Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Delirium/Depression
a r t i c l e
Keywords:
Delirium
C-reactive protein
Risk factors
Multivariate analysis
i n f o
a b s t r a c t
Background and Purpose: Delirium is thought to be associated with systemic inammatory response.
However, its association with the most widely used inammatory biomarker C-reactive protein (CRP) has not
been well established. We aimed to examine whether CRP on intensive care unit (ICU) entry was associated
with subsequent development of delirium.
Design and Setting: This prospective observational study was conducted in a mixed 24-bed ICU in a tertiary
teaching hospital.
Methods: All patients admitted to the ICU from February 2011 to June 2012 were screened for eligibility.
Demographic data and clinical characteristics of included patients were recorded. Patients were screened for
the presence of delirium by using the tool Confusion Assessment Method for the ICU (CAM-ICU). C-reactive
protein was obtained on ICU entry and 24 hours thereafter. Eligible patients were followed up for 28 days or
until death. Univariate and multivariate analyses were performed to evaluate independent risk factors for
delirium. Clinical outcome included the length of stay (LOS) in the ICU, 28-day mortality, and duration of
mechanical ventilation. Two-tailed P b .05 was considered statistically signicant.
Results: A total of 223 patients were included during study period. In univariate analysis, patients with
delirium showed signicantly higher CRP values than those without (120.5 vs 57.5 mg/L; P = .0001). By
adjusting for confounding variables (including age, sex, Acute Physiology and Chronic Health Evaluation II,
intubation, living alone, physical restraint, alcohol drinking, smoking, type of medical condition, and hospital
LOS before ICU admission) in logistic regression model, CRP remained an independent predictor of delirium
(odds ratio, 1.07; 95% condence interval, 1.01-1.15). As compared with nondelirious patients, those with
delirium showed longer LOS in ICU (13 vs 5 days; P b .001) and duration of mechanical ventilation (6 vs 1
days; P b .001). An increase in CRP greater than 8.1 mg/L within 24 hours was associated with 4-fold increase
in the risk of delirium (odds ratio: 4.47, 95% condence interval, 1.28-15.60).
Conclusion: C-reactive protein measured on ICU entry and its changes within 24 hours are risk indicators of
delirium. Further studies exploring the treatment of delirium according to CRP levels are warranted.
2014 Elsevier Inc. All rights reserved.
1. Introduction
Delirium is a frequently encountered clinical syndrome that is
characterized by altered mental status, inattention, disorganized
thinking, and uctuation of mental status. It is estimated that the
incidence of delirium in the intensive care unit (ICU) ranges between
20% and 80% across different study populations and institutions [1-5].
In analogy to acute kidney injury, delirium is now regarded as a
syndrome of acute brain dysfunction resulting from multiple risk
factors, and it imposes signicant negative impact on patient
outcomes [6,7]. Thus, the prevention and management of delirium
are of vital importance in the treatment of critically ill patients.
89
and RASS [21,22]. There are 4 features in the CAM-ICU tool: acute
onset or uctuating course, inattention, altered level of consciousness,
and disorganized thinking. Patients with delirium were categorized
into hyperactive (persistent positive RASS values: +1 to + 4 during
all assessments), hypoactive (persistently neutral or negative RASS
values: 0 to 3 during all assessments), and mixed types of delirium
(episode with both positive and negative RASS values) according to
RASS assessments obtained during delirium episodes [23]. Treatment
of delirium included nonpharmacologic and pharmacologic strategies.
The former included timely removal of catheter and physical
restraints, early mobilization, repeated reorientation of patients,
minimization of unnecessary noise/stimuli, and use of scheduled
pain protocol, and the latter included the use of haloperidol.
2.4. Statistical analysis
Quantitative data were expressed as mean SE or median and
interquartile range, as appropriate. Categorical data were expressed as
the number and percentage. Comparisons between delirious and
nondelirious groups were performed using t test or Wilcoxon rank
sum test, as appropriate. 2 Test was used for categorical data.
