You are on page 1of 5

RESEARCH

RESEARCH

A prospective before-and-after trial of a medical emergency team


Rinaldo Bellomo, Donna Goldsmith, Shigehiko Uchino, Jonathan Buckmaster, Graeme K Hart,
Helen Opdam, William Silvester, Laurie Doolan and Geoffrey Gutteridge

MOST HOSPITALS have cardiac arrest


ABSTRACT
teams that respond to in-hospital cardiac
arrests using modern technology and Objective: To determine the effect on cardiac arrests and overall hospital mortality of
standardised protocols. However, sur- an intensive care-based medical emergency team.
The Medical Journal of Australia ISSN:
vival 0025-729X
to hospital15 discharge in patients
September 2003 179 6 Design and setting: Prospective before-and-after trial in a tertiary referral hospital.
with 283-287
in-hospital cardiac arrests has Patients: Consecutive patients admitted to hospital during a 4-month “before” period
©The Medical
remained stable Journal of Australia
at between 2003
14.7% (May–August 1999) (n = 21 090) and a 4-month intervention period (November 2000 –
www.mja.com.au
(United States) and 16.7% (United
Research February 2001) (n = 20 921).
Kingdom) for 30 years.1 As several stud-
ies of in-hospital cardiac arrests suggest Main outcome measures: Number of cardiac arrests, number of patients dying after
that signs of clinical and physiological cardiac arrest, number of postcardiac-arrest bed-days and overall number of in-
instability may precede the arrest,2-4 hospital deaths.
introducing an intensive care-based hos- Results: There were 63 cardiac arrests in the “before” period and 22 in the
pital-wide preventive approach (a medi- intervention period (relative risk reduction, RRR: 65%; P < 0.001). Thirty-seven deaths
cal emergency team [MET]) might were attributed to cardiac arrests in the “before” period and 16 in the intervention
decrease the incidence of cardiac arrests period (RRR: 56%; P = 0.005). Survivors of cardiac arrest in the “before” period
and, consequently, hospital mortality. required 163 ICU bed-days versus 33 in the intervention period (RRR: 80%;
We tested this hypothesis by conducting P < 0.001), and 1353 hospital bed-days versus 159 in the intervention period (RRR:
a prospective trial comparing these out- 88%; P < 0.001). There were 302 deaths in the “before” period and 222 in the
come measures before and after intro- intervention period (RRR: 26%; P = 0.004).
ducing a MET.
Conclusions: The incidence of in-hospital cardiac arrest and death following cardiac
arrest, bed occupancy related to cardiac arrest, and overall in-hospital mortality
decreased after introducing an intensive care-based medical emergency team.
METHODS
MJA 2003; 179: 283–287
Hospital
The Austin and Repatriation Medical Cardiac arrest procedure recorded, and the information is
Centre comprises two major teaching The hospital’s cardiac arrest team (cor- entered into a computerised database.
hospital campuses (affiliated with the onary care nurse, cardiology registrar, The coronary care team also separately
University of Melbourne), one for acute- ICU registrar and anaesthesia registrar) records cardiac arrests, and the switch-
care and the other for longer-term, less is activated via the switchboard opera- board operators record all Code Blue
seriously ill patients. The acute-care tor, who calls for “Respond Blue to Ward calls. During our study, a dedicated
campus, where our study was con- X” via the internal public address and research nurse separately collected car-
ducted, admits about 60 000 patients per paging communication system. All diac arrest information. All these
year and has 21 intensive care unit (ICU) wards are equipped with resuscitation sources were used for verification of
beds. About 1700 patients are admitted trolleys with resuscitation drugs and data accuracy.
to our ICU each year. defibrillators. All cardiac arrests are
Medical emergency team
MJA Rapid Online Publication: 18 August, 2003 The MET system was structured so that
See also page 313
any member of the hospital staff could
activate it. The MET included the duty
Austin and Repatriation Medical Centre, Melbourne, VIC. intensive care fellow and a designated
Rinaldo Bellomo, MD, FJFICM, Director of Intensive Care Research; Donna Goldsmith, RN, Research intensive care nurse. If available, the
Nurse Intensive Care Unit; Shigehiko Uchino, MBBS, Research Fellow Intensive Care Unit;
receiving medical registrar was encour-
Jonathan Buckmaster, FJFICM, FANZCA, Staff Specialist in Intensive Care; Graeme K Hart, FJFICM,
FANZCA, Deputy Director of Intensive Care Unit; Helen Opdam, FJFICM, FRACP, Staff Specialist in aged to attend. An ICU consultant was
Intensive Care; William Silvester, FJFICM, FRACP, Staff Specialist in Intensive Care; available from 08:00 until 20:00, and
Laurie Doolan, FANZCA, Director of Operating Theatres; Geoffrey Gutteridge, FJFICM, FANZCA, attended if requested. After hours, an
Director of Intensive Care Unit. intensive care consultant was available
Reprints will not be available from the authors. Correspondence: Professor Rinaldo Bellomo, Austin and within 15–30 minutes for attendance if
Repatriation Medical Centre, Studley Road, Heidelberg, VIC 3084. Rinaldo.Bellomo@austin.org.au
required.

