Transcript Determinants of Post-Intensive-Care

At the boundaries of the ICU:
Triage and Post-ICU Mortality
Elie AZOULAY, MD, PhD
Medical ICU, Saint-Louis Hospital
Paris, France, Europe
The patient’s journey
… Mortality / Morbidity / Sequelae / QOL ...
Triage
Admission
Discharge
ICU stay
Hospital stay
PIM
End-Of-Life
DFLST
Withdrawing/Withholding
Short-term outcome
One-year mortality
Morbidity
Neurocognitive disorders
Psychiatric symptoms
Long-term sequelae
Quality of life
The spectrum of triage...
Patients/
families
Selftriage
ER
Pre-ICU
triage
ICU
ICU
triage
Hospital
Post-ICU
triage
Home
Levin PD, Intensive Care Med 2001
Triage to the ICU
When evaluating a patient with a severe acute
illness, the ICU physician must assess:
(i) the diagnosis, prognosis, and treatment:
reversibility, time from treatment onset ...
(ii) whether or not ICU admission is warranted
(iii) and if it is, whether the patient, if competent,
consents to ICU admission.
Wanzer SH. N Engl J Med 1984.
Triage to the ICU
The answer to the second question is a daily dilemma
for ICU physicians.
Its determinants have been reported as related to:
(i) the number of beds available in the ICU
(ii) patient characteristics and comorbidities
(iii) and the characteristics of the acute illness (severity,
reversibility, and predicted residuals and quality of life after ICU discharge)
(“ reasonable prospect of substantial recovery ”) SCCM. Crit Care Med 1999
Evaluating prognosis is difficult

Poor prediction of individual outcomes with statistical
models
Chang, Lancet 1989;2:143-146

Doctors need help to critically appraise studies on
prognostic information. Randolph Crit Care Med 1998;26:767-72

Prognostic judgements depend on the experience of the
doctors.
Poses Arch Intern Med 1990; 150: 1874-1878

Doctors are often not completely honest with their
patients.
Vincent, Intensive Care Med 1998;24:1237-8
Discriminating ability of the APACHE
II and SAPS II
AUC APACHE II
AUC SAPSII
0.60
0.67
Sculier et al. Crit Care Med 2000;28:2786-2792
North American and European recommendations
- Because the decision to recommend ICU admission is
based on complex criteria, and in order to ensure
appropriate utilization of ICU resources
-To avoid depriving patients of a chance to recover
Recommendations for ICU admission, triage, and
discharge have been established for:
- cardiac and trauma patients,
- patients with gastrointestinal hemorrhage
- patients with any critical disorder
Crit Care Med 1999;27:1073-79
Priorities for ICU admission
 Priority
1: critically ill patients requiring invasive
treatments or monitoring (MV, vasopressors, dialysis …)
 Priority
2: intensive monitoring for unstable patients
 Priority
3: critically ill or unstable patients with little chance
of recovery (comorbidities, nature of the acute medical
disease)
 Priority 4: patients with little benefit expected from the ICU
 management without ICU admission: not severe enough for the ICU
 dying patients
Alternatives to ICU admission

