Long term survival following a medical emergency team call at an Australian regional hospital

Abstract

Objective: To investigate the long term survival of medical emergency team (MET) patients at an Australian regional hospital and describe associated patient and MET call characteristics.

Design: Retrospective cohort study. Data linkage to the statewide death registry was performed to allow for long term survival analysis, including multivariable Cox proportional hazards regression and production of Kaplan–Meier survival curves.

Setting: A large Australian regional hospital.

Participants: Adult patients who received a MET call from 1 July 2012 to 3 March 2020.

Main outcome measures: Survival to 30, 90 and 180 days; one year; and 5-years after index MET call.

Results: The study included 6499 eligible patients. The cohort median age was 71 years, and 52.4% of the patients were female. Surgical (39.6%) and medical (36.9%) patients comprised most of the cohort. Thirty-day survival was 86.5% one-year survival was 66.1%. Among patients aged < 75 years, factors independently associated with significantly higher long term mortality included age (hazard ratio [HR], 3.26 [95% CI, 2.63–4.06]; for patients aged 65-74 v 18–54 years), male sex (HR, 0.71 [95% CI, 0.61–0.83]; for females) and pre-existing limitation of medical therapy (HR, 2.76; 95% CI, 2.28–3.35). Among patients aged ≥ 75 years, factors independently associated with significantly higher long term mortality included age (HR, 1.46 [95% CI, 1.29–1.65]; for patients aged ≥ 85 years), male sex (HR, 0.74 [95% CI, 0.66–0.83]; for females), and altered MET criteria (HR, 1.33; 95% CI, 1.03–1.71).

Conclusions: Long term survival probabilities of MET call patients are affected by factors including age, sex, and limitation of medical therapy status. These data may be useful for clinicians conducting end-of-life discussions with patients.

A rapid response system facilitates the timely identification of hospitalised patients experiencing clinical deterioration and the provision of a proportionate clinical intervention by a rapid response team.1, 2, 3 One form of rapid response team — the medical emergency team (MET) — comprises doctors and nurses with specific training, skills and resources, which enable them to respond appropriately to clinically deteriorating patients outside of the intensive care unit (ICU).1, 2, 3 MET calls are primarily activated by ward staff in response to a patient’s vital signs moving outside of a prespecified range or to concern for the patient’s clinical state.1, 2, 3, 4

The short term mortality of patients who experience a MET call has been well described.⁵ A 2017 systematic review found that the median in-hospital mortality of MET patients was 26% (range, 12–60%).⁵ Furthermore, factors such as increasing age and presence of a limitation of medical therapy appear to increase in-hospital MET patient mortality.6, 7

While there is substantial literature on short term mortality following MET calls, long term survival among MET patients is less well understood.⁸ The long term trend of in-hospital mortality of major surgical candidates before and after the implementation of a rapid response system has been described elsewhere,⁹ and a prospective cohort study has identified the one-year mortality of MET patients.⁷ However, no studies have identified correlates of long term survival. Furthermore, regional hospital rapid response systems are understudied.¹⁰ Knowledge of long term survival after MET activation could be beneficial for patients and clinicians, providing realistic survival probabilities to help frame informed end-of-life care discussions.

In this study, we describe MET call and patient characteristics in an Australian regional hospital setting, describe long term survival following MET calls, and identify correlates of long term survival.

Methods

Design

We conducted a retrospective cohort study, using local hospital data linked to the statewide death registry.

Setting

Bendigo Hospital is a 300-bed facility located in the Australian state of Victoria, 150 km north-west of the state’s capital, Melbourne. The hospital treats about 49 000 patients per year and offers most acute medical and surgical services, except specialties such as neurosurgery, cardiothoracic surgery, and neurology. The Bendigo Hospital ICU established its MET in 2005. This physician-led MET is separate from the hospital’s code blue team. Both teams operate 365 days per year, 24 hours per day. It is mandatory for staff to call the MET when patients breach clinical parameters (Online Appendix, appendix 1).

