Which features of EM impacts on safety?

Most patient safety incidents are caused by a series of individual factors contributing to a chain of events leading to the incident. It is important to understand and appreciate how each contributing factor is related to the final event, even if individual factors may seem insignificant when looked at in isolation. Methods such as root cause analysis (see online supplementary appendix 1) can identify contributory factors and themes leading to safety incidents; these factors may be classified generically15 or in an ED-specific context.16

Adverse events related to missed diagnosis, for example, are typically the result of failures in the diagnostic process and are usually due to several contributory factors.17 Given the uniquely complex nature and high-risk9 operating characteristics of EM, combined with often incomplete patient information, it is not difficult to appreciate how these make the speciality one which may be described as prone to error.18 Although there have been no studies which identify specific features of the ED which make it particularly susceptible to error, several authors have described common factors which intuitively make errors more likely to occur when compared with other care settings.19 ,20 Some of these factors are common to other non-medical settings—the similarities to the aviation industry being commonly cited.21 ,22 Table 1 summarises the factors that have been suggested in the literature and solutions that have been proposed or trialled to address them.

To the outsider, the ED can appear to be a chaotic and stressful environment, particularly in the context of patient safety. It is constantly in a state of flux, with inconsistent patient flows and widely variable patient acuity.27 Patients may move between several locations during a single episode, for example, from the waiting room to a cubicle, radiology and observation unit all in the space of a few hours. ED staff shifts often overlap, so a number of healthcare providers may interact with a patient during an attendance with multiple handoffs. In addition, visitors, ambulance staff and transient ED and inpatient clinicians are often present in the main ED at any given time. Strong team-working, high physician accessibility and flexible task allocation have naturally evolved in EDs to adapt to the unpredictability and variability in demand and working environment. Teamwork training has been shown to reduce errors24 but the demands placed on staffing EDs mean that clinician team members are often unfamiliar with each other.28 Relatively fixed additional specialist staff can impact on safety, for example, pharmacists working in the ED are associated with a reduction in medication incidents.25

ED patients are mainly high acuity, and this is associated with a higher incidence of safety incidents.4 ,23 Undifferentiated or non-specific patient presentations as well as atypical symptoms make the diagnostic process particularly challenging. An increase in non-urgent use of the ED also contributes to the likelihood of error as it contributes to the increasing volume of attendances and increases the demand on fixed resources, thereby reducing the time available per patient encounter overall.29

Information gaps are common in patients presenting as emergencies, as primary care, nursing home or inpatient notes are often unavailable on presentation.30 ,31 This uncertainty, coupled with the time limits for assessment and decision-making, increases the likelihood of error, as often clinicians only have a fraction of the history on which to base decisions.10 System-wide integrated patient information have been proposed as a solution; however, these have only been implemented in few settings.

EPs often simultaneously deal with a multiplicity of tasks and often manage several patients of varying acuity at the same time as supervising the delivery of patient care by more junior staff. Along with high patient volumes, time pressures and cognitive load, the huge number of decisions a EP needs to make during a shift leads to a high likelihood of error occurring.18

Handoffs are recognised as a potential hazard32 with the care of an individual patient potentially involving several handoffs due to crowding and prolonged ED length of stay (LOS). Interruptions are common in the ED and may lead to increased risk to patients.33 ,34 Combined with workload stress, this can cause a breakdown in safety and risk management mechanisms.35 Finally, shift work and resulting sleep deprivation have been shown to lead to cognitive impairment related to disruption of normal circadian rhythms.27 Interventions using electronic tools36 to assist with handoffs and reduce cognitive load have been proposed. Professional societies have also recognised the importance of improved working conditions in sustaining EPs careers and in safety.37

Many EDs are staffed predominantly by training grade physicians with a range of levels of experience. Increasingly, staff shortages mean that the use of temporary short-term staff is common in some settings. Staff inexperience is thought to be associated with preventable error.4 ,18 Most emergencies are relatively rare; therefore, exposure and practice is accumulated over a longer period than more routine and common conditions. The quality of an ED clinician's decision-making is correlated with experience but the high workload and increasing dependence on few trained EPs to deliver the majority of care means that clinical supervision of inexperienced staff may be compromised. Enhanced supervision of trainees is unsurprisingly associated with better patient safety outcomes.38 Comprehensive induction of clinical staff is well established in EDs and this has intuitively led to similar but abbreviated systems of induction for locum staff. Many departments have developed standard guidelines as well as safety net mechanisms for high-risk presentations, with interventions such as telephone follow-up23 and trainee feedback shown to reduce adverse incidents.

