Smoking causes lung cancer. That has not been in scientific dispute since the early 1950s. What is less widely understood is that the cigarette itself was redesigned in response — filters, ventilation holes, falling tar yields — and that this changed which lung cancer people develop and where in the lung it sits. Central squamous carcinoma, which announced itself with coughing up blood and could be seen down a bronchoscope, has given way to peripheral adenocarcinoma that a plain chest X-ray can miss and a standard bronchoscope cannot reach. UK rates have fallen by around a tenth since the early 1990s — by almost two-fifths in men, while rising by around a third in women, tracking a twenty-five-year difference in when each sex began to give up. This is a consultant thoracic surgeon’s account of that arc, from 82% of British men smoking in 1948 to the Tobacco and Vapes Act 2026, and of what it changed in the operating theatre. Mr Lawrence Okiror is a Consultant Thoracic and Robotic Surgeon at Guy’s and St Thomas’, with private practice at London Bridge Hospital and The Lister Hospital Chelsea. Self-referrals welcome.
Last reviewed: July 2026 · Mr Lawrence Okiror FRCS(CTh) FRCSEd(CTh) · GMC 6150382
Smoking causes lung cancer. Established by Doll and Hill from 1950, confirmed by the Royal College of Physicians in 1962 and the US Surgeon General in 1964, and never seriously overturned since.
The 2014 Surgeon General’s report concluded that rising adenocarcinoma results from changes in cigarette design and composition since the 1950s. The tumour moved outwards, and the operation moved with it.
Stopping four weeks before an operation reduces complications. Stopping after a lung cancer diagnosis is associated with 6.6 versus 4.8 years median survival. Neither window is closed.
Almost everything written about smoking and the lungs draws on evidence gathered from people who smoked unfiltered, high-tar cigarettes. That product no longer exists. Tar yields in the UK were capped at 15 mg from 1992, 12 mg from 1998, and 10 mg under the 2001 European directive, and average nicotine had already fallen from about 2 mg in 1955 to about 0.9 mg by 1996. The exposure changed, and the disease changed with it.
What did not change is the causal relationship. The redesign altered which cancer people get and where it sits — not whether they get it. Understanding that distinction is what allows a scan, a symptom, or a nodule to be interpreted against the disease as it presents now rather than as it presented in 1960, and it is the work of the consultation rather than the article. Request an appointment within 2–3 working days →
I begin here because everything that follows is a refinement of this sentence, and refinements are easily mistaken for retractions. The remainder of this piece describes how the cigarette was redesigned, how the disease it produces changed shape and position as a result, and how the whole apparatus of finding and treating that disease had to change with it. None of that softens the opening claim. The redesign altered which lung cancer people get and where it sits. It did not alter whether they get it.
That distinction matters clinically, because the version of the story that circulates most widely — that cigarettes became “safer” — was a marketing proposition rather than a medical one, and the people who acted on it are among the patients I now operate on.
The scale is difficult to picture now. The highest recorded level of smoking among men in Great Britain was 82% in 1948, of whom 65% smoked manufactured cigarettes [1]. The remainder smoked pipes or cigars as well as, or instead of, cigarettes. Both figures are worth holding: 82% is the proportion using tobacco in any form, and 65% is the proportion smoking the manufactured cigarette that carried most of the subsequent risk.
Cigarettes had been issued in service rations through both world wars. They were cheap, socially universal, and marketed without restriction. A generation of men returned from the Second World War with an established habit and no reason to question it.
The pattern in women was entirely different, and it is the reason for one of the most persistent misunderstandings about lung cancer today. Smoking prevalence among women in 1948 was 41%. It then remained broadly constant until the early 1970s, peaking at around 45% in the mid-1960s [1]. Women had taken up smoking later and in smaller proportions, and — critically — they gave up later too. That offset is picked up in section 10.
For contrast: around 5.3 million adults in the UK, or 10.6%, smoked cigarettes in 2024, the lowest proportion since the current series began [2]. The proportion of adults who have never smoked rose from 37.4% in 1974 to 60.4% in 2019 [1]. This is one of the largest sustained behavioural changes in British public health, and it is the reason the lung cancer figures look as they do.
The design mattered as much as the result. Lung cancer had been rising sharply through the first half of the twentieth century, and the two explanations favoured by most senior clinicians at the time were improvements in diagnosis and atmospheric pollution — both of which, incidentally, still circulate today. Bradford Hill set out to build a study large enough to distinguish between them.
