A prospective randomized study was undertaken to compare conventional study model-based manual Peer Assessment Rating (PAR) scoring with computer-based automated scoring using scanned study models or intraoral scanning.
The sample consisted of 67 patients, mean age 15.03 (range 11-37) years. Sixty-seven patients underwent alginate impression-taking and intraoral scanning (CS 3600; Carestream Dental, Stuttgart, Germany) at a single appointment in a randomized order. For each patient, a weighted PAR score was calculated manually by a calibrated examiner using study models and a PAR ruler (conventional group), and automatically using Carestream Dental CS Model+ software and data from scanned study models (indirect digital group) or intraoral scans (direct digital group). All procedures were timed, and each patient completed a binary questionnaire relating to their experience.
There were no significant differences between methods for calculated mean weighted PAR score ( P = 0.68). Mean (standard deviation) chairside time for impression-taking was 5.35 (± 1.16) minutes and for intraoral scanning, 7.76 (± 2.76) minutes ( P <0.05). Mean (standard deviation) times taken to calculate weighted PAR scores were 2.86 (± 0.96), 5.58 (± 2.33), and 4.58 (± 2.18) minutes for conventional, indirect digital, and direct digital groups, respectively ( P >0.05). A total of 61 patients (91%) preferred intraoral scanning to impression-taking.
Automated PAR scoring using cast study models or intraoral scanning is valid, though both methods take longer than conventional scoring. Patients prefer intraoral scanning to impression-taking.
The protocol was not published before study commencement.
Automated digital Peer Assessment Rating scoring is as accurate as manual.
Automated scoring takes longer.
Patients prefer intraoral scanning to dental impressions.
The Peer Assessment Rating (PAR) is an occlusal index developed to provide an objective and standardized measure of static occlusion at any stage of treatment using cast dental study models. It was originally formulated over a series of 6 meetings by a group of 10 experienced orthodontists in the United Kingdom and quantifies the deviation of teeth from their ideal position by assigning numerical scores within 5 categories: (1) contact point displacement in the anterior segments, (2) buccal occlusion, (3) overjet, (4) overbite, and (5) dental centerlines. The higher the overall score, the greater the deviation from ideal. Conventional PAR scoring is performed manually by a trained and calibrated individual using a specific clear acetate ruler and is a contractual requirement for the assessment of treatment complexity and outcome for orthodontic providers within the government-funded National Health Service. The process of PAR scoring can be time-consuming and despite calibration, some aspects of the assessment have been associated with low reliability. Moreover, the routine use of cast dental stone study models is on the decline. They pose a considerable storage problem for orthodontic providers, and the use of digital study models is now an accepted alternative.
Digital study models are a virtual 3-dimensional (3D) representation of the dentition and occlusion that can be viewed and manipulated on a computer screen in a similar manner to traditional cast stone study models. A common method for producing these virtual models is by imaging existing cast study models, a process known as indirect digitization. The technology is nondestructive, and the original models are preserved during the process. , Retrieval of data is almost instantaneous, and transmission to dental laboratories and other care providers can be performed electronically. More recently, the advent of high resolution intraoral scanners with the capability of scanning the dentition directly has led to a significant increase in the use of digital technology in orthodontics. It is now possible to fabricate a multitude of orthodontic appliances using direct digitization from the dentition including study models representing the dentition and occlusion. Current evidence would suggest that trueness and precision are superior for the indirect method of virtual study model generation. , However, patient-focused studies have shown that the avoidance of dental impressions associated with intraoral scanning and direct digitization is favored by patient groups.
The validity and reliability of PAR scoring using digital study models have been previously investigated. These studies all used indirect rather than direct digitization, and the software required the operator to visually identify relevant landmarks on the display screen followed by individual measurement of component dimensions within the PAR analysis. The measurements were taken separately and then collated by the software to yield the final PAR score. None of these studies found any significant differences between PAR scores derived from conventional or digital models but concluded that digital scoring was less time-efficient and had poorer intraexaminer reliability owing to larger discrepancies in digital landmark identification. ,
Carestream Dental have recently developed the CS Model+ software program, which automatically detects dental landmarks and allows automatic calculation of weighted PAR scores, without the need for separate measurement of contact point displacements, overjet, overbite, and other occlusal features. The software is marketed for use with the commercially available Carestream Dental CS 3600 intraoral scanner (Carestream Dental, Stuttgart, Germany).
Specific objectives and null hypothesis
The aim of this study was to investigate the validity of digital PAR scoring by comparing conventional manual assessment using cast dental study models with automated scoring using scanned dental study models and intraoral scanning. The null hypothesis was that there is no difference in mean calculated PAR score between the 3 methods.
