Document resource
WHAT IS ALREADY KNOWN ON THIS TOPIC Incidence rates for several cancers are rising among younger adults in England and many other countries. Whether changes in risk factor prevalence explain these trends is unclear and has not been systematically assessed using population-level data.WHAT THIS STUDY ADDS In England (2001–2019), nine of eleven cancers with established behavioural factors that are increasing in younger adults are also rising in older adults where the burden is greatest. Colorectal and ovarian cancers are exceptions. Rates for several cancers in younger adults have stabilised or declined in recent years. Except for body mass index (BMI), most behavioural risk factors show stable or declining trends. While BMI is a key contributor, it alone is unlikely to explain rising early-onset cancer rates.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY This study highlights the need to investigate novel and emerging exposures, while strengthening prevention strategies targeting known risk factors, particularly obesity, across all ages. There is also a need to assess the impact of changing diagnosis and screening practices on incidence trends.Introduction The incidence of cancer in young adults has been increasing for several cancer types in the UK and many other countries, raising questions on potential underlying causes and concerns about the public health implications. 1–5 In the USA, many obesity-related cancers that increased in younger adults between 1995 and 2014 were also rising in older adults, though often at a slower rate.3 6 A notable exception is colorectal cancer, for which incidence has declined in older adults in the USA and has shown stable or decreasing trends in several countries, including the UK.6 However, how incidence trends for other cancers in the UK compare between younger and older adults is less clear. These comparisons can provide valuable insights into the potential drivers of these increases, for example, whether younger adults are more susceptible to certain exposures or whether specific risk factors have emerged or intensified at a particular point in time, leading to cohort effects.1 3Multiple established and emerging risk factors have been proposed to explain rising cancer incidence in younger adults.1 2 These include established behavioural risk factors such as alcohol use, obesity, physical inactivity, red and processed meat consumption and low fibre intake. Additionally, there are growing concerns about suspected risk factors such as ultra-processed foods, sweetened beverages, environmental exposures (eg, air pollution, water contaminants, poly and perfluoroalkyl substances (PFAS), changes in sleep patterns (eg, light at night), and prenatal and early-life exposures (eg, antibiotic use during infancy)). Despite the extensive set of proposed risk factors for rising cancers in younger adults, there has been no formal evaluation of how temporal trends in these factors align with observed increases in cancer incidence in the UK.To address these gaps, we analysed cancer incidence trends in England from 2001 to 2019, comparing patterns in younger (age 20–49 years) and older (age ≥50 years) adults. We then focused on cancers with rising incidence in younger adults and established behavioural risk factors to assess whether exposure trends could explain increases in incidence. Finally, we evaluated the evidence for suspected risk factors by assessing available population data and theoretical calculations. Our aim was to assess the plausibility and potential impact of specific risk factors on rising cancer incidence, and to inform future research directions and prevention strategies.Methods A brief description of the methods used is given, with more details available in the online supplemental methods.SP110.1136/bmjonc-2025-000966.supp1Supplementary dataData sources Cancer incidence data Age-specific cancer incidence data were obtained from the National Disease Registration Service (NDRS) for England from 2001 to 2019 ( online supplemental table 1) stratified by age group (20–49, ≥50 years) and sex (men and women). Cancer incidence rates within the two age groups were age-standardised using the 2013 European Standard Population (ESP).7 International Classification of Diseases (ICD Version 10) codes were grouped into 25 total cancer sites, 22 in women and 21 in men, using groupings similar to Globocan8 (details in online supplemental materials).Behavioural risk factor data Established behavioural risk factors were defined as those classified as Group 1 carcinogens by the International Agency for Research on Cancer (IARC) or having strong evidence for cancer risk associations by the World Cancer Research Fund (WCRF) ( online supplemental table 2). These included cigarette smoking, elevated body mass index (BMI), alcohol consumption, red