Early enteral nutrition in critically ill patients: authors’ response

Gordon S. DoigFiona Simpson

Early enteral nutritionin critically ill patients:authors’ response

Accepted: 31 January 2010Published online: 6 March 2010� Copyright jointly held by Springer andESICM 2010

This reply refers to the comment availableat: doi:10.1007/s00134-010-1788-6.

Dear Dr Koretz: Thank you for yourletter regarding our published meta-analysis on the topic of early enteralnutrition (EN) [1]. I trust this briefresponse will clear up some of yourmisunderstandings.The definition of ‘methodologicallyunsound’ trials that we used to iden-tify studies for exclusion was basedon criteria established by the BMJGroup publication ‘ClinicalEvidence’ (http://clinicalevidence.bmj.com/ceweb/about/appraisal.jsp,Accessed 27 Jan 2010). We note thatunsound studies were not included inour systematic review and meta-analysis. Where the methodology ofincluded studies was ‘questionable’,we use terminology that clearly indi-cates to the reader the presence ofuncertainty. Although we cannot (anddo not) claim the included studies are‘high-quality’, because studies withmajor flaws were excluded, theincluded studies do constitute the‘best available evidence’ on thistopic.

With regards to the sentencedescribing the patient allocation pro-cess used in the Chiarelli trial [2], wenote that we have explicitly describedthe allocation process as ‘unclear’.We agree that it is possible the

authors did not maintain allocationconcealment, however, it is also pos-sible that the authors selected theterm ‘case–control’ to refer to the useof blocking to achieve balance. If thisis the case, allocation concealment islikely to have been preserved. Indeed,all primary authors of the Chiarellipaper appear to be native Italianspeakers, so the awkward wording inthe sentence you refer to likely rep-resents a harmless translation artifact.Until such time that additional infor-mation is received from the authors ofthe Chiarelli paper, the use of theterm ‘unclear’ to describe this sen-tence is informative and does notmake any assumptions. Using themethodology outlined by the BMJGroup journal ‘Clinical Evidence’, itis appropriate to include ‘unclear’papers.

We have no concerns regardingthe Pupelis trial [3]. Your calcula-tions are based on the theoreticalframework commonly referred to as‘sampling with replacement’ [4].The underlying assumptions ofsampling with replacement are notapplicable to a randomised con-trolled trial (RCT) that uses anyform of blocking. With an RCT thatuses any form of blocking, thetheoretical framework known assampling without replacementapplies. In essence, if Pupelis et al.had used sequentially numberedopaque sealed envelopes to allocatepatients to treatment or controlgroups, they would have created andshuffled 30 ‘standard care’ enve-lopes with 30 ‘early EN’ envelopes.Use of a block size of 60 in a 60patient trial guarantees equal bal-ance between standard care (30) andstudy treatment (30) at the end ofthe trial, after 60 patients have beenrandomised, but if the trial is stop-ped early, balance is not guaranteed[5]. Thus, the fact that we seeimbalance in an interim report [6],and balance at the end of the trial isunremarkable.

As we report in our manuscript,parenteral nutrition (PN) use differsbetween trials [1]. We have reviewedthe literature and are familiar withthe dogma and evidence surroundingthe use of PN in the critically illpatient [7]. In the context of a meta-analysis on the use of early EN, ifthe concomitant use of PN intro-duces any form of ‘confounding’, wecould expect to see a quantifiabledifference in the treatment effectattributable to early EN in the RCTswhere PN is not used, compared tothe treatment effect observed inRCTs where PN is used. In otherwords, we would expect some formof observable treatment interaction.The I2 test was specifically designedto detect treatment interactions in ameta-analysis. Since our I2 testreturned the value of zero, we are ascertain as we can be that no treat-ment interactions exist. Thus, there isno observable evidence that PN useis a ‘confounder’.

In conclusion, we wish toacknowledge the imperfections of theevidence we were able to find in orderto address the question framed in ourmeta-analysis. We agree that moreand better evidence is needed in thisfield. In the meantime, what are we todo for our patients?

