Spatio‐temporal trends in caries: A study on children in Berlin‐Mitte

Abstract Background Significant inequalities in caries distribution among children in Germany have been reported, but small‐scale areas remain understudied. Aim To examine spatio‐temporal trends in children's dental caries at the small‐area level in Berlin‐Mitte. Design Routinely collected data from Berlin's annual Health Examination Surveys were used, which also include information on age, sex, country of origin, and residential area. The study population consists of 14,866 children aged 5 to 7 between 2006 and 2014 in the district of Berlin‐Mitte. Outcome variables are the dmft (decayed, missing, and filled teeth), the presence of any caries experience, untreated caries, and caries risk. The outcomes are summarized descriptively and graphically presented for 10 quarters and 41 communities within Berlin‐Mitte. Results Relevant gaps in children's dental caries were discovered between the quarters of Mitte. Three quarters in the northeast part of Mitte have consistently indicated the lowest oral health status in all four outcomes, and children having high caries risk have been increasingly concentrating in this area over time. Despite the continuous improvements in the southern part, the averages in total of Mitte for all outcomes have risen. Conclusion Our findings confirm the spatiotemporally mounting disparities in children's oral health between the quarters in Berlin‐Mitte and that particular quarters need urgent attention. The small‐area approach made it easier and more effective to reveal the spatial distribution of children's dental caries at the local level. The small‐area analysis should be strongly encouraged in future caries research to narrow the inequalities in children's oral health.

Berlin has its own administrative system. Each of the 12 districts is divided into several subdistricts (Prognoseräume), which are then formed by quarters (Bezirksregionen). The smallest administrative unit are communities (Planungsräume). Mitte consists of four subdistricts (Zentrum, Moabit, Gesundbrunnen, and Wedding), 10 quarters, and 41 communities (Figure 1). The majority of dental studies, including studies from Germany (Jordan & Micheelis, 2016;Splieth et al., 2017), have been based on a large-area approach by focusing on a country, a region, or a city. Most existing health reports from Berlin also conducted their analyses at the levels of districts, or occasionally of subdistricts (Bettge & Oberwöhrmann, 2015;Lakes & Burkart, 2016).
Recently, the small-area approach is being used frequently in other disciplines (Lothrop, Hussaini, Billheimer, & Beamer, 2017), but it is rarely applied in dental research. The small-area analysis reveals specific problems in specific areas, and as such identifies high-risk areas more accurately than research at national or regional levels, helping to ensure that appropriate interventions are implemented where they are most needed (Piel et al., 2020). Identifying spatial disparities in oral health within small-scale areas can be an important basis to find hidden risk factors contributing to oral health inequalities.
Therefore, the purpose of this exploratory study is to examine spatiotemporal trends in dental caries among first-year schoolchildren at the level of small areas in Berlin-Mitte. We investigate the spatial distribution of dental caries within Berlin-Mitte, the temporal patterns from 2006 to 2014, as well as the spatio-temporal dynamics.

