AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 78:79-92 (1989) Caries Prevalences Among Geochemical Regions of Missouri C.F. HILDEBOLT, M. ELVIN-LEWIS, S. MOLNAR, J.K. McKEE, M.D. PERKINS, AND K.L. YOUNG Department o f Radiology, Mallinckrodt Institute of Radiology, St. Louis, Missouri 63110 (C.FH.1; Department ofsiomedicine, School ofDental Medicine (C.RH., M.E.-LJ, Department ofAnthropology (S.M.), and Medical Computing Services GroupIDepartment of Anthropology (J.K.M.), Washington University, St. Louis, Missouri 63130; Bureau ofDental Health, State of Missouri, Jefferson City, Missouri 65102 (M.D.P., K.L. Y ) KEY WORDS: Epidemiology, Geochemical factors, Trace elements, Fluoride, Water fluoridation ABSTRACT Our objectives were to determine how the prevalences of caries in elementary school children vary between geochemically defined regions of the state of Missouri and to compare this variation with that found for prehistoric Missouri inhabitants (Hildebolt et al.: Am. J. Phys. Anthropol. 75:l-14, 1988). Caries data on 6,584 school children were used in the study of second and sixth graders drinking optimally and suboptimally fluoridated water. Geochemical regions were based on maps recently published by the United States Geological Survey. Differences in mean caries scores and proportions of children with caries were tested by analysis of covariance, analysis of variance, Student t, and chi-squared tests. We found that caries prevalences do vary between the geochemical regions of the state. In the total sample, however, there were no significant differences between those children drinking optimally fluoridated water and those drinking suboptimally fluoridated water. We conclude that there is variation in caries rates among geochemically defined regions of the state and that geochemical factors associated with young parent materials may be antagonizing the action of fluoride. The caries-beneficial property of fluorine is well established, but other geochemical factors in the environment are also likely to have a n influence on caries rates (Curzon and Cutress, 1983). Much of the data that suggest the influence of these other factors come from recent epidemiological surveys that point to several trace elements and to the buffering capacity of water as likely caries-beneficial agents (Barmes et al., 1970; Bowen et al., 1977; Glass et al., 1973; Schamschula et al., 1978). Because of the multifactoral etiology of caries and the potential for numerous interactions between fluorides and geochemical factors, it is reasonable to consider not a single but multiple geochemical factors in caries surveys (Featherstone, 1983; Helle and Haavikko, 1977; Hopps and 0’Dell, 1981). As a n initial step in such a survey, distinct geographic divisions can be defined on the basis of geochemical combinations Of soils snd waters contained within (c) 1989 ALAN R. LISS, INC. the divisions. Such divisions have been made for the state of Missouri as the result of data collected during a recent geologic survey. This geochemical survey of Missouri is the most comprehensive study of its kind ever conducted in the nation (Cannon, 1978). During the study, 7,000 samples were analyzed for 30 to 40 geochemical factors, including concentrations of trace elements, hydrogen ions, nitrates, sulfates, and carbonates. So far, over 150 geochemical maps have been published as a result of this survey and are available through the United States Geological Survey (USGS). These maps were used to validate the geochemically defined regions that were used in our study (Hildebolt et al., 1988). With these regions and recent caries Received November 9, 1987; accepted August 10, 1988. 80 C.F. HILDEBOLT ET AL. survey data for Missouri school children (second and sixth grades), we examined the degree of variation in caries prevalences that occurs among the geochemically defined divisions of the state. An additional aim was to compare any discovered variations with those that we found for prehistoric inhabitants of the state, who showed significant differences in dental disease rates between these same regions (Hildebolt et al., 1988). MATERIALS AND METHODS During 1983 and 1984, the Bureau of Dental Health for the State of Missouri conducted a caries survey of second and sixth grade Missouri school children. These children lived mostly in rural communities; children in large metropolitan areas such as Kansas City and St. Louis were not surveyed. The geographical locations of the cities in which children were surveyed are illustrated in Figure 1A. The length of time that a child had resided within a school district was determined by a consent fordquestionnaire that was completed by the parent, as well as by personal interview with the child. Based on information contained on the questionnaire, a determination was made concerning whether the child's home was serviced by the same public water supply that was used by the school. Only life-long residents of school districts and children for whom it could be determined whether they were drinking water with fluoride levels 1) a t or above or 2) below recommended dosage schedules for optimal caries protection were included in the data. Fluoride levels whether natural or artificial were considered in making this decision. Data on the specific levels of fluoride in the drinking waters of the children surveyed were not collected. In this article, we refer to water at or above optimal fluoride levels as optimally fluoridated and that below optimal levels as suboptimally fluoridated (for Missouri, optimal fluoride levels vary between 0.8 and 1.2 parts per