Study reference

Study characteristics

Patient characteristics

Intervention (I)

Comparison / control (C)

Follow-up

Outcome measures and effect size

Comments

Strijbos, 2018

PS., study characteristics and results are extracted from the SR (unless indicated with (#))

SR and meta-analysis of  RCTs, retrospective reviews, cohort and  case-control studies.

Literature search up to May 2017

A: Silas, 2005

B: Laasch, 2003

C: Grant, 2009

D: Möller, 2009

E: La Nauze, 2012

F: Allen, 2013

G:Laskaratos, 2013

H: Rustom, 2006

I: Galaski, 2009

J:Blondet, 2010

K: Chio, 2004

L: McAllister, 2013

M: Desport, 2005

N: McDermott, 2015

O: Rio, 2010

P: Elliot, 1996

Study design:

A: Retrospective

B: Prospective, nonrandomized

C: Prospective, nonrandomized

D:  Retrospective

E:  Retrospective

F:  Retrospective

G:  Retrospective

H:  Retrospective

I:  Retrospective

J:  Retrospective

K:  Retrospective

L:  Retrospective

M:  Retrospective

N: Prospective

O: Retrospective

P: Prospective, nonrandomized

Setting and Country:

A: USA, single centre

B: UK, multi centre

C: UK, multi centre

D: Sweden, single centre

E: Australia, single centre

F: USA, single centre

G: UK, single centre

H: UK, single centre

I: Canada, single centre

J: France, single centre

K: Italy, single centre

L: UK, single centre

M: France, single centre

N: UK, multi centre

O: UK, single centre

P: UK, single centre

Source of funding and conflicts of interest#:

A: not reported

B: not reported

C: not reported

D: not reported

E: funding not reported, authors declared to have no conflict of interest

F:

G:

H: funding not reported, authors declared to have no conflict of interest

I: not reported

J:

K: funding not reported, authors declared to have no conflict of interest

L: funding not reported, authors declared to have no conflict of interest

M:

N: funded by MND association UK and the Sheffield institute for translational neuroscience, authors reported their interests (see article)

O: no funding, authors declared to have no conflict of interest

P: not reported

Inclusion criteria SR:

Randomized clinical trials and

retrospective

reviews/cohort/case-control studies were included.

Types of studies;

All eligible studies

Types of participants; Adults, all indications

Types of interventions;

PEG versus PRG

Types of outcome measures;

Tube related complications, infectious complications, procedure related mortality, 30day mortality.

(In discussion section it is mentioned that only studies are included which analysed data on a direct comparison between PEG and PRG. Also studies of low quality were excluded based on the ROBINS tool.)

Exclusion criteria SR:

Not reported

16 studies included

Important patient characteristics at baseline:

N (PRG/PEG)

A: 193/177

B: 50/25

C: 50/121

D: 94/12

E: 97/80

F: 51/57

G: 40/53

H: 28/40

I: 30/44

J: 22/21

K: 25/25

L: 89/21

M: 20/30

N: 124/163

O: 121/21

P: 45/33

Age in years (PRG/PEG)#

A: 63±14/68±15

B: 68 (27–86)/73 (19–96)

C: 65/61

D: 64 (20-92)/48 (29-73)

E: 61 (16-92)/61 (16-94)

F:

G:

H: 64.8/63.6

I: 65±19/55±21

J:

K: 68.9±9.5/65.1±10.3

L: not reported

M:

N:63.6±9.8/ 64.2±11.7

O:  64 (55-70), total group

P: not reported

Sex: (%, Male, PRG/PEG)#

A: 59.1/56.5

B: not reported

C: 77/65

D: 52/50

E: 71/66

F:

G:

H: 19/31

I: 70.5/63.3

J:

K: 48/52

L: not reported

M:

N: 51/55

O: not reported

P: not reported

Indication

A: Various

B: Various

C: Head and neck cancer

D: Various

E: Various

F: Motor neuron disease

G: Various

H: Head and neck cancer

I: Various

J: Motor neuron disease

K: Motor neuron disease

L: Head and neck cancer

M: Motor neuron disease

N: Motor neuron disease

O: Motor neuron disease

P: Various

Groups comparable at baseline?

A: differences with regard to indications for gastrostomy

B: differences with regard to indications for gastrostomy

C: differences with regard to tumour site and stage.

D: median age was higher in PRG group.

E: small differences with regard to indications for gastrostomy

F:

G:

H: no significant differences in demographic characteristics (other factors not reported by treatment group)

I: differences with regard to indications for gastrostomy

J:

K: No significant differences

L: no baseline characteristics reported

M:

N:significant differences between groups with regard to forced vital capacity and number of patients using non-invasive ventilation

O: baseline characteristics are not reported by treatment group

P: no baseline characteristics reported

Describe intervention:

percutaneous radiologic gastrostomy (PRG)

Describe  control: percutaneous endoscopic gastrostomy (PEG, pull method)

End-point of follow-up (#):

A: 108/174 days

B: not reported

C: 6 months

D: not reported

E:  367 (9-939)/491 (4-923) days

F:

G:

H: not reported

I: not reported

J:

K: not reported

L: 30 days postoperatively

M:

N: 12 months postoperatively

O: not reported

P: not reported

For how many participants were no complete outcome data available (#)?

(intervention/control)

A: not reported

B: not reported

C: complete data was available for 178 (76%) patients.

D: not reported

E: at 1 year, 44 patients were deceased. For 28 of them mortality data was missing.

F:

G:

H: not reportedI: not reported

J:

K: not reported

L: not reported

M:

N: no assessment at 3 months: n=9; no assessment at 12 months: n=93

O: not reported

P: not reported

30-day mortality

Effect measure: RD (95% CI):

A: not reported

B: not reported

C: not reported

D: -0.02 (-0.32-0.28)

E: -0.02 (-0.17-0.13)

F:  0.08 (-0.11-0.27)

G:  0.34 (0.14-0.55)

H: 0.19 (-0.05-0.43)

I: 0.26 (0.02-0.49)

J: -0.01 (-0.31-0.29)

K: 0.00 (-0.28-0.28)

L: not reported

M: -0.05 (-0.33-0.23)

N: 0.00 (-0.11-0.12)

O: 0.01 (-0.23-0.24)

P: -0.07 (-0.30-0.15)

Pooled effect (random effects model): 0.06 (95% CI -0.01 to 0.13) favoring PEG.

Heterogeneity (I²): 12%

Subgroup various population

(included studies: P, E, I, D, G)

Pooled effect (random effects model): 0.10 (95% CI -0.07 to 0.27) favoring PEG.

Subgroup Head & Neck cancer (included studies: H)

Pooled effect (random effects model): 0.19  (95% CI -0.05 to 0.43) favoring PEG.

Subgroup Motor Neuron disease

(included studies: K, M, O, J, F, N)

Pooled effect (random effects model): 0.01 (95% CI -0.07 to 0.09) favoring PEG.

