Journal of Critical Care 29 (2014) 230–235

Contents lists available at ScienceDirect

Journal of Critical Care

journal homepage: www.jccjournal.org

The effects of discharge to an intermediate care unit after a critical

illness: A 5-year cohort study☆,☆☆,★
Otavio T. Ranzani, MD a,⁎, Fernando Godinho Zampieri, MD a, b, Leandro Utino Taniguchi, MD, PhD a, c,
Daniel Neves Forte, MD, PhD a, c, Luciano César Pontes Azevedo, MD, PhD a, c, Marcelo Park, MD, PhD a, c
a
Intensive Care Unit, Emergency Medicine Discipline, Hospital das Clínicas, University of São Paulo, São Paulo, Brazil
b
Intensive Care Unit, Hospital Alemão Oswaldo Cruz, São Paulo, Brazil
c
Research and Education Institute, Hospital Sírio-Libanês, São Paulo, Brazil

a r t i c l e i n f o a b s t r a c t

Keywords: Purpose: The impact of the intermediate care unit (IMCU) on post–intensive care unit (ICU) outcomes is
Critical care controversial.
Discharge planning Materials and Methods: We analyzed admissions from January 2003 to December 2008 from a mixed ICU in a
Long-term survivors teaching hospital in Brazil with a high patient-to-nurse ratio (3.5:1 on the ICU, 11:1 on the IMCU, 20-25:1 on
Patient discharge
the ward). A retrospective propensity-matched analysis was performed with data from 690 patients who
Wards
were discharged after at least 3 days of ICU stay.
Results: Of the 690 patients, 160 (23%) were discharged to the IMCU. A total of 399 propensity-matched
patients were compared: 298 were discharged to the ward and 101 were discharged to the IMCU. Ninety-day
mortality rate was similar between the IMCU and ward patients (22% vs 18%, respectively, P = .37), as was the
unplanned ICU readmission rate (P = .63). In a multivariate logistic regression, discharge to the IMCU had no
effect on the 90-day mortality rate (P = .27).
Conclusions: In a resource-limited setting with a high patient-to-nurse ratio, discharge to IMCU had no impact
on 90-day mortality rate and on unplanned readmission rate. The impact of discharge to the IMCU on the
outcome for critically ill patients should be evaluated in further studies.
© 2014 Elsevier Inc. All rights reserved.

1. Introduction recovering from a critical illness [4]. The supposed objective of

Up to one third of the total mortality that occurs after a critical
illness occurs after a successful discharge [1,2]. This high mortality
rate after a discharge from the intensive care unit (ICU)
emphasizes the need for additional attention to this complex
decision-making process. Optimizing the discharge is one of the
many challenges that should be addressed by future research
studies [3].
Intermediate care units (IMCUs) have been proposed to deliver
transitional care between the ICU and the ward for patients
☆ Declaration of interests: The authors declare that no competing interests exist in
relation to the content of this study and the manuscript.
☆☆ Funding: This work has no funding.
★ Author's contributions: OTR, FGZ, LUT, DNF, LCPA, and MP conceived the study and
participated in its design. OTR and MP performed the statistical analysis. OTR, LCPA, and
MP collected the data for the study. OTR and FGZ drafted the manuscript. LUT, DNF,
LCPA, and MP critically revised the manuscript for important intellectual content. All
authors read and approved the final manuscript.
⁎ Corresponding author. Rua Dr Enéas de Carvalho Aguiar, 255, Fifth Floor,
Emergency Medicine Discipline, São Paulo, Brazil, 05403-000. Tel.: +55 11 2661 6457.
E-mail address: otavioranzani@yahoo.com.br (O.T. Ranzani).
the IMCU is to continuously provide patient care with a reduced
bed to nurse ratio. Two mechanisms may be related to the impact
of IMCU on prognosis. First, IMCU may allow earlier discharge of
critically ill patients, thereby reducing ICU length of stay (LOS) and
costs, without changing mortality [4-7]. Secondly, IMCU may
provide a longer better care after discharge, reducing postdischarge
mortality. Nevertheless, there are no randomized controlled
studies, and the current literature is controversial, being that they
are mainly before and after studies or post hoc analyses without
further adjustments [4,8-11]. Indeed, implementation of an IMCU
has not been associated with reduced costs [5] and may add patient
stress due to the additional transfer process [12]. Furthermore,
discharge to an IMCU was recently described as an independent
risk factor for an ICU readmission [8,9,13]. The impact of IMCU may
be even more relevant on resource-limited settings with a high
patient-to-nurse ratio.
One of the possible modifiable factors involved in a post-ICU
prognosis is the destination facility. The impact of a transfer to an
IMCU after an ICU discharge is still unknown. The aim of the present
study was to evaluate if an IMCU admission, instead of a ward transfer,
after a critical illness was associated with post-ICU outcomes.

