Home / Resources / Clinical Gems / Practical Diabetes Care, 3rd Ed., Excerpt #3: Management of Inpatient Diabetes Part 2 of 2

Practical Diabetes Care, 3rd Ed., Excerpt #3: Management of Inpatient Diabetes Part 2 of 2

Feb 16, 2015

David Levy, MD, FRCP


Patients in the Intensive Care Unit


Hyperglycemia, whether associated with known diabetes or not, carries a poor prognosis in critically ill patients through multiple mechanisms, including vascular disease, polyneuropathy, an increased tendency to infection, dyslipidemia, and abnormal anti-inflammatory and coagulation responses. Consistent near-normoglycemia has therefore been a target for surgical patients in the ICU since van den Berghe reported significant reductions in mortality and morbidity by targeting blood glucose levels in the range 4.5–6.0 mmol/L (81–108 mg/dL; achieved mean 6.1 mmol/L, 110 mg/dL) compared with a target of 10–11 mmol/L (180–198 mg/dL) in patients spending more than 5 days in ICU [17]. Similar results were subsequently seen in medical ICU patients, though without the mortality benefits. Two major recent studies replicating van den Berghe’s protocol have moderated nearly 10 years of euglycemic enthusiasm. Both highlighted a significantly increased risk of severe hypoglycemia, one a possible trend towards longer ICU stays, the other a significantly increased death rate (NICE-SUGAR [18]). The whole field is currently unclear again, but blood glucose levels below 10 mmol/L (180 mg/dL), avoiding hypoglycemia, are now generally considered reasonable in critically ill patients. Variable-rate insulin infusions are required, especially with enteral or parenteral nutrition….

Non-critically Ill Patients
There is no RCT evidence in general medical patients for benefits or otherwise of tight glycemic control, but severe hypoglycemia must be avoided: inpatient mortality significantly increases with any hypoglycemic episode with glucose below 1.7 mmol/L (30 mg/dL), and both inpatient length of stay and 1-year mortality increases with increasing numbers of hypoglycemic episodes where glucose is below 2.8 mmol/ L (50 mg/dL) [19]. Although these are not RCT data, the figures are compelling, particularly in the light of other recent RCTs, and certainly more convincing than the evidence for near-normglycemia in general medical patients. There is no reason for blood glucose targets to be different from those in critically ill patients (i.e. <10 mmol/L, 180 mg/dL), while rigorously avoiding hypoglycemia and taking immediate action to prevent further episodes. A blood glucose target below 10 mmol/L is difficult to achieve with non-insulin agents or twice-daily biphasic insulin, but if the patient is eating, a temporary change to a basal-bolus regimen will usually be satisfactory. If not eating, then use a continuous variable intravenous insulin infusion together with 5 or 10% glucose, but this is uncomfortable, labour-intensive and carries risks of hypoglycemia, hyponatremia and hypokalemia. A recent trend in the UK is to alternate glucose or saline infusions depending on an arbitrary blood glucose level, together with a continuous variable intravenous insulin infusion. It has no sound basis, and carries an increased risk of both hypoglycemia and inadequate volume replacement. Use glucose and insulin for control of blood glucose, and infuse other solutions as necessary for volume replacement, as in hyperglycemic emergencies (see Chapter 2).

The degree of hyperglycemia at presentation of stroke is related to initial infarct volume, early infarct progression, and poor short- and medium-term clinical outcomes, but as in many aspects of glycemia and general medicine, these robust observational data are not complemented by RCT evidence. Hyperglycemia is as common in stroke as in myocardial infarc- tion (40% of survivors have normal glucose tolerance, 40% IGT, 20% diabetes). Studies are in progress to determine the benefits of acute control of hyperglycemia (target: 4–7 mmol/L, 72–126 mg/dL). Current European guidelines propose glucose levels below 10 mmol/L (180 mg/dL), though caution is required in patients with impaired consciousness. Even without insulin treatment, glucose levels tend to fall in the first 8 hours after admission with stroke, so do not rush to start an intravenous regimen.

