The 1950 study that first tied smoking to lung cancer
Two British researchers interviewed hundreds of lung-cancer patients and found almost all of them were smokers — one of the first hard signals that cigarettes were deadly.
From “Smoking and Carcinoma of the Lung” · Doll & Hill (1950)
The 30-second version
In post-war Britain, lung cancer was rising fast and no one knew why. Doll and Hill interviewed patients with lung cancer and compared them to similar patients with other illnesses. The lung-cancer group was overwhelmingly made up of smokers, and the heaviest smokers had the highest rates. It was an observational study — it couldn't prove cause by itself — but the link was so strong it helped launch decades of research that established smoking as a cause of cancer.
Evidence grade
How much to trust this study — broken down, not a black box.
A large, carefully designed case-control study with a strikingly strong association — excellent for its era — but observational, so it pointed strongly at smoking without proving causation on its own.
By the numbers
The figures that matter
The core group of men with lung cancer who were interviewed.
Almost every male lung-cancer patient was a smoker — only two were not.
The heaviest smokers showed the highest rates, strengthening the link.
What they found
The key findings
Nearly every lung-cancer patient smoked
High confidenceAmong the men with lung cancer, all but a tiny handful were smokers — far more than in the comparison group.
Why it matters: It flagged smoking as a prime suspect at a time when its dangers were dismissed.
More cigarettes, higher risk
High confidenceRisk rose with the amount smoked — a 'dose-response' pattern that makes a causal link more believable.
Why it matters: Dose-response is one of the classic signs that an exposure may actually cause harm.
It couldn't prove cause by itself
Moderate confidenceBecause it observed people rather than running an experiment, the study could show a strong link but not definitively prove smoking caused the cancer.
Why it matters: Honest about its limits — causation took years more evidence to nail down.
The process
How the study worked
The researchers identified hospital patients with lung cancer, interviewed them about their smoking, and compared their answers to a matched group of patients with other diseases.
- 1
Find the cases
Identify hospital patients diagnosed with lung cancer.
- 2
Pick comparison patients
Match each to a patient with a different illness, by age and sex.
- 3
Interview both groups
Ask detailed questions about smoking history.
- 4
Compare the rates
See whether smoking was more common among the lung-cancer group.
The data
What the numbers actually show
The contrast was stark: smokers dominated the lung-cancer group, and the proportion of non-smokers was tiny compared with the comparison patients. The risk climbed with heavier smoking — a pattern that's hard to explain away as coincidence, even if a single observational study can't seal the case.
Non-smokers in each group
Share of patients who had never smoked.
Takeaway: Non-smokers were far rarer among lung-cancer patients than among comparison patients.
A critical eye
Strengths & limitations
✓ What it did well
- Large sample for its time.
- Striking, dose-dependent association.
- Careful matching of comparison patients.
- Honest about being observational.
! What to keep in mind
- Observational — can't prove causation on its own.
- Possible recall bias in interviews.
- Hospital patients may not represent the general population.
- Confounding factors weren't fully eliminated.
Why you should care
So what?
This was a turning point in public health. The link it surfaced — later confirmed by many studies — underpins every cigarette warning label, smoking ban, and anti-tobacco campaign since. It also became a textbook example of how to reason carefully from observational data.
What’s next
Questions this opens up
Could the link be explained by some other shared factor?
Would tracking people forward in time confirm it? (Later cohort studies did.)
What is the biological mechanism connecting smoke to cancer?
Plain-English glossary
Jargon buster
- Case-control study
- A study that compares people who already have a disease to similar people who don't, looking back at past exposures.
- Dose-response
- When more of an exposure leads to more of an effect — a clue that the link may be causal.
- Confounding
- When a hidden third factor is responsible for an apparent link between two things.
- Recall bias
- When people with a disease remember or report past habits differently than those without it.