MJA Vol 179 15 September 2003 283


RESEARCH

The criteria for MET activation medical (no operation performed)


1: Criteria for initiation of a MET call*
(Box 1) were displayed promi- and analysed separately.
nently in each ward. The MET If one of these is present call 7777 and ask for the MET The primary reason for the
was activated by a pager call and ■ Staff member is worried about the patient MET call and the time of day of
by a public announcement inter- ■ Acute change in heart rate to <40 or > 130 beats/min the call were listed for each patient.
nal communication call “Medical ■ Acute change in systolic blood pressure to <90 mmHg ICU and hospital bed-days were
emergency team to Ward X”. The ■ Acute change in respiratory rate to < 8 or > 30 breaths/min obtained from the ICU and hospi-
MET carried an emergency pack ■ Acute change in pulse oximetry saturation to < 90%, tal electronic databases, and the
with drugs and equipment for despite oxygen administration number of in-hospital deaths was
resuscitation and endotracheal ■ Acute change in conscious state obtained from the hospital elec-
intubation. ■ Acute change in urine output to < 50 mL in 4 hours. tronic admission and discharge
database. Patients transferred to
*Criteria were listed on a large red poster placed prominently the long-term care campus were
Study design in all wards. considered as discharged from the
All patients admitted to the hospi- acute-care campus.
tal were considered as participants.
The study design was that of a prospec- Cardiac arrest was defined as the Ethical approval
tive before-and-after intervention trial, sudden onset of all of the following: We obtained approval from the Austin
with three periods: ■ Lack of palpable pulses;
and Repatriation Medical Centre Ethics
■ A 4-month “before” period (1 May ■ No detectable blood pressure; Committee to implement the MET and
1999 – 31 August 1999) during which the ■ Unresponsiveness; and to collect data related to the study. The
outcome measures were studied under ■ Documented initiation of basic life need for informed consent was waived
the normal operating conditions of the support. because consent to receive care accord-
hospital. All patients electronically recorded as ing to hospital emergency protocols was
■ A preparation and education period
admitted to the acute-care campus were considered implicit for each admission.
(1 September 1999 – 31 August 2000) included in the denominator for the
to introduce the MET. During this study. As surgical and medical patients
p e r i o d , ex te n s i ve an d r e p e ate d might be affected differently, patients Statistical analysis
presentations and discussions were held were identified as surgical (operation A computerised statistical package
with all members of the medical, nurs- performed during the admission) or (Statview) was used for data analysis
ing and paramedical staff. The MET
was then implemented (1 September
2000), and a run-in period of 2 months
2: Hospital population, patients having major surgery* and types of
was allowed. surgery, before and after introducing the medical emergency team
■ A 4-month “after” or intervention (MET). Data are number of patients or procedures (95% CI), unless
period (1 November 2000 – 28 Febru- otherwise indicated
ary 2001) during which the outcome
measures were studied under the new Before MET After MET
(availability of a MET) operating condi- (1/5/99 – 31/8/99) (1/11/00 – 28/2/01)
tions of the hospital. Medical admissions 8 974 (8 834–9 114) 8 377 (8 239–8 516)
To assess the effect of seasonal varia- Surgical admissions 12 116 (11 976–12 256) 12 544 (12 405–12 683)
tion, we obtained data on cardiac arrests Patients receiving major surgery 1 127 (1 065–1 189) 1 067 (1 006–1 128)
and hospital deaths for the same 4 months
Men 660 (628–692) 613 (582–644)
of the year as the intervention period 2
Women 467 (434–500) 454 (422–484)
years before introduction of the MET
(November 1998 – February 1999). Mean age (years) 60.7 (59.5–61.9) 60.2 (59–61.4)
Patients > 75 years 315 (286–344) 281 (253–309)
Outcome measures Major surgical procedures 1 369 (1 301–1 437) 1 313 (1 246–1 380)