Material and methods
Study design and any changes after study commencement
This prospective randomized study investigated the validity of PAR scoring using conventional manual assessment from cast dental study casts (conventional group) with automated scoring using data derived from indirect scanning of dental study models (indirect digital group) and direct intraoral scanning (direct digital group). Ethical approval was granted by the United Kingdom Health Research Authority (Integrated Research Application System identification: 230630), and written informed consent was received from all participating patients, including parents or guardians of children participating in the study. This project is registered at ClinicalTrials.gov ( NCT03405961 ), and study details are available at clinicaltrials.gov/ct2/show/NCT03405961?term=03405961&rank=1 . No changes were made to the methods after study commencement.
Participants, eligibility criteria, and setting
Eligibility for participant inclusion in this study comprised patients aged between 11 and 50 years attending for record-taking as part of routine orthodontic treatment. After informed consent, patients underwent intraoral scanning and conventional alginate dental impression-taking at a single appointment. These procedures were timed, and patients completed a short binary questionnaire relating to their experience. All clinical procedures and time measurements were undertaken by a single operator (SL).
Participants were recruited from patients attending the Department of Orthodontics at Kingston Hospital NHS Trust between November 02, 2018 and December 14, 2018. Data collection took place during routine appointments for clinical record-taking.
Each patient received conventional dental impressions and intraoral scanning at the same appointment in an order determined through random allocation.
Dental impressions were taken using plastic stock trays (LockTight Impression trays; DB Orthodontics, Silsden, United Kingdom), and alginate impression material (Create Alginate; DB Orthodontics) manually mixed by a qualified dental nurse following manufacturer instructions. A wax record of maximum intercuspation was also taken using dental modeling wax (Anutex; Kemdent, Swindon, United Kingdom) with the patient sitting upright. The impressions and wax bite were inspected by the operator, disinfected, and cast in dental stone (Herculite 2; Saint-Gobain Formula GmbH, Göttingen, Germany) by a commercial laboratory. Scanning of cast study models and direct intraoral scanning was carried out using the Carestream Dental CS 3600 scanner (provided courtesy of Carestream Dental, Stuttgart, Germany), which operates on a continuous video image capture system with a manufacturer-reported total scanning time of approximately 5 minutes for maxillary and mandibular dental arches, including bite registration. The mean trueness and precision for CS 3600 via indirect digitization have been reported as 45.8 and 24.8 μm, respectively. For intraoral scanning, the teeth were dried for 60 seconds with a 3-in-1 syringe before scanning of the maxillary and then mandibular arch occlusal, buccal, and lingual surfaces. Finally, 4 buccal bite images were taken with the patient in maximum intercuspation capturing the first permanent molars and canines on each side. Any areas of deficiency or inaccuracy in the scan were re-imaged. A digital stopwatch was used to time the process of impression-taking and intraoral scanning, and each patient was given a short binary questionnaire composed of 6 questions relating to themselves and their perception of these procedures ( Supplementary Fig 1 ).
Stone dental study models were PAR scored directly by the same calibrated individual (SL) using a conventional PAR ruler (Victoria University of Manchester, United Kingdom) and weighted (conventional group).
The scanned 3D dental study model and intraoral images were imported and viewed, and PAR scores automatically generated using CS Model+ software, version 2.0.5 (indirect digital and direct digital groups, respectively) run on an Intel Core i7 desktop computer (Intel, Santa Clara, Calif) ( Fig 1 ). The total time taken to generate the automated PAR score was measured using a stopwatch.
The same operator (SL) performed all procedures and scoring. All PAR scoring was repeated after 14 days to test intraexaminer reliability.
The primary outcome measure for this study was mean weighted PAR score. Secondary outcomes included mean chairside time for impression-taking and intraoral scanning, patient perception of which procedure was quicker, mean time to calculate a weighted PAR score using the different methods, and patient preference for impression-taking or intraoral scanning. There were no changes to study outcomes after the study had commenced.
Sample size was determined with a power calculation performed using G*Power version 3.1 based on the findings of a previous investigation of PAR scoring and digital study model accuracy. Assuming a correlation of 0.95 between conventional and digital PAR scores, a sample size of 66 subjects was required for 80% power to test using a 2-tailed test at 0.05 level of significance.
A simple randomization sequence for the order in which patients underwent either impression-taking or intraoral scanning was generated centrally at King’s College London using Statistical Package for the Social Sciences software, version 24.0 (IBM, Armonk, NY). This was independent from the clinical operator and communicated before the record-taking for each participant.
The nature of the study meant that blinding to intervention was not possible for participants during data acquisition. All data analysis was carried out blinded using a coded data set.
Data were analyzed using Statistical Package for the Social Sciences software, version 24.0. All data were tested for normality and equality of variances and transformed where necessary before applying parametric tests. Outliers were checked using sensitivity analysis. The significance level for all tests was set at 0.05. The difference between mean PAR scores of the 3 methods was tested with 1-way analysis of variance. Patient preference was assessed with chi-square and t tests. Time efficiency related to the 2 clinical procedures was evaluated using paired t test and to the 3 different scoring methods with 1-way analysis of variance. Intraexaminer reliability was examined with paired t test, Bland-Altman analysis, and intraclass correlation coefficient.