Plaster casts as the medium for data collection in orthodontic studies pose disadvantages. In this study, we aimed to assess the validity and reliability of using 3-dimensional (3D) photographs instead of plaster casts to determine the Index of Orthodontic Treatment Need (IOTN) score.
Data were collected retrospectively from the clinical records of 91 subjects. The IOTN grades were independently determined first from plaster casts, then from 2-dimensional (2D) and 3D photographs only, and then from 2D and 3D photographs combined with radiographs. IOTN grade agreement was assessed using kappa statistics and percentages of agreement.
The percentages of agreement between both photographic sets and the plaster casts varied among the different occlusal traits from 63.7% to 93.4%. Agreement between the IOTN grades obtained from 2D and 3D photographs only and the IOTN grades obtained from plaster casts was fair (K = 0.35). The reliability of using 2D and 3D photographs instead of plaster casts was improved when those were combined with radiographs.
In general terms, orthodontic treatment need can be assessed from 2D and 3D pictures; however, the individual occlusal traits are sufficiently assessed only when these pictures are combined with radiographs. Plaster casts remain the preferred method compared with 3D pictures for assessment of the IOTN.
We evaluated orthodontic need based solely on photographic records.
High-precision 3-dimensional (3D) surface imaging was used.
Orthodontic need was determined sensitively and specifically by 2D and 3D pictures.
This enables orthodontists to meet higher scientific demands of validity and reliability.
Malocclusions have varying prevalences in different countries but are a common oral health problem worldwide. Despite the extensive applications of orthodontics, large-scale population-based epidemiologic studies, such as prospective cohort studies, that investigate the possible beneficial effects of orthodontic treatment, are scarce. Whether orthodontic treatment can reduce susceptibility to dental caries, periodontal disease, temporomandibular disorders, traumatic dental injuries, or psychosocial problems is still inconclusive. In general, the matter of treatment may eventually be better described as correcting a deviation from the norm than in healing an acute disease, and the impact on oral health may occur after many years in which oral habits, parafunction, and other influences on the occlusion might develop. This nature of the orthodontic specialty challenges epidemiologic research such as long-term evaluation of treatment, but it also highlights its importance.
One of the most important parts of epidemiologic studies is systematic, accurate, and credible data collection. Occlusal indexes have been proposed as a means of acquiring descriptive orthodontic data, such as the Index of Orthodontic Treatment Need (IOTN). The IOTN is valuable as a simple and quick measure for scientific studies because it needs to measure only the worst features of the malocclusion. In addition, intraexaminer and interexaminer variabilities are reduced when calibrated examiners use the IOTN to assess orthodontic treatment need.
The IOTN grade is usually obtained during a clinical examination in combination with intraoral and extraoral photographs, radiographs, and plaster casts. The agreement of the IOTN applied to plaster casts and the IOTN applied during oral examinations is reported to be high. However, both clinical examinations and taking plaster casts provide practical constraint to the collection of orthodontic data, for example, in clinical trials or large-scale observational cohort studies. Although direct intraoral examination is the gold standard to apply the IOTN in clinical practice, this cannot meet the prerequisite of gathering scientific data in terms of repetition and validation. The dental impression process in turn is time-consuming and unpleasant for study participants.
Application of the IOTN solely based on photographic records, which burdens study participants less, would simplify the conduct of large-scale studies. In dental practice, 2-dimensional (2D) photographs are already an important part of orthodontic planning but are valid for occlusion determinations only when used in combination with plaster casts and radiographs. The advent of 3-dimensional (3D) imaging techniques may provide opportunities for orthodontic large-scale studies without requiring plaster casts. Recently, intraoral scanners were introduced to dental practice to avoid the disadvantages of traditional impression processes, such as patient burden and information errors. However, for large population-based cohort studies, with an enormous amount of data collection outside the orthodontic practice and from diverse clinical specialities, such as observational cohort studies, these scanners are too time-consuming. A cheaper, faster, and even more comfortable alternative for epidemiologic studies might be highly accurate extraoral surface imaging; however, this possibility of 3D photographs has not yet been evaluated in the literature.
In this study, we aimed to investigate the validity and reliability of determining orthodontic treatment need with the IOTN based on 2D and 3D extraoral photographs of the dentition. For this purpose, we compared the grades of the IOTN-dental health component (DHC) applied to 2D and 3D photographs—once combined with radiographs and once without radiographs—with the grades of the IOTN-DHC applied to plaster casts from clinical records.
Material and methods
This validation study was reviewed and approved by the Medical Ethical Committee of the Erasmus MC, Rotterdam, The Netherlands (MEC-2013-098). Informed consent was given by the participants’ parents before the photographs were taken. The study was carried out in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki) for experiments involving human subjects.
The study population was a convenience sample recruited from one orthodontic practice in Capelle aan den Ijssel, The Netherlands. All eligible study participants were children when their orthodontic treatment began. Children with clefts or other dentofacial deformities were excluded. We conducted a power calculation based on the proposed method by Cicchetti and obtained a minimal required sample size of 50 subjects. In total, 91 children with a mean age of 11.77 ± 1.39 years were included in the study. All of them were in the mixed dentition stage.
For this validation study, the data were retrospectively collected by means of plaster casts and photographs. No data were acquired from direct clinical examinations. Except for the 3D photograph, all materials used in this study were obtained as part of standard clinical procedures. All assistants were trained and calibrated in taking the photographs and dental impressions.
