Total Parenteral Nutrition in the Critically Ill Patient
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Caring for the Critically Ill Patient
Total Parenteral Nutrition
in the Critically Ill Patient
A Meta-analysis
Daren K. Heyland, MD, FRCPC, MSc; Shaun MacDonald MD, FRCSC;
Laurie Keefe, RD; John W. Drover, MD, FRCSC
Context.—Nutritional support has become a standard of care for hospitalized barrier structure and function, aug-
patients, but whether total parenteral nutrition (TPN) affects morbidity and mortality menting the inflammatory response to
is unclear. illness and resulting in greater infectious
Objective.—To examine the relationship between TPN and complication and morbidity.3-5 As a consequence, nutri-
mortality rates in critically ill patients. tional support has become a standard of
care for hospitalized patients.
Data Sources.—Computerized search of published research on MEDLINE from Because intestinal stimulation from
1980 to 1998, personal files, and review of relevant reference lists. luminal nutrients helps maintain gastro-
Study Selection.—We reviewed 210 titles, abstracts, and papers. Primary intestinal mucosal structure and func-
studies were included if they were randomized clinical trials of critically ill or surgi- tion,6-9 enteral nutrition may have some
cal patients that evaluated the effect of TPN (compared with standard care) on advantage over total parenteral nutri-
complication and mortality rates. We excluded studies comparing TPN with enteral tion (TPN). Compared with TPN, ran-
nutrition. domized trials of critically ill patients
Data Extraction.—Relevant data were abstracted on the methodology and out- have demonstrated that enteral nutri-
comes of primary studies. Data were abstracted in duplicate, independently. tion administered within the first 24
Data Synthesis.—There were 26 randomized trials of 2211 patients comparing hours of admission to the intensive care
unit (ICU) results in better wound heal-
the use of TPN with standard care (usual oral diet plus intravenous dextrose) in ing,10 a decrease in gastrointestinal tract
surgical and critically ill patients. When the results of these trials were aggregated, mucosal permeability,11 and lower infec-
TPN had no effect on mortality (risk ratio [RR], 1.03; 95% confidence interval [CI], tion rates.12-14 Where possible, enteral
0.81-1.31). Patients who received TPN tended to have a lower complication rate, feeding is preferred to parenteral feed-
but this result was not statistically significant (RR, 0.84; 95% CI, 0.64-1.09). We ex- ing.15 However, some patients with an
amined several a priori hypotheses and found that studies including only malnour- intact gastrointestinal tract do not tol-
ished patients were associated with lower complication rates but no difference in erate enteral feeds or do not receive suf-
mortality when compared with studies of nonmalnourished patients. Studies pub- ficient intake enterally or orally to meet
lished since 1989 and studies with a higher methods score showed no treatment their energy and protein requirements.
effect, while studies published in 1988 or before and studies with a lower methods Total parenteral nutrition is used as a
supplement or as the sole source of nu-
score demonstrated a significant treatment effect. Complication rates were lower trition in these patients16,17; however,
in studies that did not use lipids; however, there was no difference in mortality rates previous evidence supporting this prac-
between studies that did not use lipids and those studies that did. Studies limited tice seems to be lacking.18,19 Since these
to critically ill patients demonstrated a significant increase in complication and mor- studies were reviewed in 1987,19 addi-
tality rates compared with studies of surgical patients. tional randomized trials have been pub-
Conclusions.—Total parenteral nutrition does not influence the overall mortal- lished. The purpose of this article is to
ity rate of surgical or critically ill patients. It may reduce the complication rate, es- review systematically, appraise criti-
pecially in malnourished patients, but study results are influenced by patient popu- cally, and aggregate statistically studies
lation, use of lipids, methodological quality, and year of publication. evaluating the effect of TPN in critically
JAMA. 1998;280:2013-2019 ill patients.
METHODS
From the Departments of Medicine (Drs Heyland and MALNUTRITION among hospitalized
MacDonald) and Surgery (Dr Drover), Queen’s Univer- patients has been associated with in- Search Strategy
sity and Nutritional Services, Kingston General Hospital
(Ms Keefe), Kingston, Ontario. creased morbidity, prolonged hospital We conducted a computerized biblio-
Dr Heyland is a Career Scientist of the Ontario Minis- stay, and increased costs to the health graphic search of MEDLINE (including
try of Health. care system.1,2 Several studies have pre-MEDLINE) for studies from 1980
Reprints: Daren K. Heyland, MD, FRCPC, MSc, Angada
3, Kingston General Hospital, 76 Stuart St, Kingston, On- documented that “bowel rest” is associ- to April 1998 to locate all relevant ar-
tario, Canada K7L 2V7 (e-mail: dkh@post.queensu.ca). ated with a disruption of the mucosal ticles. The terms randomized controlled
JAMA, December 16, 1998—Vol 280, No. 23 Total Parenteral Nutrition in the Critically Ill Patient—Heyland et al 2013
©1998 American Medical Association. All rights reserved.Table 1.—Criteria Used to Assess Methodologic Quality* blinding the administration of TPN, we
Score
only awarded points for studies that
blinded the adjudication of study end
0 1 2 points. We also evaluated the extent to
Randomization ... Not concealed Concealed randomization which consecutive, eligible patients
or not sure
were enrolled in the trial, whether
Blinding Not blinded ... Adjudicators blinded
groups were equal at baseline, if cointer-
Analysis Other ... Intention to treat
ventions were adequately described,
Patient selection Selected patients or Consecutive eligible ...
unable to tell patients whether objective definitions of infec-
Comparability of groups No or not sure Yes ... tious outcomes were used, and whether
at baseline all patients were properly accounted for
Extent of follow-up ,100% 100% ... in the analysis (intention-to-treat analy-
Treatment protocol Poorly described Reproducibly described ... sis) (Table 1).
