Background: DKA carries risk of morbidity and mortality but management relies on expert opinion in the absence of large randomised controlled trials (RCTs). 

Aims: To review original studies investigating inpatient management of DKA specifically i. Choice of intravenous fluids, ii. Method of insulin administration, and iii. Rate of potassium replacement.

Methods: Ovid Medline searches were conducted including terms [Diabetic Ketoacidosis.mp. or Diabetic Ketoacidosis/], [Rehydration.mp. or Fluid Therapy/], [Insulin/ or insulin infusion rate.mp.], [Potassium/ or Potassium replacement.mp.], with the limit "all adult (19plus years)” applied to each search. Reference lists from national consensus-statements were also reviewed. Eligibility was determined by single reviewer’s assessment of title, abstract, and full-text availability.

Results: 65 eligible articles published between 1973-2015. 

i.                Crystalloids are favoured due to cost and potentially increased mortality with starch containing colloids. Choice between ‘balanced electrolyte solutions’ (BES) and saline is contentious. Consensus-statements favour saline due to familiarity, availability, and pre-mixed potassium. One small RCT along with case-series and retrospective cohort studies highlight potential adverse effects of saline, particularly hyperchloraemic metabolic acidosis (HMA). This may be avoided with BES. Rehydration must avoid pulmonary or cerebral oedema and hypoperfusion, though data is lacking in adult populations. 

ii.               Regular insulin is favoured due to cost and failure to demonstrate superiority of analogs. Intravenous infusion is favoured over subcutaneous and intramuscular routes in small randomised and retrospective studies due to rapid onset and similar cost when ICU admission is avoided. Consensus-statements recommend fixed weight-based insulin infusions, although comparative trials are lacking. Initial insulin bolus was superfluous in a RCT with rates of 0.14units/kg/hour.

iii.              Hypokalaemia may be fatal though randomised trials comparing rates are not feasible. Consensus suggests 10-20mmol/hour is reasonable but hospital, regional, and country specific policies should be followed considering urine-output and cardiac-monitoring.

Conclusions: 

National consensus-statements guide management of DKA, though analysis of original data reveals limitations. Saline is favoured as the rehydration fluid at a rate avoiding fluid overload and hypoperfusion. Regular insulin infusion rates need to be assessed for effectiveness, and rates <0.14units/kg/hour may benefit from initial bolus. Physiological levels of potassium should be maintained with attention to local safety policies.