Nutrition in Critical Care

E ENTERAL
ARLY
NUTRITION AND OUTCOMES
OF CRITICALLY ILL PATIENTS
TREATED WITH
VASOPRESSORS AND
MECHANICAL VENTILATION
By Imran Khalid, MD, Pratik Doshi, MD, and Bruno DiGiovine, MD, MPH
Objective To determine the effect of early enteral feeding on
the outcome of critically ill medical patients whose hemody-
namic condition is unstable.
Methods Prospectively collected data in a multi-institutional
medical intensive care unit database were analyzed retrospec-
tively. A total of 1174 patients were identified who required
mechanical ventilation for more than 2 days and were treated
with vasopressor agents to support blood pressure. The patients
were divided into 2 groups: those who received enteral nutri-
tion (n = 707) within 48 hours of the start of mechanical venti-
lation, termed the early enteral nutrition group, and those who
did not (n = 467), termed the late enteral nutrition group. The
primary end points were overall intensive care unit and hospital
mortality. Subgroup analyses were used to evaluate potential
confounding variables. The data were also analyzed after adjust-
ments for confounding by matching for propensity score.
Results Intensive care unit and hospital mortality were lower
in the early enteral nutrition group than in the late enteral group:
22.5% vs 28.3%; P = 03; and 34.0% vs 44.0%; P < .001, respec-
tively. The beneficial effect of early feeding was more evident
in the sickest patients, that is, those treated with multiple vaso-
pressors (odds ratio, 0.36; 95% confidence interval, 0.15-0.85)
and those without early improvement (odds ratio, 0.59; 95%
confidence interval, 0.39-0.90). When adjustments were made
for confounding by matching for propensity score, early feeding
was associated with decreased hospital mortality.
This article is followed by an AJCC Patient Care Page Conclusion Early enteral nutrition may be associated with
on page 269.
reduced intensive care unit and hospital mortality in patients
whose hemodynamic condition is unstable. (American Journal
©2010 American Association of Critical-Care Nurses of Critical Care. 2010;19:261-268)
doi: 10.4037/ajcc2010197

www.ajcconline.org AJCC AMERICAN JOURNAL OF CRITICAL CARE, May 2010, Volume 19, No. 3 261

Downloaded from ajcc.aacnjournals.org by guest on February 5, 2015

 N
utritional support is an essential component in critical care. Malnutrition has
been associated with poor outcomes among patients in intensive care units
(ICUs), as indicated by increased morbidity, mortality, and length of stay.1-5
Evidence suggests that in patients with a functional gut, nutrition should be
administered through the enteral route, largely because of the morbidity asso-
ciated with other routes of feeding. Parenteral nutrition is especially associated with increased
infectious complications.6 In critically ill patients without severe sepsis, compared with enteral
nutrition, parenteral nutrition results in longer ICU stays and a greater chance that severe sep-
sis or septic shock will develop.7 Other favorable effects of enteral nutrition include better
substrate utilization, prevention of mucosal atrophy, preservation of the integrity of gut flora,
and preservation of immunocompetence.8-16 Early enteral feeding can also ameliorate oxidative
stress after surgery17 and can decrease postoperative mortality in patients who have gastroin-
testinal surgery.18 Therefore, physicians have become interested in feeding patients as soon as
possible. In a recent study,19 early feeding significantly reduced ICU and hospital mortality in
medical ICU patients treated with mechanical ventilation. However, the investigators did not
specifically analyze the data on patients who were receiving vasopressors.

In healthy adults, enteral nutrition is associated sidered a relative or absolute contraindication to early
with an increase in blood flow to the gut.20,21 In patients enteral feeding.26-28
whose hemodynamic condition is unstable, enteral Despite the guidelines, many clinicians continue
nutrition has been considered problematic mainly to feed patients whose hemodynamic condition is
for 2 reasons. The first reason is gut unstable. Thus, we sought to examine the impact
Enteral nutrition ischemia, and the best data indicating of early enteral nutrition in critically ill medical
an increase in ischemia with feeding patients in unstable hemodynamic condition as
should be used if were obtained in a study of rats with indicated by treatment with vasopressors. We
occlusion of the mesenteric artery.22 The hypothesized that early enteral feeding would be
the patient has a relevance of this model to patients associated with lower mortality.
functional gut. without occluded arteries has been
questioned.23 The second reason is the Methods
“steal” phenomenon, an increase in splanchnic Data were obtained from a large, multi-
blood flow without an increase in overall cardiac institutional critical care patient data set (Project
output.24,25 The impact of this phenomenon on clini- Impact Critical Care Data System, Society of Critical
cal outcomes is not clear. Nevertheless, because of Care Medicine, Des Plaines, Illinois; see http://www
these concerns, hemodynamic instability has been con- .trianalytics.com/programs_pi.html for project details).
For the Project Impact data set, coordinators at each
of the participating sites collected the data prospec-
About the Authors tively from patients’ charts. Data for the study
Imran Khalid is a staff physician, Division of Pulmonary
and Critical Care Medicine, John D. Dingell VA Medical reported here were acquired in January 2003, after
Center in Detroit, Michigan, and a consultant intensivist approval of the study protocol by the Project Impact
at King Faisal Specialist Hospital and Research Center in study committee. The research design was approved
Jeddah, Saudi Arabia. Pratik Doshi is a staff physician,
Division of Emergency Medicine and Pulmonary Critical by the appropriate institutional review board.
Care Medicine, Henry Ford Hospital in Detroit. Bruno For the study reported here, data were requested
DiGiovine is associate chair and chief quality and medical on all nonsurgical patients admitted to an ICU who
officer, Department of Internal Medicine, and an associate
professor, Division of Pulmonary and Critical Care Medi- received mechanical ventilation during their ICU stay
cine, Wayne State University School of Medicine, in Detroit. whose hemodynamic condition was unstable at
the time mechanical ventilation was started. Patients
Corresponding author: Bruno DiGiovine, MD, MPH, Division were considered in unstable hemodynamic condition
of Pulmonary and Critical Care Medicine, Wayne State
University School of Medicine, 4201 St. Antoine, Ste 2E, if they were given the vasopressor agents norepi-
Detroit, MI 48201 (e-mail: brunod1@comcast.net). nephrine, epinephrine, dopamine, or phenylephrine

