Intervention Review

You have free access to this content

Systemic antimicrobial prophylaxis for percutaneous endoscopic gastrostomy

  1. Allyson Lipp*,
  2. Gail Lusardi

Editorial Group: Cochrane Wounds Group

Published Online: 14 NOV 2013

Assessed as up-to-date: 30 AUG 2013

DOI: 10.1002/14651858.CD005571.pub3


How to Cite

Lipp A, Lusardi G. Systemic antimicrobial prophylaxis for percutaneous endoscopic gastrostomy. Cochrane Database of Systematic Reviews 2013, Issue 11. Art. No.: CD005571. DOI: 10.1002/14651858.CD005571.pub3.

Author Information

  1. Department of Care Sciences, University of South Wales, Faculty of Health, Sport and Science, Pontypridd, Rhondda Cynon Taff, UK

*Allyson Lipp, Faculty of Health, Sport and Science, Department of Care Sciences, University of South Wales, Glyn Taff Campus, Pontypridd, Rhondda Cynon Taff, CF37 1DL, UK. allyson.lipp@southwales.ac.uk.

Publication History

  1. Publication Status: New search for studies and content updated (no change to conclusions)
  2. Published Online: 14 NOV 2013

SEARCH

 
Characteristics of included studies [ordered by study ID]
Ahmad 2003

MethodsRandomised controlled trial.


ParticipantsMale and female patients.
Over 18 years.
PEG tube inserted for CVA (n= 38) CNS disorders (n= 20), oropharyngeal cancer (n= 18) and miscellaneous (n= 26) (Taken from Table 1 - 102 patients evaluable on a per protocol analysis).
Excluded if no consent or suspected/confirmed allergy to antibiotic used.
Total number of patients randomised; n = 141.


Interventions'Pull' technique
Group 1: cefuroxime 750 mg IV 30 min prior to PEG (n = 50);
Group 2: saline placebo IV 30 min prior to PEG (n = 51);
Group 3: receiving antibiotics before and during the study (n = 40).


OutcomesPeristomal infection.
Complications.


NotesGroup 3 excluded from analysis as not randomised.Used Jain et al criteria for wound assessment.
Country of origin: Wales.
Power calculation performed.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskPredetermined computer-generated randomisation scheme.

Allocation concealment (selection bias)Unclear riskReview authors unable to contact study authors for clarification.

Blinding (performance bias and detection bias)
Blinding patient
Low riskPatients received intravenous antibiotics or saline.

Blinding (performance bias and detection bias)
Blinding care provider
Low riskProcedure by doctor or nurse not involved in the study.

Blinding (performance bias and detection bias)
Blinding outcome assessor
Low riskBlinded to treatment group.

Incomplete outcome data (attrition bias)
Drop outs/withdrawals
Unclear risk8 (5.6%) of 141 patients excluded because of incomplete data.

Comment: A 5.6% drop out rate makes the study at unclear risk of attrition bias. However the study reports 133 patients "were analysable on an intention to treat analysis".

Selective reporting (reporting bias)Low riskPrespecified outcomes reported.

Baseline imbalance? (Was the study free of baseline imbalance?)Low riskBalanced for gender, age, reason for PEG, underlying conditions.

Timing of outcome assessment similar in all groups?Low riskIn all groups outcomes were assessed immediately after procedure plus three, five and seven days.

Akkersdijk 1995

MethodsRandomised controlled trial.


ParticipantsConsecutive male and female patients.

PEG tube inserted for oropharyngeal cancer (n= 56), neurological (n= 32) and other (n= 12).
Inclusion and exclusion criteria not stated.
Total number of patients randomised; n = 100.


Interventions'Pull' and 'push' techniques.
Group 1: Pull, augmentin 1.2 g IV 3 doses given, 1st dose 30 min prior to procedure two doses administered over 24 hours (n = 37);
Group 2: pull, no placebo, no antibiotic (n = 34);
Group 3: push, no placebo, no antibiotic (n = 29).


OutcomesPeristomal infection.
Major and minor complications.
Comparison between push and pull technique.


