Figure 1

Used Markov-model. We followed posterior teeth in a male patient initially aged 18 years over his lifetime. The prevalence of caries lesions was assumed to differ between risk groups. The probability of a sound surface developing a shallow dentinal lesion was determined by p_develope. Depending on the patient’s utilization of dental services (p_ utilization), the patient attended for a dental checkup, where the dentists detected and invasively treated the lesion with a certain probability (p_detection). Treatment at this stage resulted in a shallow occlusal composite restoration, which failed according to its transition probability (p_fail_composite), resulting either in repair or refill according to the respective allocation probability. If failing a second time, re-treatment was assumed. Progression of a shallow lesion (according to p_progress) was assumed to lead to a deep dentinal lesion, which was subsequently treated by one of three caries excavation strategies. Transition probabilities in follow-up stages were modelled as described elsewhere [15]. Bold variables were found to differ according to an individual’s risk group, and were used to separately model low- and high-risk patients.