- Published
- Updated
In observational research, a comparator considered at a similar clinical decision point can provide more meaningful context than people on very different care pathways. An active comparator is not, however, an automatic solution that eliminates all confounding or makes selection equivalent to randomisation.
Define the question and assumptions
Examine with clinical specialists when each option is chosen and for whom it is considered appropriate. The same indication label can hide differences in severity, previous failures or access. Eligibility, time origin and covariate measurement should connect both groups to the same underlying scientific question.
Hypothetical study example
In a hypothetical study, two interventions treat the same condition, but one is usually selected after earlier options fail. The team does not match solely on diagnosis codes. It evaluates treatment line and selection reasons, explicitly acknowledging residual confounding where those factors cannot be adequately measured in the available source.
Design review checks
- Map each option within the care pathway and its practical eligibility conditions.
- Assess the adequacy of prior-treatment and disease-severity information before comparison.
- Justify comparator selection before inspecting which choice produces a favourable difference.
Record the decision and its limits
An accepted design explains why this group was chosen and to whom inference applies. Statistical balance does not prove identical clinical decision processes. Carry unmeasured selection reasons into sensitivity assessment and reporting limitations. Preserve the observational nature of the research rather than presenting comparator selection as the equivalent of random allocation.
Frequently asked questions
- Does an active comparator create randomisation?
- No. Clinical selection and unmeasured reasons can still influence outcomes.
- Is sharing an indication sufficient?
- No. Treatment line, severity and access conditions may still differ.