Transfer to the ward
SCCM. Crit Care Med 1999

Management by the ER
Rausch PG. Ann Emerg Med 1991
Pavey B. Ann Emerg Med 1992

Palliative care
Alleviate symptoms
Adequate use of technology
Improvement in communication skills
Management of family needs, respect for religious values
Sedation, analgesia, oxygen, tracheal aspiration
Ethical consultation
Caregivers available
Adam J. BMJ 1997
Falk S. BMJ 1997
In practice …prevalence of refusal
38% Hong Kong
Joynt GM, Intensive Care Med 2001
28% Israel
Sprung CL, Crit Care Med 1999
26% United Kingdom
Metcalfe MA, Lancet 1997
22% France
Azoulay E, Crit Care Med 2001
Determinants of ICU refusal
ISRAEL
FRANCE
* age
* severity scores
* bed availability
* surgical status
* admission diagnosis
* age
* chronic health status
* comorbidities (cancer, COPD, CHF)
* admission diagnosis
Hong Kong
* age
* severity scores
* cardiac arrest
* neuro-surgical status
* comorbidities (cancer)
Sprung CL, Crit Care Med 1999 27:1073-9
Azoulay, Crit Care Med 2001;29:2132-6
Joynt GM, Intensive Care Med 2001
Cumulative survival estimates of triaged patients
who were admitted, admitted later, and never
admitted (controlling for APACHE II scores)
RR = 2.5; p < .01
Sprung CL et al. Crit Care Med 1999;27(6):1073-1079
Recommendations
Patient’s opinion
Expected cost of treatment i f patient admitted
Family’s opinion
Bed available in another ICU
General practitioner’s opinion
Severity of underlying disease
QOL as evaluated by the fa mily
Reversibility of underlying disease
QOL as evaluated by the patient
Reversibility of acute condition
QOL as evaluated by the intensi vist
Ti me to optimal treatment
Religious values o f the patient
Potential handicap
ICU admission for an iatrogenic event
Expected chance of sur vival i f patient admitted
Burdens and a dvantages o f a new treatment
Consequences for the fa mily
Investigations might better explain the event Social context
JAMA 1994;271:1200-1203
Crit Care Med 1994;22:358-62
SCCM pejorative diagnoses
Does the patient have any of the following?
Refractory leukemia associated with multiple organ failure
I__I
Persistent vegetative status or brain death without possibility of organ donation
I__I
Acute respiratory failure associated with refractory leukemia
I__I
Chronic respiratory or cardiac failure, or metastatic cancer without therapeutic resource
I__I
JAMA 1994;271:1200-1203
Crit Care Med 1994;22:358-62
The example of cancer patients

Carlon GC.
Just say no …

Schuster D.P.
Everything that should be done-not everything that can be done.

G.D. Rubenfeld
Withdrawing life support from
MV recipients of BMT

Brunet F.
Is ICU justified for hematological
patients?
Crit Care Med. 1989;17:106-7
Ann Intern Med 1996;125;625-30
Am Rev Respir Dis 1992;145:508-9
Intensive Care Medicine 1990;16:291-7
Autologous PBSCT Patients
Variables
1991 - 1993
1993 - 1995
1995 - 1997
1997 - 1999
Patients
Hosp. survival
7 (9)
0 (0)
13 (17)
2 (15)
25 (33)
7 (27)
32 (42)
11 (35)
No or 1-organ failure
Patients
7 (100)
Hosp. Survival 0 (0)
11 (85)
2 (18)
18 (72)
7 (39)
24 (75)
10 (42)
Two-organ failure
Patients
0 (0)
Hosp. Survival 0 (0)
2 (15)
0 (0)
7 (28)
0 (0)
8 (25)
1 (13)
Khassawneh et al. Chest 2002;121:185-188
Critically Ill Cancer Patients
% deaths
Number of patients
100
1
80
0.8
60
0.6
40
0.4
20
0.2
0
0
1990
1991
1992
1993
1994
1995
1996
Year of ICU admission
1997
1998
1999
Patient selection (1)
1992-1995
1996-1998
n=41 (%)
n=34 (%)
Knaus scale C or D
26 (66.5)
13 (38.2)
0.02
Stage III disease
34 (83)
21 (62)
0.03
SAPS II score at admission
54 (38-70)
64 (43-82)
0.05
Dialysis
9 (22)
15 (44)
0.04
NIMV
2 (5)
7 (20.6)
0.03
31 (75.6)
12 (35)
0.0008
Myeloma patients
P
Need for :
30-day mortality
Changing use of ICU … Azoulay et al. Intensive Care Medicine 1999;25:1395-1401
Patient selection (2)
1991-1995
n=132 (%)
1996-1998
n=105 (%)
Complete remission
28 (21.2)
34 (32.3)
0.04