and processed meat consumption, low fibre intake and physical inactivity. Data on these risk factors were obtained from population-based surveys in England according to age and sex (online supplemental tables 1 and 3), and relative risks for cancer associations were updated from Brown et al.9 following a literature search (online supplemental table 4).Statistical analyses Cancer incidence and risk factor trends Temporal trends in age-standardised cancer incidence rates and risk factor prevalence in England were visualised using time series plots on a log scale to reflect relative changes over time. To quantify changes, we estimated average annual percentage change (AAPC; 2001–2019) with corresponding 95% CIs by age and sex groups for 25 cancer sites. 10 On a log scale, parallel lines of cancer incidence rates indicate similar AAPCs across age groups. The most recent APC was calculated for cancer sites with significant and positive AAPCs in younger adults, which also had established associations with behavioural risk factors (as defined earlier). Risk factor AAPCs were calculated over the timeframe with available data that varied for different risk factors. AAPCs were estimated by age, sex and index of multiple deprivation (IMD) groups.11 All AAPC and APC calculations were done using the Joinpoint regression software.12 To assess whether incidence rates in younger adults were increasing faster than in older adults, we conducted a one-sided test comparing the AAPCs between the two age groups. For all statistical tests, we used an alpha level of 0.05 to indicate statistically significant findings.Population attributable fractions We estimated the population attributable fraction (PAF), that is, the proportion of cancer cases attributable to a given risk factor in the population, for the cancer sites that met the criteria for APC calculation (defined earlier). PAFs were calculated by age group and sex according to the following formula:PAF=∑pn(RRn−1)1+∑pn(RRn−1)wherepn = population prevalence for the nth level of the risk factorRRn = relative risk for the nth level of the risk factor.PAFs aggregated across risk factors for each cancer site assumed independence between risk factors and were calculated by combining risk factor-specific PAFs, with each additional risk factor applied only to the remaining cancer cases not attributed to a previous exposure. To account for the expected delay between risk factor exposure and cancer diagnosis, a 10-year time lag was applied in estimating attributable cases.4 9 13 The 95% CIs for PAF estimates were estimated using Monte Carlo simulation methods.Incidence rates attributable and non-attributable to risk factors From the earliest available data to 2019, yearly PAFs were calculated for each cancer site and risk factor combination if the risk factor showed a statistically significant (p<0.05) and increasing AAPC in the 20–49 years age group for men or women. Risk factor-specific attributable and unattributable incidence rates were calculated across periods with available data, focusing on risk factors whose prevalence significantly increased in younger adults (AAPC>0 and p<0.05). For multilevel risk factors, trends were assessed using the prevalence of the highest exposure categories (ie, current smoking, moderate/heavy drinking and obesity). AAPCs were estimated for both serial datasets using Joinpoint software to calculate the relative change in attributable and non-attributable cancer incidence across periods with available data. 12Code and data accessibility All code and publicly available datasets used in this analysis are publicly available at https://doi.org/10.5281/zenodo.18789718. Individual-level data on risk factors can be obtained through an End User Licence Agreement with the UK Data Service (UKDS). Links to all used datasets are available in online supplemental table 1.Patient and public involvement Patients and members of the public were not directly involved in the design, conduct or reporting of this study. However, the national health surveys and cancer registration data utilised are developed and maintained with ongoing patient and public involvement, including advisory boards and consultations that inform data collection and governance. The research questions addressed also reflect priorities informed by public engagement in related studies on cancer risk and prevention. Future work will incorporate input from public and patient representatives to guide dissemination and further research on early-onset cancer.Results Cancer incidence trends The incidence for 16 of 22 cancers in younger women and 11 of 21 cancers in younger men increased significantly in England during the period 2001– 2019 (AAPC>0, p<0.05) ( online supplemental figure 1). Further analyses focused on cancers with significantly increasing trends in younger adults