NotesCriteria for wound assessment: minor infection present if redness with or without purulent discharge, major infection judged by physician as those requiring antibiotics.
Country of origin: Netherlands.
No power calculation performed.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskNot stated.

Allocation concealment (selection bias)Unclear riskNot stated.

Blinding (performance bias and detection bias)
Blinding patient
Unclear riskNot stated; no placebo given.

Blinding (performance bias and detection bias)
Blinding care provider
Unclear riskNot stated.

Blinding (performance bias and detection bias)
Blinding outcome assessor
Unclear riskNot stated.

Incomplete outcome data (attrition bias)
Drop outs/withdrawals
Low riskPEG placement failed in four patients (4%).

Selective reporting (reporting bias)Low riskAll prespecified outcomes reported.

Baseline imbalance? (Was the study free of baseline imbalance?)Low riskAll groups were balanced for age, reason for PEG, underlying conditions.

Timing of outcome assessment similar in all groups?Low riskOutcomes assessed in all groups twice weekly for one month.

Blomberg 2010

MethodsRandomised controlled trial


ParticipantsMale and female patients.

PEG tube inserted for ear, nose or throat cancer (127), neurological disease (42), oesophageal cancer (n= 30), stroke (n= 11), dementia (n= 1), gastric cancer (n= 1), and other (n= 22).

Included if able to consent to participation in the study after receiving oral and written information
or did not meet exclusion criteria and no contraindications for PEG.

Excluded if ongoing antibiotic treatment, illness too severe to allow the patient to participate, allergy to any of the antibiotic alternatives.

Total number of patients randomised; n = 234


InterventionsPull technique.
Group 1: Co-trimoxazole (800mg sulfamethoxazole &160mg trimethoprim) in 20ml deposited in the PEG immediately after insertion (n=116).
Group 2: Cefuroxime 1.5g IV given 1 hour before PEG insertion (n= 118).


OutcomesPrimary outcome - wound infection

Clinically evident peristomal infection at follow up appointment infection within 7- 14 days after insertion of the PEG catheter.
Secondary outcomes - Positive bacterial culture or blood biochemistry.


NotesCriteria for wound assessment: a clinically identifiable wound infection, as judged by a red zone around the catheter or occurrence of pus, subcutaneous swelling, and pain on palpation in the area around the catheter.
Secondary outcomes objective signs of infection, including a positive bacterial culture, high levels of highly sensitive C reactive protein, and a high white blood cell count.
Country of origin: Sweden.
Power calculation performed for non-inferiority.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk"The randomisation process was conducted by personnel at a hospital department not engaged in the care of the included patients"
"Personnel were contacted by telephone and on request opened a closed envelope, taken from a pre-prepared block (50 envelopes in each block) of equally distributed and mixed envelopes, containing a randomisation sheet with information on the drug to be used"

Allocation concealment (selection bias)Low risk"The randomisation process was conducted by personnel at a hospital department not engaged in the care of the included patients"
"Personnel were contacted by telephone and on request opened a closed envelope, taken from a pre-prepared block (50 envelopes in each block) of equally distributed and mixed envelopes, containing a randomisation sheet with information on the drug to be used"

Blinding (performance bias and detection bias)
Blinding patient
Low risk"the blinding of the patients was accomplished by using intravenous fluid and manipulating the newly inserted PEG catheter in all patients. This sham manoeuvre was facilitated by the use of sedation"

Blinding (performance bias and detection bias)
Blinding care provider
Unclear riskNot stated.

Blinding (performance bias and detection bias)
Blinding outcome assessor
Low risk"The nurses who evaluated the patients at follow-up visit were not involved in insertion of the PEG catheter, including the administration of antibiotics"

Incomplete outcome data (attrition bias)
Drop outs/withdrawals
High riskOut of 34 dropouts (14.5%) or withdrawals, a total of twelve participants in each group did not undergo PEG placement for anatomical reasons. Other attrition included five deaths, one patient pulled out PEG, three lost to follow-up and one received co-trimoxazole after being randomised to cefuroxime.

Comment: 15% drop out rate makes the study at high risk of attrition bias, but ITT analysis undertaken.

Selective reporting (reporting bias)Low riskAll prespecified outcomes reported.

