←  All research The New England Journal of Medicine · 2009

A 19-item checklist halved surgical deaths

Surgeons with fifteen years of training halved their patients’ death rate because somebody handed them a list. This is the strongest evidence anywhere that structure outperforms expertise — and it is about a piece of paper.

Design: prospective before-and-after study, October 2007 to September 2008. Sites: 8 hospitals in Seattle, Toronto, London, Auckland, Amman, New Delhi, Manila and Ifakara. Sample: 7,688 patients — 3,733 before the checklist, 3,955 after.
1.5%→0.8%
Inpatient death rate before and after the checklist, across 8 hospitals
1.5%0.8%BEFOREAFTERINPATIENT DEATH 11%7%BEFOREAFTERCOMPLICATIONS
Dashed outline marks the pre-checklist level. The same surgeons, the same hospitals, the same operations — the only change was a 19-item list read aloud at three points during the procedure.

What they did

The World Health Organization put a 19-item safety checklist into eight hospitals chosen to span the range of global healthcare: high-income teaching hospitals in Seattle, Toronto, London and Auckland, and hospitals in Amman, New Delhi, Manila and Ifakara, Tanzania.

The list is aggressively unremarkable. Confirm the patient’s identity and the surgical site. Check the pulse oximeter. Ask about known allergies. Introduce the team by name and role. Confirm antibiotics were given. Count the instruments and sponges at the end.

Nothing on it is knowledge a surgical team lacks. Every item is something they already know.

8 HOSPITALS · 4 CONTINENTS · ONE PROTOCOLSEATTLETORONTOLONDONAUCKLANDAMMANNEW DELHIMANILAIFAKARA
Rich and poor systems alike: the same 19 items produced the same direction of result at every site.

What happened

Across 7,688 patients, the inpatient death rate fell from 1.5% to 0.8%. Major complications fell from 11.0% to 7.0%. The improvement appeared at every one of the eight sites, in rich hospitals and poor ones alike.

BEFORE → AFTER THE CHECKLISTDEATHS1.5%0.8%COMPLICATIONS11.0%7.0%
Deaths fell from 1.5% to 0.8%; major complications from 11.0% to 7.0% — in the same operating rooms, with the same surgeons.
7 688PATIENTS19CHECKLIST ITEMS−47%DEATHS−36%COMPLICATIONS
The cost of the intervention: a laminated sheet of paper and two minutes of talking.

Why it worked

The checklist did not add competence. It removed the assumption that competence would surface on its own, under pressure, every single time.

That is the entire finding, and it generalises further than surgery. Expertise does not fail because the knowledge is missing. It fails because retrieval is unreliable when you are tired, rushed, or certain you already did the thing. A list is not a substitute for skill. It is a substitute for remembering perfectly.

If a checklist can do this in an operating theatre — the highest-stakes, highest-training environment there is — the argument that you personally are too experienced or too smart to need one does not survive contact with the data.

WHY A LIST BEATS A MEMORY✕✕✕Under load, experts skip steps they know by heart. The list exists for the three you would have skipped today.
Expertise was never the missing piece — coverage was. The checklist closes the holes attention leaves.
What to build from this

Turning the finding into a mechanic

  • Write the routine down once. Do not re-derive it every morning from memory.
  • Keep the list short enough to actually run. Nineteen items covered an entire operation.
  • Put the check at a fixed moment, not ‘whenever I remember’. The WHO list runs at three specific points.
  • The list is for your worst day, not your best one. That is when it earns its place.
Find your missing mechanism See the systems
IN PRACTICE
LEVEL UP SYSTEM treats your day the same way: a pre-set quest list you run, not a memory you trust.
See LEVEL UP SYSTEM →
Haynes AB, Weiser TG, Berry WR, et al. (2009). A Surgical Safety Checklist to Reduce Morbidity and Mortality in a Global Population. New England Journal of Medicine 360(5), 491–499. DOI 10.1056/NEJMsa0810119.
Read the original paper →