| MATERIALS AND METHODS
The district Mitte accounts for 4.4% (39.47 km 2 ) of the total area of Berlin and had about 380,000 inhabitants, while Berlin had a total population of 3.7 million in 2018 (Amt für Statistik Berlin-Brandenburg, 2018). Mitte is the second smallest of the 12 districts in Berlin, but the most preferred among foreigners without German citizenship such that 17.6% of the foreigners living in Berlin settled down in Mitte, and the percentage of Mitte's foreign residents is increasing annually. In 2018, Mitte was the district with the highest proportion (33.6%) of foreign inhabitants in Berlin and had the second-highest population density (Amt für Statistik Berlin-Brandenburg, 2018).
The anonymized data for our secondary analysis was derived from routinely collected data, which were generated by the annual health examination mandatory for children entering primary school in Berlin. Lately, routinely collected data (RCD) have been used more often in the biomedical field due to their advantages such as representativeness and generalizability, coverage of large populations, low cost, and longitudinal structure (Benchimol et al., 2015). But there are also several limitations in RCD, such as uncertain validity and lacking information on key confounders or risk factors due to the finite number of assessed characteristics (Jorm, 2015). Our data was gathered through a dental examination and a questionnaire answered by parents. Children were clinically examined by qualified dentists working at the Health Center of Mitte. This cross-sectional survey is repeated annually with a target population consisting of children between 5 and 7 years living in Mitte. We analyzed the data of Berlin-Mitte from 2006 to 2014. Furthermore, children who did not provide residency information were excluded from the analysis. The data comprised 10 quarters and 41 communities (Figure 1).
In the original survey a variety of variables was collected. Of importance for this research were demographic variables such as age, sex, and country of origin. The latter was classified into several groups: Germany, Arab countries, Turkey, Eastern European countries, Western European countries, other countries, and we additionally investigated observations with missing information on their country of origin. Since 2010, non-response to the question on the country of origin has not occurred due to systematic changes in the survey process. However, we included this group in our exploratory study, as it could provide important information for further detailed analysis. The assignment of origin was carried out by examiners according to the criteria proposed by the Berlin Senate Administration for Health and Social Affairs (Bettge & Oberwöhrmann, 2015). The anonymized dataset containing no personal information was obtained from the Bezirksamt Berlin-Mitte in accordance with the Raw-Material-Use Regulations. As this research uses anonymized data, ethical approval was not necessary. This study is described following the STROBE checklist recommendations.
The study investigated four outcome variables measuring dental caries. The first outcome is the dmft (decayed, missing, and filled teeth in primary dentition) index as a discrete variable, which was suggested by the World Health Organization (Petersen, Baez, & WHO, 2013).
The second outcome is the presence of any caries experience as a binary variable (yes/no). It was used to judge the existence of life-long caries experience in primary or permanent teeth by whether the sum of dmft and DMFT (decayed, missing, and filled teeth in permanent dentition) was greater than or equal to zero. Children whose dmft and DMFT scores are all zero are called caries-free. Germany is aiming at raising the proportion of caries-free children among 6-year-olds up to 80% by 2020 (Ziller, Micheelis, Oesterreich, & Reich, 2006). Based on this cut-off, areas in which 20% or fewer children have any caries experience are considered as being in a good oral health condition.
The third outcome is untreated caries as a discrete variable, which is defined as the sum of dt and DT (decayed teeth in permanent dentition) components of dmft/DMFT index. It indicates the total number of untreated decayed teeth at the time of examination. Caries risk as a binary variable (yes/no) represents the fourth outcome, measured by the criteria of the German Working Group for Youth Dental Care (DAJ, 2000). According to this criteria, 5-year-olds with dmft >4 and 6-7-year-olds with dmft/DMFT >5 or DT > 0 were identified as being at high risk of caries. For children belonging to this group in Germany, the participation in the intensive prophylaxis program is strongly recommended. Therefore, the outcome of caries risk can be used to identify children needing substantial treatment.
Characteristics of children are summarized with descriptive statistical methods. We report frequencies and percentages, as well as the mean and standard deviation (SD) for the outcomes and explanatory variables of this research and visualize them through maps and graphs to see the spatial, temporal, and spatio-temporal trends. All analyses and visualizations were performed using R (version 3.5.2) (R Core Team, 2018) and the R packages tmap (v.2.2) (Tennekes, 2018), and tidyverse (v.1.2.1) (Wickham et al., 2019).

| RESULTS
Our study included a total of 14,866 children examined between 2006 and 2014, with an annual average of 1,652 (Table 1 and Appendix Table). Out of 14,933 original participants, 67 had to be excluded from the analysis as they were not living in Berlin-Mitte or did not give any information about their living quarter and community. The mean age of participants was 5.6 years, the number of 7-year-olds was less than 1% of all children. The sex ratio was relatively well-balanced, there were slightly more boys than girls every year except for 2010, with an average of 50.9% for boys and 49.1% for girls. In terms of the children's country of origin, the proportion of Turkish origin was the highest (31.8%), children of German origin were about a quarter (25.6%), followed by Arab origin (15.1%).
The total mean dmft was 2.87, the average proportion of children with any caries experience was 63% (Table 1). The mean number of untreated decayed teeth was 1.33, the average proportion of children with caries risk was 25.2%. Boys reported a higher dmft than girls (mean: 3.07 vs 2.65) ( Table 1). In tendency, the older the children, the higher the dmft, as it was 3.43 among 7-year-olds in contrast to values of 2.77 among 5-year-olds and 2.93 among 6-year-olds.
The spatial distribution of the four outcomes during the entire period is shown in Figure 2. The maps depict the mean dmft, the aver-   In 2014, the difference widened further such that dmft of Osloer Strasse reached 4.06, in contrast to 1.00 of Government Quarter. In other words, there were great differences in mean dmft between quarters and the discrepancies were becoming even greater over time.
The dmft of Tiergarten South and Government Quarter, which had fewer participants, were fluctuating.
A similar tendency is found in the presence of any caries experience (Appendix Figure 2) and untreated caries (Appendix Figure 3).

