Inpatient Insulin: The Evidence Behind the Protocol

It is two in the morning on the medical ward. The intern rings you about Mr Shetty, a 62-year-old retired postmaster from Kundapura with type 2 diabetes. He is on haemodialysis on Mondays, Wednesdays and Fridays, and he was admitted three days ago with cellulitis of the leg. He takes Regular insulin before breakfast and lunch and premix at night. “Sir, his fasting sugar has been above 230 for two days. His HbA1c is only 6.5 per cent. Shall I increase the night dose?” This is an illustrative case, not a real patient.

It is the most natural question in the world, and the honest answer is: not yet. By the end of this article you will know why. You will know when each insulin acts, and when its low comes. You will know how one reading at 3 AM tells you whether a high morning sugar means too little insulin or too much. And you will know why dialysis days run low, and why that reassuring HbA1c may be wrong.

The Inpatient Insulin Protocol keeps every timing, dose, threshold and caution it uses. What it no longer carries is the reasoning behind four of them. That reasoning has moved here.

1. When each insulin acts, and when its low comes

Every insulin has three numbers that matter at the bedside. The onset tells you how long before the meal to give it. The peak tells you when the sugar falls fastest, and so when a low is most likely. The duration tells you how long the cover lasts, and so when the next dose is due. This is the table the tool carried, as one table.

In the tool, the same table now appears as two smaller ones, the insulin in both: onset and peak in the first, duration in the second. Every figure is unchanged. A table of four columns does not fit a phone screen held in one hand. Every one of these columns is something you act on, so none could be dropped.

“Why does the peak matter more than the onset?” Because the peak is when the harm happens. Look at NPH: it peaks at 4 to 8 hours. Given at ten at night, it is working hardest between two and six in the morning, when the patient is asleep and not eating. Hold on to that. It is the whole of the next section.

The same table explains the commonest timing mistake on our wards. Regular insulin has an onset of 30 to 60 minutes. Given at the moment the plate arrives, it is still getting started while the meal is being absorbed. So the sugar spikes after the meal. The insulin then peaks after the food has gone, and brings a low before the next meal. That is why the tool says Regular insulin goes 30 to 45 minutes before the meal. A rapid analogue has an onset of 10 to 15 minutes, which is why it can go just before eating.

2. A high morning sugar has two opposite explanations

Back to Mr Shetty. His fasting sugar is high. The reflex is to give more insulin at night. Sometimes that is right. Sometimes it is exactly wrong, and the patient pays for it.

There are two reasons a morning sugar can be high after a normal evening. The tool carried their causes side by side, and here they are in its words.

In plain words: in the Somogyi effect, the night insulin has taken the sugar too low in the small hours. The body defends itself with its stress hormones, and they push the sugar up by morning. The morning reading is high because of too much insulin. In the dawn phenomenon, nothing went low. The body’s normal morning hormones make it resistant to insulin for a few hours, and there was not enough insulin to meet them.

“Then how do I tell them apart?” With one reading at 3 AM. If it is low, it is the Somogyi effect, and the night dose comes down. If it is normal or high, it is the dawn phenomenon, and the night dose can go up or move closer to bedtime. The tool keeps that reading, and what each answer means, in its Evidence tab.

This is where residents most often go wrong, and it is easy to see why: a high number looks like a call for more insulin. Increasing the night dose in a patient who is already going low at 3 AM makes the low deeper, and the morning sugar may go up again. Repeated lows also blunt the warning signs of sweating, palpitations and tremor, so the next low may arrive as confusion or a seizure without any warning at all. Check before you increase.

3. Why dialysis days run low

Mr Shetty dialyses three days a week, and on those days his insulin needs are not the same. The tool says that the commonest pattern is hypoglycaemia during the session and for up to 6 hours after it. The reason has three parts. Glucose is removed across the membrane while the session runs, meals are often missed for it, and insulin sensitivity improves as uraemia clears.

Put simply, the machine takes some glucose out of the blood. The patient often misses a meal to go for the session. And as the waste products are cleared, the body becomes more sensitive to the insulin already given. All three push the same way. So the tool asks you to reduce the dose that acts across the session, to check the glucose at the end of the session, and to check again 2 hours later.

“Does peritoneal dialysis do the same?” No, it does the opposite. The dialysate itself contains glucose, and much of it is absorbed. The tool works it out: a 2 L exchange of 2.5% dextrose instils 50 g, roughly 60% of it is absorbed over a standard dwell, and four such exchanges deliver about 120 g of glucose a day. That is a steady load spread across the whole day, which is why a basal-heavy or premix regimen suits it better than a meal-linked one.

4. Why the HbA1c misleads on dialysis

Now the HbA1c of 6.5 per cent that reassured the intern. The HbA1c measures how much of the haemoglobin has picked up glucose over the life of the red cell. It assumes the red cells live their normal span. On dialysis they do not. Red cell survival is shortened, transfusions dilute the glycated fraction, and erythropoietin increases the young red cell population.