17 (12.8)
48 (36.6)
25 (23.8)
43 (40.9)
0.02
0.50
SAPS II score at admission
55 (38-70)
62 (43-82)
0.005
Need for :
Vasopressors
Dialysis
NIMV
Conventional MV
End-of-life decision
63 (47.7)
25 (18.9)
19 (14.4)
113 (85.6)
25 (18.9)
55 (52.4)
29 (27.7)
29 (27.6)
76 (72.4)
17 (16.2)
0.59
0.11
0.01
0.01
0.58
30-day mortality
108 (81.8)
64 (60.9)
0.0003
P
MV patients

BMT
Neutropenia
Azoulay et al. Crit Care Med 2001;29:519-525
Patient selection (3)
1995 - 1997
1998 -2000
(n=34)
(n=54)
21 (61.8)
31 (57.4)
0.79
Poor chronic health status (Knaus scale C or D) 11 (32.3)
9 (16.6)
0.05
Hospitalization >48h before ICU admission
31 (91.1)
40 (74)
0. 03
LOD Day 1
8 (6-11)
7 (4-9)
0.49
DLOD
0 (-0.25-0.25) 0 (-0.75-0.25) 0.20
SAPSII score at ICU admission
72 (53-86)
65 (45-93)
0.33
Mechanical ventilation
31 (93.9%)
37 (68.5%)
0. 005
Noninvasive ventilation
4 (11.8)
8 (14.8)
0.12
Volume of crystalloid infused on day 1 (ml)
141 (0-500)
444 (0-1200)
0. 03
End-of-life decisions
6 (17.6)
11 (20)
0.54
30-day mortality
27 (79.4)
30 (55.5)
0. 01
Remission of the malignancy
P value
Larché J, Azoulay E. Submitted
Compliance with Triage-to-ICU
Recommendations (1)
Number of recommendations observed
20
16
P=0.0003
12
Admission denied after patient examination
8
Admission denied over the phone
4
0
Compliance with Triage-to-ICU
Recommendations (2)
Number of recommendations observed
20
16
P<0.0001
12
8
4
0
Beds available
Full unit
Who needs recommendations ?
Patients/
families
Selftriage
ER
Pre-ICU
triage
ICU
ICU
triage
Hospital
Post-ICU
triage
Home
Levin PD, Intensive Care Med 2001
The patient’s journey
… Mortality / Morbidity / Sequelae / QOL ...
Triage
Admission
Discharge
ICU stay
Hospital stay
PIM
End-Of-Life
DFLST
Withdrawing/Withholding
Post ICU Mortality

Mortality after discharge from intensive care has been
reported to range from 6.1 to 27%.
Latour J, Intensive Care Med 1990; 16:125-7.
Goldhill DR, CritCareMed1998;26:1337-45.

Rowan KM, Bmj 1993; 307:972-7.
Munn J, Anaesthesia 1995; 50:1017-21.
Death after ICU discharge can be related to factors
occurring:
before
Bion J. Bmj 1995; 310:682-3.
Ryan DW. Bmj 1996; 312:654.
or
after the ICU stay
Wallis CB, Anaesthesia 1997; 52:9-14.
Dragsted L, Eur J Anaesthesiol 1989; 6:385-96.
Post ICU Mortality

In-hospital mortality was higher among patients
discharged at night.
Goldfrad C, Rowan K. Lancet 2000; 355:1138-42.

Keeping at-risk patients in the ICU for another 48
hours may reduce mortality after ICU discharge
by 39%.
Daly K. Bmj 2001; 322:1274-6.
Determinants of
Post-Intensive-Care-Unit
Mortality (PIM)
The OUTCOMEREA Study Group
http//www.outcomerea.org
Patients and Methods

Prospective multicenter study to report determinants
of PIM, with particular attention to EOL decisions in
the ICU.

OUTCOMEREA DATABASE®
– 2-year study period in 6 medical and surgical ICUs
– All adult patients consecutively hospitalized for longer than
48 hours were enrolled starting in January 1997.
– Variables were collected daily on standardized forms. All
patients were screened for end-of-life decisions, ICU
discharge, or death. After ICU discharge, all patients were
screened on the wards for hospital length of stay and for
death or hospital discharge.
Patient management and
end-of-life decisions

Management policies were identical in the 6 ICUs.