Conventional logistic regression model was tted to adjust for the
odds ratio (OR) of the development of delirium. All variables that
might interact with delirium and CRP were added to the model. These
variables included age, sex, APACHE II score, tracheal intubation,
physical restraints, living alone, history of drinking and smoking, and
length of hospital stay before ICU admission. The nal model was
assessed for its validity in aspects of interaction, collinearity, and
goodness of t [24]. Interaction term was created for CRP and age
because the impact of CRP on the development of delirium may vary
across different age groups. Collinearity was assessed using variance
ination factor, which is an indicator of how much of the ination of
the SE could be caused by collinearity. A variance ination factor
greater than 10 indicates signicant collinearity. Model t was
assessed by using Hosmer-Lemeshow goodness-of-t test [25]. The
CRP was assessed for its linkage with the development of delirium
by using CRP less than or equal to 9.8 mg/L as the reference group,
and the effect was adjusted for all potential confounders in logistic
regression model, as previously described. Change in CRP values was
classied into quartiles, and patients in the lowest quartile were used
as the reference group for the comparison with groups of other
quartiles. Diagnostic performance of CRP in predicting the development of delirium was assessed using the receiver operating
characteristic curve. All statistical analyses were performed using
the software Stata 11.0 (College Station, TX). Two-tailed P b .05 was
considered statistically signicant.
A total of 223 consecutive patients were eligible during the 16month study period. The overall prevalence of delirium in our cohort
was 24.2% (54/223). Characteristics of included patients are shown in
Table 1. The mean age of the patients was 57.2 years, and 141 (63.2%)
of them were men. The median APACHE II score was 13 with an
interquartile range of 9 to 17. There were 93 medical patients (41.7%),
74 surgical patients (33.2%), 30 cardiac surgery patients (13.5%), and
26 neurosurgical patients (11.7%). There were 120 patients (54.1%)
with tracheal intubation. Thirty-ve patients (15.7%) lived alone at
home before this ICU admission. One hundred six patients (47.8%) had
physical restraint before the development of delirium. Fifty-eight
patients (26.1%) had a history of heavy alcohol drinking, and 58
(26.1%) reported a history of cigarette smoking. C-reactive protein
levels were not normally distributed (P b .01 for skewness; P b .001 for
Kurtosis), and they were expressed as median and interquartile range
(67.7; 21.5-128 mg/L).
90
Table 1
Characteristics of included patients
Parameters
Age (y)
Male, n (%)
APACHE II
Type of medical condition,
n (%)
Medical
Surgical
CPB surgery
Neurological
Tracheal intubation
Living alone
Physical restraint
Drinking
Smoking
Hospital LOS before
ICU admission (d)
CRP (mg/L)
Outcome measures
LOS in ICU (d)
Duration of mechanical
ventilation (d)
28-d mortality
Table 2
The main logistic regression model displaying independent risk factors for delirium
Total
(n = 223)
Patients with
delirium
(n = 54)
Patients
without
delirium
(n = 169)
57.2 17.3
141 (63.2)
13 (9-17)
64.5 18.1
37 (68.5)
17 (13-27)
54.9 16.3
104 (61.5)
11 (8-15)
b.001
.354
b.001
93 (41.7)
74 (33.2)
30 (13.5)
26 (11.7)
120 (54.1)
35 (15.7)
106 (47.8)
58 (26.1)
65 (29.4)
6 (0-15)
26 (48.2)
14 (25.9)
1 (1.9)
13 (24.1)
39 (73.6)
16 (29.6)
40 (75.5)
20 (37.0)
18 (34.0)
11 (0-16)
67 (39.6)
60 (35.5)
29 (17.2)
13 (7.7)
81 (47.9)
19 (11.2)
66 (39.1)
38 (22.6)
47 (28.0)
5 (0-13)
.270
.193
.004
.001
.001
.001
b.001
.036
.404
.029
67.7 (21.5128)
120.5 (45.9156)
57.5 (16.6100.2)
5 (3-11)
1 (1-6)
13 (6-21)
6 (1-14)
5 (3-8)
1 (1-2)
34 (15.5)
11 (21.2)
23 (13.7)
.0001
b.001
b.001
.193
Data were expressed as mean SE, median and interquartile range, or number and
percentage as appropriate.