Baseline imbalance? (Was the study free of baseline imbalance?)Low riskBoth groups were balanced for age,smoking, diabetes, indications for PEG.Slight gender imbalance 42 females in group 1 versus 31 females in group 2.

Timing of outcome assessment similar in all groups?Low risk7-14 days for both groups.

Dormann 2000

MethodsRandomised controlled trial.


ParticipantsMale and female patients aged over 18 years requiring enteral feeding via PEG tube > six weeks.

PEG tube inserted for neurological disease (n= 145), tumour (n= 63) and other (n= 8). (Taken from Table 1 - 216 patients evaluable as 21 dropouts).
Excluded if signs of infection, peritonitis ascites, peritoneal malignancy, prior gastric/bowel disease, granulocytopenia, previous radio/chemotherapy, antibiotic treatment within previous 72 h, clotting/platelet disorders, sensitivity to ceftriaxone.
Total number of patients randomised; n = 237.


Interventions'Pull' technique:
Group 1: ceftriaxone 1 g IV 30 min prior to PEG (n = 106);
Group 2: no placebo, no antibiotic (n = 110).

Numbers originally randomised into groups not given in trial report.


OutcomesPeristomal infection.
Mortality.
Cost of antibiotic therapy.
Post-intervention antibiotic therapy.


NotesUsed modified Jain et al criteria for wound assessment.
Data from four previous studies incorporated into this study.
Data from previous study (1999a), sponsored by Hoffman La Roche, incorporated into this study.
Conducted in 12 secondary and tertiary medical centres.
Country of origin: Germany.
Power calculation performed.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskRandomised in blocks of four using Rancode 3.1.

Allocation concealment (selection bias)Unclear riskNot stated.

Blinding (performance bias and detection bias)
Blinding patient
Unclear riskNot stated.

Blinding (performance bias and detection bias)
Blinding care provider
Unclear riskNot stated.

Blinding (performance bias and detection bias)
Blinding outcome assessor
Unclear risk'Study monitors' involved but specific role not stated.

Incomplete outcome data (attrition bias)
Drop outs/withdrawals
Low risk21 drop outs (8.9%).

Selective reporting (reporting bias)Low riskPrespecified outcomes addressed.

Baseline imbalance? (Was the study free of baseline imbalance?)Low riskBoth groups balanced for gender, age, reasons for PEG tube, BMI,indications for PEG, underlying conditions.

Timing of outcome assessment similar in all groups?Low riskOutcomes in both groups were assessed at one, two, four and ten days (mean 8.7).

Gossner 1999

MethodsRandomised controlled trial.


ParticipantsMale and female patients with proportionately more males (n = 243) than females (n = 93).

PEG tube inserted for malignant disease (n= 210), neurological disorders (n= 97). (Taken from Table 1 - 307 patients evaluable as 40 dropouts).

Included if had functional resorption and digestive capacity with temporary, or permanent, dysphagia.
Excluded if severe clotting or wound healing disorder, pronounced immune deficiency, paralytic ileus, Billroth II procedure, peritonitis, ascites, sensitive to antibiotics used.
Total number of patients randomised; n = 347.


InterventionsModified 'pull' technique.
Group 1: cefotaxime 2 g by infusion 30 min prior to PEG (n = 101);
Group 2: piperacillin 2 g + 0.5 g tazobactam by infusion (n = 100);
Group 3: no placebo, no antibiotic (n = 106).
Infusion given over 20 min.

Numbers originally randomised into groups not given in trial report


OutcomesPeristomal infection.
Peritonitis.
Mortality.


NotesUsed Jain et al and Shapiro et al criteria for wound assessment.
Included data from a previous study.
Country of origin: Germany.
No power calculation performed.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Unclear riskNot stated.

Allocation concealment (selection bias)High riskRandomly assigned to group after consent taken, and at least one day before procedure.

Blinding (performance bias and detection bias)
Blinding patient
Unclear riskNot stated.

Blinding (performance bias and detection bias)
Blinding care provider
Unclear riskNot stated.

Blinding (performance bias and detection bias)
Blinding outcome assessor
Unclear riskNot stated.

