Filetype PDF | Posted on 04 Jan 2023 | 2 years ago
Authentication
digital resources
145x Filetype PDF File size 0.16 MB Source: jamanetwork.com
Caring for the Critically Ill Patient
Total Parenteral Nutrition
in the Critically Ill Patient
AMeta-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. illnessandresultingingreaterinfectious
Objective.—To examine the relationship between TPN and complication and morbidity.3-5 As a consequence, nutri-
mortality rates in critically ill patients. tional supporthasbecomeastandardof
DataSources.—ComputerizedsearchofpublishedresearchonMEDLINEfrom care for hospitalized patients.
1980 to 1998, personal files, and review of relevant reference lists. Because intestinal stimulation from
Study Selection.—We reviewed 210 titles, abstracts, and papers. Primary luminalnutrientshelpsmaintaingastro-
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-
complicationandmortalityrates.WeexcludedstudiescomparingTPNwithenteral tion (TPN). Compared with TPN, ran-
nutrition. domized trials of critically ill patients
DataExtraction.—Relevantdatawereabstractedonthemethodologyandout- have demonstrated that enteral nutri-
comes of primary studies. Data were abstracted in duplicate, independently. tion administered within the first 24
DataSynthesis.—Therewere26randomizedtrialsof2211patientscomparing hoursofadmissiontotheintensivecare
the use of TPN with standard care (usual oral diet plus intravenous dextrose) in unit(ICU)resultsinbetterwoundheal-
10
surgical and critically ill patients. When the results of these trials were aggregated, ing, adecreaseingastrointestinaltract
11
mucosalpermeability, andlowerinfec-
TPNhadnoeffectonmortality (risk ratio [RR], 1.03; 95% confidence interval [CI], 12-14
0.81-1.31). Patients who received TPN tended to have a lower complication rate, tion rates. Where possible, enteral
feeding is preferred to parenteral feed-
butthisresultwasnotstatisticallysignificant(RR,0.84;95%CI,0.64-1.09).Weex- 15
ing. However, some patients with an
aminedseveralapriori 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 erateenteralfeedsordonotreceivesuf-
mortality when compared with studies of nonmalnourished patients. Studies pub- ficientintakeenterallyororallytomeet
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
score demonstrated a significant treatment effect. Complication rates were lower supplement or as the sole source of nu-
trition in these patients16,17
in studies that did not use lipids; however, there was no difference in mortality rates ; however,
previousevidencesupportingthisprac-
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- 19
tality rates compared with studies of surgical patients. studies were reviewed in 1987, addi-
tional randomizedtrialshavebeenpub-
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,andaggregatestatisticallystudies
lation, use of lipids, methodological quality, and year of publication. evaluatingtheeffectofTPNincritically
JAMA.1998;280:2013-2019 ill patients.
FromtheDepartmentsofMedicine(DrsHeylandand MALNUTRITIONamonghospitalized METHODS
MacDonald)andSurgery(DrDrover),Queen’sUniver- patients has been associated with in- Search Strategy
sityandNutritionalServices,KingstonGeneralHospital creased morbidity, prolonged hospital Weconductedacomputerizedbiblio-
(Ms Keefe), Kingston, Ontario. graphicsearchofMEDLINE(including
DrHeylandisaCareerScientistoftheOntarioMinis- stay, and increased costs to the health
try of Health. care system.1,2 Several studies have pre-MEDLINE) for studies from 1980
Reprints:DarenK.Heyland,MD,FRCPC,MSc,Angada documentedthat“bowelrest”isassoci- to April 1998 to locate all relevant ar-
3, KingstonGeneralHospital,76StuartSt,Kingston,On- ated with a disruption of the mucosal ticles.Thetermsrandomizedcontrolled
tario, Canada K7L 2V7 (e-mail: dkh@post.queensu.ca).
JAMA, December 16, 1998—Vol 280, No. 23 Total Parenteral Nutrition in the Critically Ill Patient—Heyland et al 2013
©1998AmericanMedicalAssociation. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 01/04/2023
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
012
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 groupswereequalatbaseline,ifcointer-
Analysis Other . . . Intention to treat ventions were adequately described,
Patient selection Selected patients or Consecutive eligible ... whether objective definitions of infec-
unable to tell patients tious outcomeswereused,andwhether
Comparability of groups No or not sure Yes . . . all patientswereproperlyaccountedfor
at baseline
Extent of follow-up ,100% 100% . . . intheanalysis(intention-to-treatanaly-
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 Twoofus(D.K.H.andS.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-
†Theextenttowhichantibiotics,enteralnutrition,ventilation,oxygen,andtransfusionswereappliedequallyacross agreementbyconsensus.Notallstudies
groups. reportedcomplicationrates.Somestud-
trial,doubleblindmethod,clinicaltrial, differences that might exist between ies reported total complications per
placebo, and comparative study were thesepatientsinthesubgroupanalysis. group but not on a per-patient basis.