Procedure related mortality

Effect measure: RR, RD, mean difference (95% CI):

A: 0.00 (-0.10-0.10)

B: not reported

C: 0.11 (-0.05-0.28)

D: 0.00 (-0.29-0.29)

E: not reported

F: -0.11 (-0.30-0.08)

G: 0.01 (-0.19-0.22)

H: 0.19 (-0.05-0.43)

I: 0.06 (-0.17-0.28)

J: 0.00 (-0.30-0.29)

K: 0.00 (-0.27-0.27)

L: not reported

M: 0.03 (-0.25-0.30)

N: 0.01 (-0.11-0.12)

O: -0.20 (-0.43-0.03)

P: 0.06 (-0.16-0.28)

Pooled effect (random effects model): 0.01 (95% CI -0.04 to 0.06) favoring PEG.

Heterogeneity (I²): 0%

Subgroup various population

(included studies: P, A, I, D, G)

Pooled effect (random effects model): 0.01  (95% CI -0.06 to 0.09) favoring PEG.

Subgroup Head & Neck cancer (included studies: H, C)

Pooled effect (random effects model): 0.14  (95% CI 0.00 to 0.27) favoring PEG.

Subgroup Motor Neuron disease

(included studies: K, M, O, J, F, N)

Pooled effect (random effects model):  -0.04 (95% CI -0.12 to 0.04) favoring PRG.

Infectious complications

Defined as peristomal infection and peritonitis.

Effect measure: RD (95% CI):

A: 0.15 (0.05-0.25)

B: -0.30 (-0.49- -0.10)

C: 0.16 (-0.01-0.32)

D: -0.15 (-0.45-0.15)

E: 0.01 (-0.14-0.16)

F: not reported

G: not reported

H: 0.24 (0.01-0.48)

I: -0.06 (-0.29-0.17)

J: not reported

K: not reported

L: -0.08 (-0.31-0.16)

M: 0.03 (-0.25-0.30)

N: 0.07 (-0.04-0.19)

O: 0.07 (-0.16-0.30)

P: 0.05 (-0.14-0.28)

Pooled effect (random effects model): 0.03 (95% CI -0.05 to 0.11) favoring PEG.

Heterogeneity (I²): 39.12%

Subgroup various population

(included studies: P, B, A, I, D, E)

Pooled effect (random effects model): -0.03 (95% CI -0.17 to 0.11) favoring PRG.

Subgroup Head & Neck cancer (included studies: H, C, L)

Pooled effect (random effects model): 0.11 (95% CI -0.06 to 0.28) favoring PEG.

Subgroup Motor Neuron disease

(included studies: M, O, N)

Pooled effect (random effects model): 0.07 (95% CI -0.03 to 0.17) favoring PEG.

Tube related complications

Defined as dislocation,

leak, obstruction, need for replacement, and defects (e.g., rupture, break down).

Effect measure: RD (95% CI):

A: 0.10 (0.00-0.20)

B: 0.16 (-0.08-0.40)

C: 0.16 (-0.01-0.32)

D: 0.05 (-0.24-0.34)

E: 0.47 (0.32-0.62)

F: not reported

G: not reported

H: 0.25 (0.00-0.49)

I: 0.05 (-0.19-0.28)

J: not reported

K: 0.00 (-0.28-0.28)

L: 0.06 (-0.18-0.30)

M: not reported

N: not reported

O: not reported

P: 0.17 (-0.05-0.40)

Pooled effect (random effects model): 0.16 (95% CI 0.06 to 0.26) favoring PEG.

Heterogeneity (I²): 59.83%

Subgroup various population

(included studies: P, B, A, I, D, E)

Pooled effect (random effects model): 0.18 (95% CI 0.03 to 0.33) favoring PEG.

Subgroup Head & Neck cancer (included studies: H, C, L)

Pooled effect (random effects model): 0.15 (95% CI 0.04 to 0.27) favoring PEG.

Subgroup Motor Neuron disease

(included study: K)

Pooled effect (random effects model): 0.00 (95% CI -0.28 to 0.28).

Author’s conclusion

Overall, no significant differences were found between PEG and PRG with regard to mortality or infectious complications.

Patients with PEG had lower risk of tube-related complications compared with PRG. This results from higher rate of tube dislocation

in the PRG group. Subgroup analysis revealed PEG to

be favorable over PRG in HNC patients based on lower rates of procedure-related mortality, infectious, and tube-related complications.

In MND, no differences were observed.

Considering the small differences in mortality and

complications between both techniques, the local experience and availability of the technique should be taken into account in the shared decision process. In order to improve the level of evidence, future studies should be performed in a randomized controlled manner, with uniform definitions for

complications and standardized follow-up. The issue of tumor seeding remains to be answered.

Sensitivity testing for infectious complications in HNC was performed, thereby excluding the study by McAllister (2013) because the authors expected underreporting by patients, nurses, and medical staff. RD for HNC increased from a nonsignificant 0.11 (−0.06 to 0.28) to significant 0.19

(0.05-0.32) in favor of PEG. In the overall study population,

RD increased from −0.03 (−0.05 to 0.11) to 0.04 (−0.04 to 0.12), both nonsignificant.

Antibiotics were only administered in PEG placement in three studies (McDermott, 2015 ;Galaski, 2009; Blondet, 2010) One did not properly report infectious complications (Blondet, 2010). The other 2 were excluded

in sensitivity analysis because of the expected influence on

the infectious complication rate. The RD in the overall

study population remained nonsignificant in the overall

pooled study population as well as in the subgroup.

Study reference

Study characteristics

Patient characteristics

Intervention (I)

Comparison / control (C)

Follow-up

Outcome measures and effect size

Comments

Park, 2019

Type of study:

Retrospective cohort study

Setting and country:

Five university hospitals, South Korea

Funding and conflicts of interest:

Funding is not reported, authors declared to have nothing to disclose.

Inclusion criteria:

All patients who underwent initial PEG or PRG tube insertion between January 2010 and December 2015 at five university hospitals in Korea.

Exclusion criteria:

Patients who

underwent gastrostomy and had a follow-up of <30 days were excluded.

N total at baseline:

Intervention: 94

Control: 324

Important prognostic factors²:

age (years± SD):

I: 66.2 ± 13.3

C: 66.7 ± 14.8

Sex (% M):

I: 58 (61.7)

C: 234 (72.2)

Indication (%)

Medical diseases

I: 2 (2.1)

C: 6 (1.9)

Neurological diseases

I: 56 (59.6)

C: 240 (74.1)

Surgical diseases

I: 28 (29.8)

C: 60 (18.5)

Tube diameter (Fr.± SD)

I: 14.3 ± 1.9

C: 21.3 ± 1.7

Antibiotic prophylaxis (%)

I: 81 (86.2)

C: 290 (89.5)

Groups comparable at baseline?

Groups are not fully comparable, with regard to sex, tube choice, tube diameter and procedure duration.

Percutaneous radiological gastronomy (PRG)

PRG tubes were inserted under fluoroscopic

guidance using a modified Seldinger technique by two experienced radiologists.

All patients were routinely sedated with intravenous midazolam (and/or propofol), 5–10 mL of 1% lidocaine was used

for local anaesthesia. The procedures were performed in the ICU, or in an endoscopic or radiological unit. All patients were placed in the supine position during PRG.

Percutaneous endoscopic gastronomy (PEG)

PEG tubes were

inserted using the pull-type or introducer-type technique by an experienced endoscopist with the assistance of another

endoscopist.