0883-9441/$ – see front matter © 2014 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcrc.2013.10.013
 O.T. Ranzani et al. / Journal of Critical Care 29 (2014) 230–235
2. Materials and methods
2.1. Population
The patient population in this study included all living patients
discharged from an adult ICU after at least 72 hours of stay from
Hospital das Clínicas, a tertiary teaching hospital in São Paulo, Brazil,
from January 2003 to December 2008. Patients discharged to another
ICU or to another hospital were excluded.
This study protocol followed the statements of the Declaration of
Helsinki. The institutional review board, called the Comissão para
Análise de Projetos de Pesquisa, reviewed and approved this study
(protocol no. 107 443). The requirement for written informed consent
was waived because there was no intervention, and only a database
that had guaranteed confidentiality was used.
2.2. Intensive care
Our study included patients discharged from a mixed ICU with
7 beds. This unit followed current standard-of-care practices,
including sedation, nutritional, mechanical ventilation, and hemo-
dynamic monitoring protocols. An intensivist was available on site
24 hours a day, 7 days a week (24/7). The staff comprised 1 senior
physician, 1 critical care fellow, and 3 residents from the internal
medicine program. At night, there was 1 senior physician and 1
resident. The health staff comprised 2 nurses and 3 nurse
assistants on a 24/7 schedule in addition to a respiratory therapist
who was on a 12/7 schedule.
2.3. Intermediate care unit
The referral IMCU at our hospital, at the time of this study, was an
11-bed unit with 1 nurse, 2 nurse assistants, and an intensivist
consultant who was available 24/7. The staff comprised 1 senior
physician, 2 critical care fellows, and 5 residents from the internal
medicine program. At night, there was 1 senior physician on-call and
2 residents, 1 of which was from the critical care program. The
intensivists evaluated the patients at the IMCU on a daily basis. A
respiratory therapist consultant was also available on a 24/7 schedule.
All of the decisions about the IMCU patients were shared with the
IMCU team. The senior physician of the ICU and IMCU were from the
same department and were in contact about the transfer process and
follow-ups, thereby minimizing the loss of any important information
regarding a patient's condition.
Patients on the IMCU received routine nurse care, basic vital
signs monitoring (continuous electrocardiogram, oxygen satura-
tion, and automated blood pressure measurements), frequent
reassessments of neurologic status (at least 4 times a day),
monitoring of urinary output and fluid balance, and 2 sessions of
respiratory therapy a day. Mechanical ventilation was not allowed
on the IMCU. Need for oxygen supplementation of more than 50%
through Venturi mask was considered an indication for transfer to
the ICU. Brief periods of noninvasive ventilation (b2 hours per
period) were allowed. Continuous infusion of vasopressors, contin-
uous antiarrhythmic agents, or vasodilators (including nitroprus-
side) was not allowed on the IMCU. Intermittent venous
hemodialysis was allowed on the IMCU.
2.4. General wards
Our hospital contains wards for medical and surgical specialties.
The mean nurse-to-patient ratio, at the time of this study, was 1 nurse
for every 20 to 25 patients and 1 nurse assistant for every 4 to 6
patients. The staff for each ward was variable, consisting of mainly
residents and medical students who were under the supervision of an
231
assistant physician. On the weekends, the availability of the ward staff
was lower than during the weekdays.
2.5. Discharge criteria
The patients were discharged to the ward or to the IMCU at the
discretion of the attending intensive care physician. In brief, a
discharge from the ICU was only considered if the reason for
admission was solved or controlled. Patients who demanded any
dose of vasopressors, inotropes, or intravascular vasodilators and/or
intravenous antiarrhythmic agents were not discharged because
those therapies were not allowed on the IMCU (see before). The
need for noninvasive ventilation for more than 2 hours per day period
(ie, morning, afternoon, and night) was considered a contraindication
to an ICU discharge.
2.6. Data collection
All data were recorded prospectively with a computerized
physician order entry system. The admission data included age, the
reason for admission, the physiological data, the Sequential Organ
Failure Assessment (SOFA) score, and the Acute Physiology and
Chronic Health Evaluation II (APACHE II) score [14,15]. The daily SOFA
score and physiological variables were also collected. The patients
were followed up to 90 days after their ICU discharge with the
hospital online system.
2.7. Outcomes
The primary end point for this study was the patient mortality rate
over a 90-day period after an ICU discharge. The secondary end points
included the hospital mortality rate and the unplanned ICU read-
mission rate.
2.8. Statistical analysis
Patients were categorized into 2 groups according to their
discharge destinations. The discharge destinations were either the
ward or the IMCU (these groups are described as the ward group and
the IMCU group, respectively). Categorical and continuous data are
presented as percentages and as the mean ± SD (or median and 25th-
75th percentile), respectively. Categorical variables were compared
using the χ 2 or Fisher exact tests, as appropriate. The quantitative
continuous variables were compared using the unpaired Student t test
or the Mann-Whitney U test for parametric and nonparametric
variables, respectively.
Because the discharge process was not randomly assigned in our
study population, a selection bias was accounted for by using the
propensity scores [16-19] and a standard multivariate logistic
regression. For more information on propensity score and multivar-
iate analysis, see the online supplement.
The Kaplan-Meier method, with a log-rank test, was used to
analyze the time to discharge from the hospital for living patients. The
association between an unplanned ICU readmission and discharge to
either the IMCU or the ward was assessed according to the Fine and
Gray model [20]. This model extends the Cox regression model by
taking into account any competing risk data and by considering the
hazard function associated with the cumulative incidence function
(CIF). Thus, the informative censoring was taken into account, that is,
after dismissal from the ICU, a patient could be discharged from the
hospital (censored), readmitted to the ICU, or die in the hospital
without an ICU readmission (informative censoring, that is, the
survival time of an individual does depend on censoring). Therefore,
the CIF and the Fine and Gray model allowed for handling of both the
time-to-event and the informative censoring [21,22].
232 O.T. Ranzani et al. / Journal of Critical Care 29 (2014) 230–235

