Main

The PAR Index and IOTN have now become widely used in the UK as orthodontic audit tools. They have provided valuable data to inform political and clinical debate on the quality of GDS orthodontics. The introduction of IOTN in the General Dental Services is currently being considered. In order to assess treatment inputs and outcomes using IOTN and PAR, two different measurement protocols must be learned and this duplication of effort is inefficient.

The PAR index has been accused of both undue leniency on poor finishes1 and undue harshness on treatments with limited aims.2,3 Experience with IOTN has shown that the need for treatment does not necessarily equate to the complexity of the treatment. It is important to make an assessment of complexity for the following clinical reasons (Richmond et al4):

  • To identify the most appropriate setting in which a patient should receive treatment.

  • To allow meaningful assessment of treatment outcomes.

  • To identify cases that are likely to take longer to treat.

  • To inform the patient of the likely success.

Richmond et al4 found that the factors which may determine the professional perception of orthodontic treatment complexity include:

  • Cost of treatment

  • Number of appointments

  • Length of appointments

  • Age of patient when treatment commenced

  • Initial PAR Score

However they discounted most of these treatment factors as little use in the formulation of a complexity index because they were too dependent on the choices and efficiency of the clinician. The PAR score was suggested to be an interim possibility in the development of a complexity index.

The ICON index5 has been developed recently and claims among other things, to evaluate orthodontic treatment complexity. ICON is based on the subjective judgements of 97 orthodontists from nine countries on 240 initial and 98 treated models.5 The index is described in full in the appendix. In brief, the index comprises five weighted measurements, and owes some of its structure to IOTN and PAR. The measured traits include:

  • Dental aesthetics as measured by the aesthetic component of IOTN.6

  • The presence of a crossbite

  • Anterior vertical relationship ie deep bites and open bites as measured by PAR.7

  • Upper arch crowding/spacing on a five point scale

  • Buccal segment antero-posterior relationship as measured by PAR.7

The sum of the weighted scores is interpreted using cut-off values and score ranges to indicate the treatment need, complexity and degree of improvement. The cut-off values are to give the index meaning and link the numerical ICON score to the original subjective description of the 97 orthodontists. It is claimed that using ICON to make these assessments is more efficient (than PAR and IOTN) because it requires only a single measurement protocol. ICON should prove to be a useful tool for research and audit, but if used alongside PAR and IOTN, could be an additional burden in a busy hospital or practice setting.

PAR and IOTN have been validated against UK dental opinion7,8 and in an increasingly global environment, perhaps a more internationally recognised standard should now be sought. ICON reflects the views of 97 orthodontists from eight European countries and the USA, and may provide an acceptable alternative to IOTN and PAR.

The aim of this paper is to investigate any link between PAR, IOTN and ICON with respect to orthodontic treatment need and outcome.

Method

A retrospective sample of 55 consecutively treated cases was identified from the orthodontic department of Middlesbrough General Hospital. Cases were treated by clinical assistants, specialist registrars or one of three consultant orthodontists. Discontinued cases were included in the sample.

Pre-treatment and post-treatment study models were scored using IOTN, PAR and ICON by a calibrated examiner from another hospital (TG). The examiner did not know which operator was responsible for treating the case when the models were scored. Pre- and post-treatment models were scored at random

Data analysis

Descriptive statistics were calculated to locate the central tendency and spread. Intra-examiner reliability was examined by repeating the examination of a random subset of 15 pre-treatment and post-treatment cases. Reliability was expressed using the weighted Kappa statistic for ordinal scores (IOTNDHC, IOTNAC, ICON complexity). The reliability of ICON scores and PAR were expressed using Root Mean Square to assess random error and a Students t–test for systematic bias. The relationship between ICON and the other indices was examined using non-parametric (Spearman) correlation.

The correlations were calculated for the pre-treatment and post-treatment data separately.

Outcome assessment has been traditionally made by assessing either the percentage change or absolute change in occlusal index scores. Assessment of occlusal improvement with ICON is made by subtracting 4x post-treatment score from the pre-treatment score,5 with the range of possible results nominally given five categories of improvement. These five categories can be thought of as similar in purpose to the three PAR improvement grades as determined from the PAR nomogram.

Bivariate correlations were calculated for PAR reduction versus ICON improvement and per cent PAR reduction versus ICON improvement.

Results

The reliability assessment of the indices is given in Tables 1a and 1b and descriptive statistics are given in Tables 2a and 2b. When used as an ordinal scale to express treatment complexity, ICON has marginally lower but acceptable reliability as the use of IOTN to assign treatment need categories. When used as a quasi-continuous variable (similar to PAR) it has a similar level of precision. The RMS value appears to be three times higher for ICON compared with PAR. This is because the minimum score increment for ICON is three times higher than PAR.