million.). Data on decayed, filled, and missing (or indicated for extraction) primary and permanent teeth of 6,584 school children constituted the data set, which was kindly made available to us for our comparative study by the Missouri Department of Health. Dental examinations were performed by Bureau of Dental Health dentists. These dentists were calibrated with regard to the repeatability of their determinations. A dental mouth mirror, explorer, and high-intensity light were used in the survey. No dental radiographs were taken, and data were not collected by tooth type (incisor, cuspid, premolar, and molar). Two types of geochemical maps were selected for use in our study: those depicting significant variations in the levels of geochemical factors between 1) water regions and 2) soil divisions (Fig. 1B,C, respectively). The rationale for creating and using these map divisions are given elsewhere (Erdman et al., 1976; Feder, 1979; Hildebolt et al., 1988; Miesch et al., 1976). They are based on the hypotheses that levels of geochemical factors in waters vary according to the underlying bedrocks, and levels in soils vary according to vegetative areas. The geochemical regions that were used in our study resulted from overlaying the water region maps with the soil division maps (Fig. 1D). Caries data were available for children living in 15 of these regions. These regions and their abbreviations are listed in Table 1 and are illustrated in Figure 1B-D. The mean caries scores and proportions of children with caries were calculated for four groups of children living within each resulting geochemically defined region: 1) second graders drinking optimally fluoridated water, 2) second graders drinking suboptimally fluoridated water, 3) sixth graders drinking optimally fluoridated water, and 4) sixth graders drinking suboptimally fluoridated water. Mean caries scores per region and group were determined by adding the numbers of carious, filled, and missing teeth to those requiring extraction and dividing this number by the number of children in the group. Missing primary teeth or unerupted permanent teeth were not included in the calculations. Proportions of children with caries were simply the percentages of children within groups with one or more carious teeth. The advantsges and disadvantages of these indices (and others) are described elsewhere (Hildebolt et al., 1988). The average age of the second graders was 7.5 years and for the sixth graders was 11.5 years. Thus the second graders had a mixed dentition of primary and secondary teeth, whereas the sixth graders had almost exclusively permanent teeth. Differences in means and proportions between regions were tested using analysis of variance (ANOVA), analysis of covariance, Student t, and chi-squared tests (SAS Institute Inc., 1985a). Plots of means and proportions by region were created with 81 CARIES PREVALENCES IN MISSOURI LACIAL DEPOSITS CAMBRIAN AND ;AMBRIAN AND ORDOVICIAN STRATA (SW MISSOURI) CRETACEOUS AND POPULATION CENTERS . CITIES WHERE CHILDREN A WERE SURVEYED B GEOHYDROLOGIC UNITS EAMBRIAN AND ORDOVICIAN VEGETATIVE AREAS STRATA (SW MISSOURI1 1 / Lf /CRETACEOUS AND TERTIARY STRATA I. GLACIATED PRAIRIE GEOCHEMICAL REGIONS 2. UNGLACIATED PRAIRIE 3 . OAK - HICKORY FOREST 4 . OAK - HICKORY- PINE FOREST 5. CEDAR GLADES 6 . FLOODPLAIN FOREST C D Fig. 1. A: Locations of cities in which school children were surveyed. B: Water Regions (Geohydrologic Units) o f Missouri (adapted from Feder, 1979).C: Soil Divisions (Vegetative Areas) o f Missouri (Adapted from Erdman et al., 1976). D: The Geochemical Regions of Missouri that result from overlaying the water region maps (B) with the soil division maps (C). 82 C.F. HILDEBOLT ET AL. TABLE 1. Geochemical regions used in study Quaternary Alluvium, Flood Plain Forest Missouri River Middle, Unglaciated Prairie Missouri River Middle, Oak Hickory Forest Missouri River North, Glaciated Prairie Missouri River North, Oak Hickory Forest Glacial Deposits, Oak Hickory Forest Glacial Deposits, Glaciated Prairie Deep Glacial Deposits, Glaciated Prairie Deep Glacial Deposits, Oak Hickory Forest Pennsylvanian, Oak Hickory Forest Mississipian, Oak Hickory Forest Cambrian Ordovician Southwest, Cedar Glades Cambrian Ordovician Southwest, Oak Hickory Forest Cambrian Ordovician Southeast, Cedar Glades Cambrian Ordovician Southeast, Oak Hickory Forest QAFF MRMUGP MRMOHF MRNGP MRNOHF GDOHF GDGP DGDGP DGDOHF PENNOHF MSOHF COSWCG COSWOHF COSECG COSEOHF MEAN CARIES SCORES FOR SECOND GRADE CHILDREN FOR MEAN CARIES SCORES SECOND GRADE CHILDREN i LUORIDATED DRlhh'lAC N U N F I X O R I D A ~ ~ EDPIVA'I O VL ~ ~ A T E R 309 347 273 340 MRNOHF MRNGP xUGDGP 144 .. i\ 3 36 GDOHF 278 MRNOHF *GDGP 7 q.5 . A I45 MRMOHF $\ I1ATEP 2,8 DGOGP - - ._ ~. COSEOHF .> B A MEAN CARIES SCORES FOR SIXTH GRADE CHILDREN MEAN CARIES SCORES FOR SIXTH GRADE CHILDREN N U N F L U U R I D A T f D D R I N K I N G II'ATER FLUORIDATED DRINKINC WATER 259 185 xUMRNOHF MRNGP I48 xMRNOHF 2.18 GDGP 2 09 +*GDOHF 83 CARIES PREVALENCES IN MISSOURI TABLE 2. Means and percentages for second and sixth grade children who have caries and are drinking optimally and suboptimally* fluoridated water Geochemical region Grade 2 No fluoride QAFF MRMUGP MRMOHF MRNGP MRNOHF GDGP GDOHF PENNOHF MSOHF COSWCG COSWOHF COSECG COSEOHF Fluoride QAFF MRMUGP MRMOHF MRNOHF GDOHF GDGP DGDGP DGDOHF COSEOHF Grade 6 No fluoride QAFF MRMUGP MRMOHF MRNGP MRNOHF GDOHF GDGP PENNOHF MSOHF COSWCG COSWOHF COSECG COSEOHF Fluoride CJAFF GRMUGP MRMOHF MRNOHF GDOHF GDGP DGDGP COSEOHF No. Mean Percent with caries 103 23 25 53 34 217 93 38 385 61 96 102 318 3.51 