Reliability and discriminatory power of methods for dental plaque quantification

Objective This in situ study evaluated the discriminatory power and reliability of methods of dental plaque quantification and the relationship between visual indices (VI) and fluorescence camera (FC) to detect plaque. Material and Methods Six volunteers used palatal appliances with six bovine enamel blocks presenting different stages of plaque accumulation. The presence of plaque with and without disclosing was assessed using VI. Images were obtained with FC and digital camera in both conditions. The area covered by plaque was assessed. Examinations were done by two independent examiners. Data were analyzed by Kruskal-Wallis and Kappa tests to compare different conditions of samples and to assess the inter-examiner reproducibility. Results Some methods presented adequate reproducibility. The Turesky index and the assessment of area covered by disclosed plaque in the FC images presented the highest discriminatory powers. Conclusions The Turesky index and images with FC with disclosing present good reliability and discriminatory power in quantifying dental plaque.

In order to improve the quality of research in this field, methods of plaque quantification should have good discriminatory power and reliability.
Some appropriated indices to assess the association of plaque with periodontal disease have presented good reproducibility 5,12 , and few manuscripts have demonstrated their discriminatory validity 1,19 . However, there is still no research on the evaluation of the feasibility of methods for quantifying the dental plaque formed under high frequency of sucrose exposition. This kind of plaque is probably more prone to provoke dental caries 13 . Therefore, studies should be conducted to test the reliability and discriminatory power of methods of plaque quantification in these conditions. to distinguish mature from immature plaque is the two-tone disclosing agent, which stains the mature plaque in blue purple and the immature plaque in red 3,11 . Indeed, a visual index has previously been described in order to make this kind of distinction 5 . Nevertheless, comparison between this visual index and the laser fluorescence camera in detecting mature plaque has not been assessed yet.
The aim of this in situ study was to evaluate the reliability and discriminatory power of visual methods using two-tone dye and laser fluorescence camera in quantifying dental plaque formed under high frequency of sucrose exposition. It was also verified whether the presence of dental plaque showing red autofluorescence with the FC could be correlated with the plaque stained in blue purple with the two-tone disclosing dye. The null hypothesis tested was that there is no difference among methods regarding evaluation of dental plaque and reliability.

MATERIALS AND METHODS
This in situ study was approved by the local Research Ethics Committee, and volunteers' written consent was obtained.  In order to evaluate the association between red autofluorescence of plaque using the FC device and the dental plaque disclosed in blue purple with the two-tone dye, one examiner (FMM) evaluated the area covered by red fluorescent plaque assessed using the FC device without disclosing, and the area disclosed in blue purple in digital photographs.

Statistical analysis
The percentage of area covered by plaque The inter-examiner reproducibility with the indices was firstly evaluated using a Cohen's Kappa test, 6 and quadratic weighted Kappa test. 10 For the methods that evaluated the area covered by plaque, the inter-examiner reliability was calculated using the intraclass correlation coefficient (ICC) and 95% Confidence Interval

RESULTS
The discriminatory power of the different methods for dental plaque quantification is presented in Table 1. All methods showed difference at least among two groups. However, the Turesky index and the percentage of area covered by plaque evaluated using the FC after disclosing demonstrated differences among the three groups, showing better discriminatory power than the other methods (Table 1).
Regarding the reliability, the three indices presented similar inter-examiner reproducibility using Cohen's Kappa analysis, but the value of Turesky index was improved when the weighted approach was used ( Different letters indicate statistically significant differences among the groups within the same row (p < 0.05). * Mean of scores standard deviations of each index. ** Mean of percentage area of block surface covered by plaque ± standard deviations. FC = Fluorescence camera.
Reliability and discriminatory power of methods for dental plaque quantification value similar to Turesky index using the weighted Kappa test ( Table 2).
The area of red autofluorescence of dental plaque observed using the FC was correlated to the area of dental plaque disclosed in blue purple with the two-tone disclosing dye (Rs = 0.727; 95% CI = 0.524 -0.852, p < 0.0001). However, the area of plaque disclosed in blue purple (mean = 0.297) was statistically significant higher than the area of red autofluorescent plaque (mean = 0.216, p = 0.0008).

DISCUSSION
The majority of studies that assessed the feasibility of methods for dental plaque quantification emphasize its relationship to periodontal disease 1,4,5,9,12,19,20  were not proposed to be used in square blocks, but in dental surfaces, the specimens were positioned in the removal appliance aiming to mimetic different regions of a dental surface.
Thus, we assessed the power discrimination and reliability of methods for dental plaque quantification under these conditions.

