Published on 

Ministers Donelly and Donohoe publish: Hospital Performance: An Analysis of Unscheduled Care Activity 2017 – 2022, as part of Spending Review 2023

Minister for Health, Stephen Donnelly, and the Minister for Public Expenditure, NDP Delivery and Reform Paschal Donohoe publish: Hospital Performance: An Analysis of Unscheduled Care Activity 2017 – 2022, as part of Spending Review 2023


Minister for Health, Stephen Donnelly, and the Minister for Public Expenditure, National Development Plan (NDP) Delivery and Reform, Paschal Donohoe, today (Thursday 21 September), have published: Hospital Performance: An Analysis of Unscheduled Care Activity 2017 – 2022, as part of the 2023 Spending Review process. The aim of this review is to better understand the composition of Emergency Department activity to inform our response to this demand.

Commenting on the publication, Minister for Health, Stephen Donnelly said:

“I am delighted to announce the publication of this paper which provides an unparalleled overview of activity within Emergency Departments and Local Injury Units across Ireland. These facilities are often the first port of call for people trying to access healthcare and I am very focused on delivering improvements. Examining patient level data offers deep insight into the operation of unscheduled care, including who goes to Emergency Departments, how unwell they are, when they arrive, how long they wait and whether they are admitted. 

It also identifies areas where change is possible – for example treating less acute patients, with little need of hospital admission, in the primary and community setting. This paper will inform and shape future response to pressure on our Emergency Departments including a new multi-annual year plan which the HSE are currently developing. This body of research shows how my department is actively improving performance oversight and activity forecasting.”

This paper provides an analysis of Unscheduled Care Activity at a National and Regional level over the period 2017-2022. The use of the Patient Experience time dataset has enabled this analysis to evaluate characteristics of patients presenting to Emergency Departments and Local Injury Units. In publishing this paper, it provides a detailed analysis of the Unscheduled Care system, presenting analysis which has never been published before including GP Out of Hours/ GP referral admission rates, the potential for alternative pathways such as Local Injury Units to offset emergency department pressures, and how a patient’s time of arrival can influence their wait times and their likelihood of admission. 


Notes to editor:


Paper Summary

  • This paper provides an analysis of patient level Unscheduled Care data recorded in the Patient Experience Time dataset over the period 2017-2022.


  • The paper provides numerous insights into patient characteristics and outcomes in Unscheduled Care environments, including in relation to their age, admission probability, clinical need, mode of arrival and whether they have a referral. The findings of the paper either have overt or indirect implications for strategic policy development for Unscheduled Care in Ireland.


  • The paper identifies data gaps for Unscheduled Care in Ireland, as well as opportunities for future research to further develop understanding. Where possible, the PET dataset should be leveraged to provide bespoke ex-ante evaluation of interventions to improve outcomes and reduce pressures in Unscheduled Care. In addition, the data environment should be improved with the provision of better data on patient clinical characteristics (via Urgency Related Groups), rectification of data gaps, and the provision of more extensive data on interventions and costs used within each Winter / In-year Unscheduled Care plan.



  • Unscheduled Care Demand: Unscheduled Care Presentations have risen from 1.1m in 2010 to 1.59m in 2022. Cumulatively, this represents a 45% increase in demand over this period, with the largest periods of growth between 2010 and 2012 (an increase of 200,000 presentations) and an increase of 250,000 presentations between 2021 and 2022.


  • Utilisation Rates: Per capita utilisation of emergency departments is concentrated among those aged 0-5 (54 visits per 100) and those ages 76+ (53 per 100 visits, or greater) relative to a presentation rate of under 30 visits per 100 for persons aged 6-75.


  • Admission Probability: Admission probability has a direct relationship with age. For example, a person aged 20 has a 14% rate of admission, compared to someone aged 80 with a 50% rate of admission.


  • Triage Score: Probability of admission is directly related to triage score. Less than 10% of patients at triage score 4 and 5 (standard and non-urgent) are admitted. Triage category 3 patients (urgent) are admitted 23% of the time. This has implications for which patients are likely best suited to treatment within community settings rather than in the ED.


  • Wait Times:  There is a strong relationship between wait times and age. For example, median wait times for under 5s are 3.5 hours, compared to wait times of 9 hours for persons over 85. This is likely driven by the triage scores of patients by age, as persons of a younger age are less likely to be admitted than older patients and are therefore able to leave an Unscheduled Care environment after initial assessment (which takes place within a maximum of four hours from arrival).


  • Hour of Arrival: There is some evidence that probability of admission and waiting times in ED are dependent on hour of arrival. Patients who arrive with a triage score of immediate (patients of highest acuity) wait on average one hour longer for admission / discharge if they arrive at night than during the day. Equally, patients with a triage score of immediate are 20% less likely to be admitted when arriving during the day, versus a nighttime arrival. 


