Articles

Low versus standard calorie and protein feeding in

ventilated adults with shock: a randomised, controlled,
multicentre, open-label, parallel-group trial (NUTRIREA-3)
Jean Reignier, Gaetan Plantefeve, Jean-Paul Mira, Laurent Argaud, Pierre Asfar, Nadia Aissaoui, Julio Badie, Nicolae-Vlad Botoc, Laurent Brisard,
Hoang-Nam Bui, Delphine Chatellier, Louis Chauvelot, Alain Combes, Christophe Cracco, Michael Darmon, Vincent Das, Matthieu Debarre,
Agathe Delbove, Jérôme Devaquet, Louis-Marie Dumont, Olivier Gontier, Samuel Groyer, Laurent Guérin, Bertrand Guidet, Yannick Hourmant,
Samir Jaber, Fabien Lambiotte, Christophe Leroy, Philippe Letocart, Benjamin Madeux, Julien Maizel, Olivier Martinet, Frédéric Martino,
Virginie Maxime, Emmanuelle Mercier, Mai-Anh Nay, Saad Nseir, Johanna Oziel, Walter Picard, Gael Piton, Jean-Pierre Quenot, Florian Reizine,
Anne Renault, Jack Richecoeur, Jean-Philippe Rigaud, Francis Schneider, Daniel Silva, Michel Sirodot, Bertrand Souweine, Fabienne Tamion,
Nicolas Terzi, Didier Thévenin, Guillaume Thiery, Nathalie Thieulot-Rolin, Jean-Francois Timsit, Francois Tinturier, Patrice Tirot,
Thierry Vanderlinden, Isabelle Vinatier, Christophe Vinsonneau, Sebastian Voicu, Jean-Baptiste Lascarrou, Amélie Le Gouge, for the
NUTRIREA-3 Trial Investigators and the Clinical Research in Intensive Care and Sepsis (CRICS-TRIGGERSEP) Group

Summary
Lancet Respir Med 2023; Background The optimal calorie and protein intakes at the acute phase of severe critical illness remain unknown. We
11: 602–12 hypothesised that early calorie and protein restriction improved outcomes in these patients, compared with standard
Published Online calorie and protein targets.
March 20, 2023
https://doi.org/10.1016/
S2213-2600(23)00092-9 Methods The pragmatic, randomised, controlled, multicentre, open-label, parallel-group NUTRIREA-3 trial was
See Comment page 580
performed in 61 French intensive care units (ICUs). Adults (≥18 years) receiving invasive mechanical ventilation and
Movement, Interactions,
vasopressor support for shock were randomly assigned to early nutrition (started within 24 h after intubation) with
Performance, UR 4334, Nantes either low or standard calorie and protein targets (6 kcal/kg per day and 0·2–0·4 g/kg per day protein vs 25 kcal/kg
Université, Nantes, France per day and 1·0–1·3 g/kg per day protein) during the first 7 ICU days. The two primary endpoints were time to
(Prof J Reignier MD, readiness for ICU discharge and day 90 all-cause mortality. Key secondary outcomes included secondary infections,
J-B Lascarrou MD); Médecine
Intensive Réanimation, CHU de
gastrointestinal events, and liver dysfunction. The trial is registered on ClinicalTrials.gov, NCT03573739, and is
Nantes, Hôtel-Dieu, Nantes, completed.
France (Prof J Reignier,
J-B Lascarrou); Service de Findings Of 3044 patients randomly assigned between July 5, 2018, and 8 Dec 8, 2020, eight withdrew consent to
Médecine Intensive
Réanimation, Centre
participation. By day 90, 628 (41·3%) of 1521 patients in the low group and 648 (42·8%) of 1515 patients in the
Hospitalier d’Argenteuil, standard group had died (absolute difference –1·5%, 95% CI –5·0 to 2·0; p=0·41). Median time to readiness for ICU
Argenteuil, France discharge was 8·0 days (IQR 5·0–14·0) in the low group and 9·0 days (5·0–17·0) in the standard group (hazard ratio
(G Plantefeve MD); Service de
[HR] 1·12, 95% CI 1·02 to 1·22; p=0·015). Proportions of patients with secondary infections did not differ between
Médecine Intensive
Réanimation, Hôpital Cochin, the groups (HR 0·85, 0·71 to 1·01; p=0·06). The low group had lower proportions of patients with vomiting (HR 0·77,
Groupe Hospitalier Paris 0·67 to 0·89; p<0·001), diarrhoea (0·83, 0·73 to 0·94; p=0·004), bowel ischaemia (0·50, 0·26 to 0·95; p=0·030), and
Centre-Université Paris Cité, liver dysfunction (0·92, 0·86–0·99; p=0·032).
AP-HP, Paris, France
(Prof J-P Mira MD); Service de
Médecine Intensive Interpretation Compared with standard calorie and protein targets, early calorie and protein restriction did not
Réanimation, Hôpital Edouard decrease mortality but was associated with faster recovery and fewer complications.
Herriot, Hospices Civils de
Lyon, Lyon, France
Funding French Ministry of Health.
(Prof L Argaud MD); Service de
Médecine Intensive
Réanimation, CHU Angers, Copyright © 2023 Published by Elsevier Ltd. All rights reserved.
Angers, France
(Prof P Asfar MD); Service de
Médecine Intensive
Introduction death.3,4 International guidelines recommend starting
Réanimation, Hôpital Européen Critical illness requiring organ support is associated with nutritional support within 48 h after ICU admission, via
Georges Pompidou, AP-HP, both mortality and prolonged recovery in survivors. A key the enteral route if not contraindicated, with targets of
Paris, France period is the acute phase characterised by organ failure, 20–25 kcal/kg per day and 1·2–2 g/kg per day of protein
(Prof N Aissaoui, MD); Service de
Médecine Intensive
anorexia, metabolic disorders, endocrine dysfunction, at the acute phase.5,6 These targets are rarely achieved in
Réanimation, Hôpital Nord and major hypercatabolism with severe muscle wasting.1,2 patients with severe critical illnesses, who frequently
Franche Comté, Trevenans, Nutritional support is crucial at this phase. Greater calorie experience gastroparesis responsible for intolerance to
France (J Badie MD); Service de and protein deficits have been shown to be associated enteral nutrition.7
Médecine Intensive
Réanimation, Centre
with higher risks of health-care-associated infections, Data from studies conducted in the past 10 years
Hospitalier de Saint Malo, intensive care unit (ICU)-acquired weakness, prolonged challenge the appropriateness of these standard calorie
Saint-Malo, France invasive mechanical ventilation, long ICU stays, and and protein targets during the acute phase of critical

602 www.thelancet.com/respiratory Vol 11 July 2023

Downloaded for Anonymous User (n/a) at Industrial University of Santander from ClinicalKey.com by Elsevier on July 12,
2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
 Articles

