Do combinations of clinical parameters related to caries activity status predict progression more accurately than individual parameters?

Few studies have addressed the predictive power of the clinical parameters used in assessing caries lesion activity. This study assessed the predictive validity of evaluating clinical parameters that are related to caries lesion activity status, individually and combined, in a long-term analysis. The occlusal surfaces of primary molars (1361 surfaces) were examined in 205 children according to the following clinical features: potential for plaque stagnation, colour, luster, cavitation, texture, and clinical depth. Cavities with frankly exposed dentine were excluded from this sample. After 1 year, 148 children (828 surfaces) were re-evaluated using the International Caries Detection and Assessment System to assess caries lesion progression. Progression was set as an outcome to verify the predictive power of the initially assessed clinical parameters. Different combinations of two or more parameters were also tested to check for any association with caries progression. Multilevel Poisson regression analyses were performed and the relative risk for each parameter/combination tested was calculated by considering a confidence interval of 95%. Forty percent of the reassessed surfaces presented caries progression after 1 year. Despite their surface integrity, dentine caries lesions were approximately 10-fold more likely and enamel lesions were approximately three-fold more likely to progress than sound surfaces. Similarly, cavitated lesions showed the highest risk of progression compared to sound/non-cavitated lesions. When only non-cavitated surfaces were considered, roughness proved to be a risk factor for caries progression. In conclusion, the lesions presenting clinical involvement of the dentine and even those cavitations clinically involving only the enamel had a higher risk of progression compared to sound or non-cavitated surfaces. For these lesions, the evaluation of other conjoint parameters seems unnecessary. Nevertheless, surface roughness can be a useful feature in predicting the risk of non-cavitated caries lesion progression.

117 surface would be included per child. A minimum sample size of 126 surfaces was 118 calculated and this number was increased by 20% to compensate for parent or 119 children's refusal to participate and for possible dropouts. Hence, a sample of 151 120 surfaces was required. Because more than one occlusal surface could be included in 121 the sample, to compensate for the clustering effect, we assumed a factor of correction 122 of 1.4. We determined that at least 212 surfaces were needed for our sample.
123 Clinical Examination at Baseline 124 The children were examined in a dental unit and the examiners used a plane 125 dental mirror, a ball-ended probe, and a three-in-one syringe. Before examination, 126 teeth were gently cleaned with a rotating-bristle brush and pumice/water slurry.

127
In the baseline, an examiner experienced in caries diagnostic research (MMB) 5 129 each surface by an external researcher on the basis of the highest ICDAS score found 130 on the respective surface. Sites were recorded using a specific illustration in the 131 participant's file to guide next stages.

132
Sites were classified according to clinical characteristics of caries lesions that are 133 generally understood to be associated with caries lesion activity status: potential of 134 plaque stagnation on the basis of morphology and position of the surface, colour, 135 lustre, surface integrity, depth, and texture [2]. The examiner did not use any specific 136 system, but classified the sites as described in a previous report [5].
137 Clinical Examination at the 1-year Follow-up 138 One year after the first examination, an examiner (IF), different from that 139 responsible for baseline assessments, reassessed the children using the ICDAS [12].
140 In addition, restorations and teeth that had been extracted (because of caries) were 141 also recorded. The clinical examination was performed under the aforementioned 142 conditions. For this evaluation, the examiner followed pre-signaled charts with the 143 occlusal sites evaluated at the baseline.

145
Caries progression was set as the outcome in further analysis and the evaluated 146 sites were dichotomized into those that presented caries progression and those that did 147 not. Each of the clinical features related to the activity status of caries lesions (potential 148 for plaque stagnation, colour, lustre, surface integrity, texture, and lesion depth) was 149 tested as an independent variable.

150
To evaluate caries progression, the baseline and follow-up assessment results 151 were compared. Caries progression was considered when cavities with dentine 152 exposure and/or teeth were restored or extracted because of caries as progression.
153 Progression not related to cavitation exposing the dentine (e.g., ICDAS score 1-to 154 ICDAS 2) was not considered for analyses. To verify the association of caries 155 progression with the independent variables (clinical features), multilevel Poisson 156 analyses were performed, considering the tooth and the child as the levels. Univariate 157 analyses were performed both for the full sample and for non-cavitated lesions. For 158 these analyses, we alternatively considered caries progression excluding cases that 159 were restored after 1 year.
6 162 the effect of another related variable in a multiple model if they are strongly associated 163 with each other, despite being equally crucial individually in explaining the outcome.
164 However, the interactions among some of the variables was tested to evaluate the 165 possible benefit of combining these variables in assessing caries lesion activity.

166
Besides, to simulate the use of systems that combine these characteristics, we 167 created other 2 independent variables regarding lesions activity status. First, we 168 assumed that an initial or established active lesion would be whitish/yellowish, with no 169 lustre and with rough enamel. If a lesion did not present these three features at the 170 same time, it was classified as inactive. Second, we assumed that an active lesion 171 should present at least two of the three aforementioned clinical features.