The 1950 preliminary report established the association [3]. The 1952 paper strengthened it [4]. The British Doctors Study then did something the case-control design could not: it followed a defined population forward in time, reporting at ten years, twenty years, forty years, and finally at fifty years in 2004 [5, 6]. Doctors were chosen because they were easy to trace on the medical register. They also, as a group, began giving up early — which turned the cohort into an unintended natural experiment on the effects of cessation.
The British contribution continued. In 2000, Peto, Darby, Doll and colleagues combined national statistics with two case-control studies conducted around 1950 and 1990 to describe what had happened across the intervening half-century [7]. The finding was that prolonged smoking carried a considerably greater risk than the original studies had been able to detect, and that prolonged cessation conferred a correspondingly greater benefit. The accompanying editorial summarised the national effect: widespread cessation had halved UK lung cancer mortality.
The Royal College of Physicians published Smoking and Health in 1962, the first major report by a national medical body to state the case plainly and to call for policy action. It circulated widely, including in Washington.
Two reasons, one respectable and one not.
The respectable objection came from Ronald Fisher, the most influential statistician of the age, who argued that the association might be confounded — that some constitutional or genetic factor might predispose both to smoking and to lung cancer, with no causal arrow between them. This was a serious argument made in good faith about a genuine methodological problem, and answering it took time and data. It is also the reason the debate produced the modern framework for causal inference in epidemiology, which Bradford Hill set out in 1965 — after the American report rather than before it. The criteria by which medicine now decides whether an association is causal were forged in this argument.
The disreputable reason was a sustained industry campaign to manufacture uncertainty, most memorably summarised in an internal tobacco company memorandum in 1969 with the observation that doubt was the product being sold. The strategy did not require winning the scientific argument. It required only keeping it open.
A version of the same manoeuvre persists in this field. When the shift from squamous carcinoma to adenocarcinoma is attributed instead to changes in pathological classification, it is worth knowing that much of that analysis has come from industry-linked sources. Classification changes are real and account for part of the shift, as section 7 sets out. They do not account for all of it.
The report itself was 387 pages and was released on a Saturday — timed to limit the effect on the stock markets while catching the Sunday newspapers. The Kennedy administration had convened the committee two years earlier, prompted in part by the circulation of the Royal College of Physicians report from London. British evidence produced an American document, which then changed behaviour and law on both sides of the Atlantic.
Three consequences followed, and two of them produce the trends described later in this piece.
A Gallup survey in 1958 found that only 44% of Americans believed smoking caused cancer. By 1968 the figure was 78% [8]. In the space of a decade, a contested scientific claim became general knowledge. This is the mechanism by which the report acted on prevalence, and prevalence is what eventually acts on cancer.
The report called for “appropriate remedial action” without specifying what that action should be, leaving the question to legislators. Congress passed the Federal Cigarette Labeling and Advertising Act in 1965, requiring a health warning on cigarette packages, and the Public Health Cigarette Smoking Act in 1969, which banned cigarette advertising in broadcast media from September 1970 [8]. Comparable measures followed in the UK and across Europe over the succeeding decades, culminating in the smokefree legislation of 2006–07 and the Tobacco and Vapes Act 2026.
The 1964 Committee described smoking as a cause of lung cancer in men but only a probable cause in women. The evidence base at the time was overwhelmingly male, because the disease at the time was overwhelmingly male. Set that hesitation against the fact that female smoking prevalence stayed flat for another decade, and the modern divergence in incidence between the sexes becomes less surprising. The document that turned the male curve was equivocal about women, and the female curve did not turn for another generation.
The third response to 1964 was commercial. Rather than the selling of cigarettes stopping, the product was reformulated and sold as reassurance.
Filters had been introduced in the 1950s and spread rapidly. Tar and nicotine yields were progressively reduced, and brands were marketed on their machine-measured numbers. Ventilation holes were perforated into filters, which lowered the yield recorded by a smoking machine without necessarily lowering the dose received by a person, because fingers and lips cover the holes and because smokers compensate by inhaling more deeply and drawing larger puffs.
Regulation ratified and accelerated a trend the industry had begun voluntarily some thirty years earlier. The Cigarettes (Maximum Tar Yield)(Safety) Regulations 1992 came into force on 30 November 1992, enacting European Directive 90/239/EEC and capping tar at 15 mg per cigarette. This was reduced to 12 mg from 1 January 1998 [9]. The 2001 Tobacco Products Directive then set maximum yields of 10 mg tar, 10 mg carbon monoxide and 1 mg nicotine. Average nicotine had already fallen from about 2 mg in 1955 to about 0.9 mg by 1996 [1].
