One line per American hospital, drawn from how often the sepsis bundle gets started to how often it gets finished. Every line falls. The whole band falls 16.8 points, and it does not matter who owns the hospital or how big it is.
A travel nurse at Adventist Health in Bakersfield spent late September 2024 taking patients' IV opioids and charting them as given. The hospital ran machine-learning software built to catch exactly that pattern. It fired. Managers ignored the alerts, auditors found.
Nobody publishes how often diversion alerts get ignored — hospitals aren't required to say they run the software, let alone report when it fails. But CMS does publish, for 3,084 hospitals, what happened after a different clinical trigger fired. Here is all of it, at once.
Each hairline is one hospital. The navy end is how often the first move landed — the 3-hour severe sepsis bundle: lactate, cultures before antibiotics, antibiotics, fluids. The red end is how often the whole thing got finished and closed out. Watch it fall.
Sepsis detection is a solved problem. The trigger logic is decades old, it is baked into every major EHR, and by CMS's own numbers it works about four times in five. Diversion detection is a solved problem too — at Bakersfield the software fired for weeks.
What both stories have in common is that everything downstream of the firing was unowned and uninstrumented. Somebody has forty alerts in a queue, thirty-eight are noise, and the queue becomes a thing you clear rather than a thing you read.
If your product can be fully ignored and never tell you, you don't have telemetry. You have a dashboard.Builder's Briefing, August 26, 2026
A third of hospitals are missing. The CMS file holds 4,660 hospitals. Only 3,106 have a SEP_1 score at all; 1,554 are blank — footnoted “too few cases,” “not available,” or “suppressed.” The hospitals most likely to have an escalation problem are the ones least likely to produce a publishable number.
The extremes are arithmetic, not care. Toggle “hide small hospitals” above. Below 30 reported sepsis cases the standard deviation of completion is 22.8 points and 3.3% of hospitals score exactly 0% or exactly 100%. At 300 cases or more the SD is 14.0 and no hospital scores 0 or 100. Same care, quieter denominator.
The drop survives every cut. That is the point of the toggle. The spread collapses; the gap does not. Nonprofit 16.6. For-profit 16.6. Government 17.6. Under 30 cases 19.0. Over 300 cases 15.4.
This is chart abstraction, not behaviour. SEP_1 asks whether an element was documented inside a time window. It does not know whether a clinician read an alert, dismissed it, or never saw it. There is no CMS measure — not one of the 24 in this file — for time-to-acknowledge, close-without-comment rate, or override volume. Every number here describes the outcome of an escalation path. None of them describe the path.
| Denominator | Hospitals | Started | Finished | Drop | SD |
|---|---|---|---|---|---|
| Under 30 cases | 275 | 77.7 | 58.6 | 19.0 | 22.8 |
| 30–99 | 867 | 79.8 | 63.3 | 16.5 | 16.5 |
| 100–299 | 1,613 | 79.7 | 62.9 | 16.8 | 14.8 |
| 300 or more | 329 | 81.1 | 65.6 | 15.4 | 14.0 |
If you are building clinical AI: the detector is the easy half. Who receives the alert, what their shift looks like when it lands, what the false-positive rate does to their trust by week six, whether closing it requires a reason, whether anyone audits the closures. None of that is in your model card. All of it decides whether your model does anything at all.