Skip to Main Content

Dr. Atul Gawande is best known as a surgeon and a writer, but it’s his experience running a Boston-based health research program that may be most relevant to his new role as chief executive of the health care venture spun out of Amazon, Berkshire Hathaway, and JPMorgan Chase.

Ariadne Labs, a partnership between a major Boston hospital and Harvard University, has since 2012 served as a testing ground for projects to improve the quality of health care around the world. The project’s track record, and how Gawande has run it, may be indicative of what’s in store for the company he’ll be heading up starting July 9.


A STAT examination of Ariadne’s work shows that Gawande and his team are willing to push ahead with bold ideas that have the potential to save patients’ lives but don’t always end up working as well as intended in the complex world of health care. It paints a picture of Gawande as a big thinker, ready to empower others and reluctant to micromanage.

“Our tools do work to improve the quality of care, but, in some cases, don’t achieve the result hoped for,” said Deborah O’Neil, director of communications for Ariadne. “We recognize there is more to do.”

Much of Ariadne’s work has focused on incorporating checklists — widely used in the airline and other industries to standardize safety practices and promote communication among teams — into medicine. Ariadne has spread the idea from surgical suites in Toronto to maternity clinics in India. Named for a goddess in Greek mythology who helped Theseus out of a labyrinth, Ariadne takes on a wide range of projects, some focused on the U.S. and some abroad.


At the Aspen Ideas Festival Spotlight Health program last weekend, Gawande said that “a fundamental principle” of his work with Ariadne has been to freely share its tools and insights.

“If you discover a new surgical procedure, we tell everybody about it,” Gawande said. “We teach people how to do it. We don’t tell them, OK, now you have to give me 10 percent of your future earnings as a surgeon because I taught you how to do this work. That is a basic, I think, principle of what we have to do in medicine as well.”

Ariadne Labs is a hybrid, with one foot in academia and the other in the messy, real world of an operating medical center. Legally, it is a part of Brigham and Women’s Hospital and the Harvard T.H. Chan School of Public Health, and its approximately 90 staffers are employees of one of those two organizations. Financially, it’s separate and self-sustaining, funded by grants, contracts, and philanthropic donations. Last fiscal year, it brought in about $16 million and spent about $18 million.

Its staff sit at multipronged cubicle islands in a 1929 art deco building that originally housed a Sears, Roebuck and Co. department store, Gawande equal among them.

And while the surgeon is its founder, former staffers and outside researchers who have collaborated with Ariadne emphasize that Gawande’s co-founder, Dr. Bill Berry, a distinguished doctor without his partner’s celebrity status, also plays a large role in the day-to-day operations. Gawande himself is eager to give credit where it’s due, according to those who’ve worked with him.

“If you talk to him, he’ll be very quick to say, ‘This is [another researcher’s] work.’ He gets a little bit uncomfortable when we say, ‘the work of Dr. Atul Gawande,’” said Dr. Robert Fine, the clinical director of the Office of Clinical Ethics and Palliative Care for Baylor Scott & White Health. The Dallas health system paid Ariadne to teach some of its doctors how to use a script for end-of-life conversations that Gawande’s group developed.

“He’s not a micromanager. He’ll decide something important, participate in the vision, empower … and then be accessible,” said Sara Goldhaber-Fiebert, a Stanford anesthesiologist who works with Ariadne on a checklist for emergency situations.

Many of Ariadne’s projects focus on checklists — chief among them, the “safe surgery” checklist, which has its roots in research that Gawande began before Ariadne’s official founding. He developed the checklist along with Berry, and in 2007 and 2008 they tested it in eight hospitals around the world.

The goals of the checklist are twofold, Berry said. “[It was] a conscious attempt to combine both process checks and prompts for team communication.”

At Aspen, Gawande said that checklists are a first step in moving health care from an individual to a team exercise.

“We are changing a culture from individual delivery of stuff — operations, devices, drugs — to team delivery of outcomes, and that’s a radically different place,” he said.

The results of the eight-hospital study, published in 2009 in the New England Journal of Medicine, appeared stunning. After the checklist was implemented, deaths during the hospital stay dropped by about 50 percent. Based on those findings, the checklist spread quickly around the world.