The final regulatory word on the exercise came in 2014, when the revised European directive removed the tar, nicotine and carbon monoxide figures from cigarette packs altogether, on the evidence that the numbers misled consumers into believing that some brands were safer than others [1]. The measurement that had defined the product for forty years was withdrawn as actively deceptive.
It is the sequence that gives this section its point. The redesign was a commercial response to Doll and Hill, sold to the public as a safety improvement. It did not make smoking safe. It changed the disease.
This is not a soft association. It is a formal causal conclusion by the same institution that indicted the cigarette fifty years earlier — the report that named the product, and then, half a century later, the report that named its redesign [10].
A 2017 review in the Journal of the National Cancer Institute examined filter ventilation specifically and set out the case that it contributed materially to the rise in adenocarcinoma [11]. In United States data, adenocarcinoma overtook squamous cell carcinoma at around 1990, and now accounts for roughly 60% of non-small cell lung cancer [11]. That figure is American; the direction of travel in the UK is the same, but I am not aware of a comparably clean published UK crossover date and do not assert one here.
The most directly relevant study for a surgeon is not about histology at all but about position. Reviewing CT scans, chest radiographs and records for 330 smokers with lung carcinoma, Stellman and colleagues found a direct association between the tar yield of the cigarette smoked and where the tumour sat — lower-yield cigarettes were associated with more peripheral tumours [12]. The relationship holds for location independently of cell type. That is the finding that ends up mattering in theatre.
Two honest caveats belong here. First, the deeper-inhalation mechanism is not the only candidate. Changes in tobacco curing and blending altered the nitrosamine content of smoke, and tobacco-specific nitrosamines have a particular affinity for the peripheral lung. Both mechanisms may operate. Second, successive revisions to the World Health Organization classification of lung tumours, together with immunohistochemistry, have reassigned many tumours formerly labelled large-cell or undifferentiated into the adenocarcinoma category. Analysis of United States registry data attributes part of the apparent rise in adenocarcinoma since 2005 to precisely this reclassification.
Reclassification is real. It is not sufficient. A change in how pathologists label tumours does not move those tumours from the centre of the chest to the periphery, and the peripheral shift is what section 8 and section 9 are about.
Consider what that list actually consists of. Six investigations, four of them surrogates for something nobody could see directly — and every one of them calibrated to a cancer that grew in or around a large airway.
Films were taken in two projections, posteroanterior and lateral, and it was frequently the lateral film that carried the finding — the retrocardiac space, the hilar structures, the mediastinum. A routine lateral chest film is now an uncommon request. That is a reasonable adaptation to a disease that has moved outwards, but it is worth recognising it as an adaptation rather than an improvement in isolation.
Rigid bronchoscopy was the instrument of the 1950s; flexible bronchoscopy arrived from the late 1960s and extended the range considerably. Both, however, work by travelling down airways. A tumour growing into the lumen of a main bronchus can be seen, biopsied, and photographed. A four-millimetre nodule in the outer third of an upper lobe cannot be reached by either instrument. The bronchoscope did not become less useful; the tumour moved out of its range.
Coughing up blood was a classical presentation of lung cancer because a tumour eroding the lining of a large airway bleeds into a tube that leads directly to the mouth. Peripheral adenocarcinoma sits in lung tissue that has no such connection, and it is characteristically silent until it is large or has spread. The disappearance of a symptom is not the disappearance of a disease, and it is one reason a patient can now have a substantial cancer with no symptoms at all. Patients seeking guidance on symptoms should see the pages on coughing up blood and persistent cough.
The evening before an operation, part of the thoracic registrar’s job was to review the barium swallow performed as part of the pre-operative workup, looking for an indentation on the oesophagus — because if something was pressing on it from behind, that probably meant enlarged nodes in the mediastinum. This was described to me by Professor Peter Goldstraw, recalling his own training in the late 1970s; I record it here as his account rather than as a citable source.
Beyond that inference there was cervical mediastinoscopy, which in that era meant peering down the instrument by eye rather than working from a screen. And there was a serum calcium, taken as a surrogate for bony metastatic disease. Each of these is an attempt to infer, from an indirect signal, something that could not be looked at.
King George VI was found in exactly this way in 1951. He was a heavy smoker, one of the 82%. X-rays and a bronchoscopy demonstrated a malignant tumour in the left lung, described in contemporaneous reporting as an obstruction in the left main bronchus. Central. Endobronchial. Found by the two investigations that worked, precisely because of where it sat.