(N-V Botoc MD); Service

Research in context d’Anesthésie Réanimation
Chirurgicale, Hôpital Laënnec,
Evidence before this study Added value of this study CHU de Nantes, Nantes, France
To prepare the study protocol, we searched PubMed for The NUTRIREA-3 study is to our knowledge, the largest, (L Brisard MD); Service de
studies assessing outcomes of severe critical illness according multicentre, randomised, controlled trial, to date, comparing Médecine Intensive
Réanimation, CHU de
to calorie and protein intakes. We applied no date or language the effects of low-calorie low-protein feeding and standard-
Bordeaux, Bordeaux, France
restrictions. The main search terms were “protein”, calories”, calorie standard-protein feeding during the acute phase of (H-N Bui MD); Service de
“shock”, “mechanical ventilation”, “intensive care unit”, critical illness. In keeping with the latest literature on the topic, Médecine Intensive
“critical illness”, “muscle weakness” and “recovery”. During the the hypothesis was that low-calorie low-protein feeding would Réanimation, CHU de Poitiers,
Poitiers, France
trial, we updated the search until the writing of the current improve outcomes of patients with severe critical illness
(D Chatellier MD); Service de
report. Many observational studies available at the time compared with the standard calorie and protein intakes Médecine Intensive
NUTRIREA-3 was designed suggested poorer patient recommended in current guidelines. The strengths of this trial Réanimation, Hôpital de la
outcomes when enteral calorie and protein intakes during the include the large sample size achieved by recruitment at Croix Rousse, Hospices Civils de
Lyon, Lyon, France
acute phase of critical illness were lower than recommended 61 centres, careful pragmatic design, strict adherence to the (L Chauvelot MD); Service de
by guidelines (20–25 kcal/kg per day and 1·2–2·0 g/kg per day, protocol, and calorie and protein intakes close to targets in Médecine Intensive
respectively). However, randomised trials did not confirm both arms. Moreover, we included a well defined and Réanimation, Sorbonne
these findings. Moreover, strategies designed to achieve representative population of critically ill patients who required Université, Inserm, UMRS
1166-ICAN, Institute of
calorie or protein targets in critically ill patients intolerant to at least invasive mechanical ventilation and vasoactive drugs Cardiometabolism and
enteral feeding failed to prove beneficial. In the EPaNIC trial, and were thus at high risk for death or protracted recovery. Our Nutrition, Hôpital Pitié–
in patients with below-target calorie intakes due to enteral population was therefore likely to benefit from improved early Salpêtrière, AP-HP, Paris,
feeding intolerance, supplemental parenteral nutrition was nutritional support. The calorie and protein targets differed France (Prof Alain Combes MD);
Service de Médecine Intensive
associated with a longer stay in the intensive care unit and a substantially between the two groups. These conditions Réanimation, Centre
higher frequency of infection. Ancillary studies of EPaNIC data support the validity of our findings that early calorie and Hospitalier d’Angoulême,
suggested improved recovery with lower macronutrient protein restriction does not decrease mortality but diminishes Angoulême, France
intakes. These findings prompted randomised trials of the risk of complications and expedites recovery, compared (C Cracco MD); Université Paris
Cité, Service de Médecine
permissive or trophic underfeeding, which showed no harms with standard calorie and protein intakes. Intensive Réanimation, CHU
but also no benefits. Limitations of these trials included an Saint Louis, AP-HP, Paris,
Implications of the available evidence
insufficient difference in calorie intakes between groups and France (Prof M Darmon MD);
NUTRIREA-3 provides a high level of evidence that patients Service de Médecine Intensive
failure to achieve targets set by the trial protocol or
benefit from restricted calorie and protein intakes during the Réanimation, Centre
recommended in guidelines. Protein intakes during the acute Hospitalier Intercommunal
acute phase of critical illness. The NUTRIREA-3 results are
phase of critical illness have also been investigated. Higher André Grégoire, Montreuil,
consistent with those of EPaNIC, which focussed only on calorie
intakes seem to be associated with worse muscle weakness France (V Das MD); Service de
intake. Thus, NUTRIREA3 provides an important evidence of Médecine Intensive
but, as with calories, patient outcomes did not differ. Recent
nutritional support in the critically ill and could change nutrition Réanimation, Centre
reviews have emphasised the uncertainty about optimal Hospitalier de Saint Brieuc,
practices in the intensive care unit. Moreover, these results
calorie and protein intakes at the acute phase of severe critical Saint Brieuc, France
might help health-care providers in other settings to improve
illness. Moreover, no large randomised trial has compared very (M Debarre MD); Service de
patient management and to build further studies on nutrition. Réanimation Polyvalente,
low to standard intakes of both calories and protein.
Centre Hospitalier Bretagne-
Atlantique, Vannes, France
(A Delbove MD); Service de
illness.8,9 In randomised trials, increasing the enteral We designed the NUTRIREA-3 multicentre randomised Réanimation Polyvalente,
calorie intake did not improve outcomes.10–12 Adding trial to evaluate whether, in critically ill patients receiving Hôpital Foch, Suresnes, France
(J Devaquet MD); Service de
parenteral nutrition to enteral nutrition to increase invasive mechanical ventilation and vasoactive drugs, Médecine Intensive
intakes was associated with longer ICU stays and higher low-calorie low-protein feeding decreased day 90 Réanimation, Hôpital Louis-
infection rates.13,14 Higher protein intakes during the mortality or ICU length of stay, or both parameters, Mourier, AP-HP, Colombes,
acute phase might be linked to greater muscle wasting compared with standard calorie and protein supplies. France (L-M Dumont MD);
Service de Médecine Intensive
and ICU-acquired weakness.1,15 Intentionally supplying Réanimation, Centre
fewer calories than recommended, even down to 400 kcal Methods Hospitalier de Chartres,
per day (trophic feeding), did not adversely affect patient Study design Chartres, France (O Gontier MD);
outcomes.16,17 Thus, low calorie and protein intakes might NUTRIREA-3 was a 1:1 randomised, controlled, Service de Médecine Intensive
Réanimation, Centre
have benefits. A major limitation of available studies of multicentre, open-label, parallel-group, superiority trial. Hospitalier de Montauban,
calorie restriction is failure to reach the standard target The study protocol was approved by the competent ethics Montauban, France
in the control groups, which decreased the ability to committee (Comité de Protection des Personnes Sud- (S Groyer MD); Service de
detect a significant difference between groups. Moreover, Méditerranée 2, number 2018-A00424-51). Before each Médecine Intensive
Réanimation, CHU Bicêtre,
no studies compared low versus standard protein intakes inclusion, the patient or next of kin provided written AP-HP, Paris, France
at the acute phase. Thus, the optimal calorie and protein informed consent. If the patient was unable to receive (L Guérin MD); Sorbonne
intakes at the acute phase of severe critical illness remain information and no next of kin could be contacted during Université, Inserm, Institut
unknown.9,18,19 screening for the study, trial inclusion was completed as Pierre Louis d’Epidémiologie et

www.thelancet.com/respiratory Vol 11 July 2023 603

Downloaded for Anonymous User (n/a) at Industrial University of Santander from ClinicalKey.com by Elsevier on July 12,
2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
 Articles

