The Patient Who Said No

An increasing number of patients are refusing ambient AI recording, and many health systems are treating this refusal as a problem to be managed, seeing these patients as a stubborn minority who do not understand the advantages. This perspective is backward. Patients that refuse to be recorded by a technology that is introduced to them with minimal information are not blocking progress. They are completing the risk benefit analysis that the health system failed to conduct.

Viewing refusal as resistance indicates a lack of introspection and a lack of understanding of the patient. It means the health system believes the technology is good, the data would be handled in a safe manner, and the patient would understand the benefit of the technology if only it were explained better. None of these are self-evident to a patient in a situationally coercive recording environment who is being asked to give consent to a device that would make an omnipresent recording of the most private conversation of the week. In this case, refusal is not a barrier to progress. In fact, the refusal is the only rational action that is taking place.

This is important now as the reasons for patients to proceed with caution were not present a few years ago. Trust in institutions has been at its lowest in decades and the erosion of trust is not different for the healthcare industry. Patients are no longer passive viewers. When a health system as large as Sharp HealthCare faces a class action lawsuit for allegedly capturing conversations through ambient recordings with no consent and retaining records that claim consent, the cautious patient is not being paranoid. They are informed.

Refusal is information, not obstruction

A patient who says no is providing data to the organization, if anyone is actually reading it. A refusal to consent is an indication that the consent process was ineffective, the patient was not adequately informed of the purpose of the tool, or the patient does not have confidence in the destination of their words. All of these are critical pieces of information. An organization that views refusal as merely a consent gap to be filled, misses the most honest feedback it has ever received.

There is a less obvious angle. The patient who declines is making a choice that the health system should have made first. Before a technology gets implemented in an exam room, someone in the health system should have made the choice whether the system is even capable of protecting patient data and whether the captured data is worth the potential benefit for the patient. When the patient raises questions at the bedside, they are making up for a failure of the health system to have that discussion prior. The awkwardness that results is actually beneficial, as it indicates the work that was neglected.

Consent Theater

Most consent today is a performance. Patients are handed tablets containing dense, legalistic language, and asked to tap agree while the next appointment is starting, with no explanation of what data are being captured, where the data will go, and what the consequences are for opting out. There is rarely a clear alternative, which shows the patient that declining is not an option. It is not consent. It is a signature that is collected under mild duress, and is intended to protect the institution, not the patient.

Patients feel the difference in approach, even when the difference is not articulated. A process that is designed to elicit agreement feels different from a process designed to help a decision be made. One is primarily focused on satisfying the process by getting a box checked. The other is prepared to continue the decision-making process, even if it means hearing no. When patients feel that the process is primarily focused on getting agreement, the rational choice is to withhold, and that is what many are doing.

The skepticism is earned

It could be easy to ignore patient refusals if worries were not based on reality. They are based on reality. Claims that documents were created with permissions where no permissions were given is the kind of thing that creates distrust amongst the public regarding the documentation. The technology is not without fault either. AI created clinical documentation had been studied and in some cases, such documents had hallucinations and omissions. Some of these errors were of such serious nature that the subsequent healthcare had been negatively impacted. A patient does not lack justification in believing the system to be less than adequate.

When all these factors are taken into consideration, it is apparent that patient refusals are based on more than just irrational fear. The refusal is a form of judgment. Patients are unable to review the vendor contracts or the data retention policies. However, these patients can decline to contribute their data to a system that is unverified and that they are unable to evaluate. This is a logical conclusion given the inadequate information the patients usually receive.

Making no a real option

If refusals are logical then patients should not be persuaded to change their refusal. This means that a process should be developed that truly accepts the refusal and builds a process that earns a yes. This means that patients should first be informed in clear understandable terms what data is being collected, what the data is used for, how long the data is kept, and who has access to the data. It also means that patients should be informed of the other alternatives which means that the healthcare provider should be prepared to proceed with the healthcare visit even if the patient chooses to refuse. Further, a refusal should not be treated as a hindrance.

It means understanding that consent creates an ongoing relationship, not a one-time agreement. A brief communication about the functionality of the tool, and the reason behind its use, helps build a relationship based on trust, more than any form or template. It is also about keeping the consent record separate from the system to which consent is being given, to make sure that the consent was given by the patient and not the system on behalf of the patient. None of this requires a high expenditure. What is most needed is the acceptance of the design to hear a no and to provide a solution that hears a no instead of creating a system that only knows how to hear a yes.

Earned trust moves further than speed

The patient trust, rather than the speed of deployment or the volume of registrations, will determine the success or failure of an organization during this phase. A patient who chooses and understands the implications of their agreement is more valuable than ten patients who completed a task on a system without reading the information and far more valuable than a patient who learned that a choice was not truly theirs and that their interaction was recorded.

Christopher Hutchins Founder and CEO, Hutchins Data Strategy Consultants