A seven-part interactive documentary based on the landmark 1981–1984 McMaster University series — one of the most cited publications in the history of Canadian medicine, written in Hamilton, where you studied.
Published in Hamilton — your city — this series became the founding document of an intellectual movement ranked by the BMJ among the 15 most important milestones in modern medicine, alongside antibiotics and vaccines.
David Lawrence Sackett (1934–2015) arrived in Hamilton at age 32 to build something that had never existed anywhere in the world: a department of clinical epidemiology focused on the individual patient at the bedside, not population statistics in the abstract.
Before Sackett, clinical decisions were governed by expert authority, tradition, and anecdote. He used a favourite story: George Washington, age 68, developed epiglottitis. His physicians — following expert consensus — prescribed bloodletting. Eight pints. The experts chose exsanguination. Washington likely died of their treatment.
“The EBM movement started in 1981 when a group of clinical epidemiologists at McMaster University, led by David Sackett, published the first of a series of articles in CMAJ advising physicians how to appraise the medical literature.”
— Thoma & Eaves, Aesthetic Surgery Journal, 2015The term “evidence-based medicine” was coined a decade later by Gordon Guyatt — Sackett’s own mentee at McMaster. But the intellectual engine had been running since 1981, in seven articles whose print copies are still in a box in Hamilton.
Sackett received the Order of Canada (2001), the Canada Gairdner Wightman Award (2009), and was inducted into the Canadian Medical Hall of Fame (2000). In 1994 he left McMaster to found the Centre for Evidence-Based Medicine at Oxford — exporting Hamilton’s revolution to the world. Until his death in 2015, he still lectured at McMaster.
“This series, one of the most frequently cited ever in CMAJ, changed medical practice worldwide.”
— CMAJ Editor-in-Chief Paul C. Hébert, 2007Click any episode to expand its full content summary, original key concepts, and documentary angles for the AK format.
“Can I trust what I’m reading — and where do I even begin?”
The opening article established both the stakes and the method. Most published clinical research is flawed, misleading, or irrelevant — and most readers cannot tell the difference. The article’s central gift was a three-question framework that structures all critical appraisal and applies to every study type examined in Parts II–VII.
The three master questions: (1) Are the results valid? Was the study designed well enough to generate trustworthy data? (2) What are the results? What did the study actually find, precisely and quantitatively? (3) Will the results help? Are the findings applicable to real people in real settings? Deceptively simple, profoundly demanding — these are the spine of the entire series.
The article also introduced the study design hierarchy: why a randomised trial answers different questions than a cohort study, and why expert opinion — however august — sits at the bottom of the evidence pyramid.
Historical note: This article appeared in March 1981, the public-facing manifesto of the McMaster revolution. The entire subsequent architecture of EBM — the Users’ Guides to the Medical Literature, the Cochrane Collaboration, NICE guidelines, CADTH reviews — rests on the three questions this single article introduced.
“Does this test actually tell you what it claims to tell you?”
A masterclass in the mathematics of uncertainty. A positive test result does not mean you have the disease. What it means depends entirely on the test’s sensitivity (how reliably it catches true cases), its specificity (how reliably it excludes non-cases), and the pre-test probability — the probability you had the disease before the test was done at all.
The article introduced likelihood ratios as the practical tool for translating a test result into a revised probability — more useful than sensitivity and specificity alone, because they work regardless of disease prevalence. It warned against spectrum bias: validating a test in a population of clear-cut sick vs. clearly-well patients, then applying it in clinical reality where most patients fall between extremes.
A valid diagnostic study requires blind comparison with an independent gold standard. Without this, a diagnostic test is merely an unverified assertion dressed in laboratory clothes.
Historical note: Cited in over 130 PubMed-indexed studies. The concepts of sensitivity and specificity this article popularised for clinicians became household terms during the COVID-19 pandemic — four decades later — when the public suddenly had to evaluate rapid antigen tests. Sackett had taught this framework to doctors in 1981. The public finally needed it in 2020.
