The Chart Doesn't Add Up

A 58-year-old with known type 2 diabetes on metformin and basal insulin has a laboratory panel drawn on a routine morning. The same-day fasting plasma glucose comes back at 158 mg/dL, no flag. A1c on the same specimen batch reads 6.4%. Run the standard conversion, estimated average glucose (eAG) equals 28.7 times A1c minus 46.7, and 6.4% maps to an eAG of about 137 mg/dL. That is 21 mg/dL below the fasting result drawn the same morning, and well below the two-week home fasting log the patient brought in, which runs 138 to 172 mg/dL with two afternoon readings over 220 mg/dL after missed insulin doses.

Nothing here points to an obviously wrong result. The glucose method has no flag. The A1c analyzer is NGSP-certified and its current interference listing shows no effect for this patient's recorded hemoglobin phenotype. Both numbers can be analytically correct and still describe two different pictures of the same patient, because they are not measuring the same thing over the same span of time.

This patient also has chronic kidney disease (CKD) stage G4, an estimated glomerular filtration rate near 22 mL/min/1.73m^2, anemia with a hemoglobin of 9.6 g/dL, and is on erythropoiesis-stimulating agent (ESA) therapy. Every one of those facts belongs in the discordance workup before anyone calls a laboratory error or a clinical error.

Reconciling short-term glucose data, A1c, and the conditions that make the two disagree calls for a laboratory response that is accurate, bounded, and useful to the person who ordered the test.