Diabetic ketoacidosis

Prevalence

• hyperglycemia hyperglycaemia (> 14 mmol/l)
• metabolic acidosis (pH < 7.35 or bicarbonate < 15 mmol/l)
• High anion gap (Anion gap = Na- + KHC0 ₃ ) diabetic_ketoacidosis_Schwab_99
• ketonemia ketonaemia

• Blood glucose is higher (often > 33 mmol/l).
• No acidosis
• One plus ketonuria at the most on urine dipstick.
• Higher Na (often > 150 mmol/l)

Causes

• infection
• treatment error
• new-onset diabetes
• other medical illness

Why? The cause of many cases of DKA is unknown cause diabetic_ketoacidosis_Johnson_80 diabetic_ketoacidosis_Snorgaard_89 diabetic_ketoacidosis_Basu_93 diabetic_ketoacidosis_Westphal_96 prevalence unknown 19% to 38% infection 27% to 38% treatment error 12% to 28% other medical illness 11% pancreatitis 2% to 3% myocardial infarction 1% to 7% heart failure 2% GI bleed 1% affected by drugs or alcohol 9% newly diagnosed diabetes 10% to 27%

Clinical features

• known diabetes mellitus diabetic_ketoacidosis_Johnson_80 diabetic_ketoacidosis_Snorgaard_89

• previous episodes of DKA diabetic_ketoacidosis_Johnson_80 diabetic_ketoacidosis_Snorgaard_89

• current medication diabetic_ketoacidosis_Egger_97 - and any recent changes or mistakes diabetic_ketoacidosis_Johnson_80 diabetic_ketoacidosis_Snorgaard_89

• recent illness diabetic_ketoacidosis_Johnson_80 diabetic_ketoacidosis_Snorgaard_89

• polyuria, polydipsia and weakness ^d

Why? Patients who have intensive insulin therapy are at increased risk of ketoacidosis. However there is no clear effect on mortality. diabetic_ketoacidosis_Egger_97 In particular, patients on continuous insulin infusions are at increased risk of ketoacidosis. There is no clear increase in ketoacidosis for patients on multiple daily injections. diabetic_ketoacidosis_Egger_97 Intensive insulin regimens particularly insulin pumps increases the risk of DKA Patient Treatment Comparison Outcome CER OR (95% CI) NN T (95% CI) insulin-dependent diabetes diabetic_ketoacidosis_Egger_97 intensified insulin regimen standard insulin regimen diabetic ketoacidosis at 2 to 6 years 6.4% 2.88 (2.38 to 3.48) 4 (3 to 5) insulin-dependent diabetes insulin pump standard insulin regimen diabetic ketoacidosis at 2 to 6 years 0.86% 5.76 (2.88 to 11.5) 26 (12 to 64)

• dehydration

• infection (e.g. lobar pneumonia, urinary tract infection)

• associated disease (e.g. myocardial infarction, pancreatitis)

• Think about acidosis in any hyperventilating patient. ^d

Investigations

• Take a capillary blood glucose.

• Take a urine sample and test for
  • ketones diabetic_ketoacidosis_Schwab_99
  • leukocytes or nitrites - if abnormal send for culture ^d

Why? No ketones on urine dipstick make diabetic ketoacidosis very unlikely patient target disorder and reference standard diagnostic test LR+ (95% CI) post-test probability LR- (95% CI) post-test probability suspected diabetic ketoacidosis diabetic_ketoacidosis_Schwab_99 (pre-test probability: 14%) diabetic ketoacidosis (elevated glucose, metabolic acidosis and ketonemia ketonaemia) positive urine ketone dipstick 3.2 (2.9 to 3.7) 35% 0.015 (0.0021 to 0.10) 0.24%

• blood glucose

• urea and electrolytes, creatinine diabetic_ketoacidosis_Soler_72

Why? 40% of patients with diabetic ketoacidosis have abnormal potassium levels: 28% hyperkalemia hyperkalaemia ; 12% hypokalemia hypokalaemia . diabetic_ketoacidosis_Soler_72

• ketones

• pH from venous blood diabetic_ketoacidosis_Brandenburg_98

Why? Venous blood pH and bicarbonate levels correlate closely in patients with diabetic ketoacidosis. diabetic_ketoacidosis_Brandenburg_98

• bicarbonate diabetic_ketoacidosis_Schwab_99
• Calculate the anion gap (Na- + KHC0 ₃ ) diabetic_ketoacidosis_Schwab_99