Cointerventions† Not described Described but not equal Well described and
or not sure all equal Data Extraction
Outcomes Not described Partially described Objectively defined
Two of us (D.K.H. and S.M.) extracted
*The first 3 questions and the last 2 questions had a possible score of 0, 1, or 2. The middle 3 questions had a data for analysis and assessment of the
possible score of 0 or 1. The highest possible score was 14. Ellipses indicate data not applicable. methodologic quality; we resolved dis-
†The extent to which antibiotics, enteral nutrition, ventilation, oxygen, and transfusions were applied equally across
groups. agreement by consensus. Not all studies
reported complication rates. Some stud-
trial, double blind method, clinical trial, differences that might exist between ies reported total complications per
placebo, and comparative study were these patients in the subgroup analysis. group but not on a per-patient basis.
combined with explode parenteral nu- We excluded studies of pediatric or neo- When data were missing, unclear, or not
trition, total. Citations were limited to natal patients. reported on a per-patient basis, we at-
English-language studies reporting on We included only studies that evalu- tempted to contact the primary investi-
adult patients. Reference lists of rel- ated the use of supplemental TPN in pa- gators and requested them to provide
evant review articles and personal files tients receiving enteral feeds or studies further information if the article had
were also searched. evaluating the use of TPN in patients been published in the last 5 years.
who were not receiving TPN or enteral
Study Selection Criteria nutrition. There are several randomized Prior Hypotheses Regarding Sources
Initially, 2 of us (D.K.H. and S.M.) trials of surgical patients that examine of Heterogeneity
screened all citations and classified them the effect of amino acid infusion (without When conducting a systematic re-
as primary studies, review articles, or additional nonprotein energy or lipids) view, heterogeneity (major differences
other. We then retrieved and reviewed on clinical outcomes. Such therapy is not in the apparent effect of the interven-
independently all primary studies. Pri- a standard of care in the critically ill pa- tions across studies) is often found.
mary studies were selected for inclusion tient, whereas TPN (with or without lip- When heterogeneity is present, it weak-
in this overview if the study’s (1) re- ids) is commonly administered to criti- ens inferences that can be made from the
search design was a randomized clinical cally ill patients. For the purpose of this results. The possible sources of varia-
trial; (2) population consisted of surgical review, we excluded studies that used tion in study results include the role of
or critically ill human adult subjects; (3) only amino acid infusions as the inter- chance or differences across studies in
intervention included any form of TPN vention. As the scope of our review was population, intervention, outcome, and
(protein, source of nonprotein energy defined by our research question, we also methods. We developed several hypoth-
with or without lipids) compared with excluded studies that compared TPN eses that might explain heterogeneity of
standard care (oral diet plus intravenous with enteral nutrition or other forms of study results.
fluids); and (4) outcome measures in- TPN. Finally, studies that evaluated the First, we considered that the premor-
cluded complications, length of stay, and impact of TPN only on nutritional out- bid nutritional status of study patients
mortality. comes (ie, nitrogen balance, amino acid was a possible cause of variation in re-
Because studies in which treatment is profile) were not included in this article. sults. Where possible, we grouped the
allocated in any method other than ran- While these end points may explain un- results of studies that included only pa-
domization tend to show larger (and fre- derlying pathophysiology, we consid- tients who were malnourished and com-
quently false-positive) treatment effects ered these as surrogate end points23 and pared them with the results of studies
than do randomized trials,20 we elected we only included articles that reported that included patients who were not mal-
to include only randomized trials in this on clinically important outcomes (mor- nourished at entrance into the study.
review. We defined critically ill patients bidity and mortality). When possible, we used the definition of
as those who would routinely be cared malnourished provided in each study. If
for in a critical care environment. Pa- Methodologic Quality none was provided, we assumed patients
tients undergoing major surgery may of Primary Studies who had greater than 10% weight loss to
not always be cared for in a critical care We assessed the methodologic quality be malnourished.
environment but share similarities in of all selected articles in duplicate, inde- Second, we hypothesized that study
their response to illness, a hypercata- pendently, using a scoring system that results may be related to the methodo-
bolic state characterized by weight loss, we have used previously24 (Table 1). logic quality of the study. We planned a
loss of body fat, and accelerated break- Even in randomized trials, failure to pre- separate analysis comparing the effect
down of body proteins.21 Previous sys- vent foreknowledge of treatment as- of studies with an overall methodologic
tematic reviews have incorporated data signment can lead to an overestimation quality score to those with a score less
from surgical patients and critically ill of treatment effect.25 Accordingly, we than 7 (median score, 7).
patients.15,22 Therefore, we opted to com- scored higher those studies that re- Third, since the practice of providing
bine studies of surgical patients and ported that their randomization schema nutritional support and managing criti-
critically ill patients and to explore any was concealed. Given the difficulties of cally ill patients has evolved over time
2014 JAMA, December 16, 1998—Vol 280, No. 23 Total Parenteral Nutrition in the Critically Ill Patient—Heyland et al
©1998 American Medical Association. All rights reserved.(included studies range from 1976 to with bias and instability associated with To better understand our findings, we
1997), we divided the studies into equal RR estimation in sparse data, we added proceeded to examine our a priori hy-
groups comparing studies published in one half to each cell.35 In the meta-analy- potheses. We compared trials that in-
1988 or earlier with studies published sis, we used maximum likelihood meth- cluded only malnourished patients with
since 1989 (halfway point of the study ods of combining RR across all trials and other trials. No difference in mortality
range). examined the data for evidence of het- existed (Figure 3) for studies of malnour-
Fourth, since some studies adminis- erogeneity within groups.36 The Mantel- ished patients (RR, 1.13; 95% CI, 0.75-
tered amino acids and a carbohydrate Haenszel37 method was used to test the 1.71) or in studies that included ad-
source of energy while others adminis- significance of treatment effect. We used equately nourished patients (RR, 1.00;
tered amino acids, carbohydrates, and a random effects model to estimate the 95% CI, 0.71-1.39; P = .64 for differences
lipids, we separated trials into those that overall RR.38,39 For the test of heteroge- between subgroups). The rate of major
included lipids and those without. We neity across subgroups, we used the t complications was significantly lower
hypothesized that there may be adverse test for the difference between the 2 sub- among malnourished patients receiving
effects caused by lipid use.26 groups. We considered P,.05 to be sta- TPN (RR, 0.52; 95% CI, 0.30-0.91). No
Finally, we speculated that differ- tistically significant. difference existed in complication rates
ences in patient populations (surgical vs among studies of adequately nourished
critically ill) may account for different patients (RR, 1.02; 95% CI, 0.75-1.40).
