Table III. —Components of protocol using the target trial emulation framework^{5, 6, 20, 21} applied to estimating the effect of a type of rehabilitation unit on the risk of fracture in older adults with stroke.

#	Protocol component	Specification of target trial	Emulation of target trial using observational data
1	Eligibility criteria	• Inclusion criteria: Older adults (aged ≥65 years) with a diagnosis of stroke or transient ischemic attack • Exclusion criteria: Adults with subarachnoid hemorrhage or with preexisting hemiparesis (for those with transient ischemic attack)	• Same as specification • Required in observational data: age, diagnosis of stroke or transient ischemic attack, history of subarachnoid hemorrhage, pre-existing hemiparesis
2	Treatment strategies	• Dedicated units for inpatient stroke rehabilitation (units with geographically distinct, stroke-dedicated beds and dedicated therapists).22 This represents one model of service delivery in Ontario • Non-dedicated units for inpatient stroke rehabilitation, representing a different model of service delivery	• Same as specification • Required in observational data: dedicated versus non-dedicated units for inpatient stroke rehabilitation
3	Assignment procedures	• Participants will be randomly assigned to dedicated units of inpatient stroke rehabilitation or non-dedicated units of inpatient stroke rehabilitation at baseline • Participants will be aware of treatment assignment (cannot be blinded due to the nature of the intervention)	• Different from specification (i.e., no randomization) • Eligible persons will be assigned to dedicated versus non-dedicated units of inpatient stroke rehabilitation based on what was documented in observational data at the time of eligibility
4	Follow-up period	• Begins at randomization (baseline) • Ends at fracture, loss to follow-up, death, or 2 years after baseline	• Same as specification • Required in observational data: date of fracture, date of loss to follow-up, date of death
5	Outcome	• Low-trauma fracture that occurred within 2 years of index stroke, defined as any fracture of the femur, forearm, humerus, pelvis, or vertebrae and excludes fractures from trauma, traffic collisions, falls from a height, or in people with active cancer6	• Same as specification • Required in the observational data source: Fracture, including type of fracture, cancer data
6	Causal contrasts of interest	• Intention-to-treat effect • Per-protocol effect	• Observational analogue of intention-to-treat effect
7	Analysis plan	• Comparison of fracture risk among participants assigned to dedicated units versus non-dedicated units of inpatient stroke rehabilitation • Intention-to-treat or per-protocol analysis • All analyses adjust for baseline confounders/prognostic factors	• Analyses adjusting for baseline confounders/prognostic factors • Required in observational data: All baseline confounders/prognostic factors