|BSG Guidelines in Gastroenterology 2007
ANTIBIOTIC PROPHYLAXIS IN GASTROINTESTINAL ENDOSCOPY
MC Allison, JAT Sandoe, R Tighe, IA Simpson, RJ Hall, TSJ Elliott
1.0 EXECUTIVE SUMMARY
1.1 Infective endocarditis is an extremely rare complication following endoscopy. Because its sequelae can be devastating, it is recommended that antibiotics are given before endoscopy in certain circumstances even though there is no good evidence that antibiotics can prevent this complication.
1.2 Antibiotic prophylaxis is recommended for all endoscopic procedures in patients at high cardiac risk for endocarditis. Evidence Grade III, Recommendation Grade B.
1.3 Antibiotic prophylaxis should also be given to patients at moderate risk of endocarditis, or those with a history of recent vascular grafting or stenting, undergoing a procedure with a high risk of bacteraemia. Procedures with high risk of bacteraemia are oesophageal dilatation, variceal injection, and thermal procedures such as laser or argon beam ablation of a tumour. Oesophageal stenting and interventional procedures done under endoscopic ultrasound guidance are included in this category even though studies on associated bacteraemia are lacking. Evidence Grade IV, Recommendation Grade C.
1.4 The recommended regime for adults is a single slow intravenous injection of amoxicillin (1g for adults), followed by a single slow intravenous injection of gentamicin (1.5mg/kg). The post-procedure dose of amoxicillin has been abandoned. Teicoplanin should be given instead of amoxicillin if there is a history of penicillin allergy. Evidence Grade IV, Recommendation Grade C.
1.5 Antibiotic prophylaxis is also recommended in circumstances where immunosuppression and/or neutropenia might cause symptomatic bacteraemia to become a potentially life-threatening event. Evidence Grade IV, Recommendation Grade C.
1.6 Microbiological advice based on any recent positive cultures should be taken into account when deciding on antibiotic prophylaxis regimens. Evidence Grade IV, Recommendation Grade C.
1.7 Patients having a percutaneous endoscopic gastrostomy (PEG) should normally receive a single dose of intravenous cefuroxime or co-amoxiclav during the hour before the procedure. Evidence Grade Ia, Recommendation Grade A.
1.8 Those already receiving broad spectrum antibiotics do not require additional prophylaxis for PEG unless endocarditis cover is indicated. Evidence Grade III, Recommendation Grade B. The choice of antibiotic for patients having PEG with a history of serious penicillin sensitivity has not been established, but clindamycin can be given.
1.9 Antibiotic prophylaxis is recommended for all patients with pancreatic pseudocyst undergoing ERCP. Patients with ongoing cholangitis should have already been established on antibiotics. Routine prophylaxis for ERCP is not necessary in most other circumstances provided that adequate biliary decompression can be achieved. Evidence Grade III, Recommendation Grade B.
1.9 There are specific circumstances where antibiotic prophylaxis should be given routinely to cover ERCP. These include patients with immunosuppression (e.g. following liver transplant), and those biliary disorders, such as primary sclerosing cholangitis or hilar cholangiocarcinoma in whom it can be anticipated that complete biliary drainage will be difficult or impossible to achieve during one procedure. Evidence Grade III, Recommendation Grade B.
1.10 When prophylaxis for ERCP is given, oral ciprofloxacin or parenteral gentamicin is recommended. A combination of parenteral amoxicillin and gentamicin should normally be used to cover patients with endocarditis risk factors and evidence of biliary obstruction. Evidence Grade IIa, Recommendation Grade B.
1.11 The recommended antibiotic regimen for ERCP prophylaxis and persisting biliary obstruction post-ERCP may need to be altered locally in the light of epidemiological patterns in isolates of microorganisms resistant to these agents. Evidence Grade IV, Recommendation Grade C.
1.12 Patients with suspected variceal bleeding, or patients with decompensataed liver disease who develop acute gastrointestinal bleeding, should have already been established on intravenous antibiotics before undergoing endoscopy. The choice of antibiotics will normally be guided by regional liver unit practice. Evidence Grade Ia, Recommendation Grade A.
