporating at least the organizational or public policy level regarding the level of influence.

In our study, we found that the timing of the intervention, from birth onward, may play a role in the effectiveness of the intervention, especially among the most disadvantaged groups. This is supported by a previous study that concluded that targeting children at a younger age (before 24 months) could lead to greater reductions in early childhood caries (Hirsch et al. 2012). Also, another study revealed the importance of intervening during the early years of a child’s life for future oral health (Chen et al. 2007). Besides, in older children with permanent dentition, disparities might be already wider due to long-term poor lifestyle behaviors that could already be present at an early age and continued throughout the years (Harris et al. 2006; Verlinden et al. 2019; Lioret et al. 2020). This theory is also supported by the WHO, which supports a good start to life for every child as a key strategy to limit the effects of social disadvantage on health during childhood and to reduce the risk of several chronic and lifestyle diseases (Marmot et al. 2012). Furthermore, the interventions with frequent contact moments had a higher potential to reduce oral health inequalities than less intensive health programs. The idea that repeated exposure to a beneficial health behavior increases health improvement over time, and thus that the intensity and frequency of an intervention program is an important element of an intervention to succeed, has been described earlier in the intervention mapping approach (Whitlock et al. 2002). Last, our review indicates that the level of influence may be a potential factor that affects the direction of interventions on oral health inequalities. The level of influence is based on an ecologic perspective model, which is advised to use for health promotion programs and when health behaviors need to be changed (McLeroy et al. 1988; Rimer and Glanz 2005). Most of the interventions that relied on only one of the levels of influence were equally effective in both social groups, whereas intervention programs that aimed to influence multiple levels, but at least and especially the broader organizational or public policy level (such as fluoridated water or provision of fluoride at school), showed a reduction in dental caries in particular among children from the most disadvantaged groups. This is supported by a recent review of Shen et al. (2021), which reported that interventions that aim to target the whole population are more likely to reduce oral health inequalities among children. Previous literature already indicated that when interventions solely rely on the individual voluntary behavior change, so-called downstream interventions, they may increase health inequalities (Mechanic 2002; Lorenc et al. 2013). This is due to the individuals’ lack of resources to adopt preventive and healthy behaviors, such as knowledge, self-efficacy, and social network (McLaren et al. 2010). On the other hand, “upstream” interventions try to influence the social or policy environment and are often recognized as interventions that are more likely to reduce health inequalities (Lorenc et al. 2013). “Upstream” interventions address the underlying cause of health disparities, by making adaptations to the social or physical environment of individuals. Therefore, “upstream” interventions will most likely reach the whole population, and especially the most disadvantaged groups will benefit since few individual-level actions or resources are required (Whitehead 2007; McLaren et al. 2010). The WHO supports this by setting goals to limit health disparities, which underlines that all levels from the wider society and system should be involved when taking action to improve equal health to the whole population (Marmot et al. 2012).

A strength of this scoping review is the comprehensive literature search in multiple databases. We aimed to include all relevant articles without language or publication year limits. Furthermore, we also included nonrandomized studies in this review. Although these individual studies may be more prone to bias, by including these studies, we were able to evaluate the effect of certain policies or regulations implemented at schools or communities and thereby provide a more comprehensive overview of potential measures that could be implemented to reduce oral health inequalities in the future. On the other hand, there are also limitations to consider. A limited number of relevant studies could be included in this scoping review, as we had to exclude 63 studies because of the nonstratification of social groups. Also, of the limited studies that were included, 6 were of low quality. Therefore, the results of this scoping review should be interpreted with caution. However, it clearly identifies the need for high-quality studies when aiming to reduce dental caries across different social groups. If randomization is not possible, a proper description of the study population, including an overview of baseline characteristics, or taking measures to reduce loss to follow-up, could ensure that the internal validity of individual studies will not be highly affected. We only included studies from European countries, although other studies have been performed that evaluated the effect of interventions across different social groups in other continents. Nevertheless, we decided to only include European studies since these will be more homogeneous regarding fluoridated water laws, health care and school systems, and the cultural and socioeconomic backgrounds of the population. Consequently, the results of this review are applicable for and generalized to a European population setting. Another limitation is that each study determined socially disadvantaged groups differently. Also, the determination and comparison of the lowest and highest groups might differ per study. For example, when the most extreme social groups are compared (e.g., lowest 10% versus highest 10%), a much larger difference will be revealed than comparing the lowest 50% versus the highest 50% of a total population, and this might influence the conclusion regarding the potential impact on oral health inequalities per study. Moreover, in our study, we evaluate the potential impact on oral health inequalities using the (significance of) effect estimates per social group, by only including studies that reported the effect of the intervention in groups of children with different social backgrounds separately. It should be noted that other methods exist to evaluate the impact on oral health inequalities potentially leading to a divergent conclusion (King et al. 2012). Furthermore, our scoping review mainly relies on the statistical significance of the individual study results to evaluate the differential impact of the intervention in each social group, which is highly influenced by the sample sizes of each study group (Appendix Table 2). Since most studies that were included in this review were initially not designed to study the effect of the intervention in different social groups of children separately, the sample size of each social group might be underpowered. Therefore, we cannot exclude the possibility that interventions with an effect size that clearly showed an absolute mean difference would have shown statistically significant results and a potential impact on oral health inequalities if it would have larger power.