All patients were routinely sedated with intravenous midazolam (and/or propofol), 5–10 mL of 1% lidocaine was used

for local anaesthesia. The procedures were performed in the ICU, or in an endoscopic or radiological unit. All patients were placed in the supine position during PEG.

Length of follow-up (months (IQR):

At least ≥ 30 days

I: 400 (20–2287)

C: 335 (10–2122)

Loss-to-follow-up:

Intervention:

Not reported

Reasons: patients having follow-up <30 days were excluded

Control:

Not reported

Reasons: patients having follow-up <30 days were excluded

Incomplete outcome data:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Outcome measures and effect size (include 95%CI and p-value if available):

Overall complications (%)

I: 28 (29.8)

C: 88 (27.2)

P-value: 0.61

Mortality (%)

I: 20 (28.6)

C: 78 (28.3)

P-value:0.95

Wound infection, abscess (%)

I: 2 (2.1)

C: 18 (5.6)

P-value: 0.14

Pain (%)

I: 0 (0)

C: 7 (2.2)

P-value: 0.15

Peritoneal leakage (%)

I: 3(3.2)

C: 3 (0.9)

P-value: 0.10

Multivariate analysis (PEG versus PRG)

Total complications

OR (95%CI): 3.13 (0.46–21.34)

Major complications

Defined as aspiration, pneumonia, peritonitis, and massive haemorrhage

OR (95%CI): 1.31(0.02–77.80)

Minor complications

Defined as pneumoperitoneum,

peritoneal leakage, wound infection, tube dislodgement,

tube blockage, diarrhea, abdominal pain, PEG site ulcer, and ileus.

OR (95%CI): 1.98 (0.13–28.2)

Cherian, 2019

Type of study:

Retrospective cohort study

Setting and country:

Australia

Funding and conflicts of interest:

Funding is not reported, authors declared to have nothing to disclose.

Inclusion criteria:

All-cause gastrostomy

insertions at a quaternary Australian Hospital, Royal Brisbane and Womens’ Hospital (RBWH) between January 2012 and August 2015 with follow-up to 1 year.

Exclusion criteria:

Not reported

N total at baseline:

Intervention: 95

Control: 307

Important prognostic factors²:

age (years, IQR):

I: 65 (35-89)

C: 62 (18-91)

Sex (n, M):

I: 65

C: 242

Indication (n)

Head and neck cancer

I: 42

C: 211

Cancer (other)

I: 24

C: 20

No malignancy

I: 29

C: 76

Groups comparable at baseline?

Groups are not comparable, with regard to age, sex and indication for tube insertion.

Radiologically inserted gastrostomy (RIG)

Standard peri-procedural antibiotic use included 1-2g IV cephazolin or 0.5g vancomycin for patients allergic to cephalosporins or known to be colonised with MRSA.

Radiological tubes were performed using gastric

insufflation with three-suture gastropexy.

Percutaneous endoscopic gastrostomy (PEG)

Standard peri-procedural antibiotic use included 1-2g IV cephazolin or 0.5g vancomycin for patients allergic to cephalosporins or known to be colonised with MRSA.

Endoscopic tubes were inserted under deep sedation.

Length of follow-up:

12 months

Loss-to-follow-up:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Incomplete outcome data:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Outcome measures and effect size (include 95%CI and p-value if available):

Survival (%)

I: 67

C: 88

P= <0.0001

(none of the patients has been died due to procedure, according to the authors)

Minor complications (%)

Defined as bleed minor, minor infection (no hospital admission, PO antibiotics), pain

simple, pain strong oral, pain strong IV.

I: 54 (56.8)

C: 227 (73.7)

Major complications <48 hours of admission(%)

Defined as major infection – IV and hospital admission, pain requiring radiological

investigation:

I: 3 (3.2)

C: 12 (3.9)

Major complications and prolonged hospital admission (%)

Peritonitis requiring surgical intervention, perforation, pain requiring surgical intervention, hemoperitoneum no surgery. Major bleeding requiring endoscopy.

I: 5 (5.3)

C: 9 (2.9)

Overall P = 0.0085

PEGs are removed more often

than RIGs (P < 0.0001)

RIGs replaced more often than

PEGs P < 0.0001

Death in RIG group more often than PEG P < 0.0001 (no numbers reported)

Infection was independently analysed with minor

infection (no hospital admission, PO antibiotics) and

major infection (major infection – IV and hospital admission, peritonitis requiring surgical intervention) separately classified. There was no statistically significant

difference in infection rates between groups stratified to

time (Periop, 1,2,3,6,9 and 12 months), P = 0.2139, using an asymptotic generalised Cochran–Mantel–Haenszel test.

Skin-related complication rates were independently analysed (hypergranulation, minor skin degradation/pressure injury, major skin degradation, fistula developed) with no significant difference between groups

stratified to time (Periop 1, 2, 3, 6, 9 and 12 months),

P = 0.937.

Mahawongkajit, 2019

Type of study:

Retrospective cohort study

Setting and country:

Thailand

Funding and conflicts of interest:

Funding is not reported, authors declared to have nothing to disclose.

Inclusion criteria:

Patients

referred to the department of surgery with advanced esophageal

cancer and who underwent placement of feeding tube between January 2016 and December 2017, and who were

identified in the hospital electronic documentation system

Exclusion criteria:

Not reported

N total at baseline:

Intervention: 28

Control: 36

Important prognostic factors²:

age (years±SD):

I: 63.9 ± 9.7

C: 59.4 ± 12.0

Sex (%, M):

I: 82.1

C: 83.3

Staging (%)

IVA

I: 35.7

C: 44.4

IVB

I: 64.3

C: 55.6

Groups comparable at baseline?

Yes, baseline characteristics did not differ significantly between groups.

Percutaneous endoscopic gastrostomy (PEG, introducer, push)

PEGs were performed by introducer technique (IDEAL PEG Kit; MD-43430; Akita Sumitomo Bakelite Co. Ltd., Akita, Japan) under the clear focus of the

intra-gastric endoscopic view.

The introducer PEG was chosen for advanced esophageal cancer patients, except the patients who had previous

abdominal surgery, could not pass endoscopy to stomach or afford the introducer PEG tube and device price. (The PEG and open gastrostomy procedures are covered by health insurance but the introducer PEG tube and device did not include with public health insurance).

Open gastrostomy

The procedure of open gastrostomy was performed with the Stamm technique.

Length of follow-up:

At least 30 days

Loss-to-follow-up:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Incomplete outcome data:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Outcome measures and effect size (include 95%CI and p-value if available):

30-day mortality

I:0

C: 0

P=-

Pain score (mean±SD)

I: 2.4 ± 1.5

C: 5.9 ± 1.3

P= <0.001

Surgical site infection (%)

I: 0 (0)

C: 3 (8.3)

P= 0.083

Tube leakage (%)

I: 0 (0)

C: 5 (13.9)

P= 0.023

Tube dislodgment %)

I: 0 (0)

C: 2 (5.6)

P= 0.160

Strijbos, 2019

Type of study:

Retrospective cohort study

Setting and country:

Tertiary referral centre, the Netherlands

Funding and conflicts of interest:

Funding is not reported, authors declared to have nothing to disclose.