Enteral Feeding (nasogastric, percutaneous endoscopic gastrostomy)
Malnutrition in hospital is very common, especially in people with diabetes, and renal and neurological impairment, and patients taking less than 60% of their requirements after 7–14 days should be assessed for specialized nutritional support. While the benefits of nutritional support (reduced infection rates and length of stay) are evident, enteral nutrition in people with diabetes may cause marked hyperglycemia, which may neutralize some of the benefit. Diabetes-specific feed formulations, i.e. low carbohydrate, high monounsaturated fatty acids (up to 30% of total calories) and high fibre, give smaller increases in blood glucose and lower peak values than standard feeds. Overall insulin requirements may be lower, medium-term glycemia may be better, and where available these formulations should be used [20]. However, the clinical problem remains the rapid rise in blood glucose levels after the start of the feed. Oral agents are usually ineffective, including metformin, the only drug available as a liquid formulation (see Chapter 6). Discuss the feed regimen with the nursing staff, dietitian and the hospital nutrition team, and then devise an appropriate insulin regimen, for example the following two options.

  • Isophane insulin (NPH) or the long-acting analogue insulin glargine (Lantus), where the dose of insulin glargine is the same as the total previous insulin dose. This simple regimen is reported to be successful in controlling blood glucose without hypoglycemia up to 7 days. In patients new to insulin, start at the conservative dose of about 0.5 units/kg body weight per day, but be prepared to increase the dose rapidly.
  • A high mixture (e.g. Humalog Mix 50) at the start of the feed, and 8 hourly thereafter.

Many feeding regimens alternate with water for 6–10 hours out of 24; take this into account when planning insulin injection times. High insulin doses are often required, so adjust them every day according to capillary blood glucose (CBG) values. Apart from the few patients requiring parenteral nutrition, avoid intravenous insulin wherever possible.

Glucocorticoid Treatment
Acute high-dose glucocorticoid treatment (e.g. prednisolone 20–30 mg daily) or ‘neurological’ doses of dexamethasone (e.g. 12–16 mg daily) nearly always causes marked and rapid deterioration in glycemic control in known diabetic patients, especially postprandially. The mechanism is increased peripheral insulin resistance, but this fact is of little practical help, because steroid-induced hyperglycemia rarely responds to insulin-sensitizing drugs. Sulphonylureas and insulin are the mainstays of treatment, but runaway symptomatic hyperglycemia is still common, and vigilance is needed. It would be wise to request an HbA1c and perform once- or twice-daily postprandial CBG measurements in people without known diabetes, but at high risk, as soon as the decision is made to start high-dose steroid treatment.

There are no clear guidelines for the management of steroid-induced hyperglycemia, and it is not known in what proportion of patients it remits once treatment is stopped. The aim in short-term steroid treatment is to avoid severe hyperglycemia (e.g. blood glucose > 15 mmol/L, 270 mg/dL) and hyperglycemic symptoms. The usual criteria for good control would apply in long-term treatment.

Patients with Known Diabetes Insulin-Treated Diabetes
Anticipate increased insulin requirements of 25–50% and increase insulin doses immediately steroid treatment starts. Involve the diabetes team as soon as possible.

Diabetes Treated with Oral Hypoglycemic Agents

  • Start or increase the dose of sulphonylurea according to CBG measurements. If patients are already taking near-maximum effective doses of sulphonylureas (e.g. glibenclamide 10–15 mg/day, gliclazide 160–240 mg/day, glimepiride 2–3 mg/day), further increases are likely to be ineffective (Chapter 6). Plan for early insulin treatment, and certainly when CBG levels are consistently above 12–15 mmol/L (216–270 mg/dL), at which level symptoms are likely.
  • Other glycemic drugs can be continued, but do not attempt to increase doses: they act too slowly to be effective in this situation.
  • Start with twice-daily biphasic insulin (e.g. 12–16 units before breakfast, 8–12 units before evening meal), or Humalog Mix 50 three times daily (e.g. 6 units with breakfast, 8 units with lunch, 8–10 units with the evening meal), and be prepared to increase each dose daily by at least 2 units. Basal bolus regimens may be useful in long-term treatment (add basal bedtime insulin to the 2- or 3-times daily biphasic insulin).

