A Potential Model for the Interaction of Enamel Fluoride and Plaque in the Development of Dental Caries FRED L. BOOKSTEIN and PAUL F. DEPAOLA
Forsyth Dental Center, Boston, Massachusetts 02115, USA
We have previously reported an epidemiological study of the relationship between caries experience and surface enamel fluoride in man.1 In this work, fluoride was measured as loge mass F corrected to a standardized depth, and caries experience was expressed as V\/DMFS. The rationale for the logarithmic and square root transformations has been described in detail.1 2 We found a simple linear relationship between log0 mass F and \/DMFS. The multiple correlation, taking age into account, was R - 0.40. In this same investigation, the dental debris of the subjects was measured using the Greene and Vermillion Simplified Oral Hygiene Index.3 The present study represents a continuation of the work in which an attempt is made to bring plaque into the model via a regression of caries experience on fluoride and plaque
nonfluoridated areas were assessed for caries, enamel fluoride, and dental debris. The subjects were examined for dental caries utilizing the conventional DMFS index. Enamel surface fluoride was collected and chemically analyzed as described by Brudevold,4 utilizing a standardized sampling area of 20.1 square millimeters.5 An adjustment procedure was applied to the biopsy data to compensate for uncontrolled differences in sample weight.2 This procedure permitted the adjustment of 1,447 of the 1,829 F assay values. Not all of the values could be corrected since the adjustment techniques were niot applicable to subjects whose biopsies were unusually heavy or light, and not all subjects examined for caries had a suitable surface for the biopsy; however, the age distribution by community was not seriously disturbed by this diminution in sample size. The relation between dental caries and enamel fluoride was determined by regression analysis. At this point, an attempt was made to incorporate plaque into the model. As in the case of the DMF survey, the plaque assessment was made by one examiner in Boston, Danvers, and part of Charlotte. A
simultaneously. Materials and Methods The methods of the original epidemiological investigation have been described, and will be summarized briefly. Eighteen hundred and twenty-nine 12- to 16-year-old life-time residents in selected fluoridated and
BLE 1 COMMUNITIES AND SUBJECrS Community
Boston, Massachusetts Danvers, Massachusetts Charlotte, North Carolina Charlotte, North Carolina Kalamazoo, Michigan Stickney, Illinois Midland, Texas Total
*reduced
F status of water supply
No. of subjects
0.06 ppm fluoridated 0.08 ppm fluoridated fluoridated fluoridated
289
5-7 ppm*
301 118 208 305 304 304 1829
Mean age in years
13.34 13.52 13.34 13.43 13.45 13.38 13.41 13.42
(1.04) t (1.13) (0.99) (1.06) (1.06) (1.07) (1.06) (1.06)
to 0.32 ppm 10 months prior to study. are standard deviations
tfigures in parentheses
C49
BOOKSTEIN & DE PAOLA
j Dent Res Special Issue C
Downloaded from jdr.sagepub.com at PENNSYLVANIA STATE UNIV on May 27, 2015 For personal use only. No other uses without permission.
ENAMEL FLUORIDE, PLAQUE & CARIES
Vol. 56 1977
MEAN iDMFS BY
C41
F BY TOTAL PLAQUE
LOGe
3 PLAQUE CATEGORY 0
MEAN \JDMFS
2
I
11.7512.00
-12.50
-13.00
13.50
2-3
2-6
4-6
7+
LOGe FLUORIDE
FIG 1.-Mean
V/DMFS by Loge F by plaque.
different examiner worked in all other communities. Independent examinations of subgroups indicated good interexaminer agreement with respect to the caries evaluation, so that the DMF data of the two investigators could be pooled. However, inspection of the OHI data revealed a marked difference in the recording level of the two examiners. A variety of transformations failed to
reconcile the findings. Eventually, it was necessary to eliminate the Boston/Danvers/ Charlotte data of the one examiner, reducing the sample size from 1,447 to 1,088. The attempt to incorporate plaque into the model required a series of exploratory analyses in order to obtain a successful outcome.
Downloaded from jdr.sagepub.com at PENNSYLVANIA STATE UNIV on May 27, 2015 For personal use only. No other uses without permission.