Why? A normal anion gap makes DKA unlikely, and a low bicarbonate makes it very likely patient target disorder and reference standard diagnostic test LR+ (95% CI) post-test probability LR- (95% CI) post-test probability suspected diabetic ketoacidosis diabetic_ketoacidosis_Schwab_99 (pre-test probability: 14%) diabetic ketoacidosis (elevated glucose, metabolic acidosis and ketonemia ketonaemia) anion gap > 16 mmol/l 6.3 (5.1 to 7.6) 51% 0.096 (0.049 to 0.19) 15% suspected diabetic ketoacidosis (pre-test probability: 14%) diabetic ketoacidosis (elevated glucose, metabolic acidosis and ketonemia ketonaemia) serum bicarbonate < 15 mmol/l 100 (42 to 240) 94% 0.16 (0.11 to 0.26) 2.6%

• blood count ^d

• cardiac enzymes diabetic_ketoacidosis_Johnson_80 diabetic_ketoacidosis_Snorgaard_89

• amylase diabetic_ketoacidosis_Johnson_80 diabetic_ketoacidosis_Snorgaard_89

• blood cultures ^d

• chest X-ray diabetic_ketoacidosis_Johnson_80 diabetic_ketoacidosis_Snorgaard_89

• 12-lead ECG and ECG monitoring diabetic_ketoacidosis_Johnson_80 diabetic_ketoacidosis_Snorgaard_89

• Repeat electrolytes and glucose levels diabetic_ketoacidosis_Soler_72 diabetic_ketoacidosis_Harris_90 frequently ^d until biochemical normality is achieved. A chart for vital signs, laboratory results and fluid balance is helpful. ^d

Therapy

• Resuscitate and seek help if required. ^d

• Give intravenous fluids - initially 0.9% saline diabetic_ketoacidosis_Harris_90 (e.g. 1 litre over 30 min, 1 litre over 1 h, 1 litre over 2 h, 1 litre over 4 h).
• If none of the following are present, fluids can safely be given more slowly if necessary ^{ddiabetic_ketoacidosis_Adrogue_89}
  • circulatory shock
  • oliguria (<30ml/hr) during the first 4 hours of admission
  • renal insufficiency (urea > 21mmol/l or creatinine > 350

Why? If there is no evidence of severe dehydration, normal saline given at 600ml/hour for 4 hours followed by 250 ml/hour for 4 hours does not clearly affect time to normalised biochemistry than normal saline 1L/ hour for 4 hours followed by 600ml/hr for 4 hours. ddiabetic_ketoacidosis_Adrogue_89

• In dehydrated or comatose patients, consider ^d
  • a urinary catheter
  • a central venous line

• Monitor electrolytes diabetic_ketoacidosis_Soler_72 diabetic_ketoacidosis_Harris_90 and capillary glucose frequently. ^d . Give potassium supplementation after insulin therapy has begun if K+ <5.5 mmol/l ^d . Provide 10-30 mmol/h diabetic_ketoacidosis_Soler_72

Why? Potassium abnormalities are common - 28% have hyperkalemia on admission (95% CI: 10% to 46%) and 12% have hypokalaemia (95% CI: 0% to 25%) diabetic_ketoacidosis_Soler_72 Patients required on average 30-40 mmol of potassium per litre of fluid to keep serum potassium normal during rehydration. diabetic_ketoacidosis_Soler_72 A patient whose serum sodium concentration falls or fails to rise during rehydration is at increased risk of developing cerebral oedema. A failure to rise suggests rehydration with excess free water. diabetic_ketoacidosis_Harris_90 A failure of sodium to rise on rehydration increases the risk of cerebral oedema Patient Prognostic factor Outcome CER RR (95% CI) NNF+ (95% CI) DKA diabetic_ketoacidosis_Harris_90 no rise in serum sodium on rehydration not independent cerebral oedema at ? 2.3% (0.0% to 5.5%) 6.56 (1.56 to 27.53) 8 (2 to 76)

• Give broad-spectrum antibiotics if there is evidence of infection

• Give soluble insulin in low-doses (e.g. 5 to 10 units per hour) diabetic_ketoacidosis_Kitabchi_76 intravenously ^d at regular intervals or continuously ^d