262 AJCC AMERICAN JOURNAL OF CRITICAL CARE, May 2010, Volume 19, No. 3 www.ajcconline.org

Downloaded from ajcc.aacnjournals.org by guest on February 5, 2015

during the first 2 days of ventilatory support. Study with vasopressors for more than 2 days after the
variables included age, sex, race, admitting diagnosis, start of intubation. Also, because the main analysis
medications received, Mortality Prediction Model at was based on vital status at discharge regardless of
time zero (MPM-0) score, Simplified Acute Physiologic time, further analyses were done by using only 28-
Score (SAPS) II, and Acute Physiologic and Chronic day outcomes. Finally, the subgroup of patients for
Health Evaluation (APACHE) II score. Ventila- whom a decision may have been made to forgo
tor-associated pneumonia (VAP) in the Project aggressive therapy in the ICU who
Impact database was defined as new or progressive did not die in the ICU were Early enteral
infiltrate, consolidation, cavitation or pleural effu- defined as patients whose ICU dis-
sion, and any of the following: new onset of puru- charge condition was “moribund.” nutrition started
lent sputum or change in character of sputum;
organism isolated from blood culture; isolation of Statistical Analysis
within 48 hours of
pathogen from specimen obtained by tracheal aspi- SAS software (SAS Institute Inc, mechanical venti-
rate, bronchial brushing, or biopsy; and diagnosis Cary, North Carolina) was used for
of pneumonia based on histopathological findings. all statistical analysis. Baseline lation onset.
Data on expected (based on the SAPS II score) and characteristics of the early and late
observed mortality rate for each ICU were also col- enteral nutrition groups were compared by using
lected. These data were used to calculate a standard- unpaired t tests for normally continuous variables
ized mortality ratio for each ICU. and the Kruskal-Wallis test for nonnormally distrib-
The primary outcome variables were ICU and uted data. The χ2 test was used for dichotomous vari-
hospital mortality. Secondary outcome variables ables. Kaplan-Meier survival analysis was used to
included occurrence of VAP and ICU length of stay, evaluate the impact of early feeding on mortality;
ventilator-free days, and vasopressor-free days. Ven- the time from mechanical ventilation to death was
tilator-free days were defined as the number of days compared in the 2 groups by using a log-rank test.
within the first 28 days after initial intubation that Logistic regression was used to evaluate the effect
a patient was breathing independently of the venti- of early enteral nutrition on ICU and hospital mortal-
lator. Vasopressor-free days were defined as the ity after adjustments were made for important con-
number of days within the first 28 days after initial founders. In order to control for severity of illness,
intubation that a patient was alive and not receiv- 3 separate models were developed; 1 model used
ing vasopressor agents. No information allowing APACHE II scores, 1 used SAPS II values, and 1 used
identification of individual patients, hospitals, or MPM-0 scores. In each model, adjustments were made
physicians was supplied. for age, sex, race, source of admission, standardized
The cohort was divided into 2 groups according mortality ratio for the ICU, and admitting diagnosis,
to when enteral nutrition was first started. The early because these variables were considered important
enteral nutrition group included patients who were confounders. Thus, the variables were included in all
started on enteral feeding within 48 hours of the models regardless of the P values
start of mechanical ventilation. The remainder of associated with the variables. Odds
the patients were the late enteral nutrition group. ratio were computed from the coeffi- Patients in
Patients who died or were extubated within 2 days cients in the logistic model, and 95% the early enteral
of the start of mechanical ventilation were excluded. confidence intervals were calculated
Attempts were made to exclude patients who most for all variables. Findings were con- nutrition group
likely had an absolute or relative contraindication sidered significant at α < .05. In order had a lower
to enteral feeding at the time of admission. Thus, to determine the effect of feeding on
patients admitted because of gastrointestinal the risk of death, Cox proportional hospital mortality.
obstruction or bleeding, intestinal ileus, gastropare- hazard analyses were used. Again, 3
sis, acute pancreatitis, peritonitis, ischemic colitis, different models were constructed, each with a differ-
or esophageal rupture were excluded. Patients who ent severity-of-illness score and controls for all of the
received total parenteral nutrition before treatment variables included in the logistic regression.
with mechanical ventilation also were excluded. Logistic regressions were also used to evaluate
In order to ensure that the impact of feeding the effect of early enteral nutrition on VAP. In these
was not valid solely in the patients given minimal analyses, the models also included the use of drugs
vasopressor treatment, analyses were done on 2 that might have confounded the analysis, specifi-
subgroups: patients who were given 2 or more cally, histamine2-blockers, proton pump inhibitors,
vasopressor agents and patients who were treated narcotics, and paralytic agents.