The impressions to make plaster casts were taken by the orthodontic practice staff before the start of orthodontic treatment. An orthopantomogram, a cephalogram, and 3 intraoral 2D photographs were also taken at the start of treatment. With a camera (Lumix DMC-TZ7; Panasonic, Osaka, Japan), intraoral photographs were taken from 3 perspectives: frontal, left buccal, and right buccal views. The children were asked to show their teeth with the help of cheek retractors. We also obtained a 3D photograph using the 3dMD imaging system and the 3dMD vultus viewer software (3DMD Imaging Equipment, Atlanta, Ga). The photograph was taken of the face while the child made the teeth visible with cheek retractors. The Figure shows a complete radiographic set of a subject without the lateral 2D photographs.
The IOTN was used to assess orthodontic treatment need and malocclusions. The IOTN recognizes 5 grades of orthodontic treatment need, ranging from no need to very great need. There are 2 IOTN components: the aesthetic component and the DHC. We focused on the DHC, which categorizes the detrimental effects of various deviant occlusal traits. When using the DHC of the IOTN, only the most severe grade and specifications of the deviant occlusal trait are typically recorded (final IOTN-DHC grade). However, we documented additionally every applicable trait specification to identify the particular benefits of 3D photographs as well as the possible sources of disagreement in retrospect.
The IOTN-DHC uses a systematic and reliable scoring technique with a hierarchical scale based on the following order of severity of the deviant occlusal traits: missing teeth, overjet, crossbite, displacement of contact points (also called crowding), and overbite (including deepbite and open bite).
For each participating child, the IOTN grade was assessed 3 times using the IOTN-DHC scoring based on different materials. First, each patient’s plaster cast was assessed by a calibrated examiner (L.K.) in combination with the patient file. Second, the IOTN-DHC grade was determined on intraoral 2D photographs combined with the 3D photograph (2D-3D set) by 2 other calibrated examiners (A.M.H., E.M.O.). Third, the IOTN-DHC grade was assessed on the 2D-3D set combined with the radiographs by the same 2 examiners (radiographic set). Randomly selected subgroups were reassessed by the same examiner and the other examiners as well as the same set and the other photographic set.
Statistical analyses were performed with SAS software (version 9.3; SAS Institute, Cary, NC).
We used linear weighted kappa statistics for intrarater reliability (between the same photographic sets), test-retest reliability (between the same photographic sets), and intermethod reliability (between different photographic sets) among the 5 grades of the IOTN. Linear weighted kappa was also used for the comparison between grades from the plaster casts with the 2D-3D set or the radiographic set, respectively. Linear weights were applied, because with regard to the accuracy of the method, the difference between the second and third IOTN grades has the same importance as the difference between the third and fourth grades or between the fourth and fifth grades. The IOTN developers recommended analysis of kappa agreement with linear weights for the DHC.
The ability to assess different occlusal traits using both sets of photographs was evaluated based on the unweighted Cohen kappa coefficient because the specifications are nominal data. The occlusal traits per group of the hierarchical scale of the IOTN-DHC that were identified from the 2 photographic sets were each compared with the occlusal traits determined using the plaster casts. The agreement for all kappa correlations was evaluated using the guidelines suggested by Landis and Koch.
After we categorized the different IOTN grades into treatment need (grade >3) and no treatment need (grade <3), we calculated sensitivity, specificity, positive predictive value, negative predictive value, and positive and negative likelihood ratios for both photographic sets with the plaster casts as the reference. Treatment need was determined using a cutoff value of greater than grade 3 as suggested by Roberts and Richmond.
Table I shows the frequencies of the highest IOTN grades when using each of the 3 different assessment methods. With the IOTN grades obtained from plaster casts, we found an orthodontic treatment need prevalence of 71.4%.
|Plaster casts||2D-3D set||Radiographic set|
|IOTN grade 2||2||8||4|
|IOTN grade 3||24||24||23|
|IOTN grade 4||50||47||50|
|IOTN grade 5||15||12||14|
|Displacement of contact points||90||84||84|
Table II presents the comparison between the final IOTN grade obtained from both sets of photographs and the final IOTN grade obtained using the plaster casts. Table III presents the comparison of deviant occlusal traits identified from both sets of photographs and with those from plaster casts.
|2D-3D set vs plaster casts||Radiographic set vs plaster casts|
|Total||91||0.35 (0.21-0.50)||53.7||91||0.44 (0.29-0.59)||62.6|
|Examiner 1||49||0.34 (0.12-0.56)||61.2||51||0.50 (0.29-0.71)||70.6|
|Examiner 2||42||0.37 (0.19-0.54)||47.6||40||0.42 (0.25-0.60)||52.5|
|n (n ∗ )||2D-3D set||Radiographic set|
|Missing teeth||91 (15)||0.28 (0.06-0.56)||85.7||0.59 (0.36-0.82)||90.1|
|Overjet||91 (64)||0.67 (0.51-0.82)||84.6||0.69 (0.53-0.84)||85.7|
|Crossbite||91 (17)||0.65 (0.44-0.86)||90.1||0.68 (0.48-0.89)||91.2|
|Displacement of contact points||91 (90)||0.24 (0.0-0.62)||93.4||0.24 (0.0-0.62)||93.4|
|Overbite||91 (69)||0.30 (0.14-0.46)||68.1||0.22 (0.07-0.38)||63.7|
|Other||91 (12)||0.41 (0.11-0.70)||89.0||0.53 (0.31-0.75)||86.8|