Reliability of light microscopy and a computer-assisted replica measurement technique for evaluating the fit of dental copings

Abstract The aim of this in vitro study was to assess the reliability of two measurement systems for evaluating the marginal and internal fit of dental copings. Material and Methods Sixteen CAD/CAM titanium copings were produced for a prepared maxillary canine. To modify the CAD surface model using different parameters (data density; enlargement in different directions), varying fit was created. Five light-body silicone replicas representing the gap between the canine and the coping were made for each coping and for each measurement method: (1) light microscopy measurements (LMMs); and (2) computer-assisted measurements (CASMs) using an optical digitizing system. Two investigators independently measured the marginal and internal fit using both methods. The inter-rater reliability [intraclass correlation coefficient (ICC)] and agreement [Bland-Altman (bias) analyses]: mean of the differences (bias) between two measurements [the closer to zero the mean (bias) is, the higher the agreement between the two measurements] were calculated for several measurement points (marginal-distal, marginal-buccal, axial-buccal, incisal). For the LMM technique, one investigator repeated the measurements to determine repeatability (intra-rater reliability and agreement). Results For inter-rater reliability, the ICC was 0.848-0.998 for LMMs and 0.945-0.999 for CASMs, depending on the measurement point. Bland-Altman bias was −15.7 to 3.5 μm for LMMs and −3.0 to 1.9 μm for CASMs. For LMMs, the marginal-distal and marginal-buccal measurement points showed the lowest ICC (0.848/0.978) and the highest bias (-15.7 μm/-7.6 μm). With the intra-rater reliability and agreement (repeatability) for LMMs, the ICC was 0.970-0.998 and bias was −1.3 to 2.3 μm. Conclusion LMMs showed lower interrater reliability and agreement at the marginal measurement points than CASMs, which indicates a more subjective influence with LMMs at these measurement points. The values, however, were still clinically acceptable. LMMs showed very high intra-rater reliability and agreement for all measurement points, indicating high repeatability.


Introduction
The fit of dental restorations has been subjected to numerous investigations. A poor marginal fit is associated with secondary caries 6 , which is among the most common causes of fixed partial-denture loss 28 . Both marginal and internal fit measurements are used to evaluate new materials and manufacturing procedures for dental restoration [9][10][11][12][13]21,26 .
Many measurement methods have been established to investigate the fit of dental restorations. To confirm a method's validity, however, two methods should be studied and compared 22 . A commonly used method is the internal silicone replica technique, described by Holmes, et al. 5 (1989), which enables the investigation of both marginal and internal gaps. Light-body silicone replicas fill the space between the restoration and the die. They are coated from the inner or outer side with a heavy-body silicone of various colors.
Another possibility for determining the replica's thickness is a computer-assisted technique that measures the optically captured replicas digitally 14,20 .
The validity of the replica technique for the predictable reproduction of cement thickness, regardless of the measurement point location, has been proven 15 .
However, evidence regarding the reliability and repeatability of the conventional light microscopy replica technique is sparse 21 , and none are available for the more recent computer-assisted replica technique.
The reliability of a measurement method is determined by comparing the measurements performed by several investigators (inter-rater reliability and agreement), whereas repeatability is calculated by repeated measurements by the same investigator. Thus, repeatability can be referred to as intra-rater reliability and agreement, which is done throughout this paper.
This study focused on analyzing the reliability and repeatability of the conventional light microscopy replica technique by determining both inter-rater and intra-rater reliability and agreement for specific measurement point locations. As a second step, the study aimed to compare the conventional light microscopy replica technique to the more recent computer-assisted replica technique by means of the respective inter-rater reliability and agreement for specific measurement point locations.
The hypotheses for this in vitro study were that: (1) the conventional light microscopy replica technique for the analysis of dental coping fit shows high intrarater reliability and agreement; (2) the conventional light microscopy method and the computer-assisted replica technique for the analysis of dental coping fit show high inter-rater reliability and agreement; and (3) the intra-rater and inter-rater reliability and agreement are independent of the specific measurement point location for the conventional light microscopy technique (intra-rater and inter-rater) and the computer-assisted replica technique (inter-rater) for analyzing dental coping fit.

Manufacture of copings and replicas
A prepared (chamfer) stainless steel maxillary canine (FDI 13) and its computer-aided design (CAD) surface model served as the master die (height 7.8 mm, cone angle 4°, bucco-oral diameter 10 mm at the margin). Using the CAD software (Surfacer ® V.10.0; Imageware Inc., Ann Arbor, MI, USA), the parameters were modified to create eight different CAD models resulting in varying fit. Therefore, the following parameters were modified: four CAD models showed high data density (point clouds of 123,029) and four CAD models showed low data density (point clouds of 8,513). Both data density groups were modified with regard to the fit by enlarging the originally sized data (1 st CAD model) in z-direction (height; 2 nd CAD model), in x-/y-direction (circumference; 3 rd CAD model) and x-/y-/z-direction (height and circumference, 4 th CAD model). Two titanium copings were manufactured for each of the eight different CAD models, resulting in 16 copings. For each titanium coping, five silicone replicas were produced for the light microscopy measurements

Accordance between measurements at both sides of each sectional cut (LMMs)
The accordance between the two measurements at both sides of each sectional cut (LMMs) was checked.

Results
Accordance between measurements at both sides of each sectional cut (LMMs) The mean ± SD (minimum value, maximum value) of the difference (absolute value) between the two measurements at each sectional cut was 5±7 µm (0,      Table 1 for each measurement point.

Discussion
The ICC for the intra-rater reliability of the LMMs Although the bias for the ax-d and inc measurement points is still close to zero, the ma-b bias is almost 6 times higher and the ma-d bias is almost 16 times higher than the bias for the intra-rater agreement.
The first investigator systematically determined lower marginal values (ma-b, ma-d) than the second investigator, which may have been due to the difficulty identifying the exact marginal measurement point.

The light-body material thins toward the margins, in
contrast to the thicker layers that are found axially and incisally. Thus, the Bland-Altman bias gave a more differentiated view than the ICC, as has been previously claimed 2 . In summary, the results indicate a subjective influence for the LMMs at the ma-b and ma-d measurement points. The inter-rater reliability of the LMMs still seems acceptable, however, given the clinically acceptable marginal fit values 7,29 . Thus, the study's second hypothesis can be accepted, whereas the third hypothesis must be rejected for inter-rater evaluation using LMMs.
The inter-rater reliability for the CASMs shows very high ICC values with a rather narrow 95% CI.
The bias values are closer to zero than the LMMs' Clinical factors, e.g. saliva or blood contamination, during dental impression making or margin geometry, lead to unintended modifications in coping fit 18  However, they did not specify how they calculated the mean difference (absolute values or not: "bias") and the coefficient of correlation. They also did not clarify whether the pairwise measurements had been performed by a single investigator (intra-rater reliability) or different investigators (inter-rater reliability). Neither the measurement points nor the inlay material used for the correlation analyses was specified. Thus, our study is the first to have evaluated both intra-and inter-rater reliability

Conclusion
The following conclusions can be drawn for the internal replica technique for evaluating the marginal and internal fit of dental copings.
The light microscopy replica measurements showed high intra-rater reliability and agreement (repeatability) and somewhat worse, but still clinically acceptable, inter-rater reliability and agreement at the marginal measurement points.
The computer-assisted replica measurement was slightly more objective than the light microscopy replica measurement and was independent from the measurement point.