2.48 1.44 3.48 3.09 3.40 2.95 3.61 3.71 3.33 3.54 2.81 3.37 70 65 36 79 62 68 75 63 68 66 63 66 66 263 47 84 26 301 319 146 26 67 1 4.42 2.55 1.45 2.73 3.36 2.78 2.78 2.58 3.31 81 66 44 58 66 67 67 65 72 121 20 17 40 41 88 177 86 313 80 88 106 405 2.53 0.90 2.94 1.85 2.59 1.81 1.91 2.07 2.30 1.35 1.52 1.99 2.02 64 35 77 63 81 62 67 65 66 46 52 68 67 268 32 89 29 270 373 112 633 2.36 0.81 1.93 1.48 2.09 2.18 2.54 2.03 44 57 52 67 68 71 67 71 *For abbreviations,see Table 1. Fig. 2. The height of the peaks represent the mean :oronal caries scores for A: second grade children drinkng suboptimally fluoridated water; B: second grade chilken drinking optimally fluoridated water; C: sixth grade :hildren drinking suboptimally fluoridated water; and D: sixth grade children drinking optimally fluoridated water. For abbreviations, see Table 1. #, Larger of two means if they are meaningfully or significantly different; +, P < 0.1; *, P < 0.05. 84 C.F. HILDEBOLT ET AL PER CENT SECOND GRADE CHILDREN WITH CARIES PER CENT SECOND GRADE CHILDREN WITH CARIES tLUDI?ID,ITED DRln'tiI,\C POA'fl L 0 P I D A T I . U DRIA'tiJA'C IIATLP 6176 7925 6820 MRNGP GDGP 7 5 27 +nGDOHF 5769 MRNOHF A COSWCG 6646 GDGP d illlFP 66 12 +GDOHF 4423 . 6712 COSECG B A PER CENT SIXTH GRADE CHILDREN WITH CARIES A'ONt LL'OPIUATEU DRIA'KIPC ItATER Fig. 3. The height of the peaks represent the proportion with caries for A second grade children drinking suboptimally fluoridated water; B: second grade children drinking optimally fluoridated water; C: sixth grade children drinking suboptimally fluoridated water; and SAS/GRAPH (SAS Institute Inc., 1985b). (In this article, the term mean refers to the mean caries score, and proportion refers to the proportion of children with one or more decayed, missing, or filled teeth, a s explained above.) Multiple comparisons of pairs of means and proportions were used where appropriate. In addition, differences in means and proportions for children drinking optimally fluoridated versus suboptimally fluoridated waters were tested. These tests were done separately for second and sixth grade children. PER CENT SIXTH GRADE CHILDREN WITH CARIES ,VLUORlDATED DRI,VKlNC WATER D sixth grade children drinking optimally fluoridated water. For abbreviations, see Table 1. #, Larger of two proportions if they are meaningfully or significantly different; f , P < 0.1;*, P < 0.05. RESULTS Table 2 is a list of the mean caries scores and proportions for the four survey groups. Table 3 is a list of multiple comparisons by pairs of means and proportions for the suboptimally fluoridated groups versus the optimally fluoridated groups for each geochemical region. In this article, to demonstrate general trends, we consider P values less than 0.1 but greater than or equal to 0.05 as meaningful (noticeable, marked; see 85 CARIES PREVALENCES IN MISSOURI TABLE 3. Results ofpairwise comparisons of means and proportions for second and sixth graders drinking optimally and subopitmally fluoridated water* Grade 2 6 2 6 2 6 2 6 2 6 2 6 2 6 Geochemical region QAFF QAFF MRMUGP MRMUGP MRMOHF MRMOHF MRNOHF MRNOHF GDOHF GDOHF GDGP GDGP COSEOHF COSEOHF Group with highest mean or proportion Chi-squared test Student t test F (P < 0.03) F ( P = 0.02) NO F ( P < 0.09) NO F ( P < 0.03) NO F ( P = 0.01) NO F ( P = 0.09) NO F ( P < 0.05) F ( P = 0.07) *Blanks indicate P values that are more than 0.1.For abbreviations, see Table 1. F = optimally fluoridated water, NO F = suboptimally fluoridated water. Blalock, 1979:158; Sokal and Rohlf, tween some of the regions when pairwise 1981:164).P values less than 0.05 are consid- comparisons were made. These differences ered significant. The mean caries scores were were at several levels and sometimes varied plotted by region and are displayed in Figure depending on whether means or proportions 2A-D, and the proportions are in Figure 3A- were compared. Figures 2 and 3 show the D. Tables of statistical results are available ranges of caries prevalences by region. Cerfrom the authors upon request. tain groups show more clearly defined regional differences than others, as Comparisons of caries prevalences for demonstrated by their P values and as dechildren drinking optimally versus scribed below. suboptimally fluoridated water For the total sample there are no meaningful or significant differences between means Second graders drinking suboptimally for second or sixth grade children drinking fluoridated water optimally versus suboptimally fluoridated water. There are, however, meaningful difAmong regions, there was a marked overferences in means and proportions within re- all variation in proportions for children with gions (P < 0.1). Fourteen comparisons were caries (P < 0.07, chi-squared), whereas the made using both t and chi-squared tests. Re- overall variation in mean caries scores was sults are presented in Table 3. Of these, no- not meaningful (P > 0.1, ANOVA). Accordticeable differences were obtained for eight ing to the results of multiple comparisons of comparisons. Children receiving optimally pairs of regions, there were meaningful diffluoridated water had markedly lower caries ferences between means and proportions for prevalences in the GDGP, MRMOHF, some of the geochemical regions. For examMRNOHF, and GDOHF; however, surpris- ple, in general, the children of the MSOHF ingly, the second grade children receiving have markedly higher proportions than those optimally fluoridated water had significantly of the other regions, and the children of the higher caries prevalences in the QAFF re- MRNGP have noticeably higher means (P< gion of southeast Missouri, as reflected by 0.1). The children of the QAFF also have both means and proportions ( P < 0.03 and P high proportions and means, although these = 0.02, respectively). Moreover, the second rates are not meaningfully higher. Prevagraders drinking optimally fluoridated water lences of caries are lowest in the MRMUGP, in the COSEOHF of south-central Missouri MRMOHF, and MRNOHF. In the COShad a noticeably higher proportion of caries. WOHF, children have a high mean and a low proportion. This mean is not markedly differGeochemical regional differences by grade ent from the low means, but the proportion and water fluoridation is noticeably less than the high proportions Within each of the study groups, there were (P < 0.1). Children in the other regions tend differences of means and proportions be- to have medium rates of caries. 