Background: Diabetic ketoacidosis (DKA) is a serious complication of diabetes with significant morbidity and mortality. Infection is a common precipitating factor of DKA. Investigation for an infective focus with chest radiography (CXR) is present in all DKA management protocols but the benefit of CXR in finding the precipitating cause of DKA is unknown.

Aim and rational: To audit the number of CXRs performed in DKA evaluation, the indications for CXR, and its yield of a clinically significant finding.

Methods: A retrospective audit of all DKA admissions from 1^st January 2015 to 31^st December 2015 at Flinders Medical Centre was conducted. Admissions were identified from the electronic medical system using the search term “diabetic ketoacidosis”. Admissions under the adult endocrine unit with confirmed DKA defined as a plasma pH of <7.30, glucose of >11.0 mmol/L, and positive serum or urinary ketones were included in the study. CXR requests and results were obtained from the electronic medical records, and the precipitating factor for DKA was identified from the electronic discharge summary.

Findings: A total of 92 DKA admission episodes were analysed in this study. Seventy two CXRs were performed with only 13% requests documenting respiratory symptoms or abnormal chest examination findings. 29% CXRs reported abnormal findings but only 4% (n=3) identified infective cause.  Two patients had confirmed pneumonia and both had significant oxygen desaturation on oximetry that was not documented in the radiology request form.

Conclusions:  Routine CXR without detailed history and thorough clinical examination may have a low yield for identifying an infective focus in DKA patients without respiratory symptoms.  Consideration should be given to revising DKA guidelines, to remove the recommendation to perform routine CXR in this context in favour of a more targeted approach based on presenting features, thereby reducing health costs and unnecessary radiation to DKA patients.