#181 1991 · Intermountain Healthcare / LDS Hospital (Brent James, Alan Morris) · Healthcare / clinical quality
Doctors treating the same disease were quietly running different protocols, and the hospital fixed it without taking away their judgment
the problem
Expert practitioners doing the same job were producing wildly different, unmeasured variation, and neither rigid rules nor pure discretion had fixed it
background
In 1991, pulmonologist Alan Morris was running an NIH-funded trial at LDS Hospital in Salt Lake City testing a new artificial-lung treatment for acute respiratory distress syndrome (ARDS), and needed a stable standard-of-care control arm to compare it against. Instead he found that expert intensivists, all treating the same condition, were setting mechanical ventilators in substantially different, unstandardized ways from physician to physician — variation nobody had previously measured because no one had ever tracked it at that resolution.
The two obvious fixes both had real costs: mandating one rigid ventilator protocol for every patient would strip out clinically necessary judgment calls for genuinely atypical patients, while leaving the variation untouched meant patients' odds depended on which physician happened to be on shift rather than on the best available evidence.
what everyone would do
The two available fixes were mandating a single rigid protocol every physician had to follow, or leaving ventilator management to individual clinical judgment as before. Rigid protocols get resented and quietly routed around by experts who encounter genuine edge cases the rule didn't anticipate, while untouched discretion is exactly what produced the unmeasured variation Morris found in the first place — neither option had actually solved the problem anywhere it had been tried.
what they saw
James and Morris saw that standardization and clinical judgment weren't actually opposed — the real fix wasn't choosing between a mandatory rule and free discretion, it was making every deviation from a documented default visible and explained, which turned physicians' judgment calls into data instead of an untracked wildcard indistinguishable from arbitrary variation.
the move
Working with quality officer Brent James, Intermountain built a documented default ventilator-management protocol embedded directly into clinical workflow as the automatic path, but let any physician override it whenever a specific patient's condition called for it — with every deviation recorded and fed back through a continuous improvement loop, so the aggregate pattern of exceptions became data the protocol itself could be refined against, rather than either a mandate or a suggestion.
why it works
Embedding the current best-known protocol as the automatic default means most patients get the current best evidence-based care without any extra physician effort, while any physician who judges a specific patient doesn't fit that default can override it — with the override and its reason logged instead of happening silently and unrecorded as before. Aggregating those logged deviations over time reveals which overrides recur and which actually improve outcomes, feeding directly back into revising the default protocol itself, so genuine clinical expertise gets captured and compounded into the system rather than discarded, while the harmful kind of variation — outcomes depending on which physician happened to be on shift — gets squeezed out because it's no longer invisible.
the payoff
Under the standardized ARDS ventilator protocol, survival for patients meeting ECMO entry criteria rose from about 9.5% to about 44%, cost per case fell from roughly $160,000 to roughly $120,000, and physician time spent per case fell by about half — improved outcomes and lower cost arriving together rather than trading off against each other.
where it breaks
The mechanism needs enough case volume and tracking time to accumulate a statistically meaningful pattern in the logged deviations — a low-volume or short-lived process won't generate enough data to meaningfully refine the default. It also requires real organizational capacity actually mining that deviation data and updating the protocol, not just recording exceptions that pile up unused, and it depends entirely on physicians trusting that overriding the default is genuinely safe rather than a trigger for punitive review — if practitioners fear that logging a deviation invites blame, they either stop deviating even when clinically warranted or stop logging honestly, and either failure collapses the system back into rigid conformity or hidden, unmeasured variation.
what came after
Brent James generalized the approach into Intermountain's broader 'shared baselines' methodology, applied across more than 100 standardized clinical processes system-wide, and it became a widely cited model in health-policy and quality-improvement circles for how to standardize expert practice without eliminating the clinical judgment that standardization is usually assumed to cost.
references
- [1]In Conversation with...Brent C. James, MD, MStatAHRQ Patient Safety Network, 2009psnet.ahrq.gov
- [2]Brent James on Expanding Precision Medicine and Reducing Unwarranted VariationsSiemens Healthineers, 2018siemens-healthineers.com