86 C.F. HILDEBOLT ET AL. TABLE 4. Summary for children drinking suboptimally fluoridated water* Geochemical region Caries incidence QAFF MSOHF MRNOHF High High High MRNGP COSECG Medium Medium PENN0HF Medium GDGP GDOHF COSEOHF MRMUGP Medium Medium Medium Low COSWOHF Low COSWCG Low MRMOHF None assigned Comments These regions contain both second and sixth graders who consistently have high means and proportions Sixth graders have high caries prevalences. The second graders have medium to low prevalences; however, based on the city with the largest sample size (N = 22) for this region, the second grades have high prevalences These children have medium caries prevalences For the sixth graders, the mean is medium and the proportion is high. For second graders, the mean is low, and the proportion medium The sixth graders have a rather high mean and a medium proportion, whereas the second graders have a high mean and a low proportion The children of these regions generally have medium prevalence of caries Both second and sixth graders tend to have low caries prevalences Although the second graders in this region have a high mean, the proportion is low. For the sixth graders, both the mean and proportion are low Both proportion and mean are low for the sixth graders. For the second graders, the mean and proportion are medium The sixth graders have high caries prevalences, and the second graders have low caries prevalences. The numbers of children surveyed in each city is low (N < 13) For abbreviations, see Table 1 Sixth graders drinking suboptimally fluoridated water There are highly significant overall variations in means and proportions between regions ( P < 0.001, ANOVA and chi-squared). The highest prevalences of caries are in the children of the MRMOHF, QAFF, and MRNOHF. Children living in the COSWOHF, COSWCG, and MRMUGP have the lowest prevalences of caries. The remaining regions form a middle range of caries prevalences. In general, the high and low means and the high and low proportions are markedly different from those for other regions ( P < 0.1). Second graders drinking optimally fluoridated water There are highly significant overall variations in means and proportions between regions (P < 0.0001). Prevalences of caries are highest in the children of the QAFF and lowest in those of the MRMOHF. Both means and proportions for these regions are meaningfully (and for most comparisons significantly) different from those of the other regions. When both means and proportions are considered, the children in the other regions tend to have medium prevalences of caries that are not significantly different from one another. Sixth graders drinking optimally fluoridated water There are significant overall variations in the means (P < 0.001, ANOVA) and proportions (P < 0.02, chi-squared). Prevalences of caries are highest in the children of the DGDGP and QAFF. In the DGDGP, all of the children were surveyed in one city. Water for this city comes from deep wells and is naturally fluoridated (1.35 ppm). Prevalences of caries for the children in the QAFF and DGDGP are significantly higher than the prevalences for children in the other regions. Significantly lower prevalences occur in the CARIES PREVALENCES IN MISSOURI children of the MRNOHF and MRMUGP. Based on information available through the Bureau of Dental Health for the State of Missouri, the cities that make up the MRMUGP sample are in a natural-fluoride belt that runs through the state. Prevalences of caries for the children of the GDGP, GDOHF, and COSEOHF are in the middle of the range. DISCUSSION Second graders drinking suboptimally fluoridated water In general, for this group, small numbers (N < 30) of children were surveyed in cities that are spread geographically far apart within regions. Although most geochemical variation occurs between regions, there is also variation within regions (Erdman et al., 1976; Feder, 19791, and we believe that this may be partially responsible for the results that represent a general lack of significant differences in caries scores between regions for these children. This does not mean that differences do not exist, but that given the low-resolution maps (available at this stage of our study) and the small sample sizes with wide geographical distribution (for this particular group of children), statistically significant differences cannot be demonstrated. The sample sizes are particularly small for the MRMUGP, MRMOHF, and MRNOHF (N < 34). In addition, two or more cities were surveyed in each of these regions (a list of the cities surveyed by caries scores is available from the authors upon request). Therefore, we tend to view the results concerning this group of children with caution. We can, however, make several observations concerning this group. The second