Inclusion criteria:

All adult patients receiving initial PRG (January 2010 until April 2016) and PEG or PEG-J (January 2008 until April 2016) placement at the Maastricht University Medical Centre. Patients

with PEG placements with additional duodenal/jejunal extension were included in the PEG group. Patients with initial and successful PEG or PRG were included.

Exclusion criteria:

In case of incomplete

follow-up data, patients were excluded from the analysis. When PEG or PRG was unsuccessful, impossible or if there was high risk, patients were excluded. Children and patients with multiple gastrostomy procedures were also excluded.

N total at baseline:

Intervention: 469

Control: 291

Important prognostic factors²:

age (years±SD):

I: 62,41±14,94

C: 63,38±10,97

Sex (%, M):

I: 65

C: 59.5

Indication (%)

ALS

I: 46 (9.8)

C: 8 (2.7)

Head and neck cancer

I: 328 (69.9)

C: 113 (38.8)

Other malignancy

I: 18 (3.8)

C: 21 (7.2)

Neurological disease

I: 13 (2.7)

C: 41 (14)

Comorbidities

(≥ 2 primary malignancies, %)

I: 41 (8.7)

C: 10 (3.4)

Groups comparable at baseline?

Groups were not fully comparable, differences with regard to indication for tube insertion and comorbidities.

Percutaneous radiological gastronomy (PRG)

PRG placement was performed according to standard placement

using a Wills-Oglesby Percutaneous

Gastrostomy Set.

Neither standard prophylactic antibiotics nor sedatives were

given. Only local anaesthetics were administered at the puncture site. The day after placement, saline fluids were administered through the tube for 3 to 6 hours. Feeding was started thereafter and the patient was discharged. Ten to 14 days after the procedure, patients returned to the radiology department to remove the anchors.

Percutaneous endoscopic gastrostomy (PEG)

Conscious sedation using midazolam (low-dose, mean 2.5mg) and a fast-acting opioid was administered. Oral amoxicillin/clavulanic

acid 1200mg was given 30 minutes before the procedure (in case of allergy, a substitute was given). PEG placement

was performed according to the standard Pull-method. In several cases (n = 26),

a jejunal extension tube was placed through the PEG tube, and pulled distally from the papilla of Vater with a grasping forceps.

Afterwards, patients were observed for at least 1 hour at the day care unit. Immediately after placement, water was administered through the tube. Feeding was started the next day.

Length of follow-up:

Loss-to-follow-up:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Incomplete outcome data:

Intervention:

None

Reasons: In case of incomplete follow-up data, patients were excluded from the analysis.

Control:

None

Reasons: In case of incomplete follow-up data, patients were excluded from the analysis.

Outcome measures and effect size (include 95%CI and p-value if available):

30-day mortality (%)*

I: 5.1

C: 10.7

OR (95%CI):  0.771 (0.374 – 1.590)

P= 0.481

Procedure related mortality (%)

I: 2 (0.4)

C: 5 (1.7)

OR (95%CI): not reported

P= 0.113

Early tube related complications**

Including dislocation, leak, blockage

I: 124 (26.4)

C: 8 (2.7)

OR (95%CI): 0.061 (0.026-0.139)

P= <0.001

Late tube related complications (%)**

Including dislocation, leak, blockage

I: 139 (29.6)

C: 23 (7.9)

OR (95%CI): 0.252 (0.155-

0.411)

P= <0.001

Early pain (requiring conservative treatment, %)

I: 43 (9.2)

C: 11 (3.8)

OR (95%CI): 0.389 (0.197 – 0.768)

p-value: 0.005

Late pain (requiring conservative treatment, %)

I: 1 (0.3)

C: 5 (1.1)

OR (95%CI): not reported

P= 0.415

Early peristomal infection

I: 7 (1.5)

C: 5 (1.7)

OR (95%CI): not reported

P= 0.808

Early peritonitis (%)

I: 2 (0.4)

C: 0 (0)

OR (95%CI): not reported

P= 0.527

Early abscess (%)

I: 2 (0.4)

C: 0 (0)

OR (95%CI): not reported

P= 0.527

Late abscess (%)

I: 2(0.4)

C: 1 (0.3)

OR (95%CI):

P= 1

*corrected for age, gender, weight, diagnosis, and comorbidities.

**corrected for baseline differences.

Vidhya, 2018

Type of study:

Retrospective cohort study

Setting and country:

Medical centre, Australia

Funding and conflicts of interest:

No funding received, authors declared to have nothing to disclose.

Inclusion criteria:

Patients undergoing initial PEG and RIG insertion at a tertiary health institution between February 2013 and February 2015.

Exclusion criteria:

Not reported

N total at baseline:

Intervention: 52

Control: 85

Important prognostic factors²:

age (years, median):

I: 64

C: 65

Sex (%, M):

I: 65.4

C: 63.5

Indication (n)

Head and neck cancer

I: 21 (40.4)

C: 30 (35.3)

Oesophageal cancer

I: 8 (15.4)

C: 0

CVA

I: 11 (21.2)

C: 27 (31.8)

Neuromuscular pathology

I: 0

C: 9 (10.6)

Trauma

I: 3 (5.8)

C: 6 (7.1)

Other

I: 9 (17.3)

C: 13 (15.3)

Prophylactic Antibiotics*

I: 2 (3.8)

C: 50 (58.9)

*cases in which use was documented, others are not given or unknown.

Groups comparable at baseline?

Groups are not fully comparable, differences with regard to use of prophylactic antibiotics and indication for tube insertion.

Radiologically inserted gastrostomy (RIG)

All patients were fasted for 12 h prior to the procedure. All PEG insertions were performed by a gastroenterologist using light sedation (midazolam ± fentanyl) or general anaesthesia and

occurred in the endoscopy suite or emergency operating theatre. As

per standard practice, a Kimberly Clark gastrostomy feeding tube with an internal flange tip PEG was placed under direct vision,

using a pull through technique.

Percutaneous endoscopic gastrostomy (PEG)

All patients were fasted for 12 h prior to the procedure. Hospital guidelines recommend pre procedure antibiotics (2G IV Cephazolin) prior to PEG gastrostomy insertion. An interventional radiologist in the fluoroscopic procedural

suite performed all RIG insertions with an 18F balloon tipped Cook

Medical Gastrostomy Tube. The procedure was performed using

local anaesthetic and light sedation (midazolam ± fentanyl).

Length of follow-up:

At least until discharge, mortality data was derived from a state based register.

Loss-to-follow-up:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Incomplete outcome data:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Gastrostomy discontinuation data were available for 77/137 patients. There will be patients lost to follow-up/have incomplete data due to loss of correspondence and follow up due to transfer of patients to other centres including rehabilitation

services for further care.

Outcome measures and effect size (include 95%CI and p-value if available):

30-day mortality (%)

I: 4 (7.7%)

C: 4 (4.7%)

P= 0.50

(none of the patients had been died due to the procedure)

1 year mortality (%)

I: 26 (50%)

C: 20 (23.5%)

P= <0.05

Procedure-related mortality (%)

I: 0

C: 0

Tube dislodgement (%)

I: 14 (26.9)

C: 2 (2.4)

P=<0.001

Failed insertion

I: 2

C: 6

*patients with failed PEG or RIG insertion subsequently underwent respectively successful RIG or PEG.

