Endoscopy Practice and Sedation: Antibiotics, Devices, Capsule, and Adverse Events

0:00Welcome to Board Pearls. This is episode four of four of the Endoscopy Practice and Sedation chapter, in the Endoscopic Procedures module. In this episode we cover antibiotic prophylaxis, device management, capsule endoscopy, and the adverse event lexicon: endocarditis prophylaxis is NOT given for GI endoscopy with narrow specific indications for procedural antibiotics, pacemakers and ICDs require electromagnetic interference management with monopolar cautery, capsule endoscopy contraindications and patency capsule, and the ASGE adverse event lexicon that grades severity from mild to fatal.

0:37The four items in this episode look unrelated on first inspection. They are not. Each one tests a single recognition: when the reflexive answer is wrong because the underlying principle has shifted. Antibiotics for a prosthetic device sound right and are wrong. Pacemaker reprogramming for every polypectomy sounds right and is wrong. Pacemakers as a capsule contraindication sound right and are wrong. Grading a complication by what it felt like sounds right and is wrong. Each of these has a clean principle that replaces the reflex, and that principle is the teaching.

1:17Start with antibiotic prophylaxis, because the misconception this domain is built around is the prosthetic device. The 2007 American Heart Association endocarditis prophylaxis guideline removed gastrointestinal procedures from the list of indications. That is the headline. Endocarditis prophylaxis is not given for GI endoscopy. That covers prosthetic heart valves alone, pacemakers, implantable defibrillators, synthetic vascular grafts, prosthetic joints, and a history of endocarditis. The reasoning is that the transient bacteremia of endoscopy carries an absolute endocarditis risk too low to justify routine antibiotics in an immune-competent patient. The brief bacteremia that does occur is clinically insignificant. The prosthesis itself does not change the endoscopic prophylaxis decision. That is the rule the boards test repeatedly. The reflex stem describes a prosthetic hip and a planned EGD with biopsy and asks about antibiotics. The answer is no antibiotics.

2:19Once the endocarditis-prophylaxis instinct is set aside, the remaining indications organize around one mechanism: closed-space infection. Prophylaxis is recommended only when the procedure introduces gut or skin flora into a space the immune system cannot clear. That principle predicts the list.

2:38The first scenario is ERCP with anticipated incomplete biliary drainage. Primary sclerosing cholangitis with multiple intrahepatic strictures that cannot all be drained qualifies. Hilar Klatskin tumors qualify, because cholangiography opacifies segments that may not be stented. Liver transplant patients with anastomotic strictures qualify. Complex post-surgical anatomy qualifies for the same reason. The mechanism is that retained contrast or bile in an undrained duct is a closed space, and instrumentation seeds that space with enteric flora. Coverage targets gram-negatives and enterococci, typically a fluoroquinolone plus metronidazole or ampicillin-sulbactam, and prophylaxis is continued after the procedure when drainage remains incomplete. If drainage is complete, prophylaxis is not needed, because there is no longer a closed space.

3:29The second scenario is EUS-guided fine-needle aspiration of cystic lesions of the pancreas or mediastinum. Same mechanism. The FNA introduces oral and skin flora into a previously sterile cyst, the cyst lacks the natural drainage of a duct or vessel, and infection there is morbid. Pancreatic cyst infection and mediastinitis are the consequences. Prophylaxis is a single pre-procedure dose plus three to five days of oral coverage. FNA of solid lesions is different. The post-FNA tract is not a closed space. The small bacterial inoculum gets cleared by the immune system. Solid-lesion FNA does not need prophylaxis. Lymph nodes, mediastinal masses, pancreatic solid tumors. No antibiotics. The closed-space principle predicts which FNAs need coverage and which do not.

4:20The third scenario is PEG and PEJ placement. The tube traverses oral flora and a percutaneous puncture site, and a single dose of cefazolin one gram intravenous about thirty minutes before the procedure covers both. One dose. Not a course. The tract heals over the subsequent days and ongoing prophylaxis is not needed.