Incomplete outcome data (attrition bias)
Drop outs/withdrawals
Low risk40 drop outs (11.5%).

Selective reporting (reporting bias)Low riskPrespecified outcomes reported.

Baseline imbalance? (Was the study free of baseline imbalance?)Low riskAll groups were balanced for age, gender, weight, Karnofksy index, reasons for PEG.

Timing of outcome assessment similar in all groups?Low riskOutcomes in all groups were assessed daily for seven days. Those discharged were assessed over the telephone or via the outpatient department, but the group was not specified.

Jain 1987

MethodsRandomised controlled trial.


ParticipantsMale and female patients (distribution not stated) who had given consent.

PEG tube inserted for CVA (n= 53), oropharyngeal cancer (n= 27), CNS trauma (n= 8), CNS infection (n= 3), CNS degenerative disease (n= 10) and miscellaneous (n= 6).
Excluded if: allergic to cefazolin, refused signed consent, technical reasons for PEG placement.
Total number of patients randomised; n = 107.


InterventionsTechnique not stated, but presumed to be 'pull' as the authors state 'the feeding tube traverses the mouth and pharynx'.
Group 1: receiving antibiotics before and during the study and were randomly assigned to:
Group 1 a: cefazolin 1 g IV 30 min prior to PEG (n = 25);
Group 1 b: saline placebo IV 30 min prior to PEG (n = 27);

Group 2 not receiving antibiotics were randomly assigned to:
Group 2 a: cefazolin 1 g IV 30 min prior to PEG (n = 27);
Group 2 b: saline placebo IV 30 min prior to PEG (n = 28).


OutcomesPeristomal infection.


NotesCriteria for wound assessment scoring devised by Jain et al which used indicators previously used by Shapiro et al (1982).
Group 1 excluded from grouped analysis.
Country of origin: USA.
Power calculation performed.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskRandomisation schedule generated by Hewlitt-Packard HP-67 pocket calculator.

Allocation concealment (selection bias)Low riskOne author (KPC) was responsible for (and the only one aware of) assignment and did not evaluate the wounds.

Blinding (performance bias and detection bias)
Blinding patient
Low riskStated as 'double blind'.

Blinding (performance bias and detection bias)
Blinding care provider
Low riskStated as 'double blind'.

Blinding (performance bias and detection bias)
Blinding outcome assessor
Low riskOutcomes assessed by team members blind to placebo allocation.

Incomplete outcome data (attrition bias)
Drop outs/withdrawals
Low riskNo drop outs.

Selective reporting (reporting bias)Low riskPrespecified outcomes addressed.

Baseline imbalance? (Was the study free of baseline imbalance?)Unclear riskAll groups were balanced for gender, underlying conditions, reasons for PEG. In group 1 the mean age was slightly but significantly less than that of patients who did not receive prophylaxis.

Timing of outcome assessment similar in all groups?Low riskOutcomes in all groups were assessed daily for seven days.

Jonas 1985

MethodsRandomised controlled trial.


ParticipantsAll male patients with dysphagia, consent, and functionally-intact gastrointestinal tract.

PEG tube inserted for underlying malignancy (n= 18) and neurological (n= 15). (Taken from Table 4 - 33 patients evaluable as 4 dropouts).
Excluded if had gastric ulcer/cancer, active infection requiring antibiotics, peritonitis, ascites, extensive abdominal surgery, contraindications to endoscopy, hypersensitivity to cephalosporins, or refused consent.
Total number of patients randomised; n = 37.


Interventions'Pull' technique:
Group 1: cefoxitin 1 g IV 30 min prior to PEG (2 further doses given at 6 h intervals) (n = 17);
Group 2: saline placebo IV 30 min prior to PEG (n = 16).

Numbers originally randomised into groups not given in trial report


OutcomesPeristomal infection.
Effect of gastric pH.
Microbiology of oropharyngeal flora and tip of tube.
Effect of underlying disease (neurological/malignancy).


NotesCriteria for wound assessment: red, tender, indurated area at exit site with pain ± systemic signs of leukocytosis and fever.
Country of origin USA.
No power calculation performed.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskRandom sequence generated by pharmacist.

