Methicillin-resistant Staphylococcus aureus infection of percutaneous endoscopic gastrostomy sites


Dr M. Hull, Division of Medicine, Clinical Sciences Building, St James’s University Hospital, Leeds LS9 7TF, UK. E-mail:



Antibiotic prophylaxis for percutaneous endoscopic gastrostomy insertion remains controversial. The bacteriology of peristomal infection following percutaneous endoscopic gastrostomy insertion has been poorly studied, leading to uncertainty regarding the optimum choice of antibiotic for prophylaxis.


To investigate the bacteriology of peristomal infection following percutaneous endoscopic gastrostomy insertion and to determine the contribution of methicillin-resistant Staphylococcus aureus.


Nasal and pharyngeal swabs were taken from a consecutive series of patients prior to percutaneous endoscopic gastrostomy insertion over a 6-month period. Bacterial colonization and infection at the peristomal site were prospectively evaluated at days 2/3 and 7 post-insertion.


Thirty-one patients underwent percutaneous endoscopic gastrostomy insertion (mean age, 68 years; cerebrovascular disease, 52%). Naso-pharyngeal colonization by methicillin-resistant Staphylococcus aureus (35%) invariably led to peristomal colonization following percutaneous endoscopic gastrostomy insertion. Peristomal infection occurred in eight (26%) cases (seven (88%) methicillin-resistant Staphylococcus aureus- positive). Peristomal infection was significantly more likely to occur in patients with naso-pharyngeal methicillin-resistant Staphylococcus aureus colonization (odds ratio, 10.8; 95% confidence interval, 1.6–70.9).


Naso-pharyngeal methicillin-resistant Staphylococcus aureus colonization invariably predicts peristomal methicillin-resistant Staphylococcus aureus colonization following percutaneous endoscopic gastrostomy insertion, and is associated with an increased peristomal infection rate. Currently recommended antibiotic prophylaxis regimens may be inappropriate in institutions with significant methicillin-resistant Staphylococcus aureus colonization rates.


A percutaneous endoscopic gastrostomy (PEG) is commonly used for the provision of enteral nutrition in hospitalized patients. However, the standard PEG ‘pull’ insertion technique is not without associated morbidity and mortality, which includes peristomal infection in 5–30% of patients.1 Antibiotic prophylaxis for PEG insertion has now been advocated by the European Society of Gastrointestinal Endoscopy,2 American Society for Gastrointestinal Endoscopy3 and, more recently, by the British Society of Gastroenterology,4 on the basis of several randomized controlled trials5–11 and a meta-analysis.12 However, the relevance of the data from these trials to hospital practice remains controversial, because of conflicting data from the individual randomized trials,5–11 and the fact that these studies have varied markedly with respect to the antibiotic(s) used, the patient populations studied and the criteria used for the diagnosis of peristomal infection.5–11 A particular weakness of previous trials has been a relative lack of characterization of causative organism(s). It has been assumed that, because the internal PEG positioning device passes through the oro-pharynx during insertion, antimicrobial chemoprophylaxis should be specifically aimed at oro-pharyngeal bacteria, such as Staphylococcus aureus and Streptococcus spp.13 In keeping with this, several studies have shown that Staphylococcus aureus is the most commonly implicated organism in peristomal infection.5, 9, 13, 14 Infection with methicillin-resistant Staphylococcus aureus (MRSA) is becoming increasingly common in hospitalized patients in England and Wales, as indicated by a recent analysis of bacteraemia isolates between 1990 and 1998.15 Therefore, we investigated the frequency of naso-pharyngeal MRSA colonization in patients undergoing PEG insertion, and the contribution of MRSA to subsequent peristomal infection.


Data collection

We studied a consecutive series of 31 patients undergoing insertion of a 16 Fr (except for two patients who had 20 Fr devices inserted for palliative gastrointestinal decompression) CorFlo PEG (Merck Pharmaceuticals, Alton, Hampshire, UK) within a 6-month period (January–June 2000 inclusive) at St James’s University Hospital. The pre-procedure protocol did not include the use of routine antibiotic prophylaxis during this period. Nasal and pharyngeal swabs were obtained just prior to PEG insertion. On days 2/3 and 7 post-insertion, the PEG entry site was inspected and the presence of peristomal infection was determined prospectively using a modification of the PEG site scoring system devised by Jain et al.5 The presence of erythema (score 0, nil; 1, < 5 mm; 2, 6–10 mm; 3, 11–15 mm; 4, > 15 mm), induration (score 0, nil; 1, < 10 mm; 2, 11–20 mm; 3, > 20 mm) and exudate (score 0, nil; 1, serous; 2, serosanguinous; 3, sanguinous; 4, purulent) was determined by an individual blind to the nasal and pharyngeal microbiology results. A combined score greater than or equal to six, on either or both days, was taken as indicative of peristomal infection. Patients with ‘purulent’ exudate, but no other sign of infection, were not considered to be infected in order to avoid misclassification of patients with simple reflux of enteral feed through the PEG insertion site. Finally, a swab was taken from the peristomal area on each occasion. In addition, details of antibiotic administration pre- and post-PEG insertion were obtained, including name(s), dose, duration and indication.