RESULTS
results. To test this hypothesis, we The difference in complication rates be-
planned a separate analysis comparing Study Identification and Selection tween these subgroups was of border-
studies of surgical patients with studies A total of 153 citations were identified line significance (P = .05).
of critically ill patients. through a computerized bibliographic We compared trials with a methodo-
database search. Our personal files and logic quality score of less than 7 with tri-
Analysis review of reference lists yielded 57 ad- als with a score of 7 or better (Figure 3).
The primary outcome was periopera- ditional articles for consideration. Initial Trials with the higher methods score
tive mortality (death within 30 days of op- eligibility screening resulted in 46 ar- demonstrated no effect of TPN on mor-
eration) or mortality reported at dis- ticles selected for further evaluation. Of tality (RR, 1.17; 95% CI, 0.88-1.56). We
charge from hospital. The secondary these potentially eligible studies, 26 met noted a trend toward a lower mortality
outcome was the rate of major complica- the inclusion criteria. rate in studies with a lower methods
tions. We defined major complications as We reached 100% agreement on the in- score (RR, 0.76; 95% CI, 0.49-1.19). The
pneumonia, intra-abdominal abscess, sep- clusion of articles for this systematic re- difference between these 2 subgroups
sis, line sepsis, myocardial infarction, pul- view. Reasons for excluding relevant ran- was short of conventional levels of sig-
monary emboli, heart failure, stroke, re- domized studies included studies not nificance (P = .12). With respect to com-
nal failure, liver failure, and anastomotic generalizable to critically ill patients40; plication rates, studies with a higher
leak. Minor complications were defined as studies that evaluated different kinds of methods score demonstrated no treat-
wound infection, phlebitis, urinary tract TPN41-43; studies that evaluated amino ac- ment effect (RR, 1.13; 95% CI, 0.86-1.50).
infection, and atelectasis. In 4 studies, the ids only44-47; pseudorandomized studies Studies with a lower methods score
data were not portrayed in a fashion that (not true randomization)48-52; studies du- showed a significant reduction in com-
allowed us to report major complication plicated in other publications34,53,54; stud- plication rates associated with TPN
rates, so we reported total compli- ies not reporting clinically important out- (RR, 0.54; 95% CI, 0.33-0.87). The differ-
cations27-29 and total infectious complica- comes55-57; studies available in abstract ence in complication rates between these
tions.30 Reporting methods of individual form only58; and a study that also random- subgroups was significant (P = .02).
studies did not allow us to disaggregate ized patients to anabolic steroids.59 We next compared trials published in
infectious from noninfectious complica- 1988 or earlier with trials published in
tions. One study31 randomized patients to Impact of TPN on Mortality 1989 or later (Figure 3). Trials published
3 groups (control vs standard TPN vs and Complications Rates in 1988 or earlier demonstrated a trend
TPN with branch-chain amino acids). We There are 26 randomized trials involv- toward a lower mortality rate associated
only included data from the control group ing 2211 patients that compare the use of with TPN (RR, 0.70; 95% CI, 0.44-1.13).
and the standard TPN group. Two other TPN with standard care (usual oral diet Trials published since 1989 demon-
studies randomized patients to 3 groups plus intravenous fluids) in patients under- strated no treatment effect (RR, 1.18;
(control vs TPN without lipids vs TPN going surgery,27-34,60-74 patients with pan- 95% CI, 0.89-1.57). Differences between
with lipids), and we included both experi- creatitis,75 patients in an intensive care these 2 subgroups were short of conven-
mental groups in the analysis.32-34 One unit,76 and patients with severe burns.77 tional levels of statistical significance
study included reports of 2 trials.34 The The details of each study, including the (P = .07). There were significantly fewer
second trial was presumed to include pa- methodologic quality score, are described major complications associated with
tients from the first trial and was there- in Table 2. When the results of these trials TPN reported in studies that were pub-
fore excluded. We also reported on dura- were aggregated, there was no effect on lished in 1988 or earlier (RR, 0.49; 95%
tion of hospital stay, although these data mortality (RR, 1.03; 95% CI, 0.81-1.31) CI, 0.29-0.81), while in studies published
were not aggregated because of infre- (Figure 1). The test for heterogeneity was since 1989 there was no effect of TPN on
quent and variable reporting methods. not significant (P = .59), although a visual complication rates (RR, 1.19; 95% CI,
Agreement between reviewers on in- inspection of Figure 1 suggests that the 0.93-1.53). The P value for the difference
clusion of articles was measured by k treatment effects are variable. between these subgroups was signifi-
with quadratic weights. Twenty-two studies reported major cant (P = .005).
