The First Try

Core idea

The World Health Organization asks Gawande to lead a global effort to reduce avoidable harm from surgery — 230 million operations a year, an estimated 7 million people left disabled, at least 1 million dead. He convenes a Geneva working group, settles on the checklist as the only intervention simple, measurable, and transmissible enough to work everywhere, and assembles a first draft. He brings it back to his own operating room — and within hours it is dead. The list is too long, too ambiguous, and the nurse silently fills it in on paper instead of running it as a verbal team protocol. The topic is the story of what a checklist is not, and the discovery that the structural features (pause-points, verbal team execution, name-and-role introductions, the "activation phenomenon") are at least as important as the content.

Why it matters

Surgery as a global public-health problem

The numbers Gawande's team compile reframe surgery from a specialty into a public-health crisis: ~230 million major operations a year — more than global childbirth, with a death rate ten to one hundred times higher. Complication rates run 3–17 percent. At least half of complications, U.S. studies suggest, are preventable. Two billion people lack access to a surgeon, and the surgery that does exist is often performed by under-trained staff using missing or broken equipment.

Why guidelines and pay-for-performance won't reach

The conventional moves all fail this scale test. Training is too expensive to deploy widely. Pay-for-performance pilots show 2–4 percent improvements and rely on self-reported data adjusted for case mix. The WHO basement is stacked, waist-high, with shrink-wrapped expert guidelines that have changed almost nothing at the bedside.

The criteria the intervention must meet: simple, measurable, transmissible — what business calls a large ROI and what Archimedes would have called leverage.

Soap as the precedent

Stephen Luby's Karachi study is Gawande's clinching analogy. Procter & Gamble's Safeguard soap, distributed with instructions to wash hands in six specific situations, dropped child diarrhea by 52 percent, pneumonia by 48 percent, impetigo by 35 percent. The soap was already in most households — what changed was the systematic use. The intervention was, in effect, a hand-washing checklist with the soap as its delivery vehicle.

What the prior checklists had figured out

Three independent teams — Columbus Children's Hospital, Toronto's Reznick, Johns Hopkins's Makary, plus Kaiser in Southern California — had already built surgical checklists. All of them mixed the same two ingredients: task checks for the simple killers (infection, bleeding, anaesthesia), and a team briefing for the fourth and irreducible killer, "the unexpected." All of them included a name-and-role introduction at the start — a step Gawande initially dismissed as hokey.

The activation phenomenon

Brian Sexton's psychology research at Johns Hopkins showed two findings the introductions were tracking. First, operating-room staff who did not know each other's names — about half the time — rated communication during the case much lower than those who did. Second, when nurses were given a chance to say their names and concerns at the start of a case, they were significantly more likely to speak up later about problems and offer solutions. Sexton called it the activation phenomenon: a brief structured opportunity to speak early turns silent specialists into a team that will speak up when it matters. The introductions were not social grease. They were a permission slip.

The first try fails

Gawande prints the consolidated checklist, takes it into his operating room, hands it to the circulating nurse, and within fifteen minutes the experiment is over. The nurse, sensibly, has ticked the boxes silently on paper because the list does not say it is verbal. The questions are ambiguous — "confirm allergies" could mean "verify they're known" or "say them aloud." The list is too long; everyone is exasperated; the patient on the table asks if anything is wrong. By the end of the day, they have abandoned it. Forget making it work around the world; it isn't even working in one operating room. The failure is a gift — it is what teaches Gawande, in the next topic, where to go for help.

Key takeaways

Mental model

Copy sourceZoom

Practical application

1. Match the intervention to the scale

If the goal is reach, the criteria are non-negotiable: simple, measurable, transmissible. A 200-page guideline does not clear those bars. A laminated card with three pause-points does. Choose accordingly.

2. Build pause-points into the workflow

Aviation's contribution to the consolidated checklist was the pause-point — a moment in the workflow at which the team must stop and run a defined set of checks before proceeding. The WHO draft has three: before anaesthesia, before incision, before the patient leaves the room. Pause-points are how a checklist anchors into a process that does not naturally stop.

3. Use the introduction as an activation device

Asking each team member to state their name and role at the start is not a social courtesy. It is a measured intervention: it raises communication ratings and increases the chance that the nurse will speak up later about a problem. Treat the introduction as a clinical step.

4. Expect the first draft to fail

Gawande's first attempt died in an afternoon. The list was too long, the verbal/written distinction was unspecified, the questions were ambiguous. The failure was the data point that made the next iteration possible. Plan the first run as a learning instrument, not a deployment.

Example

A site-reliability team adopts a new pre-deployment checklist drafted by a senior engineer. The first day, the on-call goes through it on her own and ticks the boxes in a Google Doc; the deploy proceeds; an incident follows. In review, two items on the list were ambiguous — "verify cache is warm" could mean "look at the dashboard" or "page through a test request" — and the on-call interpreted them the easier way. The next iteration adds three changes Gawande's experience predicts: (1) the checklist runs at a defined pause-point — a five-minute call before the release window — and is spoken aloud; (2) each item is rewritten as an unambiguous yes/no with a concrete verification; (3) the most junior engineer on the call reads each item and the senior engineer responds, restoring the "activation phenomenon" effect Sexton measured in the OR. None of this required new content. It required the surrounding ritual.

Related lessons