Why good healthcare software goes unused

The most expensive outcome in healthcare technology is not a system that fails. It is a system that works, passes testing, goes live, and is then quietly worked around. The cost is fully incurred and the benefit is not.

The arithmetic of a click

A workflow that adds four clicks to a task performed sixty times a shift adds two hundred and forty interactions to a day that was already full. That is the entire explanation for a large share of adoption failures, and it is usually known to the people doing the work well before go-live.

What makes this hard to catch is that each individual addition is defensible. A required field improves data quality. A confirmation step prevents an error. A second screen supports a legitimate secondary use. None of them is unreasonable in isolation, and their sum is a system nobody wants to use.

Every requirement was justified. The total was not survivable. Both statements are true, and only the second one determines adoption.

Where the signal is available earlier

• Observation of the task as performed, including the shortcuts people already use
• Counting interactions per task before and after, and treating an increase as a design defect rather than an acceptable cost
• Testing with the busiest users rather than the most enthusiastic ones
• Watching for parallel processes — a workaround maintained alongside the system is the clearest possible signal
• Adoption measured at month three, not week one, since launch numbers are inflated by mandate

The fourth is the most reliable indicator and the easiest to miss, because workarounds are rarely reported. Staff do not experience them as system failures. They experience them as how the job gets done.

Designing against the constraint

Treating clinician attention as the scarcest resource in the design changes decisions in a specific direction: fewer required fields with better defaults, information surfaced where the work already happens rather than in a separate application, and secondary data needs met by derivation rather than by asking a person.

That last principle carries most of the weight. A substantial amount of what healthcare systems ask users to enter is recoverable from data already captured elsewhere. Every field eliminated that way is a permanent reduction in the cost of using the system — and unlike training, it does not need to be repeated with every new member of staff.