- Published
- Updated
A historical control may reflect a different period of care from the current trial. Changes in diagnosis, supportive treatment and outcome documentation can make it inappropriate to attribute the entire observed difference to the investigational intervention.
Define the question and assumptions
Map care pathways, concomitant practices and outcome assessment for the compared periods. Calendar year alone does not measure care equivalence. Work with clinical specialists to identify affected participants and outcomes, explicitly recording uncertainty about changes that cannot be measured in the available historical records rather than assuming unchanged practice.
Hypothetical study example
In a hypothetical comparison, an older control period used a different diagnostic method for a complication. Current trial methods permit earlier and more frequent detection. The team assesses observation differences before comparing event rates; including calendar year in a statistical model is not treated as sufficient adjustment for the changed process.
Design review checks
- Document changes in standard care and measurement practices by comparison period.
- Distinguish changes in event detection from changes in underlying event frequency.
- Justify period selection before examining which choice gives a favourable difference.
Record the decision and its limits
Accepted assessment presents strengths and weaknesses together. More recent records may still be unsuitable because centre selection and patient profiles can differ. When uncertainty is substantial, narrow the claims that the comparison supports. A difference against historical records should not be described as conclusive causal evidence without addressing these potential alternative explanations.
Frequently asked questions
- Are recent records always a suitable control?
- No. Care, population and measurement processes still require assessment.
- Is adjustment for calendar year sufficient?
- Not alone. The clinical and recording mechanisms of change matter.