www.ajcconline.org AJCC AMERICAN JOURNAL OF CRITICAL CARE, May 2010, Volume 19, No. 3 263

Downloaded from ajcc.aacnjournals.org by guest on February 5, 2015

 Table 1
Baseline characteristics of the
study population (N = 1174)a
et al32 by using an SAS software macro described by
Enteral nutrition group Parsons.33 Basically, a randomly selected patient
who was fed early was selected from the popula-
Early Late
Characteristic (n = 707) (n = 467) P tion. Then, data on all of the patients who were not
fed early were searched to find a match on the
Age, mean (SD), y 64.8 (15.2) 62.8 (17.2) .04 propensity score (within 0.01 on a scale from 0 to
Sex, No. (%) of patients 1). This procedure was continued until all possible
Male 384 (54.3) 260 (55.7) pairs were identified. The success of the matching
.65
Female 323 (45.7) 207 (44.3) was determined by evaluating differences in individ-
Race, No. (%) of patients ual demographic data.
White 580 (82.0) 360 (77.1) After matching was completed, the new matched
African American 89 (12.6) 82 (17.6)
data set was evaluated to assess the effect of feeding
Hispanic 19 (2.7) 12 (2.6) .10
Other 15 (2.1) 7 (1.5) on ICU mortality, hospital mortality, ICU length of
Unknown 4 (0.6) 8 (1.7) stay, duration of mechanical ventilation, and occur-
Admission source, No. (%) of patients
rence of VAP. All the analyses were done by using
Outpatient 374 (52.9) 261 (55.9) methods that accounted for the matched design.
General care floor 258 (36.5) 162 (34.7) Continuous variables were compared by using a
Another hospital’s ICU 55 (7.8) 28 (6.0) .56 matched t test (Proc Mixed in SAS), and dichoto-
Extended care facility 17 (2.4) 10 (2.1) mous outcomes were compared by using conditional
Unknown 3 (0.4) 6 (1.3)
logistic regression (Proc Phreg in SAS). Kaplan-Meier
Reason for ICU admission, No. (%) of patients survival analyses were conducted without accounting
Respiratory problem 383 (54.2) 171 (36.6)
for matching; however, survival was also analyzed
Sepsis 98 (13.9) 79 (16.9)
Cardiac problem 97 (13.7) 125 (26.8) <.001 by using Cox proportional hazard methods with
Central nervous system disorder 66 (9.3) 32 (6.9) adjustments for matching.
Others 63 (8.9) 60 (12.8)
Severity scores, mean (SD) Results
APACHE II 23 (6.8) 24.6 (8.1) .002 Patients
MPM-0 0.39 (0.23) 0.42 (0.26) .07 At the time of the query, 1174 patients in the
SAPS II 52 (15.2) 55.3 (15.6) <.001 Project Impact database met our inclusion criteria.
Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; ICU,
Among these, 707 (60%) received early enteral
intensive care unit; SAPS II, Simplified Acute Physiology Score II; MPM-0, Mortality nutrition and 467 (40%) did not. At least 1 severity-
Prediction Model at time zero.
a Because of rounding, not all percentages total 100. of-illness score was available in each patient’s record:
APACHE II in 924 records (78.7%), SAPS II in 1045
(89.0%), and MPM-0 in 1008 (85.9%). The means
Matching by Propensity Score of the APACHE II, SAPS II, and MPM-0 values were
As in any nonrandomized protocol, the 2 groups 23.6 (SD, 7.4), 53.3 (SD, 15.4), and 0.60 (SD, 0.15),
might have had inherent differences. Therefore, analy- respectively. We found no difference (P = .06) in the
ses were done to specifically control for potential lowest mean arterial pressure in the first 24 hours
confounding variables (see preceding). However, even between the early feeding group (mean, 54.6; SD,
with these methods, residual bias may 12.4) and late feeding group (mean, 53.2; SD, 12.4).
Enteral nutrients exist. Propensity score methods have Patients in the early enteral nutrition group
were older than those in the late enteral nutrition
been proposed to control for such
increase blood biases.29-31 Use of propensity scores group and were more likely to be admitted to the
enables better control for the likeli- ICU with a respiratory diagnosis (Table 1). Addi-
flow to the gas- hood of being assigned to a group. tionally, the early feeding group had a small but sta-
trointestinal tract In this study, the likelihood of tistically significant lower severity of illness as
being fed early was modeled by using indicated by the SAPS II and APACHE II values, but
and decrease bac- logistic regression. SAPS II value, age, not by the MPM-0 score.
terial translocation. sex, site of origin, standardized mor-
tality ratio for the ICU, and APACHE Mortality Analysis
II score at admission were used in the regression. In an unadjusted analysis (Table 2), patients in
This analysis allowed calculation of the probability the early enteral nutrition group had a lower ICU
of being fed for each patient. Matching was done mortality than did those in the late enteral nutrition
according to the procedure described by Connors group (22.5% vs 28.3%; P = .03) and lower hospital