combined with explode parenteral nu- Weexcludedstudiesofpediatricorneo- Whendataweremissing,unclear,ornot
trition, total. Citations were limited to natal patients. reported on a per-patient basis, we at-
English-language studies reporting on Weincluded only studies that evalu- temptedtocontacttheprimaryinvesti-
adult patients. Reference lists of rel- atedtheuseofsupplementalTPNinpa- gators and requested them to provide
evant review articles and personal files tientsreceivingenteralfeedsorstudies further information if the article had
werealsosearched. evaluating the use of TPN in patients beenpublishedinthelast5years.
Study Selection Criteria whowerenotreceivingTPNorenteral Prior Hypotheses Regarding Sources
nutrition.Thereareseveralrandomized of Heterogeneity
Initially, 2 of us (D.K.H. and S.M.) trials of surgical patients that examine
screenedallcitationsandclassifiedthem theeffectofaminoacidinfusion(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 onclinicaloutcomes.Suchtherapyisnot in the apparent effect of the interven-
independently all primary studies. Pri- astandardofcareinthecriticallyillpa- tions across studies) is often found.
marystudieswereselectedforinclusion tient,whereasTPN(withorwithoutlip- Whenheterogeneityispresent,itweak-
in this overview if the study’s (1) re- ids) is commonly administered to criti- ensinferencesthatcanbemadefromthe
searchdesignwasarandomizedclinical callyill patients. Forthepurposeofthis results. The possible sources of varia-
trial; (2) populationconsistedofsurgical review, we excluded studies that used tion in study results include the role of
orcritically 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.Asthescopeofourreviewwas population, intervention, outcome, and
(protein, source of nonprotein energy definedbyourresearchquestion,wealso methods.Wedevelopedseveralhypoth-
with or without lipids) compared with excluded studies that compared TPN esesthatmightexplainheterogeneityof
standardcare(oraldietplusintravenous with enteral nutrition or other forms of studyresults.
fluids); and (4) outcome measures in- TPN.Finally,studiesthatevaluatedthe First,weconsideredthatthepremor-
cludedcomplications,lengthofstay,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-
Becausestudiesinwhichtreatmentis profile)werenotincludedinthisarticle. sults. Where possible, we grouped the
allocated in any method other than ran- Whiletheseendpointsmayexplainun- results of studies that included only pa-
domizationtendtoshowlarger(andfre- derlying pathophysiology, we consid- tientswhoweremalnourishedandcom-
quentlyfalse-positive)treatmenteffects eredtheseassurrogateendpoints23and pared them with the results of studies
than do randomized trials,20 we elected weonly included articles that reported thatincludedpatientswhowerenotmal-
to include only randomized trials in this on clinically important outcomes (mor- nourished at entrance into the study.
review.Wedefinedcriticallyillpatients bidity and mortality). Whenpossible,weusedthedefinitionof
as those who would routinely be cared Methodologic Quality malnourishedprovidedineachstudy.If
for in a critical care environment. Pa- of Primary Studies nonewasprovided,weassumedpatients
tients undergoing major surgery may whohadgreaterthan10%weightlossto
notalwaysbecaredforinacriticalcare Weassessedthemethodologicquality bemalnourished.
environment but share similarities in ofallselectedarticlesinduplicate,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-
bolicstatecharacterizedbyweightloss, we have used previously24 (Table 1). logic quality of the study. We planned a
loss of body fat, and accelerated break- Eveninrandomizedtrials,failuretopre- separate analysis comparing the effect
down of body proteins.21 Previous sys- vent foreknowledge of treatment as- of studies with an overall methodologic
tematicreviewshaveincorporateddata signmentcanleadtoanoverestimation quality score to those with a score less
from surgical patients and critically ill of treatment effect.25 Accordingly, we than7(medianscore,7).
