Open Journal of Nursing, 2012, 2, 289-300 OJN
http://dx.doi.org/10.4236/ojn.2012.223043 Published Online November 2012 (http://www.SciRP.org/journal/ojn/)
The clinical utility of preoperative surgical risk indices and
ICU bed allocation on outcomes of noncardiac surgical
patients: A cohort study
Demetrios J. Kutsogiannis1, Sean Norris1, Becky K. L. Leung2
1Division of Critical Care Medicine, Faculty of Medicine and Dentistry, University of Alberta, Edmonton, Canada
2Division of Cardiology, Faculty of Medicine and Dentistry, University of Alberta, Edmonton, Canada
Received 5 September 2012; revised 8 October 2012; accepted 20 October 2012
ABSTRACT
Summary statement: In non-cardiac surgical patients,
respiratory failure index and intensivists’ (expert)
opinion predicted postoperative mortality and respi-
ratory failure. Intermediate risk patients allocated to
postoperative ICU care vs. surgical high intensity
care demonstrated increasing lengths of hospital stay.
Background: No guidance exists for allocating post-
operative ICU resources for patients undergoing non-
cardiac surgery. We determined the predictive value
of preoperative risk sores and “expert opinion” in
predicting postoperative mortality and complica-
tions. Methods: A cohort study involving 403 adults
undergoing elective noncardiac surgery and being as-
sessed in a preoperative clinic within a university af-
filiated tertiary care hospital. Postoperative outcomes
included 30-day mortality, respiratory failure at 48-
hour, unplanned intubation, cardiac composite score,
hospital length of stay, hypotension, hypertension,
and delirium. Results: Preoperative respiratory fail-
ure index (PRFI) predicted 30-day mortality (OR
1.11, 95% CI 1.04 to 1.19). An intensivist’s opinion
predicted respiratory failure 48-hour postoperatively
(OR 28.70, 95% CI 7.44 to 110.70). Patients with an
equivalent PRFI risk had a longer hospital stay (17.2
v. 8.9 days, P = 0.01), increased respiratory failure
risk (P = 0.009), hypertension (P = 0.009), hypoten-
sion (P = 0.005) and delirium (P = 0.05) if allocated to
an ICU bed versus a high-intensity bed. Conclusions:
PRFI predicts 30-day postoperative mortality and
cardiac events. A decision to allocate an ICU bed pre-
dicted the development of postoperative respiratory
failure. Patients with an intermediate PRFI risk and
allocated to an ICU demonstrated increasing lengths
of hospital stay and morbidity.
Keywords: Risk Factors; Comorbidity; Postoperative
Complications
1. INTRODUCTION
The aging of the United States population in the next two
decades will increase the burden of acute and chronic
illness and the demand for critical care services [1]. In
2004, 33% of Medicare hospitalizations had intensive
care unit (ICU) or coronary care unit care representing an
annual increase in costs of 36% to $32.3 billion from
1994 [2]. In Ontario, the crude incidence of mechanically
ventilated adults with noncardiac surgical and medical
diagnoses between 2000 and 2026 is projected to in-
crease 31% from 222 to 291 per 100,000 adults. As a
significant proportion of critical care is devoted to the
care of postoperative patients, improved efficacy is ex-
pected in the delivery of critical care services to elective
postoperative patient [3]. Several statistical models have
been validated to risk stratify patients undergoing non-
cardiac surgery and using both cardiac and respiratory
outcomes [4,5]. Two of the most robust models include
the preoperative respiratory failure risk index (PRFI) and
the revised cardiac risk index (RCRI) [6,7]. In an effort
to reduce the heterogeneity of post-operative morbidity
and mortality across hospitals, post-operative events such
as respiratory failure, myocardial infarction and surgical
site infections represent benchmarks for measuring the
variation in quality of hospitals [8,9]. However, despite
large numbers of patients undergoing noncardiac opera-
tions worldwide, there are no randomized trials demon-
strating the effectiveness of ICU care for subgroups of
noncardiac surgical patients [10].
The primary objective of this cohort study was to de-
termine the predictive value of the preoperative respira-
tory failure index (PRFI) the revised cardiac risk index
(RCRI), and the “expert opinion” of a group of intensive
care physicians after consultation with anesthesiologists,
in predicting the development of post operative respira-
tory and cardiac events and mortality in a population of
elective noncardiac surgical patients. The secondary ob-
jective of this study was to explore whether there existed

D. J. Kutsogiannis et al. / Open Journal of Nursing 2 (2012) 289-300 290
any difference in complication rates and length of hospi-
tal stay between those patients within similar PRFI cate-
gories allocated to and receiving a postoperative ICU bed
versus those patients not allocated to nor receiving a
postoperative ICU bed by “expert opinion”.
2. MATERIALS AND METHODS
From January 2002 to December 2004, patients were
consented in a preoperative clinic of a 550 bed, univer-
sity affiliated tertiary care hospital in Edmonton, Canada.
The study was approved by the Human Research Ethics
Board, University of Alberta, and 403 provided written
informed consent to participate. Inclusion criteria in-
cluded all patients undergoing abdominal, vascular, tho-
racic and head and neck procedures and requiring the
preoperative consultation of at least one non-surgical
specialist. For those undergoing general, orthopaedic or
neurosurgical procedures, an additional requirement of at
least one comorbid medical condition was required for
enrolment into the study. Orthopaedic surgical patients
were only included during the first year of the study
(Appendix 1).
