LENS Learning Engineering for Next-Generation Systems
Case Studies
Part 1 · Healthcare & Patient Safety · What worked

Keystone ICU / Pronovost Checklist

2004 – present healthcare

Impact. Central-line-associated bloodstream infections (CLABSI) reduced to near zero across 103 Michigan ICUs; ~1,500 lives saved in 18 months; ~$175M saved; sustained at ten years

Ask AI about this case Opens a new chat prefilled with this case — explore it in your own context.

Peter Pronovost’s central-line checklist has five items, and not one of them was news to the physicians it governed. The question was never what to do — it was whether it got done every time, and whether anyone junior could say so when it didn’t. Keystone paired the checklist with something harder to engineer than a list: a nurse’s standing authority to halt the procedure when a step was skipped. Across 103 Michigan ICUs, bloodstream infections fell to near zero and stayed there for a decade. The capability that mattered wasn’t knowledge — it was permission, built into the system.

In brief

Peter Pronovost's central-line checklist has five items — wash hands, use full barrier precautions, clean the skin with chlorhexidine, avoid the femoral insertion site, and remove unnecessary catheters — and not one of them was unknown to the physicians it governed. The question was never what to do, but whether it would be done every time. The Keystone project, launched across 103 Michigan ICUs in 2004, paired the checklist with a cultural change: nurses were not merely permitted but required to stop the procedure if a step was skipped. That authorization is the element the case treats as load-bearing — though it rode inside a multi-component bundle (clinician education, a dedicated line cart, daily catheter-review goals, and monthly infection-rate feedback) that the trial tested as a whole and could not decompose. Central-line-associated bloodstream infections fell to near zero, an estimated 1,500 lives and $175 million were saved in eighteen months, and the effect was sustained at ten years.

The case in five beats

  1. Central-line infections persisted because nurses lacked procedural path to intervene across the authority gradient
  2. Pronovost paired a five-item sterile checklist with a required nurse stop authority
  3. Requirement, not permission, plus an impersonal trigger made the stop usable against senior physicians
  4. CLABSI rates fell near zero across 103 ICUs; effect sustained at ten years
  5. Packaged as the AHRQ CUSP toolkit, replicated in forty states and internationally as paired design
The Learning Engineering Lens

LE insight

Keystone is the clearest evidence in healthcare that a technical intervention without authority intervention produces no durable change, and authority intervention without a technical artifact produces no measurable change. Both are necessary. The empirical record of Keystone is the strongest available argument for designing interventions as pairs — and treating the cultural half as engineering, not aspiration.

LENS approach

LENS uses Keystone in LEN 4 and LEN 10 as the canonical worked example of measurement linked to intervention. Studio projects require students to specify both halves of the pair, name the authority that authorizes the cultural half, and identify the measurement signal that will confirm or refute the effect. The course treats "is the cultural change actually authorized?" as falsifiable, not stated.

The LENS competency this case exercises
  1. 1 Systems Analysis
  2. 2 Iterative Development
  3. 3 Human-System Collaboration
  4. 4 Test & Evaluation
  5. 5 Sociotechnical Constraints

Problem type · PT3

Problem type
D3/PT3
Induced
4.1
CLO
3