Table 1.

Utilization of the estimands framework to address intercurrent events (such as COVID-19 diagnosis in a trial patient)

Strategy name	Approach	Limitations
Treatment policy strategy	The trial is largely conducted on the “intention-to-treat” principle.	The outcome must be acquired irrespective of IE (e.g., Mortality would be an acceptable outcome, given that it can be determined in patients with or without COVID-19. However, re-stenosis following angioplasty/stenting at 7 days may not be feasible in some critically ill COVID-19 patients who are too unstable to undergo repeat imaging).
Hypothetical strategy	Investigator imputes outcome data based on absence of IE for a given patient.	The outcome(s) may not accurately reflect a response or association with the exposure due to confounding by the IE (e.g., Intracranial hemorrhage risk can be estimated for a patient, assuming they did not develop COVID-19).
Composite strategy	The IE can be woven into one (or more) other outcomes into a composite endpoint (e.g., development of COVID-19 or hemorrhagic transformation following recanalization).	This approach is less effective when the composite outcome includes events which do not occur naturally, and when the IE is not anticipated to influence the outcome of interest (e.g., delayed extreme hypertension [systolic blood pressure >180 mm Hg] 90 days after stroke and/or COVID-19).
	This is most useful when an IE can affect the outcome of interest, and a pre-specified outcome can be selected knowing that the IE can influence it (e.g., 90-Day functional dependence is influenced by COVID-19 status).
While on treatment strategy	The outcome variable should occur prior to the IE. This is advantageous when there are repeated measures over time (e.g., Serial NIHSS scores may be collected until the day the patient becomes symptomatic of COVID-19, at which point remaining events are censored).	This strategy is less effective when an IE may occur prior to the index or primary outcome event (e.g., Readmission within 30 days may be influenced by COVID-19 status).
Principal stratum strategy	This approach involves planned subgroup analysis of patients based on IE. This is particularly useful if the relationship between the exposure and outcome should exist within a particular population (e.g., pre-specification of a subgroup analysis based on patients who never develop COVID-19 during the study period).	Sample size estimates need to account for expected IE rate(s), which may require more funding or follow-up in clinical trials.

COVID-19, coronavirus disease 2019; IE, intercurrent event; NIHSS, National Institutes of Health Stroke Scale.