4:42Two more closed-space indications exist outside the standard procedural list. Cirrhotic patients with acute upper GI bleeding receive ceftriaxone one gram intravenous daily for the duration of the bleed. The mechanism is bacterial translocation from a gut barrier that hemorrhage and hypoperfusion have disrupted, in a host whose cirrhotic immune dysfunction allows seeding of ascites or bacteremia. That is spontaneous bacterial peritonitis prophylaxis on the front end of the bleed, with a documented mortality benefit behind it. The detailed mechanism lives in the upper GI bleeding and cirrhosis chapters. Peritoneal dialysis patients undergoing lower endoscopy receive ampicillin plus an aminoglycoside before the procedure. Translocation across an instrumented colon into peritoneal dialysate is the same closed-space risk in a different anatomic space.

5:32The non-indications follow from the same logic. Diagnostic EGD and colonoscopy without high-risk features. EUS-FNA of solid lesions. Variceal band ligation in a non-bleeding cirrhotic patient. None of these create a closed space the immune system cannot clear. None get prophylaxis. The variceal point is worth pausing on, because the cirrhotic-on-ceftriaxone instinct from the bleeding patient bleeds into the non-bleeding case. It should not. The procedure does not drive the prophylaxis decision in cirrhotics. The bleeding episode does. Elective band ligation in a stable patient does not need antibiotics. The bleeding patient does, for the duration of the bleed.

6:13That covers the antibiotic domain. The principle is closed-space infection, the indications follow the principle, and the most-tested wrong answer is the prosthetic device. Move on to cardiac devices, because the same kind of obsolete reflex governs how device management was once taught.

6:30The principle for cardiac implantable electronic devices is electromagnetic interference, and the modern point is that the interference risk is small for routine endoscopic polypectomy because of how the current is delivered. Monopolar electrosurgery sends current from the active electrode through the patient's body to a return electrode somewhere on the skin. That current path can cross the device. Bipolar electrosurgery contains current between two electrodes within the snare or coagulation device itself, so the body is not in the circuit and the device is not exposed. Modern polypectomy with snare-based monopolar cautery does cross the body, but the return electrode placement and shorter current pulses limit the exposure enough that modern pacemakers tolerate routine endoscopic polypectomy without reprogramming.

7:16That tolerance is the testable point. Pacemakers do not need to be reprogrammed to asynchronous VOO mode for routine endoscopic polypectomy. The only requirement is placement of the return electrode, the grounding pad, away from the heart and the device. The lower back or the thigh works. The reasoning is that current passing from the active electrode at the polyp base to a pad on the thigh travels a path that does not cross the chest. Stray current cannot pass directly across the pacemaker. Routine reprogramming adds complexity without benefit. The exception is the pacemaker-dependent patient with monopolar electrosurgery in the cardiac field, near the gastroesophageal junction during APC or hemostasis for a high upper-body lesion. Those cases warrant a backup pacing strategy and cardiology input. The rare scenario where monopolar interference is misread as intrinsic activity could inhibit pacing in a patient with no intrinsic rhythm to fall back on.

8:15ICDs require an additional layer. The pacing function of an ICD behaves like a pacemaker and is fine. The arrhythmia detection algorithm is the problem. Electrosurgical interference looks like high-frequency electrical noise. The device's tachyarrhythmia detection can misread that noise as ventricular tachycardia or ventricular fibrillation and deliver a shock. Inappropriate ICD discharge is the complication. The fix is to suspend arrhythmia detection for the duration of electrosurgery. Two options accomplish that. The first is magnet placement over the device, which inhibits anti-tachycardia therapies on most ICDs as long as the magnet is in place. The second is formal device interrogation and reprogramming by an electrophysiology or device team, which disables arrhythmia detection explicitly. External defibrillator pads go on the patient before the procedure as a backup, in case a real arrhythmia happens during the disabled window. Arrhythmia detection is reactivated immediately after the procedure. The pacing function is not touched.