■ The number of cardiac arrests (pri- Cardiac surgery 188 (163–213) 141 (119–163)
mary outcome measure); Thoracic surgery 142 (120–164) 117 (97–137)
■ The number of patients who died General surgery 288 (259–317) 318 (288–348)
from cardiac arrest; Orthopaedic surgery 253 (225–281) 285 (256–314)
■ The number of in-hospital deaths; Vascular surgery 160 (137–183) 132 (111–153)
■ The number of ICU bed-days occu-
Neurosurgery 147 (125–169) 111 (92–130)
pied by survivors of cardiac arrest;
Plastic surgery 77 (61–93) 84 (67–101)
and
■ The number of hospital bed-days Other (includes liver transplantation) 114 (94–134) 125 (104–146)
occupied by survivors of cardiac *Major surgery is defined as any surgery requiring a hospital stay > 48 hours.
arrest.

284 MJA Vol 179 15 September 2003


RESEARCH

and descriptive statistics.5 Fisher’s exact


3: Changes in number of cardiac arrests, bed-days and mortality, before
test was used for comparisons between
and after introducing the medical emergency team (MET)
the “before” and “after” periods, and
the ␹2 test was used for three-way com- Before After Difference Relative risk ratio
parison of the “before” period, the MET MET (95% CI) (95% CI)
“after” (intervention) period, and the No. of cardiac arrests 63 22 41 (23–59) 0.35 (0.22–0.57)
additional seasonal control period. Deaths from cardiac arrest 37 16 21 (7–35) 0.43 (0.26–0.70)
No. of days in ICU after 163 33 130 (110–150) 0.20 (0.13–0.33)
cardiac arrest
RESULTS
No. of days in hospital after 1353 159 1194 (1119–1269) 0.11 (0.09–0.13)
Box 2 shows the number and distribu- cardiac arrest
tion of medical and surgical admissions Inpatient deaths 302 222 80 (37–123) 0.74 (0.70–0.79)
during the period before and after the ICU = intensive care unit.
MET was introduced. There were no
significant differences.
sure, 35; heart rate change, 20; ranged from 1 to 31 days, with 10 patients
Cardiac arrests (Box 3) respiratory rate change, 18; and oliguria, staying for 3 days or less.
In the “before” period, there were 8974 2). Of the 24 patients who died after a
The MET attended each call within a MET call, 10 were designated “not for
medical admissions, compared with
mean (SD) period of 4.5 ( 2.2) minutes, resuscitation” before the call, and two
8377 in the intervention period. There
and was in attendance for a mean (SD) were so designated after the MET call.
were 33 cardiac arrests among medical
period of 19 (18) minutes. Different Three patients had a cardiac arrest at
patients in the “before” period, com-
units within the hospital activated the the time of the MET call and died
pared with 11 in the intervention period
MET in a relatively uniform way (Box during the call. The other nine patients,
(relative risk reduction [RRR], 66%;
5), but the MET was significantly more who were for full resuscitation, died a
P = 0.002).
There were 12 116 surgical admissions frequently activated during the evening median of 19 days after the call (range,
in the “before” period, compared with (48 calls from 16:00 to midnight versus 2–57 days). Considering “not for resus-
12544 in the intervention period. The 31 from 08:00 to 16:00 versus 20 from citation” orders and cardiac arrests sep-
number of cardiac arrests in surgical midnight to 08:00; P < 0.001). arately, survival after a MET call was
patients decreased from 30 to 11 (RRR, 89.6%, and none of the patients who
63%; P = 0.003). Therefore, the total MET procedures and outcomes were for full resuscitation died within 24
reduction in the number of cardiac arrests hours of a MET call.
The MET initiated and completed a vari-
was from 63 to 22 (RRR, 65%; ety of therapeutic, investigational and pro-
P < 0.001). None of the patients suffering cedural interventions (Box 6). Of the 99 In-hospital deaths
a cardiac arrest and receiving treatment MET calls, 18 resulted in an emergency There were 37 in-hospital deaths related
had “do not resuscitate” orders explicitly ICU or high dependency unit (HDU) to cardiac arrests in the “before” period
written in the patient progress notes. admission, with a total of 109 days and 18 and 16 in the intervention period (RRR
In the same 4-month period (seasonal days spent in ICU and HDU, respectively. for cardiac arrest deaths, 56%; P = 0.008).
control period) 2 years before the intro- The ICU/HDU stay for these patients There were a total of 302 inpatient deaths
duction of the MET, there were 51
cardiac arrests, which was not signifi-
cantly different from the number of 4: Number of cardiac arrests in the study periods
cardiac arrests in the “before” period 25
Number of cardiac arrests

