Wanted: Less Prevalence of and More Data on Early Childhood Caries

The 2016 Global Burden of Disease Study reported a worldwide total of 486 095 000 (95% uncertainty interval: 398 110 000–573 474 000) cases of caries in deciduous teeth (i.e., nonpermanent teeth), making caries in deciduous teeth the fifth-highest incident cause (out of 328) of disease and injury.¹ The study by El Tantawi et al. (p. 1066) provides new insight into the distribution of this scourge.

In their study, which focused on 193 United Nations (UN) member countries, the authors estimated the prevalence of early childhood caries (ECC) among children younger than 36 months and among those 36 to 71 months of age; their results showed mean rates of 23.8% and 57.3%, respectively. The authors then assessed these rates (as well as the availability of ECC data in the first place) with respect to their correlation with five macro-level factors: three factors related to health care systems (i.e., number of physicians per 100 000 population, number of dentists per 100 000 population, and universal health coverage) and two factors that were economic in nature (i.e., growth in per capita gross national income and growth in per capita health spending).

A NEW DATA REPOSITORY

The authors have provided a great service through their extensive literature review of early childhood caries data, aimed at circumnavigating the globe. They combined the literature review with existing data repositories (e.g., the country oral health profiles housed at Malmö University) to create a systemic accounting of available ECC data, country-level gaps in these data, and estimated ECC rates among children younger than 36 months and 36 to 71 months of age. Among other discoveries, the authors found that 39 of 193 UN member countries had data available from 2007 to 2017 for ECC among children younger than 36 months (another 13 countries had data predating 2007), whereas 86 member countries had data for children 36 to 71 months old (another 18 had data predating 2017). Eighty-five countries had no data available for either age group. The amount of work accomplished is to be applauded, and the end product will be a useful resource for future research on global ECC.

MOSTLY NULL FINDINGS

Because the authors were examining a very complex social ecological topic, drawing conclusions from simple univariate regressions or nonsignificant multivariate regression results can be tricky. None of the health system factors (number of physicians per 100 000 population, number of dentists per 100 000 population, universal health coverage) and neither of the economic factors (growth in per capita gross national income, growth in per capita health spending) were significantly associated with availability of ECC data for either age group in the authors’ multivariate logistic regressions.

With respect to the association of the health and economic factors with ECC prevalence, again there were no significant relationships among children younger than 36 months (according to multivariate linear regressions). However, growth in per capita gross national income was significantly associated with a higher ECC prevalence among children 36 to 71 months of age. As the authors’ most important finding, it is worth further discussion in future studies. El Tantawi et al. correctly question whether, rather than a true increase in ECC prevalence, the association is due to additional data gathering enabled by a country’s growing wealth.

VALUE ADDED

Although significant findings are always interesting to explore, they are not the only elements that add to our understanding. The authors have moved us forward on a relatively new line of inquiry, one that explicitly acknowledges exo- and macro-level influences on early childhood caries. Several other fields are finding that population-based approaches can be effective.² When applied to the ECC field, this broader scope has the potential to uncover additional insights or interventions that could improve outcomes, which in turn could also minimize the “blame the victim” phenomenon that arises when exclusive attention is focused on individual-level analyses.

Moreover, the authors insightfully incorporated a strongly salient topic from the global health and development sector, that of universal health care. Universal health care has reached such standing as a potential means to accomplish various health and well-being outcomes that all UN member states have agreed to attempt to achieve it by 2030 as part of the Sustainable Development Goals.³ Clearly, more research connecting ECC and universal health care is needed, as nonsignificant findings can help clarify future research questions (e.g., using insights from previous analyses to formulate new hypotheses and choose new variables) rather than provide definitive answers.

Finally, given the large disparity in ECC data availability between children younger than 36 months and those 36 to 71 months of age, the authors’ recommendation to disaggregate future studies according to these age groups is both logical and actionable.

MORE IS NEEDED

As is typical with intriguing research, El Tantawi et al. raise many new questions: If ECC data were obtained for the majority of countries (versus 22% of countries with data for children younger than 36 months and 46% with data for those 36 to 71 months old, as identified by El Tantawi et al.), would the findings change? Was universal health care not significantly associated with any dependent variable because the oral and medical health sectors are still too siloed? What additional exo- and macro-level factors might be important contributors to ECC data availability or to ECC prevalence?

Future research might adjust these associations for more confounding variables for which data exist. Dietary intake, for example, is essential to tooth condition. Among children younger than 36 months, breastfeeding prevalence would be appropriate to explore,⁴ whereas sugar or corn consumption might be a good option to assess among children 36 to 71 months of age. Other macro-level factors worth considering are a mother’s educational level, the under-5 mortality rate,⁵ healthy life expectancy at birth,⁶ and the sociodemographic index developed through the Global Burden of Disease Study.⁶

A balance has to be struck between bias introduced with each additional predictor and inclusion of predictors for which theory offers a credible argument. In such cases, even if these additional control variables are not statistically significant in a final model, this can be an important signal to policymakers and future researchers that such predictors have been taken into account. There are ever more creative ways to explore this balancing act. For example, the sociodemographic index is a summary indicator that combines per capita income, years of schooling, and the total fertility rate. Thus, this index could save degrees of freedom in the analysis (thus preserving statistical power), help avoid overfitting of a model, and potentially solve the earlier-mentioned complexity issue related to measures of wealth.

Given that dental caries can lead to systemic infections and even death,⁷ we must continue to insist that our country surveillance systems improve measurement of ECC and that our health (including oral health) systems improve both their reach and their effectiveness.