As disclosure requirements have solidified, the default institutional response has been to add language to existing consent documents. A paper published this month argues that this is the least effective possible implementation, and possibly a counterproductive one.

The authors' argument is straightforward. Admission consent is signed under time pressure, in volume, by patients who are frequently unwell, and is understood by essentially everyone involved as a formality. Placing meaningful information there is a reliable way to ensure it is not read.

Their proposed alternative is point-of-decision disclosure: a short, specific statement at the moment a model-informed recommendation is delivered, by the clinician delivering it. This is harder to operationalize and considerably harder to audit, which the authors concede is why institutions do not do it.

The paper has drawn measured pushback from health system counsel, who point out that point-of-decision disclosure creates documentation obligations that current EHR workflows cannot satisfy. The authors' response — that a requirement being inconvenient is not an argument that it is wrong — has not fully settled the disagreement.