Study selection
A total of 995 citations were initially identified after an electronic database search (994 articles) and a manual search (1 article). The selection process is presented as a flow diagram (Fig. 1). Ultimately, 21 articles of the 16 latest studies were included (Tables 1a-b and Table 2), of which 9 articles were related to 4 different series of studies and 1 article compared infiltration and sealing to the control group individually [19-23, 33-48].
Characteristics of the included studies
The data of the included studies are summarized in Tables 1 and 2. All the studies were split-mouth RCTs. A total of 774 patients (ranging from 4.6 to 41 years) were enrolled in 16 clinical trials. There were 1994 non-cavitated proximal lesions in the trials. A total of 5 studies were included that assessed lesions in primary dentition [23, 33, 36-39], and 11 studies assessed lesions in permanent dentition [19-22, 34, 35, 40-48]. The interventions included resin infiltration (14 articles) [33-39, 41-47] and sealant (8 articles) [19-23, 40, 41, 48]. The follow-up duration ranged from 12-84 months. In terms of caries risk levels, 2 studies reported high risk [21, 46, 47], 4 studies reported moderate to high risk [33, 34, 37, 39, 40], 1 study reported low to moderate risk [38], 2 studies reported low to high risk [36, 42], 3 studies reported mixed risk levels [23, 41, 43-45] and 4 studies did not report caries risk in the articles [19, 20, 22, 35, 48]. Four caries risk statuses were included in the subgroup analysis: low [42], low to moderate [38], moderate to high [33, 34, 37, 39, 40] and high [21, 42, 46, 47]. All trials used radiographic lesion progression as the primary outcome. Methods for evaluating lesion progression included independent reading of radiographs, pairwise reading of radiographs and DSR. For data analysis, the most sensitive outcome was recorded if two or more evaluation methods were used in a study (outcomes obtained by DSR>pairwise reading>independent reading).
Risk of bias within studies
The risk of bias within studies was summarized in Fig. 2 and Fig. 3. All of the studies had a low risk of random sequence generation and blind outcome assessment. For allocation concealment, 9 studies had an unclear risk for insufficient information on the methods of concealment [19-23, 33, 36, 39-41, 48]. Except for 6 studies with mock treatment of unclear risk [35, 38, 41-47], the remaining 9 studies were ranked as high risk because of the lack of blinding for patients and personnel. Eight studies with drop-out rates greater than 25% were regarded as having unclear attrition bias [19-21, 23, 34, 36, 39, 44, 47]. A total of 2 studies had selective reporting of data [37, 41]. Except for 1 study with no caries risk assessment at baseline [22], 1 study with unbalanced lesion allocation [37] 1 study with unbalanced lesion severity [40], and 1 study with no reporting on Score 2 [48], the other 11 studies showed a low risk of other biases.
Meta-regression analysis
The meta-regression analysis results revealed that different research durations (ranging from 12-84 months) did not influence caries progression (P>|t|: 0.997, 95% CI: -0.020 to 0.020). Thus, we chose caries progression at the longest follow-up times for continuous RCTs, similar to previous reviews [3, 24, 26].
Efficacy of infiltration and sealing for non-cavitated proximal caries
Sixteen RCTs were enrolled to assess the efficacy of infiltration and sealing for non-cavitated proximal caries. A fixed-effects model was used for the analysis, as there was no significant heterogeneity between studies (I2 = -25.78%, Fig. 4). The overall intervention effects of infiltration and sealing were significantly different from the intervention effects of control treatment (OR = 0.23, 95% CI: 0.17 to 0.29). We analysed the two different measures (infiltration and sealing) using subgroup analysis, and we found that both invention measures reduced the odds of lesion progression compared with the control group (infiltration vs non-invasive treatments: OR = 0.20, 95% CI: 0.15 to 0.29; sealing vs placebo: OR = 0.27, 95% CI: 0.18 to 0.41).
Sixteen RCTs were related to infiltration and sealing of primary dentition or permanent dentition. There was no significant heterogeneity of the included RCTs (I2 = -25.78%, Fig. 5). Non-cavitated proximal lesions were reduced when measures were taken in primary dentition and permanent dentition (primary dentition: OR = 0.30, 95% CI: 0.20 to 0.45; permanent dentition: OR = 0.19, 95% CI: 0.13 to 0.27, Fig. 5).
Eight RCTs were analysed for the efficacy of infiltration and sealing at different caries risk levels (Tables 1a-b). There was no significant heterogeneity among the eight RCTs (I2 = -12.04%, Fig. 6). The overall effects of infiltration and sealing were significantly different from the overall effects of control treatment (OR = 0.19, 95% CI: 0.14 to 0.27). For patients with different caries risk levels, there were significant differences between micro-invasive treatments and non-invasive treatments (low risk: OR = 0.23, 95% CI: 0.07 to 0.75; low to moderate risk: OR=0.38, 95% CI: 0.18 to 0.81; moderate to high risk: OR=0.17, 95% CI: 0.10 to 0.29; and high risk: OR=0.14, 95% CI: 0.07 to 0.26). Six RCTs were related to infiltration at different caries risk levels. There was no significant heterogeneity among the seven RCTs (I2 = -12.06%, Fig. 7). Significant differences in the progression rate were found among patients who were treated with infiltration and non-invasive treatments (low risk: OR=0.07, 95% CI: 0.02 to 0.22; low to moderate risk: OR=0.38, 95% CI: 0.18 to 0.81; moderate to high risk: OR=0.19, 95% CI: 0.09 to 0.38; and high risk: OR=0.14, 95% CI: 0.07 to 0.29). Two RCTs were related to sealing across different caries risk levels. Due to insufficient patient information in terms of caries risk levels in the sealing group, no subgroup analysis was conducted.
Publication bias
Begg’s test and Egger’s test results indicated that the included studies had no obvious publication bias (Begg’s test: Z = -0.12, P > 0.05; Egger’s test: Z = 0.45, P > 0.05), as shown by the funnel plot in Fig. 8.
Quality of evidence
Based on this study, infiltration or sealing arrested progression in 287 lesions per 1000 treated lesions. Infiltration arrested progression in 286 lesions per 1000 treated lesions. Sealing arrested progression in 288 lesions per 1000 treated lesions. All evidence was graded as moderate (Appendix 2).