During each week, one of seven rotating intensivists
was responsible for allocating a postoperative ICU bed,
surgical high intensity bed or surgical ward bed during
the patient’s preoperative risk assessment based on their
“expert opinion”. Intensivists were not provided with
PRFI and RCRI scores. Preoperative cardiac risk was de-
termined using the RCRI which previously utilized a de-
rivation cohort of 2893 patients [7]. In the validation co-
hort of 1422 patients, a RCRI score of 0 (RCRI-I), 1
(RCRI-II), 2 (RCRI-III), and ≥3 (RCRI-IV) predicted a
0.4%, 0.9%, 6.6%, and 11.0% frequency of major car-
diac complications. The history of angina and congestive
heart failure were recorded using the Canadian Cardio-
vascular Society Classification and the New York Heart
Association Classification respectively [11,12]. Postop-
erative respiratory risk was determined using the PRFI
which has been previously derived using a cohort of
81,719 patients undergoing noncardiac surgery. In the
validation cohort of 99,390 patients, a PRFI score of ≤10
(PRFI-I), 11 - 19 (PRFI-II), 20 - 27 (PRFI-III), 28 - 40
(PRFI-IV) and >40 (PRFI-V) predicted a 0.5%, 2.2%,
5.0%, 11.6% and 30.5% frequency of respiratory failure
respectively [6]. Within the PRFI score, patients under-
going abdominal aortic aneurysm or thoracic surgery
score the highest number of points followed by neuro-
surgery/upper abdominal/peripheral vascular and neck
surgery. Body mass index was defined as the weight in
kilograms divided by the square of the patient’s height in
meters. Forced expiratory volume in one second was
measured either by formal pulmonary function testing or
using bedside spirometry. If the “expert opinion” of the
intensivist deemed that the patient required the assign-
ment to an ICU bed, the surgery was not undertaken
unless an ICU bed was available on the morning of the
planned surgery. The surgical high-intensity unit was
staffed with a 1:2 nurse-to-patient ratio with continuous
arterial blood pressure monitoring, pulse oximetry, and
non-invasive ventilation if necessary. However, the high-
intensity unit could not provide mechanical ventilation or
inotropic/vasopressor support. The surgical ward did not
offer continuous blood pressure monitoring or continu-
ous oximetry, non-invasive or mechanical ventilation, or
inotropic/vasopressor support.
Predictor (independent) variables for the primary post-
operative outcomes included “expert opinion”, RCRI,
PRFI, body mass index (BMI), and forced expiratory
volume in 1 second (FEV1). The primary outcome (de-
pendent) variables included mortality at day 30, postop-
erative respiratory failure (defined as requiring either
mechanical or non-invasive ventilation 48 hours post-
operatively), unplanned intubation, and a cardiac com-
posite score. A hypothesis generating exploratory analy-
sis was also undertaken comparing those allocated to the
ICU by “expert opinion” versus those not allocated to the
ICU and stratified by RFRI categories to assess the sec-
ondary outcomes of hospital length of stay, hypotension,
hypertension, and delirium. Postoperative respiratory
failure was defined according to the original validation
of the PRFI, as the requirement for mechanical ventila-
tion or non-invasive ventilation for more than 48 hours
after surgery [6]. A cardiac composite score was created
which was calculated as the sum of the presence of each
of myocardial infarction, pulmonary oedema, ventricular
fibrillation, primary cardiac arrest, and complete heart
block. Myocardial infarction was defined as a rise in se-
rum troponin above the upper limit of normal with or
without associated ECG changes (Q-waves) and docu-
mented as such on the medical record. Pulmonary oe-
dema was defined as respiratory distress with evidence
of fluid accumulation in the lungs by clinical exam, chest
X-ray, or invasive monitoring. Hypotension and hyper-
tension was defined as having a systolic blood pressure
≤90 mmHg or ≥160 mmHg and requiring intervention or
deemed clinically significant by the attending physician.
Delirium was defined as a confusional state marked by a
prominent disorder of perception, terrifying hallucina-
tions and vivid dreams, a kaleidoscopic array of strange
and absurd fantasies and delusions, inability to sleep,
tendency to convulse, and intense emotional disturbance
[13]. The Logistic Organ Dysfunction (LOD) score was
calculated as a measure of illness severity in those pa-
tients who were admitted into the ICU postoperatively
[13,14]. All outcome measures were ascertained both
prospectively and by a retrospective chart review.

D. J. Kutsogiannis et al. / Open Journal of Nursing 2 (2012) 289-300 291
3. STATISTICAL ANALYSIS
Frequency distributions, means, and medians were de-
termined for all variables and postoperative complica-
tions. Significant differences in the distribution of vari-
ables existed if the two-tailed P value was <0.05 as de-
termined by a chi-square test, or Fisher’s exact test for
categorical variables, and the t-test, Wilcoxon rank sum
test, or test for the equivalence of medians for continuous
variables. The RCRI, PRFI, BMI, and FEV1 have been
previously found to predict postoperative cardiac and
pulmonary complications, and these in addition to “ex-
pert opinion”, were included as independent variables in
a stepwise logistic regression analysis using the follow-
ing dependent variables: 1) mortality 30 days postopera-
tively; 2) respiratory failure 48 hours postoperatively; 3)
unplanned intubation; 4) cardiac composite score or res-
piratory failure; and 5) cardiac composite score or mor-
tality 30 days postoperatively. A significance level of
≤0.05 was selected for variable retention in the final mo-
del after backwards selection. The analysis was perfor-
med in S-Plus version 6.01 and STATA version 10.
4. RESULTS
4.1. Patient Demographics and Disposition
A total of 403 patients were included in the analysis. The
majority of patients were in the age group of 70 to 79
years, male, and undergoing a general surgical procedure.
Fourty-nine patients (12.1%) and 18 patients (4.5%) had
a PRFI score of 28 - 40 (PRFI-IV) and >40 (PRFI-V).
Eighty patients (19.9%) and 26 patients (6.5%) were in
categories RCRI-III and RCRI-IV respectively (Table 1).
Sixteen patients (4.0%) returned to the operating theatre
for a repeat operation and 13 of these operations were
related to a complication from the initial surgery. A total
of 46 patients (11.4%) were allocated an ICU bed preop-
eratively. Of these, 40 patients went to the ICU post-
operatively and 6 patients were deemed stable enough to
be transferred directly the high intensity unit. Twelve
additional patients (3.0%) were not allocated an ICU bed
preoperatively but were transferred to an ICU bed from
the operating theatre or surgical ward. Eighty-six patients
(21.3%) had a high intensity bed allocated preoperatively
where they were cared for postoperatively. An additional
21 (5.2%) patients did not have a high intensity bed al-
located preoperatively, but were transferred to the high
intensity unit from the operating theatre or surgical ward.
Ultimately 52 (12.9%), 113 (28.0%), and 238 (59.1%)
patients required ICU, high intensity and surgical ward
care as their highest level of postoperative care [Appen-
dix 2]. Five of 403 patients (1.2%) were lost to follow-up
after hospital discharge and were not included in the
30-day mortality multivariable analysis.