de Santé Publique, Service de an emergency procedure by the ICU physician, in for chronic bowel disease; dying patient, not-to-be-
Médecine Intensive compliance with French law. The electronic case-record resuscitated order, or other treatment-limitation decision
Réanimation, Hôpital Saint
Antoine, AP-HP, Paris, France
form and database organisation were approved by the at ICU admission; pregnancy, recent delivery, or lactation;
(Prof B Guidet MD); CHU de appropriate committees as required by French law. The adult under guardianship; and correctional facility inmate.
Nantes, Inserm, Nantes study protocol has been published elsewere.20
Université, Anesthesie Randomisation and masking
Reanimation, CIC 1413, Nantes,
France (Y Hourmant MD);
Participants All patients treated with invasive mechanical ventilation
Service de Réanimation The trial was conducted in the 61 French ICUs listed in and vasoactive drugs for shock within 24 hours after ICU
Chirurgicale, Hôpital Saint-Eloi, the appendix (pp 3–8). Adults (≥18 years) admitted to any admission were screened for eligibility by the ICU
CHU de Montpellier, of the participating ICUs were eligible if they were physicians and clinical research nurses, around the clock
Montpellier, France
(Prof S Jaber MD); PhyMedExp,
receiving invasive mechanical ventilation, with an and 7 days a week. Investigators at each centre used a
Inserm, CNRS, Montpellier, expected duration of at least 48 h after inclusion and secure, computer-generated, interactive, web-response
France (Prof S Jaber); Service de initiation either in the ICU within the past 24 h or before system to randomly assign consecutive eligible patients
Médecine Intensive ICU admission having occurred within the past 24 h, in a 1:1 ratio to one of the two early nutritional-support
Réanimation, Centre
Hospitalier de Valenciennes,
concomitantly with vasoactive therapy (adrenaline, groups defined by low versus standard calorie and
Valenciennes, France dobutamine, or noradrenaline) for shock, and if nutritional protein targets. Randomisation was stratified by centre
(F Lambiotte MD); Service de support was expected to be started within 24 h after using permutation blocks of variable sizes. The
Médecine Intensive
intubation (or within 24 h after ICU admission when investigators could not access the randomisation list and
Réanimation, Centre
Hospitalier Emile Roux, Le Puy- intubation occurred before ICU admission). Exclusion were unaware of block size. Given that the volume and
en-Velay, France (C Leroy, MD); criteria were specific nutritional needs, such as pre- delivery of enteral and parenteral feeding preparations
Service de Médecine Intensive existing long-term home enteral or parenteral nutrition cannot be masked, masking of physicians and nurses
was not feasible. The electronic case-report form was a
10 767 patients assessed for eligibility secure, interactive, web-response system that was
available at each study centre and was provided and
managed by the biometrics unit of the Tours University
7723 excluded
5507 met exclusion criteria
Hospital (CIC INSERM 1415, Tours, France).
1651 started invasive mechanical ventilation >24 h
earlier or started nutritional support before Procedures
invasive mechanical ventilation or vasoactive
drug initiation The calorie and protein targets for the acute phase,
581 had an expected invasive mechanical defined as the first 7 days after ICU admission, were
ventilation duration of <48 h
3017 had treatment-limitation decisions
6 kcal/kg per day and 0·2–0·4 g/kg per day in the low
172 had pre-existing nutritional support group versus 25 kcal/kg per day and 1·0–1·3 g/kg per day
39 women were pregnant in the standard group. On day 8, the targets were changed
47 were under guardianship or were correctional
facility inmates to 30 kcal/kg per day for calories and 1·2–2·0 g/kg per
2216 were eligible but not randomised day for protein in both groups. The daily nutritional
808 patients or relatives could not receive intake required to meet the assigned calorie target was
information about the study or refused
participation calculated based on bodyweight. In patients with obesity
702 were inadvertently omitted from the study (BMI >30 kg/m²), the bodyweight yielding a BMI of
inclusion process
352 were admitted at a time when no research
30 kg/m² was used. In patients whose BMI was below
staff were available or trial organisation 18·5 kg/m², we used the corrected bodyweight computed
difficulties were occurring as half the sum of the ideal and actual bodyweights. The
85 were excluded by the physician in charge
269 were enrolled in another trial calorie to protein ratios of nutrient preparations currently
available in French hospitals ensured that the protein
intake complied with the assigned target.
3044 randomly assigned
In both groups, the assigned feeding strategy was
initiated as soon as possible after randomisation and no
later than 24 h after intubation (or after ICU admission
1525 allocated to the low group 1519 allocated to the standard group in patients intubated before ICU admission). It was
continued until extubation and withdrawal of vasoactive
drugs, death, or end of day 7 from admission, whichever
4 included under the emergency 4 included under the emergency
consent procedure did not confirm consent procedure did not confirm occurred first. Patients who were reintubated within
their willingness to participate their willingness to participate 7 days after inclusion were managed until the end of the
acute phase using the feeding strategy they were assigned
1521 included in the primary outcome 1515 included in the primary outcome to initially. Nutritional support was started at the flow
analysis analysis rate (mL/h) that achieved the calorie target on day 1 and
was delivered continuously over the 24 h cycle, with no
Figure 1: Trial profile interruptions.

604 www.thelancet.com/respiratory Vol 11 July 2023

Downloaded for Anonymous User (n/a) at Industrial University of Santander from ClinicalKey.com by Elsevier on July 12,
2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
 Articles

Réanimation, Centre
During the acute phase, bedside physicians determined control and insulin-therapy protocols were applied. Hospitalier Jacques Puel, Rodez,
the best feeding route each day, according to clinical Additional water, electrolytes, vitamins, and trace elements France (P Letocart MD); Service
considerations, to ensure that the calorie target was were given intravenously according to the needs of each de Médecine Intensive
Réanimation, Centre
achieved.21,22 After the acute phase, enteral feeding patient, as assessed by the physician in charge, using the
Hospitalier de Bigorre, Tarbes,
remained the preferred route in patients without standard preparations and protocols available at each France (B Madeux MD); Service
contraindications.5,6 At each centre, the usual blood-glucose centre. de Médecine Intensive
Additional details of the interventions and protocols for Réanimation, CHU Amiens-
Picardie, Amiens, France
Low group Standard group providing nutritional support, including measures
(Prof J Maizel MD); Service de
(n=1521) (n=1515) designed to evaluate tolerance, have been published Médecine Intensive
Age, years 66 (13) 66 (13) previously.20 All participating ICU staff members Réanimation, CHU de la
attended training in the study procedures and protocols Réunion, Saint-Denis,
Sex
La Réunion, France
Men 1010 (66·4%) 1026 (67·7%) for providing nutritional support and managing (O Martinet MD); Service de
Women 511 (33·6%) 489 (32·3%)
intolerance to enteral nutrition.20 Médecine Intensive
Réanimation, CHU de la
McCabe score
Guadeloupe, Abymes,
0—no fatal underlying 1051 (69·5%) 1054 (69·7%) Guadeloupe, France
Low group Standard group
disease (F Martino MD); Service de
(n=1521) (n=1515)
1—death expected 389 (25·7%) 378 (25·0%) Médecine Intensive
within 5 years (Continued from previous page) Réanimation, Hôpital Raymond
2—death expected 73 (4·8%) 81 (5·4%) Ongoing treatments§ Poincaré, AP-HP, Garches,
within 1 year France (V Maxime MD); Inserm
Prone position 90 (6·0%) 93 (6·1%)
U 1173, Université de Versailles-
Pre-existing illness at ICU 1059 (70∙2%) 1066 (70·9%) Sedative agents 1368 (90·5%) 1351 (89·3%) Saint Quentin en Yvelines,
admission
NMBA 483 (32·0%) 497 (32·8%) Versailles, France (V Maxime);
Pre-existing illness at ICU admission Service de Médecine Intensive
Insulin 593 (39·2%) 664 (43·9%)
Chronic renal failure 184 (12·2%) 180 (12∙0%) Réanimation, CHU de Tours,
Antiulcer medication 656 (43·4%) 686 (45·3%)
CRICS-TRIGGERSEP Network
Liver disease 139 (9·2%) 150 (10·0%)
Prokinetic agents 39 (2·6%) 52 (3·4%) Tours, France (E Mercier MD);
Cardiovascular disease 292 (19·5%) 298 (19·9%) Service de Médecine Intensive
Antimicrobial treatment 1298 (85·8%) 1290 (85·1%)
Chronic respiratory 195 (13·0%) 155 (10·4%) Réanimation, Centre
Dialysis 161 (10·6%) 183 (12·1%)
failure Hospitalier Régional d’Orléans,
Other parameters Orléans, France (M-A Nay MD);
Neurological disease 209 (13·9%) 192 (12·8%)
FiO₂ 50·0 (40·0–80·0) 60·0 (40·0–90·0) Médecine Intensive-
Cancer or immune 395 (26·2%) 407 (27·1%)
PEEP, cmH₂O 7·0 (5·0–10·0) 7·0 (5·0–10·0) Réanimation, CHU Lille, France
deficiency
(Prof S Nseir MD); CNRS, Inserm,
Oesophageal, gastric, or 116 (7·7%) 99 (6·6%) Glucose, mmol/L¶ 8·8 (6·7–12·0) 9·5 (7·1–12·8)
UMR 8576– U1285, Unité de
duodenal ulcer Serum creatinine, 128·5 (84·0–215·0) 130·0 (85·0–211·0) Glycobiologie Structurale et
Diabetes 374 (24·9%) 384 (25·6%) µmol/L|| Fonctionnelle, Université de
Weight, kg 76·5 (65·0–89·5) 77·0 (65·3–90·0) Lactate, mEq/L 2·4 (1·5–4·6) 2·7 (1·6–4·9) Lille, France (Prof S Nseir);
Service de Médecine Intensive
BMI, kg/m² 26·7 (23·0–31·1) 27·0 (23·0–31·5) Bilirubin, µmol/L 13·0 (8·0–25·0) 13·0 (8·0–26·0)
Réanimation, Hôpital
Pre-existing malnutrition* C-reactive protein, mg/L 137·0 (42·0–266·0) 144·0 (51·0–273·8) Avicenne, AP-HP, Bobigny,
None 1362 (90·0%) 1379 (91·2%) Pre-albumin, g/L 0·10 (0·06–0·15) 0·11 (0·06–0·17) France (J Oziel MD); Service de
Albumin, g/L 25·0 (20·0–29·6) 25·0 (20·0–29·9) Médecine Intensive
Moderate 64 (4·2%) 62 (4·1%)
Réanimation, Centre
Severe 87 (5·8%) 71 (4·7%) Time from intubation to 14·8 (8·7–20·0) 15·1 (8·8–19·8)
Hospitalier de Pau, Pau, France
randomisation, h
SAPS II† 60 (48–74) 61 (48–74) (W Picard MD); Service de
SOFA score‡ 10 (8–13) 10 (8–13) Data are mean (SD), n (%), or median (IQR). Denominators might vary slightly Médecine Intensive
according to missing data. ICU=intensive care unit. SAPS II=Simplified Acute Réanimation, CHU de
Medical diagnosis at 1253 (82·8%) 1258 (83·1%) Besançon, Besançon, France
Physiology Score version II. SOFA=Sequential Organ Failure Assessment.
admission (Prof G Piton MD); Université de
NMBA=neuromuscular blocking agent. FiO₂=fractional concentration of oxygen
Acute illness at ICU admission in inspired air. PEEP=positive end-expiratory pressure. *Severe pre-existing Franche Comté, Equipe EA
Cardiac arrest 185 (12·2%) 209 (13·8%) malnutrition was defined as BMI <18∙5 kg/m² or weight loss >10% within the past 3920, Besançon, France
6 months and moderate pre-existing malnutrition as BMI 18∙5–19∙9 kg/m² or (Prof G Piton); Service de
Acute heart failure 232 (15·3%) 253 (16·7%)
weight loss >5% within the past 6 months; patients who did not meet these Médecine Intensive
Acute central nervous 109 (7·2%) 112 (7·4%) criteria were classified in the none category. †SAPS II values can range from Réanimation, CHU François
system failure 0 (lowest level of critical illness) to 163 (most severe level of critical illness with Mitterrand, Dijon, France
Acute respiratory failure 686 (45·3%) 654 (43·2%) 100% predicted mortality); a score of 50 predicts a 46∙1% risk of death; the SAPS II (Prof J-P Quenot MD); Lipness
was determined 24 h after ICU admission. ‡SOFA scores can range from 0 (no Team, Inserm, LabExLipSTIC
Trauma 25 (1·7%) 28 (1·8%)
organ failure) to 24 (most severe level of multi-organ failure); the SOFA sub-score France (Prof J-P Quenot) and
Miscellaneous 278 (18·3%) 258 (17·7%) values at ICU admission are reported in the appendix (p 18). §Antimicrobial Inserm Centres d’Investigation
Cause of shock treatments included antibiotics, antiviral drugs, and antifungal drugs; antiulcer Clinique, Département
treatments included proton-pump inhibitors and histamine-2 receptor d’Epidémiologie Clinique
Cardiac 254 (16·8%) 283 (18·7%)
antagonists; prokinetic agents were metoclopramide and erythromycin. (Prof J-P Quenot), Université de
Sepsis 889 (58·7%) 874 (57·7%) ¶To convert glucose values to mg/dL, multiply by 18∙02. ||To convert creatinine Bourgogne, Dijon, France;
Other 372 (24·5%) 357 (23·6%) values to mg/dL, multiply by 0∙113.
Service de Médecine Intensive
(Table 1 continues on next page) Table 1: Baseline characteristics of the participants
Réanimation, CHU de Rennes,
Rennes, France (F Reizine MD);