Data

Bendigo Hospital MET call database

Bendigo Hospital has recorded structured MET call data since 2012. Each MET call that occurs at the hospital is attended by a data scribe (typically the MET nurse, an ICU nurse), who collects relevant data on a hard-copy standardised MET call data collection form (Online Appendix, appendix 2). Data on the form are entered into an electronic database by ICU liaison nurses. Data are cross-checked to ensure optimal data integrity. Incomplete or missing fields are reviewed by ICU liaison nurses at the time of data entry.

Local hospital data

For each MET call, we extracted the patient’s full name, date of birth, and date of death (where applicable) from the hospital’s patient administration system, using the unique patient unit record number as a linkage key. This linkage identified if any patients had a recorded in-hospital death, negating the need for linkage with the statewide death registry.

The Victorian Death Registry

The Victorian Death Registry (VDR) records details of all deaths that are registered in the state of Victoria. The full name and date of birth of each eligible MET patient (based on the eligibility criteria in the following subsection and the absence of in-hospital death data) were securely supplied to the VDR for probabilistic linkage to death records up to 19 August 2020. The VDR returned a list of death dates corresponding to a subset of patients for whom a match was made. The final database included individual-level MET call data, date of birth, and date of death, where applicable.

Inclusion and exclusion criteria

Adult patients (aged ≥ 18 years) who received at least one MET call at Bendigo Hospital between 1 July 2012 and 3 March 2020 (inclusive) were included in the study. Patients who had a listed residential postcode outside the state of Victoria were excluded, as their death would likely not be recorded in the VDR. Patients whose recorded data were incorrectly entered (ie, entry in the wrong field), such that data linkage was impossible, were excluded. Patients who did not survive to 30 days after their index MET call were excluded for the purposes of long term analysis, but were included in the calculation of demographic descriptive statistics and survival rates by the time since the MET call. The decision to exclude these patients from long term analysis was made because the characteristics of patients dying within 30 days of their MET call could obscure patient characteristics that were associated with long term survival. Furthermore, multiple previous studies have investigated the 30-day mortality of MET patients.11, 12, 13, 14, 15, 16 A patient’s first recorded MET call (the “index” MET call) within the study period was used to conduct our analyses, with subsequent MET calls for that patient excluded. A flow diagram showing the derivation of the cohort is included in the Online Appendix (appendix 3).

Definitions

In this study, “long term” refers to the time from 30 days after a patient’s index MET call to their date of death, or the study censor date of 19 August 2020, whichever came first. We defined the geriatric population as patients aged ≥ 75 years to resolve violation of the proportional hazards (PH) assumption of Cox PH regression, described in the next subsection.

Statistical methods

MET dose

The MET dose was calculated by summing the total number of MET calls, with no exclusion criteria applied, per financial year (July–June), and dividing by the number of multiday admissions to Bendigo Hospital in the same year. This quotient was multiplied by 1000 to obtain the MET dose.

Patient and MET call characteristics

The following analyses were performed for the eligible patient cohort (ie, all eligible MET patients’ index MET call), irrespective of survival and age.

Cohort and MET call characteristics were described. Categorical variables were summarised in terms of their frequency (percentage), and age was also summarised in terms of the median and interquartile range (IQR). The proportions of patients surviving from their index MET call to follow-up at 30, 90 and 180 days; one year; and 5 years were described. If a given patient’s survival time was insufficient to meet one or more of these metrics, then the patient was designated as being censored for the particular time points.

Long term survival

The following analyses were performed for eligible patients who survived to 30 days after their index MET call.

Kaplan–Meier survival curves for the entirety of this cohort, and stratified by age 18–74 years (inclusive) and age ≥ 75 years, were produced. These were compared with survival curves for the general Australian population matched on age, sex and year. Mortality data to produce matched survival curves were sourced from the Australian Bureau of Statistics.¹⁷

Cox proportional hazards regression was used to identify factors associated with increased risk of long term mortality. Hazard ratios (HRs), 95% confidence intervals (CIs) and P values were calculated. P values < 0.05 were considered indicative of statistical significance.

In the first instance, univariable models were fitted for the Cox proportional hazards regression analysis. This was followed by fitting a multivariable model, using all variables significant in the univariable model. Since mortality showed no consistent trend over time, year of MET call was coded as an annual categorical variable rather than a grouped categorical or continuous variable. Two theoretical confounding factors, age and sex, were included regardless of statistical significance.