Allocation concealment (selection bias)Low riskAllocation concealment as 'foil-covered vials of identical appearance and equivalent volume' by pharmacy.

Blinding (performance bias and detection bias)
Blinding patient
Low riskStated: 'foil-covered vials of identical appearance and equivalent volume'.

Blinding (performance bias and detection bias)
Blinding care provider
Low riskStated: 'foil-covered vials of identical appearance and equivalent volume'.

Blinding (performance bias and detection bias)
Blinding outcome assessor
Low riskKnown to pharmacist only.

Incomplete outcome data (attrition bias)
Drop outs/withdrawals
Low riskFour drop outs (10.8%).

Selective reporting (reporting bias)Low riskPrespecified outcomes reported.

Baseline imbalance? (Was the study free of baseline imbalance?)Low riskBoth groups were balanced for malignancy, neurological disease.

Timing of outcome assessment similar in all groups?Low riskOutcomes in both groups were assessed at more than three days (endpoint not stated).

Panigrahi 2002

MethodsRandomised controlled trial.


ParticipantsConsenting male and female patients, who had not received antibiotics in the preceding three days.

PEG tube inserted for neurological (n= 49), Cancer (n= 4) and miscellaneous (n= 5). (Taken from Table 1 - 58 patients evaluable as 17 dropouts).
Excluded if allergic to penicillin, immunocompromised, had severe clotting disorder, end-stage renal failure, or gastrointestinal surgery which precluded site of PEG.
Total number of patients randomised; n = 75.


InterventionsTechnique not stated.
Group 1: co-amoxiclav (dose not stated) IV 15-30 min prior to PEG (n = 29);
Group 2: saline placebo IV 15-30 min prior to PEG (n = 29).

Numbers originally randomised into groups not given in trial report


OutcomesPeristomal infection.


NotesCriteria for wound assessment: modified ASEPSIS scoring system used.
Country of origin England.
No power calculation performed.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk'Provided by the hospital pharmacy'.

Allocation concealment (selection bias)Low riskAdministered by 'endoscopist 15-30 min prior to endoscopy with no prior knowledge of the sequence in the randomisation'.

Blinding (performance bias and detection bias)
Blinding patient
Low riskStated as 'double blind'.

Blinding (performance bias and detection bias)
Blinding care provider
Low riskStated as 'double blind'.

Blinding (performance bias and detection bias)
Blinding outcome assessor
Low risk'The infection control team was blinded to the randomisation'.

Incomplete outcome data (attrition bias)
Drop outs/withdrawals
High risk17 in total, seven died, eight drop outs, two lost to follow-up (22.6%).

Comment: A 15% drop out rate makes the study at high risk of attrition bias.

Selective reporting (reporting bias)Low riskPrespecified outcomes reported.

Baseline imbalance? (Was the study free of baseline imbalance?)Low riskBoth groups were balanced for age, gender, reasons for PEG.

Timing of outcome assessment similar in all groups?Low riskOutcomes in both groups were assessed daily for seven days, and then day 28.

Preclik 1999

MethodsRandomised controlled trial.


ParticipantsConsenting male and female patients with dysphagia, 18 years and over.

PEG tube inserted for malignancy (n= 55) and neurological disease (n= 29) (Taken from Table 1 - 84 patients evaluable on a per protocol analysis).
Excluded if had: allergy to penicillin; received antibiotics in past 4 days; neutropenia < 500 cells/microlitre; creatinine < 300 micromols/litre or contraindication to PEG.
Total number of patients randomised; n = 106.


InterventionsThread 'pull' technique.
Group 1: co-amoxiclav 2.2 g, by short infusion 30 min prior to PEG (n = 46).
Group 2: saline placebo, by short infusion 30 min prior to PEG (n = 47).

Numbers originally randomised into groups not given in trial report


OutcomesPeristomal infection.
Mortality.
Adverse effects.


NotesIncluded data from a previous study.
Used criteria by Jain et al for wound assessment.
Conducted in six hospitals.
Country of origin: Germany.
Power calculation performed.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskPermuted block design with separate sequences of random numbers for each centre to ensure equal numbers.