Statistical analysis

The association between naso-pharyngeal colonization and the presence of peristomal infection was tested by Fisher’s exact test. The fold increase in risk of peristomal infection in patients with naso-pharyngeal MRSA colonization was determined by calculation of the odds ratio (OR) and 95% confidence intervals (CI). Logistic regression analysis was used to determine which factors (naso-pharyngeal MRSA colonization, naso-pharyngeal colonization by coliforms and previous antibiotic therapy) predicted subsequent peristomal infection.


Patient details

The mean age of the patients was 68 ± 3 years (± standard error of the mean). The indications for PEG insertion were cerebrovascular disease (n=16; 52%), enteral nutrition supplementation (7; 23%), chronic neurological disease (5; 16%), palliative gastrointestinal decompression (2; 6%) and carcinoma (1; 3%). Thirteen patients (42%) received antibiotic therapy within 7 days of PEG insertion (Table 1). There were three deaths (10%) within 7 days of PEG insertion which were all caused by pneumonia.

Table 1.   Details of antibiotic use and duration within 7 days of percutaneous endoscopic gastrostomy (PEG) insertion Thumbnail image of

Nasal and pharyngeal isolates

Eleven patients (35%) had an MRSA-positive nasal swab prior to PEG insertion (Figure 1). Other organisms cultured from nasal swabs included S. aureus (n=4), Pseudomonas aeruginosa (n=3), coliforms (n=4), Acinetobacter spp. (n=2), Enterococcus spp. (n=1) and viridans Streptococcus (n=1). Ten patients (32%) had a sterile nasal culture. Nine of the 11 patients (82%) with nasal MRSA colonization had simultaneous MRSA colonization of the pharynx.

Figure 1.

 Flow diagram of the outcome of patients depending on the nasal swab result prior to percutaneous endoscopic gastrostomy (PEG) insertion. MRSA, methicillin-resistant Staphylococcus aureus.

Peristomal colonization and infection

All patients had at least one PEG site evaluation and swab (no day 2/3 evaluation in three cases (at the weekend); no day 7 evaluation in five cases (three deaths and two discharged)). In every case in which naso-pharyngeal colonization with MRSA was detected, it was subsequently cultured from the PEG site (Figure 1). By contrast, an MRSA-positive PEG site swab was obtained in only five out of 20 cases (25%) if naso-pharyngeal MRSA colonization was not detected (Figure 1).

Peristomal infection occurred in eight (26%) patients. MRSA was cultured from peristomal swabs in every case but one (88%), in which S. aureus and a coliform were the organisms detected. In addition, more than one organism was cultured from the PEG site in seven cases. Swabs from infected PEG sites revealed the following organisms: MRSA, n=7; coliforms, n=6; S. aureus, Enterococcus spp. and Group A and G Streptococcus, all n=1. Patients with nasal MRSA colonization had a significantly higher number of subsequent peristomal infections (n=6; 55%) compared with those patients with MRSA-negative nasal cultures (n=2; 10%; P=0.012, Fisher’s exact test; Table 2 and Figure 1) giving an odds ratio (95% CI) of 10.8 (1.6–70.9). By contrast, there was no association between naso-pharyngeal colonization with coliforms (n=13) and the risk of subsequent peristomal infection (n=4 coliform-positive vs. n=4 coliform-negative; P=0.69, Fisher’s exact test; OR 1.6 (0.3–7.9)). Six peristomal infections occurred in patients who had not received prior antibiotic therapy (n=18; six nasal MRSA-positive) compared with two infections in patients (n=13; five nasal MRSA-positive) who had been given systemic antibiotics within 7 days of PEG insertion (P=0.41, Fisher’s exact test; OR 0.4 (0.1–2.2)). Logistic regression analysis confirmed that only naso-pharyngeal MRSA colonization (t=3.1, P=0.005) was significantly associated with subsequent peristomal infection.

