Table 1 Characteristics of included studies

1. Montecalvo 1992

38

Med/Surg Intensive Care Unit (ICU). Patients anticipated to require ≥ days of nutrition

Small bowel versus Gastric (no comment gastrokinetic drugs)

(ii) endoscopy

(iii) 14/19 (74%)

(iv) 0.3 +/- 0.9 days; mean (SD)

Randomised Control Trial (RCT) Multicenter (2 sites)

(ii) No

(i) Administration of nutrient

(ii) Pneumonia Mortality Duration of mechanical ventilation (MV) Duration of ICU stay

Blinded New Chest x-ray (CXR) changes + 3 of:

(i) sputum > 25 leukocytes < 10 epithelial and numerous bacteria;

(ii) sputum > 25 leukocytes < 10 epithelial and nosocomial pathogen present;

(iii) Temperature > 101.4° F (38.6°C); OR

(iv) White Cell Count > 10,000 (units)

Conceal: Uncertain Intension To Treat (ITT): No (analyzed according to location of feeding tube, rather than intention to treat) Blinding: No (Score 8)

2. Kortbeek 1999

80

Trauma Likely to require mechanical ventilation (MV) > 48 hrs, and enroled < 72 from admission, and Injury Severity Score (ISS) > 16 Exclusion traumatic pancreatitis and physiologic instability precluding transportation for fluoroscopic placement of a duodenal tube

Small bowel versus Gastric (no comment gastrokinetic drugs)

(ii) fluoroscopy

(iii) not reported

(iv) 30 minutes (15 to 120 minutes); med (range)

RCT Multicenter (2 sites)

(ii) No

(i) Administration of nutrient

(ii) Mortality Pneumonia Duration of MV Duration of ICU stay Duration of hospital stay

Blinded New CXR changes and 2 of: Temp > 38.5°C,;

(ii) WCC > 10,000 or < 3,000 (units);

(iii) purulent sputum; or

(iv) pathogenic bacteria cultured from bronchoalveolar lavage (BAL)

Conceal: Yes ITT: Yes Blinding: No (Score 11)

3. Taylor 1999

82

Traumatic Brain Injury (TBI), MV, Glasgow Coma Scale (GCS) score > 3, and at least one reactive pupil at some time during the first 24 hrs, as well as suitable for EN. Exclusion criteria included presence of any other organ failure

Small bowel versus Gastric (no comment gastrokinetic drugs)

(ii) 'blind'

(iii) 34%

(iv) not reported

RCT Single-center

(ii) No

(i) Neurological outcome at 6 months

(ii) Mortality (6 months) and pneumonia

Diagnosis of pneumonia not described

Conceal: Uncertain ITT: Yes Blinding: No (Score 10)

4. Kearns 2000

44

Medical ICU, MV and EN ≥ days Excluded patients with pancreatitis and ileus

Small bowel versus Gastric (gastrokinetic drugs allowed but not reported)

(ii) blind placement with metoclopramide

(iii) 21/21 (100%), but three required fluoroscopy

(iv) < 10 minutes except if required fluoroscopy - time not reported in those patients

RCT Single-center

(ii) No

(i) Pneumonia

(ii) Mortality, duration of ICU and hospital stay, and nutrient administered

Not blinded New CXR changes and and 2 of: Temp > 38.5°C;

(ii) WCC > 10,000 or

(iii) positive glucose test or blue discolouration in endotracheal secretions.

Conceal: Uncertain ITT: Yes Blinding: No (Score 9)

5. Minard 2000

27

Trauma GCS⁷ 3 to 10 Excluded patients with sepsis, kidney or respiratory failure or requiring vasoconstricting drugs

Small bowel versus Gastric (no comment gastrokinetic drugs)

(ii) endoscopy

(iii) 13/15 (87%)

(iv) not reported

RCT Single-center

(ii) No

(i) Length of ICU stay

(ii) Mortality, pneumonia duration of MV and hospital stay

Not blinded CXR changes+ Purulent sputum+ Temp > 101°F+ WBC > 12,000 OR BAL > 100,000 CFUs

Conceal: Uncertain ITT: No Blinding: No (Score 7)

6. Esparaza 2001

54

Medical ICU Inclusion criteria not reported

Small bowel versus Gastric (gastrokinetic drugs, erythromycin or metoclopramide- allowed in both groups. Reported as administration per patient-fed days)

(ii) electromyographic guided with erythromycin, metoclopramide and/or fluoroscopy

(iii) 21/27 (78%)

(iv) not reported

RCT Single-center

(ii) No

(i) Aspiration events based on radiolabelled 'meal' and gamma camera)

(ii) Mortality, administration of nutrient

Pneumonia not reported

Conceal: Uncertain ITT: Yes Blinding: No (Score 8)

7. Boivin 2001

80

Med/Surg ICU, MV in 98% Enteral Nutrition (EN) ≥72 hrs Excluded: pancreatitis, burns, severe head injury

Small bowel versus Gastric (all patients in both groups received erythromycin)

(ii) blind placement with erythromycin, with fluoroscopy for failed attempts

(iii) 28/40 (71%)

(iv) 304 minutes

RCT Single-center

(ii) No

(i) Administration of nutrient

(ii) Mortality

Pneumonia not reported

Conceal: Uncertain ITT: No Blinding: No (Score 6)

8. Day 2001

25

Primary neurological diagnosis and expected to receive EN for ≥ 3 days Patients were excluded who had gastroparesis