The proportional hazards assumption was found to be violated by the “age” variable after conducting the scaled Schoenfeld residuals test. Splitting the cohort into two distinct age groups, ≥ 75 years and < 75 years, resolved this issue. All subsequent Cox proportional hazards regression analyses were conducted on these two cohorts individually. The year variable for the cohort aged ≥ 75 years was not included in the multivariable model despite obtaining a significant result for one year, due to its lack of clinical relevance and the insignificant results for other years.

Data preparation was conducted using Microsoft Excel 2019 (version 1808; Microsoft Corporation). Data cleaning and statistical analysis were conducted using the programming software R Version 4.0.2 (R Foundation for Statistical Computing).

Ethics

Ethics approval was provided by Bendigo Health (reference: HREC/19/BHCG/58947) and Monash University (reference: 23961).

Results

During the study period, 6499 unique patients received an index MET call at Bendigo Hospital (Table 1). Females comprised 52.4% of the entire cohort. The median age was 71 years, with 41.8% of patients aged ≥ 75 years (Table 1).

Table 1.

Demographic characteristics of the entire cohort (n = 6499)

Characteristics and variables	n (%)*
Total number of patients	6499
Sex
Male	3045 (46.9%)
Female	3407 (52.4%)
Missing	47 (0.7%)
Age, years (continuous), median (IQR)	71.3 (55.7-82.1)
Age, years (categorical)
18-54	1578 (24.3%)
55-64	882 (13.6%)
65-74	1318 (20.3%)
75-84	1562 (24.0%)
≥ 85	1159 (17.8%)
Time of MET
In office hours† (08:00-17:00)	3072 (47.3%)
Outside office hours (17:00-08:00)	3411 (52.5%)
Missing	16 (0.2%)
Admitting unit
Surgical	2573 (39.6%)
Medical	2395 (36.9%)
Obstetrics and gynaecology	415 (6.4%)
Oncology	335 (5.2%)
Psychiatry	229 (3.5%)
Cardiology	214 (3.3%)
Rehabilitation	179 (2.8%)
Other‡	136 (2.1%)
Missing	23 (0.4%)
LOMT in place before MET
No LOMT recorded before	4748 (73.1%)
Not for CPR, intubation or ICU	770 (11.8%)
Not for CPR or intubation	443 (6.8%)
Not for CPR	415 (6.4%)
2 LOMTs recorded	58 (0.9%)
≥ 3 LOMTs recorded	42 (0.6%)
Not for CPR, intubation, ICU or MET	23 (0.4%)
LOMT instituted by MET
No LOMT recorded	6306 (97.0%)
Not for CPR, intubation, ICU or MET	69 (1.1%)
Not for CPR, intubation or ICU	65 (1.0%)
Not for CPR or intubation	25 (0.4%)
Not for CPR	25 (0.4%)
3 LOMTs recorded	5 (0.1%)
2 LOMTs recorded	4 (0.1%)
Reason for MET (MET trigger)
Hypotension (SBP < 90 mmHg)	2145 (33.0%)
Tachycardia (> 130 beats/min)	1169 (18.0%)
Other concern	801 (12.3%)
GCS decreased by 2 or more	543 (8.4%)
Tachypnoea (> 30 breaths/min)	535 (8.2%)
Hypoxia (SpO2 < 90%)	507 (7.8%)
AKI (urine output < 50 mL over 4 h)	389 (6.0%)
Bradycardia (< 40 beats/min)	167 (2.6%)
Multiple MET reasons§	93 (1.4%)
Bradypnoea (< 8 breaths/min)	74 (1.1%)
PACT	34 (0.5%)
Parent unit not present	19 (0.3%)
Missing	23 (0.4%)
Immediate outcome of MET
Remained on ward	5445 (83.8%)
Other outcome¶	366 (5.6%)
Altered MET criteria	330 (5.1%)
Transferred to ICU	320 (4.9%)
Upgrade to code blue	38 (0.6%)
Event occurring before MET call
No event recorded	5811 (89.4%)
Operation 6 h before	561 (8.6%)
ICU 24 h before	127 (2.0%)

CPR = cardiopulmonary resuscitation; GCS = Glasgow Coma Scale; ICU = intensive care unit; IQR = interquartile range; LOMT = limitation of medical treatment; MET = medical emergency team; PACT = Patient/Consumer Activated Care Team (this is a Bendigo Health initiative whereby patients or their family can activate a MET call); parent unit = team of clinicians who are primarily in charge of the patient’s hospital stay; SBP = systolic blood pressure.