Table 2.   Peristomal infection in nasal methicillin-resistant Staphylococcus aureus (MRSA)-positive and MRSA-negative patients Thumbnail image of

Treatment of peristomal MRSA colonization and infection

In all cases of MRSA colonization, a decontamination regimen was commenced, incorporating topical 10% povidone-iodine to the PEG site, nasal 2% mupirocin, 0.2% chlorhexidine mouth-wash, bathing in 2% triclosan (instead of soap) and 1% chlorhexidine powder application to the axillae/groins. Systemic antimicrobial chemotherapy was only required for peristomal infection in one case (MRSA- and Group G Streptococcus-positive).


This prospective study, set in a large UK teaching hospital, has highlighted the importance of naso-pharyngeal MRSA colonization in subsequent MRSA colonization and infection of PEG sites in the first 7 days following PEG insertion. A high concordance between naso-pharyngeal and peristomal colonization following PEG insertion by the ‘pull’ technique is not surprising as the internal gastrostomy positioning device is passed through the pharynx prior to placement adjacent to the gastric wall. A similar concordance (88%) between oro-pharyngeal and gastrostomy tube bacterial colonization has also been noted in the only other study investigating pharyngeal and PEG site microbiology concurrently following PEG insertion.6S. aureus has been noted to be the commonest organism in peristomal isolates in four5, 9, 13, 14 out of the five5, 8, 9, 13, 14 studies which have reported the microbiology of early peristomal infection following PEG insertion. In addition, a study of gastrostomy site isolates in residents of long-term care facilities in the USA found that S. aureus (including MRSA) was the commonest organism.16

The profile of naso-pharyngeal isolates obtained in our study was similar to that of the only other study of patients undergoing PEG insertion in which oro-pharyngeal swabs were taken prospectively.6 The high prevalence of MRSA and coliforms in the naso-pharynx of the patients in our study is consistent with the fact that the majority of patients who underwent PEG insertion were frail, debilitated and had already been hospitalized for days to weeks, often having received antibiotics for previous infections (42% within 7 days prior to PEG insertion). Cerebrovascular disease was the commonest indication for PEG insertion in our study, whereas malignancy was a minor indication (10% of cases). Patient characteristics may explain the discrepancy between our study and that of Preclik et al.,8 in which malignancy was the indication for enteral feeding in 65% of cases and in which S. aureus was isolated from peristomal infection sites in only 8%. We and others believe that studies (such as that reported here) in which chronic neurological disease is the commonest indication for PEG insertion are probably most relevant to hospital practice in the UK and elsewhere.17, 18 In support of this, MRSA has been found to be the predominant organism cultured from peristomal swabs in another centre in the UK.19

The European Society of Gastrointestinal Endoscopy2 and American Society for Gastrointestinal Endoscopy3 guidelines recommend the use of cephalosporin prophylaxis and the more recent British Society of Gastroenterology4 guidelines recommend cephalosporin or co-amoxiclav prophylaxis for PEG insertion. These antibiotic regimens would have reduced efficacy in institutions with significant MRSA carriage and infection rates. This may account for the marked variation in antibiotic prophylaxis practice in the UK prior to publication of the British Society of Gastroenterology guidelines.20 Indeed, variability in bacterial flora (and hence antibiotic sensitivity) between different units could account for the conflicting data regarding the benefit of antibiotic prophylaxis which have been published.5–11 The alternative strategy of reduction of naso-pharyngeal flora using antiseptic agents prior to PEG insertion has been recommended by some authors,1 but has never been subjected to a randomized trial. Our data clearly suggest that a study of naso-pharyngeal MRSA eradication, prior to PEG insertion, is warranted.

The prevalence of naso-pharyngeal MRSA carriage in our study was high (35%). The patients in the study came from a variety of medical wards, but not the intensive care unit (ICU). Data on MRSA prevalence on general medical wards in UK hospitals are scarce. Therefore, the applicability of our data to other UK institutions remains unclear. However, the MRSA prevalence in the ICU at St James’s University Hospital is 10.2%, which is similar to other ICUs.21 This may imply indirectly that the MRSA prevalence in this study may be representative of that in PEG patients in other general hospitals.

Following identification of MRSA in the peristomal area after PEG insertion, a topical antisepsis regimen was used. This may explain the limited use of systemic antibiotic therapy in our series, despite a similar peristomal infection rate to that reported by several other groups.1 Therefore, an alternative (or complementary) approach to reduce peristomal infection rates, especially in institutions with a high MRSA prevalence, could be the use of screening for MRSA carriage and subsequent peristomal antisepsis following PEG insertion for those patients with MRSA colonization.


We thank Gill Wyllie, Gillian Hodgson and Anne Pollard for their contribution to the study. Mark Hull is an MRC (UK) Clinician Scientist Fellow.