Professional Documents
Culture Documents
1989 The Development of An Index of Orthodontic Treatment Priority
1989 The Development of An Index of Orthodontic Treatment Priority
SUMMARY The aim of this study was to develop a valid and reproducible index of orthodontic
treatment priority.
After reviewing the available literature, it was felt that this could be best achieved by using two
separate components to record firstly the dental health and functional indications for treatment,
and secondly the aesthetic impairment caused by the malocclusion.
A modification of the index used by the Swedish Dental Health Board was used to record the
need for orthodontic treatment on dental health and functional grounds. This index was modified
by defining five grades, with precise dividing lines between each grade.
An illustrated 10-point scale was used to assess independently the aesthetic treatment need of
the patients. This scale was constructed using dental photographs of 12-year-olds collected
during a large multi-disciplinary survey. Six non-dental judges rated these photographs on a
visual analogue scale, and at equal intervals along the judged range, representative photographs
were chosen.
To test the index in use, two sample populations were defined; a group of patients referred for
treatment, and a random sample of 11 -12-year-old schoolchildren. Both samples were examined
using the index and satisfactory levels of intra- and inter-examiner agreement were obtained.
study models could be used in place of the out of 118 re-tests and in all cases the disagree-
recordings taken at the school visits. In practice, ment was only by one grade. Guidelines for the
as many of the children were only just commenc- interpretation of the Kappa statistic (Landis and
ing treatment, the gradings were little changed. Koch, 1977) are shown in Table 2.
Inter-examiner agreement ranged from 0.731-
0.797. In total there were 21 out of 154 measure-
Results
ments that were not agreed. There were only 2
Reproducibility of the index cases where the error was by more than one
grade.
Dental health component
Intra-examiner agreement ranged from a Kappa SCAN component
value of 0.837 for the referred population seen For the referred sample there were three raters;
under ideal conditions, to 0.754 for the non- the patient and the two examiners. This gave two
referred population. In total there were 14 errors patient ratings (PI and P2), two ratings by
81:4 PETER H. BROOK AND WILLIAM C. SHAW
05
1-5
45
25
Figure 1 The SCAN Scale. Originally presented in colour in a horizontal arrangement. 0.5 extreme left. 5 extreme right.
Table 2 Guidelines for the interpretation of Kappa. Table 3 Examiner variability. Pearson's correlation
coefficients for SCAN.
Kappa statistic Strength of agreement
Referred Population
<0.00 Poor
0.00-0.20 Slight PB2 PI P2 WCS
0.21-0.40 Fair
0.41-0.60 Moderate PB1 0.87 0.50 0.36 0.71
0.61-0.80 Substantial (72) (82) (72) (82)
0.81-1.00 Almost perfect
PB2 0.45 0.40 0.73
(72) (72) (72)
PI 0.67 0.37
(72) (82)
examiner 1 (PB1 and PB2), and one rating by P2 0.29
examiner 2 (WCS). The examiner reproducibi- (72)
lity, and the comparability of patient and exa-
miner ratings, were investigated using Pearson's
correlation coefficient. The results are listed in tists were quite high, they were poorer than those
Table 3 with the number of repeat examinations obtained by Evans and Shaw (1987), where self-
in parenthesis. retaining lip retractors were used during the
Whilst the correlations between the orthodon- orthodontists' and the patients' assessments. For
ORTHODONTIC TREATMENT PRIORITY 313
this reason, self-retaining cheek retractors were in Table 5, and Figs. 2 and 3. The SCAN Index
used routinely for the examination of the non- scores are illustrated in Table 6, and Figs. 4 and
referred sample. 5.
