Who is less likely to die in association with improved National Emergency Access Target (NEAT) compliance for emergency admissions in a tertiary referral hospital?

Many studies have shown that overcrowding in emergency departments (ED) is associated with prolonged length of ED stay and increased mortality for patients admitted acutely via the ED.^1–4 Postulated causal mechanisms that have attracted remedial interventions include delayed or missed investigation results, medications or therapeutic interventions,^5–7 delayed time to consultant review and definitive management^8,9 and poor transitions of care across the ED–in-patient interface.^10,11 In a recent study, we demonstrated an association between improved compliance with the National Emergency Access Target (NEAT), defined as discharge from the ED within 4 h of presentation, and lower all-cause mortality among patients admitted from the ED to in-patient wards of a large adult tertiary hospital.¹² As NEAT compliance improved from 32% to 62% in response to multiple interventions over a 9-month period, mean length of ED stay for admitted patients decreased significantly from 637.8 to 404.4 min. This reduction in time spent in the ED was associated with a significant decrease in crude mortality rate for admitted patients from 2.3% to 1.0% and a fall in the hospital standardised mortality rate for emergency admissions (eHSMR) from 93 to 72.

Other studies, both overseas and in Australia, have confirmed a similar association. For example, Kelman et al.¹³ examined hospitals across 155 UK National Health Service (NHS) trusts and found that as the percentage of hospitals achieving more than 98% compliance with the <4 h target increased from 1.24% to 59.4%, the mean in-hospital mortality rate for emergency admissions across all hospitals decreased from 1.17% to 0.84%. In Western Australia, Geelhoed et al.¹⁴ studied six hospitals in Perth (three tertiary and three secondary hospitals) following the introduction of the 4-h rule and found that improvements in access block and ED overcrowding were associated with an overall significant decrease in mortality rate from 1.12% to 0.98% in tertiary hospitals. Secondary hospitals had no significant improvement in access block or mortality.

As far as we are aware, no study to date has examined which subsets of patients demonstrate the most benefit in association with improved NEAT compliance in terms of reduced risk of in-hospital death. Much of the literature has focused on patient outcomes associated with access block in the ED of >8 h,^15,16 which may be less relevant to the current target of <4 h.^13,14,17

The aim of the present study was to explore patient and system of care predictors of in-patient death for emergency admissions following the introduction of the 4-h rule at a 640-bed tertiary hospital.

We retrospectively analysed routinely collected data from the ED Information System (EDIS) for the calendar years 2011 to 2013 to identify and characterise patients admitted via the ED and nominated years 2011 and 2013 as the pre- and post-NEAT intervention periods, respectively, with 2012 being classed as a transitional year. Routinely collected in-patient mortality data from 2011 to 2013 were obtained on all emergency admissions according to their medical record number (MRN). Provisional diagnoses at time of ED discharge to in-patient ward were generated from the Primary Diagnosis International Classification of Diseases version 10 (ICD-10) codes recorded in EDIS. Diagnostic and palliative care codes at time of death were derived from ICD-10 codes and acute or non-acute care classification recorded in hospital discharge abstracts.

Patients eligible for study were those admitted via the ED to an in-patient unit. Patients who were admitted to the ED short-stay ward and then discharged home, those who died in, or were discharged home from, the ED or those who were transferred from the ED to another hospital were excluded from the study. Patients were classified as medical or surgical according to the destination in-patient unit. Patients admitted to the psychiatric unit were excluded from the study. Patients were categorised according to: (1) age (<64, 65–74, 75–84, 85+ years); (2) time of arrival at the ED (0000–0559, 0600–1159, 1200–1759, 1800–2359 hours); (3) day of arrival (weekdays, weekend); (4) mode of arrival to the ED (ambulance, walk-in, other); (5) emergency triage category (Australian Triage Scale (ATS), www.acem.org, verified 11 November 2014); (6) type of clinical presentation (undifferentiated or differentiated); and (7) major ICD-10 diagnosis-related group (DRG) category (e.g. diseases of the circulatory system (Code I), diseases of the respiratory system (Code J). Two experienced triage nurses reviewed the ED triage presentation notes for each patient who died and categorised the presentation as differentiated (where there was a clear and obvious diagnosis such as multitrauma or subarachnoid haemorrhage confirmed by imaging) or undifferentiated (where clinical features were non-specific and in need of diagnostic work-up, such as fever or confusion). Cases involving disagreement between the two triage nurses were classified as undifferentiated.

The primary outcome variable was death of admitted in-patients as defined according to coding, with deaths coded as occurring in palliative care patients excluded from the analysis on the basis that many of these are expected deaths, care goals for these patients differ markedly from those of non-palliative patients and NEAT improvement strategies are aimed at reducing avoidable deaths in acutely ill patients who do not have a terminal illness.

