Whole-of hospital occupancy management through the ED
First published on Australian Hospital + Healthcare Bulletin – September 5th, 2016
With millions of patient records now sitting in the healthcare system, health informatics researchers have been able to transform this data into an effective patient management system, incorporating emergency department admission predictions, patient flow throughout the hospital and even predicting readmission rates for regular patients known as ‘frequent flyers’.
Starting at the door
Approximately 20-25% of patients start their journey through the health system by arriving at an Emergency Department in an ambulance. Ambulance services often maintain separate databases from hospital admissions services, and identify patients in different ways5, which is where the research first began.
The Patient Admission and Prediction Tool (PAPT) was developed at the Australian e-Health Research Centre by CSIRO in partnership with Queensland Health, Griffith University and Queensland University of Technology.
The web-based PAPT delivers real-time forecasting based on admissions data from previous years in similar circumstances – for example, prediction of admissions for the Easter long weekend would be based on what admissions had been received at the target’s own site and surrounding medical facilities for up to 10 years. Allowances are made for other factors such as specific events (large public events), current or additionally forecasted outbreaks of communicable illnesses and commonly repeating patterns (day of week, other identifiers that hospital bed managers can easily pinpoint for their facilities). The Gold Coast University Hospital tested the system at Schoolies Week to great success.
After developing the model at two Queensland public hospitals over 2002-2007, the research team ran a further five-year study across 27 additional Queensland facilities representing approximately 95% of the ED presentations across the state. Mean daily presentation rates ranged from 60 to 190 patients/day and it was in the regional facilities with lower ED presentations that something important was discovered: to forecast a particular category of interest, there needed to be roughly more than 10 admissions or presentations per day.
It was also found that poorer forecast performance was experienced over winter months, particularly the winter of 2009 (and to a lesser extent 2007), which correlates with significantly increased influenza-like ED presentations experienced across this season…