Surrogate vs Patient-Important Outcomes

A surrogate outcome is a lab value, image, or physiological measure (e.g. blood pressure, HbA1c, ejection fraction, ectopic-beat count) used as a stand-in for a patient-important outcome — death, disability, symptoms, or quality of life. Surrogates make trials faster and smaller, but only count if a change in the marker reliably tracks a change in the outcome that matters.

The causal chain a surrogate trial relies on: intervention → better marker → assumed better patient outcome. The branch makes the danger explicit — the same marker improvement can sit above either real benefit or real harm.

Trace the arrows downward. A trial that stops at “surrogate improves” has not shown the bottom row. Ask whether the surrogate has been validated — i.e. proven in earlier studies to lie on the causal pathway and to capture the treatment’s full effect on the patient outcome.

Surrogates routinely mislead. A drug can move the number in the “right” direction while doing nothing for — or actively harming — patients. In the ED and acute care, decisions should rest on outcomes patients care about, not on a reassuring lab trend that may not translate.

• Surrogate = marker (BP, HbA1c, ectopy, EF)
• Patient-important = death, function, symptoms, QoL
• A surrogate is trustworthy only once validated for that outcome

CAST — Cardiac Arrhythmia Suppression Trial (NEJM 1989/1991) — class I antiarrhythmics (encainide, flecainide) were given after myocardial infarction to suppress ventricular ectopy, a surrogate thought to predict sudden death. The drugs suppressed the ectopy — yet the encainide/flecainide arms had roughly 2.5× the mortality of placebo (total mortality ~7.7% vs ~3.0%), and the arms were stopped early. The surrogate moved the “right” way while patients died more. Suppressing a marker is not the same as helping the patient — the textbook case against unvalidated surrogates.

• Accepting an unvalidated surrogate as proof of patient benefit.
• Assuming a better marker means a better patient — the surrogate can improve while the hard outcome doesn’t (or worsens).
• Missing a surrogate hidden inside a composite endpoint, driving an apparently positive result.