that also had established behavioural risk factors as defined by IARC (Group 1) or WCRF (strong evidence). Eleven cancers (three female-specific) met these criteria: thyroid, multiple myeloma, liver, kidney, gallbladder, colorectal, pancreatic, endometrial, oral, breast and ovarian cancers (online supplemental methods and figure 2).Figure 1 illustrates the temporal trends in age-standardised incidence rates for the 11 selected cancers by age groups and sex. Except for colorectal and ovarian cancers, cancers with significant increases in younger adults also increased significantly in older adults (ie, AAPC>0, p<0.05; online supplemental table 5). While colorectal cancer rates in older adults were stable (p≥0.05), ovarian rates were significantly decreasing by 1.31%/year in older women. For all other cancers increasing in both age groups, five of nine (endometrium, kidney, pancreatic, multiple myeloma and thyroid cancer) increased significantly faster in younger than older women, and one of seven (multiple myeloma) increased faster in younger than older men (online supplemental table 5). Examination of the most recent APCs revealed that several cancers that increased in younger adults over the 2001–2019 period had stable or declining trends in the most recent period, namely breast (APC2013-2019=0.21, p=0.62) and ovarian (APC2001-2019=–0.57, p=0.51) cancers in women, and kidney (APC2016-2019=–0.67, p=0.84), liver (APC2014-2019=–3.52, p=0.36) and oral (APC2014-2019=–1.07, p=0.06) cancers in men (online supplemental table 5).Figure 1Temporal trends in 11 selected cancer sites with incidence rates increasing in younger adults (20–49 years) in England (2001–2019), and with established behavioural risk factors, by age groups for women (A) and men (B). Plots have different y-scales and are sorted from highest to lowest incidence in younger adults. Straight lines represent the Joinpoint annual percent changes (APCs). Estimates of the most recent APC and average APC (AAPC) from 2001 to 2019 are shown in online supplemental table 5 .Behavioural risk factor trends We assessed trends for seven established behavioural risk factors associated with the eleven selected cancers increasing in younger adults ( figure 2, online supplemental table 6). All of these cancers (except oral) are obesity-related, while six are linked to smoking (liver, colorectal, oral, pancreas, kidney, ovary), four with alcohol (liver, colorectal, oral, breast), three with physical inactivity (colorectal, breast, endometrial) and one with dietary factors (colorectal). Of note, colorectal cancer is linked to all factors examined. Except for BMI, trends in these risk factors over the past one to two decades were stable or improving for younger adults. Cigarette smoking had a ~2%/year relative decrease in prevalence in younger men and women since 1995. In 2019, a lower percentage of younger women than men were current smokers (~20% vs 25%, respectively). Younger adult drinking trends were decreasing or stable across sex and exposure levels, except for light drinking in younger men. Overall, moderate and heavy drinking prevalence remained lower in women than men. Specifically, the prevalence of heavy alcohol drinking in younger men had a relative decline of 3.6%/year between 2011 and 2019 from a prevalence of 7% to 5%. Moderate drinking was more common (~20% of younger women, 30% of younger men in 2019), with levels also declining since 2011 by 3.0%/year for younger women and 2.2%/year for younger men. Similarly, physical inactivity decreased across sex groups from 2003 to 2012. By 2016, based on updated guidelines, levels of physical inactivity remained lower in younger men (20%) than women (35%). Although risk factor levels varied by age groups, the temporal trends were generally similar.Figure 2Temporal trends in established behavioural risk factors in England (1995–2019) for 11 selected cancer sites with raising incidence rates in younger adults, by age and sex groups. Straight lines represent the Joinpoint annual percent changes (APCs). Estimates of the most recent APC and average APC (AAPC) from earliest available year to 2019 are shown in online supplemental table 6. Risk factors: fibre intake deficiency (A), red and processed meat consumption (B), alcohol consumption (C), cigarette smoking (D), body mass index (BMI) (E) and physical inactivity (F). *150 min/week of moderate or 75 min/week of vigorous activity. **150 min/week of activity.Data on dietary consumption trends were available from 2008 to 2018 and showed sharp declines in the consumption of processed and red meat across all age and sex groups. The largest reductions were observed in red meat consumption, with relative annual decreases of 7.4% in younger men and 7.1% in younger women. This resulted in the median amount of red meat consumed per day for younger