“What actually causes what — and how do we know?”
The question that drives most popular health journalism and most public misunderstanding: the difference between association and causation. Two things being correlated — even strongly, consistently, over decades — does not mean one causes the other. The article gave readers a practical framework for deciding when an observed association justifies a causal inference.
The McMaster team drew on the Bradford Hill criteria — nine viewpoints developed by English statistician Sir Austin Bradford Hill in 1965: strength of association, consistency, specificity, temporality, dose-response gradient, plausibility, coherence, experimental evidence, and analogy. No single criterion is sufficient; all must be weighed together.
The article examined cohort studies as the primary design for etiologic research, and confronted the ever-present spectre of confounding: the third variable that creates the illusion of causation between two others.
Historical note: The tobacco industry’s decades-long campaign to deny the causal link between smoking and lung cancer was a systematic exploitation of every weakness this article teaches readers to defend against — cherry-picking, manufacturing confounders, demanding impossible proof thresholds, questioning the Bradford Hill criteria piecemeal. The article is, in part, an answer to that specific campaign of epistemic manipulation.
“When can an experiment settle the question that observation alone cannot?”
Part IV extends causation by introducing the randomised controlled trial (RCT) as the gold standard for establishing cause — and by confronting the reality that RCTs are often impossible or unethical. You cannot randomly assign people to smoke for 30 years. The article explored alternatives: case-control studies and natural experiments.
The RCT’s power lies in two features: allocation concealment (random assignment that distributes all confounders — known and unknown — equally between groups before the study begins) and blinding (preventing the expectation of benefit from creating the appearance of benefit). The article carefully separated randomisation from random sampling — a confusion that persistently distorts lay reporting of research.
Case-control studies identify people who developed disease and compare them to similar people who did not, looking for differences in past exposure. Efficient and ethical, but structurally vulnerable to recall bias.
Historical note: The thalidomide catastrophe (1957–1962) — in which a sedative prescribed to pregnant women caused severe limb defects in thousands of children — accelerated global regulatory demand for RCT evidence before drug approval. The framework this article codified is the intellectual foundation of every drug approval process in the world today. Canada’s own Frances Oldham Kelsey held thalidomide off the US market by demanding exactly this kind of evidence.
“Does this treatment actually work — and how much, for whom?”
The most practically urgent of the seven parts. Part V structured the evaluation of therapeutic studies around the three master questions from Part I, now applied specifically to treatment: Are the results valid? What were the results? Can the results be applied to patients like mine?
The article introduced the critical, often-abused distinction between absolute risk reduction (ARR) and relative risk reduction (RRR). A drug reducing risk from 2% to 1% has an RRR of 50% but an ARR of 1%. The number needed to treat (NNT) — how many patients must receive the treatment for one to benefit — makes the absolute benefit undeniable and honest in a way that relative risk alone is not.
Most memorably: internal mammary artery ligation, a surgery widely performed for angina in the 1950s. When tested in a blinded RCT against a sham procedure (chest opened and sutured without ligating the artery), both groups improved equally. Thousands underwent needless open-chest surgery based on nothing but uncontrolled clinical impression.
Historical note: Part V is cited by over 140 indexed studies. Its framework became the direct template for the JAMA “Users’ Guides to the Medical Literature” series (1993–2000) — the canonical EBM reference that reshaped clinical training globally. The therapy checklist in Part V was adopted verbatim by teaching hospitals across North America within five years of publication.
“What will happen — and how honest is the prediction?”
Prognosis — the science of predicting outcomes — is the question patients ask most urgently and that science answers most poorly. Part VI argued that valid prognostic research requires an inception cohort: patients gathered at a common, reproducible, early point in their disease. Starting with survivors inflates the prognosis. Starting at specialist referral depresses it. The inception point is not arbitrary; it is the whole argument.