Transferred patients were discharged to the source
wards. When ICU admission occurred directly
(emergency room or mobile emergency medical
system), discharge was to a same-hospital ward.

The EOL decision procedure included at least 2
deliberations and involvement of more than 2 ICU
physicians and the patient’s nurse or of all
caregivers
Patient characteristics
1385 patients
were discharged alive
DFLST implemented
in 80 patients
P I M = 10.8%
47
died
1235 were discharged
alive from the Hospital
150 died during their
post-ICU hospital stay
Comparison between post-ICU decedents
and post-ICU survivors
Discharged from
the ward
n=1235 (89.2%)
Age (years)
63 (47-73)
Female gender, n (%)
458 (37)
Transfer from the ward, n (%) 548 (44.4)
Type of patients, n (%)
Medical
791 (64.1)
Surgical
130 (10.4)
Scheduled surgery
314 (25.4)
Died in the
ward
n=150 (10.8%)
72 (61-79)
47 (31.3)
99 (66)
Chronic Disease, n (%)
485 (39.3)
94 (62.7)
<0.0001
Mc Cabe 1, n (%)
734 (59.4)
41 (27.3)
<0.0001
116 (77.3)
7 (4.6)
27 (17.8)
P value
<0.0001
0.05
<0.0001
0.11
P
Discharged
Died in the
value
from the ward ward
n=1235 (89.2%) n=150 10.8%)
Severity and organ failure scores
At admission
 SAPS II
 LOD
 SOFA
At discharge
 SAPS II
 LOD
 SOFA
Omega Score of Workload
Omega 1
Omega 3
Patients discharged Fri-Sunday
Patients with end-of-life decisions
35 (25-47)
4 (3-5)
3 (1-5)
42 (34-56)
5 (3-6)
5 (3-8)
0.0001
0.0001
<0.0001
22.5 (18-27)
1.5 (1-2)
1.5 (1-2)
35 (28-42)
3 (2-5)
3 (1-5)
<0.0001
0.0005
<0.0001
9 (1-18)
48 (24-115)
12 (6-24)
126 (28-370)
0.001
<0.0001
308 (24.9)
29 (19.3)
0.13
33 (2.6)
47 (31.3)
<0.0001
Multivariate analysis: Independent
predictors of Post-ICU-mortality
Odds ratio 95% CI
Mc Cabe class 1
0.388
[0.258-0.583]
Transfer from the ward
1.89
[1.27-2.80]
SAPS II score at admission > 36
1.57
[1.60-2.33]
SOFA score at discharge
1.11
[1.03-1.18] per point
End-of-life orders
9.64
[5.75-16.16]
Determinants of
Post-Intensive-Care-Unit Mortality
in Critically Ill Septic Patients:
An International study
For the European Sepsis Group
Flow-chart of patients
14 364 ICU admissions
3 034 patients infected
at ICU admission
11 330 patients not infected
at ICU admission
868 patients with ICUacquired infection
P I M = 10.4%
3902 patients
with first infection
1913 removed
1989 alive at ICU discharge
1782 alive at hospital discharge
207 dead at hospital discharge

