Cancer prevalence in the UK: results from the EUROPREVAL study

D. Forman1,2,+, D. Stockton3, H. Møller4,5, M. Quinn6, P. Babb6, R. De Angelis7 and A. Micheli8

1 Northern and Yorkshire Cancer Registry and Information Service, Leeds; 2 University of Leeds, Leeds; 3 Scottish Cancer Intelligence Unit, Information and Statistics Division, National Health Service Scotland, Edinburgh; 4 Thames Cancer Registry, King’s College, London; 5 London School of Hygiene and Tropical Medicine, London; 6 National Cancer Intelligence Centre, Office for National Statistics, London, UK; 7 Istituto Superiore di Sanità, Rome; 8 Istituto Nazionale per lo Studio e la Cura dei Tumori, Milan, Italy

Received 26 August 2002; revised 25 November 2002; accepted 13 January 2003


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Background:

Cancer incidence, mortality and survival statistics for the UK are routinely available; however, data on prevalence, which is generally regarded as an important measure for health planning and resource allocation, are relatively scarce.

Materials and methods:

Eight cancer registries in the UK, covering more than half the population, provided data based on >1.5 million cases of cancer. Total prevalence was calculated using methods developed for the EUROPREVAL study, based on modelling incidence and survival trends. The prevalence of cancers of the stomach, colon, rectum, lung, breast (in females), cervix uteri, corpus uteri and prostate, melanoma of skin, Hodgkin’s disease, leukaemia and all malignant neoplasms combined, was estimated for the UK for the end of 1992.

Results:

Overall, ~1.5% of males and 2.5% of females in the UK population at the end of 1992 were living with a diagnosis of cancer. These proportions increased steeply with age, with ~7.5% (7.3% and 7.8%, in males and females, respectively) of people aged >=65 years living with a diagnosis of cancer. Of the individual cancers, by far the highest prevalence (almost 1%) was seen for breast cancer in females; more than one in three of all living female cancer patients had been diagnosed with breast cancer. For males, around half of prevalent cases had been diagnosed >5 years previously and 30% >10 years previously; for females, these figures were both higher, at ~60% and 40%, respectively.

Conclusions:

The estimates of prevalence presented here comprise: recently diagnosed patients in need of treatment and monitoring; long-term survivors, some of whom will nevertheless eventually die from the cancer, while others may be cured of the disease; and patients in the terminal phase who are dying from the cancer. Further work should attempt to identify the proportions of patients in the different phases of care in order to optimise the use of prevalence estimates in health care planning.

Key words: cancer, cancer registry, prevalence, UK


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The prevalence of cancer in a population—the number or proportion of people alive at a specific date who have been diagnosed with cancer prior to that date—is generally regarded as an important measure for health planning and resource allocation purposes. While reliable cancer incidence, mortality and survival statistics for the UK are now routinely available from cancer registries, the Office for National Statistics (ONS) and the Scottish Cancer Intelligence Unit [13], prevalence data are relatively scarce. This is largely a result of methodological difficulties involved in the estimation of prevalence. To provide accurate figures, it is necessary to have either incidence data for very long time periods together with reliable follow-up for death, or incidence data combined with good estimates of survival for the specific cancer sites (or types).

Three- and ten-year cancer prevalence figures have recently been published for England and Wales for patients alive in January 1993 and diagnosed in 1990–1992 and 1983–1992, respectively [1], using incidence data and follow-up for death or emigration. This is the directly observed 3- or 10-year prevalence. Total prevalence is all those cases alive at a given date, i.e. those directly observed by a cancer registry plus those that were diagnosed before the registry started. The methodology developed by the collaborative EUROPREVAL study [4] provided an opportunity to estimate the total UK prevalence to the same follow-up date (the end of 1992) for 11 sites of cancer and for all malignant cancers combined. This enabled the estimation of the total burden of cancer within the UK by sex, age group and time since diagnosis that is presented in this paper. The EUROPREVAL estimates were validated by comparing them with the directly observed figures for England and Wales.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Data
Eight cancer registries in the UK (seven regional registries from England together with the Scottish registry) provided the necessary incidence and survival data needed to estimate prevalence. The analysis of cancer prevalence in the UK was based on >1.5 million cases of cancer that occurred in the population of >30 million people in the areas covered by the eight registries (Table 1). All the registries provided data for which a high level of completeness could be assured and follow-up for vital status was complete until 31 December 1992 (the index date for the estimation of prevalence). The cases shown in Table 1 includes patients who died before the index date (and are, therefore, not included in the prevalence estimate).