The statistical analyses were made in SPSS 19.0 and with the R Interestingly, nighttime dismissal was more frequent in the ward
project (version 2.4 Patched, www.r-project.org). The significant P group (27% vs 8%; P b 0.001), and weekend discharge was more
value was .05 for all comparisons. common in the IMCU group (37% vs 27%; P = .014).

3. Results 3.2. The baseline characteristics after propensity matching

3.1. The baseline characteristics before propensity matching
The general cohort consisted of 690 living patients who were
discharged after at least 3 days of an ICU LOS. Five hundred thirty
patients (77%) were discharged to the ward, and 160 (23%), to the
IMCU. The patients discharged to the ward or to the IMCU had
comparable ages, sex, and source of admission (Table 1). However, the
patients discharged to the ward more frequently had systemic arterial
hypertension and chronic kidney failure. Chronic liver failure was
more common in the patients discharged to the IMCU (Table 1). In
addition, lactate levels were higher in the IMCU group (1.88 vs 1.66
mmol/L; P = .002) (Table 1), and this group had higher initial APACHE
II (17 vs 15; P = .002) and SOFA (5 vs 4; P = .028) scores (Table 2). In
addition, the SOFA score was higher for a longer duration of time in
the IMCU group during their ICU stays (Table 2).
The main reason for an ICU admission was similar between the
groups (P = .74) as well as the LOS before an ICU admission (P = .88)
and the ICU LOS (P = 0.34). Vasoactive support during an ICU stay was
similar between the groups (39% vs 41%, P = .69, between the IMCU
and the ward groups, respectively), and renal replacement therapy
was more common for the patients discharged to the IMCU (P b .001).
However, there was a trend toward a more frequent use of mechanical
ventilation in the ward group (P = .074) (Table 2).
Following the propensity score matching procedure (model 1),
399 matched patients remained for analysis, 298 (75%) of which were
in the ward group and 101 (25%) were in the IMCU group. The main
characteristics of this matched cohort remained similar to the crude
patient cohort, including the demographic, epidemiological, severity,
and discharge features. The proportion of the patients discharged to
the IMCU and the ward was maintained in the matched cohort
(Tables 1 and 2).
In a sensitivity analysis, 455 patients (123 [27%] in the IMCU group
and 332 [73%] in the ward group) were matched using model 2, and
466 patients (122, 26% in the IMCU group, and 344, 74% in the ward
group) were matched using model 3. The main characteristics of these
matched cohorts remained similar to the crude patient cohort.
After analyzing the matched cohorts, the agreement of the selected
patients between the models was 57%, 59%, 64%, and 44% for models 1
and 2; models 1 and 3; models 2 and 3; and models 1, 2, and 3,
respectively. Of note, 597 patients (87%) from the entire cohort were
represented in these matched analyses.
3.3. The destination facility and the patient outcomes
In the propensity-adjusted logistic regression model, the mortality
rate over a 90-day period was not different between the ward and