| Comparison with other studies and a possible explanation for findings
Our results are consistent with a previous study (Watt & Sheiham, 1999), which reports large district and regional discrepancies in caries experience of children, and are in line with recent research suggesting that the place of residence is one of the social determinants of health and can thus help predicting health levels to some extent (Johnson, Hines, Johnson 3rd, & Bayakly, 2014;Marmot, 2010). There have been a number of studies indicating that socio-demographic factors highly correlate with dental caries among children. According to earlier findings, strong associations were established between socio-economic inequalities (i.e., in terms of ethnicity, wealth, parental education, and employment) and caries experience in children (Kramer, Petzold, Hakeberg, & Ostberg, 2018). An increase was also observed in the association between deprivation levels and dmft in 5-year-olds over time (comparing 2013 with 2003) (Masood, Mnatzaganian, & Baker, 2019). A systematic review supports the idea of an association between the parental educational and occupational background, income and having any caries lesions or experience (Schwendicke et al., 2015). A significant association between children's oral health and a migration background has also been noted in an earlier study (Cvikl et al., 2014).
The three quarters (Osloer Strasse, Brunnenstrasse North, Wedding Center) are the very areas where the percentages of unemployment, welfare recipients, and child poverty have been noticeably high (Foundation SPI, 2017). Above all, the share of children with a migration background in these quarters was exceptionally higher than in other quarters. An overwhelmingly large proportion (55-64%) of children from the three quarters were of Turkish and Arab origins (Appendix Figure 4). Their oral health status was at a lower level compared with other origin groups (Appendix Figure 5).

| Strengths and limitations
Our study analyzes dental caries in children based on a large, representative sample for Berlin-Mitte. The use of routinely collected data covering the whole population of first-year schoolchildren in Mitte facilitated the spatio-temporal analysis at a small-area level. Focusing on small areas at quarter and community levels helps to transfer the conventional perspective in dental research to the new dimension of small areas. As our findings reveal, the caries pattern at a small-area level can be completely different from that at national or regional levels. By observing a small area, the study provides more specific information on where disadvantaged children in terms of oral health are located. Furthermore, using routinely collected data from a repeated cross-sectional survey, changing characteristics of the district Mitte over time could be explored. As data of all children who participated in the health examination were analyzed, the risk of selection bias was lower compared to data obtained by other sampling methods.
One limitation of the study is that the demographic variables were limited to sex, age, and country of origin. Other socio-economic factors, such as parental education, income or employment status were not controlled for in this study. Another limitation is that the numbers of children by quarter originally invited to the health examination were not available due to German data privacy regulations, and consequently response rates by quarter could not be calculated. However, except for the subdistrict Zentrum containing quarters of a very low population density, the numbers of participants from the other quarters were steady overall during the study period. For that reason and the high sample size of our study, we did not consider this as biasing spatial distribution of dental caries.

| Policy implication
Caries in primary dentition is an important indicator for predicting future caries in permanent dentition (Li & Wang, 2002). Interventions to combat health inequalities should already begin in early childhood, during which higher returns on investment are expected than in adolescence (Marmot, 2010). In this context, a more effective and efficient approach for narrowing the oral health gap in Mitte would be to establish tailored programs focusing on those areas where high-risk children live. First, preventive programs could be implemented in kindergarten or school settings, which are recognized as the best places to perform group prophylaxis (WHO, 2003).