Medicine 1950

Smoking and Carcinoma of the Lung

Richard Doll & Austin Bradford Hill

Men with lung cancer were almost all smokers — counting backwards exposed the cause.

Choose your version

In depth · the introduction

Lung-cancer deaths had risen fifteenfold in a generation — so two researchers asked a simple question: of the men dying of it, how many had never smoked? The answer was almost none.

The big idea

You cannot ethically give people cancer to test what causes it. So Doll and Hill ran the experiment backwards. They found a large group of patients who already had lung cancer (the “cases”) and, for each one, a similar patient without it (a “control”) — same sex, same age, same hospital. Then they simply asked everyone about their smoking.

The gap was enormous. Among the men with lung cancer, almost every single one was a smoker — only 2 in 649 had never smoked. Among the matched healthy men, far more were non-smokers. Stack those numbers up and smokers were roughly fourteen times more likely to be in the cancer group. That kind of comparison — cases versus controls — became one of medicine's most powerful tools.

How it came about

After the Second World War, British doctors were alarmed: deaths from lung cancer were climbing relentlessly. Two suspects led the field — the soot and exhaust of modern cities, and cigarettes, which had become wildly popular. Richard Doll, a young physician, teamed up with Austin Bradford Hill, the statistician who had just pioneered the randomised clinical trial, to find out which.

Through 1948 and 1949 they sent trained interviewers into twenty London hospitals to question more than a thousand cancer patients and just as many matched controls. Doll himself smoked when the study began; the results were so clear that he quit partway through. They published in the British Medical Journal in 1950 — the very same week an American team reported the same finding.

Why it mattered

Cigarettes were everywhere — advertised by doctors, handed out to soldiers, woven into daily life. To claim they were killing people took unusually strong evidence, and a single study could be dismissed. But this one, confirmed by a parallel American study and then by Doll and Hill's own long-term follow-up of 40,000 doctors, slowly turned the tide. It underpinned the warning labels, advertising bans, and public-health campaigns that followed — and it taught medicine how to hunt for the causes of disease when you cannot run an experiment.

A way to picture it

Imagine a fire keeps breaking out in one neighbourhood. You cannot start fires to test theories, so instead you visit every house that burned and an equally ordinary house next door that did not, and ask each what was in the kitchen. If nearly every burned house had a faulty heater and the unburned ones mostly did not, you have your suspect — even without ever lighting a match. That backwards comparison, burned versus unburned, is exactly what a case–control study does: cancer patients versus healthy ones, asking who smoked.

Where it sits

This is one of the foundation stones of modern epidemiology. Bradford Hill had already given medicine the randomised trial (its way of testing treatments); here he and Doll gave it the case–control study (its way of finding causes). The work sits alongside the Library's other turning points in medicine, and its method — comparing the sick with the well and weighing the odds — now lies behind almost every headline about what is good or bad for you.

The original document

Original source text

Richard Doll & A. Bradford Hill · British Medical Journal 2(4682): 739–748 · September 30, 1950

The increase in lung cancer

The number of deaths attributed to cancer of the lung provides one of the most striking changes in the pattern of mortality recorded by the Registrar-General. For example, in the quarter of a century between 1922 and 1947 the annual number of deaths recorded increased from 612 to 9,287, or roughly fifteenfold. This remarkable increase is, of course, out of all proportion to the increase of population — both in total and, particularly, in its older age groups.

The method of the investigation

Twenty London hospitals were asked to co-operate by notifying all patients admitted to them with carcinoma of the lung, stomach, colon, or rectum. … On receipt of the notification an almoner, engaged wholly on research, visited the hospital to interview the patient, using a set questionary.

As well, however, as interviewing the notified patients with cancer of one of the four specified sites, the almoners were required to make similar inquiries of a group of “non-cancer control” patients. … for each lung-carcinoma patient visited at a hospital the almoners were instructed to interview a patient of the same sex, within the same five-year age group, and in the same hospital at or about the same time.

The result

It will be seen that the vast majority of men have been smokers at some period of their lives, but also that the very small proportion of those with carcinoma of the lung who have been non-smokers (0.3%) is most significantly less than the corresponding proportion in the control group of other patients (4.2%).

[ … ]

This is not necessarily to say that smoking causes carcinoma of the lung. The association would occur if carcinoma of the lung caused people to smoke or if both attributes were end-effects of a common cause. The habit of smoking was, however, invariably formed before the onset of the disease … so that the disease cannot be held to have caused the habit ; nor can we ourselves envisage any common cause likely to lead both to the development of the habit and to the development of the disease 20 to 50 years later.

We therefore conclude that smoking is a factor, and an important factor, in the production of carcinoma of the lung.

Richard Doll · A. Bradford Hill · Statistical Research Unit of the Medical Research Council, London · 1950