Table 1
Patient's characteristics at ICU admission

Unmatched cohort Propensity matched cohort

Ward group (n =530) IMCU group (n =160) P Ward group (n =298) IMCU group (n =101) P

General characteristics
Age, y (mean ± SD) 51 ± 20 49 ± 18 .28 50 ± 20 49 ± 19 .80
Male, n (%) 272 (51) 86 (54) .59 155 (52) 51 (51) .79
Origin of patients, n (%) .58 .97
Ward 160 (30) 52 (33) .64 86 (29) 30 (30) .87
Emergency department 328 (62) 91 (57) .26 186 (62) 64 (63) .87
Other ICU 21 (4) 8 (5) .57 13 (4) 4 (4) 1.00
Operating room 11 (2) 6 (4) .230 6 (2) 2 (2) 1.00
Step-down unit 3 (1) 2 (1) .33 1 (1) – 1.00
Other 7 (1) 1 (1) .69 6 (2) 1 (1) .68
Comorbidities, n (%)
Systemic hypertension 278 (53) 68 (43) .027 142 (48) 48 (48) .98
Diabetes mellitus 117 (22) 32 (20) .58 57 (19) 22 (22) .56
Chronic renal failure 97 (18) 16 (10) .013 46 (15) 11 (11) .26
Chronic heart failure 91 (17) 19 (12) .109 40 (13) 12 (12) .69
Chronic coronary disease 57 (11) 19 (12) .69 33 (11) 11 (11) .96
Chronic obstructive pulmonary disease 48 (9) 11 (7) .39 22 (7) 5 (5) .50
Chronic liver disease 8 (2) 8 (5) .016 7 (2) 1 (1) .69
AIDS 13 (2) 6 (4) .30 9 (3) 3 (3) 1.00
Cancer 54 (10) 17 (11) .87 32 (11) 7 (11) .97
Hour of discharge, n (%)
Night (19.00-06:59) 141 (27) 12 (8) b.001 46 (15) 10 (10) .166
Day of discharge, n (%)
Weekend 142 (27) 59 (37) .014 82 (28) 33 (33) .32
Laboratorial data at discharge, median (IQR)
pH 7.42 (7.39; 7.45) 7.41 (7.38; 7.44) .042 7.42 (7.39; 7.45) 7.41 (7.38; 7.44) .220
pCO2 (mm Hg) 36 (32; 41) 37 (33; 41) .111 36 (32; 41) 37 (33; 41) .77
BE (mEq/L) −0.3 (−2.6; 1.9) −0.7 (−3.8; 2.35) .50 −0.3 (−2.8; 2.2) −0.8 (−4.1; 2.6) .50
Lactate (mmol/L) 1.66 (1.22; 2.22) 1.88 (1.44; 2.44) .002 1.66 (1.22; 2.33) 1.77 (1.33; 2.55) .228
Hemoglobin level (g/dL) 9 (8; 11) 9 (8; 11) .33 9.4 (8.1; 11.2) 9.4 (8.1; 11.1) .97
WBC (units × 103/mm3) 9.6 (6.8; 12.9) 10.4 (6.9; 13.8) .195 9.6 (6.8; 12.7) 10.1 (7.3; 13.7) .238
Platelets (units × 103/mm3) 243 (165; 349) 240 (166; 332) .47 234 (160; 327) 256 (176; 342) .29
Creatinine (mg/dL) 0.80 (0.60; 1.20) 0.83 (0.61; 1.45) .32 0.80 (0.60; 1.30) 0.80 (0.61; 1.35) .82
Albumin (g/L) 27 (23; 31) 27 (22; 33) .62 27 (23; 31) 28 (24; 33) .35
CRP at discharge (mg/L) 44 (21; 102) 51 (24; 114) .157 46 (23; 100) 50 (22; 113) .68

IQR indicates interquartile range; BE, base excess; WBC, white blood cell count; CRP, C-reactive protein.
 O.T. Ranzani et al. / Journal of Critical Care 29 (2014) 230–235 233