^*Unless otherwise stated.

^†In this context, “office hours” refers to the time 08:00-17:00 every day of the week.

^‡Other admitting units included admissions to palliative care/hospice, unequivocal admissions (eg, admissions under both psychiatry and rehabilitation), and units specified as “other” by the attending MET.

^§Multiple MET reasons refer to MET calls that were called for more than one parameter being breached for the patient.

^¶Other immediate MET outcomes include patients who died at the MET call, patients with multiple recorded outcomes, patients transferred to the operating theatre, and patients transferred to the emergency department.

Patients who survived to 30 days after the MET call comprised 86.5% of the entire cohort (Table 2). The survival rates at 180 days and one year were 77.2% and 66.1% respectively.

Table 2.

Frequency and percentage of patients surviving until discrete time measures after their index medical emergency team (MET) call

	Day of MET call	30 Days	90 Days	180 Days	1 Year	5 Years
Total number of patients	6499	6499	6499	6499	6499	6499
Survived	6413 (98.7%)	5622 (86.5%)	5293 (81.4%)	5014 (77.2%)	4297 (66.1%)	1221 (18.8%)
Died	86 (1.3%)	877 (13.5%)	1206 (18.6%)	1467 (22.6%)	1789 (27.5%)	2676 (41.2%)
Censored	-	0 (0.0%)	0 (0.0%)	18 (0.3%)	413 (6.4%)	2602 (40.0%)

For both the younger (< 75 years) (Table 3) and older (≥ 75 years) (Table 4) cohorts, factors independently associated with significantly increased risk of long term mortality in the multivariable model included increasing age, male sex, medical ward admission, and a limitation of medical therapy in place before, or as a result of, a MET call. For the cohort aged < 75 years only, an operation in the 6 hours preceding the MET call and tachypnoea were independently associated with significantly reduced risk of long term mortality (Table 3). For the cohort aged ≥ 75 years only, altered MET criteria and hypoxia were independently associated with significantly increased risk of long term mortality, and a patient’s discharge from the ICU during the preceding 24 hours was independently associated with significantly reduced risk of long term mortality (Table 4).

Table 3.

Univariable and multivariable Cox proportional hazards regression modelling results for patients aged 18-74 years (inclusive) who received a medical emergency team (MET) call at Bendigo Hospital (N = 3467)