We combined data from all studies to complications in study patients. Aggre- We then compared studies that pro-
estimate the common relative risk of gation of these results revealed a trend vided intravenous lipids as a component
mortality and complications and associ- toward reducing complication rates in of TPN administration with studies that
ated 95% confidence intervals (CIs). We patients receiving TPN (RR, 0.84; 95% did not include lipids. In studies that
summarized the treatment effect using CI, 0.64-1.09) (Figure 2). The test for het- used lipids (RR, 1.03; 95% CI, 0.78-1.36)
risk ratios (RRs). To avoid the problem erogeneity was significant (P = .003). and studies that did not (RR, 0.98; 95%
JAMA, December 16, 1998—Vol 280, No. 23 Total Parenteral Nutrition in the Critically Ill Patient—Heyland et al 2015
©1998 American Medical Association. All rights reserved.Table 2.—Randomized Studies Evaluating Total Parenteral Nutrition (TPN) in Critically Ill Patients*
% of
Methods Malnourished
Source, y Score Patient Population (No.) Patients Intervention
Veterans Affairs,27 1991 10 Thoracoabdominal surgery (395) 100 TPN with lipids 14 d before surgery
Fan et al,28 1989 10 Esophageal cancer surgery (40) 75 TPN with lipids 7-15 d before surgery
Figueras et al,29 1988 7 Gastrointestinal surgery (49) 0 TPN without lipids after surgery
Sandstrom et al,30 1993 10 Major surgery/trauma (300) 22 TPN with lipids after surgery
Reilly et al,31 1990 7 Liver transplant (18) 100 TPN with lipids after surgery
Hwang et al,32 1993a§ 5 Gastric surgery (42) ... TPN with lipids after surgery
Hwang et al,32 1993b§ 5 Gastric surgery (42) ... TPN without lipids after surgery
Muller et al,33 1982 3 Gastrointestinal surgery (125) 60 TPN without lipids 10 d before surgery
Muller et al,34 1986 4 Gastrointestinal surgery (105) ... TPN with lipids 10 d before surgery
Jimenez et al,60 1986 5 Gastrointestinal surgery (75) 100 TPN without lipids after surgery
Brennan et al,61 1994 8 Pancreatic resection (117) ... TPN with lipids after surgery
Askanazi et al,62 1986 3 Radical cystectomy (35) ... TPN with lipids after surgery
Thompson et al,63 1981 4 Gastrointestinal surgery (21) 100 TPN without lipids 5 d before surgery
Fan et al,64 1994 7 Hepatocellular cancer surgery (124) 26 TPN with lipids 7 d before surgery
Abel et al,65 1976 4 Cardiac surgery (44) 100 TPN without lipids after surgery
Bellatone et al,66 1988 6 Gastrointestinal surgery (100) 100 TPN with lipids 7 d before surgery
Smith and Hartemink,67 1988 7 Gastrointestinal surgery (34) 100 TPN without lipids 10 d before surgery
Holter and Fischer,68 1977 5 Gastrointestinal surgery (56) 100 TPN without lipids 3 d before surgery
Meguid et al,69 1988 4 Gastrointestinal surgery (64) 100 TPN with lipids 9 d before surgery
Woolfson and Smith,70 1989 10 Thoracoabdominal surgery (122) ... TPN with lipids after surgery
Von Meyenfeldt et al,71 1992 7 Gastrointestinal surgery (101) 29 TPN with lipids 10 d before surgery
Yamada et al,72 1983 3 Gastric surgery (62) ... TPN with lipids after surgery
Gys et al,73 1990 7 Colorectal surgery (20) 0 TPN with lipids after surgery
Freund et al,74 1979 8 Gastrointestinal surgery (35) 0 TPN without lipids after surgery
Sax et al,75 1987 8 Pancreatitis (54) ... TPN with lipids after admission
Chiarelli et al,76 1996 6 Neurology ICU (24) ... TPN after admission; both groups received EN
(unknown lipids)
Herndon et al,77 1989 7 Burns on .50% of body (49) ... TPN without lipids after admission; both groups
received EN
*Ellipses indicate data not available; EN, enteral nutrition; ICU, intensive care unit.
†Presented as mean ± SD or (range).
‡No range was specified.
§Control group is the same for both criteria.
CI, 0.49-1.95), there was no difference in
mortality. (P value for the difference be- Abel et al,65 1976 Holter and Fischer,68 1977
tween subgroups = .89). Complication Holter and Fischer,68 1977 Freund et al,74 1979
Freund et al,74 1979 Thompson et al,63 1981
rates in studies that used lipids demon- Thompson et al,63 1981 Muller et al,33 1982
strated no effect (RR, 0.96; 95% CI, 0.69- Muller et al,33 1982 Yamada et al,72 1983
Yamada et al,72 1983 Brennan et al,61 1994
1.34). In studies that did not use lipids, Brennan et al,61 1994 Askanazi et al,62 1986
the complication rate was significantly Askanazi et al,62 1986 Muller et al,34 1986
Muller et al,34 1986 Sax et al,75 1987
lower (RR, 0.59; 95% CI, 0.38-0.90). The Sax et al,75 1987 Bellatone et al,66 1988
P value for the difference between these Bellatone et al,66 1988 Smith and Hartemink,67 1988
Meguid et al,69 1988 Fan et al,28 1989
subgroups was just short of significance Smith and Hartemink,67 1988 Figueras et al,29 1988
Fan et al,28 1989 Woolfson and Smith,70 1989
(P = .09). Figueras et al,29 1988 Gys et al,73 1990
Finally, we compared studies of criti- Herndon et al,77 1989 Veterans Affairs,27 1991
Woolfson and Smith,70 1989 Von Meyenfeldt et al,71 1992
cally ill patients with studies of primarily Gys et al,73 1990 Hwang et al,32 1993a
surgical patients. The mortality rate of Reilly et al,31 1990 Hwang et al,32 1993b
Veterans Affairs,27 1991 Fan et al,64 1994
critically ill patients was higher among Von Meyenfeldt et al,71 1992 Jimenez et al,60 1995
those receiving TPN (RR, 1.78; 95% CI, Hwang et al,32 1993a Chiarelli et al,76 1996
Hwang et al,32 1993b
1.11-2.85), while studies of surgical pa- Sandstrom et al,30 1993 Overall Risk Ratio
tients showed no treatment effect (RR, Fan et al,64 1994
0.001 0.01 0.1 1 10 100
Jimenez et al,60 1995
0.91; 95% CI, 0.68-1.21). The difference be- Chiarelli et al,76 1996 TPN TPN
tween these subgroups was statistically Overall Risk Ratio Beneficial Harmful
significant (P = .03). The complication rates 0.001 0.01 0.1 1 10 100 Risk Ratio (Log Scale)
in the studies of critically ill patients (only TPN TPN
2 studies reported complication rates) Beneficial Harmful
Figure 2.—Risk ratios and associated 95% confi-
showed a trend toward an increase in com- Risk Ratio (Log Scale) dence intervals for the effect of total parenteral nu-
plications (RR, 2.40; 95% CI, 0.88-6.58), trition (TPN) on major complications.
while studies of surgical patients were as- Figure 1.—Risk ratios and associated 95% confi-
sociated with lower complication rates dence intervals for effect of total parenteral nutrition
(RR, 0.76; 95% CI, 0.48-1.0). The P value (TPN) on mortality.
for the difference between these sub- the variability in duration of stay and
groups was significant (P = .05). ported median stay and 9 reported variability of reporting methods, we did
Only 14 studies reported the effect of means. In 8 studies, the duration of stay not statistically aggregate these results,
TPN on duration of hospital stay; 5 re- was shorter in the control group. Due to but they are displayed in Table 2.