Tube discontinuation status %)

Removed

I: 12 (23.1)

C: 21 (24.7)

Ongoing

I: 10 (19.2)

C: 34 (40)

Unknown

I: 30 (57.7)

C: 30 (35.3)

Stoma site infection (%)

I: 7 (13.5)

C: 9 (10.6)

P= 0.62

Peritonitis (%)

I: 1 (1.9)

C: 2 (2.4)

P= 0.72

Ljungdahl, 2006

Type of study:

Prospective, randomized trial

Setting and country:

University hospital, Sweden

Funding and conflicts of interest:

Funding nor disclosures are reported.

Inclusion criteria:

Patients with impaired swallowing

and a need for long-term (>4 weeks) enteral feeding, irrespective of

the cause, were considered for this study. For inclusion, both techniques

(PEG and SG) had to be feasible.

Exclusion criteria:

Patients with previous surgery

in the upper gastrointestinal tract and those for whom endoscopy was

not possible because of obstructions from tumors in the pharyngoesophageal

region were excluded.

N total at baseline:

Intervention: 35

Control: 35

Important prognostic factors²:

age (years, IQR):

I: 69 (22–82)

C: 65(19–88)

Sex (n, M):

I: 21

C: 12

Indication (n)

Stroke

I: 17

C: 13

Neurologic disease

I: 10

C: 13

Oropharyngeal cancer

I: 4

C: 7

Cerebral trauma

I: 4

C: 2

Groups comparable at baseline?

Yes, baseline characteristics did not differ between treatment groups.

Percutaneous endoscopic gastrostomy (PEG)

Preoperatively, all the patients were administered an intravenous single 1.5-g dose of cefuroxime.

Feeding through the gastrostomy tube was delayed 24h for all the patients, according to our routines.

An upper gastrointestinal endoscopy was carried out to confirm normal

anatomy and stomach emptying. After light sedation, the appropriate

puncture site was located. Local anesthetics were applied after sterile

washing, and the puncture cannula was advanced into the stomach

under endoscopic control. The guidewire was inserted through the cannula, grasped with biopsy forceps, and drawn out together with the

gastroscope. The oral end of the wire was attached to the PEG tube, which was pulled into the stomach

and out through the abdominal wall. The tube was secured, and

the dressings were applied. In selected cases only, the endoscope was

reintroduced to verify that the tube was correctly positioned.

Surgical gastrostomy (SG)

Preoperatively, all the patients were administered an intravenous single 1.5-g dose of cefuroxime. Feeding through the gastrostomy tube was delayed 24h for all the patients, according to our routines.

The SG procedure was performed through a short upper midline incision. Two purse-string sutures were placed around the intended

entrance site on the stomach wall, and a 22-Fr gastrostomy tube was introduced through a

stab wound in the left subcostal area. The tube was inserted into the

stomach, and by tying of the purse-string sutures, a short serosal tunnel was created over the tube. Finally, the adjacent part of the stomach was attached to the abdominal wall by interrupted sutures,

and a skin suture was secured the tube at the exit site.

Length of follow-up:

7 to 30 days after the procedure, concerning mortality patients were followed for at least 6 months.

Loss-to-follow-up:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Incomplete outcome data:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Outcome measures and effect size (include 95%CI and p-value if available):

Procedure related mortality (%)

I: 1 (2.9)

C: 2 (5.7)

Perioperative mortality (%)

I: 0

C: 0

30-day mortality (%)

I: 2 (5.7)

C: 5 (14.3)

P= 0.238

30-day complications (%)

I: 15 (42.9)

C: 26 (74.3)

P=0.007

Major complications

(P=0.562)

Pneumonia (%)

I: 0

C: 2 (5.7)

Local peritonitis (%)

I: 1 (2.9)

C: 0

Minor complications

(P=0.054)

Peristomal infection (%)

I: 6 (17.1)

C: 6 (17.1)

Stomal leakage (%)

I: 4 (11.4)

C: 7 (20.0

Dislodged tube (%)

I: 2 (5.7)

C: 6 (17.1)

Bankhead, 2005

Type of study:

Retrospective cohort study

Setting and country:

University hospital, USA

Funding and conflicts of interest:

Funding nor disclosures are reported.

Inclusion criteria:

Patients undergoing gastrostomy tube placement

in the operating room at Temple University

Hospital between July 1, 2000, and June 30, 2001,

were included in the study.

Exclusion criteria:

All patients who had previous gastric surgery or a concomitant intraabdominal

procedure at the time of gastrostomy tube placement

were not included.

N total at baseline:

Intervention: 23

Control 1: 39

Control 2:29

Important prognostic factors²:

age (years, median):

I: 57.1±15.9

C1: 56.1±18.3

C2: 60.5±16.5

Indication

Ventilator-dependent respiratory failure (n=45), dysphagia (n=30), head and neck cancer (n=9) and altered mental status (n=7).

Groups comparable at baseline?

Not fully clear, since not all relevant baseline characteristics were reported for different treatment groups.

Percutaneous endoscopic gastrostomy (PEG)

A preoperative dose of prophylactic antibiotics

was administered intravenously (IV) to all patients.

The PEG method used was the Ponsky, or “pull,” technique. This method involved the execution of a complete esophagogastroduodenoscopy (EGD), followed by insufflation of the stomach with air,

transillumination of the anterior abdominal wall,

piercing of the anterior abdominal wall with a needle

under endoscopic guidance, and passage of a wire through this needle into the stomach. The wire was then captured with a polypectomy snare and

pulled up through the esophagus to exit the patient’s mouth. The enteral feeding tube was tied to the wire and pulled down through the esophagus into the

stomach to exit the abdomen at the point of needle

puncture.

Surgical gastrostomy (open or laparoscopic (LAP))

A preoperative dose of prophylactic antibiotics

was administered intravenously (IV) to all patients.

LAP (=C1) insertion was performed through 2 port

sites, one in the umbilicus introduced after induction

of pneumoperitoneum via a Verees needle and

another in the left upper quadrant of the abdomen.

The 10-mm umbilical port allowed for the camera

placement and the 5-mm left upper quadrant port

for grasping of the stomach and tube placement.2

Using this technique, the stomach was insufflated

with air via a nasogastric tube and then grasped and

pulled up through the 5-mm port site. The stomach

was then held against the anterior abdominal wall

by placing “T” fasteners.  Once the stomach was secured to the

abdominal wall, a 5-mm trochar was exchanged for a dilator, which in turn was exchanged for dilators of

larger sizes. Once the track was dilated up to 14 Fr,

a balloon tube was placed through a peel-away

introducer.

The OPEN (=C2) procedure was performed through a

laparotomy by way of the Stamm technique. After

exposure through a small (5 cm) upper midline

incision, pursestring sutures were placed on the

anterior wall of the stomach low in the body and close to the greater curvature. The feeding tube was drawn through a stab wound on the abdominal wall and then inserted into the stomach. The stomach was fixed to the abdominal wall by tacking sutures that traverse the seromuscular layer of the stomach and the peritoneum around the exit site of the catheter.