C42
BOOKSTEIN & DE PAOLA
j Dent Res Special Issue C
Results The communities, the fluoride status of their water supplies, the initial number of subjects in each group, and the mean ages
category the highest single debris score assigned to each subject in place of the conventional aggregate OHI score. The rationale for this strategy is given in the Discussion. This version of the plaque index successfully ameliorates most of the blurring evident at the intermediate levels of the previous figure. There is clearly apparent a crest in the fluoride/caries curve which moves consistently along the fluoride scale as the plaque score increases. The significance of this pattern was tested by postulating the model DMFS = Fg * loge + error (before the crest) DMFS = (Fg + Fer) * loge F + error (beyond the crest) with errors independent, identically normally distributed. Fg designates the general effect of fluoride level, and Fcr, the effect of location at or beyond the crest. This model can be estimated using a multiple regression of V/DMFS on logeF together with a new term, the "slope-change
are shown in Table 1. The first analyses involving plaque revealed that, in linear form, the conventional Simplified OHI did not aid in predicting caries experience, suggesting either no relationship between plaque and DMF, or a relationship of a nonlinear nature; that is, a relationship which changes its form as a function of plaque category. An exploration of the caries-fluoride dependency as a function of plaque category, utilizing various cutting points applied to the fluoride scale, led eventually to the situation depicted in Figure 1. For each level of plaque, there appears to be a "crest" level of fluoride. (Note that the class intervals for fluoride have been systematically varied in order to highlight the effect.) The crest is nearly a monotonic function of plaque except in the case of plaque category 2-3 and possibly plaque category 2-6. In Figure 2, the analysis has been repeated, utilizing for plaque MEAN \DMFS BY
LOG.
FLUORIDE BY HIGHEST PLAQUE SCORE
PLAQUE
2.5
I
12 5°
2,5
-
2.5 ,
2.5
1.5
\the "crest" i 12.12512.375 12.0012.25
-12.50
i -12.625
i
i -12.875 -12.75
i
-13.375
-13.125 -13.00
-13.25
FIG 2.-Mean V/DMFS by Loge F by highest plaque
-13.50
score.
Downloaded from jdr.sagepub.com at PENNSYLVANIA STATE UNIV on May 27, 2015 For personal use only. No other uses without permission.
ENAMEL FLUORIDE, PLAQUE & CARIES
Vol. 56 1977
C43
TABLE 2 VARIANCE ANALYSIS STEMMING FROM POSTULATED MODEL INVOLVING PLAQUE
Anova Summary
Variable c loge F
Slope-change var. Regression constant 12.13
Coefficient -0.82
-0.31
t-test
-8.77 -2.23
df 1085
1085
Significance
Forsyth Dental Center, Boston, Massachusetts 02115, USA
We have previously reported an epidemiological study of the relationship between caries experience and surface enamel fluoride in man.1 In this work, fluoride was measured as loge mass F corrected to a standardized depth, and caries experience was expressed as V\/DMFS. The rationale for the logarithmic and square root transformations has been described in detail.1 2 We found a simple linear relationship between log0 mass F and \/DMFS. The multiple correlation, taking age into account, was R - 0.40. In this same investigation, the dental debris of the subjects was measured using the Greene and Vermillion Simplified Oral Hygiene Index.3 The present study represents a continuation of the work in which an attempt is made to bring plaque into the model via a regression of caries experience on fluoride and plaque
nonfluoridated areas were assessed for caries, enamel fluoride, and dental debris. The subjects were examined for dental caries utilizing the conventional DMFS index. Enamel surface fluoride was collected and chemically analyzed as described by Brudevold,4 utilizing a standardized sampling area of 20.1 square millimeters.5 An adjustment procedure was applied to the biopsy data to compensate for uncontrolled differences in sample weight.2 This procedure permitted the adjustment of 1,447 of the 1,829 F assay values. Not all of the values could be corrected since the adjustment techniques were niot applicable to subjects whose biopsies were unusually heavy or light, and not all subjects examined for caries had a suitable surface for the biopsy; however, the age distribution by community was not seriously disturbed by this diminution in sample size. The relation between dental caries and enamel fluoride was determined by regression analysis. At this point, an attempt was made to incorporate plaque into the model. As in the case of the DMF survey, the plaque assessment was made by one examiner in Boston, Danvers, and part of Charlotte. A
simultaneously. Materials and Methods The methods of the original epidemiological investigation have been described, and will be summarized briefly. Eighteen hundred and twenty-nine 12- to 16-year-old life-time residents in selected fluoridated and
BLE 1 COMMUNITIES AND SUBJECrS Community
Boston, Massachusetts Danvers, Massachusetts Charlotte, North Carolina Charlotte, North Carolina Kalamazoo, Michigan Stickney, Illinois Midland, Texas Total
*reduced
F status of water supply
No. of subjects
0.06 ppm fluoridated 0.08 ppm fluoridated fluoridated fluoridated
289
5-7 ppm*
301 118 208 305 304 304 1829
Mean age in years
13.34 13.52 13.34 13.43 13.45 13.38 13.41 13.42
(1.04) t (1.13) (0.99) (1.06) (1.06) (1.07) (1.06) (1.06)
to 0.32 ppm 10 months prior to study. are standard deviations
tfigures in parentheses
C49
BOOKSTEIN & DE PAOLA
j Dent Res Special Issue C
Downloaded from jdr.sagepub.com at PENNSYLVANIA STATE UNIV on May 27, 2015 For personal use only. No other uses without permission.
