A patient-important outcome is one the patient can feel or that changes how long or how well they live: death, stroke, heart attack, hospitalization, pain, or loss of function. A surrogate outcome is a laboratory measurement, imaging finding, or physiological marker used as a stand-in for a patient-important outcome, on the assumption that changing the marker changes the outcome. Common surrogates include blood pressure, LDL cholesterol, blood glucose, tumor size on imaging, and viral load.
The assumption is not always true. A treatment can improve a surrogate while leaving the patient-important outcome unchanged or even worse. The classic illustration is a drug that lowers a laboratory marker through a mechanism that also causes harm, so the marker improves while mortality rises. When a trial reports benefit only on a surrogate, the honest reading is that the effect on the outcome the patient cares about is unproven, and confidence in applicability should be lower than if the trial had measured that outcome directly.