Reconciling a Reassuring A1c With a Dangerous Glucose Pattern
A 66-year-old man with type 2 diabetes for 14 years, hypertension, and CKD stage 3b (eGFR 32 mL/min/1.73 m², urine albumin-to-creatinine ratio 640 mg/g) returns for fatigue. Medications: metformin 1000 mg twice daily, glyburide 10 mg daily, empagliflozin 10 mg daily, lisinopril 40 mg daily, atorvastatin 40 mg daily. His A1c is 7.4%. His home meter, however, shows fasting glucose 150 to 180 mg/dL and three readings below 70 mg/dL in the past month, two of them around 3 a.m. He reports no symptoms with these lows and says he 'just sleeps through it.' BMI 31, BP 138/78. Potassium 4.9 mEq/L, bicarbonate 21 mEq/L, hemoglobin 10.8 g/dL.
1. The A1c looks acceptable and the meter looks alarming. How do you decide which one to trust, and what is the clinical significance of the discrepancy?
Read the worked answer
Neither number is wrong - they answer different questions.
- A1c estimates a 2-to-3-month average and can be distorted in advanced CKD by shortened red cell survival, recent transfusion, erythropoietin use, or iron deficiency. A hemoglobin of 10.8 in a patient with eGFR 32 raises the real possibility that the A1c is falsely low.
- The meter shows the thing an average cannot: variability. Fasting values of 150 to 180 coexisting with readings under 70 mean wide swings, not a stable 7.4%.
- The asymptomatic lows are the most important finding. Hypoglycemia unawareness in a 66-year-old with long-standing diabetes is a safety emergency, not a footnote. He is not waking for his own lows, so the next one may be severe.
- Do not accept a reassuring average as reassurance. Confirm the A1c with a measure less dependent on red cell turnover - fructosamine or, better, continuous glucose monitoring for 10 to 14 days - and check iron studies and a reticulocyte count while you are at it.
Framing for the patient: the goal is not a lower number, it is fewer highs and fewer dangerous lows. This reframe is what makes the medication changes acceptable to him.
2. Which medications do you change today and on what grounds? Be specific about renal dosing rationale.
Read the worked answer
Stop glyburide - first and unequivocally. Its active metabolites accumulate in CKD and produce prolonged, recurrent hypoglycemia. In a patient with eGFR 32, nocturnal lows, and unawareness, this drug is the leading suspect and has no compensating advantage.
Reduce metformin. At eGFR 30 to 45, metformin should not be newly started, but if already tolerated it may be continued at reduced dose (commonly capped at 1000 mg total daily) with renal function checked every 3 to 6 months. Drop him to 500 mg twice daily and hold during acute illness or dehydration. Stop entirely if eGFR falls below 30.
Continue empagliflozin. Glycemic potency falls at this eGFR, but the renal and cardiovascular benefits that matter for a patient with an albumin-to-creatinine ratio of 640 persist at eGFR ≥20. Do not stop it because the glucose effect is modest - it is not being used for glucose.
Continue lisinopril as the albuminuria agent, but check potassium and creatinine 1 to 2 weeks after any change. Potassium 4.9 and bicarbonate 21 warrant monitoring for hyperkalemia and metabolic acidosis.
Replace, don't just subtract. Removing glyburide leaves a gap. Choose a glucose-lowering agent with low hypoglycemia risk that fits CKD - a GLP-1 receptor agonist is a reasonable fit given BMI 31, or basal insulin at low dose if his fasting values remain 150 to 180 after the other changes.
Justify the sequence: glyburide out first because it is actively causing harm; metformin dose-adjusted because harm is conditional; SGLT2 inhibitor untouched because benefit is independent of glucose.
3. Design the monitoring and education plan that will tell you in four weeks whether this plan is working.
Read the worked answer
Objective measures at 2 and 4 weeks:
- Basic metabolic panel including creatinine and potassium - after metformin change and to track the ACE inhibitor.
- CGM download if available, or structured seven-point fingerstick profiles. Target metric: time below range under 4%, with no reading under 54 mg/dL. Time in range 70 to 180 is the outcome that matters here, not the A1c.
- Repeat A1c in 3 months alongside fructosamine now, so you can see whether the A1c is tracking the sensor data.
Safety education, delivered concretely:
- Sick-day rules: hold metformin and empagliflozin during vomiting, diarrhea, or poor oral intake; specify who to call.
- Euglycemia can feel like hypoglycemia during de-escalation; warn him so he does not treat phantom symptoms and overshoot.
- Check for and remove hypoglycemia contributors he may not mention: alcohol, skipped meals, sulfa antibiotics, and any over-the-counter products.
- Review his meter technique and whether he is testing at the times that would actually reveal his pattern.
Escalation triggers: any reading below 54 mg/dL, any severe event requiring assistance, potassium above 5.5, creatinine rise greater than 30% from baseline, or bicarbonate falling below 18.
Evaluation anchor: success is not a lower A1c. It is elimination of nocturnal lows, time in range above roughly 70%, stable renal function, and a patient who understands why the plan changed.