Sunday, February 27, 2011

The Checklist Manifesto, by Atul Gawande

Atul Gawande has been writing on medical practice, provocatively and informatively, for the New Yorker for several years. His comments on price differences and what drives disparities between different areas garnered a lot of attention, but that area hasn't been his main focus. His recent book, The Checklist Manifesto, is closer to the main line of his writings. In this short book, Gawande presents his argument that medicine as currently practiced is far from a rigorous, science-driven field. He shows how aviation and construction are at least as complex and time-constrained as medicine, and that both have benefited from the use of checklists to help practitioners get the details right when performing complex operations.

Individual errors and mistakes of coordination are far more common in medicine than in other modern highly-technical fields. If medicine followed the standards of professional practice common in other areas, there would be a dramatic improvement in our overall health. Gawande discusses how checklists are constructed and used in aviation, another field where routine work is occasionally suddenly interrupted by situation requiring split second decision making at a rapid pace in a distracting environment with enormous consequences. Safety experts in aviation have learned how to put together checklists that can be found quickly, and that enable professionals to correctly address situations that arise in one flight in a million.

The safety record in aviation world-wide is amazingly good. I've long ascribed that difference between aviation and medicine to the fact that accidents in aviation are scrutinized thoroughly, and every mistake drives new corrective processes that quickly make it less likely that the same thing will happen again anywhere in the world. Every airframe manufacturer ensures that all of its vehicles are quickly updated with the most up-to-date procedures. In medicine, individual hospitals sometimes conduct reviews, but any knowledge gained is used sporadically and locally. There is also no standard for how to conduct these inquiries, so some investigations are derailed by politics, infighting, or a desire to deflect blame (which is exacerbated by medical malpractice risks.) The inquiries conducted in aviation have been designed to find correctable causes, and not to place blame. Focusing on the checklists that result from these inquiries would be a big improvement on what we have in medicine now. It's even plausible that a coordinated process for producing checklists would drive an improvement in the checklists based on measured effectiveness.

The book is very readable. The main story is about how Gawande led a task force for WHO testing out some simple checklists for a few common surgical procedures with high rates of routine errors. The results were spectacular, leading to a 36 percent reduction in complications and 47 percent drop in deaths from a variety of hospitals in rich and poor communities all over the world. Getting doctors and hospitals to actually adopt this simple improvement is a far harder task than getting a pilot program demonstrating its effectiveness. The side trips Gawande makes into aviation and construction to show how checklists work there and how they're constructed are engaging.

Monday, February 14, 2011

Going Inside, John McCrone

John McCrone's Going Inside provides a lot of insightful observations about how the brain works, but fails to tie them together into a cohesive picture. McCrone focuses on recent findings from new brain scanning technologies, and is particularly fascinated by timing studies that give details on how long it takes us to process incoming information, the specific times at which decisions are made, and how our subjective experience of when choices happen comport with the underlying brain circuitry. In particular, the studies show that it takes a half second to react to new information, even when we're expecting it, but our subjective experience is that the decision is made instantaneously at the end of that period. The experiments that show this are ingenious: by cutting off or distracting the process at various points, we can compare the reported subjective feel about the decision state that was reached with the brain scanner's details of how far into the process the brain actually got. All this works experimentally, because repeated sessions show that there's a lot of consistency in the information processing, so the scientists can pinpoint when the incoming information started being processed, and how long it would have taken to reach a choice.

The problem with the presentation is that McCrone doesn't provide an overview of the whole picture until the closing chapter, so as a reader, I had no framework onto which to attach all the facts as he presented them so as to build up a cohesive picture. I was left with the feeling that he'd presented good evidence that seemed to bear on the issue he was investigating, but I didn't know how it fit as I encountered it, so each tidbit vanished as I encountered the next one. I'm not sure things would be much improved on a re-reading. With a familiarity with the whole story, I could figure out how most of the pieces buttress the argument, but I'd have to make up my own argument structure for why his is the best explanation for the workings of the entire system.

McCrone also flubs up on the evolutionary explanation. At various points, he attempts to show why evolution would have produced just the structures and relationships that he has revealed, but his descriptions are unconvincing--he sometimes speaks as if evolutionary pressures are pushing toward a known result, rather than explaining why some abilities would have been selectively favored and why random mutations could have produced the effects. I think the correctly formed arguments could have been constructed, but McCrone's failed attempts were distracting.

One of McCrone's goals is to show how quickly brain scanning technologies advanced over the last few decades. The best tools for peering into the deepest details of timing and interaction in the brain have only recently been developed, so the insights that are most crucial to the book's argument aren't presented until a third of the way into the book, when he has set the historical context. Benjamin Libet did a series of experiments on patients (who were getting brain surgery already) that showed that direct electrical stimulation of the brain wasn't noticed unless it continued for a full half-second. If the stimulation was cut off earlier the patient wouldn't notice anything; if it continued for longer, the subject would report that they had been aware of it from the beginning. Later experiments showed that a second stimulus could mask the first one, as much as a third of a second later. This is pretty convincing evidence that processing inputs takes us up to half a second, and our experience of the present is cobbled together after the fact. Explaining "Libet's half-second" and figuring out what it implies about consciousness occupies the bulk of the book.

Libet did other studies later in which subjects were asked to notice the position of a rotating second hand at the moment they made a decision to lift a hand or take a similar action. With these and other similar experiments by other researchers it became clear that there isn't a precise moment at which decisions are made. The state of the brain changes somewhat continuously over a period from a half second to a full second, and subjects report somewhat arbitrary times as "the moment" of decision.

I would have to re-read the whole book (I've re-skimmed about the first half) in order to provide a detailed synopsis of it. I did feel like McCrone brought quite a few fascinating and important insights to light that would clarify an understanding of brain mechanisms, but the organization makes it hard to put them together on first reading. Maybe someone else will (or has) pull the material together in a better order, and that will be a more worthwhile book.