Diabetic Ketoacidosis with a Low Serum Potassium
A 19-year-old woman with type 1 diabetes managed on basal–bolus insulin is admitted with 24 hours of vomiting, abdominal pain, and polyuria after a viral illness; she has been unable to keep insulin down. She is tachycardic and has deep, rapid respirations with a fruity breath odor. Point-of-care glucose is 486 mg/dL. Laboratory studies: arterial pH 7.12, PaCO2 18 mmHg, bicarbonate 8 mEq/L, anion gap 26, sodium 132 mEq/L, potassium 3.1 mEq/L, chloride 98 mEq/L, creatinine 1.0 mg/dL, serum ketones strongly positive. ECG shows a flattened T wave and a prominent U wave. The intern has written orders for an insulin infusion at 0.1 units/kg/hr to start immediately.
1. Should the insulin infusion start now? Justify your answer with the potassium physiology.
Read the worked answer
No - the insulin must wait until the potassium is above roughly 3.3 mEq/L, and she should get potassium chloride while receiving isotonic fluid.
The reasoning chain: in DKA, insulin deficiency plus hyperosmolality and acidemia drive potassium out of cells, so the serum potassium overstates total body stores. Total body potassium is profoundly depleted by osmotic diuresis and vomiting. This patient's serum potassium is already low at 3.1, meaning her intracellular stores are severely depleted.
Insulin is a powerful K⁺-shifting hormone - it activates Na⁺/K⁺-ATPase and drives potassium into cells. Starting an insulin infusion now would drop serum potassium further and risk fatal ventricular arrhythmia (the flattened T waves and U waves are already a warning). The correct sequence is fluid resuscitation, potassium repletion (typically 20–40 mEq/L of replacement fluid once urine output is confirmed), then insulin once K⁺ > 3.3 mEq/L.
This is the classic judgment call: in DKA, giving the definitive therapy (insulin) before the permissive condition (adequate potassium) is met is what kills patients.
2. The resident asks whether bicarbonate should be given at a pH of 7.12. Explain your recommendation and the reasoning behind it.
Read the worked answer
Do not give bicarbonate at a pH of 7.12. Bicarbonate is generally reserved for pH < 6.9.
The reasoning is that the acidosis here is a ketoacid acidosis that insulin will correct at its source - insulin shuts off lipolysis and ketogenesis and allows the ketones to be metabolized back into bicarbonate. Bicarbonate administration adds risks that outweigh a modest pH bump:
- Paradoxical CNS acidosis. Bicarbonate generates CO₂, which crosses the blood–brain barrier freely whileHCO₃⁻ does not, transiently worsening intracellular and CSF pH.
- Hypokalemia from intracellular potassium shift - dangerous in a patient already at 3.1 mEq/L.
- Hypocalcemia from increased calcium binding to albumin, causing tetany and impairing cardiac contractility.
- Left shift of the oxyhemoglobin dissociation curve, reducing tissue oxygen delivery exactly when perfusion is poor.
- Overshoot alkalosis and a sodium load.
The exam-relevant point is that treating the cause (insulin, fluids, potassium) - not the number - is the correct approach, and that the threshold is deliberately low (pH < 6.9) precisely because the therapy is not benign.
3. Twelve hours later the glucose is 240 mg/dL, bicarbonate has risen to 15, and the anion gap has closed to 10, but serum ketones remain positive. How do you interpret this, and what does it imply about when to stop the insulin infusion?
Read the worked answer
Do not stop the insulin infusion based on a closed anion gap.
Two things have happened. First, the fall in glucose reflects insulin-mediated glucose uptake and dilution by fluids - not resolution of ketogenesis. Second, and more subtly, the anion gap can close while ketoacidosis persists, because the kidney excretes ketone anions (as sodium and potassium salts) and the retained chloride substitutes for them. The patient is moving toward a hyperchloremic non–anion gap metabolic acidosis, which is expected and benign, but the underlying ketoacid production may still be ongoing.
What to do: add dextrose (D5) to the fluids once glucose falls below ~200–250 mg/dL and continue the insulin infusion until ketoacid production has clearly resolved - ketones are negative or near-negative and bicarbonate has recovered. Insulin is what turns off ketogenesis; dextrose is added so that insulin can keep working without causing hypoglycemia.
The judgment being tested is that "gap closed" ≠ "ketoacidosis resolved," and stopping insulin at that moment is a well-described cause of rebound ketoacidosis.