The operation that followed is described in section 9. The point here is that the workup was adequate to the disease. Put the same patient in a clinic today with a peripheral adenocarcinoma and that workup finds nothing at all.
Computed tomography from the 1980s, positron emission tomography from the 1990s, endobronchial ultrasound for nodal staging, and video-assisted mediastinoscopy each replaced an inference with a measurement. Navigational bronchoscopy is the terminal point of that sequence: it exists precisely and only because a standard bronchoscope cannot travel far enough into the lung to reach the lesions that now predominate. The full nodule pathway is set out in the guides to lung nodules and central airway interventions.
There is a satisfying closure to this. Professor Goldstraw — the registrar reading barium swallows for oesophageal indentation — went on to found and chair the International Association for the Study of Lung Cancer Staging Project, which built the international nodal map and the data-driven revisions behind the seventh and eighth editions of the TNM classification. The surrogate was replaced by the measurement, by the same person.
This is the part of the story that is rarely told, and it is the part I see most directly.
On 23 September 1951, Clement Price Thomas removed King George VI’s left lung in a room at Buckingham Palace. Emergency lighting was installed for the purpose and the changing of the guard was moved to St James’s so that the team would not be disturbed. Palace statements described an operation for a “structural abnormality”; the word cancer was avoided. The King died on 6 February 1952, four and a half months later.
A left pneumonectomy was the appropriate operation for a tumour in a left main bronchus, and it remained the standard answer to central disease for decades. I still use a retractor that carries Price Thomas’s name.
Contrast the operation now performed for the disease that predominates. A sub-centimetre or one-to-two centimetre peripheral nodule cannot be seen from outside the lung and often cannot be felt through a port. It has to be localised before it can be removed, frequently by placing a dye marker at the time of biopsy. It is then removed by anatomical segmentectomy — dividing the segmental artery, vein and bronchus and developing the intersegmental plane — which preserves the remainder of the lobe.
The evidence for that operation is recent. Two randomised trials established segmentectomy for peripheral tumours of two centimetres or less with no nodal involvement, and both are set out in detail in Lung Cancer Surgery in 2026. The relevant point here is that those trials could only be conducted once the disease they studied had become common enough to randomise.
Filters and falling tar yields moved the tumour outwards. A tumour in the periphery cannot be found by a plain film or reached by a bronchoscope, which is why navigation and cross-sectional imaging became necessary. A tumour in the periphery does not require a whole lung to be taken, which is why lung-sparing anatomical resection became both possible and appropriate. Magnified stereoscopic vision and articulated instruments are useful in proportion to how precise the dissection has to be, which is why the robotic platform became the natural instrument for it.
In 2024–25 my unit at Guy’s and St Thomas’ performed 892 anatomical lung cancer resections, of which 71.3% were robotic, with a pneumonectomy rate of 1.5% and an operative survival rate of 99.59%. A pneumonectomy rate of one and a half per cent is not primarily a statement about surgical technique. It is a statement about what has happened to the disease.
The overall national picture is one of decline. Cancer Research UK reports that UK lung cancer incidence rates have fallen by around a tenth since the early 1990s [13]. Within that overall fall, the two sexes have moved in opposite directions.
| UK lung cancer incidence rates | Since early 1990s | Last decade |
|---|---|---|
| All persons | Down about a tenth (11%) | Down a twentieth (5%) |
| Males | Down almost two-fifths (37%) | Down a seventh (14%) |
| Females | Up around a third (32%) | Up a twentieth (5%) |
Source: Cancer Research UK, figures for 2019 and 2021–22 [13].
Female rates have not merely held steady. They have risen substantially, and were still rising over the most recent decade. Lung cancer is now the second most common cancer in women in the UK, with around 24,700 new cases a year, against around 25,500 in men — near parity in absolute numbers, in a disease that was overwhelmingly male within living memory [13].
Lay the two histories side by side. Men: 82% in 1948, falling continuously thereafter. Women: 41% in 1948, still 41% in 1974, having peaked at 45% in the mid-1960s [1]. Roughly twenty-five years separate the point at which each sex began to give up in earnest. Apply the two-to-three-decade interval between exposure and diagnosis, and the male curve should turn a generation before the female curve. It did.