The demands on the ED have no upper limit, and it has been described as being ‘infinitely expansible’.10 Crowding has become a universal problem faced by EDs and is one manifestation of increased demands placed on the whole healthcare system. ED crowding has been shown to be associated with increased inpatient mortality, unplanned returns and preventable errors.39–42 Boarding of admitted patients in the ED has been well described and is associated with an increase in safety incidents.43–45 EDs are dependent on the efficiency of other parts of the healthcare system for their efficient functioning. The ability to arrange follow-up in primary care, access investigations or admit patients is reliant on other providers and mostly out of the control of EPs. This lack of control can make transfer of care problematic, and cause delays in ongoing care with an increase in the risks involved. ED-based solutions such as regular clinical rounds26 may reduce the risk of adverse incidents. However most solutions to crowding and boarding are system-based, with economic incentives along with bed occupancy and demand management the commonly used methods.44 Observation units have been widely used to address issues with crowding and access block, but there is no evidence that they improve safety.46 Time targets which focus on the ED in isolation have similarly failed to demonstrate an improvement in patient safety.47

Handover and referral in particular has been shown to be a particular area for potential safety incidents.48 ,49 Paradoxically, the relative dependence on other providers also results in a lack of feedback on the outcomes of ED care which compromises learning, adoption and maintenance of safety practices and skills.50 Structured handover51 and educational interventions48 to improve care transitions have been proposed to minimise safety risks associated with these transitions. Mandatory routine follow-up of patients by trainee EPs is common in some EM training programmes, but this is not universal.

Given the range of factors identified which impact on ED safety and the demonstrated variability in measures reported in the published literature, it would seem to follow that additional measures which reflect contributory factors may provide more useful information on ED safety than is currently available. Professional societies, such as the UK Royal College of Emergency Medicine, have developed safety toolkits which set about to describe the structures, processes and skills which characterise a ‘safe’ ED,52 with metrics such as missed diagnosis rate, staff vacancy rates, information and communication technology system reliability and occupancy/boarding rates proposed to provide objective measures of ED safety. Regulatory bodies such as the US Joint Commission and Centres for Medicare and Medicaid Services have used measures such as notes availability, LOS and evaluation times in a similar way.53 In many cases, however, quality standards and indicators are used as surrogate measures of safety. For example, in the UK54 and USA, quality measures such as the aforementioned LOS, unplanned return visits and left without being seen (LWBS) rates have been used as markers of safety. Examination of the validity of these markers demonstrates that although LOS is associated with a higher risk of short-term adverse events, LWBS is not.55 Data suggest that less than 5% of all return visits are associated with a safety incident. More recently, LOS has been challenged as a safety metric56 underlining the need for a fuller understanding of indicators and potential high-impact interventions.

Error, harm and safety incidents

Although the patient safety movement has been active for over a decade, there is still much variability in how information on incidents is captured, reported and classified. This is partly due to differences in how safety events are defined across different settings and countries and in the nomenclature surrounding patient safety. International consensus on the terms and classification of safety has been published,57 though many reports still use traditional terms. This variability in terminology and reporting methods makes comparing and aggregating data from across different systems challenging.

In the UK, patient safety terms have been further simplified to allow consistency (box 1).

Much of the ED patient safety literature has focussed on error, with the assumption that error intuitively leads to harm. However, it is increasingly recognised that strategies to improve patient safety should focus on preventing errors per se as well as on preventing adverse events. Not all errors occurring in the ED lead to harm and not all harm is due to errors. Studies have consistently demonstrated a much higher incidence of error in EDs than of actual adverse events related to these errors.61–63 In other words, the majority of ED errors do not result in an adverse event. This is reflected in the experience of the airline industry, where upwards of two-thirds of flights have reported errors but significant adverse events are rare.64

There is also the perspective that recognition of small recoverable errors is important in building resilience into systems, as they provide insight into where the boundaries of safe performance lies.65 ED clinicians may make several small errors during a procedure or patient assessment, which have no negative outcome, but recognising such errors alerts them to the need to improve and to avoid more costly errors in the future.

Between 55% and 82% of adverse events occurring in the ED can be judged to be preventable or avoidable.3 ,66 This figure is higher than reports from other areas of clinical practice3 ,67 and supports the idea of EDs focussing and learning from safety incidents as well as from errors which may not have resulted in an adverse event. This approach relies on using reactive methods of analysis such as root cause analysis and significant event audit58 which facilitates the allocation of resources towards mitigating or preventing incidents assessed to be of the highest potentially preventable risk to patients. Proactive approaches such as failure mode and effect analysis are promoted by some authors as a means of preventing safety incidents identified by the evaluation of a process.68 These techniques tend to be resource-intensive and error-focussed but may identify some risks which are not revealed by reactive analysis. A summary of the above methods can be found in online supplementary appendix 1. Complementary data from other sources such as complaints, medicolegal or inquest cases may enhance understanding and identification of safety issues.17 ,69

How safe is the ED?

