The electronic medical record (EMR) was supposed to make clinical documentation more efficient. Yet, in some health systems, physicians are increasingly being told how to document—not just what needs to be documented.
Instead of writing a progress note in free text, where clinical reasoning can flow naturally, physicians are required to use structured problem-list entries, with a separate box under each diagnosis to document the plan. In some settings, physicians have even been explicitly instructed not to document their assessment and plan in free text.
The rationale is understandable. Structured documentation makes diagnoses easier to capture, map to ICD-10 codes, and process downstream.
But what works for data extraction doesn’t always work for clinical reasoning.
Consider what happens during a hospitalization:
The clinical picture evolves. Today’s problems may resolve, new ones emerge, and existing diagnoses may need to be revised. The problem list becomes a moving target.
Problems are interconnected. A single clinical plan may address several related problems. Forcing that plan into separate boxes fragments the reasoning.
Priorities change. The relative importance of diagnoses shifts throughout the hospitalization, requiring physicians to continually reorganize the problem list.
These aren’t merely inconveniences. They introduce friction into a process that should help physicians communicate their clinical reasoning clearly and efficiently.
Structured data absolutely has a role in the EMR. But should data extraction dictate how physicians think through and communicate a patient’s care?
If a physician can produce a thorough, organized, and compliant assessment and plan in free text, why should the interface force that reasoning into rigid, diagnosis-specific boxes simply to make downstream data processing easier?
The goal should be to structure the data without constraining the thinking.
Physicians should be able to practice at the top of their medical license—not spend their days rearranging diagnoses and redistributing plans across a series of disconnected text boxes.
If our documentation systems make data easier to extract but clinical reasoning harder to express, we need to question whether we’ve optimized the right thing.
Technology should adapt to the way clinicians reason—not force clinical reasoning to adapt to the limitations of the technology.