In Question 13 from the second paper of 2001, the candidates were presented with an elective ICU admission who happens to also have an AICD. "How does this device affect your management" they asked. Conditions which pose as indications for the insertion of an AICD are farm more threatening than the AICD itself, in my opinion. However, there are implications to having an angry electric eel in your chest. There is a beautiful and freely available article by Sergio Pinski (2000) which lists not only the common AICD-related problems, but also the solutions to them. This article is paraphrased here to simplify revision.
Total device failure: there is no pacing or AICD activity. The device appears dead for all intents and purposes. There are several possible causes for this:
Pacing failure: the device seems to be working (pacing spikes are seen on ECG) but there is no capture. This usually means something has happened to its interface with the myocardium.
Failure to defibrillate VT or VF: the patient is clearly dying but the AICD for some reason refuses to rescue them. Why might that be? It is usually some sort of programing error. For instance:
Overenthusiastic defibrillation: the device is shocking the patient relentlessly.
Inappropriately normal function: the device missed the family conference, and does not realise the patient is being palliated. In these situations the AICD should be disabled. Ethical issues arise if the patient has an underlying complete heart block or something similar (in which you might want to merely disable the defib function).