Background

Diabetic ketoacidosis (DKA) contributes to mortality and morbidity in type 1 diabetes.  The incidence of DKA is increasing in Australia (4.6-13.4 per 1000 diabetic cases/year).

Aims

• Describe characteristics of inpatients with DKA in tertiary care.
• Compare outcomes for inpatients with DKA in Royal Perth Hospital (RPH) under the care of Endocrinologists to patients admitted under General Physicians in Fremantle Hospital (FH); both sites utilised variable dose insulin infusion protocols.

Methods

An observational, retrospective audit of 136 DKA admissions in 115 patients (88 admissions to RPH from 2008 to 2009 and 48 admissions to FH in 2010). DKA diagnosis established via clinical coding and confirmed by medical record review.

Results

The mean±SD age of patients presenting with DKA was 35.8±15.5 years, median [IQR] diabetes duration 9.0 [2.5-15.0] years and HbA1c 10.8 [8.7-12.5] %.  49.1% of patients experienced a previous episode of DKA.  Main precipitants of DKA included infection (33.1%), insulin omission (31.5%), alcohol (3.1%) and illicit drug use (1.5%).  At presentation, median pH was 7.14 [6.98- 7.24] and bicarbonate was 8.0 [5.0- 12.9] mmol/L.   Mean trough potassium (K) level was 3.4±0.5 mmol/L and hypokalaemia (K< 3.5mmol/L) occurred in 58% of admissions.  Median length of stay (LoS) was 3 [2-6] days; there was no difference in LoS or trough potassium levels (p≥0.37) between sites. 86.1% of patients had a Diabetes Outpatient appointment scheduled following admission but only 51.4% of patients attended. Prior DKA (OR(95%CI): 5.6(1.4- 22.1)) and lower trough K level (0.2(0.1- 0.8)) were associated with recurrent DKA during the audit in multiple logistic regression analysis.

Conclusion

Infection and insulin omission were the most common precipitants of DKA.  Prior DKA was associated with increased risk of recurrent DKA indicating that sick day education could be optimised.  The low attendance at follow-up clinics highlights the importance of inpatient engagement.