Abstract
Non-carious cervical lesions (NCCLs) present restorative challenges due to substrate sclerosis and marginal stress concentration, making the adhesive strategy a key determinant of long-term performance. This systematic review and frequentist random-effects network meta-analysis, conducted in accordance with PRISMA 2020 and the PRISMA-NMA extension and prospectively registered in PROSPERO, compared restorative strategies defined by etching approach, adhesive category, and restorative material, with marginal adaptation and retention loss as the primary outcomes. PubMed, Web of Science, Cochrane Library, Embase, and Scopus were searched without restrictions (25 January 2026), supplemented by alternative retrieval methods. Randomized clinical trials in permanent teeth evaluating at least two etching-based strategies (etch-and-rinse, self-etch, selective-etch, and/or resin-modified glass ionomer cement (RMGI)) were included. Risk of bias was assessed using RoB 2 and certainty of evidence with CINeMA. Seventy-four trials were eligible. Connected networks were established for marginal adaptation (57 studies; 28 interventions; 6798 patients; 1772 events) and retention loss (61 studies; 33 interventions; 7338 patients; 584 events). Selective-etch with a universal adhesive and nanocomposite reduced marginal adaptation failure compared with RMGI, whereas certain self-etch and non-universal combinations increased risk. For retention loss, selective-etch and etch-and-rinse protocols combined with universal adhesives and nanocomposites showed lower failure rates, while some self-etch or non-universal adhesive strategies performed less favorably. Overall confidence was predominantly high, with downgrading mainly due to imprecision and heterogeneity. Strategies incorporating selective enamel etching or etch-and-rinse approaches combined with universal adhesives and nanocomposites demonstrated the most consistent clinical advantages.