1.13 Units should ensure that there is a routine method (such as a check list) for drawing the attention of the endoscopist to the patients to whom antibiotics should be given. A specimen pre-assessment proforma is attached.
These guidelines aim to help clinicians in deciding which patients undergoing gastrointestinal endoscopy should receive antibiotic prophylaxis. They apply to the prevention of infective endocarditis in patients with cardiac risk factors, and also to patients at low risk of endocarditis undergoing procedures that are associated with a high risk of bacteraemia.
The British Society of Gastroenterology (BSG) first published guidelines on the use of prophylactic antibiotics in 1996, and these were revised by Professor Mike Bramble in 2001. The British Cardiac Society (BCS) updated its guidelines on the prophylaxis and treatment of infecitive endocarditis in 2004. This prompted the Endoscopy Committee of the BSG to convene a further Working Party, which met in March 2006. This was chaired by Dr Robin Teague, and, in addition to members of the Endoscopy Committee, incorporated representation from the BCS and the British Society for Antimicrobial Chemotherapy (BSAC). Dr Miles Allison researched the current literature, prepared the briefing documentation, and, after the Working Party met, set about revising the previous version of the guidelines and preparing the first draft of the current guidelines. Further changes have been made in the light of comments from members of the Endoscopy Committee and the Working Party.
The members of the Working Party were not unanimous in their opinions concerning the risk of endocarditis following endoscopy. The recommendations are therefore based on a majority consensus and are in concordance with the recently published guidelines for the prevention of endocarditis published by BSAC (1).
The guidelines conform to the North of England evidence based guidelines development
Project. The grading of each recommendation is dependant on the category of evidence supporting it. Recommendations based on the level of evidence are presented and graded as:
A: requires at least one randomised controlled trial of good quality addressing the topic of
recommendation (evidence categories Ia and Ib);
B: requires the availability of clinical studies without randomisation on the topic of recommendation (evidence categories IIa, IIb and III);
C: requires evidence from expert committee reports or opinions or clinical experience of
respected authorities in the absence of directly applicable clinical studies of good quality (evidence category IV).
2.3 Scheduled review
The content and evidence base for these guidelines should be reviewed within five years of
Bacteraemia is common following some forms of gastrointestinal endoscopic therapy, such as dilatation or submucosal injection, and can occur with diagnostic endoscopy alone. Fortunately complications resulting from dissemination of endogenous bacteria are uncommon, and infective endocarditis is a very rare complication. The evidence to support the view that antibiotic prophylaxis can reduce the incidence of infective complications is somewhat limited.
The area that has attracted the most controversy in recent years has been the use of antibiotic prophylaxis in the prevention of infective endocarditis. The recommendations by the American Heart Association (2: but see Footnote 14 below) have guided the advice of the national bodies representing endoscopic practice (3,4), including the BSG (5). The traditional recommendation has been that patients at high risk of endocarditis, such as those with a prosthetic (i.e. tissue or artificial metal) valve and/or a past history of endocarditis should receive antibiotics for all endoscopic procedures. More recently the European Society of Cardiology has recommended antibiotic prophylaxis to cover therapeutic endoscopy in patients with acquired valvular heart disease (moderate endocarditis risk) (6) and the British Cardiac Society has gone even further, advising antibiotic prophylaxis for patients at moderate risk of endocarditis undergoing any endoscopic procedure (7).
The fact that current guidance is conflicting has been recognised by the British Society for Antimicrobial Chemotherapy, which at the time of writing has produced the most recent guidelines (1). The Endoscopy Committee of the BSG recognised the need for consensus on this issue, and to this end convened a Working Party in the spring of 2006. The membership, comprised physicians with a special interest in gastroenterology, gastroenterologists, cardiologists and microbiologists. The gastroenterologists and microbiologists from this Working Party also took the opportunity to review the evidence underpinning the use of antibiotic prophylaxis in other areas of endoscopic practice, in particular Endoscopic Retrograde Cholangiopancreatography (ERCP) and Percutaneous Endoscopic Gastrostomy (PEG).
