Managment of diabetic ketoacidosis

discussion of the pathogenesis of diabetic ketoacidosis. We consider:

  1. Immediate diagnosis and management
  2. Initial evaluation of the patient presenting with diabetic ketoacidosis should be directed at confirming the diagnosis. Adverse outcomes (and litigation) associated with DKA are often related to delays in diagnosis and management. DKA should obviously be suspected in any patient presenting with typical features such as dehydration, polyuria and polydipsia, nausea and vomiting, acidotic ("Kussmaul") respiration, and impaired level of consciousness, but also considered in patients with acute myocardial infarction or other acute vascular disease, and in those with severe sepsis or septic shock. Patients with ketoacidosis may have severe abdominal pain unrelated to any acute abdominal emergency!

    Typically, patients will be overtly acidotic, with lowered serum bicarbonate (commonly under 15 mmol/L), acidotic plasma, and ketones detectable in urine and serum. Glucose will usually be high (over 14 mmol/L), although it has recently been recognized that some patients may present with marked ketoacidosis and only mild hyperglycaemia, notably pregnant women with diabetic ketoacidosis. There are several other catches, the main one being that the commonly-used nitroprusside test is wholly inadequate for the task of identifying serum or urinary ketones. This test detects acetoacetate, and not beta-hydroxybutyrate [BHB], which is unfortunate because the major derangement in DKA is increased levels of BHB rather than acetoacetate. In fact, in DKA the ratio of BHB to acetoacetate increases from the normal value of ~1:1 to 10:1 or more.

    If a patient presents with marked ketosis, but a glucose under 14mmol/L, one should consider the possibility that the ketosis is related to some other cause (such as alcoholic ketosis) rather than diabetic ketoacidosis.

    Clearly, if the patient presents in extremis , initial management should be the usual "ABC" of establishing a secure airway, sustaining ventilation and maintaining a stable circulation. An integral part of early resuscitation of such cases is vigorous fluid resuscitation, with the caveats mentioned below. Early commencement of insulin therapy is also vital, but here too, there are important cautions.

  3. Fluid resuscitation
  4. It is not clear what the optimal rate or nature of fluid resuscitation is in diabetic ketoacidosis. Conventionally, normal saline is used. One can make a fairly strong physiological case for the use of lactated Ringer's solution in adults, although it can be argued that:

    Despite these theoretical concerns, the author would favour the use of Ringer's lactate as there is a good physicochemical reason why use of normal saline might worsen acidosis in patients with DKA, especially where large volumes are infused rapidly. This argument is based on acceptance of Stewart's physicochemical approach to acid-base, a topic which is still hotly debated, although no logical refutation of this approach has been published.

    It is common practice to replace approximately 10-20% of the fluid deficit over the first one to two hours, but guidelines vary, and there is controversy as to whether the remaining deficit should be replaced over 24 hours (as is often done), or a longer period. Although fluid resuscitation is possibly the most important component of DKA management, insulin therapy is also vital. We consider this next.

  5. Insulin therapy
  6. In the (distant) past, it was conventional to routinely administer vast amounts of insulin to patients with DKA (for example, 50U per hour)! We now know that such supra-physiological doses are rarely required. By far the best method of insulin administration is continuous intravenous infusion of approximately 0.1 U/kg/hour of short-acting insulin (about 6-8U/hour in an average adult). An alternative method of dosing is to give a similar amount of short-acting insulin hourly intra-muscularly (in, for example, the lateral thigh). Far inferior is intermittent IV bolusing of insulin, because the half life and effective half-life of IV insulin boluses is short. Subcutaneous administration should be avoided , owing to the variable absorption from such sites in dehydrated patients.

    It is important to note that in the rare patient with DKA who presents with initial hypokalaemia , insulin administration may further lower the potassium, resulting in profound muscular weakness, and even respiratory embarrassment. It is therefore important to obtain a serum potassium level rapidly, and before insulin therapy is started, so that immediate supplementation can be started where required.

    There is no good reason to give an initial bolus of insulin before an insulin infusion is started. Because of the short half-life of IV insulin, physiological levels should rapidly be attained with a continuous infusion (provided your infusion pump is reliable - many pumps are not at rates under 5ml/hour); conversely a bolus will result in supra-physiological levels of insulin that could cause hypokalaemia and other undesirable side-effects.