Mortality rates of patients admitted from the ED were calculated variously as rate per number of ED presentations, rate per number of admitted patients and rates per number of patients admitted to the same destination ward.

Comparisons of variables measured in 2011 and 2013 were undertaken using Chi-squared tests for categorical variables and ANOVA for continuous variables. Data analysis was performed using IBM SPSS Statistics version 22 (IBM Corp, Armonk, NY, USA). The study was approved by the Metro South Human Research Ethics Committee (HREC/12/QPAH/021).

In 2011, excluding patients admitted to the ED short-stay ward, 17 022 of 53 419 patients presenting to the ED were admitted to in-patient surgical or medical wards (admission rate 31.9%), of whom 320 (1.88%) died, compared with 17 162 of 59 623 presentations in 2013 (admission rate 28.8%), of whom 202 (1.18%) died (P < 0.001 for all comparisons). There was no significant difference in the numbers of deaths (100 vs 99) or rates of death (0.19% vs 0.17%) of patients who died in the ED between the 2 years.

No significant changes were seen in the pattern of coding of deaths, either palliative or non-palliative, of patients who died on an in-patient ward after ED admission between 2011 and 2013. In 2011 there were 70 palliative deaths (representing 17.9% of all deaths in ED patients admitted to in-patient wards) and in 2013 there were 58 (22.2%; P = 0.157).

There was a significant decrease in mortality for patients admitted to either medical wards (n = 9164 in 2011; n = 10 290 in 2013) or surgical wards (n = 7859 in 2011; n = 6872 in 2013), from 1.62% to 0.89% (45% relative reduction; P < 0.001) and from 2.15% to 1.60% (26% relative reduction; P = 0.014), respectively.

No significant differences were seen between 2011 and 2013 in the proportion of all in-hospital deaths attributed to each of the four age groups (data not shown). However, among patients admitted to medical wards, there were significant reductions in mortality across all age groups between 2011 and 2013m except for the 64–75-year-old age group (P = 0.057), with the largest decrease seen in the ≥84 years age group (from 2.8% to 1.2%; P = 0.011; Fig. 1). A similar pattern of reduced mortality in 2013 was seen across all age groups for surgical admissions, but these reductions were not significant (data not shown).

Fig. 1.  In-hospital mortality of patients admitted to medical wards according to age. Data are the mean ± s.d.

In 2011, admitted patients who presented to ED triage between midnight and 12 noon had a significantly higher mortality rate than those presenting in the remaining 12 h of the day (2.5% vs 1.5%, respectively; P < 0.001). In 2013, this diurnal variation disappeared, with respective mortality rates of 1.3% and 1.1% (P = 0.150).

Similarly, in 2011, admitted patients who presented to ED triage on weekends had a significantly higher mortality rate than those presenting on weekdays (2.23% vs 1.74%, respectively; P = 0.038). In 2013, this variation was no longer seen (1.3% vs 1.1%’ P = 0.150).

Patients who arrived by ambulance had a significantly higher mortality rate than those who arrived by other means, and this did not change over time (2.5% vs 0.8%, respectively, in 2011; 2.5% vs 0.5%, respectively, in 2013; P < 0.001 for both comparisons). Among the patients who died as an in-patient and had been brought in by ambulance, a higher proportion were in ATS Categories 1 and 2 in both 2011 and 2013 compared with deceased patients who had arrived by other means (70.8% vs 36.9%, respectively, in 2011 (P < 0.001); 67.9% vs 36.0%, respectively, in 2013 (P = 0.006)).

Patients who died following presentation with differentiated syndromes spent less time in the ED than those with undifferentiated presentations in both 2011 (difference 216 min) and 2013 (difference 173 min; P < 0.001 for both comparisons; Fig. 2).

Fig. 2.  Classification of triage presentation (differentiated, where there was a clear and obvious diagnosis; undifferentiated, clinical features non-specific and in need of diagnostic work-up) for patients who died after being admitted to an in-patient ward and emergency department (ED) length of stay (LOS). Data are the mean ± s.d.*P < 0.001 compared with 2011.

Between 2011 and 2013 there was a significant reduction in mortality rates among patients coded as having circulatory diseases (ICD-10 Code I: 3.62% vs 1.70%, respectively; P < 0.001) and respiratory diseases (ICD-10 Code J: 3.43% vs 1.48%, respectively; P < 0.001), but not for the other major groups (Fig. 3).

Fig. 3.  Mortality rates for emergency department patients admitted to an in-patient ward according to major International Classification of Diseases version 10 (ICD-10) diagnostic codes. Data are the mean ± s.d.*P < 0.001 compared with 2011.