grade children of the QAFF have high prevalences of caries, as found in the other three groups of children living in this region, all of whom obtain their water from shallow alluvial wells. The children of the MSOHF, who also have high caries prevalences, live in a city whose water source is a creek that flows through a mining area with known deposits of zinc and lead. The children of MRNGP (high caries prevalences) live in cities that get their water from the Missouri River. Therefore, it could be that geochemical factors associated with the waters of these rivers and shallow alluvial wells are affecting caries prevalences. By contrast, the children of the COSWOHF', who have one of the lowest proportions, live near 87 the children of the MSOHF. This may be explained by the fact that the cities in which the COSWOHF children live obtain their water from deep wells, which have low concentrations of geochemical factors. Both of the regions of the Cedar Glades (COSWCG and COSECG) have children with low to medium prevalences of caries. The children of the glaciated regions (GDGP and GDOHF) tend to have high prevalences of caries. In general, the highest prevalences are in the northwestern parts of these regions, and the lowest are in the eastern and southeastern parts; however, for the GDGP, children were surveyed in 16 cities, with less than 25 children being surveyed in each city. Thus suggested trends for this large geographic area are questionable, but these trends certainly warrant further survey with higher resolution maps. Sixth graders drinking suboptimally fluoridated water The highest prevalences of caries for the sixth grade children drinking suboptimally fluoridated water is in the MRNOHF. The means for the QAFF and MSOHF are also high, and the proportions are toward the middle of the range. We thus suggest that these children have high caries prevalences. The results of multiple t tests and chi-squared tests can be used to support this classification. The children of the MRMOHF also have high prevalences of caries, and those of the MRMUGP have low prevalences; however, we tend to view the results for these two groups with reservation because of the small sample sizes (N = 17 and 20) and because surveys were performed in two cities for each region. Nevertheless, it is intriguing that all of these cities are located close together, with one of the cities of the MRMUGP and one of the MRNOHF sharing the same water supply. The major difference between these regions is that the cities of the MRMOHF are located on the alluvial deposits of the Missouri river and the cities of the MRMUGP are located on unglaciated prairie. It thus seems as though the high-caries regions have cities that are located on the alluvial deposits of the Missouri and Mississippi Rivers or on glacial deposits (QAFF, MRNOHF, MRM0HF)-with the exception of the MSOHF. The sixth grade children of the MSOHF live in the same city that was surveyed for the second graders; thus they drink suboptimally fluoridated water that comes 88 C.F. HILDEBOLT ET AL. TABLE 5. Summary for children drinking optimally fluoridated water* Geochemical region Caries incidence QAFF High DGDGP High GDGP GDOHF COSEOHF MRNOHF MRMOHF MRMGP Middle Middle Middle Low Low Low Comments For both the second and sixth graders, the mean caries scores and proportion are high For the sixth graders, this region had the highest means and proportions. For second graders, the mean is medium, and the proportion is high These regions tend to have children with medium caries prevalence These regions tend to have children with low caries incidences *For abbreviations, see Table 1. from a river that flows through a mining area and that contains high levels of zinc and lead. The children of the MRNGP have medium caries prevalences and drink water obtained from the Missouri River. Within the GDOHF, the children have high caries prevalences in the cities of the northern part of the region, and in the cities of the southeastern part the children have low prevalences. Because the surveys covered 15 cities of the GDGP, the numbers of children surveyed in most cities were small, which makes it difficult to discern patterns within this region. The highest prevalences of caries, however, tend to be in the children living in the Northwest and the lowest in the children of the Southeast. The area of Missouri north of the Missouri river is covered with glacial loess, which is thought to have originated from the Missouri River drainage; therefore, for the northern part of the state, the deepest deposits of loess are in the Northwest and the thinnest in the Southeast (Ebens and Conner, 1980). In addition, when one considers that the deepest deposits of glacial till are in the northern part of the state (Ebens and Conner, 19801, it would seem that the caries prevalences within the GDOHF and GDGP vary directly with the depth of the glacial overburden. Nearly all of the public water supplies in these regions are from surface waters as opposed to deep wells. There is no clear pattern within the COSEOHF. One of the two Cedar Glades regions (COSWCG)contains children with low caries prevalences; the other (COSECG) contains children with medium caries prevalences. The children of the COSECG drink water obtained from a lake, whereas those of the COSWCG drink water from deep wells. The cities of the COSWOHF are close to the city of the COSWCG; both regions contain children with low caries prevalences. Table 4 is a list of the caries prevalences for the children drinking suboptimally fluoridated water and provides summary information concerning the