    During the entire course of insulin infusion therapy, blood glucose should be determined hourly. If the glucose level is not consistently dropping by about 3mmol/L every hour, then the rate of insulin infusion should be increased appropriately. Occasionally, due to insulin resistance, vast amounts of insulin may be required. Conversely, if the blood glucose is dropping precipitately, the rate of infusion may have to be slowed.

    Once glucose levels have dropped to about 13 mmol/L, it will become necessary to add glucose to the infusate. Note that insulin infusion should not be stopped abruptly before ketosis (and acidosis) has cleared. As soon as the patient is well-hydrated, acidosis is reversed, and oral intake is tolerated, subcutaneous insulin can be started, together with oral feeds - the insulin infusion can then be discontinued.

  7. Electrolyte replacement
  8. Potassium

    Several regimens of potassium replacement have been proposed. If the serum potassium is over about 5.5 - 6 mmol/L, no potassium is added to replacement fluids, but soon after insulin therapy starts, potassium levels often drop precipitately, and about 10-20mmol/l should be added to replacement solutions as soon as the serum potassium drops below 5.5 mmol/l. Once the serum potassium level has dropped to under 4 mmol/l, 30-40 mmol/L can be added to the replacement solution, provided serum K+ levels are diligently monitored.


    Several studies suggest that routine phosphate replacement is deleterious in DKA. However, levels should be checked several times during the first 24 hours, and if moderate to marked hypophosphataemia (< 0.5mmol/l) develops, this should be carefully corrected, as profound hypophosphataemia may result in muscular weakness, neurological dysfunction, and even haemolysis or rhabdomyolysis. Insulin therapy will usually lower serum phosphate levels, just as it lowers serum potassium.


    DKA may be associated with substantial hypomagnesaemia, and this should be looked for.


    Attention should be paid to alterations in serum sodium levels during therapy. Initially, if marked hyperglycaemia is present, then serum sodium will be spuriously low. An approximate correction for this pseudohyponatraemia is:

    	[Na]corrected  = [Na]measured  + 0.3 * ([glucose] - 6)

    All concentrations are in mmol/L. If the (corrected) serum sodium rises dramatically during therapy, then it is advisable to change to hypotonic fluids to avoid marked hypernatraemia. Many authorities recommend half normal saline (0.45%) in preference to more hypotonic fluids.

  9. Treating infection
  10. The two major features of infection (pyrexia and leukocytosis) are both obscured in DKA. Patients with DKA are commonly mildly hypothermic and rarely pyrexial even in the face of severe infection; DKA itself causes a raised white cell count. It is therefore reasonable to have a very low threshold for empirically treating infection in such patients, after appropriate cultures have been taken. Common sites of infection are urinary tract and skin, but every patient should also have a chest x-ray to look for evidence of pneumonia. Wherever possible, treatment of infection should be directed at likely organisms in a particular site, rather than "broad spectrum" shotgun therapy.

    Occasionally, DKA may be complicated by mucormycosis. The infection is rare, but those managing DKA should always be aware of it, owing to its rampant progression and the possibility of immediate surgical intervention being curative.

  11. Ongoing acidosis
  12. Fairly commonly, hyperglycaemia resolves before ketosis and acidosis. In some circumstances, ketones may become more prominent in the urine and blood as the patient improves , because acetoacetate and beta-hydroxybutyrate constitute a redox couple, so sicker patients have more beta-hydroxybutyrate which is not detected by the inadequate nitroprusside test. As they improve, acetoacetate may become more readily detectable. The treatment of acidosis is to continue with fluid and insulin, and administer enough sugar to prevent hypoglycaemia.

    If acidosis still persists, especially if other parameters are unfavourable (the patient looks clinically more ill, has substantial organ dysfunction, or has hypotension that is refractory to fluid therapy), then one should consider the possibility of severe sepsis/septic shock and treat appropriately. Occasionally in such circumstances, raised lactate levels may contribute to acidosis.