Length of follow-up:

Patients were followed until postoperative day 6.

Loss-to-follow-up:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Incomplete outcome data:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Outcome measures and effect size (include 95%CI and p-value if available):

Insertion complications (%)

I: 1 (4.3)

C1: 3 (7.7)

C2: 0

Mortality (peri- and postoperative, %)

I: 0 (0)

C1: 0 (0)

C2: 0 (0)

Maintenance complications (%)

I: 0 (0)

C1: 3 (7.7)

C2: 0 (0)

Enteral complications

I: 6 (26.1)

C1: 9 (23.1)

C2: 5 (17.2)*

Of which one patients reported pain after bolus.

Oliveira, 2016

Type of study:

Retrospective cohort study

Setting and country:

Tertiary centre, Portugal

Funding and conflicts of interest:

Funding nor disclosures are reported.

Inclusion criteria:

SGs were identified

using the procedure codes ICD-9 (43.19 - Other gastrostomy), and PEG through clinical files from the Enteral Feeding Team.

Exclusion criteria:

Patients under 18 years and with incomplete or

misclassified files were excluded.

N total at baseline:

Intervention: 509

Control1: 8

Control2: 26 *

*of which 8 were excluded due to incomplete records.

Important prognostic factors²:

age (years, median):

I: 63 ± 15.7

Ctotal: 59.6±14.8.

Sex (n, M):

I: 346

C1: 8

C2: 13

Indication (%)

Head and neck cancer

I: 37.3

C1: 100

C2: 22.2

Neurological dysphagia

I: 52

C1: 0

C2: 0

Drainage

I: 0.6

C1: 0

C2: 44.5

Other

I: 10

C1: 0

C2: 33.3

Groups comparable at baseline?

Groups were not fully comparable, differences with regard to indication of tube insertion.

Percutaneous endoscopic gastrostomy (PEG)

Surgical gastrostomy (OPEN, C2 versus laparoscopic (LAP, C1))

Length of follow-up:

At least 30 days

Loss-to-follow-up:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Incomplete outcome data:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Outcome measures and effect size (include 95%CI and p-value if available):

30-day mortality (%)

I: 68 (13.4)

C1: 0 (0)

C2: 6 (33.3)

P=0.154

(one patient in the PEG group died due to the procedure, compared to none in the LAP group. For the OPEN group, it was unclear whether deaths were due to the procedure)

Complication rate (%)

I: 12.1

C1: 0

C2:.27.8

P=0.069

Of the patients submitted to PEG, three suffered dislocation of the tube into the peritoneum with peritonitis due to leakage of gastric contents. One patient died as a consequence

of this complication. Seven other patients suffered dislocation of the tube into

the abdominal wall creating “buried bumper syndrome”,

all of which were solved with endoscopic replacement of

the PEG tube.

Complications occurred in five patients submitted

to OPEN surgery (27.8%) consisting of two cases of dehiscence (11.1%) and three with peristomal infection (16.7%).

OPEN gastrostomy was preserved for more complex situations in which gastrostomy was a secondary procedure, frequently for drainage purposes, when nutritional support was provided either by jejunostomy or total parenteral nutrition.

Edelman, 1998

Type of study:

Retrospective cohort study

Setting and country:

Medical center, USA

Funding and conflicts of interest:

Funding nor disclosures are reported.

Inclusion criteria:

Patients who underwent PEG or LAP between December 1990 and December 1992.

Exclusion criteria:

Not reported

N total at baseline:

Intervention: 17

Control: 14

Important prognostic factors²:

age (years, mean):

I: 81 (43-97)

C: 67 (20-94)

Sex (n, M):

I: 10

C: 9

Indication (n)

The indications for gastrostomy were the same for the two groups and were inability to eat, malnutrition,

and/or recurrent aspiration. However, in group 2 there were four patients who had failed at an attempted PEG; 8 patients who were not candidates for gastroscopy

because of head and neck or esophageal tumors (five), esophageal obstruction from radiation (two), or

hiatal hernia (one); and 2 patients who, because of surgeon's preference, while a concomitant tracheostomy

was being performed in the middle of the night,

had a LAP.

Groups comparable at baseline?

Groups were not fully comparable, since the LAP group had more patients with cancer.

Percutaneous endoscopic gastrostomy (PEG)

PEG was performed utilizing a modification of the Gauderer-

Ponsky method (7) as follows: With the patient in the supine, head elevated position, the endoscope was inserted into the stomach and the light was visualized through the abdominal wall. A palpating

finger was seen endoscopically and the abdominal skin was infiltrated

with local anesthesia. An 18-gauge needle was inserted into the stomach and grasped with a snare. A guidewire was inserted through the needle into the stomach; the snare was loosened slightly and

then resecured around the guidewire. The endoscope, snare, and guidewire were then withdrawn as a unit. Using the guidewire as a

tether, the tapered gastrostomy tube was pushed over the wire into the mouth and guided into the stomach and out the abdominal wall.

Alternatively, a pull technique was used, attaching a looped wire to the tube and pulling the gastrostomy through the abdominal wall. The endoscope was then reintroduced into the stomach, and under direct vision, the tube was snugged up against the gastric wall. The tube was then loosened from the wall by 1 cm, checked for ability to rotate, and secured to the abdominal wall by a bolster. The procedure involved a complete esophagogastroduodenoscopy (EGD) in all cases. Feeding was begun on the evening of the operative day.

Laparoscopic gastrostomy (LAP)

LAP was done using the technique described by Edelman and Unger. Under local anesthesia, a pneumoperitoneum was created by insufflation of 2-3 L of carbon dioxide through a Verres needle.

A 5-mm umbilical trocar was inserted and a 5-mm laparoscope was used to visualize the abdominal contents. Adhesions were lysed and metastatic or concomitant disease was identified. A second 5-ram trocar was placed in the right upper or left lower abdomen for an

atraumatic grasper to hold and elevate the midbody of the stomach. Beneath the left costal margin, a 7-cm, 18-gauge needle was inserted through the skin and into the stomach. A J-wire was threaded via the needle into the stomach. Sequential 12-French and 14-French dilators were placed over the guidewire before a 17-French peelaway sheath was inserted into the stomach. A 14-French Silastic balloon catheter was slid into the stomach and the balloon was inflated. The peel-away sheath was removed and the catheter was

secured in place after pulling the stomach up to the abdominal wall.

Alternatively, two to three T-fasteners were used to fix the stomach to the abdominal wall. An 18-gauge needle was introduced into the

stomach under direct vision and a preloaded T-fastener (COOK Surgical, Bloomington, IN) was pushed by a guidewire. Feeding was begun the following day.

Length of follow-up:

At least 30 days, since 30-day mortality is reported.

Loss-to-follow-up:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Incomplete outcome data:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Outcome measures and effect size (include 95%CI and p-value if available):

30-day mortality (%)

I: 4 (23)

C: 5 (35)

(none of the deaths were due to the procedure)

Other complications

Tube removal due to pain

I: 1

There were no wound infections, gastrocolic fistulae, bleeding, stomal leaks, or perforations in either group.

Ho, 1999

Type of study:

Retrospective cohort study

Setting and country:

Medical centre, USA

Funding and conflicts of interest:

Funding nor disclosures are reported.