Table 2
Admission reason, severity of illness, and ICU support during ICU stay

Unmatched cohort Propensity matched cohort

Ward group (n = 530) IMCU group (n = 160) P Ward group (n = 298) IMCU group (n = 101) P

Illness severity, median (IQR)

APACHE II score 15 (10-20) 17 (12-25) .002 16 (10-21) 15 (10-22) .74
SOFA at admission 4 (2-7) 5 (3-8) .028 4 (2-7) 5 (2-7) .86
SOFA D3 4 (2-6) 4 (2-6) .42 4 (2-6) 3 (2-6) .64
SOFA maximum 6 (3-9) 7 (4-11) .006 6 (4-9) 6 (3-10) .64
SOFA 72 h 2 (1-4) 3 (2-5) .017 3 (2-5) 3 (1-4) .33
SOFA 48 h 2 (1-4) 2 (1-4) .033 2 (1-4) 2 (1-4) .39
SOFA at discharge 2 (1-3) 2 (1-4) .047 2 (0-3) 2 (1-4) .087
Mean SOFA 3 (2-5) 4 (2-5) .064 3 (2-5) 3 (2-5) .57
Type of admission, n (%)
Medical admission 426 (80) 124 (78) .43 229 (77) 78 (77) .94
Syndrome at admission, n (%) .74 .71
Respiratory failure 176 (33) 44 (28) .175 99 (33) 31 (31) .64
Shock 97 (18) 28 (18) .82 45 (15) 15 (15) .95
Neurologic disorder 56 (11) 22 (14) .27 32 (11) 17 (17) .107
Acute renal failure 16 (3) 7 (4) .40 11 (4) 5 (5) .58
Electrolytic disturbances 20 (4) 5 (3) .81 11 (4) 3 (3) 1.00
Multiple trauma 16 (3) 6 (4) .65 10 (3) 6 (6) .25
Infection 263 (50) 77 (48) .74 159 (53) 61 (60) .219
Support during ICU stay, n (%)
Noninvasive ventilation 138 (26) 17 (11) b.001 66 (22) 17 (17) .26
Mechanical ventilation 262 (50) 92 (58) .074 146 (49) 55 (55) .34
Renal replacement therapy 61 (12) 41 (26) b.001 47 (16) 17 (17) .80
Vasoactive drugs 218 (41) 63 (39) .69 112 (38) 39 (39) .85
LOS, d (IQR)
Before ICU admission 2 (0-6) 1 (0-6) .88 1 (0-7) 1 (0-4) .39
ICU 7 (4-12) 6 (4-11) .34 7 (4-11) 7 (4-11) .79
Post-ICU LOS 11 (5-24) 22 (10-38) b.001 11 (5-22) 20 (11-38) b.001
ICU plus post-ICU LOS 21 (12-36) 29 (17-47) b.001 20 (12-35) 33 (22-53) b.001
Total hospital LOS 25 (15-47) 34 (21-54) .001 24 (15-42) 29 (18-45) .001

SOFA D3 indicates SOFA score at the third day of ICU stay; SOFA 72 h, SOFA score 72 hours before ICU discharge; SOFA 48 h, SOFA score 48 hours before ICU discharge.

IMCU groups (53 of 299 [18%] for the ward group vs 22 of 100 [22%]
for the IMCU group, respectively; odds ratio [OR], 1.29; 95%
confidence interval [CI], 0.74-2.25; P = .34). With regard to the
hospital mortality, there was also no association between the
discharge facility and the final outcomes (OR, 1.12; 95% CI, 0.63-
1.98; P = .71) (Fig. 1). Following utilization of the sensitivity analyses
(models 2 and 3), there was no difference between the 90-day and
total hospital mortality rates (eTable1).
The univariate analyses for the variables associated with the 90-
day mortality rate are depicted in eTables 2 and 3. In the standard
multivariate logistic regression model, a discharge of a patient to the
IMCU facility was not associated with a change in the 90-day mortality
rate (adjusted OR, 1.32; 95% CI, 0.81-2.16; P = .27) (Table 3).
In the propensity-matched cohort, the risk of an unplanned ICU
readmission was similar between the groups (OR, 1.16; 95% CI, 0.64-
2.11; P = .63; Fig. 1), and the Fine and Gray model revealed no
differences between the patients discharged to the IMCU or the ward
(P = .140; Fig. 2A). Nevertheless, the patients discharged to the IMCU
had a longer hospital LOS after their ICU discharge when compared
with the patients discharged to the ward (log-rank test, P = .004;
Fig. 2B). As might be expected, this was reflected in the total hospital
LOS difference between the IMCU and ward groups (20 [11-38] vs 11
[5-22] days, respectively, P b .001). We also examined the total
hospital LOS because the patients discharged to the IMCU could be
discharged earlier from the ICU and thus represented a falsely
increased post-ICU LOS. As depicted in Fig. 1, the total hospital LOS
remained high within the patients discharged to the IMCU.
With a sensitivity analysis, the Fine and Gray model revealed no
differences regarding the risk of an unplanned ICU readmission
between the groups (P = .07 and P = .33, for models 2 and 3,
respectively). The patients discharged to the IMCU maintained a
longer hospital LOS after their ICU discharge compared with the