Allocation concealment (selection bias)Low riskPrepared by pharmacy.

Blinding (performance bias and detection bias)
Blinding patient
Low riskStated as 'double blind'.

Blinding (performance bias and detection bias)
Blinding care provider
Low riskBlinding of investigators, study nurses, reviewers and data managers.

Blinding (performance bias and detection bias)
Blinding outcome assessor
Low riskBlinding of investigators, study nurses, reviewers and data managers.

Incomplete outcome data (attrition bias)
Drop outs/withdrawals
Unclear risk13 drop outs (12.3%). Report stated that ITT analysis was performed on 93 patients who received PEG but 106 patients were randomised and 13 people dropped out and there are no details on how this missing data were handled. The judgement is therefore unclear.

Selective reporting (reporting bias)Low riskPrespecified outcomes stated.

Baseline imbalance? (Was the study free of baseline imbalance?)Low riskBoth groups were balanced for age, gender, previous method of feeding, underlying conditions, performance status reason for PEG, Karnofsky index.

Timing of outcome assessment similar in all groups?Low riskOutcomes in both groups were assessed daily for seven days and mortality at 30 days.

Radhakrishnan 2006

MethodsRandomised controlled trial.


ParticipantsConsenting male and female patients who had not received antibiotics in the preceding 2 days.

PEG tube inserted for CVA (n= 57), dementia (n= 10), neurogenic dysphagia (n= 11) and miscellaneous (n= 18).
Excluded if on antibiotic therapy, unable to gain consent, had allergy to cephalosporin or iodine, or required prophylaxis for endocarditis.
Total number of patients randomised; n = 96.


Interventions'Pull' technique.
Group 1: cefuroxime 750 mg IV immediately prior to PEG (2 further doses given at 8 h intervals) (n = 34);
Group 2: povidone-iodine spray only (n = 28);
Group 3: povidone-iodine spray and cefuroxime 750 mg IV immediately prior to PEG (2 further doses given at 8 h intervals) (n = 34).


OutcomesPeristomal infection.
Mortality.
Logistic regression effects of diabetes, steroids, acid suppressants, age and sex.


NotesUsed Jain et al wound assessment.
Some patients received antibiotics during follow-up period.
Country of origin: England.
Power calculation not performed.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low riskClosed envelopes opened at random by an endoscopy nurse.

Allocation concealment (selection bias)Low risk'Closed envelopes that were shuffled and opened at random by an endoscopy nurse after the patient had given consent for the study'.

Blinding (performance bias and detection bias)
Blinding patient
Unclear riskNot stated.

Blinding (performance bias and detection bias)
Blinding care provider
Unclear riskNot stated.

Blinding (performance bias and detection bias)
Blinding outcome assessor
Low riskStomal site inspected by the investigator who was blinded to regime.

Incomplete outcome data (attrition bias)
Drop outs/withdrawals
Low riskNo drop outs, five patient deaths in 1 week, 1 from gastric leakage on day 6 (5.2%).

Selective reporting (reporting bias)Low riskPrespecified outcomes addressed.

Baseline imbalance? (Was the study free of baseline imbalance?)Low riskAll groups were balanced for age, gender, steroids, diabetes, size of PEG tube, reasons for PEG.

Timing of outcome assessment similar in all groups?Low riskOutcomes in both groups were assessed on day three or four and day seven.

Saadeddin 2005

MethodsRandomised controlled trial.


ParticipantsMale and female patients over 16 years of age.

PEG tube inserted for CVA (n= 61), neurological (n= 35) and miscellaneous (n= 5). (Taken from Table 2 Error in paper as these total 101. The number quoted as 'analysable' for each outcome is 83 for peristomal infection, 97 for systemic infection, and 99 for seven-day mortality).
Included if PEG was for non-malignancy.
Excluded if had antibiotics within 48 h preceding PEG insertion.
Total number of patients randomised; n = 110.


InterventionsPull technique.
Group 1: co-amoxiclav 2.2 g (n = 43), or cefotaxime 2 g (if allergic to penicillin) (n = 8) at time of PEG tube insertion.
Group 2: placebo (not stated) at time of PEG tube insertion (n = 48).