Inclusion criteria:

Patients who underwent gastrostomy between January 1990 and June 1995.

Exclusion criteria:

Not reported

N total at baseline:

Intervention: 82

Control1: 60

Control2: 214

Important prognostic factors²:

age (years, median):

I: 52

C1: 50

C2: 44

Sex (n, M):

I: 57

C1: 35

C2: 132

Indication (%)

Head and neck cancer

I: 29

C1: 32

C2: 11

Trauma/burn

I: 22

C1: 40

C2: 25

Oesophageal cancer

I: 0

C1: 8

C2: 10

Neurologic disorder

I: 35

C1: 3

C2: 11

Malnutrition

I: 1

C1: 3

C2: 14

Groups comparable at baseline?

Groups were not fully comparable, differences with regard to indication of tube insertion.

Percutaneous endoscopic gastrostomy (PEG)

All patients received preoperative intravenous antibiotics, either prophylactically or as part of ongoing therapy. All cases were performed in the operating room by surgeons.

PEG was performed mostly with local anaesthesia

and intravenous sedation using the push technique. Needle-placed Cope or Brown-Mueller T-fasteners were used to secure the gastrostomy tube.

.

Surgical gastrostomy (OPEN, C2 versus laparoscopic (LAP, C1))

All patients received preoperative intravenous antibiotics, either prophylactically or as part of ongoing therapy. All cases were performed in the operating room by surgeons.

The LAP was placed using two 5-mm ports and needle-placed

T-fasteners. Open GT was performed using a double purse string technique.

An 18-Fr gastrostomy tube was employed in most cases.

Length of follow-up:

Not reported

Loss-to-follow-up:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Incomplete outcome data:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Outcome measures and effect size (include 95%CI and p-value if available):

Perioperative mortality (%)

I: 4.9

C1: 5.3

C2: 4.2

P=0.87

7-day mortality (%)

I: 2 (2.4)

C1: 1 (1.7)

C2: 6 (2.8)

P=0.88

8-30 day mortality (%)

I: 2 (2.4)

C1: 2 (3.3)

C2: 3 (1.4)

Perioperative morbidity – wound infection (%)

I: 3 (3.7)

C1: 0

C2: 7 (3.3)

P=0.34

Perioperative morbidity – peritonitis (%)

I: 0

C1: 0

C2: 1 (0.5)

P= 0.72

Long term complications – chronic gastrostomy infection (%)

I: 9 (11)

C1: 2 (3.3)

C2: 10 (4.7)

P= 0.078

Long term complications – stoma leakage (%)

I: 7 (8.5)

C1: 5 (8.3)

C2: 13 (6.2)

P= 0.69

Long term complications – tube dislodgement (%)

I: 12 (5.7)

C1: 4 (6.7)

C2: 2 (2.4)

P= 0.44

Rustom, 2006

Type of study:

Retrospective cohort study

Setting and country:

Regional centre for head and neck cancer, UK.

Funding and conflicts of interest:

Funding is not reported, authors declared to have nothing to disclose.

Inclusion criteria:

Head and neck cancer patients receiving gastrostomy tube between February 2002 and February 2005.

Exclusion criteria:

When medical records were not available at the time of the study, patients were excluded.

N total at baseline:

Intervention1: 40

Intervention2: 28

Control: 10

Important prognostic factors²:

age (years, median):

I1: 63.6

I2: 64.8

C: 70.3

Sex (%, M):

I1: 31

I2: 19

C: 8

Groups comparable at baseline?

No significant differences with regard to demographic characteristics (information with regard to tumour site etc. is not reported by treatment groups).

Percutaneous endoscopic gastrostomy (PEG, I1) or percutaneous radiologically inserted gastrostomy (RIG, I2)

All patients received two grams of intravenous

cefotaxime or 1.2 g of co-amoxiclav one hour prior

to the procedure (teicoplanin was an alternative in

cases of penicillin allergy), except for patients receiving

surgical gastrostomy tubes, whose antibiotic

depended on the individual surgeon’s choice. Post-operatively, patients were managed according

to a standard protocol.

Surgical gastrostomy (SG), including open and laparoscopic procedures.

All patients received two grams of intravenous

cefotaxime or 1.2 g of co-amoxiclav one hour prior

to the procedure (teicoplanin was an alternative in cases of penicillin allergy), except for patients receiving surgical gastrostomy tubes, whose antibiotic depended on the individual surgeon’s choice. Post-operatively, patients were managed according to a standard protocol.

Length of follow-up:

Until 18 months from day of tube insertion.

Loss-to-follow-up:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Incomplete outcome data:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Outcome measures and effect size (include 95%CI and p-value if available):

30-day mortality (%)

I1: 0 (0)

I2: 2 (7)

C: 1 (10)

(none of the deaths in the PEG/I1 group were due to the procedure, compared to 1 in the SG/C group and 2 in the PRG/I2 group)

Major complications (%)

Defined as peritonitis, gastro-intestinal bleeding, and pulmonary, cardiac and neurological events.

I1: 0 (0)

I2: 3 (11)

C: 1 (10)

Of which e.g.:

Peritonitis

I1: 0

I2: 3

C: 0

Minor complications (%)

Defined as peri-stomal

leakage, superficial cellulitis, tube dislodgement,

tube blockage or any combination of these.

I1: 16 (40)

I2: 12 (43)

C: 4 (40)

Of which e.g.:

Peristomal leakage (%)*

I1: 7

I2: 5

C: 3

Tube dislodgement*

I1: 2

I2: 6

C: 2

*Of which some had a combination of minor complications.

Bailey, 1992

Type of study:

Retrospective cohort study

Setting and country:

Medical centre, USA

Funding and conflicts of interest:

Funding nor disclosures are reported.

Inclusion criteria:

Patients with head and neck cancer who received feeding gastrostomies.

Exclusion criteria:

Not reported

N total at baseline:

Intervention: 30

Control1:17

Control2: 28

Important prognostic factors²:

Total group

Age (years):

61

Sex (%, M):

74.7

Groups comparable at baseline?

Not clear, since baseline characteristics are not reported by treatment group.

Percutaneous endoscopic gastrostomy (PEG)

Surgical gastrostomy (either cuffed Silastic gastrostomy (CSG, C1) or Stamm technique (Stamm, C2)).

The Stamm technique was performed with a Malecot (n=10), foley (n-14) or DePezzor (n=4) catheter.

Length of follow-up:

Not reported

Loss-to-follow-up:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Incomplete outcome data:

Intervention:

Not reported

Reasons: n.a.

Control:

Not reported

Reasons: n.a.

Outcome measures and effect size (include 95%CI and p-value if available):

Mortality (n)

I:9 (30)

C1: 3 (17.6)

C2: 6 (21)

(in both groups two of the deaths were related to the procedure)

Total complications (n)

I: 24

C1: 3

C2: 45

Tube site infection (n)

I: 0

C1: 0

C2: 3

External leak (n)

I: 8

C1: 0

C2: 15

Failure to function

I: 3

C1: 0

C2: 5

. 

Hospital records were randomly selected.

Study reference

(first author, publication year)

Describe method of randomisation

Bias due to inadequate  concealment of allocation?