Two things this is not. It is not evidence that women are biologically more susceptible to tobacco — the exposure histories are sufficient to explain what is observed, and per-person susceptibility is a separate and unsettled question. Nor is it a statement about the proportion of cases occurring in each sex, which is a different measure governed by different arithmetic. These are age-standardised rates, and rates are the right instrument for this question.
Rates are falling; case numbers are not falling as fast, because the population is ageing and lung cancer is a disease of later life. Cancer Research UK projects UK incidence rates to fall by only around 2% between 2024–26 and 2038–40, with approximately 66,200 new cases each year by 2038–40 [13]. A falling rate and a rising caseload are not contradictory, and the distinction matters for anyone planning services.
Lung cancer also carries one of the steepest deprivation gradients of any cancer: incidence rates in England are 174% higher in the most deprived quintile than the least among women, and 168% higher among men [13].
This section exists because the claim that the smoking ban reduced lung cancer is common, well-intentioned, and wrong in a way that undermines the legislation rather than supporting it. The real evidence is more impressive than the imagined version.
Pell and colleagues collected data prospectively from nine Scottish hospitals accounting for 64% of national admissions for acute coronary syndrome, over the ten months before the legislation and the same period the following year. Admissions fell from 3,235 to 2,684 — a 17% reduction — compared with a 4% fall in England, which had no such legislation at that time, and against a mean annual decrease of 3% in Scotland over the preceding decade [14].
The detail that makes the result interpretable is that 67% of the reduction occurred in people who had never smoked [14]. This was an effect on secondhand smoke exposure, measured in a population that had not changed its own behaviour at all. Admissions and deaths for childhood asthma fell in the same national experiment [15].
A myocardial infarction can be precipitated within hours of an exposure, so removing the exposure produces an effect that is measurable within a year. Carcinogenesis does not work that way. The relevant interval between exposure and diagnosis is two to three decades. Any lung cancer effect of the 2006–07 legislation operates through its influence on how many people smoke and how many give up, and will be visible in incidence data of the 2030s rather than the 2010s.
The honest version is the stronger one. The legislation delivered an immediate, large, measurable cardiac and paediatric benefit, and set in motion a slower one in cancer that is still arriving.
Chronic obstructive lung disease was not a footnote to the 1964 conclusions. It was one of the three headline findings, and it belongs here for the same reason.
The most influential description of how smoking damages lung function came from a prospective study of London working men published by Charles Fletcher and Richard Peto in the British Medical Journal in June 1977 [16]. Their findings were that forced expiratory volume in one second falls gradually across a lifetime, but that in most non-smokers and many smokers, clinically significant airflow obstruction never develops. In susceptible people, smoking causes irreversible obstructive change.
The clause that matters most to patients is the next one. A susceptible smoker who stops will not recover the lung function already lost, but the subsequent rate of loss reverts to normal [16]. That is an honest statement of both halves: the damage does not reverse, and the trajectory does. In my experience it is a more persuasive thing to say to someone than any softer version, because it is true and because it makes stopping worth doing at any point on the curve.
The Fletcher–Peto model implied a single route into COPD: accelerated decline from a normal starting point. Analysis of three large cohorts published in the New England Journal of Medicine in 2015 showed that approximately half of those with moderate COPD never experienced an excess rate of decline at all, but simply started adult life with lower lung function and developed obstruction from there [17]. There is more than one road in, and a criterion built around rate of decline will miss the people who took the other one.
Severe emphysema is not only a barrier to cancer surgery. It is a treatable condition in its own right, through endobronchial valve therapy or lung volume reduction surgery, both of which are NICE-recommended and NHS-commissioned. The full account is set out in Emphysema Surgery in 2026. Where a cancer sits within destroyed emphysematous lung, cancer resection and lung volume reduction can sometimes be delivered in a single operation.
Smoking is also implicated in spontaneous pneumothorax, in several forms of interstitial lung disease, and in impaired healing after any thoracic operation. Patients told elsewhere that their lung function makes them unfit for surgery should see fitness for lung surgery and borderline lung function; structured reassessment, including regional mapping, sometimes changes the answer.
This section is the one most likely to change what a reader does this week, so I want to be precise about it.
A systematic review and meta-analysis of studies of short-term preoperative cessation concluded that at least four weeks of abstinence reduces respiratory complications, and abstinence of at least three to four weeks reduces wound-healing complications [18]. A separate systematic review and meta-analysis of randomised and observational evidence found that preoperative cessation reduces postoperative complications, with the magnitude of effect increasing with the length of the abstinent interval [19].