Table 3
Multivariate logistic regression model to predict 90-day mortality

OR 95% CI P

Discharged to an IMCU 1.32 0.81-2.16 .27

Agea 1.03 1.01-1.04 b.001
Chronic comorbiditiesa 1.25 1.06-1.47 .007
Fig. 1. The main outcomes after an ICU discharge in the propensity-matched cohort. The
Admitted from the ward 2.12 1.36-3.31 .001
mortality rate over 90 days after an ICU discharge was similar between the groups (P =
Albumin level at dischargea 0.64 0.44-0.93 .019
.34), as was the total hospital mortality rate (P = .71) and the unplanned ICU
SOFA score at dischargea 1.27 1.15-1.41 b.001
readmission rate (P = .63). Hospital LOS after an ICU discharge and the total hospital
a
stay were longer for patients delivered to the IMCU than to the ward (P b .001 and P = Odds ratio calculated per 1 unit of change; area under the curve, 0.78 (0.74-0.82); P
.001, respectively). The error bars denote the interquartile range. b .001; Nagelkerke R2 = 0.253; Hosmer and Lemeshow test (χ2 = 3.494), P = .90.
234 O.T. Ranzani et al. / Journal of Critical Care 29 (2014) 230–235
Fig. 2. The time-dependent analysis after an ICU discharge, including the readmission
and hospital discharge data. A, depicts the CIF of the ICU readmission rate from the Fine
and Gray model. B, depicts the probability of patients remaining in the hospital
following an ICU discharge.
patients discharged to the ward (log-rank test, P = .007 and P = .003,
for models 2 and 3, respectively).
4. Discussion
In this retrospective, propensity-matched cohort study, a dis-
charge to the IMCU was associated with a similar 90-day mortality
rate when compared with a discharge to the ward. The total hospital
mortality and ICU readmission rates were also similar, irrespective of
the destination after the ICU discharge. However, the IMCU provided
the availability of a discharge for severely injured patients and for
discharges on the weekends.
Our analysis suggests that patients discharged to the IMCU tended
to be more severely ill at the time of admission and during their ICU
stays than patients discharged directly to the ward. This theory was
already suggested and highlights the concept of progressive care for
critical illness [23]. Because illness severity, residual organ failure, and
residual inflammation are strongly associated with mortality after an
ICU discharge [1,24], the attending physician may discharge high-risk
patients to the IMCU more safely. Nevertheless, when patients paired
for illness severity were evaluated, discharge to the IMCU had no
impact on the mortality rate or the need for readmission. The benefit
of a discharge to an IMCU could be diminished by the risks associated
with an inappropriately early ICU discharge [10].
Interestingly, a discharge to the ward was more common during
the nighttime, which could be related to an urgent need for an ICU bed
[25] or for a delay in the transfer process. On the other hand, a
discharge to the IMCU was more common during weekends. Because
there are less nurses and hospitalists in the wards during weekends,
the IMCU might be an ideal transitional care option to a patient
recently released from the ICU. Furthermore, an IMCU could provide
scheduled discharges on the weekends.
The finding that an IMCU discharge resulted in a longer hospital
LOS should be expected because these patients were sicker than the
patients discharged to the ward. Similar results were reported when
low-risk ICU patients were compared with patients admitted to the
IMCU [11]; however, some subjective factors might have influenced
these results. Hospital stay is a surrogate of the total health care costs
in an economy, and although we did not evaluate this outcome, our
findings could explain the increased costs from critically ill patients
after the introduction of an IMCU facility into a hospital [5]. Vincent
and Burchardi [4] proposed that mixed units, with “intensive” and
“intermediate” patients, could overcome this process and take
advantage of a global care process mixed with staff interests.
The process of a discharge after a critical illness is complex. The
high mortality rates after an ICU discharge suggest that more
attention should be given to the whole discharge process, including a
patient's clinical condition and destination facility. In this context,