15,22 Third, since the practice of providing
patients. Therefore,weoptedtocom- scored higher those studies that re-
bine studies of surgical patients and portedthattheirrandomizationschema 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,December16,1998—Vol280,No.23 Total Parenteral Nutrition in the Critically Ill Patient—Heyland et al
©1998AmericanMedicalAssociation. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 01/04/2023
(included studies range from 1976 to withbiasandinstabilityassociatedwith Tobetterunderstandourfindings,we
1997), we divided the studies into equal RRestimationinsparsedata,weadded proceeded to examine our a priori hy-
groups comparing studies published in onehalftoeachcell.35Inthemeta-analy- potheses. We compared trials that in-
1988 or earlier with studies published sis, we used maximum likelihood meth- cludedonlymalnourishedpatientswith
since 1989 (halfway point of the study odsofcombiningRRacrossalltrialsand other trials. No difference in mortality
range). examined the data for evidence of het- existed(Figure3)forstudiesofmalnour-
36
Fourth, since some studies adminis- erogeneitywithingroups. TheMantel- 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- significanceoftreatmenteffect.Weused 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=.64fordifferences
lipids,weseparatedtrialsintothosethat overallRR.38,39Forthetestofheteroge- between subgroups). The rate of major
included lipids and those without. We neity across subgroups, we used the t complications was significantly lower
hypothesizedthattheremaybeadverse testforthedifferencebetweenthe2sub- amongmalnourishedpatientsreceiving
26
effects caused by lipid use. groups.WeconsideredP,.05tobesta- TPN(RR, 0.52; 95% CI, 0.30-0.91). No
Finally, we speculated that differ- tistically significant. difference existed in complication rates
encesinpatientpopulations(surgicalvs amongstudies of adequately nourished
critically ill) may account for different RESULTS patients (RR, 1.02; 95% CI, 0.75-1.40).
results. To test this hypothesis, we Study Identification and Selection Thedifferenceincomplicationratesbe-
planned a separate analysis comparing tween these subgroups was of border-
studiesofsurgicalpatientswithstudies Atotalof153citationswereidentified line significance (P=.05).
of critically ill patients. through a computerized bibliographic Wecomparedtrials with a methodo-
Analysis database search. Our personal files and logicqualityscoreoflessthan7withtri-
review of reference lists yielded 57 ad- alswithascoreof7orbetter(Figure3).
The primary outcome was periopera- ditionalarticlesforconsideration.Initial Trials with the higher methods score
tivemortality(deathwithin30daysofop- eligibility screening resulted in 46 ar- demonstratednoeffectofTPNonmor-
eration) or mortality reported at dis- ticles selectedforfurtherevaluation.Of tality (RR, 1.17; 95% CI, 0.88-1.56). We
charge from hospital. The secondary thesepotentiallyeligiblestudies,26met noted a trend toward a lower mortality
outcomewastherateofmajorcomplica- the inclusion criteria. rate in studies with a lower methods
tions.Wedefinedmajorcomplicationsas Wereached100%agreementonthein- score (RR, 0.76; 95% CI, 0.49-1.19). The
pneumonia,intra-abdominalabscess,sep- clusion of articles for this systematic re- difference between these 2 subgroups
sis,linesepsis,myocardialinfarction,pul- view.Reasonsforexcludingrelevantran- was short of conventional levels of sig-
monaryemboli,heartfailure, stroke, re- domized studies included studies not nificance (P=.12). Withrespecttocom-
generalizable to critically ill patients40
nalfailure, liver failure, and anastomotic ; plication rates, studies with a higher
leak.Minorcomplicationsweredefinedas studies that evaluated different kinds of methods score demonstrated no treat-
woundinfection, phlebitis, urinary tract TPN41-43;studiesthatevaluatedaminoac- menteffect(RR,1.13;95%CI,0.86-1.50).
infection,andatelectasis.In4studies,the ids only44-47; pseudorandomized studies Studies with a lower methods score
datawerenotportrayedinafashionthat (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- iesnotreportingclinicallyimportantout- (RR,0.54;95%CI,0.33-0.87).Thediffer-
27-29 55-57 enceincomplicationratesbetweenthese
cations and total infectious complica- comes ; studies available in abstract
tions.30 Reporting methods of individual formonly58;andastudythatalsorandom- subgroupswassignificant (P=.02).
studies did not allow us to disaggregate ized patients to anabolic steroids.59 Wenextcomparedtrialspublishedin
infectious from noninfectious complica- 1988 or earlier with trials published in
tions.Onestudy31randomizedpatientsto Impact of TPN on Mortality 1989orlater(Figure3).Trialspublished
3 groups (control vs standard TPN vs and Complications Rates in 1988 or earlier demonstrated a trend
TPNwithbranch-chainaminoacids).We Thereare26randomizedtrialsinvolv- towardalowermortalityrateassociated
onlyincludeddatafromthecontrolgroup ing2211patientsthatcomparetheuseof withTPN(RR,0.70;95%CI,0.44-1.13).