Each of these leaves younger red cells in the blood, and younger cells have had less time to pick up glucose. So the HbA1c reads falsely low. The tool gives the size of the error: a patient with an HbA1c of 6.5% on haemodialysis may be running 250 mg/dL. Judge control from the capillary profile, or from glycated albumin or fructosamine where available, and not from the HbA1c.

Back to Mr Shetty

So what do you tell the intern? Do not increase the night dose tonight. Check the sugar at 3 AM. In our illustrative case it comes back at 58 mg/dL. That is a low, so the high fasting sugar is the Somogyi effect. Treat the low now by the rule of 15: 15 g of fast-acting carbohydrate, and a recheck in exactly 15 minutes. Then the night premix comes down, or he gets a bedtime snack. Tomorrow is a dialysis day, so the dose that acts across the session is reduced as well, with a check at the end of the session and 2 hours later. And nobody writes “diabetes well controlled, HbA1c 6.5 per cent” in his notes.

If his insulin is being adjusted day by day on a Regular, Regular and premix regimen, the author’s earlier article Daily Insulin Titration: The Crossover Method explains how each dose is titrated against the reading it governs.

Try this

A 55-year-old woman on NPH twice daily has a fasting sugar of 230 mg/dL on three mornings. Her 3 AM reading is 160 mg/dL. What is happening, and what do you change? Illustrative case.

Answer. A 3 AM reading of 160 is not low, so this is not the Somogyi effect. It is the dawn phenomenon: the early-morning hormones are making her resistant to insulin, and the night NPH is not enough to meet them. Increase the night NPH, or shift it closer to bedtime so that its peak of 4 to 8 hours falls nearer the early morning.

Take-home points

• Read each insulin by its onset, peak and duration; the peak is when the low comes.
• Before increasing a night dose for a high morning sugar, check the sugar at 3 AM.
• A low 3 AM reading means the Somogyi effect: less night insulin, or a bedtime snack.
• On haemodialysis days, expect lows during the session and for up to 6 hours after; reduce the dose and check twice.
• On dialysis the HbA1c reads falsely low; judge control from the capillary profile.

Declaration: the author built and maintains the tool. It is free to use. The figures in this article are those the tool carried in its Evidence tab on 5 November 2026, each from the source the tool names. The illustrative cases follow the tool’s rules; check the tool’s result for a real patient.

References

1. American Diabetes Association Professional Practice Committee. Standards of Care in Diabetes – 2026. Section 16: Diabetes Care in the Hospital. Diabetes Care. 2026;49(Suppl 1). [Current edition]
2. Umpierrez GE, Smiley D, Zisman A, et al. Randomised study of basal-bolus insulin therapy in the inpatient management of patients with type 2 diabetes (RABBIT 2 trial). Diabetes Care 2007;30(9):2181-2186.
3. Goldberg PA, Siegel MD, Sherwin RS, et al. Implementation of a safe and effective insulin infusion protocol in a medical intensive care unit (the Yale protocol). Diabetes Care 2004;27(2):461-467.
4. Joint British Diabetes Societies for Inpatient Care. The Use of Variable Rate Intravenous Insulin Infusion in Medical Inpatients (JBDS 04). JBDS-IP; revised 2022.
5. Joint British Diabetes Societies for Inpatient Care. Management of Hyperglycaemia and Steroid (Glucocorticoid) Therapy (JBDS 08). JBDS-IP; revised January 2023. [Supersedes Roberts A, James J, Dhatariya K, et al. Diabet Med 2018;35(8):1011-1017.
6. McMahon MM, Nystrom E, Braunschweig C, et al. A.S.P.E.N. Clinical Guidelines: Nutrition Support of Adult Patients with Hyperglycemia. JPEN J Parenter Enteral Nutr 2013;37(1):23-36.
7. Kidney Disease: Improving Global Outcomes (KDIGO) Diabetes Work Group. KDIGO 2022 Clinical Practice Guideline for Diabetes Management in Chronic Kidney Disease. Kidney Int 2022;102(5S):S1-S127.
8. Research Society for the Study of Diabetes in India. RSSDI Clinical Practice Recommendations for the Management of Type 2 Diabetes Mellitus. Int J Diabetes Dev Ctries 2022;42(Suppl 1):1-143.
9. Gangopadhyay KK, Bantwal G, Talwalkar PG, Muruganathan A, Das AK. Consensus evidence-based guidelines for in-patient management of hyperglycaemia in non-critical care setting as per Indian clinical practice. J Assoc Physicians India. 2014;62(7 Suppl):6-15. PMID 25668932.
10. Joint British Diabetes Societies for Inpatient Care. The Hospital Management of Hypoglycaemia in Adults with Diabetes Mellitus (JBDS 01). JBDS-IP; revised 2023.
11. Lane WS, Cochran EK, Jackson JA, et al. High-dose insulin therapy: Is it time for U-500 insulin? Endocr Pract 2009;15(1):71-79.