2016 JAMA, December 16, 1998—Vol 280, No. 23 Total Parenteral Nutrition in the Critically Ill Patient—Heyland et al
©1998 American Medical Association. All rights reserved.studies of malnourished patients, al-
Major Complications, No. (%) Mortality, No. (%) Mean Hospital Stay, d†
though there was no mortality benefit ob-
served. Mortality and complication rates
TPN Control TPN Control TPN Control of studies published in 1988 or earlier and
49/192 (25.5) 50/203 (24.6) 31/231 (13.4) 24/228 (10.5) ... ... studies with a lower methodologic qual-
17/20 (85.0) 15/20 (75) 6/20 (30) 6/20 (30) 15 16‡ ity score showed greater treatment ef-
4/25 (16) 5/24 (20.8) 0/25 (0) 0/24 (0) 13 ± 6 11 ± 3 fect than did later studies or studies with
... ... 12/150 (8) 10/150 (6.7) ... ... a higher methods score. Studies pub-
... ... 0/8 (0) 2/10 (20) 67 ± 29 47 ± 19 lished in 1989 or later and those with a
0/12 (0) 0/16 (0) 0/12 (0) 0/16 (0) ... ...
higher methods score suggest that TPN
0/14 (0) 0/16 (0) 0/14 (0) 0/16 (0) ... ...
may be associated with increased mortal-
11/66 (16.6) 19/59 (32.2) 3/66 (4.5) 11/59 (18.6) ... ...
ity and no effect on complication rates.
17/46 (37) 19/59 (32.2) 10/46 (21.7) 11/59 (18.6) ... ...
The similarity of the subgroup results
6/60 (10) 3/15 (20) 4/60 (6.7) 1/15 (6.7) 9±6 12 ± 8
based on year of publication and methods
27/60 (45) 13/57 (22.8) 4/60 (6.7) 1/57 (1.8) 16 (7-72) 14 (6-88)
score may be partially explained by the
1/22 (4.5) 2/13 (15.4) 0/22 (0) 2/13 (15.4) 17 24‡
fact that 9 of the 13 studies that had a
2/12 (16.7) 1/9 (11.1) 0/12 (0) 0/9 (0) ... ...
methods score of less than 7 also were
22/64 (34.4) 33/60 (55) 5/64 (7.8) 9/60 (15.0) ... ...
... ... 4/20 (20) 3/24 (12.5) 19 ± 6 18 ± 6
published in 1988 or earlier. The differ-
8/54 (14.8) 22/46 (47.8) 1/54 (1.9) 1/46 (2.2) ... ...
ences between these subgroups (meth-
3/17 (17.6) 6/17 (35.3) 1/17 (5.9) 3/17 (17.6) 44 ± 13 38 ± 10
ods score ,7 and $7 and published in 1988
4/30 (13.3) 5/26 (19.2) 2/30 (6.7) 2/6 (7.7) ... ...
or earlier vs later) was significant or close
... ... 1/32 (3.1) 0/34 (0) 10 (6-30) 13 (9-30)
to conventional levels of significance, sug-
6/62 (9.7) 4/60 (6.7) 8/62 (12.9) 8/60 (13.3) 14 (9-64) 13 (9-95) gesting that these subgroup results are
6/51 (11.8) 7/50 (14.0) 2/51 (3.9) 2/50 (4.0) 36 ± 17 32 ± 22 systematically different from each other
0/29 (0) 5/28 (17.9) 0/29 (0) 1/28 (3.6) ... ... and, therefore, may explain a portion of
1/10 (10) 1/10 (10) 0/10 (0) 0/10 (0) ... ... the heterogeneity in the overall results.
0/25 (0) 0/10 (0) 0/25 (0) 0/10 (0) 17 ± 2.3 19 ± 2.9 Indeed, if the results of studies published
4/29 (13.8) 1/29 (3.4) 1/29 (3.4) 1/26 (3.8) 15 ± 4 10 ± 3 in 1989 or later with a methods score of
6/12 (50) 3/12 (25) 3/12 (25) 4/12 (33.3) 37 ± 13 41 ± 23 more than 7 are considered the best esti-
mate of treatment effect, then TPN may
... ... 10/16 (62.5) 6/23 (26.1) ... ... do more harm than good in seriously ill
patients.
There are several reports that demon-
strate that lipids may adversely affect im-
mune status and clinical outcomes.26,79,80
The results of our meta-analysis suggest
talized patients. These studies ranged in that the adverse effects of lipids may ne-
Complications size from 18 to 395 patients with the ma- gate any beneficial effect of nonlipid nu-
Malnourished
P = .05
jority of studies including fewer than 100 tritional supplementation. This is consis-
Nonmalnourished
Quality Scorepatients receiving TPN and these treat- nutritional intervention modify it? Am J Clin Nutr. creased chemotactic and random migration of leu-
ment effects may differ from the results 1988;47:351-356. kocytes during intralipid infusion. Am J Clin Nutr.