1301 dead at ICU discharge

311 alive at ICU discharge but
LOS > 28 d

301 alive at ICU discharge with
no hospital follow-up
Comparison between post-ICU
decedents and survivors (1)
Variable
All
N (% Death)
OR [95 CI]
p-value
Surgical scheduled
231
19 (8.2%)
1
Medical
1248
143 (11.5%)
1.44 [0.87 ; 2.38]
0.15
Surgical emergency
352
39 (11.1%)
1.39 [0.78 ; 2.47]
0.26
Trauma
158
6 (3.8%)
0.44 [0.17 ; 1.12]
0.09
Chronic pulmonary failure
108
18 (16.7%)
1.79 [1.06 ; 3.04]
0.03
Chronic heart failure
175
32 (18.29%)
2.10 [1.39 ; 3.17]
0.0005
Chronic renal failure
168
37 (22.02%)
2.74 [1.84 ; 4.09] < 0.0001
Cirrhosis
98
20 (20.4%)
2.34 [1.40 ; 3.91]
Immunossupression
494
82 (16.6%)
2.18 [1.62 ; 2.94] < 0.0001
Chronic Obstructive Pulmonary Disease
349
45 (12.89%)
1.35 [0.95 ; 1.92]
Admission categories
Comorbidities
0.001
0.09
Comparison between post-ICU
decedents and survivors (2)
Variable
All
N
OR
(%deaths)
[95% CI]
1
P value
Mac Cabe
1
1345
106 (7.88%)
2
521
73 (14.01%)
1.90 [1.38 ; 2.62] < 0.0001
3
123
28 (22.76%)
3.45 [2.16 ; 5.50] < 0.0001
Renal
142
25 (17.61%)
1.96 [1.24 ; 3.09]
0.004
Shock
397
52 (13.10%)
1.40 [1.00 ; 1.96]
0.05
Inotropic or vasopressor use
747
103 (13.79%) 1.74 [1.31 ; 2.33]
Ventilation support
1455 162 (11.13%) 1.36 [0.96 ; 1.92]
0.08
Renal support
213
0.11
Reason of ICU admission
29 (13.62%)
1.41 [0.92 ; 2.15]
0.0002
Comparison between post-ICU
decedents and survivors (3)
Variable
Age
Survivors
Non survivors p values
60 (43 - 71)
67 (55 - 75)
< 0.0001
3.5 (2.5 - 4.4)
4.2 (3.5 - 5.3)
< 0.0001
LOS in ICU
2 (1 – 3)
3 (2 – 4)
0.05
LOD at ICU-discharge
2 (1 – 4)
4 (2 – 6)
< 0.0001
SAPS II at the first day of infection
Comparison between post-ICU
decedents and survivors
Variable
All
N (% Death)
629
98 (15.58%)
2,32 [1,68 ; 3,20] < 0,0001
Microbiologically documented
1242
155 (12.5%)
1.91 [1.37 ; 2.65]
0.0001
Site of infection
Pulmonary
Digestive
Urinary
Primary bacteremia
Unknown
Other
1197
262
208
202
44
327
107 (8.94%)
32 (12.21%)
26 (12.50%)
32 (15.84%)
3 (6.82%)
34 (10.40%)
0.68 [0.51 ; 0.91]
1.23 [0.83 ; 1.84]
1.26 [0.81 ; 1.96]
1.73 [1.15 ; 2.61]
0.62 [0.19 ; 2.03]
0.99 [0.67 ; 1.47]
0.01
0.31
0.30
0.01
0.43
0.99
Origin of infection
Hospital-acquired
OR [95 CI]
p-value
Documentation of infection
Ho
sp
PS ita
II t l-ac
he
qu
ir
f ir
s t ed
in
d
Cl
fe
LO ay
in
ct
ic
D
o
io
al
f
at
n
in
ly
do ICU f ec
c
t
Re um -dis ion
sp
c
e
ira nte har
d
ge
to
in
ry
f
e
Su s ite c tio
n
rg
ic inf e
al
ct
sc
e
he d
du
Im
m
le
un
d
os
A
s
Ch
up
g
ro
re e
ni
s
s
c
pu
C ion
lm irrh
In
on
ot
os
ro
ar
is
pi
y
c
f
a
or
il
v a Ma ure
s
c
o
Le
Ca
ng S a pre
s s be
ur
th
eu
or
of
IC s in us
U
fe e
Di
s
sh
Su tay c tio
ar
ns
ge rgic >1
4
a
du
d
l
rin em ay
s
er
g
ge
th
e
n
Re Fu We c y
ek
s p ng
ira al i -en
n
d
to
ry f ec
tio
d
y
Re
n
na s f u s
l d nc
y s tio
n
M f un
M
ec ed
ct
ha ic a ion
ni
c a l pa
t
Pr
im l v e ient
nt
ar
IC
ila
y
Utio
ac ba
c
qu
te n
r
Co
ire
em
d
cc
ia
in
ig
fe
ra
c
tio
m
n
po
s i Dia
t
l
Sh
y
if
oc inf s is
ec
k
at
ti
ad ons
m
is
si
on
Co
m
a
SA
Bootstrap for variables selection
Results of the bootstrap technique: % selected from the 500 samples
100
90
80
70
60
50
40
30
20
10
0
Multivariate analysis:
Independent predictors of Post-ICU-mortality
variable
OR adjusted
95 IC
p value
Age
1.02
[1.01 ; 1.03]
0.001
Immunossupression
1.92
[1.37 ; 2.68]
0.0002
Cirrhosis
1.75
[1.00 ; 3.06]
0.05
Chronic Pulmonary failure
2.05
[1.16 ; 3.63]
0.01
Medical
1.86
[1.08 ; 3.23]
0.02
SAPS II at the first day of infection
1.17
[1.04 ; 1.32]
0.01
LOD at ICU discharge
1.23
[1.15 ; 1.31] < 0.0001
Hospital-acquired
1.72
[1.21 ; 2.45]
0.002
ICU-acquired
1.54
[0.96 ; 2.49]
0.07
Infection of the respiratory site
0.69
[0.50 ; 0.95]
0.02
Microbiologically documented
1.77
[1.23 ; 2.55]
0.001
Admission categories
Origin of infection
Discussion (1)