(P = 0.3), but significantly different


from the number in the intervention 20
period (P = 0.001). Monthly cardiac
15
arrest data during these periods are
presented in Box 4. 10

5
Reasons for MET calls
In the intervention period, there were 99 0
Nov Dec Jan Feb May Jun Jul Aug Sep Oct Nov Dec Jan Feb
MET calls triggered by different, and
Seasonal control Before MET MET run- After MET
sometimes multiple, criteria for physio- Nov 1998 - 1 May 1999 - in Sep - 1 Nov 2000 -
logical instability (“worried about the Feb 1999 31 Aug 1999 Oct 2000 28 Feb 2001
patient”, 46; haemoglobin desaturation Educational period
on pulse oximetry, 37; change in con- 1 Sep 1999 - 31 Aug 2000
scious state, 28; low systolic blood pres-

MJA Vol 179 15 September 2003 285


RESEARCH

reduced number of postcardiac-arrest year for new interns (a possible seasonal


5: Proportion of medical
bed-days. For our institution, this would bias against the MET), whereas the con-
emergency team (MET) calls
from the different hospital units mean a yearly decrease of close to 3500 trol period did not.
bed-days. This suggests that a MET is The reduction in cardiac arrests was
General surgery 21%
associated with major cost savings and not due to “reclassification” of cardiac
Neurology 13% arrests into MET calls. There were three
increased hospital efficiency.
Cardiology 9% true cardiac arrests which occurred dur-
Our institution was able to continue
Nephrology 8%
to implement and sustain the MET
Cardiothoracic surgery 8%
system after the study period by adding
Othopaedic surgery 6% 6: Number of interventions and
one dedicated MET fellow to the inten- procedures implemented by the
Spinal 6%
sive care staff allocation. This staff Medical Emergency Team
Plastic surgery 4%
member responds to MET calls and
Oncology 4% Interventions
General medicine 3%
collects data, making possible continu-
Nasopharyngeal/oropharyngeal
Other (vascular surgery, thoracic
ing education and auditing. suctioning and additional oxygen 21
medicine, haematology, Introduction of the MET was associ- Administration of IV fluid bolus 18
neurosurgery, liver transplant) 18% ated with a 26% reduction in overall Administration of IV frusemide bolus 11
hospital mortality (three lives/1000 Initiation of non-invasive positive
admissions). To our knowledge, this is pressure ventilation by mask 9
in the “before” period compared with 222
deaths in the intervention period (RRR, the first before-and-after study of any Nebulised salbutamol 8
26%; P = 0.004) (Box 3). In the same 4- intervention that shows an impact on all- Temporary ventilation by bag and mask 6

