Tobacco smoke carries more than seventy known carcinogens, and the damage it inflicts accumulates over decades. Epidemiology compresses that accumulation into pack-years, a quantity that multiplies cigarettes smoked per day by years of smoking and divides by twenty. Convenience of the measure conceals a structural problem: raising daily consumption while holding duration fixed necessarily raises the total dose, so a conventional log-linear model cannot separate what intensity does from what time does. Earlier attempts to prise the two apart leaned on case-control series, where smoking histories are reconstructed from memory years after diagnosis and are therefore vulnerable to recall bias. Prospective cohorts, which record exposure before disease appears, offer a firmer footing for asking whether a dose delivered quickly carries the same hazard as the identical dose delivered slowly.
Answers came from the UK Biobank. Of 502,505 participants, 491,062 with complete and internally consistent smoking records were retained. Twelve cancers identified by the US Centers for Disease Control and Prevention as causally linked to smoking were ascertained through linked national cancer registries, with person-time accrued from enrolment until diagnosis, death or the end of follow-up. Exposure was not entered as a single pack-year term. Instead, the analysts modelled ERR per pack-year and permitted smoking intensity to modify that slope, which makes low-intensity, long-duration histories directly comparable with high-intensity, short-duration histories at matched cumulative exposure. Time since cessation entered the same framework. Physical activity, diet quality and other lifestyle variables were then tested as modifiers of the exposure-response relationship, and the models were adjusted for age, sex, body mass index, alcohol use, education and other established risk factors.
Risk gradients refused to collapse onto a single pack-year scale. At equal cumulative exposure, participants who had smoked fewer cigarettes across more years and those who had smoked heavily across fewer years did not carry identical excess risk per pack-year, and the discrepancy was site-specific rather than uniform. Cessation shifted the slope: longer intervals since quitting were accompanied by lower ERR per pack-year for most of the twelve cancers, consistent with a gradual relaxation of carcinogenic drive once exposure stops. Lifestyle left its own signature. Regular physical activity and healthier dietary patterns were associated with smaller excess risks per pack-year than sedentary behaviour or poorer-quality diets, which suggests that the carcinogenic pressure of a given smoking history is partly modifiable even after the smoking itself has been quantified. The cohort also showed how much a single summary number can hide, since current smokers who developed cancer were overwhelmingly people who had smoked more than ten cigarettes a day, and former smokers who developed cancer were mostly people who had smoked for decades before quitting.
Consequences for practice follow directly. Screening programmes that admit participants on the basis of pack-years alone inevitably mis-rank people whose exposure was brief but intense against those whose exposure was light but lifelong, and the present analysis supplies a quantitative basis for correcting that ranking. Risk communication can likewise move beyond a single number: telling a patient that twenty pack-years accumulated over ten years differs materially from twenty pack-years accumulated over forty years is more informative than the aggregate alone. Prevention planning gains a similar lever, since the lifestyle modifiers identified here are actionable in populations that cannot undo their smoking history. Limitations remain. Smoking was self-reported at baseline rather than modelled as a time-varying exposure, follow-up in the cohort is still modest relative to the latency of solid tumours, and participants are healthier and older than the general population. Even so, the demonstration that intensity, duration and cessation history each alter excess risk per pack-year reframes a familiar metric and argues for rebuilding pan-cancer risk assessment around the full shape of exposure.
DOI
10.1007/s11684-026-1244-9
Regions: Asia, China, Europe, United Kingdom, North America, United States
Keywords: Science, Life Sciences