Parameters
OR
95% CI
1.07
1.002
0.67
1.13
1.50
1.75
2.80
2.23
0.94
1.54
1.01
1.01-1.15
0.97-1.04
0.27-1.62
1.06-1.21
0.56-4.04
0.51-5.94
0.99-7.90
0.84-5.98
0.32-2.79
0.99-2.37
0.97-1.05
.029
.894
.370
b.001
.425
.371
.053
.109
.917
.052
.651
All variables that may interact with CRP and delirium were included in the main model.
Logistic regression diagnostics: variance ination factor for all variables was less than 10.
The model was well tted with P = .392 in Hosmer-Lemeshow goodness-of-t test.
and those without delirium. Patients with mixed and hypoactive types
of delirium had signicantly higher CRP levels than those without
delirium (129.1 vs 57.5 mg/L, P b .001; 120.5 vs 57.5 mg/L, P = .031,
respectively). Kaplan-Meier survival curves were depicted to compare
the difference of the time to development of delirium in patients with
CRP greater than 67.7 mg/L vs patients with CRP less than or equal to
67.7 mg/L (Fig. 3). Patients who died or were transferred to general
ward without the development of delirium were considered as
censored. The result showed that patients with higher CRP values on
ICU entry were more likely to develop delirium (log-rank test, P b .05).
Furthermore, subgroup analysis by APACHE II score quartiles
indicated that the association between CRP and delirium was more
prominent in patients with higher APACHE II scores (Fig. 4).
4. Discussion
Our study showed that CRP on ICU entry was associated with
subsequent development of delirium during ICU stay. With each 10
mg/L increase in CRP, the risk of delirium increased by 7% (OR, 1.07;
95% CI, 1.01-1.05). The predictive value of CRP is more prominent in
patients with higher APACHE II scores. Because CRP is also
independently associated with subsequent development of delirium,
serial measurement of CRP may provide useful information for the
prevention and treatment of delirium.
The result of the present study was in line with the study by
Macdonald et al [14]. However, that study showed a better predictive
value for CRP. By examining the patient characteristics, we found that
the study population were much older in the study of Macdonald et al
(82.7 vs 55.8 years), suggesting that CRP performed well in older
population. However, further well-designed studies are needed to test
this hypothesis. Another study conducted by McGrane et al [15] used
delirium-/coma-free days as the study end point. They found that
higher CRP levels showed trends toward fewer delirium-/coma-free
days, but the difference was not statistically signicant. This study is
Table 3
Unadjusted and adjusted OR for development of delirium by CRP category
CRP quartiles
Unadjusted
OR (95% CI)
OR (95% CI)
1
0.53 (0.19-1.48)
0.70 (0.27-1.83)
3.18 (1.39-7.26)
.228
.469
.006
1
0.27 (0.03-2.10)
1.15 (0.26 to 5.05)
4.47 (1.28-15.60)
.210
.857
.019
Change in CRP was the difference between the CRP values measured on ICU entry and
24 hours later.
a
Variables including age, sex, APACHE II, intubation status, living alone, physical
restraint, alcohol drinking, smoking, type of medical condition, and hospital LOS before
ICU admission.
91
Fig. 3. Kaplan-Meier plot of the probability of delirium from ICU admission to day 15.
The result showed that patients with higher CRP values on ICU entry were more likely
to develop delirium (log-rank test, P b .05).
Fig. 2. Box plot showing CRP concentrations in patients with and without delirium.
Note: , CRP values were signicantly higher in patients with mixed and hypoactive
delirium than those without (129.1 vs 57.5 mg/L, P b .001; 120.5 vs 57.5 mg/L, P = .031,
respectively).
Fig. 4. Subgroup analysis by APACHE II score quartiles showing the OR for delirium
development with each 10 mg/L increase in CRP.
92
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]