Previously, we reported a near halving in in-hospital mortality rate in association with a doubling in NEAT compliance pertaining to patients admitted acutely to in-patient units via the ED.¹² In the present study, we identified subgroups of patients who appear to derive greater benefit in terms of reduced risk of in-hospital death from improved NEAT compliance, while also taking account of other non-clinical confounders that may give rise to spurious reductions in mortality rates, as suggested by other investigators. For example, Toh et al.¹⁸ raised the possibility of greater numbers of lower acuity, relatively healthy patients being admitted to meet the 4-h target, rather than being processed and finally discharged from ED, as has been observed in some reports,¹⁹ thus resulting in an artefactual lowering of in-hospital mortality.

We accounted for this by excluding patients who were admitted to the short-stay ward and noting that the admission rate to in-patient units was stable (indeed higher in 2011 than in 2013) over the 3-year period. We also found no increase in ED deaths or upcoding of in-patient deaths as palliative, both of which could contribute to spurious reductions in the number of in-hospital deaths.

We were unable to find any other reports of the diurnal variation in mortality rates according to time of ED presentation that we observed in 2011 before NEAT interventions, but which was later corrected. This correction was temporally associated with the commencement of extended night-time rostering of medical registrars (to reduce handover of yet-to-be-admitted patients from the evening peaks to the smaller number of night registrars) and more rapid processing of patients still present in the ED in the early morning hours.

Several studies have reported a ‘weekend effect’ whereby emergency patients admitted on weekends incur a higher risk of death than those admitted on weekdays.^20,21 In some studies this has been related to longer ED stays on weekends.²² Our findings revealed a similar pattern that was no longer statistically significant following the implementation of better matching of resources at the ED–in-patient interface to patient presentations.

The association of prolonged ED stays and more adverse events in older patients with multiple comorbidities has been noted by various investigators.^23,24 In the present study, older complex medical patients admitted to medical wards fared better as a result of improved NEAT compliance. Similarly, several studies have associated worse outcome and increased mortality among patients with undifferentiated or non-specific presenting complaints compared with those with clear-cut diagnoses.^25–27 Although we were only able to classify patients who died as having either differentiated or undifferentiated presentations, thus preventing calculation of all-patient mortality rates, ED stay was considerably longer in the latter compared with the former, but decreased by a greater extent following NEAT interventions. We would expect, based on the previously cited studies, that this would translate to a larger survival benefit.

Patients presenting with time-sensitive acute circulatory or respiratory diseases, such as acute pulmonary oedema or community-acquired pneumonia, are more likely to die if indicated care is not delivered in a timely fashion. Investigations at Mid-Staffordshire revealed that highly dysfunctional interfaces between ED and in-patient units were associated with more deaths concentrated among patients presenting acutely with cardiovascular disorders.²⁸ Other researchers have observed associations between measures of ED stays and overcrowding and delays in resuscitation and treatment of acutely ill patients,^29,30 higher risk of adverse cardiovascular events in patients presenting with chest pain³¹ and suboptimal timing of antibiotic administration³² and increased mortality³³ in patients presenting with community-acquired pneumonia. In the present study, NEAT interventions were associated with reduced mortality rates in patients presenting with acute circulatory or respiratory diseases.

The present study has some limitations. Because we performed an observational analysis, we cannot exclude the possibility that other factors in addition to improved NEAT compliance may have contributed to the lowered in-patient mortality. However, no other major system of care or clinical process changes occurred during the study period. We report crude rather than adjusted death rates because it was not possible to develop robust risk adjustment regression models for each of several subgroup analyses involving highly heterogeneous patients. However, we have no reason to believe that the concordance observed between reductions in crude mortality and eHSMR noted in our previous study¹³ for all admitted patients does not apply to individual subgroups.

Moreover, the differences in mortality rates for most subgroup analyses were highly statistically significant (P < 0.001), so it is likely that in regression models these subgroup predictors would fall out as independent associations. Patients coded as being palliative were excluded from analysis because many of these deaths are expected, care goals are conservative and aimed at symptom control rather than increasing survival and the effects on mortality of NEAT improvement strategies are not relevant for this group.

Finally, we were limited to measuring in-hospital mortality only, as opposed to post-discharge mortality up to 30 days after ED presentation, which may be a more accurate marker of intervention effects. However, studies suggest that most deaths relating to emergency admissions occur within 12 days of presentation.³⁴

In conclusion, the present study provides the first preliminary evidence that low-cost targeted clinical redesign of the ED–in-patient interface leading to enhanced NEAT compliance is associated with improved mortality for particular subgroups of emergency patients admitted to in-patient wards. These include those at highest risk of death: elderly patients with complex medical conditions, those presenting on weekends or after hours and those with time-sensitive acute cardiorespiratory conditions. Further research is needed using larger samples to identify and validate these and other patient characteristics associated with differential benefit and to explore causal mechanisms in greater depth.

None declared.