264 AJCC AMERICAN JOURNAL OF CRITICAL CARE, May 2010, Volume 19, No. 3 www.ajcconline.org

Downloaded from ajcc.aacnjournals.org by guest on February 5, 2015

 Table 2
Comparison of clinical outcomes in
early and late enteral nutrition groups
mortality (34% vs 44%; P < .001). To evaluate the
independent effect of feeding on ICU and hospital
Enteral nutrition group
mortality, we constructed 3 different multivariate
logistic models (see “Methods” section). This Early Late
Characteristic (n = 707) (n = 467) P
analysis revealed that regardless of the severity-of-
illness score used, early enteral nutrition was con- Intensive care unit mortality,
sistently associated with a lower risk of hospital No. (%) of patients 159 (22.5) 132 (28.3) .03
mortality (Table 3). Age, severity of illness, the Hospital mortality, No. (%) of patients 239 (33.8) 205 (43.9) <.001
ICU’s standardized mortality ratio, and source of
Ventilator-associated pneumonia,
admission were all significant factors in these No. (%) of patients 70 (9.9) 45 (9.6) .88
analyses. Kaplan-Meier analysis of survival indi-
Days in intensive care unit, mean (SD) 27 (12.2) 25.9 (12.0) .14
cated a significant improvement in survival for the
patients fed early (P < .001; Figure 1). The absolute Ventilator-free days,a mean (SD) 15.7 (9.4) 14.9 (10.3) .17
difference in survival between the 2 groups was
a Ventilator-free days are the number of days (among the first 28 days after intubation)
evident within the first week of mechanical venti-
that the patient spends breathing independently of the ventilator.
lation and remained constant throughout the first
28 days of follow-up after intubation. Cox propor-
tional hazard analyses indicated that after correc- for matching. The results indicate that being fed
tions for confounders, early feeding was associated early was associated with a 34% decreased risk of
with a 30% to 35% decreased risk of death. death (odds ratio, 0.66; 95% confidence interval,
0.49-0.89; P = .006).
Matched Analysis
Matching for propensity score yielded 1264 Subgroup Analyses
pairs of patients who were within 0.01 points on Analyses of subgroups (Table 6) indicated
this score. This matching allowed us to find 2 well- that the beneficial effect of early feeding is more
matched groups (Table 4). We found no significant evident in the sickest patients, that is, patients
differences in any of the baseline values tested. receiving multiple vasopressors (odds ratio, 0.36;
Using this well-matched subgroup of patients, 95% confidence interval, 0.15-0.85), and in patients
we again found significant differences in outcome without early improvement, that is, patients who
(Table 5). Patients in the early enteral nutrition required vasopressors for more than 2 days (odds
group had significantly (P = .01) lower hospital ratio, 0.59; 95% confidence interval, 0.39-0.90).
mortality (34.1%) than did the late enteral nutri- Also, the effect was present when data on patients
tion group (42.7%). The difference in survival is who were likely to die after ICU discharge (as evi-
also apparent in the Kaplan-Meier survival curve denced by ICU discharge condition of “moribund”)
(Figure 2). Because this analysis could not be done were excluded. Finally, the results did not change
while accounting for the matched design, we did a significantly when the outcomes were assessed at
Cox proportional hazard analysis after accounting 28 days.

Table 3
Outcomes associated with
early enteral nutrition groupa

Intensive care unit Ventilator-associated Risk of death by Cox

Severity score mortality Hospital mortality pneumonia PH model

APACHE II 0.81 (0.58-1.13) 0.65 (0.48-0.89) 0.77 (0.43-1.36) 0.70 (0.56-0.88)

SAPS II 0.77 (0.56-1.04) 0.64 (0.48-0.85) 0.91 (0.53-1.58) 0.68 (0.55-0.84)
MPM-0 0.75 (0.55-1.03) 0.61 (0.46-0.81) 0.95 (0.55-1.63) 0.65 (0.53-0.81)

Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; Cox PH, Cox proportional hazard; MPM-0, Mortality Prediction Model at time zero;
SAPS II, Simplified Acute Physiology Score II.
a Results of the different multivariable models shown as odds ratio (95% confidence interval). Each analysis was done after adjusting for each of the severity-of-
illness scores as well as age, sex, standardized mortality ratio, race, source of admission, and admitting diagnosis.

www.ajcconline.org AJCC AMERICAN JOURNAL OF CRITICAL CARE, May 2010, Volume 19, No. 3 265