A 2024 systematic review and meta-analysis confined to pulmonary resection pooled 14 studies and 50,741 patients. Cessation of more than one month before surgery reduced postoperative complications, with an odds ratio of 0.72 (95% CI 0.63–0.83). Cessation of more than two weeks but less than one month did not reduce complications (odds ratio 1.05) or pneumonia (odds ratio 0.98) [20]. Restoration of mucociliary function takes something on the order of six weeks, which offers a plausible explanation for why the shorter interval underperforms.
A belief persists, among patients and occasionally among clinicians, that stopping smoking shortly before an operation increases the risk of respiratory complications — and it is sometimes offered as a reason to continue. The evidence does not support it. Short-term cessation of less than four weeks does not appear to increase or reduce the risk of postoperative respiratory complications [18], and there is no conclusive evidence that stopping within four weeks of surgery confers a higher risk [19]. The finding that generated this idea came from a single small study in 1989 and has not been reproduced.
The practical consequence is straightforward. Earlier is better, four weeks is a sensible target where the timetable allows, and stopping later than that is still worth doing. It should never be a reason to keep smoking.
I include this section because the assumption behind not raising it — that the damage is done, and that asking someone to stop after a cancer diagnosis is both futile and unkind — is wrong on the evidence, and because between a quarter and a half of patients are still smoking at the point of diagnosis.
The prospective cohort data are from Sheikh and colleagues, published in the Annals of Internal Medicine in 2021 [21]. Progression-free survival was 54.4% against 43.8%. After adjustment for confounding variables, cessation remained associated with lower cancer-specific mortality, lower all-cause mortality, and lower risk of disease progression [21].
This is consistent with the earlier systematic review by Parsons and colleagues in the BMJ in 2010, which found that cessation after a diagnosis of early-stage lung cancer improved prognosis and reduced the risk of a second primary cancer [22], and with the 2022 meta-analysis by Caini and colleagues in the Journal of Thoracic Oncology [23]. A 2025 meta-analysis of 25 studies and 17,584 patients found a 26% lower risk of death in those who stopped at diagnosis, with the effect most pronounced in early-stage disease and greater where an active cessation intervention was provided rather than simple advice [24].
An additional two years of median survival is a larger effect than many of the systemic therapies that are, rightly, the subject of intense research investment. It is available to a substantial proportion of patients at the moment they are most receptive to changing anything, and it costs nothing.
Vaping is covered in its own right on the vaping and lung health page; what follows is only the part that bears directly on the arc described above.
Vaping now sits at the centre of this subject whether or not one wishes it to. In Great Britain in 2024, around 5.4 million adults (10.0%) used an e-cigarette daily or occasionally, overtaking the 4.9 million (9.1%) who smoked cigarettes for the first time [2].
Two things can be said with reasonable confidence, and a third cannot.
The first is that the UK public health position — sustained across successive Royal College of Physicians reports and government reviews — is that vaping is substantially less harmful than smoking, and that switching completely is a legitimate route out of tobacco for an established adult smoker. The second is that less harmful is not harmless, and that for someone who has never smoked there is no plausible health argument for starting.
The third is the honest gap. Cigarettes required roughly fifty years of accumulated observation between the first suspicions and the settled causal conclusion, and the peripheral shift described in section 7 took another forty years to become clear. Nicotine vaping at scale is a phenomenon of the last decade. Anyone offering you a confident statement about its long-term oncological risk is going beyond what the data can currently carry, in either direction. That uncertainty is not a reason for an adult smoker to keep smoking, and it is a strong reason for a non-smoker not to begin.
On the acute side: the cluster of severe lung injury reported in the United States in 2019 was traced predominantly to vitamin E acetate in illicit cannabis-containing products, rather than to regulated nicotine vaping. It remains a reminder that the contents of the device matter as much as the act of inhaling.
The arc that began with 82% of British men smoking in 1948 reaches a defined legislative endpoint five months from the date this page was written [25]. The mechanism is unusual: rather than raising the age of sale to a fixed threshold, the Act raises it continuously, so that each successive cohort remains outside the legal market permanently. The Act restricts supply rather than criminalising smoking itself.
Whether it achieves what it intends is a question for the 2050s, for the same reason the 2007 legislation cannot yet be assessed on cancer outcomes. What can be said now is that it closes the arc described in this piece. The product that was issued in wartime rations, defended by an industry campaign for four decades, redesigned in a way that changed the disease it caused, and progressively restricted from 1965 onwards, is being withdrawn from a generation by statute.