9:15One more point closes the device domain. Antibiotic prophylaxis is not given for either pacemakers or ICDs, because they are not endocarditis-prophylaxis indications under the 2007 guideline. The instinct to give antibiotics for any prosthesis attaches to cardiac devices as readily as it does to hip replacements, and the answer is the same.

9:37The third domain is capsule endoscopy, which is best taught through the procedure's failure mode. The capsule is a swallowed wireless camera that traverses the gastrointestinal tract and transmits images, and it visualizes small bowel mucosa that conventional endoscopy cannot reach. The capsule is rigid, twenty-six by eleven millimeters, and once swallowed it cannot be retrieved unless it lands somewhere a scope can reach. Failure of passage is the dominant complication, and that complication is called retention.

10:07Retention is defined as failure of the capsule to pass through the GI tract within two weeks of ingestion. The rate scales directly with the prevalence of focal narrowing. In a patient with suspected obscure GI bleeding and no known stricture, the retention rate is about one to two percent. In the populations where stricture is common, the rate climbs to five to thirteen percent. Crohn disease is the prototype, because inflammatory strictures are the dominant lesion. Prior bowel surgery, NSAID-induced enteropathy, prior abdominal radiation, and known small-bowel obstruction sit in the same elevated-risk band. The mechanism is mechanical. A focal narrowing in the small bowel allows liquid and small particulate content to pass, but the rigid capsule does not deform and gets trapped.

10:54When retention is suspected, the workup is imaging first. CT enterography or MR enterography localizes the capsule and characterizes the stricture. Retrieval is then attempted with device-assisted enteroscopy, single-balloon or double-balloon depending on location, with surgery reserved for cases where endoscopic retrieval fails or where the underlying stricture itself needs operative management.

11:17The way the system addresses retention risk in advance is the patency capsule, the Agile patency capsule specifically. It is a self-dissolving dummy. The patient swallows it before the diagnostic capsule. The patency capsule has a barium core surrounded by a lactose-and-barium body that dissolves over about thirty hours. It contains a small radio frequency identification tag that can be tracked externally or imaged on plain film. The logic is straightforward. If the patency capsule passes intact within the thirty-hour window, the gut is patent enough for the diagnostic capsule and the diagnostic study can proceed. If it does not pass, or it partially dissolves and fragments, the patient has a stricture tight enough to retain a capsule. The diagnostic study is then deferred or delivered endoscopically into the duodenum with arrangements for retrieval if needed. The patency capsule does not eliminate retention risk in Crohn patients, but it dramatically reduces it.

12:17Contraindications to capsule endoscopy have softened over the last decade in a way the boards test as a knowledge update. The classical absolute contraindications included pacemakers and implantable cardioverter-defibrillators, on the theoretical concern of electromagnetic interference between the capsule's radiofrequency transmitter and the cardiac device. Modern capsule platforms like the PillCam SB3 and beyond have been validated as safe in patients with pacemakers and most ICDs. Current ASGE guidance does not list cardiac devices as an absolute contraindication. Manufacturer verification for older platforms or unusual devices remains prudent, but the reflex that a pacemaker bars capsule endoscopy is no longer correct. Pregnancy remains a relative contraindication because of unproven safety, and elective small-bowel evaluation is typically deferred. Swallowing disorders that prevent ingestion, severe dysphagia or an esophageal stricture under ten millimeters in diameter, are workarounds-eligible rather than absolute exclusions. The capsule can be delivered endoscopically into the duodenum using a dedicated delivery device.

13:25That leaves the fourth domain, which sits over everything else in this chapter: how complications get graded once they happen. The ASGE adverse event lexicon, the Cotton lexicon, provides the standardized framework. It grades severity by the intensity of intervention required, and that is the conceptual move that makes the lexicon work. Intensity of intervention is what makes a complication consequential rather than incidental. A patient who needed an unplanned outpatient call had a mild event. A patient whose hospitalization extended to ten days had a moderate event. A patient who needed surgery had a severe event. The grade is the intervention.