3. Buchman AL, Moukarzel AA, Bhuta S, et al. Par- 1979;32:2416-2420.
in surgical patients. The results of stud- enteral nutrition is associated with intestinal mor- 27. Veterans Affairs Total Parenteral Nutrition Co-
ies evaluating the effect of TPN in sur- phologic and functional changes in humans. JPEN J operative Study Group. Perioperative total paren-
gical patients, therefore, may not be gen- Parenter Enteral Nutr. 1995;19:453-460. teral nutrition in surgical patients. N Engl J Med.
eralizable to all types of critically ill pa- 4. Border JR, Hassett J, DaLuca J, et al. The gut 1991;325:525-532.
origin septic states in blunt multiple trauma 28. Fan ST, Lau WY, Wong KK, et al. Preoperative
tients. This leaves a very limited data (TSS=40) in the ICU. Ann Surg. 1980;206:427-448. parenteral nutrition in patients with oesophageal
set on which to base the practice of pro- 5. Deitch EA, Winterton J, Li M, Berg R. The gut as cancer: a prospective randomized clinical trial. Clin
viding TPN to critically ill patients. a portal of entry for bacteremia: role of protein mal- Nutr. 1989;8:23-27.
Because some evidence shows that en- nutrition. Ann Surg. 1987;205:681-690. 29. Figueras J, Rafecas A, Bianchi A, Hernandez F,
6. Lo CW, Walker WA. Changes in the gastrointes- Pi F, Colomer J. Postoperative hypocaloric paren-
teral nutrition is superior to TPN, en- tinal tract during enteral or parenteral feeding. Nutr teral nutrition. Acta Chir Scand. 1988;154:435-438.
teral nutrition may be the preferred Rev. 1989;47:193-198. 30. Sandstrom R, Drott C, Hyltander A, et al. The
method of nutritional support for criti- 7. Hughes CA, Dowling RH. Speed of onset of adap- effect of postoperative intravenous feeding (TPN)
cally ill patients.15 Although the results tive mucosal hypoplasia and hypofunction in the in- on outcome following major surgery evaluated in a
testine of parentally fed rats. Clin Sci. 1980;59:317- randomized study. Ann Surg. 1993;217:185-195.
of our meta-analysis do not support the 327. 31. Reilly J, Mehta R, Teperman L, et al. Nutri-
use of TPN in critically ill patients, pro- 8. Ford WD, Boelhouwer RU, King WW, deVries tional support after liver transplantation: a random-
longed starvation (more than 14 days) is JE, Ross JS, Malt RA. Total parenteral nutrition ized prospective study. JPEN J Parenter Enteral
associated with poor outcomes. In a inhibits intestinal adaptive hyperplasia in young Nutr. 1990;14:386-391.
rats: reversal by feeding. Surgery. 1984;96:527-534. 32. Hwang TL, Mou SC, Chen MF. The importance
study of 300 patients undergoing major 9. Levine GM, Deren JJ, Steiger E, Zinno R. Role of of a source of sufficient protein in postoperative hy-
general surgical procedures, TPN was oral intake in maintenance of gut mass and disac- pocaloric partial parenteral nutritional support.
compared with prolonged glucose ad- charide activity. Gastroenterology. 1974;67:975-983. JPEN J Parenter Enteral Nutr. 1993;17:254-256.
ministration. There was no difference in 10. Schroeder D, Gillanders L, Mahr K, Hill GL. 33. Muller JM, Brenner U, Dienst C, Pichlmaier H.
Effects of immediate postoperative enteral nutri- Preoperative parenteral feeding in patients with
complication rates or mortality.30 How- tion on body composition, muscle function, and gastrointestinal carcinoma. Lancet. 1982;1:68-71.
ever, patients in the control group who wound healing. JPEN J Parenter Enteral Nutr. 34. Muller JM, Keller HW, Brenner U, Walter M,
were unable to take food by mouth for 1981;15:376-383. Holzmuller W. Indications and effects of preoperative
more than 14 days had a much higher 11. Hadfield RJ, Sinclair DG, Houldsworth PE, parenteral nutrition. World J Surg. 1986;10:53-63.
Evans TW. Effects of enteral and parenteral nutri- 35. Naylor AF. Small sample considerations in com-
complication and mortality rate than pa- tion on gut mucosal permeability in the critically ill. bining 2 3 2 tables. Biometrics. 1967;23:349-356.
tients receiving TPN or patients receiv- Am J Respir Crit Care Med. 1995;152:1545-1548. 36. Rothman JR. Modern Epidemiology. Boston,
ing short-term glucose administration. 12. Moore FA, Moore EE, Jones TN, McCroskey Mass: Little Brown & Co Inc; 1986:177-237.
While one can only make weak infer- BL, Petersen VM. TEN versus TPN following ma- 37. Rothman KJ, Boice JD. Epidemiological
jor abdominal trauma–reduced septic morbidity. Analysis With a Programmable Calculator. Wash-
ences from such a post-hoc analysis, it J Trauma. 1989;29:916-923. ington, DC: National Institutes of Health; 1979.
does suggest that patients who cannot 13. Moore FA, Feliciano DV, Andrassy RJ, et al. Paper 79-1649.
tolerate enteral nutrition for more than Early enteral feeding, compared with parenteral, 38. Berlin JA, Laird NM, Sacks HS, et al. A com-
2 weeks may benefit from intravenous reduces septic complications: the results of a meta- parison of statistical methods for combining event
analysis. Ann Surg. 1992;216:172-183. rates from clinical trials. Stat Med. 1989;8:141-151.
supplementation. 14. Kudsk KA, Croce MA, Fabian TC, et al. Enteral 39. Whitehead A, Whitehead J. A general paramet-
In conclusion, while TPN may have a versus parenteral feeding: effects on septic morbid- ric approach to the meta-analysis of randomized tri-
positive effect on nutritional end points ity after blunt and penetrating abdominal trauma. als. Stat Med. 1991;10:1665-1677.
and on even minor complications, the Ann Surg. 1992;215:503-515. 40. Mezey E, Caballeria J, Mitchell MC, Pares A,
15. Heyland DK, Cook DJ, Guyatt GH. Enteral nu- Herlong HF, Rodes J. Effect of parenteral amino
overall results of our meta-analysis fail trition: a critical appraisal of the evidence. Intensive acid supplementation on short-term and long-term
to support a benefit of TPN on mortality Care Med. 1993;19:435-442. outcomes in severe alcoholic hepatitis: a randomized
or major complication rates, particularly 16. ASPEN Board of Directors. Guidelines for the controlled trial. Hepatology. 1991;14:1090-1096.