hospital policies and the demand for ICU beds in overcrowded ICUs
also influence the ICU dismissal strategy. Because highly dependent
patients who do not require invasive support occupy many ICU beds,
an IMCU could unload the burden of these patients on the ICU [26]
and reduce the shortage of ICU beds in many hospitals [11]. Indeed,
an IMCU is intended to rationalize the management of an ICU, rather
than to improve patient outcomes. Our results support the concept
that an IMCU may enhance the management of an ICU rather than
the ICU performance, by bringing more flexible local care to the
attending physician.
Some of the strengths of this study should also be emphasized.
First, this is a topic that involves more than 1 area of intensive care
research, including patient outcomes, management, and quality.
Indeed, there is a growing interest in determining both the short-
and long-term prognoses after an ICU discharge as well as the ways
to predict patient outcomes [27-31]. The currently ongoing
European Mortality & Length of Intensive Care Unit Stay Evaluation
(ClinicalTrials.gov identifier NCT01422070) was designed to answer
relevant questions about the discharge process and specifically the
role of IMCUs. Second, currently, there is little evidence that IMCUs
have an effect on patient outcomes [10,11,32]. There are no
randomized controlled studies, and the current literature is
controversial, being that they are mainly before and after studies
or post hoc analyses without further adjustments. Third, this was a
study with data collected over a 5-year period and a cohort of
critically ill patients who were followed up for 90 days after their
discharge from the ICU.
This study has some shortcomings, however, that must be
acknowledged as well. First, it is a retrospective and single-center
analysis, thus subjected to local biases. Our analysis should be solely
interpreted on the context of a limited resource setting with a high
patient-to-nurse ratio. Second, we intentionally removed patients
who remained in the ICU for less than 72 hours from the analysis, to
select a subgroup of more severely ill individuals. Therefore, we could
not evaluate low-risk or postoperative patients, a subgroup of patients
with possible advantages when the fast-track approach was applied,
decreasing the generalizability of our results. Third, we were unable to
identify the cause of in-hospital deaths and the main reason for a
readmission after an ICU discharge. Fourth, this was not a randomized
controlled trial, and the use of the propensity score method to
overcome this limitation was also restricted because of unmeasured
confounders, as previously stated.
 O.T. Ranzani et al. / Journal of Critical Care 29 (2014) 230–235
5. Conclusions
In this retrospective analysis, discharge to the IMCU after a critical
illness had a similar 90-day mortality rate and a similar need for
readmission when compared with discharge to the hospital ward.
However, the IMCU may enhance the ICU management by bringing
more flexible local care to the attending physician. The exact role of
the IMCU in critically ill patient outcomes should be evaluated in
further studies.
Appendix A. Supplementary data
Supplementary material is available at Journal of Critical
Care online.
References
[1] Moreno R, Miranda DR, Matos R, et al. Mortality after discharge from intensive
care: the impact of organ system failure and nursing workload use at discharge.
Intensive Care Med 2001;27:999–1004.
[2] Goldhill DR, Sumner A. Outcome of intensive care patients in a group of British
intensive care units. Crit Care Med 1998;26:1337–45.
[3] Brett S. Critical care outcomes: current and future issues. Curr Opin Crit Care
2011;17:493–4.
[4] Vincent JL, Burchardi H. Do we need intermediate care units? Intensive Care Med
1999;25:1345–9.
[5] Solberg BC, Dirksen CD, Nieman FH, et al. Changes in hospital costs after
introducing an intermediate care unit: a comparative observational study. Crit
Care 2008;12:R68.
[6] Zimmerman JE, Wagner DP, Knaus WA, et al. The use of risk predictions to identify
candidates for intermediate care units. Implications for intensive care utilization
and cost. Chest 1995;108:490–9.
[7] Daly K, Beale R, Chang RW. Reduction in mortality after inappropriate early
discharge from intensive care unit: logistic regression triage model. BMJ