Numbers originally randomised into groups not given in trial report


OutcomesPeristomal infection.
Systemic infection.
Seven-day mortality.


NotesUsed Jain et al criteria for wound assessment.
Country of origin: England.
Power calculation performed.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk'Done in advance using a random number generator'.

Allocation concealment (selection bias)Low risk'Study assignment cards kept with medication packs in pharmacy'.

Blinding (performance bias and detection bias)
Blinding patient
Low risk'Patient blinded'. Syringe covered with opaque sleeve prepared by endoscopy nurse.

Blinding (performance bias and detection bias)
Blinding care provider
Low riskSyringe covered with opaque sleeve prepared by endoscopy nurse.

Blinding (performance bias and detection bias)
Blinding outcome assessor
Low risk'Study investigator (outcome assessor) blinded'.

Incomplete outcome data (attrition bias)
Drop outs/withdrawals
Low risk11 withdrawals (10%).

Selective reporting (reporting bias)Low riskPrespecified outcomes addressed.

Baseline imbalance? (Was the study free of baseline imbalance?)Low riskBoth groups were balanced for age, gender, weight, underlying conditions.

Timing of outcome assessment similar in all groups?Low riskOutcomes in both groups were assessed daily for seven days.

Shastri 2008

MethodsRandomised controlled trial.


ParticipantsMale and female patients 16-89 years.

PEG tube inserted for cancer (n= 93). (Taken from Figure 1 - 93 patients evaluable as 4 dropouts).
Included if had stenotic malignancy and lesions of the upper gastrointestinal tract.
Excluded if allergic to antibiotics, or had any contraindications to PEG.
Total number of patients randomised; n = 97.


InterventionsPush technique gastropexy PEG.
Group 1: ceftriaxone 2 g IV (n = 49);
Group 2: placebo ('similar looking') IV (n = 48).


OutcomesPeristomal infection.


NotesUsed Jain et al and Gossner criteria for wound assessment.
Country of origin: Germany.
Power calculation performed.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk'Block randomisation and computer-generated random numbers'.

Allocation concealment (selection bias)Low risk'Sequentially numbered, opaque, sealed envelopes'.

Blinding (performance bias and detection bias)
Blinding patient
Unclear riskNot stated.

Blinding (performance bias and detection bias)
Blinding care provider
Low riskInjections prepared by non-study staff.

Blinding (performance bias and detection bias)
Blinding outcome assessor
Low riskWound assessment by members of nutrition support team.

Incomplete outcome data (attrition bias)
Drop outs/withdrawals
Low riskFour drop outs (4.1%).

Selective reporting (reporting bias)Low riskPrespecified outcomes reported.

Baseline imbalance? (Was the study free of baseline imbalance?)Low riskBoth groups were balanced for age, gender, BMI, location of malignancy, reason for PEG.

Timing of outcome assessment similar in all groups?Low riskOutcomes in both groups were assessed daily for seven days.

Sturgis 1996

MethodsRandomised controlled trial.


ParticipantsCharacteristics of patients unclear.

PEG tube inserted for CVA and head and neck cancer (no numbers provided).
Excluded if unable to place PEG, inadequate access to follow-up by medical personnel, or refused to sign consent form.
Total number of patients randomised; n = 115.


InterventionsTechnique not stated.
Group 1: cefazolin 1 g IV within 30 min of PEG (n = 30);
Group 2: saline placebo IV within 30 min of PEG (n = 31);
Group 3: receiving antibiotics (including cephalosporins, penicillins, quinolones and sulphonamides) before and during the study which were continued for 24 h post PEG (n = 54).


OutcomesPeristomal infection.
Gastric pH.
Wound cultures.


NotesGroup 3 excluded from analysis, as not randomised.
Used Jain et al criteria for wound assessment.
Country of origin: USA.

Power calculation not performed.


Risk of bias

BiasAuthors' judgementSupport for judgement

Random sequence generation (selection bias)Low risk'Generated random sequence known only to those in the pharmacy'.

Allocation concealment (selection bias)Low riskPharmacy dispensed 'Identical vials with equivalent volume'.