4. GENERAL CONSIDERATIONS
4.1. The aims of antimicrobial prophylaxis in gastrointestinal endoscopy are to prevent:
Colonisation of vascular grafts and endovascular stents, orthopaedic and other
non- cardiac prosthetic implants.
Pancreatico-biliary infection following ERCP.
Wound infection and peritonitis following percutaneous procedures.
4.2. The potential benefits of antibiotic prophylaxis should be weighed up against:
The potential contribution to the selection of resistant bacteria, such as MRSA
The knowledge that antibiotics may fail to prevent infective endocarditis or other infective complications
The small risk of adverse events, including anaphylaxis
The practical difficulties and costs of antibiotic administration, especially in patients who are allergic to penicillin
4.3 Endocarditis risk following endoscopy
Prospective studies to determine the value of antibiotic prophylaxis of endocarditis during gastrointestinal endoscopy are not available. Such research is unlikely to be done for two reasons. Firstly there is a natural reluctance to include patients at high risk of endocarditis into a placebo group, and secondly endocarditis complicating endoscopy is extremely rare, so prospective studies would need to recruit very large numbers of subjects. Thus, although recommendations can be based on an understanding of the pathology of infective endocarditis, they are of necessity pragmatic.
4.4. Identification of at-risk patients
Units should ensure that there is a routine method (such as a check list) for drawing the attention of the endoscopist to the patients to whom antibiotics should be given. The conditions which render the patient at high risk of developing endocarditis are listed later.
5.1. Evidence for bacteraemia in gastrointestinal endoscopy
The existence of bacteraemia during upper and lower gastrointestinal endoscopy has been well established in numerous series over decades (Table 1). These studies were reviewed in the previous edition of these guidelines (5) and more recently by Nelson (8). Some published papers overestimate the rates of potentially significant bacteraemia because microorganisms which are frequent contaminants (with little or no pathogenic potential) have been included. Other series, particularly some of the older studies, give misleadingly low rates because of deficiencies in culture techniques, especially those for anaerobic bacteria.
One study of bacteraemia associated with upper gastrointestinal endoscopy in immunosuppressed patients (in whom intravascular destruction of bacteria is minimised) reported a high rate of clinically significant bacteraemia (9/47, 19%) (9).
Bacteraemia during ERCP is considered to result mainly from contrast injection and manipulation around endogenous bacteria in bile and/or pancreatic tissue of patients with pre-existing pathology such as biliary obstruction or pseudocyst. Bacteraemia during ERCP is infrequent among patients without evidence of biliary or pancreatic ductal obstruction (8).
5.2. Clinical importance of bacteraemia
Recent evidence confirms that everyday activities such as chewing or tooth brushing produce a bacteraemia of dental flora (10,11). The incidence of bacteraemia during endoscopy has been extensively studied but the incidence of symptomatic bacteraemia is less well understood. In the great majority of cases endoscopy-related bacteraemia is not associated with any recognisable symptoms.
Thus, in most instances, there would seem to be little reason to attempt to reduce the rate of endoscopy associated asymptomatic bacteraemia in the absence of delayed clinical sequelae.
The most serious potential sequelae of bacteraemia include infective endocarditis, meningitis, cerebral abscess, and infected ascites (bacterial peritonitis) in patients with cirrhosis (8). These complications, whilst rare, are more likely to follow procedures associated with the highest risk of bacteraemia, namely oesophageal dilatation or injection sclerotherapy of varices.
5.3 Prevention of bacteraemia
One study has assessed prospectively (but in an open study design) the efficacy
of antibiotic treatment in reducing bacteraemia rates during endoscopy (12). Alternate
patients aged 60 years and over undergoing gastroscopy were given antibiotics. Blood
cultures were negative in all 130 patients receiving antibiotics but positive in 13/132
controls (9.8%, p<0.001). However, the microorganisms isolated could all have been skin
contaminants, and neither the patients who received antibiotics nor the controls experienced any symptoms likely to have been associated with bacteraemia.