    There is no evidence that administering sodium bicarbonate is beneficial, even in cases where marked acidosis is present. In fact, some studies suggest that such therapy is harmful. In addition, it has been argued that rapid infusion of bicarbonate may cause a paradoxical worsening of CSF pH, owing to a rise in blood carbon dioxide levels, which gas then diffuses into the brain and lowers the pH. Another possible adverse effect of bicarbonate therapy will be to left-shift the Hb:02 dissociation curve, with diminished tissue oxygen delivery. Administration of sodium bicarbonate will eventually increase serum pH through its contribution of sodium ions, but this is only once a new equilibrium has been reached - CO 2 must first be blown off, either by the patient (who may already be hyperventilating maximally), or through mechanical ventilation. If NaHCO 3 is given, the patient must be diligently monitored, including frequent arterial blood gas determinations, with intubation and ventilation if required. One must also remember that 8.5% sodium bicarbonate is markedly hypertonic - some authors recommend dilution to approximately isotonic levels.

  13. Integrated management goals
  14. Unless management goals in DKA are well-defined right from the start, things can rapidly become chaotic, especially in an acute-care ward. Some such patients clearly belong in an ICU setting and should be managed there, especially if the patient is profoundly obtunded. Timing of investigations (including investigation for the precipitant of DKA), and therapeutic goals should be written down on a flowsheet and adhered to. All therapeutic interventions should be meticulously documented, especially fluid balances. Glasgow Coma Score should be monitored regularly right from when the patient is first seen. In the opinion of the author, it is best to have a single large flowsheet, rather than multiple scrappy pieces of paper, each of which presents a single component of the problem.

  15. Long-term management
  16. As mentioned previously, DKA is often a symptom of inadequate patient education. Poor compliance with insulin therapy is a common precipitant of DKA, and other patients who become ill inappropriately stop taking their insulin. Good follow-up education is thus vital, as is regular home glucose monitoring using an electronic glucose meter. Patients may need careful "tuning" of their insulin therapy - newer insulins such as insulin glargine (for basal therapy) and insulin lispro for boluses may be of value.

  17. References
    1. The American Academy of Family Physicians has a good public domain article on management of diabetic ketoacidosis.
      [ Kitabchi AE, Wall BM. Management of diabetic ketoacidosis Am Fam Physician 1999 Aug;60(2):455-64 ]
    2. A PDF document: UK DKA guidelines from the British Society for Paediatric Endocrinology and Diabetes. Authored by: [Julie A Edge, Claire P Smith, Jerry KH Wales, David B Dunger. URL: Accessed Friday 10 May 2002 11:00]
    3. The ISPAD DKA Treatment guidelines are also available online. [Arlan L Rosenbloom & Ragnar Hanas Diabetic Ketoacidosis (DKA): Treatment Guidelines Clinical Pediatrics 1996;35:261-266]
    4. An article in Postgraduate Medicine that's a useful 'FAQ' for several controversial issues. [Carroll MF, Schade DS. Ten pivotal questions about diabetic ketoacidosis. Answers that clarify new concepts in treatment. Postgrad Med. 2001 Nov;110(5):89-92, 95]
    5. Elliot J Krane from Stanford has written an impressive article on cerebral oedema in DKA. Well referenced, and well worth a read!
      [URL: Accessed Friday 10 May 2002 11:50]
    6. Cullen MT, Reece EA, Homko CJ, et al. The changing presentations of diabetic ketoacidosis during pregnancy. Am J Perinatol 1996;13(7):449-51
      ('Pregnant diabetics may have severe ketoacidosis with mild hyperglycaemia')
    7. Wilson HK, Keuer SP, Lea AS, et al. Phosphate therapy in diabetic ketoacidosis. Arch Intern Med 1982;142(3):517-20
      ('Phosphate therapy is ineffective').
    8. Balasubramanyam A, Zern JW, Hyman DJ, et al. New profiles of diabetic ketoacidosis: type 1 vs type 2 diabetes and the effect of ethnicity. Arch Intern Med 1999;159(19):2317-22
      ('DKA is not uncommon in type II diabetes').
    9. Endocrinol Metab Clin North Am 2000 Dec;29(4):683-705, Diabetic ketoacidosis and hyperglycemic hyperosmolar nonketotic syndrome. Delaney MF, Zisman A, Kettyle WM

    10. J Clin Endocrinol Metab 1983 Jul;57(1):177-80 A randomized study of phosphate therapy in the treatment of diabetic ketoacidosis. Fisher JN, Kitabchi AE.
      (In a small PRCT, phosphate supplementation in DKA merely lowered ionised calcium levels).