in critically ill patients. Our a priori sub- use of parenteral and enteral nutrition in adult and 41. Bower RH, Muggia-Sullam M, Vallgren S, et al.
pediatric patients. JPEN J Parenter Enteral Nutr. Branched chain amino acid-enriched solutions in the
group analyses suggest that there may 1993;17(4 suppl):ISA-52SA. septic patient: a randomized, prospective trial. Ann
be a treatment benefit in studies of mal- 17. Cerra FB, Benitez RB, Blackburn GL, et al. Ap- Surg. 1986;203:13-20.
nourished patients. However, treatment plied nutrition in ICU patients: a consensus state- 42. Vente JP, Soeters PB, von Meyefeldt MF, Rou-
benefit was limited to those studies with ment of the American College of Chest Physicians. flart MMJ, vander Linden CJ, Gouma DJ. Prospec-
Chest. 1997;111:769-778. tive randomized double-blind trial of branched chain
a lower methodologic quality score, stud- 18. Koretz R, Nutritional supplementation in the amino acid enriched versus standard parenteral nu-
ies published in 1988 or earlier, studies ICU: how critical is nutrition for the critically ill? trition solutions in traumatized and septic patients.
that did not use lipids, and studies of sur- Am J Respir Crit Care Med. 1995;151:570-573. World J Surg. 1991;15:128-133.
gical patients. Future research needs to 19. Detsky AS, Baker JP, O’Rourke K, Goel V. 43. Brown RO, Buonpane EA, Vehe KL, Hickerson
Perioperative parenteral nutrition: a meta-analy- WL, Luther RW. Comparison of modified amino ac-
define the role of supplemental TPN in sis. Ann Intern Med. 1987;107:195-203. ids and standard amino acids in parenteral nutrition
critically ill patients who cannot tolerate 20. Sacks HS, Chalmers TC, Smith H Jr. Random- support of thermally injured patients. Crit Care
sufficient energy intake and protein via ized versus historical assignment in controlled tri- Med. 1990;18:1096-1101.
the enteral route and the timing of TPN als. N Engl J Med. 1983;309:1353-1361. 44. Garden OJ, Smith A, Harris NWS, et al. The
21. Wilmore DW. Catabolic illness: strategies for effect of isotonic amino acid infusions on serum pro-
in critically ill patients who cannot tol- enhancing recovery. N Engl J Med. 1991;325:695- teins and muscle breakdown following surgery. Br J
erate any enteral intake. Finally, the 702. Surg. 1982;70:79-82.
economics of providing TPN to critically 22. Beale RJ, B’ryg DJ, Bihari D. Clinical effects of 45. Hensle TW. Protein-sparing in cystectomy pa-
ill patients needs to be carefully studied immunonutrition on intensive care patients: a meta- tients. J Urol. 1978;119:355-358.
analysis [abstract]. Intensive Care Med. 1997;23 46. Hogbin BM, Smith AM, Craven AH. An evalu-
to facilitate future practice guidelines. (suppl 1):5128. ation of peripheral essential amino acid infusion
23. Fleming TR, DeMets DL. Surrogate end points following major surgery. JPEN J Parenter Enteral
We would like to thank the anonymous reviewers in clinical trials: are we being misled? Ann Intern Nutr. 1984;8:511-514.
for their helpful comments and Xiangyao Su, PhD, Med. 1996;125:605-613. 47. Doglietto GB, Gallitelli L, Pacelli F, et al. Pro-
for his help with the statistical analysis. 24. Heyland DK, Cook D, King D, Kernerman P, tein-sparing therapy after major abdominal sur-
Bruin-Buisson C. Maximizing oxygen delivery in gery: lack of clinical effects. Ann Surg. 1996;223:357-
References
critically ill patients: a methodologic appraisal of the 362.
1. Warnold I, Lundholm K. Clinical significance of evidence. Crit Care Med. 1996;24:517-524. 48. Preshaw RM, Attisha RP, Hollingsworth WJ,
preoperative nutritional status in 215 non-cancer 25. Chalmers TC, Celano P, Sacks HS, Smith H. Todd JD. Randomized sequential trial of parenteral
patients. Ann Surg. 1984;199:299-305. Bias in treatment assignment in controlled clinical nutrition in healing of colonic anastomoses in man.
2. Dempsey DT, Mullen JL, Buzby GP. The link trials. N Engl J Med. 1988;309:1358-1361. Can J Surg. 1979;22:437-439.
between nutritional status and clinical outcome: can 26. Nordenstrom J, Jarstrand C, Wirniek A. De- 49. Heatley RV, Williams RHP, Lewis MH. Pre-
2018 JAMA, December 16, 1998—Vol 280, No. 23 Total Parenteral Nutrition in the Critically Ill Patient—Heyland et al
©1998 American Medical Association. All rights reserved.operative intravenous feeding: a controlled trial. 60. Jimenez FJJ, Leyba CO, Jimenez LMJ, et al. MMJ, Buil-Maassen MTHJ, Soeters PB. Periopera-
Postgrad Med J. 1979;55:541-545. Study of hypocaloric peripheral parenteral nutri- tive nutritional support: a randomized clinical trial.
50. Moghissi K, Hornshaw J, Teasdale PR, Dawes tion in postoperative patients. Clin Nutr. 1995;14: Clin Nutr. 1992;11:180-186.