Patients at risk of dying after ICU discharge were
those with poorer chronic health status, more
severe acute disease at ICU admission, and
prolonged severity of illness and organ failure.

The most powerful predictors of PIM were
implementation of end-of-life decisions during
the ICU stay and the importance of severity at
ICU discharge.
DFLST

Decisions to withhold/withdraw treatment have been
recorded in half the patients who died in the ICU
Ferrand, Lancet 2001,
Pochard, Crit care Medicine 2001

Patients for whom end-of-life decisions were made in
the ICU had lower ICU and hospital mortality rates
in the late 1980s than ten years earlier.
Jayes RL, Jama 1993; 270:2213-7

However, no studies have been specifically designed
to evaluate the impact of end-of-life decisions made in
the ICU on mortality between ICU discharge and
hospital discharge (post-ICU mortality).
Discussion (2)

The 10% PIM rate reported in these studies is in
agreement with previous reports.

Moreover, patients who died in the hospital after
ICU discharge had significantly higher severity-ofillness scores on the day of discharge than those who
survived.

This suggests that patients at risk for PIM may have
been discharged from the ICU with incomplete
resolution of their acute medical condition.
Discussion (3)
 The
appropriateness of established discharge
criteria depends on whether a decision to
limit therapy was taken in the ICU.
 The
incidence of end-of-life decisions taken on
the ward was not determined in this study.
Discussion (4)

A growing proportion of patients are discharged
alive from the ICU despite previous implementation
of end-of-life decisions.

In the present study, 58.8% of patients who were
discharged alive from the ICU after an end-of-life
decision died before hospital discharge. In these
patients, ICU discharge was an appropriate response
to an awareness that further intensive care would be
futile.
Conclusion

Identifying patients at risk for premature ICU
discharge may help physicians to solve the clinical
dilemma of whom to discharge to make room for a
patient requiring urgent admission to the unit.

However, the appropriateness of discharge criteria
should be evaluated in the light of the patient’s end-oflife decision status.

Further studies are needed to assess risk factors for
PIM according to whether end-of-life decisions are
implemented in the ward.
Need to focus on long-term outcomes
Post-ICU mortality
 One-year mortality
 Morbidity
 Neurocognitive disorders
 Psychiatric symptoms (depression +++)
 Intensive feeding and malnutrition
 Long-term sequelae
 Quality of life