Downloaded from ajcc.aacnjournals.org by guest on February 5, 2015

 100

90 Discussion
In this study, initiation of enteral nutrition within
Percent survival

48 hours of mechanical ventilation was associated

80 Early feeding with a reduction in the rates of hospital mortality in a
large cohort of critically ill hemodynamically unstable
70 patients treated with mechanical ventilation. This
Late feeding
reduction in mortality was apparent even after we
controlled for confounders both in multivariate mod-
60
P = .001 els and by using propensity-score matching. Further-
more, the effect was consistent across multiple
50 subgroups. In previous studies34,35 in humans, specifi-
5
0 cally critically ill surgical and trauma patients, early
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5
enteral nutrition was associated with improvement in
Days after intubation
surrogate end points, but no improvement in mortal-
ity with early feeding was reported. In a recent study,36
Figure 1 Survival of patients in the early and late enteral nutri-
tion groups. ICU patients also did not have a difference in out-
come when given early enteral nutrition. However, in
a population limited to medical patients treated with
mechanical ventilation, enteral feeding was associated
Table 4 with improved mortality.19 None of these previous
Baseline characteristics of the studies specifically addressed early enteral nutrition in
matched study population (n = 714)a patients who had an unstable hemodynamic status
Enteral nutrition group and were being treated with vasopressors.
Current clinical guidelines26,27,35 generally pro-
Early Late
Characteristic (n = 357) (n = 357) P
mote early enteral nutrition, but evidence is lacking
in patients whose hemodynamic condition is unsta-
Age, mean (SD), y 63.2 (15.8) 62.0 (17.8) .32 ble as indicated by use of vasopressors, and guide-
Sex, No. (%) of patients lines tend to recommend delayed enteral nutrition
Male 189 (52.9) 190 (53.2) .94 for these patients. These recommendations exist
Female 168 (47.1) 167 (46.8) because many published studies lack relevance to
Race, No. (%) of patients hemodynamically unstable patients, because of the
White 280 (78.4) 269 (75.4) selection of participants and outcome variables.37
African American 56 (15.7) 66 (18.5) Another hindrance to early feeding in critically ill
Hispanic 11 (3.1) 11 (3.1) .78
Other 7 (2.0) 6 (1.7)
patients in unstable condition is the notion that
Unknown 3 (0.8) 5 (1.4) these patients have limited oxygen delivery and that
by increasing gastrointestinal oxygen demand with
Admission source, No. (%) of patients
Outpatient 186 (52.1) 198 (55.5) enteral feeding, more harm may occur if intestinal
General care floor 128 (35.9) 125 (35.0) ischemia develops. However, these propositions are
Another hospital’s ICU 31 (8.7) 21 (5.9) .43 based on evidence from animal models. Because
Extended care facility 10 (2.8) 8 (2.2) studies in humans do not answer this specific con-
Unknown 2 (0.6) 5 (1.4)
cern, recommendations for early feeding in the criti-
Reason for ICU admission, No. (%) of patients cally ill are basically empiric.
Respiratory problem 156 (43.7) 158 (44.3) Gastrointestinal blood flow is reduced in patients
Sepsis 76 (21.3) 61 (17.1)
Cardiac problem 50 (14.0) 57 (16.0) .63
after various types of critical illnesses, and, more
Central nervous system disorder 28 (7.8) 30 (8.4) important, gut blood flow remains depressed despite
Others 47 (13.2) 51 (14.3) fluid replacement and normalization of blood pres-
Severity scores, mean (SD) sure and cardiac output. This reduction in blood flow
APACHE II 24.0 (6.9) 24.2 (7.9) .72 is associated with ischemic injury, bacterial translo-
MPM-0 0.59 (0.24) 0.59 (0.26) .85 cation, and multiple organ failure. The belief that
SAPS II 54.8 (15.4) 54.1 (15.0) .50 patients whose hemodynamic condition is unstable
Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; ICU, inten-
who are receiving vasopressors should not be given
sive care unit; SAPS II, Simplified Acute Physiology Score II; MPM-0, Mortality Prediction enteral nutrition is based on the concern that nutri-
Model at time zero.
a Because of rounding, not all percentages total 100. ent absorption increases oxygen demand. In patients
whose hemodynamic condition is unstable, this

266 AJCC AMERICAN JOURNAL OF CRITICAL CARE, May 2010, Volume 19, No. 3 www.ajcconline.org

Downloaded from ajcc.aacnjournals.org by guest on February 5, 2015

 Table 5
Comparison of clinical outcomes in
early and late enteral nutrition groups
increase in demand theoretically could exceed the
after matching for propensity score
supply, leading to further complications. However,
contrary to this belief, enteral nutrients increase Enteral nutrition group
blood flow to the gastrointestinal tract, a phenome- Early Late
non referred to as “postprandial hyperemic response.” Characteristic (n = 357) (n = 357) P
Studies have been done in animals to determine the
effect of postprandial hyperemia during splanchnic Intensive care unit mortality,
No. (%) of patients 77 (21.6) 95 (26.6) .12
ischemia, as may occur in critically ill patients,
especially patients treated with vasopressors. Many Hospital mortality,
No. (%) of patients 121 (33.9) 152 (42.6) .01
investigators have reported that enteral nutrition
improves splanchnic blood flow.24 In fact, although Ventilator-associated pneumonia,
feeding increased gastrointestinal oxygen consump- No. (%) of patients 39 (10.9) 35 (9.8) .63
tion, the concomitant increase in oxygen delivery led Days in intensive care unit,
to better delivery to consumption ratio in the fed mean (SD) 12.4 (8.6) 11.1 (7.7) .39
vs the unfed state.38 Furthermore, this physiological Ventilator-free days,a mean (SD) 16.0 (9.2) 15.2 (10.3) .29
process can decrease bacterial translocation and
aVentilator-free days are the number of days (among the first 28 days after intubation)
improve survival in multiple experimental sepsis mod- that the patient spends breathing independently of the ventilator.
els.24 Thus, on the basis of studies in animals, what
might happen to patients in unstable hemodynamic
condition who are given enteral nutrition is unclear.
100
Our study has some limitations. First, it is a
retrospective analysis, and we base our results on
intent-to-treat analysis, that is, whether a patient 90
was started on any enteral nutrition within 48 hours
Percent survival

