5 ideas
Gawande, a surgical resident who defends learning on patients, describes his infant son Walker needing cardiac care: when a fellow in training was assigned, Gawande asked for the attending instead. He admits that his standing as an insider let him opt out of the very training system he argues is necessary.
Surgical skill has a learning curve, and novices climb it by operating on real patients, whose risk is higher during those early attempts. Gawande argues that if every patient could decline trainees, residents would not get the practice they need, so the system tends not to disclose fully when a trainee is involved.
Gawande shows that serious errors are also made by competent, conscientious doctors as a consequence of human fallibility. Anesthesiology's systems approach, including Jeffrey Cooper's critical-incident studies and standards such as pulse oximetry, cut anesthesia deaths sharply. Gawande supports such system redesign but also defends the morbidity-and-mortality conference, where surgeons privately take personal responsibility for their errors.
Shouldice Hospital near Toronto performs only hernia repairs, and its surgeons each do hundreds a year using a single standardized technique. Its recurrence rates are around 1 percent versus roughly 10–15 percent at general hospitals, evidence that narrow repetition and routinization outperform broad individual talent.
In a Lund, Sweden study, a neural-network program read over 2,000 ECGs and identified heart attacks about 20 percent more accurately than a leading cardiologist. For narrow, repeatable judgments, a machine that applies the same criteria every time beats even skilled humans, whose performance varies with fatigue and context. Medicine resists such tools because physicians' identity rests on the idea that clinical judgment is irreplaceable art.