men decreasing from 38 g in 2008 to 17 g in 2018, and from 22 g to 10 g in younger women. Median processed meat consumption levels were consistently lower in younger women (~10 g in 2018) compared with men (~20 g in 2018). The percentage of younger adults with fibre intake deficiency was very high (>90% in 2018), but remained stable or declined across men (−0.4%/year) and women (−0.2%/year) from 2008 to 2018. Trends in younger adults were similar to those in older adults.In contrast, the prevalence of obesity has increased steadily since 1995 for all adults. The largest increases in obesity are for younger women, with about +2.6% per year relative increase since 1995. The lowest increases were for older women whose obesity rates have been only slightly increasing since 2009. By 2019, obesity prevalence was about 30% of older adults and younger women, and 23% of younger men. The prevalence of overweight has been stable or declining across all population groups, but in 2019 was still very high (29–34% in younger and older women and 40%–46% in younger and older men, respectively).To explore the impact of contextual factors, we evaluated temporal trends of established risk factors in adults by the IMD during the periods with data available. All ages were combined to avoid small sample sizes within IMD and age groups. We found decreasing trends for current smoking, physical inactivity and moderate/heavy alcohol consumption, and increasing trends of obesity across all deprivation groups (online supplemental figure 3). However, there were differences in risk factor prevalences and their rate of decrease. The clearest patterns were for smoking and obesity. Current smoking had a slower decrease in the most deprived compared with the least deprived groups. In contrast, obesity prevalence increased across all deprivation groups, with sharper increases and higher levels for the most deprived compared with the least deprived groups.PAFs and exposure-attributable incidence rates Aggregated 2019 PAFs for the seven established behavioural risk factors and BMI-specific PAFs are shown for each cancer by age groups in figure 3. Differences in PAFs by age were small with overlapping 95% CIs. While aggregated PAFs tended to be higher in men than women for most cancers, the BMI-specific PAFs tended to be higher in women than men. Differences in PAFs mainly reflected differences in exposure levels across age and sex groups, since relative risk (RRs) were assumed to be constant by age and sex, except for BMI and breast cancer risk (online supplemental table 3).Figure 3Aggregated (A) and body mass index (B) population-attributable fractions (PAFs) for behavioural risk factors in 10 cancer sites by age group and sex in England, 2019. Missing bars indicate estimates not applicable because of sex-specific cancers or lack of a BMI association with increased risk (i.e. breast cancer in younger women).In men, the cancers with the highest PAFs (>20%) in 2019 were oral (PAF=68–65% for younger and older adults, respectively), liver (42–48%), colorectal (49–53%), kidney (29–33%) and pancreatic cancers (36–34%). In women, the highest PAFs were for oral (52–45%), endometrial (35–42%), liver (44–46%), colorectal (38–42%), kidney (33–37%), pancreas (31–28%) and gallbladder (19%–24%) cancers. PAFs for specific risk factors across different cancer types are shown in online supplemental figure 4. BMI is the risk factor associated with most cancers (all except oral cancer), with 2019 PAFs ranging from 5% for ovarian to 37% for endometrial cancers.For cancers linked to BMI, the only risk factor showing significantly increasing trends, we evaluated cancer incidence attributable and non-attributable to increases in BMI from 2005 to 2019, by age and sex groups (figure 4, online supplemental figure 5 and table 7). Both BMI-attributable and BMI-non-attributable rates increased in younger adults, although BMI-attributable rates increased faster (figure 4, online supplemental table 7). For example, BMI-attributable colorectal cancer rates in younger women increased by 4.3%/year from 0.9 to 1.7 per 100 000, while BMI-non-attributable rates increased by 3.2%/year from 6.7 to 10.0 per 100 000. For endometrial cancer in younger women, BMI-attributable rates increased by 3.8%/year from 0.9 to 1.6 per 100 000, while BMI-non-attributable rates increased by 1.7%/year from 2.6 to 3.6 per 100 000. Similar patterns of BMI-attributable and BMI-non-attributable rates were seen in younger and older adults, except for colorectal cancer. BMI-attributable incidence rates for colorectal cancer increased significantly by 0.8%/year in older men and by 0.7%/year in older women, while BMI-non-attributable rates were stable with non-significant AAPCs (online supplemental figure 5 and table 7).Figure 4Cancer incidence rates in younger adults (20–49 years) in