of mechanical ventilation. The study does not take 80

Early feeding
into account patients’ total caloric intake, rate of
advancement, and whether disruptions in the feeding 70
occurred. However, our results imply a favorable out- Late feeding
come with intent to provide early enteral nutrition.
60
The second limitation is confounding by indi- P = .003
cation, that is, the decision to feed was not made
randomly. Possibly, sicker patients were not fed 50
5.0
2.5
because of their condition, and feeding may simply 0.0
0 7 14 21 28
be a marker of a less ill patient. Also, if a physician
Days after intubation
is more likely to initiate enteral nutrition early, he
or she would also be more likely to follow other
Figure 2 Survival of patients in early and late enteral nutrition
measures to improve outcome or initiatives to groups in matched analysis.
decrease rates of health care–associated infection.
We attempted to control for this possibility by using
multivariate analysis that included severity of illness be taken to minimize complications from enteral
and standardized mortality ratios for the individual feeding (eg, bowel necrosis associated with early
ICUs. We also excluded patients with absolute or jejunal feeding).39
relative contraindications to feeding. In addition,
we used propensity-score matching as has been Conclusion
used in other studies. Even with these methods, we This comparison of early vs late enteral nutrition
could not control for the unmeasured variables, and suggests that early enteral feeding is associated with
they may have provided some residual confounding. reduction in the mortality of critically ill patients
However, we think that the differences in outcomes receiving mechanical ventilation who are in an
were too large to be explained exclusively by con- unstable hemodynamic condition as indicated by
founding by indication. We recommend a prospec- use of vasopressors. The beneficial effect of early
tive randomized trial to evaluate the effect of early feeding is more evident in the sickest patients, that
enteral nutrition on the mortality of patients receiv- is, those treated with multiple vasopressors. In addi-
ing mechanical ventilation who are in an unstable tion, we found no evidence of harm due to the early
hemodynamic condition. In such a trial, care should enteral nutrition. These results provide justification

www.ajcconline.org AJCC AMERICAN JOURNAL OF CRITICAL CARE, May 2010, Volume 19, No. 3 267