www.thelancet.com/respiratory Vol 11 July 2023 605

Downloaded for Anonymous User (n/a) at Industrial University of Santander from ClinicalKey.com by Elsevier on July 12,
2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
 Articles

1·0 Low group

Baseline characteristics were recorded at inclusion. The
Standard group Simplified Acute Physiology Score (SAPS II) was
Cumulative incidence of readiness for discharge alive from the ICU

computed 24 h after ICU admission. After nutritional

0·8
support initiation, daily recordings were made of
nutritional data, treatments, nosocomial infections,
abdominal complications, laboratory data, and invasive
0·6
devices, until hospital discharge or day 90, whichever
occurred first. Vital status was recorded at ICU discharge,
at hospital discharge, and on days 28 and 90. Bowel
ischaemia and ventilator-associated pneumonia were
0·4
diagnosed according to predefined criteria, as previously
reported.20,21 ICU-acquired infections were adjudicated by
an independent masked committee, based on all available
0·2
clinical, radiological, and bacteriological data. All the study
data were stored in a logged database that was locked on
July 28, 2022, after the site investigators had responded to
0
0 7 28 90 all the queries made by the database managers.20
Time since randomisation (days)
Number at risk
(number censored) Statistical analysis
Low group 1521 904 156 3 Two interim analyses to be performed using the
(13) (34) (0) (0)
Standard group 6
Haybittle-Peto approach were scheduled, after enrolment
1515 921 146
(6) (25) (1) (0) of 1000 and 2000 patients, respectively. The significance
level associated with both interim analyses was 0·001
Figure 2: Time to readiness for ICU discharge and the significance level associated with the final
Cumulative incidence curves for patients who achieved readiness for ICU discharge. ICU=intensive care unit.
analysis was 0·049. With this method, the overall risk of
type 1 error was 5%. As shown in the appendix (pp 8), the
Service de Médecine Intensive Outcomes independent data safety monitoring board was composed
Réanimation, CHU la Cavale We assessed two primary outcomes: all-cause day 90 of a methodologist and two intensivists not otherwise
Blanche, Brest, France
(A Renault MD); Service de
mortality and time to readiness for ICU discharge. involved in the trial. For both interim analyses, the board
Médecine Intensive Patients were considered ready for ICU discharge as had access to unmasked results on day 90 mortality, time
Réanimation, Centre soon as they met all the predefined criteria, regardless to ICU discharge alive, SOFA score variations from day 0
Hospitalier de Beauvais, of availability of beds on the ward: (1) no longer in need to day 7, amount of calories and protein received daily
Beauvais, France
(J Richecoeur MD); Service de
of, or at risk for needing, invasive mechanical from day 0 to day 7, and nosocomial infections. The
Médecine Intensive ventilation; (2) no longer in need of, or at risk for results of the interim analyses were not disclosed to the
Réanimation, Centre needing, vasoactive support; (3) no agitation or investigators.
Hospitalier de Dieppe, Dieppe, consciousness alteration requiring close monitoring To estimate the required sample size, we used the
France (J-P Rigaud MD); Service
de Médecine Intensive
and management; and (4) no severe acute metabolic or mortality rates and mean ICU length of stay recorded in
Réanimation, Hôpitaux haematological disorder requiring close monitoring and survivors in the NUTRIREA-2 trial, which used similar
Universitaires de Strasbourg, management.13,14 Readiness for ICU discharge was selection criteria.21 Assuming a 43% day 90 mortality rate
Strasbourg, France
checked daily in all patients weaned off invasive in the standard group and a 5% absolute decrease to
(Prof F Schneider MD); Service
de Médecine Intensive mechanical ventilation and vasoactive drugs. This 38% in the low group, with the α risk set at 4·9% (as
Réanimation, Hôpital evaluation was performed by the bedside physician in two interim analyses were planned) and the β risk at 20%,
Delafontaine, Saint-Denis, charge of the patient. 1522 patients were needed in each group—
France (D Silva MD); Service de
Secondary outcomes included the Sequential Organ ie, 3044 patients in total. This sample size provided
Médecine Intensive
Réanimation, Centre Failure Assessment (SOFA) score; bodyweight; amount 94% power for detecting a 1·5-day difference in readiness
Hospitalier Annecy Genevois, of calories and protein delivered; vomiting; prokinetic to ICU discharge between the two groups (mean
Epagny Metz-Tessy, France drug use; diarrhoea and constipation; blood glucose; 14·5 days in the standard group vs 13·0 days in the low
(M Sirodot, MD); Service de
insulin treatment; blood concentrations of lactate, group). No corrections were planned for multiple
Médecine Intensive
Réanimation, CHU Gabriel- bilirubin, alanine aminotransferase, and aspartate comparisons.
Montpied, Clermont-Ferrand, aminotransferase; antiulcer prophylaxis; antimicrobial All statistical analyses followed a prespecified statistical
France (Prof B Souweine MD); treatments; prone positioning; dialysis during the analysis plan. Values of p no greater than 0·049 were
Service de Médecine Intensive
intervention period; day 28 mortality; ICU mortality; taken to indicate significance. Categorical variables were
Réanimation, Hôpital Charles
Nicolle, CHU de Rouen, hospital mortality; ICU length of stay; acute-care hospital described as frequencies and percentages (n[%]) and
Normandie Université, length of stay; days without life-support; ICU-acquired continuous variables as medians (IQR) or means (SD).
UNIROUEN, Inserm U1096, infections; and non-infectious complications. Information No statistical tests were performed to compare baseline
FHU REMOD-VHF, Rouen,
France (Prof F Tamion MD);
on data collection and outcome definitions was previously characteristics between groups. Patients with missing
reported.20 data on day 90 mortality were assumed to have died.

