Using emergency department data to define a ‘mental health presentation’ – implications of different definitions on estimates of emergency department mental health workload

Introduction

Mental and behavioural conditions are the most common chronic conditions, affecting over 6.7 million (26.1%) Australians.¹ According to the Australian Institute of Health and Welfare (AIHW), in 2021–22, there were 280,176 mental health-related presentations to public Australian emergency departments (EDs), accounting for 3% of all presentations.²

With rising levels of mental health-related ED attendances, there are concerns that current primary care and ED systems are unable to address the acute healthcare needs of the community.³ The increase in ED mental health attendances has been attributed to deficiencies in community care, inadequate availability of inpatient mental health beds and shortcomings in services related to housing, drugs and alcohol.⁴ Accurate measurement of mental health presentations is important to inform EDs, governments and mental health advocacy groups. A clear understanding of mental health workload in EDs is critical to plan for clinical demands and to enable equitable allocation of resources.

Various definitions are utilised to classify ED mental health presentations, most of which are based on ED diagnostic codes.^5–7 According to the AIHW, an ED presentation related to mental health is defined as one where the ED diagnosis code corresponds to a specific list of The International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Australian Modification (ICD-10-AM) codes (F00–F99).^5–7 The ‘F’ codes include presentations such as dementia, delirium, mood disorders, psychotic disorders and eating disorders.⁸

However, there are a proportion of diagnoses that may fall into the mental health category but may not be classified as such, as they cannot be attributed to a code between F00 and F99.⁸ For example, ICD-10-AM codes X60–X84 refer to intentional self-poisoning or means of intentional self-harm.⁹ Wrist lacerations, carbon monoxide poisoning and other possible self-harm presentations are often coded as the primary injury, rather than the underlying mental health concern.⁹ Other ways to define ED mental health presentations include the presenting complaint recorded at triage and/or whether individuals are referred to psychiatric services by ED staff.

As the number of patients seeking mental health assistance in EDs rises, it is necessary to review current mental healthcare models. An initial step is to establish a precise definition of what constitutes an ED mental health presentation.^4,5,9–13

This study aimed to describe and compare the proportion of patients classified as an ED mental health presentation under different definitions and examine the agreement between these definitions and the AIHW definition.

Methods

This was a retrospective cohort study of all patients presenting to the EDs of a multi-hospital metropolitan Victorian health service which comprised one tertiary hospital (Monash Medical Centre) and two urban district hospitals (Dandenong Hospital; Casey Hospital). These hospitals were part of Monash Health, Victoria’s largest health service and provided care to one-quarter of Melbourne’s population.¹⁴

Results

Over the three and half years of the study, there were 813,078 presentations to ED of which 790,777 had a presenting complaint recorded and 757,657 were given an ED diagnosis corresponding to a code from the ICD-10-AM (Fig. 1). A small proportion of patients (33,120) were unable to be included in the ICD-10 code definitions as they did not receive diagnostic codes during the visit. There were 22,301 patients who were not assigned a triage presenting complaint and hence could not be included in the analysis for the definitions utilising the presenting complaint recorded at triage. A total of 45,376/813,078 (5.6%) patients were seen by or referred to EPS services, while 34,248 (4.2%) of all ED presentations met the AIHW definition for an ED mental health presentation.

Fig. 1.

Allocation of ED presentations according to each definition of ED mental health presentation. ED, emergency department; ICD-10-AM, The International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Australian Modification. All other abbreviations (e.g. F00–F99, X60–84, R45, R46, S and T codes) are specific diagnostic codes of ICD-10-AM.

Table 2 provides an overview of demographic data related to each definition of an ED mental health presentation. All definitions resulted in similar overall demographics, including a median age between 29 and 33 years, and a slight female predominance. Approximately 30% of ED mental health presentations resulted in admission to a hospital ward outside the ED short-stay unit.

Fig. 2 demonstrates the overlap between the definitions. There were 45,376 patients who were referred to and/or seen by EPS, however, approximately half (52%, 23,615/45,376) were not classified as a mental health patient by AIHW according to the patients’ diagnostic coding (Fig. 2c).

Fig. 2.

Overlap between the AIHW definition and different definitions of an ED mental health presentation. (a) Comparison of definitions reliant on ICD-10-AM diagnostic codes. (b) Comparison with definitions based on presenting complaint recorded at triage. (c) Comparison with a definition reliant on whether or not patient was referred to or seen by EPS. (d) Comparison between AIHW definition, triage definition and mental health review definition.

Patients who presented to triage with a mental health-related complaint were seen by EPS 92% of the time (33,504/36,160), but only 57% of these patients (19,143/33,504) were assigned an ED diagnosis code according to the AIHW definition (Fig. 2d) of an ED mental health presentation.