6. INFECTIVE ENDOCARDITIS
6.1. Background and literature survey
Over recent decades the numbers of gastrointestinal endoscopic procedures have been carried out worldwide has risen exponentially. It is therefore reassuring that there is no evidence of any concomitant increase in the incidence of endocarditis. Reports of endocarditis associated with endoscopic procedures have occurred less than once per year (Table 2) and it is not clear even in this small number of cases whether the association was always causal. On the other hand not all cases of infective endocarditis following endoscopy are reported, and the association may not always be recognised.
A UK collaborative survey of 582 patients with infective endocarditis identified three patients in whom there was a history of recent gastroscopy (13). The significance of these findings has been questioned because there was no control group. The other case reports linking infective endocarditis to recent endoscopic procedures (14-30) are summarised in Table 2. Some important points arise from these cases:
High risk patients such as those with tissue or mechanical prosthetic valves are included (17,19)
There is an example of failure of antibiotic prophylaxis (17)
There are examples of patients with no prior history of cardiac disease (16,23)
Other clinical factors may have influenced the endocarditis risk (18,19)
Marked variation in time interval between endoscopy and onset of symptoms
6.2. Does antibiotic prophylaxis prevent endocarditis?
There is only limited evidence that antibiotic administration during dental or surgical procedures reduces the risk of endocarditis (31). Failures of antibiotic prophylaxis to prevent
endocarditis are well recognised (32), and include an example of failure of prophylaxis
during a gastrointestinal endoscopic procedure (17). In the rabbit model, antibiotic prophylaxis has been shown to reduce the risk of infection in damaged valves following high bacterial challenge (33). A retrospective case control study of patients at risk suggested that antibiotic prophylaxis might be effective in preventing endocarditis in dental practice (34), but a recent Cochrane review has come out against the routine use of prophylactic penicillin for invasive dental procedures (35).
Many patients with cardiac pathology are unaware that they have any abnormality. Thus ascertainment of at-risk patients will always be incomplete. One member of our Working Party believes that “the presence of moderate or high risk cardiac conditions should be excluded by a careful history and physical examination before performing any invasive investigation including endoscopy” (36). There are real practical difficulties inherent in putting such advice into practice:
Many patients referred for endoscopy could harbour previously unknown valvular disease. Over half of males aged over 65 years have systolic murmurs (37). It has been suggested that one in six patients attending for diagnostic endoscopy has a clinically detectable valvular lesion (38).
Not all patients with documented acquired valvular disease will carry an antibiotic card or be aware of the findings from previous echocardiography.
Tens of thousands of patients per year in the UK undergo direct access endoscopy under the care of a nurse endoscopist.
Even if guidelines are established, many patients may be given the wrong antibiotic or none at all.
6.3. Risk factors for endocarditis
The risk of endocarditis varies according to the nature of the underlying cardiac condition. For the large majority of patients there is worldwide consensus on the classification of endocarditis risk (2, 39-41). Nonetheless there are some differences of opinion in this area. European cardiology working parties have classified complex and/or cyanotic congenital heart disease within the high risk bracket (6,7). One of these groups also regards mitral valve prolapse to be a high risk cardiac lesion in circumstances where there is marked valve thickening and/or regurgitation (7). Given the recent advice to dentists by the British Society for Antimicrobial Chemotherapy (1) that routine antibiotic prophylaxis should be restricted to those with a history of previous endocarditis, prosthetic heart valves and surgically constructed intra-thoracic vascular conduits, the majority of the BSG Working Group agree that the high risk group should comprise the above three conditions (Table 3). This is in keeping with American guidelines, which encompass also the rare scenario of endoscopy in patients with complex cyanotic congenital heart disease (e.g. single ventricle states, transposition of the great arteries, tetralogy of Fallot ) (2,4).
The risk of endocarditis is probably influenced by the incidence (Table 2) and intensity of bacteraemia associated with the endoscopic procedure and on the potential virulence of the micro-organism(s). Indeed bacteria vary greatly in their propensity to infect damaged heart valves.
6.4. Recommendations for endocarditis prophylaxis during endoscopy (Tables 4 and 5)
Previously healthy patients not known to have cardiac lesions who undergo procedures
associated with a low incidence of bacteraemia have an extremely low risk of endocarditis. Antibiotic prophylaxis is therefore not justified in these circumstances.