Cohen's Kappa (SE) Weighted Kappa (SE) ICC (95 % CI)
Silness  In this study, Turesky index and the quantification of the area covered by disclosed plaque detected by the FC device presented the highest discriminatory power. It was previously expected that the cleaned blocks had to present less amount of plaque than the other samples, and that the enamel blocks without plastic mesh protection had to show lower amount of plaque than the specimens protected by the plastic mesh. These two methods were able to demonstrate these differences. Other methods presented significant differences between two of three groups, but no difference within the three groups. Another study claimed that indices have presented better discriminatory power than measurement of area covered by plaque 1 , since the latter is unable to detect small differences in dental plaque quantity 17 , corroborating our findings. On the other hand, another study showed that the area assessment was better in detecting higher amount of plaque than other visual indices 19 .
The higher discriminatory power emphasizes the ability of these methods in distinguishing dental plaques in different amounts. The Turesky index scores the plaque amount according to the part of the dental surface in which the plaque is found. Therefore, smaller amounts of plaque (cleaned blocks) were usually associated with lower scores. Additionally, the disclosed plaque detected by the FC probably tended to be identified easily in specimens containing higher amount of plaque than in those previously In earlier studies, previously to plaque assessment, the examiners were trained and calibrated 12,14,21 . If the training had been performed in our study, probably the agreement values would have been higher.
As the present study intended to evaluate plaque formed under high frequency of sucrose exposition, which is more related to caries lesions induction, Turesky index and measurement of area covered by disclosed plaque detected by the FC seem to be more indicated to assess dental plaque in studies of dental caries, since they presented good reliability and discriminatory power. It has to be stated that there is an increase in the cost, regarding the use of FC.
As visual index using two-tone dye and laser fluorescence devices were possibilities to identify mature plaque, the comparability between them is extremely important. A previous study evaluated the relationship between the assessment of mature plaque with a quantifying light fluorescence (QLF) and a typical one-tone dye 7 However, no comparison between fluorescence devices and two-tone dyes were performed.
A red autofluorescence of the dental plaque illuminated by a blue light (408 nm) from the QLF device has been observed. 7,16,17,25 Authors have suggested that the obligate anaerobic bacteria are the responsible for the red autofluorescence, and these bacteria are indicative of mature plaque 7,25 . In our study, we observed the red autofluorescence mainly in the plaque on the enamel blocks which were protected by the plastic mesh, in which it is probably that a more complex plaque has been formed. However, we did not evaluate microbiologically the plaque formed on the blocks, not even their mineral loss.

Reliability and discriminatory power of methods for dental plaque quantification
The method using the FC without disclosing dye did not reliably assess the dental plaque.
However, when we employed the disclosing dye and took the images with the FC, the method was suitable. Other studies using an intraoral camera capable to obtain images and assess the plaque area have been performed, but with a regular illumination 15,21 . The advantage in using the novel FC device would be the possibility to achieve images of red autofluorescent plaque without any disclosing dye, which is a plaque possibly associated to high dental caries risk 25 , and further, to obtain images of the disclosed plaque. Despite of that, our study showed poor reproducibility in using evaluation of FC images without disclosing. The poor reproducibility per se could be considered an important disadvantage of the method. However, considering the lack of previous intensive training in using the device, this parameter can be improved if examiners are previously trained. Moreover, the FC could be used for dental caries detection after the cleaning of the teeth 22 , nevertheless, this was not the aim of our research. Therefore, more studies using the FC with dental plaque and caries evaluation are also necessary.
The FC seems to have the same principles of the QLF, but studies comparing both devices have not been performed yet. Considering the FC uses the same wavelength of the QLF, but is associated with different software, the simple extrapolation of results obtained with the first one was not really appropriated. The QLF device has already been extensively studied and previous researches have shown good results in detecting disclosed and undisclosed dental plaque. 16,17,25 Furthermore, the QLF has shown good results for caries lesions assessment 2 .
Another method to differentiate mature from immature dental plaque is using a two-tone disclosing agent 3,11 .This dye stains the immature plaque in red and the old plaque in blue purple.
In the present study, we observed a significant correlation between the area of dental plaque exhibiting red autofluorescence and the area stained in blue purple. However, the mean of area stained in blue purple was significantly higher than the mean area of red fluorescent plaque. This difference could be explained by the different mechanisms to detect mature plaque. While the two-tone disclosing dye detects mature plaque due to its thickness, the phenomenon of red fluorescence is probably due to some bacterial metabolites, possibly porphyrins. Additional studies relating mature plaque detected by both methods and increased risk of oral diseases must be carried out.

CONCLUSIONS
Turesky index and the quantification of the area covered by plaque using the new FC after disclosing have good reliability and discriminatory power in quantifying dental plaque formed under high frequency of sucrose exposition.
Furthermore, there are correlation between red autofluorescence and dental plaque disclosed in blue purple, but the method with two-tone disclosing agent shows a higher area of mature plaque than the fluorescence-based method.

ACKNOWLEDGMENTS
The authors wish to thank the volunteers for