Diabetic ketoacidosis (DKA) is an acute life threatening metabolic complication of diabetes that imposes a substantial burden on healthcare resources. The main aim of this study was to assess factors that influence time to resolution of DKA. We also benchmarked length of stay (LOS) for DKA at our hospital with other tertiary Victorian hospitals.

This retrospective study surveyed patients admitted to St Vincent’s Hospital, Melbourne, between November 2010 to November 2014, coded with a discharge diagnosis of “diabetic ketoacidosis”. Main outcome measures included time to normalisation of relevant biochemical markers, precipitating factors, type of fluid rehydration, development of hypokalaemia and hospital LOS. Resolution of DKA was pre-defined as pH >7.3, bicarbonate >15, ketones <0.6. Victorian data was obtained through the Victorian Admitted Episode Dataset (VAED). Results are expressed as median (interquartile range, IQR) and number (percentage, %).

We identified 71 patients with DKA; 42 (59%) were male, 55 (78%) had pre-existing diabetes and 21 (30%) were admitted to ICU. Median age was 31(26-45) years, HbA1c 11.4% (9.6-13.0%) (101.1mmol/mol [81.4-118.6mmol/mol]), admission pH 7.20 (7.09-7.27) and admission glucose 27.6mmol (21.7-39.2). Omission of insulin was the most common precipitant for DKA (42.3%). 24 patients (34%) had hypokalaemia (<3.5 mmol/L) during admission. Time to resolution of DKA was 11 hours (6.5-16.5) and correlated to admission potassium (p<0.05) and was inversely related to admission pH (p<0.001). Normal saline was the resuscitation fluid of choice (48%). Use of Hartmann’s solution was associated with faster resolution of acidosis (p=0.02). Median LOS was 3 days correlating with time to normalisation of metabolic parameters (p<0.005). In a similar 4 year period, there were 2175 adult admissions with DKA to Victorian tertiary hospitals with median LOS 2 days (1-4). Our LOS may be longer due to the higher percentage requiring mechanical ventilation compared to other Victorian tertiary hospitals (5.1 vs. 2.2%).

Lower admission pH and higher admission potassium levels are predictors of slower time to resolution of DKA, based on normalisation of pH, bicarbonate and ketones. Our study suggests that the use of Hartmann’s solution is associated with a faster resolution of acidosis compared with other intravenous fluids.