regions. Second graders drinking optimally fluoridated water All of the children from the MRMOHF were from the same city, and, based on maps available through the Bureau of Dental Health for the state of Missouri, this city is within a belt of naturally fluoridated waters that runs through the state. The low caries prevalences of this region, especially compared with MRNOHF and QAFF samples, may be due to the children ingesting additional fluoride from multiple sources (Leverett, 1982). Another consideration is that this city gets its water from the Missouri River in contrast to the shallow alluvial wells of the QAFF and MRNOHF cities. One of the cities of the MRMUGP also gets its water from the Missouri River, and the children of this city likewise have low prevalences of caries. The children of the GDOHF and GDGP have medium prevalences of caries; however, there is a general trend for the children with high prevalences to be in the Northwest and those with low prevalences to be in the Northeast and Southeast. Thus it would appear that caries prevalences for these children may be associated with glacial factors. The children of the COSEOHF' have medium prevalences of caries, yet they are at the high end of this range. All of the cities that make up this group, at the high end of the range, are in the eastern part of the state. Two of CARIES PREVALENCES IN MISSOURI these cities are very near the QAFF and are in areas of alluvial deposits. Thus alluvial factors may be associated with the caries prevalences for the children of these two cities. The other two cities are located within the Flat River Lead Belt, where deposits of cobalt, copper, lead, nickel, and zinc are mined. One of these cities gets its water from the Flat River, which flows through this mining area, a condition similar to that which we described for the cities of the MSOHF region. The children of the DGDGP have a medium mean and a high proportion, despite fluoride ingestion. In the three cities surveyed in this region, the waters are naturally fluoridated with the following fluoride levels: 1.35, 2.03, and 2.04 ppm. We therefore suggest the possibility that geochemical factors in this region are interfering with the actions of fluoride. 89 Northwest have the highest prevalences, with the children in the East having lower prevalences and those in the Southeast the lowest. The children of the COSEOHF have a medium amount of caries; all of the cities, however, in which children were surveyed are from the eastern part of the state. Two of these are near the QAFF, and two are within the Flat River Lead Belt. Table 5 is a list of the caries prevalences for the children drinking optimally fluoridated water. Summary information concerning the regions is also given. Optimally versus suboptimally fluoridated water The second grade children of the QAFF and the COSEOHF' who are drinking optimally fluoridated water have higher caries prevalences than those children who are drinking suboptimally fluoridated water (Table 3). The cities that were surveyed for the Sixth graders drinking optimally second graders of the COSEOHF were all in fluoridated water the eastern part of the state either near or We further suggest that for sixth graders on the alluvial deposits of the Mississippi drinking optimally fluoridated water there River or in a mining area. The cities of the may be factors in the DGDGP and QAFF QAFF are all located on the alluvium of the that are interfering with the caries-protec- Mississippi River. We suggest that there are tive effects of fluoride. These factors do not factors in the alluvial deposits of the Mississeem to be affecting caries prevalences in the sippi embayment in the southeast corner of MRMOHF, MRNOHF, and MRMUGP. The the state of Missouri that may be interfering sample sizes for the MRNOHF and with the caries-beneficial effects of fluoride. MRMUGP are small (N < 32); thus we are The same factors may also be antagonizing cautious in reaching conclusions based on the actions of fluoride in the eastern part of the results from these two regions, although the COSEOHF' and in the DGDGP. we would have expected higher prevalences. We knew only whether a child was drinkIn addition, the children of the MRMOHF ing optimally or suboptimally fluoridated and MRMUGP live in cities that are within water. Knowledge of the precise levels of fluthe natural-fluoride belt. The same thing oride would be important if there were small could be said for the children of the DGDGP, differences in fluoride levels between groups yet they have high prevalences of caries. In drinking optimally and suboptimally fluorigeneral, for the GDGP, the children in the dated water. Moreover, the caries-protective TABLE 6. Summary of coronal caries patterns for prehistoric and living population* Geochemical reeion Prehistoric QAFF MRNOHF MRMOHF High High Medium PSOHF Low COSECG GDOHF Low Low *For abbreviations. see Table I Caries rate School Children High High None assigned (although possibly medium to high) Low (for MRMUGP, see test for discussion) Medium Low (see text for explanation) 90 C.F. HILDEBOLT ET AL. effect of fluoride is more pronounced on the interproximal surfaces of teeth than on the occlusal surfaces (McClure, 1970; Loesche, 1982; Nikiforuk, 1985). Because radiographs were not available to us, it is possible that we underestimated caries rates in children not receiving optimal levels of fluoride. It has also been suggested that because of intentional use of fluoride supplements, mouth rinses, and tooth pastes and the unintentional intake of fluoride from processed foods and vegetables that children