606 www.thelancet.com/respiratory Vol 11 July 2023

Downloaded for Anonymous User (n/a) at Industrial University of Santander from ClinicalKey.com by Elsevier on July 12,
2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
 Articles

Day 90 mortality was reported as the point estimate with p=0·41). The results were similar after exclusion of Service de Médecine Intensive
the 95% CIs in each group. The difference in proportions, three patients with missing data on day 90 mortality Réanimation, Université de
Grenoble-Alpes, Inserm U1042,
with 95% CIs, was also estimated. Day 90 mortality was (estimated difference –1·6%; –5·1 to 1·9; p=0·37). Time
Grenoble, France
compared between the two groups using the χ² test. to readiness for ICU discharge was 8·0 days (Prof N Terzi MD); Service de
Time to readiness for ICU discharge was evaluated (IQR 5·0–14·0) in the low group and 9·0 days (5·0–17·0) Médecine Intensive
using a Fine-and-Gray model with death in the ICU as a in the standard group (hazard ratio [HR] 1·12, 95% CI Réanimation, Centre
Hospitalier de Lens, Lens,
competing risk. Secondary outcomes expressed as 1·02 to 1·22; p=0·015; figure 2).
proportions were compared between the two groups by
applying the χ² test. Outcomes reported as cumulative
incidences were analysed using the competing-risk A B
approach, with death, ICU discharge, or hospital 40 Low group 150
Standard group
discharge as the competing event. Changes over time

Prescribed nutritional calories

Nutritional calories received
were compared between the two groups by building 30
mixed linear models, after data transformation if 100

received (%)
(kcal/kg)
necessary. Statistical analyses for secondary endpoints 20
were not adjusted for multiplicity. The findings should
therefore be interpreted as exploratory. Continuous data 50
10
were analysed by applying Wilcoxon’s nonparametric
test. SAS (version 9.4) and R (version 3.3.1) were used for
0 0
the statistical analyses. The trial is registered with
ClinicalTrials.gov (NCT03573739) under the name
C D
NUTRIREA-3.
12 50
Non-nutritional calories received

Role of the funding source

Total calories received (kcal/kg)

10 40
The funders of the study had no role in the study design,
8
data collection, data analysis, or data interpretation; 30
(kcal/kg)

writing of the report; or decision to submit for 6

publication. 20
4
10
Results 2
From July 5, 2018, to Dec 8, 2020, 3044 patients in 0 0
61 French ICUs, including 34 (55·7%) in university
hospitals (appendix p 18), were randomly assigned to E F
one of the study groups. Four patients were withdrawn in 2·5 25
each group, leaving 3036 patients for the analysis:
2·0 20
1521 in the low group and 1515 in the standard group
Protein received (g/kg)

(figure 1). Baseline characteristics were similar between

SOFA score

1·5 15
groups (table 1; appendix p 19).
By day 90, 628 (41·3%) of 1521 patients in the low group 1·0 10
and 648 (42·8%) of 1515 patients in the standard group
had died (absolute difference –1·5%, 95% CI –5·0 to 2·0; 0·5 5

0 0

G H
Figure 3: Daily calorie intake, protein intake, SOFA score, and glucose control
25 250
during the intervention period (days 0–7)
Nutritional calories received daily (A), percentages of the calorie targets
20 200
supplied (B), non-nutritional calories received daily (C), and total calories received
Glucose (mmol/L)

daily (D), during the intervention period (from day 0 to day 7), in both groups.
15 150
Insulin (IU)

(E) Protein amounts administered daily during the intervention period in both
groups. Bodyweight measured on admission was used throughout the ICU stay
to calculate calorie and protein targets. (F) Differences between the low and 10 100
standard groups for the SOFA score during the intervention period ; SOFA scores
can range from 0 (no organ failure) to 24 (most severe level of multiorgan 5 50
failure). Differences between the low and standard groups for daily blood glucose
concentrations (G) and daily insulin intake (H); box plot represents median (IQR),
0 0
and the lower whisker represents the 25th percentile minus 1·5 times the IQR and
y4
y3
y0

y2

y5
y1
y6
y4
y0

y2

y3

y5

y7
y1

y6

y7

the upper whisker the 75th percentile plus 1·5 times the IQR; if the box plot
Da

Da
Da

Da

Da
Da
Da

Da
Da

Da

Da
Da

Da

Da

Da
Da

contains no horizontal line, the median value is the same as the 75th percentile. Time since randomisation (days) Time since randomisation (days)
ICU=intensive care unit. SOFA=Sequential Organ Failure Assessment.

www.thelancet.com/respiratory Vol 11 July 2023 607

Downloaded for Anonymous User (n/a) at Industrial University of Santander from ClinicalKey.com by Elsevier on July 12,
2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
 Articles

