I teach doctors how to interpret test results – and the labels ‘normal’ and ‘abnormal’ often hide what matters most

A physician who studies diagnostic test interpretation describes a patient whose BNP level, used to screen for heart failure, came back at 14, technically 'normal' since anything under 100 is labeled that way. A treating resident dismissed the low result as consistent with heart failure in overweight patients and ordered a diuretic, standard treatment, but a supervising colleague's ultrasound found the real diagnosis: cardiac tamponade, a fluid buildup around the heart that needed more blood flow, not less. The patient survived because doctors caught the error and drained the fluid in time.

The author argues this reflects a broader problem: labs sort results into binary 'normal' and 'abnormal' categories that collapse meaningfully different values together, citing D-dimer and cholesterol thresholds where numbers just inside or outside the cutoff carry different implications than the label suggests. The piece describes research into 'interval likelihood ratios,' tested with over 600 clinicians across four academic institutions, as an alternative way to convey how strongly a result points toward or away from a diagnosis.

The full dispatch is available from the source below.