There’s an interesting article in the Journal of Hospital Medicine on what to do when a patient wants to leave the hospital against medical advice. After reading and rereading it I had to disagree with the conclusion but it took me a bit to get there because the article, with its confusing use of terms, is a masterpiece of obfuscation. The most obvious example is the oxymoronic use of the term “AMA discharge” in the title and throughout the article. If a patient leaves AMA it's not your decision. How is that a discharge? Put another way, if you discharge the patient you are making a statement that the patient is medically ready to leave the hospital. A discharge order can only mean the patient is leaving in accordance with, not against, your advice. Why, after all, would you enter an order for something that is against your own judgment?
Another example of language confusion is the authors’ statement that leaving the hospital AMA can be a process of informed consent. Quoting directly from the article:
Because all competent patients have the right to decline recommended inpatient treatment, the ethical and legal standard is that the physician obtain the patient’s informed consent to leave…
Consent to leave? That’s an inappropriate use of the word. Consent leads to adherence with the physician’s recommendation, which in this case would be to remain in the hospital. In the AMA situation the patient’s decision to leave is a demand, not a consent.
Getting past all the confusion, there were a few good points. When the patient leaves against your advice you don’t have to destroy the rapport. It doesn’t have to be an adversarial transaction. But the authors go beyond that principle by stating that when the patient leaves AMA not only should it be handled as a regular discharge but that you should not even document that the departure is against your advice. Again, form the article:
The solution to improve quality is straightforward—avoid designating discharges as AMA…
Treat all discharges similarly. Avoid designating an inpatient discharge as AMA.
That is where I have to disagree.
There's more to unpack. The authors make frequent mention of shared decision-making. Indeed shared decision-making is a is a core principle of evidence-based medicine but it is just one component. The AMA departure sometimes pushes shared decision making to the level of absurdity. How does it apply, for example, if the patient with an actively evolving myocardial infarction wants to leave the emergency room? What if the patient just swallowed antifreeze because he ran out of his beverage of choice? Where does shared decision making come in when the patient’s choice means almost certain harm? Some patient preferences and choices are simply wrong.
Finally there's the matter of legal protection. The authors make this statement:
Although clinicians may presume that the AMA designation provides protection from liability, the claim is not supported by the available literature.14,15 In these studies, which reviewed relevant case law, defendants prevailed not because of the physician’s AMA designation, but because the plaintiff was not able to prove negligence.
That’s a misrepresentation of the cited articles. Both articles (see here and here) contain statements to the effect that the AMA designation may indeed afford some legal protection.
So what should we do? Why not consider each case on its individual merits? If the patient wants to leave prematurely but the risk is low it may be reasonable to capitulate and enter a discharge order. In other situations where the patient's choice is clearly ill-advised and the risk is high a discharge order may be inappropriate and the departure should be documented as being against medical advice. Even in such cases try to work with the patient to help formulate a follow-up plan and, if appropriate, provide medication prescriptions. Assure the patient that you are not angry, respectfully ask that they reconsider and assure them that they are welcome to return. Clearly advise them about the danger of leaving but don’t threaten them or imply adverse insurance consequences.