Table 1. Characteristics of the study population.
Age N = 403
<49 49 (12.2)
50 to 59 84 (20.8)
60 to 69 94 (23.3)
70 to 79 131 (32.5)
80 to 89 44 (10.9)
>
90 1 (0.3)
Gender
Male 226 (56.1)
Female 177 (43.9)
Type of surgery
General surgery 175 (43.4)
Vascular surgery 122 (30.3)
Orthopaedic surgery 56 (13.9)
Thoracic surgery 39 (9.7)
Neurosurgery 11 (2.7)
Preoperative Respiratory
Failure Index (PRFI)
<20 207 (51.4)
20 to 27 125 (31.0)
28 to 40 53 (13.2)
>40 18 (4.5)
Revised Cardiac Risk
Index (RCRI)
Class I 98 (24.3)
Class II 199 (49.4)
Class III 80 (19.9)
Class IV 26 (6.5)
Logistic organ
dysfunction score
(N = 40 ICU patients)
<
1 12 (24.0)
2 to 4 22 (44.0)
5 to 7 11 (22.0)
8 to 10 5 (10.0)
Body mass index, kg/m2 <30 236 (58.6)
30 to 40 119 (29.5)
41 to 50 32 (7.9)
>50 16 (4.0)
FEV-1 (L), mean (SD) 2.19 (0.75)

D. J. Kutsogiannis et al. / Open Journal of Nursing 2 (2012) 289-300 292
4.2. Mortality, Respiratory and Cardiac
Outcomes
Five patients (1.2%) died within 30 days of their opera-
tion, 12 (3.0%) met the criteria for postoperative respira-
tory failure at 48-hrs, 16 (4.0%) required an unplanned
intubation at any time postoperatively and 27 (6.7%)
suffered at least one major cardiac event post-operatively.
Nine of the 16 (56.3%) who required an unplanned intu-
bation were allocated to and received an ICU bed imme-
diately postoperatively whereas the remaining 7 (43.7%)
received high intensity or surgical ward care immediately
postoperatively. An increasing PRFI was associated with
an increasing RCRI (P = 0.0001). There was a significant
association between the PRFI and 30-day mortality (P =
0.04), respiratory failure (P < 0.0001), and unplanned
intubation (P = 0.005). Both the incidence of cardiac
events as measured by the cardiac composite score (P <
0.0001) and the incidence of hypertension (P < 0.0001)
increased with each increasing level of PRFI. The me-
dian [mean] (4.0 [6.7] versus 9.2 [16.4] days) hospital
length of stay increased significantly between the PRFI-
I/II group (score 0 to 20) and the PRFI-V group (score >
40) (P < 0.001). There was no association between RCRI
score and hospital mortality (P = 0.43), 30-day mortality
(P = 0.43), or the development of respiratory failure (P =
0.21). However, there was an association between re-
quiring an unplanned intubation and increasing RCRI
score (P = 0.03). As expected, a significant association
did exist between an increasing RCRI score and both the
cardiac composite outcome (P = 0.0001) and postopera-
tive hypertension (P = 0.05) (Table 2). The 5 patients
who died within 30 days of their operation had a signifi-
cantly higher mean PRFI (35.4 versus 18.3, P = 0.001),
and were significantly more likely to have undergone an
unplanned intubation (40% versus 3.5%, P = 0.003) than
surviving patients. The etiology of death in these five
patients included myocardial infarction, peritonitis, and
multisystem organ failure and the preoperative allocation
of an ICU bed by “expert opinion” was not associated
with 30-day mortality (P = 0.19).
“Expert opinion”, RCRI, PRFI, BMI, and FEV1 were
included in logistic regression models predicting 30-day
mortality, development of respiratory failure, unplanned
intubation, mortality or cardiac composite score  1, and
respiratory failure or cardiac composite score ≥ 1. Only
the PRFI score (OR 1.11, 95% CI 1.04 - 1.19) remained
significantly predictive of 30-day mortality in the multi-
variable model. However only “expert opinion” (OR
28.70, 95% CI 7.44 - 110.70) independently predicted
postoperative respiratory failure. Revised cardiac risk in-
dex (OR 2.00, 95% CI 1.12 - 3.57), PRFI (OR 1.06, 95%
CI 1.01 - 1.11) and BMI (OR 1.07, 95% CI 1.02 - 1.12)
were independent predictors of an unplanned intubation
in the multivariable model. Both the PRFI and “expert
opinion” remained as independent predictors of the com-
bined outcomes, cardiac composite score of 1 or 30-day
mortality, and cardiac composite outcome of 1 or respi-
ratory failure at 48 hours (Table 3). Collinearity between
Table 2. (a) Outcomes by preoperative respiratory failure index (PRFI); (b) Outcomes by revised cardiac risk index (RCRI).
(a)
Outcome* PRFI-I & II (n = 207) PRFI-III (n = 125) PRFI-IV (n = 53) PRFI-V (n = 18) P-value
In-hospital mortality 0 (0.0%) 2 (1.6%) 2 (3.8%) 1 (5.6%) 0.04
30-day mortality 0 (0.0%) 2 (1.6%) 2 (3.8%) 1 (5.6%) 0.04
48-hr respiratory failure 2 (1.0%) 1 (0.8%) 6 (11.3%) 3 (16.7%) <0.0001
Unplanned intubations 3 (1.5%) 6 (4.8%) 4 (7.5%) 3 (16.7%) 0.005
Cardiac composite score  1 4 (1.9%) 8 (6.4%) 9 (17.0%) 6 (33.3%) <0.0001
Hospital length of stay,
days median (range) 4.0 (1.6 - 6.8) 5.7 (3.7 - 8.6) 7.8 (6.7 - 10.8) 9.2 (6.8 - 13.7) <0.001
(b)
Outcome* RCRI-I (n = 98) RCRI-II (n = 199) RCRI-III (n = 80) RCRI-IV (n = 26) P-value
In-hospital mortality 0 (0%) 3 (1.5%) 1 (1.3%) 1 (3.8%) 0.43
48-hr respiratory failure 1 (1.0%) 5 (2.5%) 5 (6.3%) 1 (3.8%) 0.21
Unplanned intubations 1 (1.0%) 6 (3.0%) 6 (7.5%) 3 (11.5%) 0.03
Cardiac composite score  1 2 (2.0%) 12 (6.0%) 7 (8.8%) 6 (23.1%) 0.0001
Hospital length of stay,
days median (range) 3.7 (1.8 - 6.8) 5.7 (3.6 - 8.7) 6.6 (3.1 - 8.7) 7.7 (4.6 - 13.7) 0.006

D. J. Kutsogiannis et al. / Open Journal of Nursing 2 (2012) 289-300 293
Table 3. Logistic regression models predicting 30-day mortality, respiratory failure (48-hr), unplanned intubation, and cardiac events/
30-day mortality or respiratory failure (48-hr).
