25 year trends in cancer incidence and mortality among adults aged 35-69 years in the UK, 1993-2018: retrospective secondary analysis

Principal findings

The most striking finding in this analysis of UK cancer trends among the 35-69 years age group is the substantial decline in cancer mortality rates observed in both sexes (37% decline in men and 33% decline in women) across the period examined. A decrease in mortality was reported across nearly all the specific types of cancer examined (23 in total), with only liver, oral, and uterine cancers showing an increase together with melanoma skin cancer in men and pancreatic cancer in women, both showing small increases. By contrast, the incidence trends in this age group showed varying patterns with some sites increasing, some decreasing and some remaining relatively constant. Over all sites, a modest increase was noted in cancer incidence rates of around 0.8% per annum in both sexes, amounting to an increase of 15% in men and 16% in women over the 25 year time frame.

The increase in prostate cancer incidence over this period, especially in the 35-69 years age group considered here, is very likely to be a direct result of the uptake of prostate specific antigen testing, which results in the detection of early stage disease and, to an unknown extent, indolent disease that may otherwise never have been regarded as clinically significant.2425 The results do, however, affect people diagnosed and represent a large increase in workload for clinical staff. The fact that the overall mortality trends for men show no difference whether prostate cancer is included or excluded in the analysis indicates that the incidence increase for this cancer has largely been of non-fatal disease. That the specific mortality rates for prostate cancer showed an appreciable rate of decline during this time (–1.8% per annum) also indicates improved clinical treatment of the disease or an increase in the proportion of men diagnosed with a favourable prognosis, or both.2426 However, the increase in prostate cancer incidence still results in thousands of men each year dealing with the concerns of a cancer diagnosis and the impact this may have on their lives.

Breast cancer comprehensively dominated incidence and mortality trends in female cancer. Even though the average annual incidence increase of breast cancer over this period (0.9%) was modest in comparison to the prostate cancer increase in men (4.2%), breast cancer incidence rates remained substantially higher than those for any other cancer site in either sex. Inspection of figure 4 shows that breast cancer incidence rates (age standardised) increased at a faster rate until around 2003-05 (from 194.7 in 1993-95 to 229.9 in 2003-05), a slower rate from then until 2013-15 (240.8) but have levelled off in the most recent years analysed (238.0 in 2016-18). These changes in the incidence trend likely reflect a reduced effect of the initial incidence increases brought about by mammography screening in the UK introduced from the late 1980s or a possible effect of a decline in usage of hormone replacement treatment.2728 However, the effect of hormone replacement treatment on breast cancer risk is small in comparison to other risk factors,7 and trends in this treatment has varied over time, such as changes in preferred formulations, doses, and treatment durations,293031 which may impact breast cancer risk levels.3233 As has been reported elsewhere,343536 mortality for breast cancer has declined substantially despite the incidence increase, which is indicative of improvements in early detection (including through screening37) and improved treatment.

The other two major sites of cancer in men apart from prostate cancer, namely lung and bowel cancers, showed substantial reductions in mortality. These results are likely from primary prevention (historical reduction in smoking rates)38394041 for lung cancer and earlier detection (including screening) and improved treatment for bowel cancer.424344 While lung cancer incidence substantially decreased, the incidence rates of bowel cancer remained unchanged. However, closer inspection of the bowel cancer incidence trends over the full period shows an increase from the point the bowel screening programme was first introduced from 2006 in the UK. This rate, however, has now decreased back to the observed level prior to the introduction of the screening programme. As others have shown, the introduction of bowel screening leads to an initial short-term increase in cancer incidence due to detection of as-yet undiagnosed cancer cases, followed by a decrease because of removal of adenomas.424546 Therefore, bowel cancer incidence trends can reasonably be assumed to decrease further over the coming years, unless other preventable risk factors for bowel cancer affect the trend.