Characteristics and variables	n	Univariable			Multivariable
		HR*	95% CI	P	aHR*	95% CI	P
Age (years)
18-54	1531	1.00	-	-	1.00	-	-
55-64	797	3.35	2.67-4.21	< 0.001	2.50	1.97-3.16	< 0.001
65-74	1139	4.95	4.03-6.08	< 0.001	3.26	2.63-4.06	< 0.001
Sex
Male	1522	1.00	-	-	1.00	-	-
Female	1920	0.50	0.43-0.58	< 0.001	0.71	0.61-0.83	< 0.001
Missing	25	0.58	0.22-1.54	0.274	0.89	0.33-2.41	0.823
Admitting unit
Medical†	1341	1.00	-	-	1.00	-	-
Surgical	1403	0.40	0.34-0.47	< 0.001	0.52	0.44-0.62	< 0.001
Other‡	711	0.14	0.10-0.20	< 0.001	0.31	0.22-0.44	< 0.001
Missing	12	0.76	0.24-2.37	0.638	1.37	0.43-4.31	0.593
LOMT in place before MET call
No	3181	1.00	-	-	1.00	-	-
Yes	286	5.11	4.27-6.11	< 0.001	2.76	2.28-3.35	< 0.001
LOMT instituted by MET
No	3436	1.00	-	-	1.00	-	-
Yes	31	4.65	2.87-7.52	< 0.001	1.97	1.20-3.22	0.007
Event before MET call
No event recorded	3037	1.00	-	-	1.00	-	-
Operation 6 h before	376	0.41	0.30-0.57	< 0.001	0.62	0.44-0.88	0.008
ICU 24 h before	54	0.95	0.52-1.72	0.862	0.79	0.43-1.43	0.433
Bradypnoea
No	3410	1.00	-	-	1.00	-	-
Yes	57	0.40	0.17-0.96	0.041	0.49	0.20-1.20	0.118
Tachypnoea
No	3212	1.00	-	-	1.00	-	-
Yes	255	1.38	1.07-1.78	0.013	0.75	0.56-1.00	0.046
Hypoxia
No	3244	1.00	-	-	1.00	-	-
Yes	223	1.64	1.27-2.12	< 0.001	0.98	0.74-1.30	0.894
Tachycardia
No	2869	1.00	-	-	1.00	-	-
Yes	598	1.24	1.03-1.50	0.023	0.99	0.79-1.23	0.925
Hypotension
No	2170	1.00	-	-	1.00	-	-
Yes	1297	0.74	0.63-0.87	< 0.001	0.92	0.75-1.11	0.376
Bradycardia
No	3373	1.00	-	-	-	-	-
Yes	94	0.64	0.37-1.11	0.115	-	-	-
GCS drop of ≥ 2
No	3226	1.00	-	-	-	-	-
Yes	241	1.14	0.86-1.50	0.357	-	-	-
Oliguria
No AKI§	3331	1.00	-	-	-	-	-
AKI	136	1.12	0.78-1.62	0.534	-	-	-
Parent unit not responding
Parent unit present	3456	1.00	-	-	-	-	-
Parent unit not present	11	1.40	0.45-4.36	0.558	-	-	-
PACT
No	3442	1.00	-	-	-	-	-
Yes	25	1.17	0.44-3.12	0.758	-	-	-
Other concern
No other concern	2916	1.00	-	-	-	-	-
Other concern	551	0.87	0.70-1.08	0.208	-	-	-
Time of MET call
Outside office hours	1821	1.00	-	-	-	-	-
In office hours	1638	1.07	0.92-1.25	0.357	-	-	-
Missing	8	0.65	0.09-4.62	0.666	-	-	-
Immediate MET outcome
Left on ward	2828	1.00	-	-	-	-	-
Altered MET criteria	184	0.96	0.68-1.35	0.801	-	-	-
Transferred to ICU	202	1.01	0.74-1.38	0.937	-	-	-
Other outcome¶	232	0.97	0.70-1.33	0.838	-	-	-
Upgrade to code blue	21	1.13	0.47-2.72	0.786	-	-	-
Event year
2012	191	1.00	-	-	-	-	-
2013	403	0.90	0.66-1.23	0.504	-	-	-
2014	378	0.81	0.58-1.12	0.196	-	-	-
2015	436	1.04	0.76-1.42	0.820	-	-	-
2016	415	0.85	0.61-1.19	0.341	-	-	-
2017	489	0.95	0.68-1.31	0.740	-	-	-
2018	514	0.78	0.55-1.11	0.168	-	-	-
2019	548	0.73	0.49-1.08	0.116	-	-	-
2020	93	0.72	0.29-1.82	0.492	-	-	-

aHR = adjusted hazard ratio; AKI = acute kidney injury; GCS = Glasgow Coma Scale; HR = hazard ratio; ICU = intensive care unit; LOMT = limitation of medical treatment; PACT = Patient/Consumer Activated Care Team (this is a Bendigo Health initiative whereby patients or their family can activate a MET call); parent unit = team of clinicians who are primarily in charge of the patient’s hospital stay.

^*Hazard ratios greater than 1 indicate an increased risk of mortality.

^†Cardiology, oncology and rehabilitation consolidated into “medical” level.

^‡Obstetrics and gynaecology and psychiatry consolidated into “other” level.

^§Where AKI is defined as urine output < 50 mL over 4 hours.

^¶Other outcomes include: patients who died at the MET call, patients with multiple recorded outcomes, patients transferred to the operating theatre, and patients transferred to the emergency department.

Table 4.