For the people already affected, the clinically useful part of this page is not the legislation. It is sections 13 and 14 — that four weeks of abstinence before an operation measurably reduces complications, and that stopping after a diagnosis is associated with a median survival of 6.6 years against 4.8. Those windows are open now.
Smoking causes lung cancer. Everything else in this piece is a refinement of that sentence and none of it is a retraction.
British men smoked at 82% prevalence in 1948. Doll and Hill demonstrated the link from 1950, the Royal College of Physicians stated it in 1962, and the Surgeon General’s report of 1964 converted it into public belief and then into law. The industry’s third response was to redesign the product — filters, ventilation, falling tar — and the 2014 Surgeon General’s report concluded that this redesign is what drove the rise in adenocarcinoma. The tumour moved from the central airways to the periphery.
The consequences of that migration run the whole length of the pathway. A chest X-ray and a bronchoscope, adequate to find a tumour in a main bronchus in 1951, are not adequate to find a nodule in the outer third of a lobe. The barium swallow, the mediastinoscope, and the serum calcium were surrogates for what could not be seen, and they have been replaced by CT, PET, endobronchial ultrasound and navigation. A whole lung once had to be removed; a segment now often suffices.
Rates have fallen for decades — by about a tenth overall in the UK since the early 1990s, by almost two-fifths in men. Female rates have risen by around a third over the same period, because women began giving up roughly twenty-five years later than men and the disease follows exposure by two to three decades. The smokefree legislation of 2006–07 delivered its measurable early benefit in cardiac and childhood respiratory admissions, not in cancer, whose benefit is still arriving. The Tobacco and Vapes Act 2026 closes the arc on 1 January 2027.
For anyone reading this who currently smokes, the two most useful facts on the page are these. Four weeks of abstinence before an operation reduces complications, and stopping later than that does not make things worse. And stopping after a lung cancer diagnosis is associated with a median survival of 6.6 years against 4.8 years for those who continue — an effect comparable in size to treatments that occupy far more of the research literature, available at the moment of diagnosis, and costing nothing.
Does smoking cause lung cancer?
Yes. Smoking causes lung cancer. This has been established since the early 1950s and is not in scientific dispute. Doll and Hill published the first large British case-control study in 1950, comparing 1,465 patients with lung cancer against matched controls across 20 London hospitals, and followed it with the British Doctors Study from 1951 onwards. Smoking remains the single largest cause of lung cancer in the UK. Lung cancer also occurs in people who have never smoked, which is covered separately, but that does not qualify the main finding.
How many people smoked in Britain after the Second World War?
The highest recorded level of smoking among men in Great Britain was 82% in 1948, of whom 65% smoked manufactured cigarettes, with the remainder smoking pipes or cigars as well as or instead of cigarettes. Smoking prevalence among women in 1948 was 41%, and it remained broadly constant until the early 1970s, peaking at around 45% in the mid-1960s. By 2024 around 5.3 million adults in the UK, or 10.6%, smoked cigarettes.
Why has squamous cell lung cancer fallen while adenocarcinoma has risen?
The 2014 US Surgeon General’s report concluded that the evidence is sufficient to conclude that the increased risk of adenocarcinoma of the lung in smokers results from changes in the design and composition of cigarettes since the 1950s. Filters, ventilation holes and progressively lower tar yields altered how smoke is drawn into the lung, and the resulting tumours sit further out in the periphery rather than in the central airways. In United States data, adenocarcinoma overtook squamous cell carcinoma at around 1990 and now accounts for roughly 60% of non-small cell lung cancer. The direction of travel in the UK is the same. Changes in pathological classification account for part of the shift but not for all of it.
How was lung cancer found before CT scans?
The pre-CT workup was built for a tumour sitting in the central airways. A chest X-ray was taken in two views, and the lateral film often carried the finding. Bronchoscopy allowed the tumour to be seen directly. Haemoptysis was a common presenting symptom because there was an airway to bleed into. Mediastinal lymph node involvement was inferred indirectly, including from indentation of the oesophagus on a barium swallow, and confirmed by cervical mediastinoscopy performed by direct vision. A serum calcium was used as a surrogate for bone metastases. Every one of those tests works less well as the tumour moves peripherally, which is why CT from the 1980s, PET from the 1990s, and navigational bronchoscopy more recently became necessary.
Why is lung cancer falling in men but not in women?