(unlikely/likely/unclear)

Bias due to inadequate blinding of participants to treatment allocation?

(unlikely/likely/unclear)

Bias due to inadequate blinding of care providers to treatment allocation?

(unlikely/likely/unclear)

Bias due to inadequate blinding of outcome assessors to treatment allocation?

(unlikely/likely/unclear)

Bias due to selective outcome reporting on basis of the results?

(unlikely/likely/unclear)

Bias due to loss to follow-up?

(unlikely/likely/unclear)

Bias due to violation of

intention to treat analysis?

(unlikely/likely/unclear)

Ljungdahl, 2006

Patients were randomly assigned to PEG or SG by sealed envelopes opened after their inclusion in the

study. The randomization was nonstratified, and all the envelopes were

prepared before the start of the study.

Unlikely

Sealed envelopes were used.

Unclear

Blinding of participants was not possible.

Unclear

Blinding of care providers was not possible.

Unclear

It is not reported whether outcome assessors were blinded. 

Unclear

Outcome measures are not specified in the method section; it is only reported that they will evaluate procedure duration, technical success and complications.

Unclear

Loss to follow-up is not reported.

Unlikely

Patients included in the analysis are exactly those who were randomized into the trial.

Study reference

(first author, year of publication)

Bias due to a non-representative or ill-defined sample of patients?

(unlikely/likely/unclear)

Bias due to insufficiently long, or incomplete follow-up, or differences in follow-up between treatment groups?

(unlikely/likely/unclear)

Bias due to ill-defined or inadequately measured outcome?

(unlikely/likely/unclear)

Bias due to inadequate adjustment for all  important prognostic factors?

(unlikely/likely/unclear)

Park, 2019

Unclear

The choice of the technique was based on service availability, technical issues, and physician’s preference.

Unclear

Follow-up was at least 30 days and follow-up time was not different between groups.

Unclear

Outcome measures were well defined, but due to the retrospective character of the study it is not clear whether there might have been differences in assessing the outcome. 

Likely

Analysis was adjusted for important prognostic factors, like indication of PEG, procedural factors etc. However, residual confounding might exist.

Cherian, 2019

Unclear

It is not known which criteria were used to select patients for RIG versus PEG.

Unclear

Follow-up was at least 1 year, but it is not known whether patients are lost to follow-up due to e.g. transferences to other health services.

Unclear

Outcome measures were well defined, but due to the retrospective character of the study it is not clear whether there might have been differences in assessing the outcome. 

Likely

Analysis was not adjusted for important prognostic factors, like indication of PEG, age, sex etc. 

Mahawongkajit, 2019

Unclear

PEG was chosen for advanced oesophageal cancer patients, except the patients who had previous abdominal surgery, could not pass endoscopy to stomach or afford it. However, baseline characteristics did not differ between the treatment groups.

Unclear

Follow-up time is not reported, but was at least 30 days for the outcome measure mortality (since 30-day mortality is reported).

Unclear

Due to the retrospective character of the study it is not clear whether there might have been differences in assessing the outcome. 

Unclear

Analysis was not adjusted for important prognostic factors, however, baseline characteristics did not differ between the treatment groups.

Strijbos, 2019

Unclear

Choice of one of the two techniques is empirical and primarily based on the referring physician’s past experience

and preferences. In general, patients with ALS and HNC are referred exclusively for PRG due to fear of sedation-related complications and tumor seeding, respectively. Furthermore, patients were included at a tertiary referral center.

Unclear

Follow-up time is not reported, but was at least 30 days for the outcome measure mortality (since 30-day mortality is reported).

Unclear

Due to the retrospective character of the study it is not clear whether there might have been differences in assessing the outcome. Furthermore, patients were included at a tertiary referral center, which might have led to underreporting of (minor) complications.

Unclear

Analysis was adjusted for baseline differences, however residual confounding might exist.

Vidhya, 2018

Unclear

It is not known which criteria were used to select patients for RIG versus PEG.

Unclear

Data were at least available to moment of discharge. Due to loss of correspondence and follow up due to transfer of patients to other centres, there might be incomplete data. Mortality data was however, derived from a state based database.

Unclear

Due to the retrospective character of the study it is not clear whether there might have been differences in assessing the outcome. Furthermore, patients were included at a tertiary referral center, which might have led to underreporting of (minor) complications.

Unclear

Analysis were adjusted for age, sex and diagnosis, however residual confounding might exist.

Bankhead, 2005

Unclear

The decision of which technique to use for tube placement is dependent on several factors/list of contraindications. These factors/indications are based on the personal experiences of the authors. Furthermore, it was a teaching hospital.

Unclear

Follow-up was until postoperative day 6, which might have been insufficiently long. It is not reported whether patients were lost to follow-up.

Unclear

Due to the retrospective character of the study it is not clear whether there might have been differences in assessing the outcome.

Unclear

Analysis was not adjusted for potential relevant prognostic factors. Not all relevant baseline characteristics were reported, so it is not clear what the impact is.

Oliveira, 2016

Likely

OPEN gastrostomy was preserved for more complex situations in which gastrostomy was a secondary procedure, frequently for drainage purposes. Furthermore, patients were included at a tertiary referral center.

Unclear

Follow-up time is not reported, but was at least 30 days for the outcome measure mortality (since 30-day mortality is reported). It is not reported whether patients were lost to follow-up.

Unclear

Due to the retrospective character of the study it is not clear whether there might have been differences in assessing the outcome.

Unclear

Analysis was not adjusted for potential relevant prognostic factors.

Edelman, 1998

Unclear

It is not known which criteria were used to select patients for LAP versus PEG.

Unclear

Follow-up time is not reported, but was at least 30 days for the outcome measure mortality (since 30-day mortality is reported). It is not reported whether patients were lost to follow-up.

Unclear

Due to the retrospective character of the study it is not clear whether there might have been differences in assessing the outcome.

Unclear

Analysis was not adjusted for potential relevant prognostic factors.

Ho, 1999

Unclear

It is not known which criteria were used to select patients for LAP versus PEG.

Unclear

Follow-up time is not reported, but was at least 30 days for the outcome measure mortality (since 30-day mortality is reported). It is not reported whether patients were lost to follow-up.

Unclear

Due to the retrospective character of the study it is not clear whether there might have been differences in assessing the outcome.

Unclear

Analysis was not adjusted for potential relevant prognostic factors.

Rustom, 2006

Unclear

It is not known which criteria were used to select patients for RIG versus PEG versus LAP. Furthermore, patients were included at a regional centre for head and neck cancer.

Unclear

All patients were followed for 18 months, but it is not reported whether patients were lost to follow-up.

Unclear

Due to the retrospective character of the study it is not clear whether there might have been differences in assessing the outcome.

Unclear

Analysis was not adjusted for potential relevant prognostic factors.

Bailey, 1992

Unclear

The choice of gastrostomy depends on surgeon preferences, which was – according to the authors – independent of disease stage.

Unclear

Follow-up time is not reported.

Unclear

Due to the retrospective character of the study it is not clear whether there might have been differences in assessing the outcome.

Unclear

Analysis was not adjusted for potential relevant prognostic factors.