Академический Документы
Профессиональный Документы
Культура Документы
Introduction.
Most epidemiological caries studies employ criteria in which
a tooth, or tooth surface, is only recorded as decayed when
cavitation is obvious, and which ignore all caries present at a
less severe level (Klein et al., 1938; Todd et al., 1982; Downer
and Teagle, 1979). In 1981, the Federation Dentaire Internationale (FDI) adopted a number of "global goals for oral health"
which proposed that by the year 2000 the global average for
Decayed, Missing, and Filled teeth (DMFT) in 12-year-olds
should be no more than 3 (FDI, 1982). Here, clinical caries
was to be diagnosed at the level at which clinical intervention
was required, which is commonly taken to mean when clinical
cavitation has occurred. The workshop on Epidemiology of
Dental Caries set up by the British Association for the Study
of Community Dentistry in 1982 also recommended one, relatively gross, level of diagnosis: "caries at the cavitation level,
i.e., involving dentine", for prevalence studies, since this would
conform to the criteria in current use and, they hoped, allow
for a higher degree of reproducibility between and among examiners (Palmer et al., 1984). Few studies have considered
the extent to which inclusion or exclusion of pre-cavitation
lesions ["initial" and "enamel" lesions, to use the World
Health Organization terminology (WHO, 1979)] may affect
DMF counts, particularly after the recent decline in caries
prevalence.
In cases where the presence of "early" lesions as well as
cavities has been scored and reported, it has been shown that
large differences may occur between DMFT calculated incluReceived for publication July 28, 1987
Accepted for publication November 19, 1987
Financial support for this study came from the HKU Dental Faculty
Research Committee and latterly from the Chief Scientist Organisation
of the Scottish Home and Health Department.
592
ding early lesions and the DMFT calculated when the criteria
for recognition of D teeth are limited to cavitation (Marthaler,
1963). These differences, if not considered, can hide or confuse important findings. In a fluoridation study, for example,
there appeared to be a large disparity between the total number
of caries lesions recorded in test and control groups when only
relatively gross lesions were recognized. When, however, all
enamel lesions were included in the calculations, the total numbers of lesions in both the test and control groups at the end
of the experiment were shown to have been similar (Groeneveld, 1985). This has led to a re-examination of previous concepts of the mode of action of fluoride in caries prevention.
Rather than completely preventing the formation of a number
of lesions in the test group, it would seem that fluoride acted
by preventing the progression of many of the pre-cavitation
lesions to a more severe state (Groeneveld, 1985).
A recent preliminary report of the caries status of a group
of Hong Kong University dental undergraduates (Pitts, 1983)
found a "low" caries prevalence. A mean DMFT of 2.8 was
found when the conventional diagnostic threshold (D3), which
excludes all pre-cavitation lesions, was employed. This group
of students can therefore be considered to have already exceeded the FDI goal for the year 2000. In view of the muchdocumented recent decline in caries prevalence in many countries throughout the world (Glass, 1982; Allan et al., 1983;
Renson et al., 1985), this type of "low caries" data may
become more common in future caries studies in many developed countries. The principal aim of this study was to investigate the impact of the inclusion (or, conversely, exclusion)
of "initial" and "enamel" lesions on the results of a clinical
caries examination of a low-caries-prevalence group. Secondary aims of this study were to assess the feasibility of doing
this using a microcomputer running the CARIES software
package which re-calculates the raw DMF data according to
three different diagnostic thresholds, and to examine the extent
to which the percentage of a group identified as "caries-free"
changed when the sensitivity of the diagnostic threshold employed was increased.
Vol. 67 No. 3
593
Results.
Mean DMF.-The average time for an examination (all of
which were carried out at the D1 threshold) to be carried out
was approximately three minutes, with a range of 2-5 minutes.
Table 3 shows the mean decayed, missing, and filled teeth and
surface counts from the same data (for all of the 287 students)
calculated at each of the three diagnostic thresholds. For both
DMFT and DMFS, it is evident that, as the sensitivity of the
diagnostic threshold was increased from D3 (least sensitive) to
D1 (most sensitive), the value of the calculated index almost
doubled.
When the separate components of the indices are considered,
it becomes obvious that, as expected, the decay component is
the one most affected by the imposed thresholds. An increased
sensitivity of diagnostic threshold had no effect on the M component, which remained at 0.29 missing teeth (or 0.86 missing
surfaces) per person. Similarly, filled surfaces remained constant at 2.97 per person regardless of threshold. There was a
TABLE 2
594
Discussion.