month period 2 years before the introduc- cause hospital mortality. This effect was Suctioning of tracheostomy tube 6
tion of the MET, there were 275 deaths only partly accounted for by the impact Initiation of IV glyceryltrinitrate infusion 6
(P = 0.27 compared with the “before” of the MET on cardiac arrests. The Administration of anticonvulsants 5
period; P = 0.018 compared with the MET might, therefore, confer other ben- Administration of IV vasopressors 5
intervention period). efits, such as increasing awareness of the Insertion of a Guedel airway 4
consequences of physiological instability. Administration of IV morphine 4
It is also possible that the educational Insertion of a urinary catheter 4
Bed-days program to introduce the MET had an Cardioversion 3
After cardiac arrest, and in the absence impact on the care of acutely unwell Administration of IV ␤ blockers
of any change in the cardiac arrest patients. or digoxin 3
treatment protocol, survivors in the It is important to consider our study’s Administration of IV naloxone 2
“before” period required a total of 163 limitations. First, this trial was not dou- Transfer to operating room with
ICU bed-days and 1353 hospital bed- ble blind, or placebo-controlled or ran- ongoing resuscitation 2
days, and survivors in the intervention domised. It is not possible to have a Administration of IV metoclopramide 2
period required 33 ICU bed-days double-blind MET intervention, and Administration of IV ranitidine 2
(RRR, 80%; P < 0.001) and 159 hospi- introducing “sham” intervention as pla- Administration of IV insulin or glucose 2
tal bed-days (RRR, 88%; P < 0.001) cebo was ethically untenable, and “con- Insertion of new tracheostomy tube 1
(Box 3). tamination” (so-called Hawthorne Insertion of minitracheostomy tube 1
effect) would have been inevitable. Acute transfusion of red cells 1
Finally, a traditional, patient randomi- Administration of dexamethasone 1
DISCUSSION sation study of the MET would be Administration of intravenous
magnesium 1
We found that the incidence of in- ethically, scientifically and logistically
Administration of atropine 1
hospital cardiac arrests decreased by impossible in a single hospital.
Removal of central venous catheter 1
two-thirds after the introduction of a Our favourable findings may have been
Acute investigations
MET. This reduction, in both medical due to a high incidence of cardiac arrests
Chest x-ray 14
and surgical patients, is internally con- in the control period or an abnormally
Electrocardiogram 16
sistent and suggests a widespread low seasonal incidence in the interven-
Computed tomography scan 4
impact, irrespective of admission diag- tion period. Australian data show a car-
Arterial blood gases 36
nosis. It is also consistent with previous diac arrest incidence ranging from 36 to
Urea, creatinine, electrolytes and
observations that between 50% and 51 per 10 000 admissions.2,8 In the liver function tests 40
84% of in-hospital cardiac arrests are “before” period, there were 30 cardiac Invasive procedures
preceded by physiological instability.2- arrests per 10 000 hospital admissions, IV line insertion 18
4,6,7 and there was no statistically significant
By appropriately responding to Arterial line insertion 5
physiological instability, most in-hospi- seasonal variation in the incidence of Endotracheal intubation 3
tal cardiac arrests can be prevented. cardiac arrests in our hospital. Further- Central venous catheter insertion 3
With the MET there was also a more more, the 4-month MET intervention
than 50% reduction in the number of period included, by chance, 3 months
IV = intravenous.
cardiac arrest-related deaths, and a immediately after the start of the working

