Attempts to upgrade or reorganise overcrowded emergency services worldwide come to nothing. Three researchers, including NEOMA’s Mozart Menezes, suggest the answer may lie on decision-making tools powered by new monitoring indicators based on measures of operational complexity. These indicators have been identified by analysing 145,000 visits to the emergency department of a major hospital in Canada.

How can we quantify the level of overcrowding in an emergency department (ED)? What are the most useful indicators for forecasting patient waiting times? What factors can be exploited to cut the wait while maintaining a high level of care?

Key problem: patient arrivals are hard to predict

Overcrowding is an issue that has haunted healthcare professionals, ED supervisors, public authorities and experts in organisational optimisation for decades. Attempts to improve wait times have never achieved the expected results. And ED congestion reached epic proportions during the Covid-19 pandemic.

What, according to the three researchers, lies at the root of the problem? Current strategies are based primarily on the following ratio: the number of patients expected / the number of healthcare staff required. And yet, ED footfall is by its very nature hard to predict, even though historical data can be used to give an estimate of waiting times. Any sudden influx of patients causes long-lasting disruption to the system.

Indicators that reflect the complex ED environment

The authors of the article adopted a different approach, setting out to identify indicators that factor in the highly complex management of EDs. The number of patients fluctuates from hour to hour and day to day, as do their arrival patterns. They are prioritised according to age, symptoms and level of acuity. They are then directed to different healthcare members of staff depending on their pathology, which itself decides how long the treatment will last.

The researchers argue that patient flow or congestion in an ED is influenced more by the effective management of these unpredictable events than the simple number of patients admitted. They were keen to check this hypothesis by mining a data set of 145,000 ED visits in a major hospital in Canada.

The research team decided to measure two categories of unpredictable event which, in their opinion, are the most impactful: time-related (the interval between successive patient arrivals and the length of treatment for each patient); and case-diversity (age of the patient, symptoms, acuity, etc.).

Better forecasting of patient waiting times

Based on the analysis performed on the Canadian data, these “time complexity” and “case complexity” metrics are much more effective than the number of admissions for forecasting the average wait time and the total time spent in EDs. In other words, they are a more reliable way of assessing (and predicting) the service load and, it follows, how the ED should be organised to cope with it.

The greater the time complexity, the higher the risk of overcrowding. This complexity results, for instance, either from grouped and sporadic patient admissions rather than spaced out, steady arrivals; or from the influx of cases needing long-term care rather than patients who can be treated swiftly.

Conversely, the greater the case complexity, the lower the risk of overcrowding. High case complexity means that patients attend the ED with very diverse pathologies, which can then be assigned to healthcare staff with different areas of expertise. This is a much more positive scenario than an influx of patients with a number of identical pathologies (i.e. low case complexity). When this is the case, some of the healthcare staff become overstretched while their colleagues in other specialties are under-employed.

A tool for improving patient prioritisation

The advantage of these complexity measures is that they draw on patient data that the emergency services already collect. In other words, they do not require any extra work. In addition, it is easy to incorporate these measures into the existing structure and procedures.

What is the first potential practical application? Fine-tuning the strategy for prioritising patients. Although this is always based on the severity and acuity of the pathologies observed, we can also conceive of a second criterion: the predicted wait time, which is calculated using complexity measures. Take the following example: for similar severity levels, a patient who can be treated quickly will be seen before a patient requiring more time, even if the latter was the first to arrive; this means that the member of staff will be available for another patient more quickly. More importantly, managing the diversity of different patients undergoing treatment at a same moment may reduce the waiting time to all patients waiting.

Alerts for reassigning resources before overcrowding

The second potential application would involve integrating these complexity measures into a decision-making tool connected to the registration system for admitting patients when they arrive in the ED. This tool would be powered in real time throughout the patient journey, including the waiting room, consultation, treatment, analyses, etc. It would trigger alerts when the time complexity or case complexity level approaches predetermined critical thresholds. The ED supervisor would then reallocate resources before wait times start to get out of hand: raising the priority level of some pathologies in order to coordinate the workload of healthcare staff; asking for more personnel; or tweaking the allocation of multi-skilled members of staff.

Does this mean, then, that patient / healthcare staff ratios should be discarded? The answer is no: they are still useful… but not enough in themselves to make ED overcrowding a thing of the past. Complexity measures based on the pattern and volume of patient arrivals, in tandem with the level of diversity of their pathologies, are promising levers for breaking out of the waiting time crisis.

Find out more

Taiwo, E.S., Zaerpour, F., Menezes, M.B.C. and Sun, Z. (2023), “A complexity-based measure for emergency department crowding“, International Journal of Operations & Production Management, Vol. ahead-of-print No. ahead-of-print. https://doi.org/10.1108/IJOPM-12-2022-0792