Table 1. Transition probabilities (p) used within the model.
| State | Transition probability (p) per cycle | Transition/allocation to | Probability |
| Non-invasive E2 (I) | p = 3.0984×(2α)−1.343 (95% CI: p×0.87– p×1.13) | D1 | 1.00 |
| Non-invasive D1 (I) | p = 1.652×(2α)−2.078 (95% CI: p×0.87– p×1.13) | Composite | 1.00 |
| Infiltrated E2 (I) | p = 0.4289×(2α)−1.391 (95% CI: p×0.23– p×5.15) | Infiltrated D1 | 1.00 |
| Infiltrated D1 (I) | p = 68.869×(2α)−2.078 (95% CI: p×0.23– p×4.17) | Composite | 1.00 |
| Composite (I/F) | range p = 0.011–0.019 | Composite | 0.45 |
| Crown | 0.10 | ||
| Repair | 0.10 | ||
| RCT | 0.25 | ||
| Extraction | 0.10 | ||
| Direct capping (F) | range p = 0.008–0.168 | RCT | 0.95 |
| Extraction | 0.05 | ||
| Crown on vital tooth (F) | range p = 0.019–0.041 | RCT | 0.25 |
| Recementation | 0.15 | ||
| Repair | 0.10 | ||
| Re- crown | 0.40 | ||
| Extraction | 0.10 | ||
| Root canal treatment (F) | range p = 0.014–0.022 | Non-surgical re-treatment | 0.20 |
| Surgical re-treatment | 0.30 | ||
| Extraction | 0.50 | ||
| Crown on non-vital tooth (F) | range p = 0.015–0.328 | Recementation | 0.20 |
| Repair | 0.10 | ||
| Re- crown | 0.60 | ||
| Extraction | 0.10 | ||
| Non-surgical (F) | range p = 0.013–0.117 | Surgical re-treatment | 0.25 |
| Extraction | 0.75 | ||
| Surgical (F) | range p = 0.015–0.065 | Extraction | 1.00 |
| Implant (F) | range p = 0.001–0.015 | Recementation/Refixing | 0.60 |
| Re-crown | 0.20 | ||
| Re-implant | 0.20 |
Teeth were allocated to their initial health state (I) depending on the treatment strategy and the lesion stage (left column). For non- and micro-invasively treated lesions, transition probabilities per 6-monthly cycle depended on patient’s age (α) and were calculated using hazard functions (middle column). For all follow-up states (F), transition probabilities depended on the time spent in the health state (e.g. the time since a crown had been placed), with three time plateaus being modelled (<2, 2–5, >5 years). To introduce joint parameter uncertainty, a triangular distribution of parameters between their 95% Confidence Intervals (CI) was assumed. For hazard functions, 95% CI (given in brackets) were used within scenario analyses. To simplify the table, we only present the range of follow-up transition probabilities used within the model. Full details (time-dependent mean and 95% CI probabilities) can be found within the Supporting Information. If transition occurred, teeth were allocated to follow-up states according to allocation probabilities (right columns).