14:05The four tiers are mild, moderate, severe, and fatal. A mild adverse event is unplanned but addressed with brief medical consultation or evaluation, with hospitalization under three nights and no transfusion. The planned procedure was not interrupted and no unscheduled subsequent intervention was required. A moderate adverse event requires four to ten nights of hospitalization. Or one night of ICU admission. Or transfusion of fewer than four units. Or a repeat endoscopic intervention. Moderate events meaningfully extend recovery and consume resources beyond the planned procedure. A severe adverse event requires beyond ten nights of hospitalization. Or more than one ICU night. Or four or more units transfused. Or surgery or interventional radiology consultation with intervention. Or permanent disability. Severe events change the patient's trajectory rather than merely extending recovery. A fatal event is death within thirty days attributable to the procedure or its complications.

15:07The intervention-intensity logic applies cleanly to the specific events the boards test. Post-polypectomy bleeding graded mild if managed without transfusion, moderate if requiring transfusion or repeat endoscopy, severe if requiring surgery or prolonged ICU care. Perforation is almost always moderate or severe by definition, because it virtually always requires either prolonged hospitalization or surgical or endoscopic closure. Post-ERCP pancreatitis is graded by hospitalization length. Mild if extended hospitalization runs two to three days. Moderate if four to ten days. Severe if more than ten days or with organ failure or surgical intervention. The full PEP risk-stratification and prophylaxis framework lives in the next chapter.

15:55The lexicon has one more move that the boards test, and it is the distinction between an incomplete procedure and a failure of intent. An incomplete procedure is one where the planned intervention could not be completed for any reason, often a technical or anatomic limitation. A failure of intent is a procedure that was completed but did not achieve its clinical goal. Those represent different quality concerns. An incomplete colonoscopy where the cecum was not reached is an incomplete procedure. A colonoscopy that reached the cecum but missed a polyp that became cancer on follow-up is a failure of intent. The reporting captures both because the operational lessons differ.

16:33Pull the four domains back together. Antibiotic prophylaxis turns on closed-space infection. Endocarditis prophylaxis is explicitly off the list for prosthetic valves and pacemakers and ICDs and joints and vascular grafts and a history of endocarditis. The mechanism predicts the indications. Incomplete ERCP drainage and cyst FNA and PEG and cirrhotic UGIB and PD-patient colonoscopy. Single-dose cefazolin for PEG and prolonged coverage for the closed-space scenarios. Pacemakers do not need reprogramming for routine polypectomy because monopolar containment plus return-electrode placement provides the safety margin, and only pacemaker-dependent patients in the cardiac field warrant backup strategy. ICDs need arrhythmia detection suspended by magnet or formal interrogation, with pacing function untouched and external defibrillator pads as backup.

17:31Capsule endoscopy lives or dies on retention risk, which scales with stricture prevalence and is addressed prospectively by the patency capsule, with the pacemaker and ICD contraindications retired on modern platforms. And the ASGE Cotton lexicon grades adverse events by intervention intensity, mild through fatal, with the incomplete-procedure-versus-failure-of-intent distinction sitting on top.

17:53The next chapter shifts from periprocedural framework to the procedures themselves, beginning with ERCP and EUS. The organizing questions there are how cannulation succeeds and how post-ERCP pancreatitis is prevented. Then when malignant biliary obstruction gets stented and when EUS-guided drainage takes over after failed ERCP. Then how altered anatomy changes the ERCP plan, how EUS reads the pancreatitis spectrum from acute to chronic, and how EUS tissue acquisition and celiac plexus neurolysis fit into the same toolbox.

18:28For the full chapter, the practice vignettes, and the topic-tagged question bank, head to board pearls dot com. You'll find the rest of the series on Apple Podcasts, Spotify, or wherever you listen to podcasts. That brings us to the end of episode four of four of chapter twenty eight, and I'll see you in the next one.