Patients with cardiac lesions associated with a high risk of endocarditis should be given antibiotic prophylaxis for all endoscopic procedures. Those with a moderate risk of endocarditis who undergo gastrointestinal endoscopic procedures with a known high risk of bacteraemia (42-48) have an increased risk of endocarditis and should also receive antibiotic prophylaxis. It is recognised that the American Heart Association, over a year after the meeting of our Working Party and at the time of submission of the guidelines herein, have concluded that there is no evidence to support the administration of antibiotics with the sole aim of preventing endocarditis after gastrointestinal and genitourinary procedures (see Footnote 14 below).
Recommendations for endocarditis prophylaxis are set out according to cardiac and procedure-associated risk factors in Table 4. Recommendations on the type of antibiotics, doses and timing of administration are given in Table 5. Teicoplanin has replaced vancomycin in patients who are allergic to penicllin. There are two reasons for this: firstly teicoplanin is easier and quicker to administer, and secondly there is some evidence that teicoplanin has a longer duration of action following a single dose (49).
7. ENDOSCOPIC RETROGRADE CHOLANGIOPANCREATOGRAPHY (ERCP)
7.1 Bacteraemia is well recognised during ERCP (45-48). Pancreato-biliary infection occurs after 0.4-0.8% of endoscopic biliary procedures. These episodes must always be taken seriously because fatality may occur in 8-20% of such cases (50).
7.2 It was initially believed that the failure of early studies of antibiotic prophylaxis to show benefit was because the case mix included both diagnostic and therapeutic procedures. Infection is rare after diagnostic ERCP in the absence of stones or pancreatic or biliary obstruction. In patients with obstructed bile ducts with features of previous infection, or pancreatic pseudocyst, the available data suggests a reduction in clinically significant infective complications when prophylactic antibiotics are used (47,51,52). On closer scrutiny of these papers, however, the examples of procedure-related cholangitis were almost all in patients for whom adequate biliary drainage had not been achieved during ERCP. The contention that relief of obstruction is more important than antibiotic prophylaxis is reinforced by the finding that the chief predictor of infective complications after therapeutic ERCP is incomplete bile duct drainage (53).
Although not all authorities are in agreement (54-57), the case has been made for prophylactic administration of antibiotics for patients likely to undergo a therapeutic procedure in the context of ongoing biliary obstruction and/or infection and/or pancreatic pseudocyst (58,59). There is also a suggestion that antibiotic prophylaxis is cost-effective in these circumstances (58). This begs the question as to what constitutes “biliary obstruction”. Patients presenting with bacterial cholangitis should already be established on antibiotics at the time of ERCP. Patients with jaundice secondary to obstructing common duct stones or strictures may not necessarily require antibiotics provided that the obstruction can be properly relieved at ERCP (54). Similarly non-jaundiced patients with common duct stones may not need antibiotic cover provided that the stones can be removed or drainage can be secured by means of stenting (with or without biliary sphincterotomy). These arguments have led Subhani and colleagues to propose that antibiotics can be administered immediately after ERCP if it has not been possible to decompress the biliary tree (56). This approach has not been tested in clinical practice and deserves further study.
Other factors that are important in reducing the risk of infection include (i) the optimal cleansing and disinfection of the endoscope; (ii) the employment of single use accessories down the working channel of the duodenoscope; and (iii) the use of sterile contrast medium and careful control of the volume of contrast used. Some experts advocate that the endoscopist should aspirate bile from the biliary tree in order to attenuate the rise in intrabiliary pressure following contrast injection. Some authorities add antibiotics to the contrast media prior to injection. Neither of these two strategies, however, has been shown to reduce the risk of bacteraemia or cholangitis.
The choice of antibiotic depends on the clinical context. Common causative micro-organisms in ascending cholangitis are Escherichia coli, Pseudomonas aeruginosa, Klebsiella spp., enterococci, coagulase negative staphylococci and Bacteroides spp, but many infections are polymicrobial (56).