France (D Thévenin MD); Service

de Médecine Intensive Low group (n=1521) Standard group Hazard ratio (95% CI) p value
Réanimation, CHU de Saint (n=1515)
Étienne, Saint Priest en Jarez, Clinical management
France (Prof G Thiery MD);
Intubation to feeding, h 16·7 (10·6–21·6) 17·0 (10·8–21·8) ·· ··
Service de Médecine Intensive
Réanimation, Groupe Feeding route
Hospitalier Sud Ile de France, Patients receiving enteral nutrition only 932 (61·6%) 846 (56·0%) ·· ··
Melun, France
Patients receiving parenteral nutrition only 358 (23·7%) 349 (23·1%) ·· ··
(N Thieulot-Rolin MD); Service
de Médecine Intensive Patients receiving parenteral nutrition then enteral nutrition 208 (13·8%) 296 (19·6%) ·· ··
Réanimation, CHU Bichat- Neither enteral nutrition nor parenteral nutrition 14 (0·9%) 21 (1·4%) ·· ··
Claude Bernard, AP-HP, Paris, Daily intakes
France (Prof J-F Timsit MD);
Calories, kcal/kg per 24 h* 7·4 (5·8–9·5) 22·0 (17·5–24·9) ·· ··
Université Paris-Cité, Inserm
IAME, U1137, Team DesCID, Protein, g/kg per day 0·2 (0·2–0·3) 0·9 (0·7–1·0) ·· ··
Paris, France (Prof J-F Timsit); Fluids, L 11·1 (6·1–16·4) 17·2 (9·7–23·4) ·· ··
Service de Réanimation
Gastrointestinal events, cumulative incidence
Chirurgicale, CHU Amiens-
Picardie, Amiens, France Vomiting 14·2% 21·3% 0·64 (0·54–0·76) <0·001
(F Tinturier MD); Service de Prokinetic drug therapy 13·1% 20·9% 0·60 (0·51–0·72) <0·001
Médecine Intensive Constipation† 23·8% 23·9% 0·99 (0·87–1·13) 0·92
Réanimation, Centre
Diarrhoea 19·1% 22·9% 0·81 (0·69–0·95) 0·008
Hospitalier du Mans, Le Mans,
France (P Tirot MD); Service de Blood parameters
Médecine Intensive Glucose, mmol/L
Réanimation, Groupement
Daily highest 11·6 (9·6–14·5) 13·7 (11·2–17·1) ·· <0·001
Hospitalier de l’Institut
Catholique de Lille, FMMS– Daily lowest 4·8 (3·9–5·7) 5·0 (4·1–5·8) ·· 0·002
ETHICS EA 7446, Université Insulin, cumulative incidence 63·1% 77·7% 0·74 (0·70–0·80) <0·001
Catholique de Lille, Lille, France Hypoglycaemia, cumulative incidence‡ 6·0% 4·9% 1·24 (0·91–1·69) 0·17
(Prof T Vanderlinden, MD);
Lactate, daily highest, g/L 2·8 (1·8–5·2) 3·0 (2·0–5·6) ·· 0·003
Service de Médecine Intensive
Réanimation, Centre Lactate normalisation, cumulative incidence§ 86·5% 84·1% 1·09 (1·02–1·16) 0·01
Hospitalier Départemental de Bilirubin, daily highest, µmol/L 16·0 (10·0–33·0) 15·0 (10·0–32·0) ·· 0·22
la Vendée, La Roche sur Yon,
ALAT, daily highest, IU/L 65 (33–174) 77 (36–218) ·· 0·008
France (I Vinatier MD); Service
de Médecine Intensive ASAT, daily highest, IU/L 98 (46–288) 110 (50–345) ·· 0·016
Réanimation, Centre Hypokalaemia, cumulative incidence¶ 58·6% 58·9% 0·97 (0·89–1·06) 0·48
Hospitalier de Béthune, Hypophosphataemia, cumulative incidence|| 53·6% 61·3% 0·83 (0·76–0·91) <0·001
Béthune, France
(C Vinsonneau MD); Service de
Other treatments, cumulative incidence
Médecine Intensive Prone position 13·4% 17·2% 0·77 (0·64–0·92) 0·005
Réanimation, CHU Lariboisière, Dialysis 26·9% 27·2% 0·99 (0·86–1·12) 0·82
AP-HP, Paris, France
(S Voicu MD); Inserm CIC 1415, Data are median (IQR) or n (%), unless otherwise specified. Denominators might vary slightly according to missing data. SI conversion factors: to convert glucose values to
Tours, France (A Le Gouge MD); mg/dL, multiply by 18∙02; to convert bilirubin values to mg/L, multiply by 0∙58. Volume of fluids was the total amount of fluids received from day 0 to day 7 and included in
CHU de Tours, Tours, France enteral preparations, parenteral preparations, and non-nutritional fluids. The intervention period started with the initiation of nutritional support and ended after day 7 or
(A Le Gouge) ICU discharge or death, whichever occurred first. Percentages of patients with each outcome were compared between groups using the χ² test and categorical data reported
as median (IQR) using Wilcoxon’s non-parametric test. Outcomes reported as cumulative incidences were analysed using a competing risk approach, with death and ICU
Correspondence to: discharge as competing risks. ALAT=alanine aminotransferase. ASAT=aspartate aminotransferase. *Calories in propofol and dextrose solutions were included in the total
Prof Jean Reignier, Service de calorie count. †Defined as no passage of stools from randomisation to day 6 inclusive. ‡Defined as blood glucose concentration lower than 2·3 mmol/L. §Defined as blood
Médecine Intensive lactate lower than 2 mmol/L. ¶Defined as blood potassium concentration lower than 3·5 mmol/L. ||Defined as blood phosphate concentration lower than 0·81 mmol/L.
Réanimation, CHU de Nantes,
Hôtel-Dieu, 44093 Nantes Table 2: Clinical management and outcomes during the intervention period (days 0–7)
Cedex 1, France
[email protected]
See Online for appendix Adherence to the protocol was high, with daily calorie –0·2 IU per day, 95% CI –0·05 to 0·001; p=0·065).
and protein intakes close to targets in both groups Median phosphataemia was lower in the standard group
(figure 3, appendix p 9). Feeding routes were similar compared with the low group (appendix p 10). There
between groups, and nearly 60% of patients in both was no between-group difference in blood concen­
groups received only enteral feeding (table 2, trations of potassium and magnesium or in serum
appendix p 20). Median blood glucose concentrations C-reactive protein (appendix p 11, 20). During the
and cumulative incidence of patients receiving insulin intervention period, patients in the low group had lower
were lower in the low group than in the standard group, bodyweight and lower pre-albuminaemia compared
but there was no difference in the cumulative incidence with patients in the standard group (appendix p 12, 14).
of patients with hypoglycaemia (table 2; figure 3G, daily The cumulative incidence of patients treated with prone
blood glucose concentrations 0·01 mmol/L, 95% CI positioning was lower in the low group than in the
0·001 to 0·01; p=0·016; figure 3H, daily insulin intake standard group (table 2). The low group had a higher

608 www.thelancet.com/respiratory Vol 11 July 2023

Downloaded for Anonymous User (n/a) at Industrial University of Santander from ClinicalKey.com by Elsevier on July 12,
2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
 Articles

Low group (n=1521) Standard group Absolute Hazard ratio p value

(n=1515) difference (95% CI)
(95% CI)
Primary outcomes
Day 90 mortality 628 (41·3%) 648 (42·8%) –1·5 (–5·0 to 2·0) ·· 0·41
Time to readiness for ICU discharge* 8·0 (5·0 to 14·0) 9·0 (5·0 to 17·0) ·· 1·12 (1·02 to 1·22) 0·015
Secondary outcomes
Day 28 mortality 504 (33·2%; n=1519) 533 (35·2%) –2·0 (–5·4 to 1·4) ·· 0·24
ICU mortality, cumulative incidence 29·6% 32·7% ·· 0·89 (0·78 to 1·00) 0·051
Hospital mortality, cumulative incidence 32·2% 34·5% ·· 0·93 (0·83 to 1·05) 0·24
ICU length of stay, days† 9·0 (5∙0 to 15∙0) 10·0 (6∙0 to 17∙0) ·· ·· ··
Acute-care hospital length of stay, days† 21·0 (12∙0 to 38∙0) 22·0 (14∙0 to 39∙0) ·· ·· ··
Time to weaning from vasopressor support, days 3·0 (2·0 to 4·0) 3·0 (2·0 to 4·0) ·· 1·07 (0·99 to 1·15) 0·054
Time to invasive mechanical ventilation weaning, days 5·0 (2·0 to 11·0) 6·0 (3·0 to 12·5) ·· 1·12 (1·03 to 1·22) 0·007
Received dialysis, cumulative incidence 30·1% 31·9% ·· 0·93 (0·82 to 1·05) 0·25
Infections, cumulative incidence
ICU infection‡ 15·3% 17·5% ·· 0·85 (0·71 to 1·01) 0·06
Ventilator-associated pneumonia 11·2% 10·9% ·· 0·98 (0·79 to 1·21) 0·82
Bacteraemia 4∙0% 5·5% ·· 0·73 (0·53 to 1·01) 0·06
Central venous catheter infection 1·5% 1·9% ·· 0·81 (0·48 to 1·37) 0·44
Urinary tract infection 0·7% 0·8% ·· 1·20 (0·54 to 2·67) 0·66
Soft-tissue infection 7 patients 5 patients ·· ·· ··
Other infection 1·7% 2·4% ·· 0·78 (0·48 to 1·28) 0·33
Gastrointestinal events, cumulative incidence
Vomiting 20·2% 25·5% ·· 0·77 (0·67 to 0·89) <0·001
Diarrhoea 28·9% 33·3% ·· 0·83 (0·73 to 0·94) 0·004
Constipation 27·8% 28·7% ·· 0·97 (0·86 to 1·10) 0·64
Bowel ischaemia 0·9% 1·8% ·· 0·50 (0·26 to 0·95) 0·030
Acute colonic pseudo-obstruction 8 patients 2 patients ·· ·· ··
Liver dysfunction, cumulative incidence§ 61·7% 65·8% ·· 0·92 (0·86 to 0·99) 0·032
Data are %, n (%), or median (IQR), unless otherwise specified. Percentages of patients with each outcome were compared between groups using the χ² test and categorical
data reported as median (IQR) using Wilcoxon’s non-parametric test. Outcomes reported as cumulative incidences were analysed using a competing risk approach, with
death and ICU discharge as competing risks; the only exceptions were ICU mortality and hospital mortality, for which competing risks were only ICU discharge and hospital
discharge, respectively. ICU=intensive care unit. *Analysed using a competing risks model with death as a competing risk; hazard ratio >1 means that more patients were
ready for ICU discharge during follow-up—ie, that time to ICU-discharge readiness was shorter in the low group than in the standard group. †For patients discharged alive.
‡Ventilator-associated pneumonia, bacteraemia, urinary tract infections, catheter-related infections, and other infections acquired in the ICU. §Defined as serum bilirubin
>50 µmol/L or elevation to more than three times the upper limit of normal, or both parameters, in one or more liver enzymes (γ-glutamyl transferase, alkaline phosphatase,
aspartate aminotransferase, and alanine aminotransferase).