Outcome Variables Crude Odds Ratio (95% CI)Adjusted Odds Ratio (95% CI)
Mortality 30 days postoperatively
PFRI Score 1.11 (1.04 - 1.19) 1.11 (1.04 - 1.19)
Respiratory failure 48-hours postoperatively
Expert opinion28.70 (7.44 - 110.70) 28.70 (7.44 - 110.70)
Unplanned intubation
RCRI Score 2.47 (1.49 - 4.10) 2.00 (1.12 - 3.57)
PFRI Score 1.07 (1.02 - 1.11) 1.06 (1.01 - 1.11)
BMI, kg/m2 1.05 (1.01 - 1.10) 1.07 (1.02 - 1.12)
Cardiac composite outcome score ≥ 1 or mortality
30 days postoperatively
PFRI Score 1.09 (1.06 - 1.13) 1.08 (1.04 - 1.12)
Expert opinion7.06 (3.11 - 16.01) 2.66 (1.03 - 6.82)
Cardiac composite outcome score ≥ 1 or respiratory
failure 48-hour postoperatively
PFRI Score 1.09 (1.05 - 1.12) 1.07 (1.03 - 1.10)
Expert opinion7.78 (3.57 - 16.97) 3.30 (1.35 - 8.07)
the PRFI and “expert opinion” was investigated by ob-
serving for variance inflation or significant point esti-
mate deviation when both parameters were fit into the
logistic model as compared to the models containing the
individual parameters. Combining the two variables did
reduce the magnitude but not the direction of the inde-
pendent point estimates of “expert opinion” but not of
PRFI. However no inflation of variance was noted about
the point estimates of the models containing “expert
opinion”. This indicates that there is some overlap be-
tween PRFI and the variables used within the “expert
opinion”. Interaction terms were not fit in the models
given the low mortality.
4.3. Influence of ICU Bed Allocation
Two-hundred and seven patients had a PRFI of less than
20 (51.3% PRFI-I and PRFI-II), 125 had a PRFI of 20 -
27 (31.0% PRFI-III), 53 patients had a PRFI of 28 - 40
(13.2% PRFI-IV) and 18 patients had a PRFI of >40
(4.5% PRFI-V). Within PRFI-III to V strata, an ICU bed
was not allocated nor was the patient cared for in the
ICU immediately postoperatively in 113 (90%), 27 (51%),
and 9 (50%) of patients respectively. Within all PRFI
groups, mean RCRI, BMI and FEV1 did not differ sig-
nificantly between the ICU and non-ICU groups. How-
ever, within PRFI-IV, mean PRFI was significantly high-
er in those patients who were assigned to and received
and ICU bed by “expert opinion” versus those patients
who were not assigned to and did not receive an ICU bed
(P < 0.0001). In those patients within the PRFI-III strata,
a longer mean hospital length of stay (P = 0.002) and a
higher incidence of hypotension (P = 0.008) was present
in those patients who were allocated to and received an
ICU bed by “expert opinion” versus those patients who
were not allocated to and did not receive an ICU bed.
Within the PRFI-IV strata, there was a longer mean hos-
pital length of stay (P = 0.01) and a higher incidence of
respiratory failure (P = 0.004), hypertension (P = 0.005),
and delirium (P = 0.03) in those patients who were allo-
cated to and received an ICU bed versus those patients
who were not allocated to and did not receive an ICU
bed. Within the PRFI-V strata, there was a longer mean
hospital length of stay (P = 0.02) and a higher mean car-
diac composite score (P = 0.02) in those patients who
were allocated to and received an ICU bed versus those
patients who were not allocated to and did not receive an
ICU bed (Table 4).
5. DISCUSSION
Our study of patients undergoing elective noncardiac
surgery demonstrated that the PRFI was the only inde-
pendent predictor of 30-day mortality. However, only the
“expert opinion” of preoperative allocation of a patient to
the ICU independently predicted the requirement for
mechanical ventilation at 48-hours. For combined out-
comes of postoperative cardiac events or mortality and

D. J. Kutsogiannis et al. / Open Journal of Nursing 2 (2012) 289-300 294
Table 4. Characteristics of discrepant patients by Preoperative Respiratory Failure Index category*.