In the school survey, 46 subjects were rated
twice by the orthodontist (PB) on the SCAN
desirable. Self-retaining cheek retractors are a the SCAN Index, showded a distribution skewed
useful aid to recording dental aesthetics. towards the attractive end of the scale. The
patient ratings are skewed even further towards
The referred sample the attractive end of the scale, i.e. there was a
As expected, the referred sample showed a large tendency for subjects to overrate their dental
proportion of patients scoring in the higher attractiveness.
grades of the Dental Health Index, with all but 6
per cent in the highest three grades. The ortho- Comparison of the two samples
dontists' SCAN ratings showed a similar shift to From an overview of the data recorded from the
the unattractive side. The patient ratings showed samples, it appears quite obvious that there are
this effect to a much smaller extent. significant differences between them. Indeed the
median test and the /-test demonstrate this
The school population readily for the Dental Health component and the
There was a much more even spread of patients SCAN component respectively (p< 0.0001).
amongst the grades of the Dental Health Index However, the findings from the two surveys
for this population, with approximately one are not directly comparable. Firstly, the referred
third of the subjects in grade 3, and one third sample had a wide age spread, and secondly,
either side of this. The professional ratings using when applying the Dental Health component to
ORTHODONTIC TREATMENT PRIORITY 315
the referred population, the examiners had Table 6 Distribution of SCAN ratings (Orthodontist
access to radiographs. For the school sample, it score).
was necessary to set criteria for such parameters
as unerupted or missing teeth, i.e. except for Referred Population School 1Population
incisors and first molars, all teeth were assumed SCAN
to be present, at age 11-12 years, premolars and rating Numbers Percentage Numbers Percentage
canines were assumed to be unerupted but not
0.5 0 0.0 16 4.8
impacted, missing upper lateral incisors and 1.0 2 1.0 54 16.2
lower incisors were assumed to be developmen- 1.5 19 9.4 65 19.5
tally absent whereas missing upper central inci- 2.0 17 8.4 59 17.7
sors and first molars were assumed to have been 2.5 28 13.8 62 18.6
extracted. 3.0 33 16.3 33 9.9
3.5 39 19.2 26 7.8
In an attempt to overcome this, a sub-group of 4.0 48 23.6 12 3.6
25 from the referred sample, with a similar age 4.5 8 3.9 5 1.5
range to the school sample was selected. Any 5.0 8 3.9 1 0.3
conditions that would have required radiographs Total 202 100.0 333 100.0
to confirm the diagnosis, were regraded using the
316 PETER H. BROOK AND WILLIAM C. SHAW
criteria set for the population seen without acceptable as it has been shown that most of the
radiographs. traits can be recorded with a high degree of
It was still evident that the samples were drawn
precision (Helm et al., 1975; Helm, 1977) with up
from different populations (Mest p< 0.001 for to 80 per cent agreement. However in their pure
the SCAN ratings; median test p< 0.001 for the form they do not record treatment priority.
Dental Health ratings). These tests seemingly The allocation of weighting factors to traits
validate the index, at least in terms of the can give an overall figure that is intended to
priorities of patients or dentists in bringing about
represent a score of severity, and thus treatment
referral to an orthodontist. priority. Several indices of this type have been
developed (Summers, 1971; Draker, 1960; Gra-
Comparison with previous methods of recording inger, 1967). Correlation coefficients for exa-
treatment priority miner agreement for such indices have ranged
Angle's classification (Angle, 1899) has been from a Spearman correlation coefficient of 0.903
shown to have poor reproducibility (Gravely and (Summers, 1971) to as low as 0.34 (Albino et al.,
Johnson, 1974) and has no usefulness in record- 1978) in a community screening setting. The
ing treatment priority. For epidemiological use, validity of such indices relies on acceptance of the
the registration techniques described by Bjork et authors' weightings.
al. (1964) and Baume et al. (1973) may be quite Indices based upon the classification of mor-
ORTHODONTIC TREATMENT PRIORITY 317
25-
P 19.5
E 20 - 18.6
R 17.7
C
E 15 -
N
T
A 9.9
G 10-
E 7.8
5 - 3.6
1.5
0.3
1 2 3 4 5 6 7 8 10
SCAN RATING
Figure 5 Distribution of ratings for the SCAN index obtained from examination of 333 unselected schoolchildren.
