1. | Model horizon | Lifetime | One year | Lifetime |
2. | Duration of treatment |
| Continuous | Continuous or intermittent use |
3. | Extrapolation beyond trial duration | Continued effect size while receiving treatment adjusted by withdrawal rates from observational studies | Estimates of benefit based on trial data; estimates of withdrawal and long term outcomes based on observational data | Continued effect size whilst receiving treatment adjusted by withdrawal rates from observational studies |
4. | Modelling beyond trial duration | Linear decline of effect size after treatment is continued | No benefit or harm if treatment is stopped |
5. | Synthesis of comparisons where clinical trials do not exist | Not recommended owing to uncertain validity of transitive comparisons |
6. | Outcome measures | Clinical fractures |
| OARSI—20% improvement |
| | | | Clinical adverse events |
7. | Mortality | Hazard for mortality from observational studies | Incorporate additional hazard attributable to drug based on observational data |
8. | Valuation of health (e.g. QALY) | Values from general public for policy makers; values from patients for clinicians | Values from general population using direct measurement | Values from general public for policymakers; values from patients for clinicians |
9. | Resource use | Include all associated direct medical costs in the analysis, but report indirect and non-medical costs separately |
10. | Classification and reporting of adverse events | Report adverse events with patients as the unit of analysis using common toxicity criteria (under development by OMERACT Toxicity Working Group, Woodworth) |
11. | Discontinuation of treatment | Use discontinuation rates from observational studies |
12. | Therapeutic strategies | Include modelling of most commonly used therapeutic strategy with sensitivity analysis to consider other strategies |
13. | Population risk stratification | Include clear definition of underlying population including low and high risk groups |