women (A) and men (B) for nine cancers in which higher body mass index (BMI) increases cancer risk, and partitioned into overall, BMI-attributable and BMI-non-attributable cancer incidence rates, England 2001–2019. See online supplemental figure 5 for similar figures for older adults and online supplemental table 7 for estimates of the most recent annual percentage change (APC) and the average annual percentage change (AAPC).Discussion Our analysis identified 11 cancers linked to behavioural risk factors with rising incidence in younger adults in England. For most cancers, incidence also increased in older adults at a similar pace, except colorectal and ovarian cancer, which increased only in younger adults, and some cancers that increased faster in younger than older adults. These patterns suggest that while similar risk factors across ages are likely, some cancers may have age-specific exposures, susceptibilities or differences in screening and detection practices.Common behavioural risk factors—smoking, alcohol, overweight and obesity, physical inactivity, red and processed meat consumption, and low fibre intake—could explain a substantial fraction of cancers in 2019 in England (eg, ~40–50% of colorectal, endometrium, oral or liver cancers). However, except for obesity, most factors showed stable or declining trends in the last 10-20 years. Increases in obesity at the population level may still occur even when some related behaviours, such as physical activity, are stable or improving, if other factors, such as overall diet quality, total calorie intake or sedentary time, are changing in an adverse direction. For the risk factors where we found no adverse trends over the periods with available data, their contribution to increasing cancer incidence is likely limited, unless their associations with cancer risk are underestimated, have longer lag periods or persistent effects after reductions in exposure (eg, smoking and lung cancer).14The observed increasing cancer incidence despite declining trends in several behavioural risk factors may reflect the net effect of multiple influences operating in different directions. Other contributing factors not evaluated here, for example, reproductive history, early-life or prenatal risk factors, and changes in cancer diagnosis and detection practices, may also play a role. Although incidence rates increased overall from 2001 to 2019, recent stabilising or declining trends for some cancers—for example, breast and ovarian in women, and kidney, liver and oral cancers in men—may reflect decreasing risk factor trends, for instance, declines in smoking and alcohol consumption influencing oral and liver cancer trends. Future cancer site-specific studies are needed to disentangle these complex patterns.Although overweight and obesity are linked to ten of the eleven cancers evaluated and account for a substantial proportion of cancer cases, both BMI-attributable and BMI-non-attributable incidence rates have increased—though the latter more slowly—suggesting other contributors. Given the high prevalence of overweight/obesity, reducing their prevalence could substantially lower incidence (eg, ~20–30% for endometrial, kidney, liver and gallbladder cancers in younger women, ~20% for kidney and liver cancer in younger men, and ~15% for colorectal cancers in younger men or women). The impact of BMI on incidence trends may be underestimated if the lag times exceed the typically assumed 10 years, or if the magnitude of associations estimated from epidemiological studies is substantially underestimated. Additionally, obesity-related conditions such as diabetes and metabolic syndrome are independently linked to several cancers and may also contribute,15 16 underscoring the importance of considering metabolic health more broadly.For a single risk factor to substantially impact changes in cancer incidence at the population level, it must be both common and strongly associated with risk, as illustrated in online supplemental figure 6. In addition, there must also be large shifts in risk factor prevalence over relevant periods of time. Several suspected risk factors—including ultra-processed foods, childhood obesity/physical inactivity, sedentary behaviour, antibiotic use, sweetened beverages and air pollution—have been proposed as contributors to rising cancer incidence in younger adults. While these factors are common in England, most have shown stable or declining trends in the last decade (table 1). Other emerging factors such as dysregulation of the gut microbiome and its potential interactions with diet, obesity and lifestyle factors in colorectal cancer risk17 warrant further study. For example, a somatic sequencing study identified a mutational signature in colorectal tumours linked to colibactin, a mutagen produced by some gut bacteria, which was common