What 5.9 Hours of EHR Actually Looks Like

Arndt and colleagues (Ann Fam Med, 2017) pulled the EHR event logs of 142 primary care physicians over three years. The headline number — 5.9 hours of EHR work in an 11.4-hour workday, including 1.4 hours of "pajama time" — gets cited a lot. What gets cited less is the breakdown.

Of those 5.9 hours:

• 1.4 hours on inbox management (results to acknowledge, messages to respond to)
• 1.3 hours on clinical review (chart review, looking up past visits)
• 1.0 hours on documentation (note writing)
• 0.7 hours on order entry
• 0.6 hours on chart review
• 0.5 hours on logging in, switching contexts, navigating menus

The line item nobody defends

The number that should make every clinician uncomfortable is the 0.5 hours on logging in and navigating. Half an hour per day, every workday, spent on the navigational overhead of the EHR. Not on documentation. Not on care. Just on moving between screens.

Everything else on that list is at least arguably work. Documentation is a legal and clinical requirement. Inbox management is results being acknowledged, which is care. Order entry is the doctor doing the thing the doctor is for. You can argue the software makes each of them slower than it needs to be, but you cannot argue the underlying activity is worthless.

Navigation has no such defence. It is thirty minutes a day of a trained physician's attention going into the difference between where the software is and where they need it to be. Over a working year that is roughly two and a half weeks. Nobody would sign off on a rota line that read "physician: two weeks per year, menus."

Why micro-overhead is invisible

The reason this survives is that no part of it ever appears as a delay anyone can point at.

There is no moment in the day labelled "navigation." There is a moment where you need yesterday's creatinine and it takes forty seconds instead of four, and you do not experience that as a cost — you experience it as looking something up. The same forty seconds recurs eighty times. At no point does anyone feel a forty-second wait.

What it surfaces as, instead, is the day ending later. Which brings us to the most diagnostic number in the paper.

Pajama time is the tell

1.4 hours at home. That is not extra work the physician chose to take on. It is the arithmetic remainder: the work did not fit inside the day, so it landed on the evening — unscheduled, unpaid, and invisible to every staffing model in the building.

Any efficiency loss inside the working day is silently absorbed by this overflow. That is why the micro-overhead is so stable and so hard to argue about: it never breaks anything visible. It just moves an hour of a doctor's life from their kitchen table's "own" column to its "work" column, every night, and the system registers no change at all.

The Indian version of the same tax

I cannot find a comparable Indian study with similar rigor. But the residents I've spoken with at AIIMS, RG Kar, and JIPMER describe the same fragmentation — except instead of EHR navigation, the time is spent flipping between paper charts, lab reports, ECG strips photocopied at the back of the file, and WhatsApp messages from the lab.

Count the surfaces a resident touches to answer one question — is Bed 16's potassium safe to give the next dose of spironolactone?

The paper chart for the current orders. The lab report file for the last potassium, which may or may not have been filed yet. The unit WhatsApp group, where the lab manager posted this morning's results as a photograph. Their own folded list, where they wrote it down in pencil during the round. The nurse's medication register for what was actually given, as opposed to what was prescribed. The ECG at the back of the folder, loose.

Six surfaces. None of them knows the other five exist. The physician's half-hour of menu navigation and the resident's ten minutes of paper archaeology are the same tax collected by different means, and the resident's version has the additional property that any of the six can simply be missing.

What "one surface" has to mean

The point of these numbers is not that the EHR is bad. The point is that the cumulative micro-overhead — half a minute here, two minutes there — adds up to hours, and the work that gets squeezed out is the work that has no built-in timekeeper: thinking about the patient.

What I keep coming back to as a designer is this. The tool that holds the ward list, the tool that captures vitals, the tool that records the handover — these should all be the same tool, in the same surface, with the same login. Not because integration is fashionable, but because every "open a different tool" is a working-memory tax.

Being precise about the claim matters, because "one system for everything" is how hospital software becomes unusable. The unit that has to be seamless is not the hospital. It is the loop the resident actually walks: list → patient → vitals → order → handover, and back to the list. Every context switch inside that loop is a tax. Every context switch outside it — to radiology, to theatre scheduling, to billing, to the pharmacy system — is a normal boundary between different jobs, and trying to absorb those is how a ward tool turns into a decade-long integration project that ships nothing.

So: not one system. One loop, closed properly, and honest interfaces at its edges.

Medows is built as a single surface for the doctor on rounds, not because we don't believe in specialised tools elsewhere in the hospital, but because at 2 a.m. with 28 patients, the navigation tax is the difference between catching the K+ and not.