Inpatient Screening Routine
All people with diabetes should be screened systematically. This is now usually performed at annual review in primary care; where possible, check for recent laboratory tests, and do not repeat them unnecessarily (many laboratories will not permit HbA1c measurements to be repeated within 8 weeks). Be especially vigilant in patients:

  • admitted with a hyperglycemic emergency, particularly type 1 patients with DKA;
  • who have limited contact with their primary care team, for whatever reason, including geographical mobility and multiple comorbidities that prevent their attending for primary care reviews.

Clinical Examination

  1. Weight and BMI.
  2. Peripheral pulses.
  3. Arterial bruits (femoral, abdominal, carotid).
  4. Feet.
  5. Urinalysis for ketones and proteinuria.

Laboratory Tests

  1. HbA1c.
  2. Fasting lipids.
  3. Thyroid function (if not done within the past year in type 1 patients, within the last 3–5 years in type 2 patients). Remember that sick patients with non-thyroidal illness are likely to have the ‘sick euthyroid’ picture of low FT4 and low but not suppressed thyroid-stimulating hormone (TSH), so do not rush into treatment.
  4. Early-morning urine for albumin/creatinine ratio (ACR). Although albuminuria increases in infection, fever or severe hyperglycemia, it will give a broad indication of the likely range of albuminuria out of acute illness.
  5. 12-lead ECG.
  6. Dietetic review, as required.
  7. Review insulin-taking patients with diabetes specialist nurse. If possible, identify in advance areas that need addressing, for example:
    1. The insulin regimen: did it directly or indirectly cause the admission? Can it be rationalized or simplified? Is the patient using up-to-date injection devices?
    2. Injection technique
    3. Home blood glucose monitoring: education, technique and equipment
    4. Employment, school, driving status, family psychodynamics, alcohol, smoking, drugs.

Perioperative Management

Metabolic Responses to Surgery
Surgery induces insulin resistance through several mechanisms including increased secretion of counterregulatory hormones, especially catecholamines, cortisol and growth hormone. Catecholamines themselves inhibit insulin secretion, and the combination of insulin resistance and reduced insulin secretion, in the presence of perioperative starvation, results in increased lipolysis and ketogenesis, increased protein breakdown, and hyperglycemia.

Although it is difficult clinically to demonstrate any differences in wound healing in people with diabetes, postoperative infections are more common in poorly controlled diabetes (see Chapter 7), partly as a result of defective neutrophil function.

Aims of Diabetes Management in Surgical Patients
Diabetes management in surgical patients aims to avoid:

  • excess morbidity and mortality, especially through infection;
  • severe hyperglycemia;
  • ketoacidosis in type 1 patients;
  • hypoglycemia during anesthesia and the postoperative period

Preoperative Diabetic Control
Increasing use of minimally invasive and day-case surgery means that patients are less likely, and have less need, to be admitted preoperatively for glycemic stabilization. Glycemic control must be optimized before admission, but recognize that mean HbA1c in a hospital diabetes clinic is about 8% (64 mmol/mol), representing an estimated average glucose (eAG) of about 10 mmol/L (183 mg/dL) (95% CI 8–12 mmol/L, 150–220 mg/dL). In many patients, reducing this further is difficult and may not be clinically useful in reducing perioperative complications (while increasing the risk of hypoglycemia). However, try where possible to improve on HbA1c levels of 9% (75 mmol/mol) or more, representing an eAG of about 12 mmol/L (210 mg/dL) (95% CI 9–14 mmol/L, 170–250 mg/dL). Despite everyone’s best efforts some patients remain in very poor glycemic control (HbA1c > 10%, 86 mmol/mol). Recurrent or prolonged postponement pending improved control is dispiriting for everyone, and this is the group likely to benefit from intensified input from the primary or secondary care diabetes team in the weeks before admis- sion, and from intensive inpatient preoperative management. If admitted preoperatively, start 7-point blood glucose monitoring (before meals, 1.5–2 hours after meals and at bedtime) on the patient’s usual insulin regimen. If there is enough time, change temporarily to a basal-bolus regimen if control is very poor, but it may make things worse in the short term. Intravenous insulin infusion from admission may be wise in the chronically poorly controlled. Most hospitals have their own detailed protocols, but bear in mind the following.