Similarly, lung and bowel were the other two major cancer sites for women (alongside breast cancer), and both showed reductions in mortality. The decline in lung cancer mortality was, however, not as extensive as that for men (–0.5% compared with –3.1% per annum) likely reflecting the different demographic pattern in smoking rates that led to peak smoking prevalence in women occurring around 30 years later than men, albeit at around half the peak prevalence observed in men.4047 Smoking prevalence in women has always been lower than in men.3948 The lung cancer incidence trends showed a significant increase in women of 0.8% per annum as opposed to the –2.1% per annum decrease in men. That the incidence rate in 2016-18 was still higher in men than in women again is almost certainly a reflection of historical differences in smoking patterns.394950 Bowel cancer incidence in women followed a similar pattern to men and is equally reflective of the introduction of the bowel screening programme. Bowel cancer mortality in women has declined at a similar rate to men (–2.2% compared with –2.5% per annum), indicative of the same improvements in early detection and improved treatment.

These reductions in mortality across the most common cancers in both sexes are likely a representation of considerable success in cancer prevention, diagnosis, and treatment. Further improvements are likely to be realised from the continued reduction in smoking prevalence, of which smoking prevention policies continue to contribute,51 alongside the recent move to faecal immunochemical testing in the bowel screening programme adopted throughout the UK during 2019.52 The recommended rollout of targeted lung screening is expected to further help with the earlier diagnosis of lung cancer where surgery is a viable treatment option and outcomes are vastly improved.5354

Although four major sites influenced the overall pattern of cancer incidence and mortality, increases in rates among some of the less common sites do raise concerns. Four cancers showed substantial increases in incidence (more than 2% per annum) in both sexes: liver, melanoma skin, oral, and kidney cancers. All have strong associations with established risk factors: alcohol consumption, smoking, and HPV for oral cancer;75556 overweight and obesity, smoking, alcohol, and hepatitis B and C for liver cancer;75758 ultraviolet light for melanoma;5960 and obesity and smoking for kidney cancer.616263 Increases in liver cancer incidence and mortality for both men and women are very concerning, with nearly one in two attributable to modifiable risk factors.7 With high prevalence of overweight and obesity and diabetes in the general population, other studies expect the rates to remain high.64 For oral and kidney cancer, despite the association with smoking, incidence rates have not followed the decrease seen for lung cancer incidence in men. This is likely to be due to the smaller proportion of cases attributable to smoking in these two sites. Whilst smoking accounts for around 17% of oral cancers, over one in three are attributed to alcohol consumption.7 For kidney cancer, smoking is attributable to 13% of cases whereas obesity causes around 25%, however, increasing trends in kidney mortality are shown for this age group and period.7 Therefore, the increasing incidence trends could potentially have been worse, especially in men, if the reduction in smoking prevalence had not occurred. The increased incidence of melanoma skin cancer is likely to be caused by the increased sunlight and ultraviolet exposure caused by the availability of cheaper air travel to countries with a warmer climate and insufficient regulation of tanning beds until 2010.6566

In women, uterine cancer incidence increased by 1.9% per annum; although, this increase was predominantly seen over the period 1993-2007 and since then incidence trends have increased at a slower rate. One of the main risk factors for uterine cancer is the use of oestrogen-based hormone replacement therapy,6768 and since around 2000, use has substantially declined.27 Around a third of uterine cancers in the UK are also attributed to overweight and obesity, but the increase in incidence is also likely to be caused by a decrease in the number of women undergoing hysterectomies for menorrhagia, in favour of endometrial ablation.69

Other cancers that showed increases in incidence were cancers of the pancreas, brain, and central nervous system, together with Hodgkin and non-Hodgkin lymphoma, myeloma, and leukaemia in both sexes, and oesophageal and testicular cancers in men. With the exception of pancreatic cancer, which only decreased in women, all these cancers also showed a reduction in mortality in both sexes, indicating improving treatment or earlier detection, or both. Generally, the causes of these cancers are not well understood although obesity is associated with the adenocarcinoma histological subtype of oesophageal cancer,70 especially in men,7 while a combination of smoking and alcohol is implicated in the squamous cell carcinoma subtype.71 The considerable male excess in oesophageal adenocarcinoma in comparison with squamous cell carcinoma rates,72 possibly underlined by the higher incidence of gastroesophageal reflux disease in men 73 and the protective effect of oestrogen,7475 may explain the differing trends now observed between men and women.