The most frequent mental health-related presenting complaints and diagnostic codes were similar across the six different definitions (Table 3). The top four presenting complaints in all diagnostic codes included self-harm, acute behavioural disturbance, Section 351 (a section of the Victorian Mental Health Act 2014 which provided a police officer authority to transport a person who appears to have a mental illness and appears to be at risk of harm to hospital¹⁶) and ‘other’ mental health complaints. The most common diagnoses recorded were similar across all definitions and included depression, suicide attempt, schizophrenia, psychosis, alcohol intoxication and anxiety.

Kappa statistics were calculated between the AIHW definition and other definitions. Definitions incorporating ICD-10-AM ‘F-codes’ as well as additional diagnoses had strong agreement (Definition 2: kappa (95% confidence interval (CI)) 0.84 (0.83–0.84) and Definition 3: 0.86 (0.86–0.86)). However, agreement between AIHW definitions and definitions using triage presenting complaint was lower, with Definition 4 (a broad triage definition) having a kappa (95% CI) of 0.61 (0.60–0.61) and Definition 5 (a narrower definition) 0.58 (0.58–0.59). Agreement between the AIHW definition and those referred to and/or seen by EPS was only 0.58 (0.57–0.58). Table 4 provides three examples of patients who fulfilled some, but not all definitions of an ED mental health patient.

Discussion

This study of over 800,000 ED visits has demonstrated that the definition of an ED mental health presentation currently used by the Australian Government appears to underestimate the true burden of ED mental health presentations. Nearly half of all patients referred to or seen by EPS were not classified as an ED mental health presentation using the AIHW definition, suggesting that there is also an element of misclassification.

Accurate data on ED mental health presentations is critical for a number of reasons which include appropriately informing service planning, identifying gaps outside the acute hospital system and ensuring funding is allocated to areas where the need is greatest.¹⁷ Epidemiological studies may identify changing patterns of presentation over time and have been used to highlight the increase in mental health presentations for children and adolescents in recent years.⁹

Inconsistent definitions of ED mental health presentations between studies has been highlighted as a concern in a previous systematic review.¹⁷ The review concluded that there was ‘a lack of high quality, generalisable epidemiological data available to inform service change and the development of new models of care.’¹⁷ Key recommendations from the review included ensuring a large sample of patients across a number of different EDs to better inform future health policy.

All of the definitions used that apply diagnostic codes (the AIHW definition, and Definitions 2 and 3 in our cohort) include all ‘F-code’ ICD-10-AM diagnoses. Although F codes are labelled as mental and behavioural disorders, they also include common reasons for ED attendance that are not traditionally considered mental health presentations, including dementia (F0.0, F0.1, F0.2), post-traumatic amnesia (F0.4) and acute intoxication of alcohol (F10.0).⁸

Notably, F10–F19 comprise ‘mental and behavioural disorders due to psychoactive substance abuse’, and include intoxication with alcohol, opioids, cannabinoids, sedatives or hypnotics, cocaine and other substances. It is therefore possible that, if applying the AIHW definition, a significant proportion of ED mental health presentations may relate to recreational intoxication, rather than an acute mental health condition (such as depression, suicidality or psychosis).

The use of electronic medical records is increasing across Australia.^18–21 However, even in systems without comprehensive electronic medical records, minimum datasets exist in a number of jurisdictions, for example, the Victorian Emergency Minimum Dataset¹⁵ and the mental health-focused Client Management Interface/Operational Data Store.²² Both datasets are subject to considerable scrutiny, with regular data queries and reporting. We believe that linkage of routinely collected ED data and mental health data would provide a much more accurate picture of ED mental health presentations across jurisdictions. If this is deemed not feasible in the first instance, then collection of data on patients who are referred to and/or reviewed by EPS would appear to be a robust, sensitive and specific way to identify ED mental health presentations.

Accurate identification of ED mental health presentations would then provide a clear pathway to a better understanding of epidemiology (including age, socioeconomic status, ethnicity, reason for attendance, use of restraint and sedation, destination on discharge and re-presentation rates). This data is critical to inform future health system planning and to develop reliable measurements of the processes and outcomes of care.

Our study was limited by the quality of the available administrative data. Inaccurate diagnostic coding may occur, as most ED diagnoses are entered by busy clinicians with multiple competing priorities and limited training in healthcare coding.²³

Although we had a large sample size of over 800,000 patients, our study was conducted at a single Victorian health network. It is possible that our results only reflect local practices, which may differ between hospitals and jurisdictions. To better define an ED mental health-related presentation on either a state or national level, it is important to repeat this study across a larger number of hospitals and health services across Australia.

Conclusion

The AIHW definition is a conservative measure of ED mental health presentations and may underestimate emergency psychiatry workload. There is moderate agreement between the AIHW definition and ED presentations where individuals are referred to and/or seen by EPS. Reporting of patients who are referred to and/or seen by EPS may provide a more accurate measure of mental health workload in Australian EDs.