Allocated to and receiving ICU
(Expert Opinion)
Not allocated to and not receiving ICU
(Expert opinion) P-Value†
Preoperative Respiratory Failure Index
20 - 27 (PRFI-III)
n = 7 n = 113
PRFI, mean 23.28 22.39 0.35
Revised Cardiac Risk Index, mean 1.571 1.168 0.15
BMI, kg/m2 34.10 29.30 0.20
FEV1, L 1.96 2.17 0.26
Cardiac composite score, mean 0.1428 0.0973 0.81
Cardiac composite score ≥ 1 1/7 6/113 0.88
Mean hospital length of stay, days 14.85 (median 11.7) 12.66 (median 4.75) 0.002
30-Day mortality 0/7 2/113 0.89
Ventilated at 48-hour 0/7 1/113 0.94
Unplanned intubation 0/7 5/113 0.74
Hypotension 4/7 13/113 0.008
Hypertension 0/7 8/113 0.61
Delerium 0/7 6/113 0.69
Preoperative Respiratory Failure Index
28 - 40 (PRFI-IV) n = 20 n = 27
PRFI, mean 35.65 31.55 0.0001
Revised cardiac risk index, mean 1.7 1.407 0.20
BMI, kg/m2 28.878 28.049 0.60
FEV1, L 1.804 1.937 0.36
Cardiac composite score, mean 0.20 0.222 0.88
Cardiac composite score ≥ 1 4/20 4/27 0.94
Mean (median) hospital length of stay, days 17.19 (9.75) 8.93 (6.77) 0.01
30-Day mortality 1/20 1/27 0.68
Ventilated at 48-hour 6/20 0/27 0.004
Unplanned intubation 3/20 1/27 0.30
Hypotension 7/20 5/27 0.31
Hypertension 11/20 4/27 0.005
Delerium 7/20 2/27 0.03
Preoperative Respiratory Failure Index > 40
(PRFI-V) n = 8 n = 9
PRFI, mean 46.0 47.44 0.52
Revised Cardiac Risk Index, mean 2.375 1.555 0.14
BMI, kg/m2 29.69 27.55 0.92
FEV1, L 2.272 2.415 0.75
Cardiac composite score, mean 0.875 0.111 0.02
Cardiac composite score ≥ 1 5/8 1/9 0.09
Mean hospital length of stay, days 26.45, SD 19.63 (25.26) 8.398, SD 2.641 (7.72) 0.02
30-Day mortality 1/8 0/9 0.47
Ventilated at 48-hour 3/8 0/9 0.08
Unplanned intubation 3/8 0/9 0.08
Hypotension 3/8 2/9 0.62
Hypertension 3/8 3/9 1.00
Delerium 2/8 2/9 1.00
*Two hundred and seven patients with a PRFI-I were not included in this analysis. Analysis was based on 120, 47, and 17 patients in PRFI-III, PRFI-IV, and
RFI-V with available data. †Fisher’s exact test used for categorical comparisons with fewer than 5 observations. P

D. J. Kutsogiannis et al. / Open Journal of Nursing 2 (2012) 289-300 295
postoperative cardiac events or the requirement for me-
chanical ventilation at 48-hour, both PRFI and “expert
opinion”, but not RCRI, were independent predictors.
Collectively, 50% of patients in PRFI-IV and PRFI-V
were allocated to and received an ICU bed postopera-
tively, demonstrating either the heterogeneity in decision
making among the multidisciplinary group of physicians
caring for these patients or that the PRFI omits variables
which may be relevant to allocating an ICU bed. Collec-
tively, unmeasured comorbidities, unique surgical risks,
vacancy and staffing of ICU beds, and/or random deci-
sion making may have been influencing the “expert
opinion” to allocate an ICU bed. Moreover, those pa-
tients with an intermediate or high level of PRFI risk had
a significant increase in the risk of respiratory failure,
hypotension, hypertension, delirium, and a longer length
of hospital stay if allocated to and receiving an ICU bed
postoperatively as compared to similar strata of patients
not allocated to an ICU bed. Although the design of this
study prohibits the inference of a causal link between
ICU bed allocation and increased complications, further
development of this hypothesis is warranted within fu-
ture studies. For those with intermediate PRFI risk, the
decision to allocate an ICU bed was likely made on the
assumption that monitoring a patient in the ICU would
improve her outcome as opposed to an assumption that a
high likelihood of ICU dependent intervention would be
required. Such conservative allocation to an ICU may
increase the utilization of scarce ICU and hospital re-
sources, and may have delayed the allocation of patients
at very high operative risk or prevented the admission of
emergent patients requiring ICU care. The lengthened
hospital stay was likely attributable to the a-priori “ex-
pert opinion” that mechanical ventilation was necessary
for the majority of patients assigned to postoperative
ICU care, and that pharmacological interventions were
required to maintain a patient sedated to facilitate me-
chanical ventilation. Delayed discharges from the ICU
because of a lack of beds within the high intensity unit or
surgical wards may have also contributed to the length-
ened hospital stay. Indeed the ICU co-intervention of
mechanical ventilation may, in part, be a self-fulfilling
prophecy resultant from preoperative ICU bed allocation.
The robust association between the PRFI in predicting
mortality or respiratory failure in this study is consistent
with individual variables comprising the PRFI such as
older age, albumin level, renal dysfunction, and thoracic
surgery diagnosis individually having demonstrated pre-
dictive value in other prospective studies [15-18]. Also
consistent with this study is a recent systematic review
and cohort study which demonstrated that the RCRI was
poor to moderate in discriminating postoperative cardiac
events in populations of vascular noncardiac and mixed
noncardiac surgical patients [19,20].
Heterogeneous decision making in allocating ICU
beds in noncardiac surgery patients has been described
elsewhere. In a retrospective study of 241 bariatric pa-
tients requiring either ICU or an intermediate care unit,
half of these patients were placed into these units in an
anticipatory manner with the other half admitted as un-
expected emergencies. Upon review of these cases, the
authors were unable to discern why individual patients
were preemptively placed in the ICU or an intermediate
care unit [17]. Other authors have demonstrated a re-
duced mortality with no worsening morbidities using a
strategy of up to 24-hour management in the operative
recovery room of preselected patients after elective ab-
dominal aortic surgery [21-23]. Moreover, a propensity
case matched retrospective review of 104 pairs of elec-
tive neurosurgical patients with American Society of
Anaesthesiologists Status I or II failed to demonstrate a
difference in Glasgow Outcome Score, mortality, or
complication rates between those assigned to ICU care
versus neurosurgical ward care postoperatively [24].
The increased utilization of high intensity units which
do not provide inotropic/vasopressor support or mecha-
nical ventilation should be considered as an option to
ICU care. A decrease in the rate of cancellation of major
elective operations with no more than one surgical can-
cellation for any given patient following the opening of a
new high intensity unit has been previously demonstrated
[10]. As well, the utilization of post-operative non-inva-
sive continuous positive airway pressure, which may be
available within high intensity units such as ours, has
been demonstrated to reduce the incidence of respiratory
failure, pneumonia, infection, and sepsis in selected pa-
tients [25]. Conversely, a more rational use of postopera-
tive interventions requiring ICU admission such as in-
travenous beta-blockade, pulmonary artery catheter mo-
nitoring, and goal directed hemodynamic therapy, which
have been found to be of limited benefit in subgroups of
patients, may avert the requirement for postoperative
ICU admission in such patients [26-29]. A two stage ap-
proach to risk stratification including preoperative, op-
erative and immediate post-operative parameters may be
considered to improve ICU allocation decisions. Previ-
ous studies have demonstrated that one third of postop-
erative complications and one fourth of deaths occur
within the first 48 hours after surgery [30]. These early
postoperative complications such as hypotension, hyper-
tension, tachycardia, hypoxemia, hypercapnea, a decrea-
sed level of consciousness and operations outside of nor-
mal work hours predict unplanned ICU admissions and
may add predictive value to preoperative risk factors in
the allocation of ICU beds [31-33].