Why? A low-dose insulin regimen is less likely to cause hypoglycemia hypoglycaemia or hypokalemia hypokalaemia than a high-dose one. diabetic_ketoacidosis_Kitabchi_76 There is no clear difference in the time taken to return to biochemical normality. diabetic_ketoacidosis_Kitabchi_76 A low-dose insulin regimen reduces the risk of hypoglycemia hypoglycaemia or hypokalemia hypokalaemia Patient Treatment Comparison Outcome CER RRR (95% CI) NN T (95% CI) diabetic ketoacidosis low-dose insulin high-dose insulin hypoglycemia hypoglycaemiamol/l) at 12 hours 25% 100% 4 (2 to 13) diabetic ketoacidosis low-dose insulin high-dose insulin hypokalemia hypokalaemia (< 3.4 mmol/l) at 12 hours 29% 86% (-7% to 98%) 4 (2 to 19) The route used to administer insulin in patients has no clear effect on the time taken to return to biochemical normality or the amount of insulin required. ddiabetic_ketoacidosis_Fisher_77 diabetic_ketoacidosis_Sacks_79 A continuous insulin infusion is not clearly more likely to cause a faster fall in glucose levels nor shorten the time to reach a glucose <14 mmol/l than a bolus followed by regular injections. ddiabetic_ketoacidosis_Heber_77

• Continue giving insulin by this route until diabetic_ketoacidosis_Wiggam_97
  • glucose <10 mmol/l, and
  • ketones are cleared (3-hydroxybutyrate <0.5 mmol/l)
  If glucose <10 mmol/l but ketones are still raised, continue insulin infusion with 20% glucose iv to maintain glucose 5-10 mmol/l

Why? Patients with diabetic ketoacidosis who receive an extended insulin regimen have a more rapid fall in ketones than those on a conventional regimen (~16 hours difference) diabetic_ketoacidosis_Wiggam_97

• Once patients have stabilised, swap to subcutaneous insulin. Give 5% glucose and insulin infusion with subcutaneous insulin as necessary to maintain blood glucose < 10 mmol/l until patients are eating. Give the first subcutaneous dose before stopping the infusion ^d

• sodium bicarbonate ^{ddiabetic_ketoacidosis_Morris_86 diabetic_ketoacidosis_Hale_84}

Why? Patients with severe DKA who receive bicarbonate do not clearly return more quickly to biochemical stability ddiabetic_ketoacidosis_Morris_86 diabetic_ketoacidosis_Hale_84 The effect on hypokalaemic or hypoglycaemic episodes is unclear ddiabetic_ketoacidosis_Morris_86 diabetic_ketoacidosis_Hale_84

• routine phosphate supplementation diabetic_ketoacidosis_Keller_80

Why? It reduces the risk of hypophosphataemia but increase the risk of infection, and has no clear effect on mortality diabetic_ketoacidosis_Keller_80 Patients do not recover consciousness more quickly nor leave hospital sooner diabetic_ketoacidosis_Keller_80 It has no clear effect on pH, phosphate, calcium or glucose levels at 24 h ddiabetic_ketoacidosis_Wilson_82 diabetic_ketoacidosis_Fisher_83

• hypertonic glucose ^{ddiabetic_ketoacidosis_Krentz_89}

Why? Patients with a glucose <14 mmol/l do not clearly have a faster improvement in biochemical markers following 10% glucose and insulin rather than 5% glucose and insulin ddiabetic_ketoacidosis_Krentz_89

Prevention

• Refer your patient to a diabetes team and educate your patient about diabetes. diabetic_ketoacidosis_Koproski_97

Why? It improves glycemic glycaemic control and reduces readmissions. diabetic_ketoacidosis_Koproski_97 There is no clear effect on length of hospital stay. diabetic_ketoacidosis_Koproski_97 A diabetic team improves glycemic glycaemic control and reduces hospital readmissions Patient Treatment Comparison Outcome CER RRR (95% CI) NN T (95% CI) in-patient with diabetes diabetic_ketoacidosis_Koproski_97 diabetic team intervention no intervention good glycemic glycaemic control at 4 weeks 46% 65% (28% to 110%) 3 (2 to 6) in-patient with diabetes diabetic team intervention no intervention readmission at 3 months 32% 52% (14% to 73%) 6 (3 to 22)

Prognosis

Note Half of patients have another episode number of subsequent episodes of DKA diabetic_ketoacidosis_Johnson_80 % of patients 0 58% 1 23% 2 10% 3 or more 9% 20% of women with recurrent diabetic ketoacidosis are dead within 10 years. diabetic_ketoacidosis_Kent_94 diabetes_mellitus_Tattersall_91 Two-thirds have a diabetic complication and ~75% have a pregnancy complication in this time. diabetic_ketoacidosis_Kent_94 diabetes_mellitus_Tattersall_91 Only 10% still have recurrent DKA after 10 years. diabetic_ketoacidosis_Kent_94 diabetes_mellitus_Tattersall_91