and enriched in early-onset cancers and may act early in life.18 Evaluating the role of these and other emerging factors in rising incidence trends will require sufficiently powered study designs and accurate exposure measurements over different time periods (eg, early life).Table 1Description of prevalence and temporal trends of exposures suggested to be implicated in rising cancer trends in younger adultsExposurePrevalence and temporal trendsUltra-processed food consumption23 24High levels in the UK (63% of total energy intake in 2019) with a slight decline to stable trends from 2008 to 2019. Higher in adolescents (66% of calorie intake), but slight declines over the past decade.Sweetened beverages25High levels but substantial drops in consumption from 2008 to 2016 both in adults and children. In 2016–2019, consumption was 106 g/day for adults aged 19–64 years and 142 g/day for children aged 11–18 years.Childhood obesity26Increasing in children aged 10–11 years from 17.5% prevalence in 2006–2007 to 22.7% in 2022–2023.Childhood physical inactivity26Stable or slightly declining levels in children aged 5–16 years from 30% boys/34% girls in 2017–2018 to 28% boys/31% girls in 2022–2023 being physically inactive.Antibiotic use27 28Declining prescriptions since 2006, with the steepest decline since 2013. Higher antibiotic use is associated with deprivation and geographical location. High usage in children aged 0–14 years (12% of total prescriptions in 2019–2020).Smoking and vaping in children (secondary school)26Smoking has decreased sharply in secondary school pupils from 15% boys/20% girls in 2001 to ∼3% in 2021. However, vaping doubled in girls from 2014 to 2021 (5–10% vaping) and has been stable at ∼7% in boys.Air pollution29Long-term decrease in estimated emissions of air pollutants (NO2, PM10, PM2.5, SO2, NMVOCs, ammonia) since 1990 in urban and rural areas of the UK.Poly and perfluoroalkyl substances (PFAS)30 31Ubiquitous chemicals with patterns of declining levels in the environment not yet clear despite regulations starting in the 2000s due to their persistence and historical usage. Levels in humans are high but have started to decline in North America and Europe (no data found for the UK).NMVOC, non-methane volatile organic compound; PM, particulate matter.Although cancer incidence is rising in younger adults, rates remain much higher in older adults for most cancers, with thyroid being the main exception (figure 1 and online supplemental figure 7). For most cancers, incidence is increasing across both age groups and PAFs are broadly similar, suggesting that common causes likely act across ages, thus making studies of populations with wide age ranges most informative. Colorectal cancer is a notable exception, with persistent increases limited to younger adults, pointing to causes unique to younger generations6 and the need for studies focused on early-life exposures. The NHS Bowel Cancer Screening Programme, introduced for older adults in the mid-2000s, likely contributed to declines in that group.18 However, it is unlikely to fully explain the differences by age groups given that rates have been different since the early 2000s. Other preventable cancers rising faster in younger adults—including endometrium, kidney, pancreatic, multiple myeloma and thyroid cancer—also suggest possible sex- or age-specific risk factors or differences in susceptibility that warrant investigation. Thus, while similar patterns in trends across age groups mean that studying older adults can provide insights into causes of cancer in younger populations, focused research on younger adults remains critical, particularly when incidence patterns point to unique generational or life course influences.Sex differences in incidence rates and PAFs likely reflect differing exposure prevalences in men and women that require monitoring and targeted action. Risk factors also differ by socioeconomic groups: although overall risk factor trends were similar across deprivation groups, the most deprived—defined by the IMD, which reflects income, employment, education, housing, health, crime and environment—experienced slower declines in smoking, physical inactivity and alcohol use, alongside faster rises in obesity and consistently higher exposure prevalence than the least deprived. Together, these sex and socioeconomic disparities contribute to unequal cancer burdens and underscore the need to account for both demographic and contextual factors when developing strategies to reduce health inequities.A limitation of our analyses is the lack of consistent, long-term national data for several risk factors. Physical inactivity data, for example, were available only for five timepoints between 2003 and 2012, when the UK guidelines’ definition of inactivity changed substantially. This restricted our ability to evaluate