Several cancer sites showed decreases in both incidence and mortality rates over the time period, notably stomach, larynx, and bladder cancer in both sexes, as well as cervical and ovarian cancers in women and mesothelioma in men. The changes in stomach cancer rates were of a similar magnitude and represented the largest percentage mortality decline in both sexes. This decline can probably be attributed to a combination of a reduction in the prevalence of Helicobacter pylori infection and an increase over time in fruit and vegetable consumption reducing the dependency on preserved foods.7677 Challenges in coding of stomach and oesophageal cancer before 2000 may also have had a role in shaping these trends. Laryngeal cancer is associated with tobacco use and alcohol consumption as well as occupational exposures,567879 and the decline in rates is most likely to be related to the decrease in smoking prevalence as well as decreases in occupational exposure.80 The refinement of understanding pathology for bladder cancer during this period, in which previously diagnosed malignant disease is now categorised as benign,81 is likely to have resulted in an artificial decline in incidence rates.8283 This artefact should not, however, have affected the decline in mortality rates given the benign nature of these tumours that do not cause death.81 This decline in mortality, although not as marked as that for incidence, remained appreciable. The changes in cervical cancer rates, which showed the largest percentage mortality decline amongst gynaecological cancers, are almost certainly attributed to the success of the cytological screening programme during the whole of the time period considered.8485 With the introduction of the HPV vaccination programme for girls in 200886 and the subsequent expansion to boys in 2019,87 rates of cervical cancer are expected to fall substantially over the following decades as the first cohort of vaccinated women reaches the peak age for cervical cancer incidence (aged 30-34 years). A reduction has already been shown for women aged 20-24.88 The absolute incidence rates of mesothelioma in women were small in magnitude in 1993-95 (0.8 per 100 000 per annum) and remained similar over time (0.7 per 100 000 per annum in 2016-18). The incidence rates of mesothelioma in men were considerably greater, especially in 1993-95 (around 6.3 per 100 000 per annum), due largely to occupational asbestos exposure,89 but a significant decrease was noted over time (to 3.6 per 100 000 per annum in 2016-18) resulting from both the decline in asbestos exposure and the decline in heavy industries, such as coal mining. Mortality decreased substantially in both sexes over the period for which data are available (2001-03 to 2016-18).

The conclusions that can be drawn from this analysis are, overall, positive and reassuring. Within the 35-69 year age group, cancer mortality rates have shown a substantial overall decline during the last quarter of a century in both men and women. The most probable causes are a combination of changes in the underlying risk of disease for some cancers (notably lung and stomach), in increased levels of early detection (notably breast37 and cervix90) and improved treatment (notably breast and bowel) for others. The specific circumstances leading to the increased incidence of breast cancer, of which risk factors are complex, need to be better understood and controlled. Similar results have been shown for incidence within Great Britain and mortality in the UK for some cancer sites.91 Speculated overdiagnosis, where tumours are detected that would not have caused the patient any harm during their lifetimes, has been thought to increase rates for breast and prostate cancers in particular, of which prostate is especially affected by the widespread use of prostate specific antigen testing.492 However, given the decreases in mortality across the wide set of cancer sites analysed here, improvements in early diagnosis, treatment, or both are having a positive effect for most cancer patients, although cancer mortality in this age group still needs reducing.

After accounting for the major two sites in men and women, the increase in overall incidence rates disappeared in men while it remained significant in women. This difference between sexes is due to a decrease in cancers with substantially higher initial incidence rates in men, such as lung, stomach, and bladder, resulting in a higher overall impact on male incidence, combined with an increase in incidence in uterine cancer, one of the most common cancers in women.

