I have practiced medicine for more than three decades, thirty years of obstetrics and gynecology, and now addiction medicine. For seventeen of those years, I have worked inside electronic medical records that waste my time, interrupt my thinking, and stand between me and the person in front of me.
Like a lot of clinicians, I kept notebooks, running lists of what a truly helpful record would do, scribbled after long evenings of clicking. When I finally compared my notes against the published literature, I found something both consoling and damning. A national initiative, a database of half a million clinician surveys, and colleagues on every platform have been writing the same list, in remarkably similar words, for years. The problem is not that no one knows what to build. The problem is that the people who know are not the people who buy.
That sentence is the whole disease in miniature. Electronic health records are not designed by physicians, or with meaningful physician input. They are built by software developers with little firsthand experience of everyday clinical work, then marketed and sold to the administrators who run health care organizations. The buyer is not the user. Administrators purchase for billing efficiency and regulatory compliance (legitimate concerns), but those are valued far above the needs of the providers who touch the software hundreds of times a day, and above the needs of the patients the whole enterprise exists to serve. Members of the C-suite never chart in these systems themselves. So the damage stays invisible from where they sit, and the productivity quietly destroyed every day never shows up on their dashboard.
The market compounds it. The pile of federal requirements for certified records, well-meaning but overbuilt, makes entering the market expensive, so it consists of a few large vendors with little reason to court the end user (who is not buying) and little reason to innovate (the package already sells). The result is software that, on user-friendliness, lags a decade or more behind ordinary commercial tools, behind Word, behind Excel, and further still behind the website of any large internet company. Software was invented to eliminate repetitive data entry. The electronic record, uniquely among every category of software I know, multiplies it. It is the biggest time thief in medicine, and, renewed every year through license fees, medicine’s most expensive software failure.
Real physician voices, twice a week
Free, and one click to unsubscribe.
There is really only one way to make a chief executive understand what this feels like. So here is a thought experiment.
Take away the executive assistant. Replace the sleek laptop with a machine running an antique operating system and a version of Word stripped of every modern convenience: no autosave, no autocomplete, no memory of anything typed before, no ability to pull in information that already exists elsewhere in the system. Require a full login every time the program opens. Then have the system log the executive out at random intervals, without warning, mid-sentence, displaying only the words “YOU have logged out,” with everything unsaved erased. When the executive calls IT, have IT explain that this is for security reasons. Make reformatting a paragraph take five or six checkboxes. Route every print job through a network that crashes at random, to a printer three rooms away. Let the executive work like that for one quarter, while being told, weekly, to increase output.
That is the everyday reality of the record for the people documenting in it. Then, and only then, let us talk about volume targets.
And this is not only my complaint. The data agree with the notebooks. The KLAS Arch Collaborative, the largest standing dataset of clinician record experience, with more than 500,000 standardized surveys since 2017, reports that 59 percent of over 25,000 surveyed physicians say their record does not enable efficiency. Among clinicians reporting burnout, the record is named as a contributor by 62 percent of physicians and 43 percent of nurses. In the Collaborative’s 2026 report, spanning 92 organizations and more than 121,000 clinicians, only 22 percent of physicians, and a dismal 12 percent of nurses, rate their experience at the top level.
The AMIA 25×5 initiative, funded by the National Library of Medicine, later echoed by the Surgeon General’s advisory on health worker burnout, was motivated partly by a finding that deserves to be far better known: U.S. clinicians spend 75 percent more time on documentation than clinicians in other wealthy countries, in some cases using software from the very same vendor. The burden is not an inevitable property of digitization. It is a property of American implementation choices, which means it is a property we can change.
And when Black Book Research asked 274 hospital-affiliated physicians in 2025 to name the single most urgent improvement they wanted, the winner, by a wide margin, was AI-powered documentation. The sufferers are unanimous about where the pain is. The only people who seem unable to see it are the ones holding the purchase order.
Matthias Muenzer is an obstetrician-gynecologist.
