Case Series & Case Reports

Purely descriptive accounts of the clinical course of a single patient (case report) or several similar patients (case series). There is no comparator and no denominator, so they sit at the base of the evidence pyramid. Their job is to generate hypotheses, not to test them.

Where descriptive reports sit relative to comparative designs. They are the broadest, weakest tier for proving cause and effect — one step above expert opinion — precisely because there is no control group to say what would have happened anyway.

Read the position as a warning about internal validity, not value. A case report describes one patient; a case series collects several with a shared feature. Neither has a denominator or comparator, so you cannot calculate risk, rates or any effect measure from them — only describe.

Despite the low tier, they are invaluable: they describe rare conditions and novel presentations, teach pattern recognition, and provide the first signal of a new disease or a serious adverse drug effect — signals that later, stronger studies are built to confirm or refute.

• Case report = 1 patient; case series = several
• No comparator, no denominator → hypothesis-generating
• Best for rare conditions & first adverse-effect signals

Thalidomide — McBride, Lancet (16 Dec 1961) — a one-paragraph letter reporting a cluster of newborns with the same rare limb-reduction defect (phocomelia) whose mothers had taken thalidomide in pregnancy. This handful of uncontrolled observations was enough to raise the alarm and drive the drug’s withdrawal — a textbook example of a case series delivering the first signal of a catastrophic adverse effect. The series flagged the danger; it could not prove it — causation was established afterwards by systematic epidemiology. A case series sounds the alarm, it does not return the verdict.

• Treating a case series as efficacy evidence — with no comparator it can never show a treatment works.
• Confusing a case series (descriptive, no controls) with a case-control study (analytical, has controls).
• Ignoring selection & publication bias — striking cases get written up; the unremarkable ones never do.