The HYPER in Emergency medicine
- Jesse ReeceEpidemiology and geospatial mapping of patients with HYPerglycaemia presenting
to Liverpool Hospital Emergency depaRtment – the HYPER Study
Objective: To use routinely collected hospital administrative and clinical data to investigate the epidemiology of
hyperglycaemic emergency department (ED) patients, and to use these data to develop a novel statistical model
for identification and prediction of hyperglycaemic patient characteristics.
Design, participants and setting: A 12-month retrospective period prevalence study of random blood glucose
(RBG) was performed using ED patient presentations at a tertiary hospital from 1st January 2018 to 31st
December 2018. Demographic and clinical data was acquired from hospital databases.
Main outcome measures: Prevalence of hyperglycaemia among ED patients, identification of demographic
patterns associated with these presentations, and accuracy of the predictive performance of our statistical model
to identify these patients.
Results: 64% of ED patients with an RBG result were hyperglycaemic. Patient groups more likely to present
with hyperglycaemia were culturally and linguistically diverse (OR = 1.59, 95% CI, 1.55-1.68; p < 0.0001) and
male (OR = 1.44, 95% CI, 1.39-1.50; p < 0.0001). Hyperglycaemic patients were also older than
normoglycaemic patients (median difference 14.99 years, 95% CI, 14.50-15.48; p < 0.0001). The ability of a
Generalised Additive Model to predict hyperglycaemic patients was assessed by a receiver operator curve and
produced an area under the curve of 0.72.
Conclusions: Routinely collected ED data and RBG testing administered as part of routine care identified
geographical clusters and demographic characteristics of patients more likely to attend ED with hyperglycaemia.
The predictive model provided an accurate and convenient method to estimate diabetes risk, which could assist
in effectively directing resources for long-term management in the community.