drinking water that is not optimally fluoridated may actually be receiving optimal levels of fluoride and that this is a n important focus for research (Leverett, 1982). Data on fluorosis could be used as a means of determining whether children are likely receiving optimal or suboptimal levels of fluoride (no matter what its source; see Heifetz et al., 1988). Comparisons of caries prevalences for second graders versus sixth graders As shown in Table 2 and explained in Tables 4 and 5, there is general concordance between second and sixth graders with regard to high, medium, and low caries regions. We cannot, however, explain the apparent discordance in raw caries scores between the two grades. The higher scores for the second graders appear to be opposite those found in a n earlier survey of United States children done by the National Institute of Dental Research in which it was found that the older age groups had higher incidences of caries (Brunelle et al., 1982; Bibby 1982).In another recent survey, however, the caries scores for the children of the Isle of Lewis, United Kingdom, were similar to Missouri school children in that the younger children had higher scores (Hargreaves et al., 1987). Caries has a complex etiology in which there is a n interplay between diet, genes, bacteria, and other agents (such as geochemical factors). In our study, we had data only for caries scores and geochemical factors. We cannot explain the discordance in raw caries scores between second and sixth graders. Temporal comparisons This study was undertaken to test the findings of a survey of dental disease patterns of prehistoric inhabitants of Missouri (Hildebolt et al., 1988). Table 6 contains a summary of those findings for coronal caries. It also is a comparison of coronal caries patterns for prehistoric and living populations of Missouri. Both extinct and extant populations of the QAFF and MRNOHF had high rates of coronal caries. The comparison of coronal caries patterns for the MRMOHF is more problematic because the sample sizes are small. We classified the prehistoric populations of the MRMOHF as having a medium rate of coronal caries. We did not classify the living populations, because the sixth graders had a high prevalence of caries whereas the second graders had a low prevalence (see Table 4). The sixth graders may be more comparable with the prehistoric populations in that they have their permanent dentition as opposed to the second graders who have both primary and secondary teeth. With this in mind, we classified the children of the MRMOHF as having a medium to high rate of coronal caries. The prehistoric populations of the PSOHF had low rates of coronal caries, and the school children had medium prevalences. The school children got their water from deep wells, whereas the early inhabitants used surface water. These waters are very different geochemically. Perhaps the best comparison of caries rates for extinct inhabitants of this region is with the rates for children living in the MRMUGP. This is because the children of the MRMUGP are drinking surface water and are also located in a n area that is geologically Pennsylvania stata. The comparison is not perfect but is the best that we can make until additional data are collected. The children of the MRMUGP have low caries rates, the same as prehistoric populations. For school children living in the COSECG there is a spread in caries rates. We classified these children as having a medium prevalence of caries (Table 4).The prehistoric populations had a low rate of coronal caries. We classified the children of the GDOHF as having a medium prevalence of caries (Table 4), yet, throughout our discussion, we point out that the prevalences of caries were low for children living in the southeastern part of this region. If one considers only the children of this region who lived near the prehistoric populations, their rate of coronal caries is one of the lowest for the state, and it compares well with the low rate of caries for the prehistoric populations. Thus it would seem that there is a n overall concordance between regional, coronal caries patterns for the prehistoric and contempory populations of this study. CARIES PREVALENCES IN MISSOURI The next phase of this work is a n expanded study in which larger numbers of children will be surveyed in selected regions for which higher resolution maps will be made. Data will be collected on fluorosis, caries prevalences by tooth type, specific fluoride levels in water supplies, and socioeconomic and ethnic backgrounds (factors for which we had no information in our present study). When feasible, we will also use dental radiographs. As part of our planned study, preliminary determinations will be made with regard to which specific geochemical factors or combinations of factors are affecting dental health. Geochemical data for these analyses have been made available to us by the USGS. CONCLUSIONS There are two basic objectives in epidemiology. The first is to determine the distribution and occurrence of the disease in question. After this has been accomplished, the second objective is pursued. This objective is the identification of the disease determinants. Our study was done to test the hypothesis that coronal caries rates vary among geochemically defined regions of the State of Missouri. This hypothesis was originally tested with prehistoric inhabitants of the state (Hildebolt et al., 1988). We have thus taken two initial steps in determining the distribution and occurrence of coronal caries with regard to geochemically defined regions. We suggest that caries