Microcomputers with the appropriate software enjoy a number of features which are theoretically desirable for the hanTABLE 4
PERCENTAGES OF A GROUP OF 287 HONG KONG DENTAL
STUDENTS CONSIDERED CURRENTLY FREE OF DECAY
(DECAYED TEETH = 0), OR FREE OF DECAY EXPERIENCE
(DMF TEETH = 0), AT EACH OF THE THREE DIAGNOSTIC
THRESHOLDS
DMF Teeth = 0
Diagnostic Threshold
Dacayed Teeth = 0
D3
61.0%
28.2%
D2
33.5%
15.0%
DI
14.3%
7.0%
TABLE 5
OVERALL INTRA-EXAMINER REPRODUCIBILITY FOR ALL
THREE EXAMINERS, AT THE THREE DIAGNOSTIC
THRESHOLDS, FOR A RANDOMLY SELECTED 10% OF
SUBJECrS
Diagnostic Threshold
Reproducibility Index
Coincidence Index (sound to sound)
Coincidence Index (decay to decay)
Kappa
Modified percentage reproducibility
Correlation
coefficient
D3
0.999
0.841
0.789
99.7%
0.98
D2
0.996
0.812
0.818
99.2%
0.94
D1
0.992
0.809
0.795
98.4%
0.92
Vol. 67 No. 3
There are many suggested methods of calculating reproducibility (Shaw and Murray, 1975), the choice of the most appropriate method(s) being to an extent governed by survey/
study design and objectives, as well as by personal preference.
While the use of Pearson's correlation coefficient is well-established, it was considered that the Kappa statistic and Dice's
coincidence index also have a number of advantages (Nuttall
and Paul, 1984). Many methods give few guidelines for interpretation of values. Landis and Koch (1977), however, proposed a scale for interpreting Kappa results. At each of the
three diagnostic thresholds employed here, the mean Kappa
values lie at the top of the "substantial agreement" range.
When the individual examiners are considered, it is evident
that examiner A's kappa values were at the borderline between
"moderate" and "substantial" agreement (with the D1 agreement being higher), while examiners B and C's values were
either at the borderline of, or within, the "almost perfect"
agreement range. The important point to be made, however,
is that although the individual examiners showed differences
in intra-examiner reproducibility according to the index calculated, little difference was found in their ability to diagnose
with a similar level of reproducibility at each diagnostic threshold. Fears of loss of intra-examiner reproducibility when the
more sensitive thresholds are used might, therefore, have been
overstated. Further studies of inter-examiner reproducibility
with low-prevalence groups are, however, still required.
Although the lower caries levels encountered in this group
may have played a part in the reproducibility achieved, it might
be postulated that the presence of fewer, less severe lesions
would result in a greater number of more difficult, borderline
decisions than at times of higher caries activity. There was a
slight deterioration in reproducibility from D3 to D1, but the
magnitude of this, as demonstrated in the small changes in the
various reproducibility indices and the relatively small increase
in the standard error of the mean of the DMF values (Table
3), is, we would submit, insufficient to warrant exclusion of
pre-cavitation lesions from DMF counts in all situations. It is
possible that the pre-examination self-cleaning and good examining conditions (dental light and availability of compressed
air) contributed to the reproducibility observed. These conditions can be provided quite readily -even, where necessary,
with portable equipment.
Although it is obvious that increasing the sensitivity of any
diagnostic threshold will give a different interpretation to the
results of an examination, the aim of this study was to assess
the magnitude of this effect when applied to epidemiological
caries data. The results show that the use of such thresholds
can have a dramatic effect on the reported level of dental caries. DMFT and DMFS indices were almost doubled by inclusion of "enamel" and "initial" lesions, while the "cariesfree" percentage was decreased to approximately one-quarter
of its D3 value by inclusion of these lesions (Tables 3 and 4).
In view of these findings, it may be necessary for the question
of choosing a diagnostic threshold which is appropriate to the
survey objective(s) to be re-examined and to receive greater
emphasis in future studies. There is a chance that the apparent
underestimation of disease prevalence which arises when insensitive thresholds are used may be misinterpreted by research
workers, dentists, and health planners.
From the results of this study, it would seem that it is feasible and potentially advantageous to use a microcomputer with
software suitable for storage of raw epidemiological caries data
recorded at the D1 threshold to calculate DMFT and DMFS at
differing diagnostic thresholds. It is evident that the use of
criteria which might be misinterpreted as being similar, when
in fact they use differing effective diagnostic thresholds, can
dramatically influence the reported level of dental caries, and
595
Acknowledgments.