Univariable and multivariable Cox proportional hazards regression modelling results for patients aged ≥ 75 years who received a medical emergency team (MET) call at Bendigo Hospital (N = 2155)

Characteristics and variables	n	Univariable			Multivariable
		HR*	95% CI	P	aHR*	95% CI	P
Age
75-84	1285	1.00	-	-	1.00	-	-
≥ 85	870	1.69	1.51-1.89	< 0.001	1.46	1.29-1.65	< 0.001
Sex
Male	1030	1.00	-	-	1.00	-	-
Female	1107	0.78	0.70-0.88	< 0.001	0.74	0.66-0.83	< 0.001
Missing	18	0.68	0.35-1.31	0.244	0.64	0.33-1.25	0.192
Admitting unit
Medical†	1138	1.00	-	-	1.00	-	-
Surgical	955	0.72	0.64-0.81	< 0.001	0.87	0.77-0.98	0.027
Other ‡	53	0.78	0.52-1.16	0.224	1.06	0.69-1.63	0.790
Missing	9	1.28	0.61-2.69	0.519	1.60	0.76-3.39	0.216
Time of MET call
Outside office hours	1083	1.00	-	-	1.00	-	-
In office hours	1068	1.13	1.01-1.27	0.032	1.12	1.00-1.26	0.051
Missing	4	2.34	0.75-7.28	0.143	2.34	0.74-7.42	0.148
LOMT in place before MET call
No	1266	1.00	-	-	1.00	-	-
Yes	889	2.04	1.82-2.29	< 0.001	1.75	1.55-1.98	< 0.001
LOMT instituted by MET
No	2083	1.00	-	-	1.00	-	-
Yes	72	1.94	1.50-2.53	< 0.001	1.62	1.24-2.11	< 0.001
Immediate MET outcome
Left on ward	1898	1.00	-	-	1.00	-	-
Altered MET criteria	98	1.28	1.00-1.65	0.05	1.33	1.03-1.71	0.029
Transferred to ICU	69	0.78	0.56-1.09	0.145	0.83	0.59-1.17	0.286
Other outcome §	86	0.67	0.47-0.94	0.019	0.76	0.52-1.09	0.137
Upgrade to code blue	4	0.84	0.21-3.36	0.805	0.97	0.24-3.89	0.963
Event before MET call
No event recorded	1932	1.00	-	-	1.00	-	-
Operation 6 h before	162	0.65	0.52-0.83	< 0.001	0.89	0.70-1.14	0.356
ICU 24 h before	61	0.60	0.41-0.89	0.010	0.62	0.42-0.92	0.018
Other concern
No other concern	1943	1.00	-	-	1.00	-	-
Other concern	212	0.78	0.63-0.95	0.013	0.87	0.70-1.07	0.176
Tachypnoea
No	1965	1.00	-	-	1.00	-	-
Yes	190	1.22	1.01-1.48	0.040	1.08	0.89-1.31	0.447
Hypoxia
No	1986	1.00	-	-	1.00	-	0.004
Yes	169	1.40	1.15-1.71	0.001	1.34	1.10-1.64	0.004
Bradypnoea
No	2143	1.00	-	-	-	-	-
Yes	12	0.60	0.25-1.45	0.261	-	-	-
Bradycardia
No	2076	1.00	-	-	-	-	-
Yes	79	0.95	0.70-1.30	0.749	-	-	-
Tachycardia
No	1693	1.00	-	-	-	-	-
Yes	462	0.89	0.78-1.03	0.118	-	-	-
Hypotension
No	1486	1.00	-	-	-	-	-
Yes	669	1.00	0.89-1.13	0.941	-	-	-
GCS drop of ≥ 2
No	1949	1.00	-	-	-	-	-
Yes	206	0.95	0.78-1.16	0.631	-	-	-
Oliguria
No AKI¶	1974	1.00	-	-	-	-	-
AKI	181	1.17	0.97-1.42	0.104	-	-	-
Parent unit not responding
Parent unit present	2149	1.00	-	-	-	-	-
Parent unit not present	6	0.76	0.25-2.37	0.638	-	-	-
PACT
No	2151	1.00	-	-	-	-	-
Yes	4	0.40	0.06-2.81	0.355	-	-	-
Event year
2012	109	1.00	-	-	-	-	-
2013	255	1.04	0.80-1.36	0.758	-	-	-
2014	271	1.19	0.92-1.55	0.189	-	-	-
2015	268	1.31	1.01-1.72	0.044	-	-	-
2016	261	1.31	1.00-1.72	0.054	-	-	-
2017	306	1.24	0.94-1.63	0.131	-	-	-
2018	337	1.20	0.90-1.60	0.210	-	-	-
2019	301	1.23	0.90-1.68	0.203	-	-	-
2020	47	1.78	0.96-3.29	0.068	-	-	-