Table 3: Outcomes

cumulative incidence of patients whose blood lactate mobilisation; or Medical Research Council (MRC) score
returned to normal and a faster SOFA score decrease, at ICU discharge (table 3; appendix pp 15–17, 22).
compared with the standard group (table 2, figure 3F,
SOFA mean difference –0·08; 95% CI –0·13 to –0·02; Discussion
p=0·008). In this multicentre randomised trial, day 90 all-cause
Time to weaning off invasive mechanical ventilation mortality did not differ between low versus standard
was significantly shorter in the low group than in the calorie and protein intakes during the acute phase of
standard group (table 3). Adverse gastrointestinal events critical illness requiring invasive mechanical ventilation
(including vomiting, diarrhoea, and bowel ischaemia) and vasoactive support. However, compared with
were significantly less common in the low than in the patients receiving standard calorie and protein intakes,
standard group (table 3). The cumulative incidence of those receiving low intakes had a shorter time to
patients with liver dysfunction was lower in the low readiness for ICU discharge analysed with death as a
group compared with the standard group (table 3). The competing event. The low feeding strategy was
groups were not different for 28 day, ICU, or hospital associated with a shorter duration of invasive
mortality; vasoactive drug or renal replacement therapy mechanical ventilation and fewer gastrointestinal and
use; frequency of ICU-acquired infections; out-of-bed hepatic complications.

www.thelancet.com/respiratory Vol 11 July 2023 609

Downloaded for Anonymous User (n/a) at Industrial University of Santander from ClinicalKey.com by Elsevier on July 12,
2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
 Articles

Compared with standard intakes, low intakes were not concentrations and in mortality. Finally, the larger fluid
associated with adverse outcomes. All-cause mortality intake in the standard group could have contributed to
was non-significantly lower in the low versus the the higher frequency of prone positioning and longer
standard group, in keeping with previous studies.10,16–18 duration of invasive mechanical ventilation. However,
All-cause mortality should be viewed as a marker of the two trials investigating intravenous fluid restriction in
safety of interventions in patients with severe critical patients with acute lung injury or septic shock produced
illness, and there is no evidence from our trial that low conflicting results.29,30 Moreover, two-thirds of our
calorie and protein intakes were associated with increased patients received enteral feeding only, and the effect of
risks. More specifically, infections were not more the enteral fluid intake is unknown.
common. Observational ICU studies suggested an One limitation of our trial is that neither the patients
association between calorie deficiency and infections,4 nor the health-care staff could be masked to the
but this finding was not replicated in randomised trials intervention. However, the nutritional protocols were
of permissive or trophic underfeeding.16,17 The standardised. Moreover, regarding the primary outcomes,
TARGET trial found no benefits from increasing the day 90 mortality was objective and time to readiness for
enteral calorie supply.10 In the EpaNIC randomised trial ICU discharge was determined according to predefined
of early versus late parenteral nutrition to supplement criteria and checked daily by bedside physicians. A
insufficient enteral intakes, mortality was similar in the similar strategy regarding this endpoint has been used
two groups, but benefits in the late group included fewer previously in studies on nutrition in the ICU.13,14 We also
ICU-acquired infections and shorter durations of used pre-established definitions or adjudication for the
mechanical ventilation, renal replacement therapy, and secondary outcomes. These methodological features
ICU stay.13 In our trial, ICU-acquired infections were less reduce any potential biases related to the absence of
common than in EpaNIC despite greater critical illness blinding, bed availability, and variations across ICU
severity and higher mortality of patients, perhaps because physicians in assessing readiness for ICU discharge.
all infections were adjudicated by an independent Duration of the intervention was from day 1 to day 7 in all
committee. Despite the lower values for glycaemia and patients except those weaned off invasive mechanical
proportion of insulin-treated patients in the low group, ventilation and vasopressor support before day 7.
the frequency of hypoglycaemia was not different Duration of the acute phase of critical illness can vary
between the two groups. The association of the low across patients. No clinical or laboratory criteria are
strategy with important benefits, without adverse available to determine the end of the acute phase, and the
outcomes, supports targets as low as 6 kcal/kg per day for first 7 days after ICU admission is a well accepted range
calories and 0·2–0·4 g/kg per day for protein during the used in guidelines.5 Finally, we adjusted for multiplicity
acute phase of critical illness. in the interim analyses but not for multiple outcomes in
The mechanisms involved in clinical benefits from the final analysis. Both decreasing mortality and
calorie and protein restriction are unclear. Anorexia expediting recovery are central goals of critical care
during the acute phase of critical illness is considered an medicine. In previous randomised trials, nutritional
adaptive process with benefits including a heightened interventions affected time to readiness for ICU
immune response and decreases in metabolic distur­ discharge but not mortality.13,14 In the current trial, time to
bances.23 Macronutrient restriction might contribute to readiness for ICU discharge should be viewed as an
preserve the neuroendocrine response in acute critical efficacy outcome and mortality as a safety outcome.
illness.24 Evidence exists identifying autophagy as a key Adjusting for multiple outcomes would have required an
mechanism for safeguarding cellular integrity, notably even greater number of patients than the large sample
in the muscle, and therefore making a major contribution included, thus strongly limiting trial feasibility. It is
to recovery after severe critical illness.25 Increased worth noting that the best methods for taking multiple
macronutrient intakes could suppress autophagy, outcomes into account are debated.31
thereby decreasing the clearance of damaged cell Our trial also has important strengths. The design was
components.25,26 The higher frequencies of gastro­ pragmatic. More specifically, the feeding route during
intestinal and hepatic complications in our standard the acute phase was at the discretion of the bedside
group compared with the low group can be ascribable to physicians, based on previous trials showing no outcome
a greater mismatch between oxygen needs and supply.27 differences between the enteral and parenteral routes.21,22
Another possibly relevant finding is the higher blood Standard enteral and parenteral preparations routinely
glucose concentration in the standard group, although used in ICUs were administered. These preparations
the effects of blood glucose control remain controversial.28 ensure that both calorie and protein targets are easily
Although refeeding syndrome might be considered met, as shown by our data. Complements were not given,
given the higher proportion of patients with except for electrolytes and micronutrients when required.
hypophosphataemia in the standard group, any effect These conditions, combined with the large number of
would be limited given the absence of between-group participating centres, support the external validity of our
differences in blood potassium and magnesium findings. NUTRIREA-3 shows the benefits of limiting

610 www.thelancet.com/respiratory Vol 11 July 2023

Downloaded for Anonymous User (n/a) at Industrial University of Santander from ClinicalKey.com by Elsevier on July 12,
2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
 Articles