Strengths and limitations

This study benefits from high quality cancer registry data collected by all four cancer registries in each country across the UK, which allows for the inspection of a wide range of cancer sites over 25 years. ICD-10 coding changes have been minimal, only affecting trends in cancer incidence for bladder and ovarian cancers and cancer mortality for mesothelioma, whereas challenges in coding stomach and oesophageal cancer may have affected trends for these sites. Changes in registration practice may well have had a small effect on certain cancer sites. By focusing only on the 35-69 age range, we present a clear and reliable comparative picture of cancer incidence across 25 years within the UK, which provides a reliable indicator regarding future cancer incidence trends. Understanding cancer in older people and changes in the trends of different cancers is also of interest, but subject to a different study given the increasing life expectancy over this period, impact of comorbidities, and differing interaction with health services in this age group.

Limitations include the absence of staging data to substantiate any improvements in earlier diagnosis. Due to the number of sites analysed, we also have not broken down sites by histological type, which could be beneficial in certain sites to understand the trends within cancer sites—eg, small cell and non-small cell lung cancer or oestrogen receptor-positive and oestrogen receptor-negative breast cancer. In focusing on the age group selected, we are excluding older ages where rates of cancer are higher. Although this exclusion reduces the number of cases included, providing a smaller cohort for each year, the age group selected provides a more reliable comparator for future trends given the accuracy of incidence recording and also focuses on the cancers that lead to a larger number of years of life lost. The age range included in this study has been well defined; however, other studies are indicating potentially different trends worldwide in young adults with potential increases in risk factors such as dietary risk factors playing a role.9394 The data captured across the UK registries provides a basis for further understanding to see whether different trends are observed across younger age groups and whether the causes of this can be determined. Additionally, although we have included a broad range of cancer sites, cancers that have not been included in this study could well be showing different trends, such as a more recent increase in thyroid cancer in the UK.95

This study also provides a baseline covering a 25 year period uninterrupted by covid-19. Trends in cancer incidence and mortality beyond these years will be affected and therefore understanding the causes of trends will be more complicated. Having a 25 year baseline provides the observed trend for which expected cases can be assessed against observed. This benchmark will present a comparison for the following decade as the presentation, diagnosis, and treatment of cancer have been hugely affected by rules and regulations affecting public and health service staff. Mortality trends will also be impacted with decision making regarding coding of deaths with covid-19 likely to be the underlying cause of death for people with cancer if that has directly led to the patient dying, rather than their cancer.

This study focuses on the overall sex specific trends for cancer incidence and mortality in the specified age group to observe and understand trends over the 25 year period across the entire UK. Further breakdowns have not been possible. Paucity of numbers for less common cancers precluded separate analyses for the individual UK nations while data limitations precluded analyses by other demographic characteristics, for example, ethnic group and deprivation. The main obstacle to analysing data by ethnic group is the completeness of recordings in hospitals. In England, completeness improved substantially in 2012, but prior to this, the proportion of cases with unknown ethnic group renders results over time to be incomparable. In other UK countries, completeness of ethnic group recording is still not good enough to conduct country-wide cancer incidence or mortality analyses by ethnicity. For deprivation, the measures currently available are derived within each UK nation, and a specific validated UK-wide deprivation measure does not yet exist. Given the obvious importance of looking at variation in UK trends within ethnic groups and deprivation categories, such analyses represent a priority for further research and highlights the importance of data collection across all UK nations.

Conclusions

Overall, these results substantiate the view that in this age group there is no generalised increase in cancer incidence, while there is a substantial decrease in cancer mortality in the UK over the 25 year study period. Specific concerns about individual cancer sites identified were raised, of which the most important numerically, apart from the increases in breast and prostate cancer incidence, was the need to accelerate the decrease in female lung cancer. After which, concerns about oral cancer, liver cancer, kidney cancer, uterine cancer, and melanoma skin cancer present the most pressing issues. There are also several cancer sites that showed decreases in both incidence and mortality, notably, stomach, larynx, bladder, and cervical.