The difference tracks when each sex began to give up smoking. Male prevalence fell continuously from its 1948 peak of 82%. Female prevalence was broadly flat from 1948 until the early 1970s. That is roughly a twenty-five-year offset, and with a two-to-three-decade lag between exposure and cancer it maps closely onto what is now seen. Cancer Research UK reports that since the early 1990s UK lung cancer incidence rates have fallen by about a tenth overall, with male rates down by almost two-fifths and female rates up by around a third. Over the last decade male rates fell by a seventh while female rates rose by a twentieth.
Did the smoking ban reduce lung cancer?
Not directly, and not on that timescale. Smoking was prohibited in enclosed public places in Scotland from 26 March 2006, in Wales from 2 April 2007, in Northern Ireland from 30 April 2007, and in England from 1 July 2007. The measurable near-term effects were cardiac and paediatric. Admissions for acute coronary syndrome in Scotland fell by 17% after the legislation, against 4% in England which had no such law at the time, with 67% of the reduction occurring in non-smokers. Childhood asthma admissions also fell. Lung cancer has a latency of decades, so the legislation acts on it indirectly, through denormalisation and falling prevalence, with the effect appearing twenty to thirty years downstream.
Does stopping smoking before an operation help?
Yes. At least four weeks of abstinence before surgery reduces respiratory complications, and three to four weeks reduces wound-healing complications. In lung resection specifically, a 2024 meta-analysis of 14 studies and 50,741 patients found that stopping more than a month before surgery reduced complications with an odds ratio of 0.72, while stopping between two weeks and a month did not. Importantly, stopping close to the date of surgery does not make things worse. The belief that late cessation increases postoperative respiratory risk is not supported by the evidence, and it should never be a reason to keep smoking before an operation.
Does stopping smoking after a lung cancer diagnosis make a difference?
Yes, and the size of the effect surprises most people. In a prospective cohort followed for an average of seven years, patients who stopped smoking after diagnosis had a median overall survival of 6.6 years against 4.8 years for those who continued. Five-year overall survival was 60.6% against 48.6%, and progression-free survival 54.4% against 43.8%. The benefit was comparable in lighter and heavier smokers and across earlier and later stage disease. A 2025 meta-analysis of 25 studies and 17,584 patients found a 26% lower risk of death, most marked in early-stage disease. Between a quarter and a half of patients are still smoking at the point of diagnosis, so this is a decision that remains available to a large number of people.
Mr Okiror sees private patients within 2–3 working days at London Bridge Hospital and The Lister Hospital Chelsea. NHS referrals through Guy’s and St Thomas’. Second opinion service available for patients already under another team’s care. Self-referrals welcome.
Request a consultation →Disclosures
This page describes the evidence on cigarette smoking and lung health as of July 2026. It is intended as patient and referrer information, not as medical advice for any individual case. Mr Lawrence Okiror is a Consultant Thoracic and Robotic Surgeon at Guy’s and St Thomas’ NHS Foundation Trust, with private practising privileges at London Bridge Hospital and The Lister Hospital Chelsea. He has previously received speaking fees from Pulmonx Corporation, manufacturer of Zephyr endobronchial valves, in relation to endobronchial valve therapy for persistent air leak; there is no current relationship. He has no commercial relationship with any manufacturer of tobacco, nicotine or vaping products. Historical accounts attributed in the text to Professor Peter Goldstraw are recorded as oral recollection and are not offered as citable source material. Decisions about investigation and treatment should always be made on a case-by-case basis after appropriate clinical evaluation.
The Stages I–IIIA pathway — screening, diagnosis, lung-sparing resection and perioperative therapy
Emphysema Surgery in 2026Endobronchial valves and lung volume reduction surgery for severe smoking-related emphysema
Lung Cancer in Never-SmokersThe molecular biology and treatment of lung cancer arising without a smoking history
The Lung Cancer Screening GapEligibility, symptoms, and why a normal chest X-ray is not a clear result
Lung Cancer Risk Without SmokingFor people worried about risk who have never smoked — including radon
Vaping and Lung HealthE-cigarettes, EVALI, and what is and is not yet known
Robotic SegmentectomyLung-sparing surgery for the peripheral tumours that now predominate
Lung Nodules & CancerWhat a nodule on a scan means and how it is investigated
Fitness for Lung SurgeryStructured reassessment for patients told elsewhere their lung function rules out surgery
Central Airway InterventionsTherapeutic bronchoscopy and stenting for the central airway disease of the older pattern
Specialist Second OpinionIndependent review for patients seen elsewhere — within 2–3 days