both calorie and protein intakes versus standard calorie 2 Hermans G, Van den Berghe G. Clinical review: intensive care unit
and protein targets during acute critical illness. Previous acquired weakness. Crit Care 2015; 19: 274.
3 Compher C, Chittams J, Sammarco T, Nicolo M, Heyland DK.
studies assessed restriction of either only calories or only Greater protein and energy intake may be associated with improved
protein. Of note, the calorie and protein targets differed mortality in higher risk critically ill patients: a multicenter,
markedly between our two trial groups. The standard multinational observational study. Crit Care Med 2017; 45: 156–63.
4 Villet S, Chiolero RL, Bollmann MD, et al. Negative impact of
group complied with guidelines and the restricted group hypocaloric feeding and energy balance on clinical outcome in ICU
with the hypothesis that intakes seen in individuals with patients. Clin Nutr 2005; 24: 502–09.
anorexia might provide benefits. The trial procedures 5 Singer P, Blaser AR, Berger MM, et al. ESPEN guideline on clinical
were rigorously followed and, in both groups, the calorie nutrition in the intensive care unit. Clin Nutr 2019; 38: 48–79.
6 Taylor BE, McClave SA, Martindale RG, et al. Guidelines for the
and protein intakes were near the targets. From day 8 provision and assessment of nutrition support therapy in the adult
onwards, the observed total calorie intakes were below critically ill patient: Society of Critical Care Medicine (SCCM) and
30 kcal/kg per day, indicating that overfeeding did not American Society for Parenteral and Enteral Nutrition (A.S.P.E.N.).
Crit Care Med 2016; 44: 390–438.
occur. Last, we included a well defined and representative 7 Reintam Blaser A, Starkopf J, Kirsimagi U, Deane AM. Definition,
population of critically ill patients who required at least prevalence, and outcome of feeding intolerance in intensive care:
invasive mechanical ventilation and vasoactive drugs and a systematic review and meta-analysis. Acta Anaesthesiol Scand 2014;
58: 914–22.
who were at high risk for death or protracted recovery 8 Casaer MP, Van den Berghe G. Nutrition in the acute phase of
and, therefore, likely to benefit from improved early critical illness. N Engl J Med 2014; 370: 2450–51.
nutritional support. 9 Preiser JC, van Zanten AR, Berger MM, et al. Metabolic and
nutritional support of critically ill patients: consensus and
In conclusion, for the nutritional support of patients at controversies. Crit Care 2015; 19: 35.
the acute phase of severe critical illness, calorie and 10 Target Investigators for the ACTG, Chapman M, Peake SL, et al.
protein restriction was superior to standard calorie and Energy-dense versus routine enteral nutrition in the critically ill.
protein intake, with fewer complications and a faster N Engl J Med 2018; 379: 1823–34.
11 Allingstrup MJ, Kondrup J, Wiis J, et al. Early goal-directed
recovery. nutrition versus standard of care in adult intensive care patients:
Contributors the single-centre, randomised, outcome assessor-blinded EAT-ICU
JR, GPl, J-BL, and ALG designed the study. JR, GPl, J-PM, LA, PA, NA, trial. Intensive Care Med 2017; 43: 1637–47.
JB, N-VB, LB, H-NB, DC, LC, AC, CC, MDa, VD, MDe, AD, JD, L-MD, 12 Doig GS, Simpson F, Finfer S, et al. Effect of evidence-based
OG, SG, LG, BG, YH, SJ, FL, CL, PL, BM, JM, OM, FM, VM, EM, M-AN, feeding guidelines on mortality of critically ill adults: a cluster
SN, JO, WP, GPi, J-PQ, FR, AR, JR, J-PR, FS, DS, MS, BS, FTa, NT, DT, randomized controlled trial. JAMA 2008; 300: 2731–41.
GT, NT-R, J-FT, FTi, PT, TV, IV, CV, SV, and J-BL approved the design of 13 Casaer MP, Mesotten D, Hermans G, et al. Early versus late
the study, coordinated individual sites, participated in the inclusion of parenteral nutrition in critically ill adults. N Engl J Med 2011;
365: 506–17.
study participants and collected the data. JR and ALG directly accessed
14 Fivez T, Kerklaan D, Mesotten D, et al. Early versus late parenteral
and verified all the data reported in the manuscript. ALG did the
nutrition in critically ill children. N Engl J Med 2016; 374: 1111–22.
statistical analysis. JR wrote the first draft of the manuscript with input
15 Casaer MP, Wilmer A, Hermans G, Wouters PJ, Mesotten D,
from J-BL and ALG. All authors had full access to the study data, revised
Van den Berghe G. Role of disease and macronutrient dose in the
the manuscript, and read and approved the final version before
randomized controlled EPaNIC trial: a post hoc analysis.
submission. All authors accept responsibility for submitting the final American journal of respiratory and critical care medicine 2013;
manuscript for publication. 187: 247–55.
Declaration of interests 16 Arabi YM, Aldawood AS, Haddad SH, et al. Permissive
We declare no competing interests. underfeeding or standard enteral feeding in critically ill adults.
N Engl J Med 2015; 372: 2398–408.
Data sharing 17 Rice TW, Wheeler AP, Thompson BT, et al. Initial trophic vs full
The datasets generated for this study will not be publicly available enteral feeding in patients with acute lung injury: the EDEN
owing to the limitations of participant consent and approvals in place randomized trial. JAMA 2012; 307: 795–803.
regarding data sharing between organisations involved in the study. 18 Marik PE, Hooper MH. Normocaloric versus hypocaloric feeding
The data will be held in the Nantes University Hospital, Nantes, France, on the outcomes of ICU patients: a systematic review and meta-
and will be used internally for secondary purposes. Any applications for analysis. Intensive Care Med 2016; 42: 316–23.
potential data sharing or collaboration should be made to the 19 Singer P, Cohen J. Nutrition in the ICU: proof of the pudding is in
corresponding author and will be considered. Study tools, including the the tasting. Intensive Care Med 2015; 41: 154–56.
protocol, consent forms, statistical analysis plan, definition of outcomes, 20 Reignier J, Le Gouge A, Lascarrou JB, et al. Impact of early low-
training materials, and regulatory documents are available upon calorie low-protein versus standard-calorie standard-protein feeding
request. on outcomes of ventilated adults with shock: design and conduct of
a randomised, controlled, multicentre, open-label, parallel-group
Acknowledgments trial (NUTRIREA-3). BMJ Open 2021; 11: e045041.
The NUTRIREA-3 study was supported by the Nantes University 21 Reignier J, Boisrame-Helms J, Brisard L, et al. Enteral versus
Hospital and funded by the Programme Hospitalier de Recherche parenteral early nutrition in ventilated adults with shock:
Clinique National 2017 of the French Ministry of Health (number a randomised, controlled, multicentre, open-label, parallel-group
PHRC-17-0213). We are indebted to A Wolfe, for assistance in preparing study (NUTRIREA-2). Lancet 2018; 391: 133–43.
and reviewing the manuscript; Carine Coffre and Frédérique Musset for 22 Harvey SE, Parrott F, Harrison DA, et al. Trial of the route of early
managing the database; and Diane Maugars and Manon Rouaud for nutritional support in critically ill adults. N Engl J Med 2014;
coordinating the study. We are grateful to the medical staff, nurses, and 371: 1673–84.
research nurses at the 61 participating centres for their valuable 23 Langhans W. Anorexia of infection: current prospects. Nutrition
contribution to the successful conduct of the study. 2000; 16: 996–1005.
24 Van den Berghe G. On the neuroendocrinopathy of critical illness.
References
Perspectives for feeding and novel treatments.
1 Puthucheary ZA, Rawal J, McPhail M, et al. Acute skeletal muscle
Am J Respir Crit Care Med 2016; 194: 1337–48.
wasting in critical illness. JAMA 2013; 310: 1591–600.

www.thelancet.com/respiratory Vol 11 July 2023 611

Downloaded for Anonymous User (n/a) at Industrial University of Santander from ClinicalKey.com by Elsevier on July 12,
2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.
 Articles

25 Tardif N, Polia F, Tjader I, Gustafsson T, Rooyackers O. Autophagy 29 Meyhoff TS, Hjortrup PB, Wetterslev J, et al. Restriction of
flux in critical illness, a translational approach. Sci Rep 2019; intravenous fluid in icu patients with septic shock. N Engl J Med
9: 10762. 2022; 386: 2459–70.
26 Choi AM, Ryter SW, Levine B. Autophagy in human health and 30 ARDS Clinical Trials Network, Wiedemann HP, Wheeler AP, et al.
disease. N Engl J Med 2013; 368: 651–62. Comparison of two fluid-management strategies in acute lung
27 Jenniskens M, Langouche L, Vanwijngaerden YM, Mesotten D, injury. N Engl J Med 2006; 354: 2564–75.
Van den Berghe G. Cholestatic liver (dys)function during sepsis and 31 Li G, Taljaard M, Van den Heuvel ER, et al. An introduction to
other critical illnesses. Intensive Care Med 2016; 42: 16–27. multiplicity issues in clinical trials: the what, why, when and how.
28 Ichai C, Preiser JC. International recommendations for glucose Int J Epidemiol 2017; 46: 746–55.
control in adult non diabetic critically ill patients. Crit Care 2010;
14: R166.

612 www.thelancet.com/respiratory Vol 11 July 2023

Downloaded for Anonymous User (n/a) at Industrial University of Santander from ClinicalKey.com by Elsevier on July 12,
2023. For personal use only. No other uses without permission. Copyright ©2023. Elsevier Inc. All rights reserved.