The article introduced survival analysis and Kaplan-Meier curves — methods for describing how a cohort fares over time, tracking not just whether an outcome occurs, but when, and accounting for patients who are “censored” (leave the study) before its end. The critical validity threat: loss to follow-up — if patients who do poorly are more likely to drop out, the remaining cohort appears healthier than it is.
The article required objective, blinded outcome criteria and demanded adjustment for baseline prognostic factors — without which you are comparing fundamentally different patients.
Historical note: The “5-year survival” statistic that appears in every cancer news story is a direct application of the survival analysis concepts this article taught — and the public’s persistent misunderstanding of it illustrates exactly why Part VI’s precision matters. The statistic measures survival from diagnosis, not from disease onset. The inception point matters in cancer, and in every other form of predictive reasoning.
“Who pays, who benefits — and who has the power to make that decision?”
Three years after Parts I–VI, Greg Stoddart — co-founder of McMaster’s Centre for Health Economics and Policy Analysis (CHEPA) and a global pioneer of health economics — added the dimension the first six parts had set aside: cost. Part VII introduced three forms of economic evaluation used to decide what health systems fund.
Cost-effectiveness analysis (CEA) compares the cost of achieving a unit of health outcome — lives saved, years of life gained — between alternatives. Cost-benefit analysis (CBA) assigns monetary values to health outcomes, permitting comparison across intervention types. Cost-utility analysis (CUA) uses the QALY (quality-adjusted life year) as the common currency: one year in perfect health equals one QALY. The incremental cost-effectiveness ratio (ICER) — the additional cost per additional QALY gained — is the number that health technology assessment bodies use to approve or reject coverage.
The article stressed that the analytical perspective is not neutral: a drug that saves the hospital money may cost the patient and family enormously. And it required sensitivity analysis — testing whether the conclusion changes when uncertain inputs are varied — as the intellectual honesty requirement of every economic evaluation.
Historical note: Stoddart co-authored Methods for the Economic Evaluation of Health Care Programmes (Drummond, Sculpher, Claxton, Stoddart, Torrance) — now in its fourth edition and the global reference for health economic evaluation. The QALY framework is used today by NICE (UK), CADTH (Canada), and health technology assessment bodies worldwide. When a government decides whether to cover a new cancer drug, this article’s framework is the methodology behind that decision.
This is not only a medical history documentary. It is a documentary about how knowledge is built, corrupted, and wielded — speaking directly to journalism, true crime, propaganda, and power.
Every technique for corrupting scientific evidence — cherry-picking, manufacturing confounders, exploiting the relative-vs-absolute risk ambiguity, weaponising expert authority — is also a propaganda technique. Each of the seven parts exposes one more tool in the manipulator’s methodological arsenal.
Forensic science is applied clinical epidemiology. Bite mark analysis, hair microscopy, blood-spatter modelling — all fail Part II’s diagnostic checklist. Junk causal science sends innocent people to prison. Survivor bias distorts recidivism statistics. Each episode has a courtroom hiding inside it.
Who funds the studies, who sets the endpoints, who controls journal access — every episode contains a layer about the political economy of knowledge production. Part VII makes it fully explicit: someone is always deciding whose costs count and whose benefits get measured.
A practical framework for producing all seven episodes on alexandrakitty.com, using your existing HTML/CSS skills and platform infrastructure.
The Sackett series was written to help clinicians read journals. But its three master questions — Is this valid? What do the results actually show? Does it apply to me? — are the foundational questions of every critical thinking discipline.
In an era of algorithmic information, manufactured consensus, and weaponised statistics, a seven-part series teaching people to read evidence critically is not a medical history documentary. It is a survival guide.
The original articles are freely available on PubMed Central. The print copies are in a box in Hamilton. And the ideas inside them are as sharp as the day Sackett and his colleagues put them into print in 1981 — the year evidence-based medicine was born, at a university in the city where you studied.