Although patients were analyzed within their respec-
tive risk strata in order to limit bias, residual confound-
ing, other unidentified comorbidities, and operative risk

D. J. Kutsogiannis et al. / Open Journal of Nursing 2 (2012) 289-300 296
factors may have accounted for the observed differences
in outcomes between groups. Many clinical, biological
and operative variables in these patients are complex and
it is difficult to achieve a high level of discrimination
with current predictive scoring systems in these multi-
factorial processes [34]. The cohort was limited to one
university affiliated tertiary care center with a small
number of deaths and respiratory failure events, which
may limit the generalizability of the study. Moreover the
small number of events limited any inference of potential
interactions between the predictor variables. There may
also have been ascertainment bias in measuring cardiac
outcomes, favoring the assignment of more events to the
closely observed ICU group. Also, the possibility of ICU
co-intervention bias resulting in harm to those patients
cared for in the ICU cannot be excluded and requires
further exploration. Although the CAM-ICU has recently
become a standard for measuring delirium within the
ICU, it was not in universal use at the time this study was
initiated [35,36]. Consequently, our use of a neurology-
based definition of delirium may have also added to as-
certainment bias.
There remains a requirement for well designed clinical
trials incorporating preoperative, operative and postop-
erative predictors to assess the effectiveness of ICU care
in subgroups of high risk noncardiac surgical patients.
Improved standardized protocols for fair allocation of
postoperative ICU resources should be developed, con-
cordant with the American Thoracic Society’s recom-
mendations [37,38]. An ethical framework of allocation
of ICU resources using clinical judgement including a
closed system that offers reciprocity, attention to general
concerns of justice, respect for individual variations, ex-
plicitness, and a review of the decision making process
should be advocated by the multidisciplinary group of
physicians caring for these patients [39]. Consequently,
decision support algorithms utilizing this ethical frame-
work should be explored.
6. CONCLUSION
The PRFI and “expert opinion” are robust predictors of
mortality and respiratory failure, respectively, in non-
cardiac surgical patients. Longer lengths of hospital stay
and complication rates in patients of intermediate PRFI
risk assigned to an ICU versus a high intensity unit or
ward should prompt further scientific investigation in the
form of clinical trials to aid in the decision making
process of which patients to safely allocate to high inten-
sity units versus the ICU postoperatively.
7. COMPETING INTERESTS
The authors declare that they have no competing interests in the con-
duct, analysis and interpretation of the study results.
8. AUTHORS CONTRIBUTION
DK contributed the conception and design of the study, the acquisition
of the data, the data analysis and interpretation of the data, and in
drafting and revising the manuscript. SN contributed to the interpreta-
tion of the data and in revising the manuscript for important intellectual
content. BL contributed to the primary data analysis, the interpretation
of the data analysis, and in revising the manuscript. All three authors
have given final approval of the final version to be published.
REFERENCES
[1] Angus, D.C., Kelley, M.A., Schmitz, R.J., White, A., Po-
povich Jr., J., et al. (2000) Current and projected work-
force requirements for care of the critically ill and pa-
tients with pulmonary disease: Can we meet the require-
ments of an aging population? The Journal of the Ameri-
can Medical Association, 284, 2762-2770.
doi:10.1001/jama.284.21.2762
[2] Milbrandt, E.B., Kersten, A., Rahim, M.T., Dremsizov,
T.T., Clermont, G., Cooper, L.M., et al. (2008) Growth of
intensive care unit resource use and its estimated cost in
Medicare. Critical Care Medicine, 36, 2504-2510.
doi:10.1097/CCM.0b013e318183ef84
[3] Needham, D.M., Bronskill, S.E., Sibbald, W.J., Prono-
vost, P.J. and Laupacis, A. (2004) Mechanical ventilation
in Ontario, 1992-2000: Incidence, survival, and hospital
bed utilization of noncardiac surgery adult patients.
Critical Care Medicine, 32, 1504-1509.
doi:10.1097/01.CCM.0000129972.31533.37
[4] Goldman, L., Caldera, D.L., Nussbaum, S.R., Southwick,
F.S., Krogstad, D., Murray, B., et al. (1977) Multifacto-
rial index of cardiac risk in noncardiac surgical proce-
dures. The New England Journal of Medicine, 29 7, 845-
850. doi:10.1056/NEJM197710202971601
[5] Arozullah, A.M., Khuri, S.F., Henderson, W.G. and Daley,
J. (2001) Development and validation of a multifactorial
risk index for predicting postoperative pneumonia after
major noncardiac surgery. Annals of Internal Medicine,
135, 847-857.
[6] Arozullah, A.M., Daley, J., Henderson, W.G. and Khuri,
S.F. (2000) Multifactorial risk index for predicting post-
operative respiratory failure in men after major noncar-
diac surgery. Annals of Surgery, 232, 242-253.
doi:10.1097/00000658-200008000-00015
[7] Lee, T.H., Marcantonio, E.R., Mangione, C.M., Thomas,
E.J., Polanczyk, C.A., Cook, E.F., et al. (1999) Deriva-
tion and prospective validation of a simple index for pre-
diction of cardiac risk of major noncardiac surgery. Cir-
culation, 100, 1043-1049.
doi:10.1161/01.CIR.100.10.1043
[8] Khuri, S.F., Henderson, W.G., Daley, J., Jonasson, O.,
Jones, R.S., Campbell Jr., D.A., et al. (2007) The patient
safety in surgery study: Background, study design, and
patient populations. Journal of the American College of
Surgeons, 204, 1089-1102.
doi:10.1016/j.jamcollsurg.2007.03.028
[9] Agency for Healthcare Research and Quality (2007) Na-
tional healthcare quality report. Institute US Department

D. J. Kutsogiannis et al. / Open Journal of Nursing 2 (2012) 289-300 297
of Health and Human Services, Agency for Healthcare
Research and Quality, Rockville. AHRQ Pub. No. 08-
0040
[10] Dhond, G., Ridley, S. and Palmer, M. (1998) The impact
of a high-dependency unit on the workload of an inten-
sive care unit. Anaesthesia, 53, 841-847.
doi:10.1046/j.1365-2044.1998.00522.x
[11] Campeau, L. (19 76) Letter: Grading of angina pectoris. Cir-
culation, 54, 522-523.