Blinding (performance bias and detection bias)
Blinding patient
Low riskStated as 'double blind'.

Blinding (performance bias and detection bias)
Blinding care provider
Low riskStated as 'double blind'.

Blinding (performance bias and detection bias)
Blinding outcome assessor
Unclear riskNot stated.

Incomplete outcome data (attrition bias)
Drop outs/withdrawals
Low riskNo drop outs.

Selective reporting (reporting bias)Low riskPrespecified outcomes reported.

Baseline imbalance? (Was the study free of baseline imbalance?)Low riskAll groups were balanced for age, reasons for PEG, underlying conditions.

Timing of outcome assessment similar in all groups?Low riskOutcomes in all groups were assessed daily for one week.

 
Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion

Adachi 2002Not an RCT.

Arrowsmith 1997Not an RCT but an audit.

Beales 2003A critique of two RCTs.

Chowdhury 1996Not an RCT.

Dormann 2004Not an RCT but a cohort study that did not investigate the use of antibiotics.

Gawenda 1997Abstract only: unable to obtain study data via trial authors or full-text article via British Library.

Gopal 2004Not an RCT.

Horiuchi 2006All patients received prophylactic and concomitant antibiotics.

Hull 2001Not an RCT: did not investigate the use of antibiotics.

Jafri 2007A systematic review. The patient data included were a duplicate of those in the included studies.

Kanie 2000Examined PEG feeding not placement.

Lee 2002Not an RCT.

Loser 2000Not an RCT.

Maetani 2003A comparison between push and pull methods of placement. Antibiotics were given to all patients.

Maetani 2005A comparison of the use of an over tube to reduce infection in PEG tube placement. Antibiotics were given to all patients.

Rey 1998Not an RCT.

Sharma 2000A systematic review. The patient data included were a duplicate of those in the included studies.

 
Comparison 1. Systemic antibiotic (IV) compared with placebo

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Peristomal infection8586Odds Ratio (M-H, Fixed, 95% CI)0.34 [0.22, 0.53]

 
Comparison 2. Systemic antibiotic (IV) compared with no intervention

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Peristomal infection3623Odds Ratio (M-H, Fixed, 95% CI)0.30 [0.17, 0.53]

 
Comparison 3. Systemic antibiotic (IV) compared with placebo/no intervention/skin antiseptic

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Peristomal infection121271Odds Ratio (M-H, Fixed, 95% CI)0.36 [0.26, 0.50]

 2 Allocation concealment121271Odds Ratio (M-H, Fixed, 95% CI)0.47 [0.34, 0.65]

    2.1 adequate allocation concealment
8554Odds Ratio (M-H, Fixed, 95% CI)0.58 [0.38, 0.88]

    2.2 unclear/inadequate concealment
4717Odds Ratio (M-H, Fixed, 95% CI)0.34 [0.21, 0.58]

 3 Sponsorship121271Odds Ratio (M-H, Fixed, 95% CI)0.36 [0.26, 0.50]

    3.1 Trials with no sponsorship
10962Odds Ratio (M-H, Fixed, 95% CI)0.38 [0.25, 0.56]

    3.2 Trials with sponsorship
2309Odds Ratio (M-H, Fixed, 95% CI)0.32 [0.18, 0.58]

 
Comparison 4. Systemic antibiotic compared with systemic antibiotic

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Systemic antibiotic (IV) compared with systemic antibiotic (IV)1Odds Ratio (M-H, Fixed, 95% CI)Totals not selected

    1.1 Peristomal infection
1Odds Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]

 2 Systemic antibiotic (PEG) compared with systemic antibiotic (IV)1Odds Ratio (M-H, Fixed, 95% CI)Totals not selected

 
Comparison 5. Systemic antibiotic (IV) compared with systemic antibiotic (IV) and skin antiseptic

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Peristomal infection168Odds Ratio (M-H, Fixed, 95% CI)15.78 [1.90, 130.86]

 
Comparison 6. Skin antiseptic compared with systemic antibiotic (IV) and skin antiseptic

Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size

 1 Peristomal infection162Odds Ratio (M-H, Fixed, 95% CI)15.63 [1.84, 133.09]