We are indebted to the dental students of Hong Kong University (HKU) for agreeing to participate, to all the various
members of the Faculty of Dentistry, HKU, for their cooperation in mounting the caries examinations during the busy Induction and Orientation weeks, and in particular to Dr. C.
Taylor and Dr. W. Yung (formerly HKU) for acting as examiners. We are also grateful to Mr. D. Carthy, Head of HKU's
Dental Data Processing Unit (DDPU), and to Mr. L. Pong
(DDPU) for the assistance in producing the CARIES package.
REFERENCES
ALLAN, C.D.; ASHLEY, F.P.; and NAYLOR, M.N.(1983): Caries
Experience in 11-year-old Schoolgirls between 1962 and 1981, Br
Dent J 154:167-170.
BACKER-DIRKS, 0. (1964): At What Stage Should a Carious Lesion be Recorded?. In: Advances in Fluorine Research and Dental Caries Prevention, Vol. 2, J.L. Hardwick, J.P. Dustin, and
H.R. Held, Eds., Oxford: Pergamon Press, pp. 21-23.
COHEN, J. (1960): Coefficient of Agreement for Nominal Scales,
Educ Psychol Measurement 20:37-46.
DICE, L.R. (1945): Measures of the Amount of Ecologic Association
between Species, Ecology 26:297-302.
DOWNER, M.C. and TEAGLE, F.A. (1979): SPEED-System for
Planning and Epidemiological Evaluation of Dental Services.
ESPELID, I. and TVEIT, A.B. (1986): Diagnostic Quality and Observer Variation in Radiographic Diagnoses of Approximal Caries,
Acta Odontol Scand 44:39-46.
FEDERATION DENTAIRE INTERNATIONALE (1982): Global
Goals for Oral Health by the Year 2000, Int Dent J 32:74-77.
GLASS, R.L. (1982): Conference on Declining Prevalence of Dental
Caries, J Dent Res 61 (Spec Iss): 1304.
GLASS, R.L.; PETERSON, J.K.; and BIXLER, D. (1983): The
Effects of Changing Caries Prevalence and Diagnostic Criteria on
Clinical Trials, Caries Res 17:145-151.
GORDON, M. and NEWBRUN, E. (1986): Comparison of Trends
in the Prevalence of Caries and Restorations in Young Adult Populations of Several Countries, Community Dent Oral Epidemiol
14:104-109.
GROENEVELD, A. (1985): Longitudinal Study of Prevalence of
Enamel Lesions in a Fluoridated and Non-fluoridated Area, Community Dent Oral Epidemiol 13:159-163.
HAUGEJORDEN, 0. and SLACK, G.L. (1975): A Study of IntraExaminer Error Associated with Recording of Radiographic Caries
at Different Diagnostic Levels, Acta Odontol Scand 33:169-181.
HOROWITZ, H.S. (1972): Clinical Trials of Preventives for Dental
Caries, J Public Health Dent 32:229-233.
KLEIN, H.; PALMER, C.E.; and KNUTSON, J.W. (1938): Studies
on Dental Caries. I. Dental Status and Dental Needs of Elementary
Schoolchildren, Public Health Rep 53:751-756.
LANDIS, J.R. and KOCH, G.G. (1977): The Measurement of Observer Agreement for Categorical Data, Biometrics 33:159-174.
MARTHALER, T.M. (1963): Experimental Design in Epidemiologic
Research, J Dent Res 42:192-201.
NUT-FALL, N.M. and PAUL, W.McK. (1985): The Analysis of Inter-Dentist Agreement in Caries Prevalence Studies, Community
Dent Health 2:123-128.
PALMER, J.D.; ANDERSON, R.J.; and DOWNER, M.C. (1984):
596
SHAW, L. and MURRAY, J.J. (1975): Inter-Examiner and IntraExaminer Reproducibility in Clinical and Radiographic Diagnosis,
Int Dent J 25:280-287.
TODD, J.E. and DODD, P. (1985): Children's Dental Health in
the United Kingdom 1983. London: HMSO.
TODD, J.E.; WALKER, A.M.; and DODD, P. (1982): Adult Dental
Health, Vol. 2, United Kingdom 1978. London: HMSO.
WORLD HEALTH ORGANIZATION (1977): Oral Health Surveys,
Basic Methods (2nd ed.). Geneva: WHO.
WORLD HEALTH ORGANIZATION (1979): A Guide to Oral Health
Epidemiological Investigations. Geneva: WHO.