2001;322:1274–6.
[8] Brown SE, Ratcliffe SJ, Kahn JM, et al. The epidemiology of intensive care unit
readmissions in the United States. Am J Respir Crit Care Med 2012;185:955–64.
[9] Renton J, Pilcher DV, Santamaria JD, et al. Factors associated with increased risk of
readmission to intensive care in Australia. Intensive Care Med 2011;37:1800–8.
[10] Peelen L, de Keizer NF, Peek N, et al. The influence of volume and intensive care
unit organization on hospital mortality in patients admitted with severe sepsis: a
retrospective multicentre cohort study. Crit Care 2007;11:R40.
[11] Junker C, Zimmerman JE, Alzola C, et al. A multicenter description of intermediate-
care patients: comparison with ICU low-risk monitor patients. Chest 2002;121:
1253–61.
[12] Beard H. Does intermediate care minimize relocation stress for patients leaving
the ICU? Nurs Crit Care 2005;10:272–8.
235
[13] Campbell AJ, Cook JA, Adey G, et al. Predicting death and readmission after
intensive care discharge. Br J Anaesth 2008;100:656–62.
[14] Vincent JL, de Mendonca A, Cantraine F, et al. Use of the SOFA score to assess the
incidence of organ dysfunction/failure in intensive care units: results of a
multicenter, prospective study. Working group on “sepsis-related problems” of
the European Society of Intensive Care Medicine. Crit Care Med 1998;26:
1793–800.
[15] Knaus WA, Draper EA, Wagner DP, et al. APACHE II: a severity of disease
classification system. Crit Care Med 1985;13:818–29.
[16] Rosenbaum Pr RDB. The central role of the propensity score in observational
studies for casual effects. Biometrika 1983;70:41–55.
[17] Austin PC. An introduction to propensity score methods for reducing the effects of
confounding in observational studies. Multivar Behav Res 2011;46:399–424.
[18] Ranzani OT, Ferrer M, Esperatti M, et al. Association between systemic
corticosteroids and outcomes of intensive care unit-acquired pneumonia. Crit
Care Med 2012;40:2552–61.
[19] Clec'h C, Darmon M, Lautrette A, et al. Efficacy of renal replacement therapy in
critically ill patients: a propensity analysis. Crit Care 2012;16:R236.
[20] Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a
competing risk. J Am Stat Assoc 1999:496–509.
[21] Andersen PK, Abildstrom SZ, Rosthøj S. Competing risks as a multi-state model.
Stat Methods Med Res 2002;11:203–15.
[22] Clec'h C, Gonzalez F, Lautrette A, et al. Multiple-center evaluation of mortality
associated with acute kidney injury in critically ill patients: a competing risks
analysis. Crit Care 2011;15:R128.
[23] da Silva MC, de Sousa RM, Padilha KG. Patient destination after discharge from
intensive care units: wards or intermediate care units? Rev Lat Am Enfermagem
2010;18:224–32.
[24] Ranzani OT, Zampieri FG, Forte DN, et al. C-reactive protein/albumin ratio predicts
90-day mortality of septic patients. PLoS One 2013;8:e59321.
[25] Goldfrad C, Rowan K. Consequences of discharges from intensive care at night.
Lancet 2000;355:1138–42.
[26] Weissman C. Analyzing the impact of long-term patients on ICU bed utilization.
Intensive Care Med 2000;26:1319–25.
[27] Ouanes I, Schwebel C, Francais A, et al. A model to predict short-term death or
readmission after intensive care unit discharge. J Crit Care 2012;27(4):422.e1–9.
[28] Ranzani OT, Prada LF, Zampieri FG, et al. Failure to reduce C-reactive protein levels
more than 25% in the last 24 hours before intensive care unit discharge predicts
higher in-hospital mortality: a cohort study. J Crit Care 2012;27(5):
525.e9–525.e15.
[29] Bickenbach J, Fries M, Rex S, et al. Outcome and mortality risk factors in long-term
treated ICU patients: a retrospective analysis. Minerva Anestesiol 2011;77:
427–38.
[30] Karhu J, Ala-Kokko TI, Ylipalosaari P, et al. Hospital and long-term outcomes of
ICU-treated severe community- and hospital-acquired, and ventilator-associated
pneumonia patients. Acta Anaesthesiol Scand 2011;55:1254–60.
[31] Hofhuis JG, Spronk PE, van Stel HF, et al. The impact of critical illness on perceived
health-related quality of life during ICU treatment, hospital stay, and after hospital
discharge: a long-term follow-up study. Chest 2008;133:377–85.
[32] Iapichino G, Morabito A, Mistraletti G, et al. Determinants of post-intensive care
mortality in high-level treated critically ill patients. Intensive Care Med 2003;29:
1751–6.