Table 1a Table 1a
Full size table
Table 1b Table 1b
Full size table
Table 2a Table 2a
Full size table
Table 2b Table 2b
Full size table

Table 3 shows the correlations between ICON, IOTN and PAR. The IOTN scores suggest that the sample showed a definite need for treatment on the grounds of dental health and aesthetics and that the standard of treatment was good. Figure 1 shows the plot of DHC versus ICON. For any given IOTN grade there are a range of possible ICON scores. As expected DHC and ICON scores are higher for pre-treatment cases. Both IOTN (DHC) and ICON classified two pre-treatment cases into the no treatment category, ie 98% of the sample justifies treatment on the grounds of dental health as defined by the UK usage of the DHC or according to the international standard set by ICON. Most of the post-treatment models score an acceptable end result as defined by ICON5 ie below 31. The poor correlation of ICON and DHC scores is related to the limited ordinal scale of DHC with most pre-treatment cases falling in grades 4–5 and post treatment 2–4. Many post treatment IOTN (DHC) scores were grade 4 and 3. This suggests a poor treatment result not reflected by IOTN (AC), PAR or ICON. This is because the study model protocol was applied and consequently any case with any crossbite anywhere in the mouth would have to be awarded grade 4 regardless of the size of the displacement. Most post-treatment grade 3 IOTN (DHC) scores were due to second molars not being routinely banded. This resulted in a small contact point displacement between first and second molars, which meant grade 3 had to be awarded. PAR or ICON does not pick this up.

Table 3 Table 3
Full size table
Figure 1: Scatter plot of ITON (DHC) versus ICON for pre-treatment and post-treatment casts.
figure 1

A dashed line denotes treatment thresholds. For IOTN the treatment category is for grades 4 and 5, and for ICON is any score greater than 43. The dotted line signifies the ICON value (31) below which a treatment can be considered acceptable. The possible score for DHC ranges 1–5 and 7–128 for ICON.

Full size image

Figure 2 shows the plot of IOTN (AC) versus ICON. For any given IOTN (AC) grade there are a range of possible ICON scores. According to the UK usage of IOTN (AC), 30% of this sample would not be justified for definite treatment need on the grounds of aesthetics alone (ie an AC score of 1 to 7).

Figure 2: Scatter plot of IOTN (AC) versus ICON for pre-treatment and post-treatment casts.
figure 2

A dashed line denotes treatment thresholds. For IOTN (AC) the treatment category is for grades 8 – 10, and for ICON is any scoregreater than 43. The solid line signifies the ICON value (31) below which a treatment can be considered acceptable. The possible score for AC ranges 1–10 and 7–128 for ICON.

Full size image

Figure 3 shows a plot of PAR score against ICON for pre- and post-treatment study casts. It can be seen that there is a good level of agreement between PAR and ICON.

Figure 3
figure 3

Scatter plot of PAR versus ICON for pre-treatment and post-treatment models.

Full size image

Figure 4 compares the outcome assessment of per cent PAR reduction with ICON improvement. Overall the relationship has a significant (Pearson) correlation of 0.74. Approximately 55% of the sample would be classified as greatly improved using ICON. The mean per cent PAR reduction is 85.0 and the mean PAR reduction is 32.8, suggesting the sample was very well treated compared with previous UK samples and in terms of international standards set by ICON.

Figure 4
figure 4

ICON improvement versus Percentage PAR reduction.

Full size image

Discussion

The sample used for this comparison was relatively small but reasonably diverse in as much as a fairly broad range of treatment starts were obtained and treatment was delivered by a variety of expert levels within the hospital service. The overall level of malocclusion is however more severe than previous estimates from the GDS, which typically have had mean initial PAR scores of 29.9 The sample, however, was not designed to generalise treatment outcomes in the GDS or hospital sector, but to compare the performance of the new indices.

Examination of Table 3 clearly shows that the degree of correlation between ICON and the other indices is influenced by the type of models in the sample. In this hospital sample, pre-treatment cases were moderately severe and most treatments resulted in a very minor malocclusion. For this reason a higher correlation is obtained when comparing index scores based on the pre-treatment and post-treatment dental casts. A lower level of correlation is reported for the assessment of treatment outcomes, probably because the treatment results were consistently good. (If there is little variation in a sample, poor correlations result).

The general performance of the ICON index seems to be largely in agreement with IOTN for assessing treatment need. Although ICON identified a higher proportion of the sample in need of treatment compared with IOTN (AC), on which it is heavily based, it did not over score on the proportion of cases identified for treatment on the grounds of dental health. This suggests that ICON could probably substitute IOTN (DHC) and produce largely similar results. An ICON score of greater than 43 reflects a definite need for treatment.5

The general performance of the ICON index seems to be equivalent to the PAR for assessing treatment results. Both indices identified similar proportions of the most improved cases, but ICON overall is a more stringent standard to attain a 'greatly improved' categorisation. An ICON score of 31 or less defines a clinically acceptable result.5 Like all occlusal indices, ICON will not be appropriate in assessing need and outcome in a small number of cases. This is certainly true of PAR.1,2,3 Its use does, however, seem appropriate when examining a representative sample of cases from a single practitioner or hospital department.

Conclusions

  1. 1

    ICON appears able to substitute IOTN (AC) and IOTN (DHC) as a means of measuring orthodontic treatment need. A value of greater than 43 for ICON defines a definite need for treatment.

  2. 2

    There is a close relationship between PAR score and ICON. Its use in measuring treatment outcome appears to be as appropriate as using PAR.

  3. 3

    ICON appears to require a more stringent standard than PAR to attain a 'greatly improved' categorisation. An ICON score of less than 31 defines a clinically acceptable result.