View this table:
[in this window]
[in a new window]
 
Table 1. Population covered, period of diagnosis and numbers of cases by cancer, and percentage of total cancers lost to follow-up by cancer registry
 
Cases notified only through death certification and those diagnosed at autopsy were not included in the analyses. When more than one cancer was diagnosed in a patient, only the first diagnosis was included. These and further exclusions are discussed elsewhere [5]. For cases with missing data (on average, no more than 1% of the total number of cases) some imputation procedures were adopted [5].

The data from the seven geographically widespread registries in England related to 53% of the English population; this population had a very similar age distribution to that of England as a whole. The Scottish Cancer Registry covered the whole of the Scottish population. In order to obtain UK figures, we assumed first that the cancer prevalence in the English registries covered by EUROPREVAL was representative of England and Wales as a whole. For Northern Ireland, the prevalence estimates of cancer in Scotland were applied to the population estimates for Northern Ireland, which contains only ~3% of the total UK population. The data were analysed in three age groups (0–44, 45–64 and >=65 years) to enable adjustment of the overall prevalence estimates for any differences in age structure between populations.

Calculation of observed prevalence
Observed prevalence has been computed on the basis of the longest period of data available from each single registry, by means of the counting method implemented by PREVAL software [6]. Observed prevalence was then corrected to take account of the (small) contribution of cases lost to follow-up, by assigning to them a survival probability equal to that of patients successfully followed up of the same sex and with similar age and period of diagnosis [5, 6].

Calculation of total prevalence
Total prevalence figures for each cancer site were derived by applying an estimated correction factor, the completeness index (R) [7], defined as the proportion of the total prevalence represented by the observed prevalence. This varies according to the length of the registration period and the characteristics of the cancer being considered (mainly incidence and survival trends). Total prevalence derived by the completeness index method was recently validated using data from the Connecticut Cancer Registry for a group of different cancer sites [8].

The calculation of the completeness index is based on modelling incidence and survival trends with simple parametric functions [4, 9]. A single cancer-specific incidence trend was assumed for fitting data from the UK registries, while for the survival models, registry-specific relative risks were estimated. The periods of time covered by the EUROCARE data, used for estimation of survival trends, were not adequate to correctly estimate the required incidence and survival trends by age for cancer of the cervix uteri and for Hodgkin’s disease [4]; consequently, R values and hence the total prevalence could not be produced for these two malignancies, although the method allows an adjustment of the observed prevalence that furnished estimates of the 15-year prevalence.

The completeness indices that were used in the subsequent calculations of prevalence are shown in Table 2 for the different populations and types of cancer. They were generally higher for males than for females, e.g. for all cancers the indices were ~0.8 and ~0.7, respectively.


View this table:
[in this window]
[in a new window]
 
Table 2. Observation time and completeness index by cancer site (with ICD9 code), sex and cancer registry, all ages
 
Validation
The cancer registration system in England and Wales incorporates two methods (in addition to information obtained locally by the regional cancer registries) by which the National Cancer Intelligence Centre (NCIC) at the ONS notifies deaths (or emigration) to the registries. Fuller details of the system have been published elsewhere [1]. Direct estimates of the prevalence of all malignant cancers (excluding non-melanoma skin cancer), and of 20 major cancers individually, were made at the NCIC by checking the vital status of all cases of cancer diagnosed in residents of England and Wales from 1971 (when the current system of follow-up began) to the end of 1992. For those patients who were not lost to follow up, the percentage who were still alive on 1st January 1993 was applied to the total number of cases to give the estimated number of prevalent cases.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Overall, ~1.5% of males and ~2.5% of females in the UK population at the end of 1992 were living with a diagnosis of cancer. The estimated total prevalence rates per 100 000 of the population by cancer site, sex and age group are shown in Table 3. Prevalence was higher in Scotland than in England, in both sexes, for cancers of the colon and lung and for melanoma of the skin. Prevalence was higher in England for cancers of the rectum and prostate in males and cancers of the breast and uterus in females.