Published reports of the frequency of hospital errors and adverse events are summarised in table 2. Two of the studies compared the frequency of ED errors with other specialties and in these studies the frequency of errors in the ED was lower than in surgery, medicine or obstetrics and gynaecology. However, these data were all derived from retrospective reviews of inpatient notes and thus only included those patients admitted from the ED. Events occurring in non-admitted patients, the majority of ED patients, were not analysed. Retrospective case analyses are also more likely to find errors; hindsight bias may overestimate the incidence of adverse events as the error may not have actually caused an adverse outcome. In addition, it may influence how the identified errors are attributed to the adverse outcome.65

There have been several studies looking at the frequency and incidence of errors and adverse events in the undifferentiated ED patient population (table 3). The reported adverse event rates range from 0.36% to 3.26% of all ED attendances and 0.9% to 6% for discharged patients. Higher adverse event rates (4.1%–5%) have been demonstrated in higher risk conditions, in higher acuity areas and in higher triage categories.23 ,70 Adverse events were also more common in older patients,61 ,70 although this may be related to complexity and higher risk presentations.

Overall, the most common types of adverse events occurring in hospital inpatients were related to surgery and medication,5 while in the ED setting, adverse events related to diagnostic issues and suboptimal management plans17 ,61 ,66 ,70 predominate.

The preceding data must be interpreted in the light of the challenges faced by investigators and clinicians measuring ED patient safety in deriving an accurate incident rate.75 Safety incidents in the ED are relatively uncommon occurrences even if errors are common. There are significant variations in how ED safety events are defined, identified and classified, with several studies using non-standard methods of case finding and with individual interpretation of outcomes and contributory factors. Most of the reported ED studies used a range of prospective methodologies to derive safety incident rates even though retrospective notes analyses may be well suited for determining such rates.73 Prospective studies can actively seek errors which may overestimate the likelihood of a safety incident resulting. In addition, less significant errors are more likely to be reported. For practical reasons, studies may only be carried out for a relatively short time or at intervals, hence there is a chance of missing important safety incidents.

Furthermore, most patient safety surveillance systems rely on self-reporting. It is recognised that incident reporting systems may be poor at identifying patient safety incidents, particularly those resulting in harm.76 The population at risk in the ED is relatively undefined and varies with age, acuity, presenting complaint and outcome, whether admitted or discharged. Some safety incidents may be attributed to the ED when they may have in fact involved other clinicians or external contributory factors. Studies suggest that only about half of safety incidents occurring in the ED are attributable to ED staff or processes.67 ,77 Finally, the time period from exposure to adverse event is difficult to specify or predict, although the majority become apparent within 3–10 days of discharge from the ED.23 ,25

Although most routinely collected safety incident information is dependent on self-reporting, aggregated data may use complaints, inquests and legal proceedings to provide a fuller representation of safety.78 Medicolegal claims, in particular, can provide a rich source of information.17 Analysis of health service negligence claims can provide some insight into EM claims compared with other specialities. EM accounts for 12.2% of all UK claims (table 4), which is almost identical to Australian public sector EM claims (12.1%, 2008–2013).79 EM has the lowest estimated cost per claim than all listed specialities except psychiatry, and this ranking almost exactly mirrors figures from the USA (approximate median EM payment, $90 000, median for all specialties $112 000).80 The US data also demonstrate that the proportion of EPs facing a claim (7.7%) is lower than most listed specialities (marginally higher than anaesthesiology, which has a higher proportion of claims paid). Australian estimates based on 2013 data are similar, with 8.7% of practising EPs facing a claim.79 These patterns may reflect more minor claims or perhaps are due to inherently safe practise which minimises significant harm.

The more recently available metrics seem to support the older data presented in table 1 and suggest that, when compared with other clinical areas or specialties, the ED is not particularly unsafe. The intuitive implication of this finding is that EM has evolved safe practices to mitigate against error or adverse events. It may also be that the current metrics used in measuring safety are inadequate. Complementary multi-source safety data using reactive analysis of complaints, safety incidents and legal cases combined with proactive collaborative system analysis may provide earlier, more sensitive identification of safety issues.

The variability in reported ED adverse incident rates would reinforce the idea that an understanding of the context and working environment of an individual ED is important in improving its safety profile. Measures to improve the safety of EDs must involve a collaborative approach with other parts of the acute hospital system in order to impact on safety incidents which occur in the ED but are attributed to other parts of the system. There are, however, common characteristics as discussed earlier which all EDs share that serve as a useful starting point in understanding how patient safety incidents may be addressed.

Next steps

Standardised methods and metrics for defining, identifying and measuring ED safety will facilitate the development and evaluation of interventions to improve patient safety. Research into the utility of new and existing pragmatic and contextual metrics as well as the effectiveness of implemented interventions or processes is needed. Consensus recommendations for ED safety research have been published81 ,82 which reflect these and other areas for future work. There is some suggestion that current and proposed safety indicators, particularly those used by regulatory bodies primarily as quality markers, may need further validation.