Diabetic ketoacidosis: a negative urine dipstick ruled it out

Clinical bottom line (level 4)

  1. A quarter of patients presenting in A&E with hyperglycaemia and symptoms of illness have ketones in their urine.
  2. A seventh have diabetic ketoacidosis (DKA).
  3. Patients are more likely to have diabetic ketosis or DKA if their serum bicarbonate is < 15 mmol/l (LR+45)
  4. Patients are unlikely to have diabetic ketosis or DKA if their urine ketone dipstick is negative (LR-0.24) .
  5. Patients are more likely to have diabetic ketosis if their serum bicarbonate is < 15 mmol (LR+100) .
Schwab et al: Annals of Emergency Medicine 1999; 34 (3): 342-346
Expires October 2003

The study

Setting: emergency department, university hospital, USA

697 patients (aged 18 to 96; mean 46, 53% male) with known diabetes and hyperglycaemia (glucose > 11.1 mmol/l) any complaint of illness (weakness, dizziness, nausea, vomiting, abdominal pain, malaise, altered mental status, infectious symptoms), or hyperglycaemia and symptoms of undiagnosed diabetes mellitus.

Excluded if
  • chart unavailable to obtain laboratory data
  • diabetes mellitus presenting without medical complaint (e.g. for medication refills) or for minor injuries (lacerations, ankle sprains)
  • tests not completed before patient left emergency department
  • aged < 18



    • Independent ?blinded reference standard, applied in ?all patients from a consecutive appropriate spectrum.
    Reference standard:
    • diabetic ketoacidosis
      • glucose > 11.1 mmol/l
      • metabolic acidosis (corrected pH < 7.30 or serum bicarbonate < 15 mmol/l)
      • ketonaemia positive at a 1:2 dilution or greater
      diabetic ketosis
      • glucose > 11.1 mmol/l
      • serum ketones
      • no acidosis
    Diagnostic test:
    • urine ketone dip test
    • anion gap: Na - Cl - HCO 3
    • serum bicarbonate

    The evidence

    pre-test probability of diabetic ketoacidosis or diabetic ketosis: 27%, (95% CI: 23% to 30%)
    pre-test probability of diabetic ketoacidosis: 14%, (95% CI: 12% to 17%)

    diagnostic test diabetic ketoacidosis or diabetic ketosis hyperglycaemia LR+
    (95% CI)    		 post-test probability LR-
    (95% CI)    		 post-test probability
    positive urine ketone dipstick 176 104 4.6
    (3.9 to 5.5)     		63% 0.068
    (0.037 to 0.12)     		2%
    anion gap > 16 mmol/l 135 43 8.6
    (6.4 to 12)     		76% 0.30
    (0.24 to 0.38)     		10%
    serum bicarbonate < 15 mmol/l 82 5 45
    (19 to 110)     		94% 0.56
    (0.50 to 0.64)     		17%
    total 186 511


    diagnostic test diabetic ketoacidosis diabetic ketosis or hyperglycaemia LR+
    (95% CI)    		 post-test probability LR-
    (95% CI)    		 post-test probability
    positive urine ketone dipstick 97 183 3.2
    (2.9 to 3.7)     		35% 0.015
    (0.0021 to 0.10)     		0.24%
    anion gap > 16 mmol/l 90 88 6.3
    (5.1 to 7.6)     		51% 0.096
    (0.049 to 0.19)     		1.5%
    serum bicarbonate < 15 mmol/l 82 5 100
    (42 to 240)     		94% 0.16
    (0.11 to 0.26)     		2.6%
    total 98 599

    Comments

    1. Since a low serum bicarbonate is potentially part of the definition of diabetic ketoacidosis, this test is not independent of the reference standard and therefore appears better than it performs in reality.

    Citation

    1. Schwab TM, Hendey GW, Soliz TC, et al: Screening for ketonemia in patents with diabetes. Annals of Emergency Medicine 1999; 34 (3): 342-346
    Search Terms:
    Contributor: Chris Ball and Clare Wotton, October 1999
    Reviewer: Wai-Liam Chan

    Clinical Question.
    Patient unwell and diabetes with hyperglycaemia
    Intervention or Exposure urine ketone dipstick, anion gap, serum bicarbonate
    Outcome diabetic ketoacidosis, diabetic ketosis