562. How to Succeed at Failing, Part 2: Life and Death

Episode Summary

Executive Summary: The episode examines failure in health care through tragic patient-safety cases, arguing that most harm is systemic rather than purely individual. It contrasts blame-and-punish responses with Amy Edmondson’s six-part failure spectrum and highlights how transparency, learning, and experimentation can reduce preventable harm. The show also compares a disastrous UK NHS IT rollout with Bob Langer’s iterative scientific breakthroughs to show when failure is blameworthy versus productive.

Main Topics: Patient harm and transparency in health care (Priority: 5/5): Carol Hemmelgarn describes her daughter’s death from leukemia complicated by medical errors, mislabeling, infection, and withheld information, arguing that families deserve honest communication and that hospitals must learn publicly from harm. Criminalizing frontline errors versus system failure (Priority: 5/5): The Redonda Vaught case illustrates how medication errors can arise from system design flaws, yet legal punishment may suppress reporting, create fear, and obscure organizational responsibility. Amy Edmondson’s spectrum of causes of failure (Priority: 5/5): Edmondson proposes six categories—sabotage, inattention, inability, task challenge, uncertainty, and experimentation—to distinguish blameworthy failures from those that should be treated as learning opportunities. Large-scale institutional failure in the NHS IT project (Priority: 4/5): John Van Reenen explains how a top-down, rushed, centrally imposed electronic records system in the UK National Health Service collapsed under complexity, poor engagement, and rigid contracting. Experimentation as the basis of scientific progress (Priority: 4/5): Bob Langer recounts repeated grant rejections and more than 200 experimental failures, showing that research advances through iterative trial, visible evidence, and persistence. Grit versus quit (Priority: 3/5): The episode sets up the question of when perseverance is virtuous and when it becomes wasteful, suggesting that judgment depends on scientific plausibility, advice, and potential impact.

Key Arguments: Medical harm is often rooted in systems, labeling, staffing, design, and culture—not just individual negligence. Fear of litigation encourages silence, which blocks learning and worsens patient safety. Criminal prosecution of clinicians can deter speaking up about errors and make organizations less honest. Failure should be categorized by cause; not all failures deserve blame, and some are necessary for learning. Top-down, command-and-control reform is especially risky in complex systems like health care IT. Scientific and entrepreneurial progress require experimentation, and repeated failure is often a sign that discovery is being pursued seriously. Transparency after harm matters because patients and families often want to prevent recurrence more than they want retribution.

Data Points: Age of Carol Hemmelgarn’s daughter: 9 years old - Alyssa was diagnosed with leukemia at age nine. Time from leukemia diagnosis to death: 10 days - Hemmelgarn says her daughter died 10 days after being diagnosed. Time to honest conversation with family: 3 years, 7 months, and 28 days - Hemmelgarn says this is how long it took for the organization to have its first honest conversation with her family. Year of Vanderbilt medication error case: 2019 - Redonda Vaught was prosecuted after administering the wrong drug in a hospital case. Patient age in Vanderbilt case: 75 years old - The patient who died had been admitted for a subdural hematoma. Institute of Medicine preventable harm estimate: 2% to 3% of U.S. hospital admissions - 1999 report To Err Is Human estimated preventable injury or death at this rate. Annual deaths from medical error (1999 report): 44,000 to 98,000 per year - The same report estimated this range of deaths in U.S. hospitals. Annual preventable deaths (2013 study): about 200,000 per year - A Journal of Patient Safety study estimated higher preventable hospital deaths. Failed research grants: 9 - Bob Langer said his first nine research grants were turned down. Failed experimental attempts: over 200 - Langer said he failed more than 200 times before developing a working particle delivery method. Estimated cost of NHS IT project: up to $20 billion - Van Reenen described the Connecting for Health project as a huge loss.

Pivotal Quotes: "What we see is that not everything is implemented in the system." — Carol Hemmelgarn: She explains why patient-safety reforms have not been absorbed consistently across organizations. "To criminalize this, I think, reflects an erroneous belief that by doing so, we'll preclude human error." — Amy Edmondson: She argues that punishing clinicians for complex errors discourages openness and learning. "I failed over 200 times before I finally got something to work." — Bob Langer: He describes the iterative nature of scientific experimentation and persistence.

Implications: Health systems should replace blame with diagnosis, transparency, and redesign. Better failure classification can improve safety culture, guide regulation, and help teams know when to persevere, when to experiment, and when to stop.