View this table:
[in this window]
[in a new window]
 
Table 3. Total prevalence by country, age, sex and cancer site (proportions per 100 000 population)
 
Cancer prevalence increased steeply with age, reaching values of 7.3% and 7.8% in males and females, respectively, in the >=65 years age group. In this group, the most prevalent cancers were those of the prostate, lung, colon and rectum in males, and of the breast, colon and uterus in females. The pattern was similar among the middle aged population (45–64 years), although the absolute prevalence estimates were substantially lower, especially for prostate cancer in males. For female breast cancer in this age group, prevalence estimates were lower than for patients aged >=65 years, but not as markedly as those for most other specific sites. The age related impact of breast cancer prevalence affected the corresponding all cancer prevalence estimates for ages 45–64 years, which were 1.7% in males and 3.8% in females. Prevalence for cervix cancer in females and for Hodgkin’s disease in males reached a maximum in this age group. In young persons (aged <45 years), the pattern of prevalence was somewhat different from that seen at older ages in males, being highest for haematological cancers and melanoma of skin. Although prevalence estimates for these cancers were also relatively high in young females, the highest estimates, as in middle aged females, were for breast and cervix cancers. The prevalence of Hodgkin’s disease in females reached a maximum in the youngest age group. Overall, the prevalence of breast cancer in females approached 1%, over four times the rate for prostate cancer, the most prevalent cancer in males.

The numbers of prevalent cases of cancers of stomach, colon, rectum, lung, uterus and prostate were all dominated by those in the oldest age group (>=65 years), >70% of cases being in this category (Table 4). This contrasts with the prevalence of melanoma, breast and cervix cancers that were more influenced by cases in middle age (45–64 years).


View this table:
[in this window]
[in a new window]
 
Table 4. Distribution of the proportions of prevalent cases of cancer in the UKa by age group
 
In males, about half the prevalent cases had been diagnosed >5 years before the index date (Table 5). In females, the corresponding figure was higher, ~60%. For the most fatal cancers, i.e. lung and stomach cancer, high proportions of the prevalent cases (~20–30%) were diagnosed within 1 year of the index date. A large proportion (~70%) of prevalent cases of prostate cancer, which predominantly occurs in the elderly, was diagnosed within 5 years of the index date. Around 30% of male and 40% of female prevalent cancer cases had been diagnosed >10 years before the index date. For males, the proportion of these ‘long-term’ survivors was highest for lung and stomach cancer (38% and 32%, respectively), reflecting the previously high and decreasing incidence rates of these two cancers over time, whereas for females, the proportion was highest for uterus cancer and melanoma of skin (48% and 39%, respectively).


View this table:
[in this window]
[in a new window]
 
Table 5. UKa cancer prevalence per 100 000 population by time since diagnosis, cancer site and sex
 
Table 6 shows a comparison of the total prevalence for all cancers and specific cancer sites at the end of 1992, of the EUROPREVAL estimates for England with the empirical ONS figures based on all cancer registrations in England and Wales made since 1971. With the exception of those for cervix cancer, where the EUROPREVAL estimates are for 15-year prevalence, the ONS rates were close to the estimates from EUROPREVAL; for leukaemia, in both sexes, the empirical figures were slightly higher. For all cancers, the direct results from ONS were reasonably close to the EUROPREVAL estimate in both males (1.45% compared with 1.50%) and females (2.3% compared with 2.5%).


View this table:
[in this window]
[in a new window]
 
Table 6. Comparison of Office for National Statistics (ONS) directly observed rates and EUROPREVAL estimates
 

    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Approximately 1.2 million people, 1.5% of males and 2.5% of females, in the UK population at the end of 1992 were living with a diagnosis of cancer. As would be anticipated, these proportions were strongly age dependent, especially in males, and increased to 7–8% in those aged >=65 years. It is probable that the prevalence has increased since 1992 for a combination of reasons. Incidence rates for some major cancers with poor survival have declined (male lung cancer and stomach cancer), while rates for those with a relatively good prognosis have increased (breast and prostate cancer) [1, 3] and, for many sites, survival rates have improved, particularly breast and colorectal cancers, which have high incidence rates [1012]. Even if age specific prevalence rates were not higher now, the gradual ageing of the population would have increased the absolute numbers of diagnosed cancer patients and hence also the numbers of prevalent cases.