286 MJA Vol 179 15 September 2003


RESEARCH

ing a MET intervention, and these or advances in medical or surgical treat- 10. Lee A, Bishop G, Hillman K, Daffum K. The medical
emergency team. Anaesth Intensive Care 1995; 23:
arrests were counted as such. Further- ment that could explain a greater than 183-186.
more, most of the interventions were 60% reduction in cardiac arrests and a 11. Hurihan F, Bishop G, Hillman KM, et al. The medical
technically “simple”, suggesting that 25% reduction in overall mortality. emergency team: a new strategy to identify and
intervene in high-risk patients. Clin Intensive Care
timely intervention (within minutes Another recent study showing a possi- 1995; 6: 269-272.
rather than hours) at the time of deterio- ble beneficial effect of a MET9 had 12. Smith GB, Nolan J. Medical emergency teams and
cardiac arrests in hospital. Results may have been
ration might also decrease the complex- several methodological shortcomings.12 due to education of ward staff. BMJ 2002; 324: 1215.
ity of care required. Together with the Despite other indirect supportive evi- 13. Goldhill DR, White SA, Sumner A. Physiological
dominance of the “worried” criterion for dence,10,11,13-17 the MET approach has values and procedures in the 24 hours before ICU
admission from the ward. Anaesthesia 1999; 54:
activation of the MET, this also suggests not yet been adopted by most hospitals 529-534.
that some relatively simple acute inter- in Australia or elsewhere, and contro- 14. McQuillan P, Pilkington S, Allan A, et al. Confidential
ventions may appear too technically versy continues concerning its safety inquiry into quality of care before admission to
intensive care. BMJ 1998; 316: 1853-1858.
demanding to junior medical or nursing and effectiveness.8,12 15. Hillman KM. Redefining resuscitation. Aust N Z J
staff, or that such staff might lack the In conclusion, introducing an ICU- Med 1998; 28: 759-760.
experience to recognise that these inter- based MET in a teaching hospital 16. Daly FFS, Sidney KL, Fatovich DM. The medical
emergency team (MET): a model for the district
ventions are needed immediately. decreased the incidence of and deaths general hospital. Aust N Z J Med 1998; 28: 795-798.
Similarly, it is possible that our overall from cardiac arrests, postcardiac-arrest 17. Henderson SO, Ballestreros D. Evaluation of a hospi-
in-hospital mortality was high during tal-wide resuscitation team: does it increase survival
bed-days, as well as overall mortality. for in-hospital cardiopulmonary arrest? Resuscita-
the “before” period and was simply Further testing of this approach is now tion 2001; 48: 111-116.
restored to standard levels by the MET, needed in a variety of hospital and (Received 18 Feb 2003, accepted 2 Jul 2003) ❏
or fell because of seasonal variation. In geographical settings.
Australia, data from other large hospi-
tals show an overall crude mortality rate
between 138 and 184 deaths per 10 000 ACKNOWLEDGEMENTS
admissions.8 Our crude mortality rate in
We thank our Medical Records Department for their
the “before” period was 143 deaths per assistance with this project. This study was funded by a
10 000 admissions, and there was no grant from the Quality Improvement Branch of the Acute
statistically significant seasonal variation Health Care section of the Victorian Department of Human
Services.
in in-hospital mortality.
Our findings within a single institu-
tion might not apply to other hospitals. COMPETING INTERESTS
Institution-specific heuristics and
unique administrative features may have None identified.

enhanced the impact of the MET


approach. However, our institution has
all the organisational, structural and REFERENCES
logistic features of a typical tertiary 1. Saklayen M, Liss H, Markert R. In-hospital cardiopul-
referral hospital. Another possibility is monary resuscitation. Medicine 1995; 74: 163-175.
that our implementation of a MET may 2. Buist MD, Jarmolowski E, Burton PR, et al. Recog-
nising clinical instability in hospital patients before
have differed from that of other institu- cardiac arrest or unplanned admission to intensive
tions,9-11 but whether implementation care. Med J Aust 1999; 171: 22-25.
has an impact on its efficacy is not 3. Franklin C, Mathew J. Developing strategies to pre-
vent in-hospital cardiac arrest: analyzing responses
known. We believe that our approach is of physicians and nurses in the hours before the
simple and low cost. It is also possible event. Crit Care Med 1994; 22: 244-247.
that the decrease in cardiac arrests was 4. Shein RMH, Hazday N, Pena M, et al. Clinical
antecedents to in-hospital cardiopulmonary arrest.
secondary to some other improvements Chest 1990; 98: 1388-1392.
in patient care between the “before” 5. Statview [computer program], version 4. Berkeley,
and “after” periods. However, there Calif: Abacus Concepts Inc, 1996.
6. Hodgetts TJ, Kenward G, Vlachonikolis IG, et al. The
were no changes in the structure, refer- identification of risk factors for cardiac arrest and
ral pattern or activity of our hospital, formulation of activation criteria to alert a medical
with the total number of admissions emergency team. Resuscitation 2002; 54: 125-131.
7. Smith AF, Wood J. Can some in-hospital cardio-
during the two study periods remaining respiratory arrests be prevented? A prospective
essentially unchanged (< 1% change in survey. Resuscitation 1998; 37: 133-137.
the denominator for the study out- 8. Bristow P, Hillman KM, Chey T, et al. Rates of in-
hospital arrests, deaths, and intensive care admis-
comes). Furthermore, there were no sions: the effect of a medical emergency team. Med
changes in “not for CPR” policy, hospi- J Aust 2000; 173: 236-240.
tal admission policy, discharge practices 9. Buist M, Moore GE, Bernard SA, et al. Effects of a
medical emergency team on reduction of incidence
or surgical casemix during the study. We of and mortality from unexpected cardiac arrests in
are also not aware of any improvements hospital: preliminary study. BMJ 2002; 324: 387-390.

MJA Vol 179 15 September 2003 287