Statistical modelling of public hospital emergency department presentations in Australia

Project overview

Demand for emergency department (ED) care has increased rapidly over recent years in Australian public hospitals, with resultant ED overcrowding and longer waiting times for patients. Due to lengthy queues, some patients leave the ED without being seen by a doctor while others leave with partially completed care, both contributing to higher re-presentation rates and poor system performance. However, the factors associated with higher rates of ED presentations were poorly understood by prior studies. Hence, the first aim of the research is to identify demographic, clinical and time-varying factors/subgroups that impact the quantity of ED presentations.

In a fiscally constrained environment, optimal resource planning is one main approach of increasing the efficiency, and therefore sustainability of current health systems. Accurate forecasting of ED presentations can assist EDs to more optimally organise their resources and improve ED patient flow. Therefore, secondly, this research aims to develop forecasting models for ED presentations by previously identified factors/subgroups and compare the accuracy of forecasts produced by different statistical and machine learning methods.

Study setting

All ED presentations to four Australian public hospital EDs in Queensland’s Metro South Hospital and Health Service from January 2009 to December 2015. The EDs belong to:

  • Logan Hospital
  • Princess Alexandra Hospital
  • Queen Elizabeth II Jubilee Hospital
  • Redland Hospital.

Outcomes

Journal paper 1

Status

Published in Q1 journal (EMA).

Positive outcomes

Time-varying factors associated with higher rates of ED presentations were identified. Daily level forecasts were produced using a range of statistical and machine learning models. These forecasting models provide insight for researchers to actively engage with new methods over conventional ED forecasting methods.

Real-world applications

The daily level forecast is useful at the hospital level for strategic resource planning, planning of programmes to support ED patients, and other local decision-making processes, and also informing broader public policies and planning.

Journal paper 2

Status

In preparation for submission to EMA.

Positive outcomes

Distinct and latent subgroups of patients presenting to EDs were identified. Thereby, clinical, socio-demographic and time-varying factors/subgroups associated with higher rates of ED presentations were identified.

Real-world applications

The utilisation of this approach may influence the development of models of care specific to the care needs of individual subgroups that support the sustainable delivery of acute healthcare.

Journal paper 3

Status

In modelling stage.

Expected outcomes

Develop forecasting models for identified subgroups in objective 2 and compare the accuracy of forecasts using different statistical methods.

Real-world applications

Inform broader public policies and planning.

Project team

  • Ms Kalpani Ishara Duwalage (PhD candidate)
  • Dr Helen Thompson (Principal supervisor)
  • Dr Gentry White (Associate supervisor)
  • Dr Ellen Burkett (External supervisor): Emergency Department, Princess Alexandra hospital & Healthcare Improvement Unit, Clinical Excellence Queensland
  • Dr Andy Wong (Collaborator): Emergency Department, Princess Alexandra hospital