aHR = adjusted hazard ratio; AKI = acute kidney injury; GCS = Glasgow Coma Scale; HR = hazard ratio; ICU = intensive care unit; LOMT = limitation of medical treatment; PACT = Patient/Consumer Activated Care Team (this is a Bendigo Health initiative whereby patients or their family can activate a MET call); parent unit = team of clinicians who are primarily in charge of the patient’s hospital stay.

^*Hazard ratios greater than one indicate an increased risk of mortality.

^†Cardiology, oncology and rehabilitation consolidated into “medical” level.

^‡Obstetrics and gynaecology and psychiatry consolidated into “other” level.

^§Other outcomes include: patients who died at the MET call, patients with multiple recorded outcomes, patients transferred to the operating theatre, and patients transferred to the emergency department.

^¶Where AKI is defined as urine output < 50 mL over 4 hours.

For all patients surviving to 30 days (Figure 1), Kaplan–Meier survival estimates were 82.7% at one year (95% CI, 81.7–83.7%), 68.2% at 3 years (95% CI, 66.8–69.5%) and 59.7% at 5 years (95% CI, 58.1–61.3%). For subgroup analysis of patients surviving to 30 days, stratified by age (Figure 2), Kaplan–Meier estimates for survival to one year were 89.8% for patients aged < 75 years (95% CI, 88.7-90.8%) and 71.5% for patients aged ≥ 75 years (95% CI, 69.6–73.4%). Estimates for survival to 3 years were 81.1% for patients aged < 75 years (95% CI, 79.7–82.5%) and 47.7% for patients aged ≥ 75 years (95% CI, 45.5–50.1%). Estimates for survival to 5 years were 76.5% for patients aged < 75 years (95% CI, 74.8–78.2%) and 33.4% for patients aged ≥ 75 years (95% CI, 31.0–36.0%).

Figure 1.

Kaplan-Meier curve for the entire cohort of patients surviving to 30 days after their index medical emergency team (MET) call

The dotted line represents the general Australian population matched on age, sex and year of MET call matched on age, sex and year of MET call occurrence (N = 5622).

Figure 2.

Kaplan-Meier curve for the cohort of patients surviving to 30 days after their index medical emergency team (MET) call, stratified by age

The dotted lines represent age subgroups for the general Australian population matched on age, sex and year of MET call occurrence (< 75 years, N = 3467; ≥ 75 years, N = 2155).

The MET dose increased over the study period from 81 MET calls per 1000 multiday admissions in 2012–13 to 114 per 1000 in 2018–19 (Online Appendix, appendix 4).

Discussion

Principal findings

To our knowledge, this study is among the first to describe long term survival after a MET call, and the first to identify correlates of long term survival. It is also among the first to document characteristics of regional hospital MET systems.10, 18 Increasing age, male sex, and presence of a limitation of medical therapy were independently associated with increased risk of long term mortality, while surgical admission was independently associated with a decreased risk of long term mortality.

Characteristics of MET calls and MET patients

Many characteristics of our patient cohort were comparable to those reported in other Australian and international MET studies.7, 19, 20, 21 Forty-two per cent of our MET patient cohort were aged ≥ 75 years and the median age was 71 years, representing a left-skewed distribution and replicating findings from Finland⁷ and New Zealand.²¹ Increasing MET dose has been used to track utilisation of the rapid response system in a hospital and has been associated with decreased incidence of in-hospital cardiac arrest;²² indeed, the incidence of in-hospital cardiac arrest has decreased at Bendigo Hospital from 0.91 arrests per 1000 admissions in the year 2000 to 0.32 arrests per 1000 admissions in 2017.²³ From 2012–13 to 2018–19, the annual MET dose increased 40% from 81 to 114 MET calls per 1000 multiday admissions. In comparison, the Australian and New Zealand Intensive Care Society Centre for Outcome and Resource Evaluation (ANZICS CORE) MET Dose Investigators found the median increase in MET dose across 23 Australian hospitals to be 90% (IQR, 40–180%).²⁴ The similarity between our patient cohort and other hospitals’ MET patient cohorts indicates that our long term survival findings may hold relevance for other sites.