172
The relative risk for the clinical features (alone or combined with one or more 173 other clinical features) was calculated, with 95% confidence interval (95% CI). The 174 level of significance was set at 5%.

192
The children reassessed after 1 year had similar caries experience on the basis 7 195 DMFS (SD) = 6.3 (6.7); p = 0.57). The number of reexamined caries-active children 196 was also similar to those who were not followed-up (p = 0.90).

197
The status of examined surfaces at the baseline and at follow-up is shown in 198 828 0 -6 = ICDAS scores; E = indicated for extraction / extracted tooth due to caries; R = restored surface. * lesions were clinically classified into dentine only if a shadow was observed under enamel (even without dentine exposure) -cavities exposing dentine were not considered in these analyses. The highlighted columns correspond to the surfaces on which were considered progression.

202
Dentine lesions had a probability of progression approximately 10 times higher 203 than that of sound surfaces. Conversely, enamel lesions were only three times more 204 prone to progression compared with sound sites (Table 2). Similarly, cavitated lesions 205 were six times more closely associated with caries progression than non-cavitated 206 surfaces. When only non-cavitated lesions were considered, dentine lesions (shadows) 207 were strongly associated with lesion progression (Table 2). 8 210 alteration in colour. Among the non-cavitated lesions, darker surfaces had a higher 211 probability of progression compared to non-stained sites (Table 2).

212
Lesions without lustre progressed twice as high than those lesions with lustre.
213 The same phenomenon was observed for surfaces presenting rough enamel compared 214 to smooth enamel. However, for the non-cavitated lesions, only texture was a risk 215 factor for caries progression (Table 2).
216 * with dentine exposure (ICDAS scores 5 and 6) ** lesions were clinically classified into dentine only if a shadow was observed under enamel (even without dentine exposure) -cavities exposing dentine were not considered in these analyses.
Similar trends were observed when restorations were not included in the outcome 222 (Table 3). However, when only non-cavitated lesions were considered under these 223 conditions, no associations were observed (Table 3).

225
226 3.60 (0.24 -53.75) * with dentine exposure (ICDAS scores 5 and 6) ** lesions were clinically classified into dentine only if a shadow was observed under enamel (even without dentine exposure) -cavities exposing dentine were not considered in these analyses.
233 for active status (Table 4). However, for non-cavitated lesions, when at least two 234 clinical features were positive for active status, the sites had a two-fold higher risk for 235 progression compared to sound surfaces (Table 4).

236
Cavitated lesions showed higher risk of progression regardless of texture or 237 lustre (Table 4). Different from smooth surfaces, rough whitish lesions were more prone 238 to progression than sound sites ( Table 4). The texture evaluation of black/brownish 239 caries lesions also seemed to improve the prediction of caries progression.
240 Furthermore, the magnitude of the association with caries progression increased when 241 texture and color were combined (Table 4). Among non-cavitated lesions, only rough 242 black/brownish samples were associated with caries progression (Table 4).
243 Additionally, no combination between luster and texture was associated with caries 244 lesion progression. 246

256
This study aimed to clarify the predictive power of these clinical characteristics of 257 active lesions for caries progression. Hence, we used a sample selected from children 258 who had sought dental treatment. Because our sample was calculated a priori, we 262 Conversely, the age group included may reflect a greater likelihood of seeking 263 treatment, being representative of the population that we aimed to study. In addition, 264 even considering this limitation, we obtained statistical power for demonstrating some 265 crucial associations in our findings.

266
Similar trends were observed both when restorations were and were not 267 considered as caries progression. Patients were followed-up but not treated by the 268 researchers. Therefore, it is reasonable to consider restoration as progression even if 269 the tooth had been restored by another professional before the established follow-up.

326
Clinically, especially in occlusal surfaces, the reversion of lesion status could be 327 slower and less evident than in aforementioned conditions. In addition, differences in 328 enamel porosity may impede the differentiation of caries lesions and other enamel 329 defects [24]. We should also consider that changes in the clinical appearance of caries 330 lesions may have been due to professional cleaning prior to examination. We believe, 331 however, that the effect of this procedure would have been low because we assessed 332 occlusal surfaces. Accordingly, to predict caries lesion progression, it seems accurate 333 and simpler to assess only the texture of non-cavitated occlusal caries lesions, instead 334 of assessing both parameters (texture and luster) together.

344
Because clinical features tend to reflect caries lesion activity at the moment of the 345 clinical examination, a static time point, the importance of evaluating some features 346 conjointly is paramount. However, our findings suggest that some of these parameters 347 could be more helpful in this task than others, which could simplify lesion activity 348 assessment.

349
In conclusion, caries with clinical involvement of the dentine as well as cavitated 350 caries lesions (even if, with clinical involvement of only the enamel) had a higher risk of 351 progression compared to sound or non-cavitated surfaces. Thus, evaluating other 352 conjoint parameters seems unnecessary. However, superficial roughness can be a