Downloaded from ajcc.aacnjournals.org by guest on February 5, 2015

 for a randomized controlled trial to further address 19. Artinian V, Krayem H, DiGiovine B. Effects of early enteral
feeding on the outcome of critically ill mechanically venti-
this controversial issue. lated medical patients. Chest. 2006;129(4):960-967.
20. Gallavan RH, Chou CC. Possible mechanisms for the initia-
ACKNOWLEDGMENTS tion and maintenance of postprandial intestinal hyperemia.
Am J Physiol. 1984;249:301-308.
The research was done at Henry Ford Hospital.
21. Norryd C, Denker H, Lunderquist A, et al. Superior mesen-
teric blood flow during digestion in man. Acta Chir Scand.
FINANCIAL DISCLOSURES 1975;141(3):197-202.
None reported. 22. Kles KA, Wallig MA, Tappenden KA. Luminal nutrients exac-
erbate intestinal hypoxia in the hypoperfused jejunum.
JPEN J Parenter Enteral Nutr. 2001;25(5):246-253.
23. Zaloga GP, Roberts PR, Marik P. Feeding the hemodynami-
eLetters cally unstable patient: a critical evaluation of the evidence.
Now that you’ve read the article, create or contribute to an Nutr Clin Pract. 2003;18(4):285-293.
online discussion on this topic. Visit www.ajcconline.org 24. Kazamias P, Kotzampassi K, Koufogiannis D, et al. Influence
and click “Respond to This Article” in either the full-text or of enteral nutrition-induced splanchnic hyperemia on the
PDF view of the article. septic origin of splanchnic ischemia. World J Surg. 1998;
22(1):6-11.
25. Revelly JP, Tappy L, Berger MM, et al. Early metabolic and
splanchnic responses to enteral nutrition in postoperative
REFERENCES cardiac surgery patients with circulatory compromise.
1. Chandra RK. Nutrition, immunity, and infection: present Intensive Care Med. 2001;27(3):540-547.
knowledge and future directions. Lancet. 1983;26(1):688-691. 26. ASPEN Board of Directors and the Clinical Guidelines Task
3 Bassili HR, Deitel M. Effect of nutritional support on weaning Force. Guidelines for the use of parenteral and enteral
patients off mechanical ventilators. JPEN J Parenter Enteral nutrition in adult and pediatric patients [published correction
Nutr. 1981;5(2):161-163. appears in JPEN J Parenter Enteral Nutr. 2002;26(2):144].
3. Haydock DA, Hill GL. Improved wound healing response in JPEN J Parenter Enteral Nutr. 2002;26(1 suppl):1SA-138SA.
surgical patients receiving intravenous nutrition. Br J Surg. 27. Jolliet P, Pichard C, Biolo G, et al. Enteral nutrition in inten-
1987;74(4):320-323. sive care patients: a practical approach. Working Group on
4. Sullivan DH, Sun S, Walls RC. Protein-energy undernutrition Nutrition and Metabolism, ESICM. European Society of
among elderly hospitalized patients: a prospective study. Intensive Care Medicine. Intensive Care Med. 1998;24(8):
JAMA. 1999;281(21):2013-2019. 848-859.
5. Robinson G, Goldstein M, Levine GM. Impact of nutritional 28. McClave SA, Chang WK. Feeding the hypotensive patient:
status on DRG length of stay. JPEN J Parenter Enteral Nutr. does enteral feeding precipitate or protect against ischemic
1987;11(1):49-51. bowel? Nutr Clin Pract. 2003;18(4):279-284.
6. Sena MJ, Utter GH, Cuschieri J, et al. Early supplemental 29. Joffe MM, Rosenbaum PR. Invited commentary: propensity
parenteral nutrition is associated with increased infectious scores. Am J Epidemiol. 1999;150(4):327-333.
complications in critically ill trauma patients. J Am Coll 30. D’Agostino RB Jr. Propensity score methods for bias reduc-
Surg. 2008;207(4):459-467. tion in the comparison of a treatment to a non-randomized
7. Radrizzani D, Bertolini G, Facchini R, et al. Early enteral control group. Stat Med. 1998;17(19):2265-2281.
immunonutrition vs parenteral nutrition in critically ill 31. Drake C, Fisher L. Prognostic models and the propensity
patients without severe sepsis: a randomized clinical trial. score. Int J Epidemiol. 1995;24(1):183-187.
Intensive Care Med. 2006;32(8):1191-1198. 32. Connors AF Jr, Speroff T, Dawson NV, et al. The effectiveness
8. Hadfield RJ, Sinclair DG, Houldsworth PE, et al. Effects of of right heart catheterization in the initial care of critically
enteral and parenteral nutrition on gut mucosal permeability ill patients. SUPPORT Investigators. JAMA. 1996;276(11):
in the critically ill. Am J Respir Crit Care Med. 1995;152: 889-897.
1545-1548. 33. Parsons L. Reducing bias in a propensity score matched-pair
9. Minard G, Kudsk KA. Is early feeding beneficial? How early sample using greedy matching techniques. Paper presented
is early? New Horiz. 1994;2(2):156-163. at: 26th Annual SAS Users Group International Conference;
10. Gianotti L, Alexander JW, Nelson JL, et al. Role of early April 22-25, 2001; Long Beach, CA. Paper 214-26.
enteral feeding and acute starvation on postburn bacterial 34. Marik PE, Zaloga GP. Early enteral nutrition in acutely ill
translocation and host defense: prospective, randomized patients: a systematic review. Crit Care Med. 2001;29(12):
trials. Crit Care Med. 1994;22(2):265-272. 2264-2270.
11. Chuntrasakul C, Siltharm S, Chinswangwatanakul V, et al. 35. Heyland DK, Dhaliwal R, Drover JW, Gramlich L, Dodek P;
Early nutritional support in severe traumatic patients. J Canadian Critical Care Clinical Practice Guidelines Commit-
Med Assoc Thai. 1996;79(1):21-26. tee. Canadian clinical practice guidelines for nutrition sup-
12. Tanigawa K, Kim YM, Lancaster JR, et al. Fasting augments port in mechanically ventilated, critically ill adult patients.
lipid peroxidation during reperfusion after ischemia in the JPEN J Parenter Enteral Nutr. 2003;27(5):355-373.
perfused rat liver. Crit Care Med. 1999;27(2):401-406. 36. Doig GS, Simpson F, Finfer S, et al; Nutrition Guidelines
13 Bortenschlager L, Roberts PR, Black KW, et al. Enteral feeding Investigators of the ANZICS Clinical Trials Group. JAMA.
minimizes liver injury during hemorrhagic shock. Shock. 2008;300(23):2731-2741.
1994;2(5):351-354. 37. Preiser JC, Chioléro R, Wernerman J; ESICM (European
14. Beier-Holgersen R, Brandstrup B. Influence of early postop- Society of Intensive Care Medicine) Working Group on
erative enteral nutrition versus placebo on cell-mediated Nutrition and Metabolism. Nutritional papers in ICU patients:
immunity, as measured with the Multitest CMI. Scand J what lies between the lines? Intensive Care Med. 2003;
Gastroenterol. 1999;34(1):98-102. 29(2):156-166.
15. Shou J, Lappin J, Minnard EA, et al. Total parenteral nutrition, 38. Purcell PN, Davis K Jr, Branson RD, et al. Continuous duo-
bacterial translocation, and host immune function. Am J denal feeding restores gut blood flow and increased gut
Surg. 1994;167(1):145-150. oxygen utilization during PEEP ventilation for lung injury.
16. Quigley EM, Marsh MN, Shaffer JL, et al. Hepatobiliary Am J Surg. 1993;165(1):188-193.
complications of total parenteral nutrition. Gastroenterology. 39. Melis M, Fichera A, Ferguson MK. Bowel necrosis associated
1993;104(1):286-301. with early jejunal tube feeding: a complication of postoper-
17. Kotzampassi K, Kolios G, Manousou P, et al. Oxidative stress ative enteral nutrition. Arch Surg. 2006;141(7):701-704.
due to anesthesia and surgical trauma: importance of early
enteral nutrition. Mol Nutr Food Res. 2009;53(6):770-779.
18. Lewis SJ, Andersen HK, Thomas S. Early enteral nutrition To purchase electronic or print reprints, contact The
within 24 h of intestinal surgery versus later commencement InnoVision Group, 101 Columbia, Aliso Viejo, CA 92656.
of feeding: a systematic review and meta-analysis. J Gastroin- Phone, (800) 899-1712 or (949) 362-2050 (ext 532); fax,
test Surg. 2009;13(9):569-575. (949) 362-2049; e-mail, reprints@aacn.org.

