Biostatistics & Research · patient case question
Which design best controls for both known and UNKNOWN confounding variables?
Patient case: Patient who refuses whitening because of a cancer claim
Patient: Female, 38 years old
Chief Complaint: Interested in tooth whitening but has read online that it causes cancer.
Background: She raises the concern at consultation. The clinician wishes to find the best available evidence on whether vital tooth whitening is associated with oral cancer risk, in order to counsel her properly.
Current Findings: No relevant medical history. Generalised extrinsic staining. Sound dentition.
Show answer and explanation
Correct answer: B. A randomised controlled trial
- A. Case-control studies are vulnerable to recall bias and selection bias, and can only adjust for confounders that were measured.
- B. CORRECT. Randomisation is the only method that distributes UNKNOWN and unmeasured confounders evenly between groups, because allocation is independent of every patient characteristic — including those nobody has thought to measure. Statistical adjustment in observational studies can only handle confounders that were identified and recorded.
- C. A cross-sectional survey cannot establish temporality at all, let alone control unmeasured confounding.
- D. A case series has no comparison group and therefore no control of confounding whatsoever.
Note: The evidence hierarchy, ascending: case report → case series → cross-sectional → case-control → cohort → randomised controlled trial → systematic review and meta-analysis.
Practice Biostatistics & Research questions Biostatistics & Research study notes