Atul Gawande · 2002

Complications

A Surgeon's Notes on an Imperfect Science

Rough AI truth score

83/100

Gawande accurately depicts medicine as skilled but fallible work performed under uncertainty, with learning curves, cognitive errors, and system failures that patients rarely see. The cases are not prevalence estimates, yet the broader conclusions align with modern patient-safety and diagnostic-error evidence.

Based on three central claims · high confidence

The three claims

01Supported

Medical error arises from interacting human, team, workflow, equipment, and organizational failures rather than from a few uniquely careless clinicians.

Patient-safety research consistently models adverse events as system phenomena involving communication, design, workload, supervision, and latent conditions. Individual negligence exists, but blame-only approaches miss recurring mechanisms and prevent organizational learning.

National Academies: To Err Is Human AHRQ Making Health Care Safer II

02Mostly supported

Clinical skill improves through supervised experience and learning curves, meaning outcomes can differ by operator volume and stage of mastery.

Procedure-outcome and learning-curve research supports experience effects for many operations and technical tasks. Volume is an imperfect proxy for competence, curves differ by procedure and team, and centralization can trade quality against access and continuity.

National Academies: Crossing the Quality Chasm AHRQ patient safety practices

03Mostly supported

Diagnosis and treatment inevitably require judgment under uncertainty, so competent clinicians can disagree or be wrong even when acting responsibly.

Diagnostic evidence is incomplete, probabilistic, time-dependent, and shaped by prevalence and test performance. Structured reasoning and follow-up reduce avoidable error, but no system can remove irreducible uncertainty or guarantee one correct decision for every preference-sensitive case.

National Academies: Improving Diagnosis in Health Care Cognitive biases associated with medical decisions

Other claims worth checking

Transparency about uncertainty can strengthen rather than destroy trust.
Innovation creates ethical tension when early patients bear learning risk.

What this number means. It is an AI-generated first-pass judgment of three central factual or causal claims—not a rating, exhaustive fact-check, or human peer review. Claim credits are 100% for supported, 75% for mostly supported, 50% for mixed, and 25% for weak, then averaged and rounded. Lower confidence means the score should move more as better evidence arrives.

Method three-central-claims/0.1.0 · checked 2026-09-01 · 3/3 selected claims assessed · method and source audit