268 AJCC AMERICAN JOURNAL OF CRITICAL CARE, May 2010, Volume 19, No. 3 www.ajcconline.org

Downloaded from ajcc.aacnjournals.org by guest on February 5, 2015

Correction
In the May 2009 supplement to the American
Journal of Critical Care, “Fecal Containment in
Bedridden Patients: Economic Impact of 2 Commer-
cial Bowel Catheter Systems,” by Areta Kowal-Vern
Looking for
and colleagues (2009;18[3]:S2-S15), the following
information should have been added to the last
an Article?
paragraph of the Procedures section on page S5:
A sample size of 67 subjects was required to
determine a 25% difference between catheter A and In its continuing efforts to GO
catheter B (α = 0.05, β = 0.20, power is 0.80) when
catheter B was not expected to exceed a 60% Green, the American Journal of Critical
increase in linen change (less than 1 unscheduled Care has eliminated its printed subject-
change per day).
The sample size was based on an alpha correc- author index.
tion to accommodate the planned interim analysis, It’s easy to find an article online at
and consists of a 10% increase to accommodate
dropout. www.ajcconline.org. Enter a keyword,
doi: 10.4037/ajcc2010769
title, or author name in the Search box
and the search engine will do the rest.

Correction
In the May 2010 article by Khalid and colleagues, “Early Enteral Nutrition and Outcomes of Critically Ill Patients Treated
With Vasopressors and Mechanical Ventilation” (Am J Crit Care. 2010;19[3]:261-268), Table 6 was omitted from the article.
The table appears below. We regret the error.
doi: 10.4037/ajcc2010765

Table 6
Effect of early enteral nutrition group on unadjusted
and adjusted hospital morality in a variety of subgroups

Unadjusted hospital mortality rates

No. of deaths/total in group (%) Adjusted hospital mortalitya

Early enteral Late enteral 95% confidence

Subgroup nutrition nutrition P Odds ratio interval

Total group 239/707 (33.8) 205/467 (43.9) .001 0.65 0.48-0.89

Censored at 28 days 213/707 (30.1) 184/467 (39.4) .001 0.69 0.50-0.93
Number of pressors
1 200/608 (32.9) 159/385 (41.3) .008 0.7 0.50-0.98
>1 39/99 (39.4) 46/82 (56.1) .03 0.36 0.15-0.85
Early improvement
Yes 100/358 (27.9) 72/221 (32.6) .24 0.67 0.42-1.08
No 139/349 (39.8) 133/246 (54.1) .001 0.59 0.39-0.90
Moribund at discharge from
intensive care unit 24/30 (80.0) 28/36 (77.8) .83 b b

Yes 215/677 (31.8) 177/431 (41.1) .002 0.65 0.48-0.90

No
a Adjustedhospital mortalities were determined after adjusting for the score on the Acute Physiology and Chronic Health Evaluation II as well as age, sex, standard-
ized mortality ratio, race, source of admission, and admitting diagnosis.
b Results
are not shown for patients who were moribund at discharge from the intensive care unit because the groups were too small for stable modeling.

488 AJCC AMERICAN JOURNAL OF CRITICAL CARE, November 2010, Volume 19, No. 6 www.ajcconline.org
Early Enteral Nutrition and Outcomes of Critically Ill Patients Treated With
Vasopressors and Mechanical Ventilation
Imran Khalid, Pratik Doshi and Bruno DiGiovine
Am J Crit Care 2010;19:261-268 doi: 10.4037/ajcc2010197
© 2010 American Association of Critical-Care Nurses
Published online http://www.ajcconline.org

Personal use only. For copyright permission information:

http://ajcc.aacnjournals.org/cgi/external_ref?link_type=PERMISSIONDIRECT

Subscription Information
http://ajcc.aacnjournals.org/subscriptions/

Information for authors

http://ajcc.aacnjournals.org/misc/ifora.xhtml

Submit a manuscript
http://www.editorialmanager.com/ajcc

Email alerts
http://ajcc.aacnjournals.org/subscriptions/etoc.xhtml

AJCC, the American Journal of Critical Care, is the official peer-reviewed research
journal of the American Association of Critical-Care Nurses (AACN), published
bimonthly by The InnoVision Group, 101 Columbia, Aliso Viejo, CA 92656.
Telephone: (800) 899-1712, (949) 362-2050, ext. 532. Fax: (949) 362-2049.
Copyright © 2010 by AACN. All rights reserved.

Downloaded from ajcc.aacnjournals.org by guest on February 5, 2015