EA. Parenteral nutrition in carcinoma of the esopha- 88-96. 72. Yamada N, Koyama H, Hioki K, Yamada T, Ya-
gus treated by surgery: nitrogen balance and clini- 61. Brennan MF, Pesters PWT, Posner M, et al. A mamoto M. Effect of postoperative total parenteral
cal studies. Br J Surg. 1977;64:125-128. prospective randomized trial of total parenteral nu- nutrition (TPN) as an adjunct to gastrectomy for
51. Collins JP, Oxby CB, Hill GL. Intravenous trition after major pancreatic resection for malig- advanced gastric carcinoma. Br J Surg. 1983;70:267-
amino acids and intravenous hyperalimentation as nancy. Ann Surg. 1994;220:436-444. 274.
protein-sparing therapy after major surgery. Lan- 62. Askanzi J, Starker PM, Olsson C, et al. Effect of 73. Gys T, Peeters R, Hubens A. The value of short-
cet. 1979;1:788-792. immediate postoperative nutritional support on term peripheral parenteral nutrition after colorec-
52. Jensen S. Clinical effects of enteral and paren- length of hospitalization. Ann Surg. 1986;203:236- tal surgery: a comparative study with conventional
teral nutrition preceding cancer surgery. Med On- 239. postoperative intravenous fluid. Acta Chir Belg.
col Tumor Pharmacother. 1985;2:225-229. 63. Thompson BR, Julian TB, Scrumple JW. Perio- 1990;90:234-239.
53. Bellatone R, Doglietto G, Bossola M, et al. Pre- perative total parenteral nutrition in patients with 74. Freund H, Hoover HC, Atamian S, Fischer JE.
operative parenteral nutrition of malnourished sur- gastrointestinal cancer. J Surg Res. 1981;30:497- Infusion of the branched chain amino acids in post-
gical patients. Acta Chir Scand. 1988;22:249-251. 500. operative patients. Ann Surg. 1979;190:18-23.
54. Herndon DN, Stein MD, Rutan TC, Abston S, 64. Fan S, Lo C, Thai ECS, et al. Perioperative nu- 75. Sax HC, Warner BW, Talamini MA, et al. Early
Lnares H. Failure of TPN supplementation to im- tritional support in patients undergoing hepatecto- total parenteral nutrition in acute pancreatitis:
prove liver function, immunity and mortality in my for hepatocellular carcinoma. N Engl J Med. lack of beneficial effects. Am J Surg. 1987;153:
thermally injured patients. J Trauma. 1987;27: 1994;331:1547-1552. 117-123.
195-204. 65. Abel RM, Fischer JE, Buckley MJ, et al. Mal- 76. Chiarelli AG, Ferrarello S, Piccioli A, et al. Total
55. Culebras-Fernandez JM, de la Hoz Riesco M, nutrition in cardiac surgical patients. Arch Surg. parenteral nutrition versus mixed enteral and par-
Garcia CV, et al. Improvement of the nutritional 1976;111:45-50. enteral nutrition in patients in an intensive care unit.
condition with hypocaloric peripheral parenteral nu- 66. Bellatone R, Doglietto GB, Bossola M, et al. Pre- Minerva Anestesiol. 1996;62:1-7.
trition (HPPN) in the immediate postoperative pe- operative parenteral nutrition in the high risk sur- 77. Herndon DN, Barrow RE, Stein M, et al. In-
riod of elective abdominal surgery. Infusionthera- gical patient. JPEN J Parenter Enteral Nutr. 1988; creased mortality with intravenous supplement
pie. 1987;14:202-208. 12:195-197. feeding in severely burned patients. J Burn Care
56. Bonau RA, Ang SD, Jeevanandam M, Daly JM. 67. Smith RC, Hartemink RH. Improvement of nu- Rehabil. 1989;10:309-313.
High-branched chain amino acid solutions: relation- tritional measures during preoperative parenteral 78. Oxman AD, Guyatt GM. Guidelines for reading
ship of composition to efficacy. JPEN J Parenter nutrition in patients selected by the prognostic nu- literature reviews. CMAJ. 1988;138:697-703.
Enteral Nutr. 1984;8:622-627. tritional index: a randomized controlled trial. JPEN 79. Seidener DL, Mascioli EA, Istfan NW, et al.
57. Neuvonen P, Salo M, Perttila J, Havia T. Lack of J Parenter Enteral Nutr. 1988;12:587-591. Effects of long-chain triglyceride emulsions on re-
modulation of postoperative immunosuppression by 68. Holter AR, Fischer JE. The effects of periop- ticuloendothelial system function in humans. JPEN
isotonic amino acid infusion. JPEN J Parenter En- erative hyperalimentation on complications in pa- J Parenter Enteral Nutr. 1989;13:614-619.
teral Nutr. 1986;10:160-165. tients with carcinoma and weight loss. J Surg Res. 80. Freeman J, Goldmann DA, Smith NE, Sidebot-
58. Simms JM, Oliver E, Smith JAR. A study of 1977;23:31-34. tom DG, Epstein ME, Platt R. Association of intra-
total parenteral nutrition (TPN) in major gastric 69. Meguid MM, Curtas MS, Meguid V, et al. Ef- venous lipid emulsion and coagulase-negative
and esophageal resection for neoplasia. JPEN J fects of pre-operative TPN on surgical risk: prelimi- staphylococcal bacteremia in neonatal intensive care
Parenter Enteral Nutr. 1980;4:422. nary status report. Br J Clin Pract. 1988;42(suppl): units. N Engl J Med. 1990;323:301-308.
59. Hansell DT, Davies JWL, Shenkin A, Garden 53-58. 81. Basttistella ED, Widergr JT, Anderson JT,
OJ, Burns HJG, Carter DC. The effects of an ana- 70. Woolfson AMJ, Smith JAR. Elective nutritional Siepler JK, Weber JC, MacColl K. A prospective,
bolic steroid and peripherally administered intra- support after major surgery: a prospective random- randomized trial of intravenous fat emulsion admin-
venous nutrition in the early postoperative period. ized trial. Clin Nutr. 1989;8:15-21. istration in trauma victims requiring total paren-
JPEN J Parenter Enteral Nutr. 1989;13:349-358. 71. Von Meyenfeldt MF, Meijerink WJHJ, Rouflart teral nutrition. J Trauma. 1997;43:52-58.
JAMA, December 16, 1998—Vol 280, No. 23 Total Parenteral Nutrition in the Critically Ill Patient—Heyland et al 2019
©1998 American Medical Association. All rights reserved.You can also read