Breast cancer dominates the pattern of cancer prevalence for females because it is the most commonly diagnosed tumour and, even before the introduction of breast screening in the UK in 1988, it had a five-year relative survival of 65% [10]. More than one in three of all living female cancer patients had been diagnosed with breast cancer. Although lung cancer has long been the most common cancer in UK males, the poor prognosis means that it makes only a small contribution (<10%) to overall cancer prevalence in males. No single site of cancer dominates the male prevalence pattern, although prostate cancer, which is the most common site overall, has by far the highest rate in the >65 years age group.

Approximately one half of all living male cancer patients had survived for >5 years and one-third had survived for >10 years. For females, these proportions were higher at ~60% and ~40%. The problems associated with cancer management are usually considered for the years immediately following diagnosis, whereas the majority of patients have been living with their disease for at least 5 years. Not all of these survivors are diagnosed with cancers that have good prognosis; thus, for both lung and stomach cancer, in both sexes, >30% of the prevalent cases had survived for >10 years. This apparent anomaly is, at least partly, explained by the relatively very high incidence rates for these cancers in the 1960s and 1970s such that a small fraction of long-term survivors can make a substantial contribution to current prevalence.

The annual number of new cancer cases diagnosed in the UK (excluding non-melanoma skin cancer) was ~252 000 in 1992 and this increased to 263 000 cases by 1998. The corresponding numbers of cancer deaths were 164 000 and 154 000, respectively. The estimated total number of prevalent cases on 31 December 1992 was 1 178 000 (applying the all cancers, all ages prevalence proportions per 100 000, provided in Table 3, of 1506 for males and 2534 for females to the 1992 UK population of 28.362 million males and 29.645 million females). The ratio of prevalence to incidence was thus ~5-fold, and the ratio of incidence to mortality ~1.5-fold in 1992. With generally rising incidence, gradual improvement in cancer care and increasing survival of cancer patients, prevalence will gradually increase; mortality from cancer will continue to decrease if the improvements in survival more than counterbalance the increases in incidence.

The large and increasing size of the population of living cancer patients emphasises the magnitude of the impact of cancer on peoples’ lives, either as patients themselves or as relatives of cancer patients, but the relevance of prevalence figures to the planning of cancer care is less obvious. The number of new cases in a year is an indicator of the need for resources for diagnostic investigations and first-line treatment. Similarly, the annual number of cancer deaths is an indicator of the need for resources for palliation and terminal care. Prevalence is a composite measure of a highly heterogeneous population comprised of: recently diagnosed patients in need of treatment and monitoring; long-term survivors, some of whom will nevertheless eventually die from the cancer, while others may be cured of the disease; and patients in the terminal phase who are dying from the cancer. Direct estimates of the average costs in each phase of cancer care have been obtained by linking incidence registry data with administrative databases [1315]. Treatment costs are strongly related to age, cancer type and stage of disease at diagnosis [13]. The costs of surveillance and monitoring activities needed in the intermediate phase are much lower than costs for first and terminal treatment but, as they may continue for a long period of time, their impact on the total cost of cancer care is not negligible.

Further work on the development of the definition of prevalence should attempt to identify subgroups of patients in the different phases of care. The disaggregation of prevalence by time since diagnosis (as in this paper) is a first step in this direction. Prevalence by stage of disease at diagnosis would be an additional and very informative indicator of the need for care. Recent developments in cancer survival analysis [16] can, for some types of cancer, distinguish long-term survivors who are cured and those who will die from their cancer. This may lead to more detailed assessments of prevalence in the future.