Our recommendations are summarised in Table 5. There are several scenarios to consider:
Patients with cardiac pathology at moderate or high risk of endocarditis. Those with moderate endocarditis risk should receive endocarditis prophylaxis if biliary obstruction and/or pancreatic pseudocyst are suspected. All patients at high risk of endocarditis should receive prophylaxis as discussed in Section 4, and Tables 4 and 5.
Patients with ongoing pancreatic or biliary sepsis. The choice of antibiotic prophylaxis is best guided by the results of recent microbiological cultures. These patients will normally already be receiving appropriate antibiotics, and the need for additional cover should be discussed with a clinical microbiologist.
Patients with first ERCP for biliary obstruction, no clinical evidence of infection and low endocarditis risk. In these circumstances it is reasonable for the endoscopist to elect not to give antibiotics pre-procedure provided that their administration is ensured as soon as possible post-procedure if adequate decompression of the biliary tree has not been achieved. The course of antibiotics should continue whilst arrangements are being made to relieve biliary obstruction as soon as possible (either by repeat ERCP or by some other means) and should last at least until this endpoint has been achieved.
The choice of antibiotics has been debated and reviewed in depth (56-58) and the role of specific antibiotics is discussed in Section 12. Most authorities recommend either oral ciprofloxacin taken 90 minutes before the procedure, or intravenous gentamicin at the time of sedation. Both ciprofloxacin and gentamicin have gaps in the cover they provide. Both have generally good activity against Gram-negative aerobic bacteria but are much less active against many Gram-positive species, including enterococci. Increasing ciprofloxacin resistance among coliforms (Enterobacteriaceae) has also been reported (60). Therefore the choice between ciprofloxacin and gentamicin may be influenced by local epidemiology in microbial resistance.
Oral ciprofloxacin is less expensive than the intravenous formulation and results in satisfactory blood concentrations. Although gentamicin does not penetrate into bile very well, and has limited activity against enterococci, it probably has broader Gram-negative activity than ciprofloxacin. Neither regimen provides particularly good cover for an obstructed biliary system where enterococci are implicated in up to 40% of infections (55). Therefore the combination of amoxicillin with continued treatment with the antibiotic chosen for prophylaxis should be considered in a patient who became febrile post-procedure.
Patients with a history of prior biliary manipulations: Bile within the biliary tree is normally sterile. ERCP with sphincterotomy and/or stenting disrupts the normal ampullary barrier to the gut, and is associated with long term bacterobilia (61,62). It is therefore logical to infer that patients needing repeat biliary intervention at ERCP are at increased risk of bacteraemia and cholangitis. In a large prospective series (as yet unpublished) cholangitis complicating ERCP was more likely to occur in patients with a history of prior ERCP (with sphincterotomy and/or stenting) (P Cotton: personal communication). Patients who have been receiving continuous antibiotic prophylaxis for the prevention of recurrent symptomatic bacteraemia following biliary stenting may have acquired resistant bacterial flora, and should be given a different antibiotic to cover further biliary endoscopic procedures such as stent changes. Because of the lack of an evidence base, we believe that the decision as to whether to use prophylactic antibiotics in immunocompetent patients undergoing repeat ERCP rests with the endoscopist, the local clinical microbiologist and the clinical team caring for the patient. When ERCP is performed in patients who have previously received full treatment courses of one antibiotic, consideration should be given to the use of an alternative antibiotic (or combination of antibiotics) to cover the procedure. For example, if a patient has been exposed to prolonged and/or frequent ciprofloxacin, a combination of amoxicillin and gentamicin, or monotherapy with a wider spectrum penicillin such as piperacillin with tazobactam, could be given.
Other settings in which prophylaxis for ERCP should be given include (i) immunosuppressed patients including those undergoing biliary intervention post liver transplant, and those with neutropenia; (ii) patients with known Caroli’s disease or primary sclerosing cholangitis, not only because bacterial cholangitis is common following biliary manipulation (63) but also because complete relief of biliary obstruction is unlikely to be achieved at ERCP; (iii) patients with Bismuth type III or type IV cholangiocarcinoma, for whom it may likewise be difficult or impossible to secure drainage of all liver segments; (iv) patients with pancreatic pseudocysts; and (v) those with a history of recent vascular graft insertion.