andthestandardTPNgroup.Twoother TPNwithstandard care (usual oral diet Trials published since 1989 demon-
studies randomized patients to 3 groups plusintravenousfluids)inpatientsunder- 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).Differencesbetween
withlipids),andweincludedbothexperi- creatitis,75 patients in an intensive care these2subgroupswereshortofconven-
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).Thereweresignificantlyfewer
secondtrialwaspresumedtoincludepa- methodologicqualityscore,aredescribed major complications associated with
tients from the first trial and was there- inTable2.Whentheresultsofthesetrials TPNreportedinstudiesthatwerepub-
foreexcluded.Wealsoreportedondura- wereaggregated, there was no effect on lished in 1988 or earlier (RR, 0.49; 95%
tionofhospitalstay,althoughthesedata mortality (RR, 1.03; 95% CI, 0.81-1.31) CI,0.29-0.81),whileinstudiespublished
were not aggregated because of infre- (Figure1).Thetestforheterogeneitywas since1989therewasnoeffectofTPNon
quentandvariablereportingmethods. notsignificant(P=.59),althoughavisual complication rates (RR, 1.19; 95% CI,
Agreementbetweenreviewersonin- inspection of Figure 1 suggests that the 0.93-1.53).ThePvalueforthedifference
clusion of articles was measured by k treatmenteffects are variable. between these subgroups was signifi-
withquadratic weights. Twenty-two studies reported major cant (P=.005).
Wecombineddatafromallstudiesto complications in study patients. Aggre- Wethen compared studies that pro-
estimate the common relative risk of gation of these results revealed a trend videdintravenouslipidsasacomponent
mortality and complications and associ- toward reducing complication rates in ofTPNadministrationwithstudiesthat
ated95%confidenceintervals(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)(Figure2).Thetestforhet- usedlipids(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
©1998AmericanMedicalAssociation. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 01/04/2023
Table 2.—Randomized Studies Evaluating Total Parenteral Nutrition (TPN) in Critically Ill Patients*
%of
Methods Malnourished
Source, y Score Patient Population (No.) Patients Intervention
27
Veterans Affairs, 1991 10 Thoracoabdominal surgery (395) 100 TPNwith lipids 14 d before surgery
28
Fan et al, 1989 10 Esophageal cancer surgery (40) 75 TPNwith lipids 7-15 d before surgery
29
Figueras et al, 1988 7 Gastrointestinal surgery (49) 0 TPNwithout lipids after surgery
30
Sandstrom et al, 1993 10 Major surgery/trauma (300) 22 TPNwith lipids after surgery
31
Reilly et al, 1990 7 Liver transplant (18) 100 TPNwith lipids after surgery
32
Hwang et al, 1993a§ 5 Gastric surgery (42) . . . TPNwith lipids after surgery
32
Hwang et al, 1993b§ 5 Gastric surgery (42) . . . TPNwithout lipids after surgery
33
Muller et al, 1982 3 Gastrointestinal surgery (125) 60 TPNwithout lipids 10 d before surgery
34
Muller et al, 1986 4 Gastrointestinal surgery (105) . . . TPNwith lipids 10 d before surgery
60
Jimenez et al, 1986 5 Gastrointestinal surgery (75) 100 TPNwithout lipids after surgery
61
Brennan et al, 1994 8 Pancreatic resection (117) . . . TPNwith lipids after surgery
62
Askanazi et al, 1986 3 Radical cystectomy (35) . . . TPNwith lipids after surgery
63
Thompson et al, 1981 4 Gastrointestinal surgery (21) 100 TPNwithout lipids 5 d before surgery
64
Fan et al, 1994 7 Hepatocellular cancer surgery (124) 26 TPNwith lipids 7 d before surgery
65
Abel et al, 1976 4 Cardiac surgery (44) 100 TPNwithout lipids after surgery
66
Bellatone et al, 1988 6 Gastrointestinal surgery (100) 100 TPNwith lipids 7 d before surgery
67
Smith and Hartemink, 1988 7 Gastrointestinal surgery (34) 100 TPNwithout lipids 10 d before surgery
Holter and Fischer,68 1977 5 Gastrointestinal surgery (56) 100 TPNwithout lipids 3 d before surgery
69
Meguid et al, 1988 4 Gastrointestinal surgery (64) 100 TPNwith lipids 9 d before surgery
70