Long term mortality

Among the MET patients in our cohort study, the one-year mortality rate was 27.5%. Only one other study is known to have reported the one-year mortality outcomes of MET patients.⁷ This prospective cohort study (N = 1372) conducted at a tertiary hospital in Finland had a one-year mortality rate of 41%.⁷ These findings are higher than reported one-year mortality in cohorts of ICU patients; a multicentre retrospective cohort study conducted in Ontario, Canada, found the one-year mortality of ICU patients to be 11%.²⁵ In addition, the reduced survival over time of our MET patient cohort compared with matched general population cohorts is observed in studies comparing the Kaplan–Meier survival estimates of ICU patient cohorts with matched non-ICU patient and general population cohorts.25, 26, 27 The high long term mortality of both ICU and MET patient cohorts emphasises the fact that both groups of patients are severely unwell. Our findings could indicate that MET patients follow a similar health care trajectory to that of ICU patients, whose mortality outcomes are poor compared with a matched general population.25, 26, 27, 28

Altered MET criteria

Various reasons could explain the finding that altered MET criteria were associated with increased risk of long term mortality in the geriatric subpopulation. Patients who receive altered MET criteria are generally more unwell than other ward patients. That a patient’s “normal range” of vital sign observations sits outside the pre-specified parameters of the MET call criteria reflects this. However, other factors, such as experience of the doctor making the alteration, should also be considered.²⁹ A retrospective, single-centre study identified that nearly one-quarter of alterations to the MET criteria were made by junior doctors, and that altered MET trigger criteria were significantly associated with ICU admission, in-hospital cardiac arrest and in-hospital death.³⁰ Our findings reiterate that great care must be taken when altering MET criteria.

Tachypnoea

It is unclear why tachypnoea was associated with a reduced risk of long term mortality in patients aged < 75 years. This unexpected finding, which is in contrast to existing literature,³¹ may be accounted for by confounding. In the univariable analysis for patients aged < 75 years (Table 3), tachypnoea was associated with an increased risk of death (HR, 1.38; 95% CI, 1.07–1.78; P = 0.013). However, in the multivariable setting, a reversed direction of association (HR, 0.75; 95% CI, 0.56–1.00; P = 0.046) was observed. This finding may suggest that tachypnoea is associated with other measured correlates of mortality. If this was indeed the case, then the univariable tachypnoea result would be confounded, and the corresponding multivariable result would be adjusted for by at least some confounding factors. Further research is needed to better understand if this finding is best explained by confounding or by another unidentified mechanism.

Limitations

Our study has several limitations. Linkage with the VDR provided dates of death for deaths registered in Victoria but not elsewhere. Therefore, the mortality data included in this study could underestimate the number of deaths among the patient cohort. Further, unmeasured factors such as comorbid conditions may have confounded associations with long term mortality. In addition, although a comparison with a matched hospital cohort who did not receive a MET call would have been more appropriate than a comparison with population controls, this was infeasible in the given study setting due to restricted access to relevant data. As a result, risk factors presented for shorter median survival refer only to a within-group comparison. Finally, the single-centre nature of the study limits generalisability to other settings, both in Australia and internationally. However, the characteristics of both the MET patients and the rapid response system at Bendigo Hospital are similar to those of other settings described in the literature.7, 19, 20, 21

Strengths

Our study has several strengths. The large cohort size enhances the power of the statistical analyses and precision of results. Further, the analysed data were recorded in an established MET system. Although more research needs to be undertaken to better understand the long term mortality of MET patients⁸ as well as other outcomes, such as frailty and quality of life, the present study provides a basis for this research gap. These data may be of use to the clinician in having realistic conversations about end-of-life care with patients, including discussions around patients’ goals of care and the potential implications of receiving a MET call on their expected long term survival.

Conclusion

To our knowledge, this cohort study is the first to investigate long term survival among patients who received a MET call at an Australian regional hospital. Patient and MET call characteristics associated with long term mortality have been identified. More research needs to be conducted to better describe outcomes among MET patients.