phological traits rely on the subjective opinion of development was based upon a full analysis of
an experienced judge to define the dividing lines the available literature (Brook, 1987) and the
between each trait (Linder-Aronson, 1974; experience gained from a longitudinal survey
Lundstrom, 1977; Malmgren, 1980). Used as (Shaw et ai, 1986), it is not felt that its validity
such, the percentage concordance ranges from can be inferior to that of other indices. The
55.9 to 74.6 per cent (Malmgren, 1980). inclusion of a separate index to record aesthetic
Subjective clinical opinion alone has agree- impairment removes the most subjective element
ment of about 80 per cent in most studies from indices of this kind. Good levels of agree-
(Bowden and Davies, 1975; Helm et ai, 1975) ment for this component have been demon-
but the validity of such judgements depends strated (Pearson's correlation coefficient values
upon the examiners' knowledge of the harmful from 0.71-0.95).
effects of malocclusion. In addition, inexper- Some support for the validity of the index
ienced examiners will find it difficult to apply comes from the observation that fewer subjects
such techniques. in the lower grades were referred for orthodontic
The orthodontic index of treatment need" advice. The extent to which it represents com-
described in this report has examiner agreement mon professional opinion is presently being
levels that compare well with any of those evaluated. However, true validity (i.e. that the
previously described (80.5-93 per cent). As its index measures what it purports to measure)
318 PETER H. BROOK AND WILLIAM C. SHAW
Table 7 Interrelationship between the SCAN and Dental Health Index scores.
(Cumulative Percentages)
must await the compilation of a greater body of tions could produce a similar percentage. An
knowledge than that which is currently available. alternative method would be to combine the
It may accurately reflect contemporary profes- scores from the two components to give an
sional opinion, but this may be erroneous. overall score, then define limits based on this
figure.
Epidemiological uses for the index Eventual definitions of cut-offs must reflect the
During testing of the index on the school sample, setting in which treatment would be provided
it was felt that it represented a simple, quick and and include a consideration of the success rate of
reasonably reproducible method of recording the treatment which would be available, the
orthodontic treatment need. As all the traits are iatrogenic risks, and the cost (Shaw, 1987).
simple to record, it may be possible for less highly
trained personnel to apply the index, following
suitable training and calibration. Conclusions
An index with two components has been devel-
Further development of the index oped to record orthodontic treatment priority.
As developed so far, the index records the dental The first of these components records need for
health need for treatment, and the aesthetic treatment on dental health and functional
impairment, and by implication the social- grounds. The second component records the
pschological need for treatment. As yet, no aesthetic impairment, and by implication, the
attempt has been made to combine these into an justification for treatment on social-psychologi-
overall assessment of treatment need, or to define cal grounds.
scores below which treatment should be with- The indices were tested on a sample of patients
held. referred for orthodontic treatment and advice,
To assist discussion in this area, a table and on a random selection of 11-12-year-old
showing the cumulative percentages of patients school children. It was easy and quick to use and
having varying combinations of the Dental had acceptable reproducibility. True validation
Health grades, and the SCAN ratings has been of such an index must await the emergence of
constructed from the non-referred sample (Table further research data on the effects of malocclu-
7). sion, but the present index can be adaptable to
Many authors quote figures of around 50 per new information. Work is currently in progress
cent for the percentage of children who would to gauge the extent to which the index reflects
benefit from orthodontic treatment (Gardiner, common professional opinion.
1956; Haynes, 1982; Foster and Walpole Day, Defining specific ranges within which patients
1973). From Table 7, it can be seen that this should, or should not be offered treatment has
number would be obtained if patients scoring 'not been attempted but a mathematical model
grade 3 or less on the Dental Health Index and has been suggested that can define combinations
2.0 or less on the SCAN scale were excluded (51.3 of the gradings that will encompass varying
per cent), leaving 48.7 per cent. Other combina- proportions of a target population.
ORTHODONTIC TREATMENT PRIORITY 319
international conference on the occasion of the 25th Shaw W C, Addy M, Ray C 1980 Dental and social effects of
Anniversary of the Orthodontic Department of the malocclusion and effectiveness of orthodontic treatment: a
University of Nymegen, The Netherlands. October 22-24, review. Community Dentistry and Oral Epidemiology 8:
1987. Editors: C F A Moorrees and F P G M van der 36-45
Linden. Pp 63-81
Summers C J 1971 The occlusal index: a system for identify-