rates do vary among the geochemical regions of the state of Missouri and that for Missouri geochemical factors associated with new parent materials, specifically glacial overburden, may be directly related to caries rates as may be geochemical factors associated with several mining areas. In addition, we believe that geochemical factors associated with the alluvial deposits of the Mississippi River in the southeast corner of the state may be interfering with the caries-protective effects of fluoride. In conclusion, the highly significant overall variations in mean coronal caries scores between regions ( P < 0.001) for three of the four groups of children surveyed is very encouraging. Moreover, although spurious associations caused by ecological fallacy and secondary association cannot be ruled out, we believe our approach affords better control than is possible by using previous geochemical-dental-epidemiological methods. 91 ACKNOWLEDGMENTS We gratefully acknowledge the cooperation of the Bureau of Dental Health, Missouri Division of Health, and the United States Geological Survey, with special thanks to Paul Reid and Roy Mendes. This research was supported in part by a n NIH Biomedical Research Support Grant. LITERATURE CITED Barmes DE, Adkins BL, and Schamschula RG (1970) Etiology of caries in Papua, New Guinea: Associations in soil, food, and water. Bull. WHO 43:769-784. Bibby BG (1982)Changing perspectives on dental caries. In E. Storey (ed.): Diet and Dental Caries: Changing Perspectives. Proceeding of a Symposium held May 89, 1982. Melbourne, Australia: University of Melbourne, pp. 1-17. Blalock HM, Jr (1979) Social Statistics. New York: McGraw-Hill, pp. 158. Bowen WH, Velez H, Aguila M, Velasquez H, Sierra LI, and Gillespie G. (1977)The microbiology and biochemistry of plaque, saliva, and drinking water from two communities with contrasting levels of caries in Colombia, S.A. J. Dent. Res. 56:C32-C39. Brunelle JA, Millar AJ, and Carlos JP (1982) Changes in dental caries prevalences in U.S. children. J. Dent. Res. (Abstr. 416) 61:225. Cannon H (1978) Rocks: The Geologic Source of Most Trace Elements. Geochemistry and the Environment. Volume 111, Distribution of Trace Elements Related to the Occurrence of Certain Cancers, Cardiovascular Diseases, and Urolithiasis. Washington, DC: National Academy of Sciences. Curzon MEJ, Cutress TW (1983) Trace Elements and Dental Disease. Boston: Wright. Ebens RJ, and Connor JJ (1980) Geochemical Survey of Missouri-Geochemistry of Loess and Carbonate Residium. United States Geological Survey Professional Paper 954-G. Washington, DC: United States Government Printing Office. Erdman JA, Schacklette T,and Keith JR (1976) Elemental Composition of Corn Grains, Soybean Seeds, Pasture Grasses, and Associated Soils From Selected Areas of Missouri. United States Geological Survey Professional Paper 954-d. Washington, DC: United States Government Printing Office. Featherstone JDB (1983) Magnesium. In MEJ Curzon and TW Cutress (eds.): Trace Elements and Dental Disease. Boston: Wright, pp. 221-235. Feder GL (1979) Geochemical Survey of Missouri-Geochemical Survey of Waters of Missouri. United States Geological Survey Professional Paper 954-E. Washington, DC: United States Government Printing Office. Glass RL, Rothman KJ, Espinal F, Velez H, and Smith N J (1973) The prevalence of human dental cries and water-borne trace elements. Arch. Oral. Biol. 18:10991104. Hargreaves JA, Wagg BJ, and Tompson GW (1987) Changes in caries prevalences of Isle of Lewis children: A historical comparison from 1937 to 1984. Caries Res. 21:277-284. Heifetz SB, Driscoll WS, Horowitz HS, and Kingman A (1988) Prevalence of dental caries and dental fluorosis in areas with optimal and above-optimal water-fluoride concentrations: A 5 year study. J. Am. Dent. Assoc. 116:490-495. 92 C.F. HILDEBOLT ET AL. Helle A, and Haavikko K (1977) The concentrations of nine macro- and microminerals in drinking water examined in relation to caries prevalence. Proc. Finn. Dent. SOC.73t76-86. Hildebolt CF, Molnar S, Elvin-Lewis M, and McKee J (1988) The effect of geochemical factors on prevalences of dental diseases for prehistoric inhabitants of the State of Missouri. Am. J. Phys. Anthropol. 75:l-14. Hopps HC, and O’Dell BL (1981) Introduction and conclusions to workshop on “Research Needed to Improve Data on the Mineral Content of Human Tissues.” Fed. Proc. 40t2112-2114. Leverett DH (1982) Fluorides and the changing prevalence of dental caries. Science 217t26-30. Loeshe WJ (1982) Dental Caries: A Treatable Infection. Springfield, IL: Charles C. Thomas. McClure FJ (1970) Water Fluoridation: The Search and the Victory. United States Department of Health, Education, and Welfare. Bethesda, MD: National Institue of Dental Research. Miesch AT, Barnett AJ, Bartel JI, Dinnin GL, and Feder GL (1976) Geochemical Survey of Missouri-Methods of Sampling, Laboratory Analysis, and Statistical Reduction of Data. United States Geological Survey Professional Paper 954-A. Washington, DC: United States Government Printing Office. Nikiforuk G (1985) Understanding Dental Caries. New York: Karger. SAS Institute Inc. (1985a) SAS User’s Guide: Statistics, ed. 5. Gary, SAS Institute, Inc. SAS Institute Inc. (198513) SAS/GRAPH User’s Guide, ed. 5. Gary, NC: SAS Institute, Inc. Schamschula RG, Adkins BL, Barmes DE, Charlton G, and Davey B (1978) WHO Study of Dental Caries Aetiology in Papua New Guinea. WHO Publication No. 40. Geneva, Switzerland World Health Organization. Sokal RR, Rohlf FJ (1981)Biometry: The Principles and Practice of Statistics in Biological Research, ed. 2. New York: Freeman.