    Acknowledgements
 
This research was supported by the EUROPREVAL Biomed-2 Programme, Contract No. BH4 983899. Members of the UK EUROPREVAL Working Group. Scotland: R. Black, V. Harris, D. Stockton (Scottish Cancer Intelligence Unit). England: T.W. Davies, S. Godward (East Anglian Cancer Registry); M.P. Coleman, S. Harris (London School of Hygiene and Tropical Medicine); E.M.I. Williams (Merseyside and Cheshire Cancer Registry); D. Forman, R. Iddenden (Northern and Yorkshire Cancer Registry and Information Service and University of Leeds); M.J. Quinn, P.J. Babb (Office for National Statistics); M. Roche (Oxford Cancer Intelligence Unit); J. Smith (South and West Cancer Intelligence Unit); H. Møller (Thames Cancer Registry); P. Silcocks (Trent Cancer Registry); G. Lawrence, K. Hemmings (West Midlands Cancer Intelligence Unit). Italy: R. Capocaccia (Project Leader), I. Corazzaiari, R. De Angelis, S. Francisci, S. Hartley. F. Valente, A. Verdecchia, A. Zappone (National Institute of Health, Rome); F. Berrino, G. Gatta, A. Micheli, E. Mugno, M. Sant (National Institute for the Study and Cure of Tumours, Milan).


    Footnotes
 
+ Correspondence to: Professor D. Forman, Arthington House, Cookridge Hospital, Leeds LS16 6QB, UK. Tel: +44-113-392-4309; Fax: +44-113-392-4164; E-mail: d.forman{at}leeds.ac.uk Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
1. Quinn MJ, Babb PJ, Brock A et al. Cancer Trends in England and Wales, 1950–1999. Studies in Medical and Population Subjects No. 66. London, UK: The Stationery Office 2001.

2. Booth H, Cooper N, Quinn MJ (eds): Cancer Statistics: Registrations, England, 1998. Series MB1 No. 29. London, UK: Office for National Statistics 2002.

3. Scottish Cancer Intelligence Unit. Scottish cancer statistics on the web. Information and Statistics Division [on-line] http://www.show.scot.nhs.uk/isd/cancer/cancer.htm (18 February 2003, date last accessed).

4. Capocaccia R, Colonna M, Corazziari I et al. Measuring cancer prevalence in Europe: the EUROPREVAL project. Ann Oncol 2002; 13: 831–839.[Abstract/Free Full Text]

5. Micheli A, Mugno E, Krogh V et al. Cancer prevalence in European registry areas. Ann Oncol 2002; 13: 840–865.[Abstract/Free Full Text]

6. Krogh V, Micheli A. Measure of cancer prevalence with a computerized program: an example of larynx cancer. Tumori 1996; 82: 1–4.[ISI]

7. Capocaccia R, De Angelis R. Estimating the completeness of prevalence based on cancer registry data. Stat Med 1997; 16: 425–440.[CrossRef][ISI][Medline]

8. Merril RM, Feuer EJ, Capocaccia R, Mariotto A. Cancer prevalence estimates based on tumour registry data in the SEER program. Int J Epidemiol 2000; 29: 197–207.[Abstract/Free Full Text]

9. Capocaccia R, Gatta G, Chessa E, Valente F. The EUROCARE 2 Study. IARC Scientific Publication No. 151. Lyon, France: International Agency for Research on Cancer 1999; 1–40.

10. Coleman MP, Babb P, Damiecki P et al. Cancer survival trends in England and Wales, 1971–1995: deprivation and NHS region. Studies in Medical and Population Subjects No. 61. London, UK: Office for National Statistics 1999.

11. Coleman MP, Babb P, Harris S et al. Cancer survival in England and Wales, 1991–1998. Health Stat Q 2000; 6: 54–63.

12. Scottish Cancer Intelligence Unit. Trends in cancer survival in Scotland 1971–1995. Edinburgh, UK: Information & Statistics Division 2000.

13. Taplin SH, Barlow W, Urban N et al. Stage, age, comorbidity, and direct costs of colon, prostate, and breast cancer care. J Natl Cancer Inst 1995; 87: 417–426.[Abstract]

14. Fireman BH, Quesenberry CP, Somkin CP et al. Cost of care for cancer in a health maintenance organization. Health Care Financ Rev 1997; 18: 51–76.[ISI][Medline]

15. Brown ML, Riley GF, Potosky AL, Etzioni RD. Obtaining long-term disease specific costs of care: application to Medicare enrollees diagnosed with colorectal cancer. Med Care 1999; 37: 1249–1259.[CrossRef][ISI][Medline]

16. Verdecchia A, De Angelis R, Capocaccia R et al. The cure for colon cancer: results from the EUROCARE study. Int J Cancer 1998; 77: 322–329.[CrossRef][ISI][Medline]