[12] Hurst, J.W., Morris, D.C. and Alexander, R.W. (1999)
The use of the New York Heart Association’s classifica-
tion of cardiovascular disease as part of the patient’s com-
plete Problem List. Clinical Cardiology, 22, 385-390.
doi:10.1002/clc.4960220604
[13] Adams, R.D. and Victor, M. (1989) Principles of neurol-
ogy. 4th Edition, McGraw-Hill Inc., New York, 1989.
[14] Le Gall, J.R., Klar, J., Lemeshow, S., Saulnier, F., Alberti,
C., Artigas, A., et al. (1996) The logistic organ dysfunc-
tion system. A new way to assess organ dysfunction in
the intensive care unit. The Journal of the American Me-
dical Association, 276, 802-810.
doi:10.1001/jama.276.10.802
[15] McNicol, L., Story, D.A., Leslie, K., Myles, P.S., Fink,
M., Shelton, A.C., et al. (2007) Postoperative complica-
tions and mortality in older patients having non-cardiac
surgery at three Melbourne teaching hospitals. Medical
Journal of Australia, 186, 447-452.
[16] Manku, K., Bacchetti, P. and Leung, J.M. (2003) Progno-
stic significance of postoperative in-hospital complica-
tions in elderly patients. I. Long-term survival. Anesthe-
sia & Analgesia, 96, 583-589.
[17] Cendan, J.C., Abu-aouf, D., Gabrielli, A., Caruso, L.J.,
Rout, W.R., Hocking, M.P., et al. (2005) Utilization of
intensive care resources in bariatric surgery. Obesity Sur-
gery, 15, 1247-1251. doi:10.1381/096089205774512681
[18] Helling, T.S., Willoughby, T.L., Maxfield, D.M. and Ryan,
P. (2004) Determinants of the need for intensive care and
prolonged mechanical ventilation in patients undergoing
bariatric surgery. Obesity Surgery, 14, 1036-1041.
doi:10.1381/0960892041975488
[19] Bertges, D.J., Goodney, P.P., Zhao, Y., Schanzer, A., No-
lan, B.W., Likosky, D.S., et al. (2000) The Vascular Study
Group of New England Cardiac Risk Index (VSG-CRI)
predicts cardiac complications more accurately than the
Revised Cardiac Risk Index in vascular surgery patients.
Journal of Vascular Surgery, 52, 674-683.
[20] Ford, M.K., Beattie, W.S. and Wijeysundera, D.N. (2012)
Systematic review: Prediction of perioperative cardiac
complications and mortality by the revised cardiac risk
index. Annals of Internal Medicine, 152, 26-35.
[21] Callaghan, C.J., Lynch, A.G., Amin, I., Fazel, M., Lindop,
M.J., Gaunt, M.E., et al. (2005) Overnight intensive re-
covery: Elective open aortic surgery without a routine
ICU bed. European Journal of Vascular and Endovascu-
lar Surgery, 30, 252-258. doi:10.1016/j.ejvs.2005.03.004
[22] Ryan, D. and McGreal, G. (2010) Why routine intensive
care unit admission after elective open infrarenal Abdo-
minal Aortic Aneurysm repair is no longer an evidence
based practice. Surgeon, 8, 297-302.
doi:10.1016/j.surge.2010.05.003
[23] Podore, P.C. and Throop, E.B. (1999) Infrarenal aortic
surgery with a 3-day hospital stay: A report on success
with a clinical pathway. Journal of Vascular Surgery, 29,
787-792. doi:10.1016/S0741-5214(99)70204-1
[24] Terada, Y., Inoue, S., Tanaka, Y., Kawaguchi, M., Hirai,
K. and Furuya, H. (2010) The impact of postoperative in-
tensive care on outcomes in elective neurosurgical pa-
tients in good physical condition: A single centre propen-
sity case-matched study. Canadian Journal of Anesthesia,
57, 1089-1094. doi:10.1007/s12630-010-9393-2
[25] Squadrone, V., Coha, M., Cerutti, E., Schellino, M.M., Bio-
lino, P., Occella, P., et al. (2005) Continuous positive
airway pressure for treatment of postoperative hypoxemia:
A randomized controlled trial. The Journal of the Ameri-
can Medical Association, 293, 589-595.
doi:10.1001/jama.293.5.589
[26] Devereaux, P.J., Yang, H., Yusuf, S., Guyatt, G., Leslie,
K., Villar, J.C., et al. (2008) Effects of extended-release
metoprolol succinate in patients undergoing non-cardiac
surgery (POISE trial): A randomised controlled trial. Lan-
cet, 371, 1839-1847.
doi:10.1016/S0140-6736(08)60601-7
[27] Lees, N., Hamilton, M. and Rhodes, A. (2009) Clinical
review: Goal-directed therapy in high risk surgical pa-
tients. Critical Care, 13, 231. doi:10.1186/cc8039
[28] Sandham, J.D., Hull, R.D., Brant, R.F., Knox, L., Pineo,
G.F., Doig, C.J., et al. (2003) A randomized, controlled
trial of the use of pulmonary-artery catheters in high-risk
surgical patients. The New England Journal of Medicine,
348, 5-14. doi:10.1056/NEJMoa021108
[29] Chopra, V. and Eagle, K.A. (2010) Perioperative beta-bloc-
kers for cardiac risk reduction: Time for clarity. JAMA,
303, 551-552. doi:10.1001/jama.2010.68
[30] Cavaliere, F., Conti, G., Costa, R., Masieri, S., Antonelli,
M. and Proietti, R. (2008) Intensive care after elective
surgery: A survey on 30-day postoperative mortality and
morbidity. Minerva Anestesiologica, 74, 459-468.
