Home >> Dgroups Rss >> AI and informed consent (2)

AI and informed consent (2)

30 January, 2024

Chris thank you for posting this paper. [ https://www.hifa.org/dgroups-rss/chatgpt-vs-uptodate-4-ai-and-informed-c... ]

My feeling after reading the article is to wonder why more and more effort seems to look for ways / things to interfere with and reduce quality time that every practitioner should spend with their patients. There seems to be an increased tendency to do everything to limit direct interaction with patients. What other activity in care is more important than speaking to, touching the patient, getting verbal and non verbal cues about the patient’s view of what the practitioner wants to carry out on the them? It used to be frowned upon to delegate giving of informed consent to anyone other than by the one to carry out the action/procedure, because frankly, how long does it take to inform your patient about what you are about to put him / her through?. Why should a very human interaction be delegated to some voice-mail kind of interaction? Why cant the time be factored into the practitioners time table and schedule, if he takes the patients welfare seriously? If the clinician does not have time to give information about his procedure, what else does he have time for? Is it for administrative efficiency? It reminds one occasions when one puts a call to a utility or supplier or the bank and expecting to talk to a fellow human being, but then you hear a machine-generated voice after a long wait. The disappointment is usually obvious even if your question is answered. And that is you are not ill or about to have something done to you, like surgery that has potentially serious side effects, including severe disability or even death!.

The authors say that i) ‘Clinicians benefit from consent delegation to LLMs by streamlining clinical workflow and improving administrative inefficiencies’; ii) that ‘A study investigating the effectiveness of an AI-powered consent agent, shows that the total time from referral to consent completion was 11 days faster via AI than human-based interactions’; and iii) that consent delegation may allow clinicians to focus their time on more complex clinical tasks and spend longer with patients who need it’. But, I ask again, where is the doctor-patient relationships in all these considerations that aim to justify delegating informed consent to AI, not even to a trained and informed junior member of the team!.

I am encouraged by the paragraph in the paper that cites a legal case, and says, ‘---- some legal systems may reject consent delegation in medicine. This was the case in Shinal v Toms8 in which the Pennsylvanian Supreme Court found neurosurgeon Steven Toms had failed to obtain valid consent on the basis that he had not personally provided sufficient information to the patient. Instead, Toms had delegated part of the consent process, including signing of the consent form, to his physician’s assistant, who failed to indicate which surgical approach the patient had chosen.8 Clearly, if procedural consent cannot be delegated to another qualified health professional, it is unlikely to be acceptable to delegate to an LLM.’.

In feel that this example is useful for any clinician who may not want to spend the time necessary to perform the ritual of giving informed consent, but rather wants to doing something else, like administration duties, to take note!. I take the view that it is not only a legal imperative, it is a moral and ethical duty of care to the patients to spend necessary time to give informed consent.

I am also drawn to the GMC guideline no. 1- (c) that says consent delegation should take account of ‘whether the patient has already developed a trusting relationship with you or the person you would delegate to— ‘. Surely, I cannot imagine that patients would ever trust AI more than their caring, human practitioner. Sounds old fashions but somethings should stay as they are for they just sound and make sense.

It is great to continue to innovate and introduce cutting-edge methodologies (even though I do not actually class AI giving informed consent as cutting edge), but it is difficult to see how reducing the benefits of spending time with patients, especially before a frightening intervention like surgery, can be considered an advance in delivering care. Surely, let’s find better ways and means to make patient care as physicians better, simpler, more effective, and even efficient (saving money), but great care needs to be taken when there is judgement is to be made whether to reduce certain human contact and replace it with machines.

Finally, one is encouraged to read the authors comment, that ‘--- there is still much left unanswered relating to how LLMs would be practically applied to the consent-seeking process’.

Joseph Ana

Prof Joseph Ana

Lead Senior Fellow/ medical consultant.

Center for Clinical Governance Research &

Patient Safety (ACCGR&PS) @ HRI GLOBAL

P: +234 (0) 8063600642

E: info@hri-global.org

8 Amaku Street, State Housing, Calabar, Nigeria.


HIFA Profile: Joseph Ana is the Lead Senior Fellow/Medical Consultant at the Centre for Clinical Governance Research and Patient Safety (CCGR&PS) with Headquarters in Calabar, Nigeria, established by HRI Global (former HRIWA)... jneana AT yahoo.co.uk