  • Mode of Arrival: – Ambulance: Arrival by ambulance versus through other pathways is associated with a 20-25% increase in admission probability. This is likely attributable to the higher acuity of patients arriving by ambulance, and the pre-triaging of presentations via this pathway by paramedics prior to conveyance to hospitals.


  • Referral Type: – GP vs Self: 73% of referrals to Unscheduled Care settings from GPs do not result in an admission. This compares to 77% of presentations not resulting in admission for patients who self-refer. GP Out of Hours referrals are not admitted 75% of the time. This may indicate that GP referrals are not sufficiently effective at triaging patients for treatment in the community versus treatment in an acute setting. The even lower rates of admission of patients with Out of Hours GP referrals suggest inappropriate referrals to ED from this pathway. 



  • Triage Score Attendances by Health Region: There is substantial variation in the triage level (acuity) of patients arriving in each Health Region, especially low acuity patients. For example, just 8% of presentations at Health Region Midwest are triage category 4 (standard) or 5 (non-urgent). This compares to Health Region Dublin Southeast (DSE), where 33% of presentations are of the same category. This may be the result of the availability of alternative treatment options in some regions, with for example Health Region Midwest having 3 operating Local Injury Units, versus other regions having more limited LIU coverage.



  • Alignment of Acute Care Capacity Resourcing with Demographic Change: Our analysis has shown that the utilisation rate and admission probability of older patient cohorts is substantially higher than for other groups. This has direct implications for forecasting future Unscheduled Care capacity requirements at a national and regional level, as population ageing will likely increase demand pressures in both Unscheduled Care settings and Inpatient facilities. 


  • Wait Times for Older Age Cohorts/ Patient Flow: Our analysis demonstrates the long waits patients experience in ED prior to admission.  Further work is required to understand the barriers to timely admission for these patients, especially those from older age cohorts, to better align outcomes with HSE general and age specific (over 75) wait time performance targets. Policymakers and practitioners could explore a range of options to improve in this area, including more efficient assignment, management and discharge of patients to beds, the provision of dedicated treatment pathways such as Medical Assessment Units for older patient cohorts, the treatment of additional patients in the community or; or the provision of additional beds in some hospitals to alleviate capacity pressures.


  • Specialised Treatment of Paediatric Patient Cohorts: 61% of persons under 16 presented to Unscheduled Care settings outside of the Children’s Hospital Group in 2022. Given the relatively high utilisation rate and low triage scores of patients of this type, policymakers should consider whether these patients can be treated in alternative settings within the healthcare system, either in the community or segregated pathways for treatment within a hospital.


  • Factors Contributing to Variation in Outcomes by Hour of Arrival:  We demonstrate variation in waiting times and admission probability for patients by hour of arrival, even controlling for patient triage score. This poses a potential patient safety and efficiency issue and is therefore worthy of further exploration. Potential contributors to this outcome could include differences in patient profiles by hour of arrival, reduced diagnostic availability during non-core hours, differences in staffing levels and the availability of senior decision-makers on a 24/7 basis.


  • Targeted Intervention to reduce presentations from low-acuity patients: Our analysis demonstrates the sizeable proportion of patients (29%) who present to Unscheduled Care settings with low-acuity needs. These patients are largely of younger ages and are concentrated in certain Health Regions (Dublin Southeast, Southwest). This presents the opportunity for the development of targeted interventions to reduce the utilisation rate of Unscheduled Care settings by these groups, instead treating them in community or primary care settings. For example, investment in alternative care pathways such as Primary Care Centres or Local Injury Units could be concentrated in areas where lower triage presentations are most prevalent. 


  • Explore strategies to reduce the level of unnecessary referrals from primary care settings into Unscheduled Care: We demonstrate that patients who self-refer to ED are admitted at only a slightly lower rate (23%) than those who attend with a referral from a general practitioner (27%). This indicates unnecessary referrals to Unscheduled Care settings from primary care practitioners in some cases, and therefore represents an opportunity to reduce unscheduled care pressures through more effective referral decisions by primary care practitioners.


  • Continuous Evaluation of Target Investment in Unscheduled Care: This paper is a novel application of the existing PET dataset to support improved monitoring and performance management within the healthcare system. Policymakers should ensure that analysis of PET data is used to maximise the impact of investments in unscheduled care, improving patient outcomes, value for money among other outcomes. Interventions to improve unscheduled care performance should be the subject of continuous evaluation, with the objectives, inputs, outputs and outcomes associated with each measure analysed after their implementation to determine their relative impact.


  • Data Improvements & Implementation of Urgency Related Groups (URGs): To further enable the use of PET for strategic policy development and evaluation, we advocate for further improvements to the Patient Experience Time dataset to be prioritised within short- and medium-term strategic planning in this area. First, we advocate for the development and implementation of Urgency Related Group classifications within Unscheduled Care settings in Ireland as this would allow identification of the resource requirement and reason for presentation to unscheduled care for each patient. Second, data improvements to existing reporting should be sought, including improvements to unclassified characteristics for some patients within the PET dataset, and the collection and reporting of PPSNs for each presentation.