[31] Rose, D.K., Cohen, M.M. and DeBoer, D.P. (1996) Car-
diovascular events in the postanesthesia care unit: Con-
tribution of risk factors. Anesthesiology, 84, 772-781.
doi:10.1097/00000542-199604000-00003
[32] Lee, A., Lum, M.E., O’Regan, W.J. and Hillman, K.M.
(1998) Early postoperative emergencies requiring an in-
tensive care team intervention. The role of ASA physical
status and after-hours surgery. Anaesthesia, 53, 529-535.
doi:10.1046/j.1365-2044.1998.00395.x
[33] Rosenberg, A.L. and Watts, C. (2000) Patients readmitted
to ICUs*: A systematic review of risk factors and out-
comes. Chest, 118, 492-502. doi:10.1378/chest.118.2.492
[34] Ware, J.H. (2006) The limitations of risk factors as prog-
nostic tools. The New England Journal of Medicine, 355,
2615-2617. doi:10.1056/NEJMp068249
[35] Ely, E.W., Inouye, S.K., Bernard, G.R., Gordon, S.,
Francis, J., May, L., et al. (2001) Delirium in mechani-
cally ventilated patients. The Journal of the American
Medical Association, 286, 2703-2710.
doi:10.1001/jama.286.21.2703

D. J. Kutsogiannis et al. / Open Journal of Nursing 2 (2012) 289-300 298
[36] Guenther, U., Popp, J., Koecher, L., Muders, T., Wrigge,
H., Ely, E.W., et al. (2010) Validity and reliability of the
CAM-ICU Flowsheet to diagnose delirium in surgical
ICU patients. Journal of Critical Care, 25, 144-151.
doi:10.1016/j.jcrc.2009.08.005
[37] American Thoracic Society (1997) Fair allocation of in-
tensive care unit resources. American Journal of Respi-
ratory and Critical Care Medicine, 156, 1282-1301.
[38] Kelley, M.A., Angus, D., Chalfin, D.B., Crandall, E.D.,
Ingbar, D., Johanson, W., et al. (2004) The critical care
crisis in the United States: A report from the profession.
Chest, 125, 1514-1517. doi:10.1378/chest.125.4.1514
[39] Hurst, S.A. and Danis, M. (2007) A framework for ratio-
ning by clinical judgment. Kennedy Institute of Ethics Jour-
nal, 17, 247-266. doi:10.1353/ken.2007.0021

D. J. Kutsogiannis et al. / Open Journal of Nursing 2 (2012) 289-300 299
APPENDIX 1. INCLUSION CRITERIA
1) All patients going for vascular, thoracic, pancreatic, or esophagectomy surgery, who are being seen by 1 Special-
ists (Bronchoscopy or carotid endarterectomies are excluded).
2) All patients going for orthopedic spinal or neurosurgery who are being seen by 1 Specialists and have any one of
the disease states listed below (Spinal includes discectomy, decompression, fusion, instrumentation).
3) All patients going for general surgery who are being seen by 1 Specialists and have any one of the disease states
listed, (two if hypertensive), or who are less than fully independent (see definitions below) (General surgery includes all
mastectomies, abdominal/laproscopic surgery [including umbilical, epigastric and inguinal hernia]).
Disease States
Myocardial Angina (Canadian Cardiovascular Society Classification)
Arrhythmia
Valvular heart disease
Myocardial Infarction
Congestive Heart Failure (New York Heart Association Classification)
Vascular Hypertension (must have another if undergoing general surgery)
Peripheral vascular disease
Cerebrovascular disease
Pulmonary Chronic obstructive pulmonary disease
Asthma
Interstitial lung disease Obstructive sleep apnea (with or without treatment)
Neurologic Dementia
Hemiplegia or paraplegia
Seizure disorder
Neuropathy
Myopathy
Any significant spinal disease [at the discretion of investigator]
RHEMATOLOGIC Any significant rhematlogic disease [at the discretion of investigator] —includes SLE, Rheumatoid
arthiritis but not Osteoarthritis
Gastrointestinal Peptic ulcer disease
Gastroesophageal reflux disease
Gastrointestinal bleeding
Inflammatory bowel disease
Mild, moderate or severe liver disease
Endocrine Diabetes, on insulin or oral hypoglycemics
Diabetes with complications High BMI (30)
Other significant endocrine disorders [at the discretion of investigator]
Renal Acute renal failure
Chronic renal failure
Cancer/Immune Any malignancy, including lymphoma and leukemia—excluded if surgical resection >5 years
Metastatic solid tumor
HIV-AIDS
Bleeding disorders and coagulpopathy
Partially Dependent: Requires equipment/devices + assistance from another person for some ADL. i.e.) patients from a nursing home, on kidney dialysis or
home ventilation support, yet maintains some independent function; Totally Dependent: Cannot perform any activities of daily living for him/her self; includes
patients who are totally dependent on nursing care, such as a dependent nursing home patient.

D. J. Kutsogiannis et al. / Open Journal of Nursing 2 (2012) 289-300 300
APPENDIX 2. PATIENT PREOPERATIVE ALLOCATION BY “EXPERT OPINION” AND
HIGHEST POSTOPERATIVE LEVEL OF CARE
ABBREVIATIONS (Includes: high-risk type of surgery, history of ische-
mic heart disease, congestive heart failure, cerebrovacu-
lar disease, preoperative treatment with insulin, preoper-
ative serum creatinine > 2.0 mg/dL);
BMI Body mass index;
FEV1 Forced expiratory volume in one
second; PRFI Preoperative respiratory failure index
LOD Logistic organ dysfunction score (Includes: type of surgery, emergency surgery, albumin
< 30 g/L, blood urea nitrogen > 30 mg/dL, partially or
fully dependent functional status, history of chronic ob-
structive pulmonary disease, age).
(Includes: Glasgow Coma Score, heart rate, systolic
blood pressure, urea, creatinine, urine output, PaO2/FIO2,
white blood cell count, platelet count, bilirubin, INR);
RCRI Revised cardiac risk index