General Principles of Perioperative Management

  • Wherever possible insulin-treated patients should be first on a morning operating list, a sensible rule that is increasingly not observed. Surgery later in the day puts patients at increased risk because of the need for prolonged preoperative glucose and insulin infusions, and because they may return to the ward late in the day.
  • Insulin infusions must always be accompanied by glucose; where necessary for fluid replacement, co-infuse other solutions. Ensure that all infusions keep running during transfer from ward to theatre and back.
  • Metformin: omit only on the day of surgery, and restart postoperatively when renal function is confirmed to be normal (or back to preoperative levels). The only caution might be patients taking modified-release preparations with their evening meal; changing to morning dosing for a few days before surgery, especially if there is any renal impairment, would probably be wise.
  • Sulphonylureas: omit glibenclamide (glyburide) the evening before surgery because of its long action. The twice-daily injected exenatide is glucose-responsive and does not cause hypoglycemia, so the evening dose can be taken in the usual way.
  • Insulin: after the evening meal the day before surgery, do not give further short-acting or biphasic insulin doses. Some recommend continuing long-acting subcutaneous insulin in type 1 patients in order to reduce the risk of DKA if intravenous insulin is interrupted; however, hypoglycemia while taking long-acting insulin is always a potential problem, and so long as continuity of insulin treatment is assured (which must be an aim of all hospital guidelines), the risks of hypoglycemia probably outweigh those of DKA. However, maintaining basal insulin treatment during surgery in patients on insulin pumps is strongly recommended where there is sufficient team expertise: basal rates can be rapidly and effectively changed or stopped.
  • Minor operations (body surface or endoscopic procedures): wherever possible avoid infusions.
  • All other procedures: maintain insulin infusion until normal eating resumes; continue intravenous insulin for 30 min after first subcutaneous injection is given.
  • Daily electrolytes: hyponatremia and hypokalemia are risks, particularly after orthopedic and urological procedures.

Colonoscopy is a challenge for people with diabetes, particularly during the 24 hours before the procedure, when only clear fluids are permitted. Frequent blood glucose monitoring, sugary drinks with normal or slightly reduced doses of insulin or oral hypoglycemic agents, and if necessary emergency contact with the diabetes team are needed; assess carbohydrate content of usual meals and convert to volumes of Lucozade or equivalent sports drink (100 mL contains approximately 20 g carbohydrate, therefore standard 380-mL bottle contains 65 g).

Perioperative Intravenous Insulin Regimens
Single-bag GIK infusions, the simplest and probably the safest for perioperative use, are no longer generally used in the UK because of restrictions relating to additions to infusion fluids (see Chapter 2). Separate 5% glucose infusions and variable-rate intravenous insulin infusions are therefore used; infuse 5% glucose containing 20 mL KCl at 125 mL/hour via an infusion pump, and give soluble insulin via a syringe pump at a variable rate determined by hourly CBG measurements (Table 3.1); 10% glucose is probably better, but in the UK is not available premixed with potassium.

Late Diabetic Complications in Relation to Surgery

  • Coronary heart disease and congestive heart failure (chest radiography, exercise testing, ECG). Even in type 2 patients without cardiac symptoms (Look AHEAD participants), nearly 8% had ST segment depression on routine exercise stress testing, and 1% each angina and arrhythmia; this is a group at high cardiac risk.
  • Diabetic nephropathy (anemia, fluid overload, acute on chronic renal failure).
  • Foot ulcers (osteomyelitis, MRSA, systemic sepsis).
  • Advanced autonomic neuropathy: increased risk of perioperative cardiovascular events, possibly resulting from arrhythmias or prolonged QTc interval (check postural drop in blood pressure and RR interval variation with deep breathing; see Chapter 10).
  • Rare but important: diabetic gastroparesis with delayed gastric emptying (risk of aspiration on intubation).

These patients require very close anesthetic monitoring.

For more information and to purchase this book, just follow this link:


David Levy, MD, FRCP, Consultant Physician, Gillian Hanson Centre, Whipps Cross University Hospital; Honorary Senior Lecturer
Queen Mary University of London London, UK

This edition first published 2011, © 2011 by David Levy. 1st edition 1998 (Greenwich Medical Media/Cambridge University Press) 2nd edition 2006 (Altman Publications)