Small bowel versus Gastric (No comment regarding gastrokinetic drugs)

(ii) blind placement

(iii) not reported

(iv) not reported

RCT Single-center

(ii) No

(i) Administration of nutrient

(ii) Pneumonia

Did not report how pneumonia was diagnosed

Conceal: Uncertain ITT: Yes Blinding: No (Score 5)

9. Davies 2002

73

Med/Surg ICU Expected to receive EN ≥ days

Small bowel versus Gastric (gastrokinetic drugs excluded)

(ii) endoscopy

(iii) 33/34 (97%)

(iv) time to placement not reported. However, time to commencing nutrition was delayed in patients receiving small intestinal feeds (81 vs. 55)

RCT Single-center

(ii) No

(i) Intolerance to enteral nutrition

(ii) Mortality Pneumonia Duration of ICU stay

Not blinded Clinical criteria, CXR changes and microbiological data

Conceal: Uncertain ITT: No Blinding: No (Score 8)

10. Neumann 2002

60

Medical ICU In need of enteral nutrition excluded gastroparesis, ileus and pancreatitis

Small bowel versus Gastric (no comment gastrokinetic drugs)

(ii) blind placement, fluoroscopy as second line - required in 7/20 attempts.

(iii) not reported

(iv) not reported

RCT Single-center

(ii) No

(i) Efficacy of nutrition

(ii) Aspiration events

Pneumonia not reported

Conceal: Uncertain ITT: Yes Blinding: No (Score 6)

11. Montejo 2002

101

Mixed ICUs EN > 5 days Patients with gastroparesis allowed to enter

Small bowel versus Gastric (no comment gastrokinetic drugs)

(ii) technique depending on local expertise but included endoscopy (n = 18), fluoroscopic guidance (n = 12), blind technique (n = 15), or by echography (n = 5).

(iii) 100%

(iv) time to small intestine 21.0 ± 9.8 vs. gastric 5.3 ± 7.9 hrs

RCT Multi-center (14 ICUs)

(ii) No

(i) Pneumonia

(ii) Mortality Duration of ICU stay

Not blinded Diagnosed according to criteria described by the Centre for Disease Control

Conceal: Not sure ITT: Yes Blinding: No (Score 6)

12. Hsu 2009

121

Medical ICU EN > 3 days Excluded intractable vomiting, severe diarrhea, paralytic ileus and acute pancreatitis

Small bowel Versus Gastric (gastrokinetic drugs, such as metoclopramide, erythromycin, cisapride, allowed but not routinely administered, administered n = 20/62 (32%) gastric and 18/59 (31%) small intestine

(ii) blind placement with endoscopy for failed cases

(iii) not reported

(iv) not reported

RCT Single-center

(ii) No

(i) Nutrient administered

(ii) Mortality Pneumonia Duration of MV ICU and hospital stay

Blinded New CXR changes and one of: T > 38 or < 36 with no other recognized cause;

(ii) WCC > 12,000 or < 4,000; or

(iii) for adults ≥ 70 years and altered mental status at least two of new or change n purulent sputum, new cough or tachypnea, worsening gas exchange, and bronchial breath sounds

Conceal: Yes ITT: Yes Blinding: No (Score 9)

13. White 2009

108

Medical ICU MV > 24 hrs

Small bowel versus Gastric (gastrokinetic drugs, metoclopramide and erythromycin, administered for GRVs > 200 mL)

(ii) 'blind' with erythromycin

(iii) 40/50 (80%)

(iv) Not reported

RCT Single-center

(ii) Yes (prospective)

(i) Time to reach goal feed rate

(ii) Mortality Pneumonia Duration of MV and ICU stay

(iii) weight-based

Not blinded New fever Leukocytosis New CXR changes, increased pulmonary secretions and clinical pulmonary infection score (CPIS) > 6

Conceal: Yes ITT: Yes Blinding: No (Score 7)

14. Acosta-Escribano 2010

104

TBI on MV Expect EN required for ≥ 5 days Glasgow coma scale (GCS) < 9, APACHE II between15-30, sequential organ failure assessment (SOFA) < 6

Small bowel Versus Gastric (metoclopramide administered for two consecutive GRV > 200 mL)

(ii) blind or fluoroscopy

(iii) 47/50 (94%)

(iv) not reported

RCT Single-center

(ii) No

(i) Pneumonia

(ii) Administration of nutrient Mortality (30 day) Pneumonia Duration of MV, ICU and hospital stay

Not blinded CPIS > 6 required for diagnosis. However, microbiological data collected in all patients and only one patient diagnosed with pneumonia did not have a pathogen isolated from the lower respiratory tract

Conceal: No ITT: Yes Blinding: No (Score 9)

15. Davies 2012

181

Mixed ICUs within 72 hrs of admission Receiving MV Receiving opiate drug via infusion Gastric residual volume (GRV) > 150 ml or > 500 ml over 12 hrs

Small bowel versus Gastric (metoclopramide ≥ erythromycin prn)

(ii) Self-migrating + erythromycin

(iii) 79/92 (87%)

(iv) 15 (7 to 32) hours; median (IQR)

RCT Multi-center (17 sites)

(ii) Yes (prospective)

(i) Energy delivery

(ii) Mortality Pneumonia Duration of MV, ICU and hospital stay

Blinded Consensus panel of three clinicians, pneumonia diagnosed by at least two members based on temp, WCC, sputum, P/F ratio, microbiological results and CXR

Conceal: Yes ITT: Yes Blinding: No (Score 11)