long-term exposure trends beyond one to two decades, depending on the risk factor. Our analysis focused on England because of its size and the availability of survey data. Future analyses are needed for other geographical regions in the UK. Another limitation is the assumption of a 10-year lag period between exposure and cancer incidence. While this might be reasonable for many cancers, we were unable to evaluate longer lag times due to limited exposure data. This limitation also applies to other studies that have suggested that these risk factors could explain increases in cancer incidence without formal evaluation. Our analysis makes the issue more explicit. online supplemental figure 8 illustrates this using an example of early-onset colorectal cancer showing available data on incidence trends and risk factor prevalence with a 10-year time lag.Self-reported surveys are prone to measurement error in exposures, such as underreporting alcohol or overreporting physical activity with rising health awareness, and these errors could differ for younger and older populations. However, the use of standardised, validated questionnaires in national surveys minimises differential bias across years. Using exposure surrogates (eg, BMI for adiposity) or broad categories like ultra-processed foods could underestimate the impact of underlying exposures. More precise measures, such as organ-specific fat deposition or finer food classifications, could improve risk estimation and assessment of the impact of risk factors on incidence trends. The PAF calculations assumed constant RRs across age and sex groups, except for obesity and breast cancer, for which the well-established inverse association in younger women and direct association in older women19 were modelled separately. This approach reflects the evidence sources (IARC and WCRF) which do not generally explicitly address effect modification by age or sex. However, if such differences exist for other risk factors, our approach may have masked differences in PAF across population subgroups. Additionally, while we evaluated the independent impact of each risk factor on cancer incidence, thus avoiding double counting in aggregated PAF estimates, multiple weakly associated exposures acting together could create cumulative effects that are difficult to quantify.Changes in detection and diagnostic practices such as population-wide cancer screening programmes, increased use of diagnostic tests or advanced imaging technologies, and broadening of disease definitions could also have played a role in the observed cancer incidence trends.20 21 For instance, in the UK, there is evidence for changes in practice and possible overdiagnosis of oral, endometrial, thyroid, kidney and breast cancers, which were found to be increasing.22 Future analyses of incidence trends by stage at diagnosis or accounting for overdiagnosis could help clarify the impact of surveillance and screening.A strength of this report is the systematic evaluation of both cancer incidence and risk factor trends in younger and older adults, using a formal approach to estimate the contribution of exposures changes to cancer incidence. While analyses are descriptive in nature and preclude causal inference, these ecological comparisons provide valuable context for assessing whether previously proposed explanations for rising incidence trends are consistent with population-level patterns. Although not exhaustive, our examples provide useful benchmarks for assessing whether a given risk factor could plausibly explain a substantial proportion of cancer cases.In conclusion, the incidence of several cancers is increasing in both younger and older adults in England. Apart from BMI, trends in established behavioural risk factors are unlikely to be substantial contributors to the increase in early-onset cancers at the population level, with incidence rates increasing despite favourable trends in several known risk factors. Illustrative scenario analyses indicate that changes in prevalence of a magnitude greater than those observed, or stronger risk associations, would be required for individual behavioural factors to explain the magnitude of the observed increase, suggesting that the combined effects of multiple factors are more likely to explain these trends. These findings underscore the potential influence of obesity and related factors on increases in cancer in younger and older adults. Further research into emerging and interacting exposures, improved measurement and continued surveillance are needed, alongside evaluation of the impact of screening and shifts in stage at diagnosis. Finally, although increases in cancer in younger adults are concerning, the absolute burden remains far higher in older adults, underscoring the public health and clinical importance of studying risk factors across all ages.