ENAMEL FLUORIDE, PLAQUE & CARIES
Vol. 56 1977
MEAN iDMFS BY
C41
F BY TOTAL PLAQUE
LOGe
3 PLAQUE CATEGORY 0
MEAN \JDMFS
2
I
11.7512.00
-12.50
-13.00
13.50
2-3
2-6
4-6
7+
LOGe FLUORIDE
FIG 1.-Mean
V/DMFS by Loge F by plaque.
different examiner worked in all other communities. Independent examinations of subgroups indicated good interexaminer agreement with respect to the caries evaluation, so that the DMF data of the two investigators could be pooled. However, inspection of the OHI data revealed a marked difference in the recording level of the two examiners. A variety of transformations failed to
reconcile the findings. Eventually, it was necessary to eliminate the Boston/Danvers/ Charlotte data of the one examiner, reducing the sample size from 1,447 to 1,088. The attempt to incorporate plaque into the model required a series of exploratory analyses in order to obtain a successful outcome.
Downloaded from jdr.sagepub.com at PENNSYLVANIA STATE UNIV on May 27, 2015 For personal use only. No other uses without permission.
C42
BOOKSTEIN & DE PAOLA
j Dent Res Special Issue C
Results The communities, the fluoride status of their water supplies, the initial number of subjects in each group, and the mean ages
category the highest single debris score assigned to each subject in place of the conventional aggregate OHI score. The rationale for this strategy is given in the Discussion. This version of the plaque index successfully ameliorates most of the blurring evident at the intermediate levels of the previous figure. There is clearly apparent a crest in the fluoride/caries curve which moves consistently along the fluoride scale as the plaque score increases. The significance of this pattern was tested by postulating the model DMFS = Fg * loge + error (before the crest) DMFS = (Fg + Fer) * loge F + error (beyond the crest) with errors independent, identically normally distributed. Fg designates the general effect of fluoride level, and Fcr, the effect of location at or beyond the crest. This model can be estimated using a multiple regression of V/DMFS on logeF together with a new term, the "slope-change
are shown in Table 1. The first analyses involving plaque revealed that, in linear form, the conventional Simplified OHI did not aid in predicting caries experience, suggesting either no relationship between plaque and DMF, or a relationship of a nonlinear nature; that is, a relationship which changes its form as a function of plaque category. An exploration of the caries-fluoride dependency as a function of plaque category, utilizing various cutting points applied to the fluoride scale, led eventually to the situation depicted in Figure 1. For each level of plaque, there appears to be a "crest" level of fluoride. (Note that the class intervals for fluoride have been systematically varied in order to highlight the effect.) The crest is nearly a monotonic function of plaque except in the case of plaque category 2-3 and possibly plaque category 2-6. In Figure 2, the analysis has been repeated, utilizing for plaque MEAN \DMFS BY
LOG.
FLUORIDE BY HIGHEST PLAQUE SCORE
PLAQUE
2.5
I
12 5°
2,5
-
2.5 ,
2.5
1.5
\the "crest" i 12.12512.375 12.0012.25
-12.50
i -12.625
i
i -12.875 -12.75
i
-13.375
-13.125 -13.00
-13.25
FIG 2.-Mean V/DMFS by Loge F by highest plaque
-13.50
score.
Downloaded from jdr.sagepub.com at PENNSYLVANIA STATE UNIV on May 27, 2015 For personal use only. No other uses without permission.
ENAMEL FLUORIDE, PLAQUE & CARIES
Vol. 56 1977
C43
TABLE 2 VARIANCE ANALYSIS STEMMING FROM POSTULATED MODEL INVOLVING PLAQUE
Anova Summary
Variable c loge F
Slope-change var. Regression constant 12.13
Coefficient -0.82
-0.31
t-test
-8.77 -2.23
df 1085
1085
Significance
