Complications: A Surgeon's Notes on an Imperfect Science by Atul Gawande
My rating: 4 of 5 stars

"We look for medicine to be an orderly field of knowledge and procedure. But it is not. It is an imperfect science, an enterprise of constantly changing knowledge, uncertain information, fallible individuals, and at the same time lives on the line. There is science in what we do, yes, but also habit, intuition, and sometimes plain old guessing. The gap between what we know and what we aim for persists. And this gap complicates everything we do."

Atul Gawande's Complications (2002) strongly resonates with me in at least two ways. My favorite field of mathematics is probability, and one of the topics that excite me the most is how bad most people (including myself) are in dealing with uncertainty, i.e., with practical consequences of randomness. Dr. Gawande dedicates quite some space to this issue in the book, as promised in the snippet from the Introduction, quoted in the epigraph above. Furthermore, when I teach software engineering, I emphasize the topics related to quality assurance. Dr. Gawande's book makes it absolutely clear that the best way to improve medicine is to put a stronger focus on quality assurance in the medical process.

The book contains many fascinating stories of medical cases, often from the author's own practice as a surgery resident. We read about the physiology and psychology of blushing, gastric-bypass surgeries, persistent nausea, evolution of theories of pain, cases of "flesh-eating bacteria" infection, sudden infant death syndrome, and several others. As captivating as these cases are, being in the math and computer science fields, I am more interested in issues of uncertainty in medicine, medical errors, and the potential method of reducing the uncertainty and the number of errors.

The existence of medical error is normal, and I am using the word in two different meanings: normal as 'expected,' 'regular,' 'frequent,' but also normal as in the mathematical term of "normal distribution" that can be illustrated by a bell-shaped curve. Very informally, the so-called Central Limit Theorem, one of the magnificent achievements in the mathematical theory of probability, states that when the outcome of an experiment or event is affected by many, many independent random factors, then the value of the outcome follows the bell-shaped curve: most outcomes are in the fat middle of the graph (these are the average, expected outcomes), and very few in its tails (these are the unexpectedly bad or unexpectedly super successful outcomes).

Consider the author's example of a fairly routine operation of laparoscopic cholecystectomy ("lap chole"). The extreme complexity of human physiology, the complexity of activities in the operating room, the complexity of psychology of the doctors and nurses on the team, etc. result in thousands, probably millions of independent random factors that influence the outcome of the operation. So the outcome must be normal (as in bell-shaped). There will always be the tails of the distribution, and one of these tails may mean the patient's death from a routine surgery. The author writes:

"[...]studies show that even highly experienced surgeons inflict this terrible injury [cutting the main bile duct] about once in every two hundred lap choles. [...] a statistician would say that, no matter how hard I tried, I was almost certain to make this error at least once in the course of my career."

One of the most fascinating fragments of the book deals with the human inability to choose the right decision when randomness and catastrophic results need to be considered. Suppose (this is my example) a patient has a condition that severely imperils the quality of life. Suppose there exists an operation, with a recorded success rate of 99%. But in the remaining 1% of cases the patient will die during the operation. We do know the probabilities but how can we estimate the numerical value of the patient's life relative to the value of their life with the debilitating condition and relative to the value of the healthy life? If we could, the mathematical problem would be simple but, of course, we can't! The additional complication is the natural human inability to understand the difference between probabilities of, say, 0.001 and 0.00001 of something very bad happening. Both events are unlikely to happen, but don't forget that both will eventually happen. To someone, maybe even us.

Dr. Gawande quotes a lot of statistic in Complications. Here's probably the most scary of them:

"How often do autopsies turn up a major misdiagnosis in the cause of death? I would have guessed this happens rarely, in 1 or 2 percent of cases at most. According to three studies done in 1998 and 1999, however, the figure is about 40%."

So is medicine doomed to fail in a high percentage of cases? Or is there a chance for the medical success statistic to improve? The author's answer is positive and he repeatedly offers his suggestion of the best medication for the ailing medicine. In the last chapter he writes:

"[...] to shrink the amount of uncertainty in medicine -- with research, not on new drugs or operations (which already attracts massive amounts of funding) but on the small but critical everyday decisions that patients and doctors (which get shockingly little funding)."

Reduction of uncertainty is the crucial step. It could be achieved by following the quality assurance guidelines from other fields of science and technology. Dr. Gawande mentions various methods and processes that are used to improve aviation safety as recommendations that could easily be adapted for the medical field; I would add the engineering disciplines in general, including software engineering and systems engineering. Standardization, uniformization, "routinization" of medicine are strongly recommended.

A highly worthwhile book! I am sure that also the readers, who are not particularly interested in the issues of uncertainty and error in medicine, will find the medical case stories captivating and valuable reads. Well-written book, accessible, and convincing!

Four-and-a-half stars.


View all my reviews