In our systematic, comprehensiveand unbiased attempts to findevidence addressing this question, itis important to understand that wewere unable to find ANY objectiveevidence supporting patient benefitsarising from the recommendation towithhold nutritional support for up to7 days [8]. As we addressed in ourDiscussion [1], we found scantevidence to support the widely heldconcept that 48 h is the appropriatetarget [9]. Although not perfect byany standards, the best availableevidence that is free from major flawsis reported objectively in our meta-analysis [1]. It does support therecommendation that critically illpatients may benefit most when

Intensive Care Med (2010) 36:1089–1090DOI 10.1007/s00134-010-1792-x CORRESPONDENCE

EN is provided within a 24 h window[10, 11].

References

1. Doig GS, Heighes PT, Simpson F,Sweetman EA, Davies AR (2009)Enteral nutrition within 24 h of ICUadmission significantly reducesmortality: a meta-analysis of RCTs.Intensive Care Med 35:2018–2027

2. Chiarelli A, Enzi G, Casadei A, BaggioB, Valerio A, Mazzoleni F (1990) Veryearly nutrition supplementation inburned patients. Am J Clin Nutr51:1035–1039

3. Pupelis G, Selga G, Austrums E,Kaminski A (2001) Jejunal feeding,even when instituted late, improvesoutcomes in patients with severepancreatitis and peritonitis. Nutrition17:91–94

4. Statistical Methods (1989) In: SnedecorGW, Cochran WG (eds) 8th Edn. IowaState University Press, Ames, Iowa

5. Doig GS, Simpson F (2005)Randomization and allocationconcealment: a practical guide forresearchers. J Crit Care 20:187–193

6. Pupelis G, Austrums E, Jansone A,Sprucs R, Wehbi H (2000) Randomisedtrial of safety and efficacy ofpostoperative enteral feeding in patientswith severe pancreatitis: preliminaryreport. Eur J Surg 166:383–387

7. Simpson F, Doig GS (2005) Parenteralvs. enteral nutrition in the critically illpatient: a meta-analysis of trials usingthe intention to treat principle. IntensiveCare Med 31:12–23

8. Martindale RG, McClave SA, VanekVW, McCarthy M, Roberts P, Taylor B,Ochoa JB, Napolitano L, Cresci G,American College of Critical CareMedicine; A.S.P.E.N. Board ofDirectors (2009) Guidelines for theprovision and assessment of nutritionsupport therapy in the adult critically illpatient: Society of Critical Care aMedicine and American Society forParenteral and Enteral Nutrition:Executive Summary. Crit Care Med37:1757–1761

9. Heyland DK, Dhaliwal R, Drover JW,Gramlich L, Dodek P (2003) Canadianclinical practice guidelines for nutritionsupport in mechanically ventilated,critically ill adult patients. J ParenterEnteral Nutr 27:355–373

10. Kreymann KG, Berger MM, Deutz NE,Hiesmayr M, Jolliet P, Kazandjiev G,Nitenberg G, van den Berghe G,Wernerman J, DGEM (German Societyfor Nutritional Medicine), Ebner C,Hartl W, Heymann C, Spies C, ESPEN(European Society for Parenteral andEnteral Nutrition) (2006) ESPENguidelines on enteral nutrition:intensive care. Clin Nutr 25:210–223

11. Doig GS, Simpson F, Finfer S, DelaneyA, Davies AR, Mitchell I, Dobb G,Nutrition Guidelines Investigators ofthe ANZICS Clinical Trials Group(2008) Effect of evidence-based feedingguidelines on mortality of critically illadults: a cluster randomized controlledtrial. JAMA 300:2731–2741

G. S. Doig ()) � F. SimpsonIntensive Care Unit, Northern ClinicalSchool, Royal North Shore Hospital,University of Sydney, Pacific Hwy,St Leonards, Sydney, NSW 2006, Australiae-mail: gdoig@med.usyd.edu.auTel.: ?61-2-99268656Fax: ?61-2-94398418

1090