Woolfson and Smith, 1989 10 Thoracoabdominal surgery (122) . . . TPNwith lipids after surgery
71
Von Meyenfeldt et al, 1992 7 Gastrointestinal surgery (101) 29 TPNwith lipids 10 d before surgery
72
Yamada et al, 1983 3 Gastric surgery (62) . . . TPNwith lipids after surgery
73
Gys et al, 1990 7 Colorectal surgery (20) 0 TPNwith lipids after surgery
74
Freund et al, 1979 8 Gastrointestinal surgery (35) 0 TPNwithout lipids after surgery
75
Sax et al, 1987 8 Pancreatitis (54) . . . TPNwith lipids after admission
76
Chiarelli et al, 1996 6 Neurology ICU (24) . . . TPNafter admission; both groups received EN
(unknown lipids)
77 1989 7 Burns on .50% of body (49) . . . TPNwithout lipids after admission; both groups
Herndon et al,
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),therewasnodifferencein
mortality.(Pvalueforthedifferencebe- 65 Holter and Fischer,68 1977
Abel et al, 1976
68 74
tween subgroups=.89). Complication Holter and Fischer, 1977 Freund et al, 1979
74 63
rates in studies that used lipids demon- Freund et al, 1979 Thompson et al, 1981
63 33
Thompson et al, 1981 Muller et al, 1982
33 72
stratednoeffect(RR,0.96;95%CI,0.69- Muller et al, 1982 Yamada et al, 1983
72 Brennan et al,61 1994
Yamada et al, 1983
1.34). In studies that did not use lipids, 61 62
Brennan et al, 1994 Askanazi et al, 1986
62 34
the complication rate was significantly Askanazi et al, 1986 Muller et al, 1986
34 75
Muller et al, 1986 Sax et al, 1987
lower(RR,0.59;95%CI,0.38-0.90).The 75 66
Sax et al, 1987 Bellatone et al, 1988
66 67
Pvalueforthedifferencebetweenthese Bellatone et al, 1988 Smith and Hartemink, 1988
69 28
Meguid et al, 1988 Fan et al, 1989
subgroupswasjustshortofsignificance 67 29
Smith and Hartemink, 1988 Figueras et al, 1988
28 70
(P=.09). Fan et al, 1989 Woolfson and Smith, 1989
29 73
Figueras et al, 1988 Gys et al, 1990
77 27
Finally, we compared studies of criti- Herndon et al, 1989 Veterans Affairs, 1991
70 71
callyillpatientswithstudiesofprimarily Woolfson and Smith, 1989 Von Meyenfeldt et al, 1992
73 32
Gys et al, 1990 Hwang et al, 1993a
31 32
surgical patients. The mortality rate of Reilly et al, 1990 Hwang et al, 1993b
27 64
critically ill patients was higher among Veterans Affairs, 1991 Fan et al, 1994
71 60
Von Meyenfeldt et al, 1992 Jimenez et al, 1995
32 76
those receiving TPN (RR, 1.78; 95% CI, Hwang et al, 1993a Chiarelli et al, 1996
32
1.11-2.85), while studies of surgical pa- Hwang et al, 1993b
30 Overall Risk Ratio
Sandstrom et al, 1993
64
tients showed no treatment effect (RR, Fan et al, 1994 0.001 0.01 0.1 1 10 100
Jimenez et al,60 1995
0.91;95%CI,0.68-1.21).Thedifferencebe- 76
Chiarelli et al, 1996 TPN TPN
tweenthese subgroups was statistically Overall Risk Ratio Beneficial Harmful
significant(P=.03).Thecomplicationrates 0.001 0.01 0.1 1 10 100 Risk Ratio (Log Scale)
inthestudiesofcriticallyillpatients(only TPN TPN
2 studies reported complication rates) Beneficial Harmful Figure 2.—Risk ratios and associated 95% confi-
showedatrendtowardanincreaseincom- 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.
whilestudiesofsurgicalpatientswereas- Figure 1.—Risk ratios and associated 95% confi-
sociated with lower complication rates denceintervalsforeffectoftotalparenteralnutrition
(RR,0.76;95%CI,0.48-1.0).ThePvalue (TPN) on mortality.
for the difference between these sub- the variability in duration of stay and
groupswassignificant(P=.05). ported median stay and 9 reported variabilityofreportingmethods,wedid
Only14studiesreportedtheeffectof means.In8studies,thedurationofstay notstatisticallyaggregatetheseresults,
TPNonduration of hospital stay; 5 re- wasshorterinthecontrolgroup.Dueto buttheyaredisplayedinTable2.
2016 JAMA,December16,1998—Vol280,No.23 Total Parenteral Nutrition in the Critically Ill Patient—Heyland et al
©1998AmericanMedicalAssociation. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 01/04/2023
no reviews yet
Please Login to review.
