Data sources: As well as using Medline, Embase and the Cochrane library to source studies, a search was made by hand of reference lists of the selected articles and dental conference proceedings, and experts were contacted to identify any unpublished research, irrespective of language in each case.
Study selection: Two reviewers independently selected the studies. The latter were chosen if randomised, quasi-randomised or controlled clinical trials that used split-mouth design or were parallel group studies, and which compared different available resin-based sealants (RBS) on permanent molars or compared different clinical protocols for the same RBS. Studies comparing RBS with ultraviolet-activated sealants, glass ionomer cements (GIC) or resin-reinforced GIC sealants were not considered and neither were resins indicated for restoration, except for flowable resins.
Data extraction and synthesis: Data was extracted by two reviewers independently. Data synthesis was conducted using the program RevMan (version 4.2; Cochrane Collaboration, Dublin, Ireland). Relative risks (RR) were calculated because the outcome was a dichotomous variable; variance was expressed using 95% confidence intervals (CI). Meta-analysis was carried out if the studies were homogeneous enough and subgroup analyses were conducted if necessary. The heterogeneity of the results was assessed using chi-squared tests, and sensitivity analyses were performed excluding unpublished studies or poor-quality studies.
Results: A total of 31 studies met the inclusion criteria. Sixteen studies compared two different types of RBS, those that were light-cured (LRBS) or auto-polymerised (ARBS), with fluoride-containing sealants (FRBS). Seventy-five per cent were split-mouth design studies. Seven studies compared LRBS with ARBS and nine studies compared LRBS with FRBS. The retention rate of ARBS and LRBS did not differ significantly. LRBS had a significantly higher retention rate than FRBS at 48 months (RR, 0.80; 95% CI, 0.72-0.89). Only 15 studies of limited quality focused on the best clinical protocol to adopt for sealant application.
Conclusions: Only FRBS had a lower retention rate than ARBS or LRBS. This review did not allow the best clinical procedure to be determined because of insufficient studies. Well-designed randomised clinical trials are still needed that focus on sealant retention and consider different clinical procedures, particularly new enamel preparation techniques such as air-abrasion or sono-abrasion.