Background: Diabetic ketoacidosis (DKA) remains one of the commonest causes of medical and intensive care unit hospitalisations. DKA is a preventable hospital admission which is associated with multiple modifiable risk factors. However, little is known about factors that can predict healthcare resource utilisation in DKA. Previous studies have explored clinical and biochemical factors affecting the length of stay (LOS) and recurrence of DKA in small cohorts with variable results. This study aims to evaluate factors predicting length of stay, readmission, and recurrence in patients with DKA in a sizeable multi-hospital, multicultural cohort of patients.

Methods: Data of patients admitted with DKA from 2015 to 2021 to Hamad Medical Corporation, a tertiary care academic health system in Qatar were extracted from the Cerner EHR platform, Millennium® Primary outcomes were the length of stay, 90-day readmission, 6-month and 1-year recurrence. Descriptive statistics and univariate and multivariate regression analyses were conducted as appropriate.

Results: From 1949 index DKA admissions, we excluded 1027 patients who did not fit the inclusion criteria and included 922 patients in this study. Average age was 35 years, with 62.3 %s males. Arabic ethnicity was the most prevalent (54.4%). 52 % were Type I and 48% type II diabetics. Median LOS was 2.6 days, and median time to DKA resolution was 18 hours. Male gender, new DM diagnosis, higher Charlson comorbidity index (CCI), lower haemoglobin and higher urea at admission, lowest sodium, and potassium during admission, longer DKA duration, and MICU admission remained significant factors predicting a prolonged LOS in multivariate logistic regression analysis. None of the factors was significantly associated with 90-day readmission. Recurrent DKA admissions at six months and 12 months occurred in 7.9% and 6.6% of patients, respectively. In multivariate logistic regression, patients with pre-existing Type 1 DM were more likely to have 6-month DKA recurrence than pre-existing T2D. Patients with a 6-month DKA recurrence, female gender, and Type 1 DM had higher odds of 12-month recurrence, whereas a consult with a diabetes educator at the index admission was associated with decreased odds of recurrence.

Conclusions: In this study, we showed that younger men with newly diagnosed T2D and comorbidities had a prolonged LOS with the index DKA admissions. Additionally, young, non-obese, insulin-non-compliant women with pre-existing T1D with a 6-month DKA recurrence were at a higher risk of 12-month recurrence. To our knowledge, this study represents the most diverse and the largest regional data concerning LOS and recurrence prediction in DKA. Results from this study may be valuable for both physicians and healthcare systems to decrease the diabetes-related healthcare burden in DKA patients.