Table 1 Outline of the target trial protocol: specification and emulation using RCT data with interventions
Protocol components | Target trial | RCT data with interventions |
|---|---|---|
Specification | Emulation [Required data] | |
Aim | To estimate the effect, if any, of reductions in received dose intensity on event-free survival across levels of histological response in patients with osteosarcoma aged 40 years or less. | Same as for TT specification. |
Eligibility criteria | Patients aged 40 years or less with a histologically confirmed diagnosis of high-grade nonmetastatic osteosarcoma in an extremity long bone.a Patients need to commence chemotherapy within 28 days after biopsy, with normal leukocyte (\(\ge 3.5\times 10^9/L\)) and platelet (\(\ge 100\times 10^9/L\)) counts. | Same as for TT specification. [Age, diagnosis type, start date of therapy, leukocyte count, platelet count] |
Treatment strategies | Chemotherapy dose intensity regimens: standard, reduced, and highly-reduced six-cycle EOI treatment.b | Same as for TT specification. [Received dose, cycle timing] |
Treatment assignments | Eligible persons will be randomly assigned to one strategy and will be aware of which strategy they were assigned to. | Eligible persons assigned to the strategy based on observed dose-intensity behavior at the end of the therapy. |
Outcomes | Death, local recurrence, evidence of new or progressive metastatic disease, second malignancy, or a combination of those events. | Same as for TT specification. [Date of death, local recurrence, or other malignancies] |
Follow-up | Start: treatment assignment. End: death, local recurrence, evidence of new/progressive metastatic disease or second malignancy, or censoring. | Same as for TT specification, except start is the end of therapy. [Date of loss to follow-up or censoring] |
Causal estimand | Intention-to-treat conditional effects (effect of being assigned to a reduced treatment across subgroups). Per-protocol conditional effect (effect of receiving a reduced treatment as indicated in the protocol across subgroups). | Observational analogue of per-protocol conditional effect across subgroup. |
Statistical analysis | Intention-to-treat analysis via landmark Cox model [28,29,30] to estimate effects of reduced exposures across levels of histological response.c | Randomization will be emulated via adjustment for pre-assignment and post-assignment confounders by inverse probability of treatment weighting. |
Conditional average effect measured as contrast of restricted mean survival times over follow-up. | Per-protocol analysis analogous to TT specification. [Pre-assignement confounders, post-assignment confounders, histological response] |