AbstractAimSelective root‐canal retreatment has been proposed as an alternative to full retreatment in multirooted, root‐canal‐filled teeth with evidence of apical pathology, where only the affected root(s) is retreated. Whilst this option may save costs initially, failures and retreatments may compensate for these initial savings. We assessed the cost‐effectiveness of full versus selective root‐canal retreatment using data from a recent clinical pilot study, employing a modelling approach.MethodologyA Markov model was constructed to follow up a previously root‐canal treated maxillary molar with apical pathology on a single root (mesio‐buccal), receiving either selective or full root‐canal retreatment. A private‐payer perspective in Dutch health care was adopted. Permanent molar teeth with apical lesions on the mesial root were simulated over the lifetime of initially 50‐year‐old patients. Teeth could have endodontic complications and require interventions such as retreatment or tooth extraction and replacement. Costs were calculated based on the Dutch dental fee catalogues. Monte‐Carlo microsimulations were performed to assess lifetime costs and effectiveness (measured as tooth retention time), and the resulting cost‐effectiveness. Probabilistic joint uncertainty and sensitivity analyses were performed, and cost‐effectiveness at different willingness‐to‐pay‐thresholds was evaluated.ResultsIn the base‐case scenario, selective retreatment was less costly (2137; 2.5%–97.5% percentiles: 1944–2340 Euro) and more effective (19.6; 18.3–20.8 Years) than full retreatment (2495; 2305–2671 Euro; 16.5; 15.2–17.9 Years) in 100% of the simulations and regardless of the willingness‐to‐pay threshold. Only in a worst case scenario was selective retreatment more costly, but remained more effective.ConclusionsSelective retreatment, when clinically applicable, is likely to be more cost‐effective than full retreatment in endodontically treated molars with persistent apical periodontitis. Our results should be interpreted with caution because the quality of the underlying data is limited.