But not all researchers were convinced. Dr. David Urbach, a professor of surgery and health policy at the University of Toronto, who works at one of the eight hospitals in the 2009 study, said that the findings “were a little too good to be true.”

“It means it prevents one out of every two deaths after surgery, which is not just extremely effective when you think about it, it’s more effective than any intervention that had ever been considered,” Urbach said.

He took issue with the study’s design. It was a “before-and-after” study, meaning that it compared the rate of deaths for a period of time before the checklist was implemented to the deaths during a period of time after the checklist was implemented. But just because the decrease in deaths coincided with the adoption of the checklists doesn’t mean the checklists were the cause.

It’s possible that medical techniques simply improved over the time period that the study was conducted. Or perhaps the hospitals that were trained on the checklist simply paid more attention to how they were taking care of patients. The study acknowledged these limitations.

Having a control group — a set of comparable hospitals that, over the same period of time, did not receive the checklist — could have tested for these confounding factors. But the checklist study did not have such a comparison group.

“Any methodologist or epidemiologist worth their salt would say this isn’t a good study design,” Urbach said.

In 2010, Urbach’s home province of Ontario mandated that all hospitals adopt some sort of surgical safety checklist. So, Urbach conducted a study of whether the introduction of a checklist was associated with reduced deaths, and found that it was not.

Gawande said that one reason the Canada study came back negative is because there weren’t dedicated training programs for staff using the checklist.

“They just mandated that everybody use the checklist,” Gawande said at Aspen. “It became regulation. No reduction in death demonstrated in Ontario.”

Berry also acknowledged that the 2009 study wasn’t perfect.

“We could have potentially used [a] randomization strategy, although it would have been much harder to do,” Berry said. “The funding that we had to do that first study was relatively limited, and we wanted to create a study that we could do relatively expeditiously.”

Berry also said the study’s funder, the World Health Organization, wanted the team to test the tool all over the world, which would have made it harder to find a suitable control group.

“The use of the checklist was associated with lower complications and lower mortality, right,” Berry said. “But we can’t say it proved it.”

That association — combined with evidence that the individual items on the checklist were worth doing anyway — was enough to convince Berry that the tool was ready for wider implementation. Berry and Gawande, along with other members of their team, partnered with the South Carolina Hospital Association to bring the intervention there.

Lorri Gibbons, who at the time was the vice president for quality and safety at the South Carolina Hospital Association and now serves as the vice president of development, said that Gawande himself was personally involved, but that Berry spent more time physically in South Carolina, along with other Ariadne staffers.

The South Carolina program was a health intervention, not a research project, said Berry, but the team decided to publish research papers.

A 2017 paper evaluating the program in South Carolina had good things to say — or maybe not, depending on who you ask.

The paper found that the hospitals that fully implemented a checklist did see a reduction in deaths, which was the paper’s primary outcome measure.

But overall, hospitals across South Carolina saw no change. Urbach, who said he was a reviewer for the paper, said that in his view, the final result was negative because the intervention did not reduce overall deaths in the state. Berry attributed this difference to the “vagaries of statistical analysis,” saying that there weren’t enough hospitals that adopted the program to pull down the average.

This isn’t the only example of an Ariadne program having mixed results.

From 2014 to 2016, Ariadne helped deploy a WHO checklist, which Gawande helped develop, in 60 Indian government health facilities, hypothesizing that it would reduce deaths among newborns and their mothers. It did not.

But the team published the results in a prestigious journal anyway, demonstrating their commitment to sound science even if it didn’t line up with their expectations.

“The fact that it was published in the New England Journal [of Medicine], I think, speaks volumes about the motivations for doing that kind of work,” said Dr. Thomas Tsai, who spent three years at Ariadne and considers Gawande a mentor. Tsai recently started a surgery fellowship at Massachusetts General Hospital.

The facilities trained with the checklist did end up completing more items on the list, though, which was seen as an improvement in the quality of care. In a press release following the publication of the study, Gawande said that “we in public health must identify the additional ingredients required to produce the complete recipe for saving lives at childbirth.”

Added Tsai, “You need to learn from both the successes and the failures … [and] I think it’s that level of nuance and transparency, which, I think, makes him unique.”

Erin Mershon contributed reporting.

Comments are closed.