Cancer Research UK : European cancer survival - is Britain the 'sick man of Europe'?

European cancer survival - is Britain the 'sick man of Europe'?

Comparing countries' survival rates for cancer always makes for strong headlines, especially as the UK has historically lagged behind its European counterparts.

In 1989, an international group of doctors and scientists set out to measure and explain the international differences in cancer survival in Europe in a project called EUROCARE. The results put the UK near the bottom of a European 'league table' of cancer survival, prompting much public outcry. The UK government was criticised for its apparent failure in cancer care and was spurred to make sweeping changes.

With publication in August 2007 of the most recent set of figures from EUROCARE, we take a look at the science behind the so-called 'league tables' of European cancer survival. Are the differences entirely down to the quality of care patients receive in different countries? Or is the picture more complicated than it seems?

Regional variations
Stage at diagnosis
Classification of disease
Registry coverage
So what difference does it make?

Regional Variations

Before the EUROCARE project, people suspected that there were variations between European countries' survival rates, but no one could actually prove it. The published survival statistics among European countries obviously weren’t the same, but no one could say for sure whether the differences were because of the quality of treatment or the way the statistics were collected and analysed. Any comparisons that scientists or politicians made were also compromised by differences between countries' definitions, classifications and groupings of cancers, their procedures for patient follow-up, their methods of data quality control, their statistical methods of survival analysis and their presentation of results.

The EUROCARE project was launched to develop a standard way of analysing raw, nationally representative cancer survival data so that people could finally make meaningful comparisons between European countries. These comparisons showed unexpectedly large variations in cancer survival between countries with similar levels of wealth and health care provision, with the UK consistently faring worse than other comparable European countries.

These results quickly caught the attention of the media and the government came under fire. "The most common conclusion that people reach about reasons for variation in survival across Europe is that patients in some countries get better treatment than patients in other countries," says Ruth Yates, Cancer Research UK’s head of statistical information.

Few other European governments felt the impact of the EUROCARE data as strongly as in the UK. In response to the findings and public opinion, the government vowed to improve the organisation and funding of cancer detection and treatment. In September 2000, the NHS Cancer Plan was published, establishing the first comprehensive national cancer programme in England. It set out a 10-year national strategy to reorganise, standardise and rejuvenate cancer services so that by 2010, five-year survival rates for cancer would "compare with the best in Europe" (Department of Health, 2000).

But was quality of treatment really the only thing that could have influenced cancer survival rates?

"The importance of treatment in explaining differences in survival will depend on the type of cancer and the methods of data collection that the country in question uses," says Yates. "There are a number of factors other than the quality of treatment that can influence cancer survival rates."

Stage at Diagnosis

For many types of cancer, the stage at which it is diagnosed has a massive impact on how likely it is that a patient can be successfully treated. In general, a doctor can treat and cure cancers more easily the earlier it is detected. Therefore a country with a higher survival rate may have better cancer screening programmes rather than better treatment.

For instance, one study found that women diagnosed in the earliest stage of breast cancer were 26 times more likely to survive than women who were diagnosed in the most advanced stages (Ugnat et al, 2004).

Screening programmes and other systems of early detection vary across Europe, so large differences in the stages at which people are diagnosed would be expected. However, cancer registries don't routinely collect the stage at which a cancer is diagnosed.

Classification of Disease

Another factor affecting survival rates is how each type of cancer is classified.

"In Europe, most countries will register cancers according to the International Classification of Disease, which is now in its tenth revision," says Yates. "This is a very comprehensive system for making sure that we have comparable data across the world. However, the more we know about the disease the more problems arise with classification, and as with any international system there is the problem of interpretation and disputes over what should be classified where."

For example, some countries have coded carcinoma in situ of the bladder as invasive bladder cancer. The in situ carcinomas have far better survival rates than other types of bladder cancer, so those countries that include these types of cancer have better survival rates for bladder cancer.

While the potential for misclassification is not a problem for most types of cancer, regional variations have been identified for head and neck, lung, thyroid and soft tissue cancers, as well as borderline malignancies of the ovary, bladder and stomach. This results in more problems when attempting to compare survival rates.

Registry Coverage

A third problem with interpreting the EUROCARE data comes when one considers how much data each country collects.

In some northern European countries – the UK and most of Scandinavia, for example – the entire population is covered by cancer registration. In other countries, especially those in central and southern Europe, coverage is less than 10 per cent. Such a small sample of regional populations cannot represent a country as a whole. Registries that cover a small population have a smaller number of cases, making their survival data more susceptible to regional bias.

Such registries may also suffer from 'selection bias' because they are associated with areas that have specialist centres. People in these areas are likely to receive better treatment and have higher survival rates than the rest of the population who don’t have access to these resources and are not included in the statistics. Despite this, these small regional samples are used to estimate national survival rates.

Large amounts of data are collected in the UK, so its national figures are highly representative of the population as a whole. In fact, UK registries alone contributed more than half of the total data to EUROCARE. So having relatively accurate statistics about cancer rather than estimating the data from a potentially biased sample could have contributed to the lower survival rates. In this case, being good at collecting data may actually have counted against the UK in the EUROCARE comparisons.

However, for registries in the Netherlands, Italy, England and Sweden, a substantial increase in population coverage between EUROCARE-2 and -3 did not alter the overall survival rankings of these countries. This suggests that the impact of coverage on survival may not be as large as some people think.

There are other problems with the EUROCARE data in the way different countries follow up patients who have been diagnosed with cancer, and whether registrations of cancer death are mandatory or not (leading to data going missing). Another issue relates to cases where the death certificate is the only indication of cancer. If the patient’s history cannot then be traced back to a cancer diagnosis, the record (so-called 'Death Certificate Only', or DCO) is excluded from the survival data. By excluding this group of cancer deaths the sample is biased towards good survivors – but it is less biased towards good survivors in countries with low DCO rates, such as the UK. All of these factors can contribute to uncertainty when directly comparing any two country’s survival rates.

So what difference does it make?

Undoubtedly, there were real differences in the early diagnosis and treatment of cancer patients in the UK compared with other European countries, but the degree of that difference remains uncertain.

Partly as a result of the political impetus provided by the first EUROCARE study, there have been large changes in the organisation of cancer treatment throughout the UK since the studies were published. Existing screening programmes, such as that for breast cancer, have been extended, and a new screening programme for bowel cancer is being introduced. Approximately 1,400 more cancer doctors have been recruited in England, and treatment is increasingly delivered to patients by multi-disciplinary teams of specialists rather than by individual clinicians.

However, we will probably not see what sort of effect these changes have had in comparison to other European countries. Data for the most recent, and possibly last, EUROCARE report – EUROCARE-4 – have now been published in Lancet Oncology, but include cases diagnosed from 1995 onwards, as much as five years before the NHS Cancer Plan was put in place.

"In interpreting results it is important to discern a pattern rather than to look at isolated data from individual registries or countries," concludes Yates. "There is little to be gained by continually presenting over-simplified league tables of cancer survival which undermine progress being made in the UK and cause anxiety among cancer patients."

References:

Berrino F., Sant M., Verdecchia A., Capacaccia R. et al. (1995). "Survival of Cancer Patients in Europe: The EUROCARE Study". IARC
Berrino F., Capocaccia R. et al. (1999). "Survival of Cancer Patients in Europe: the EUROCARE-2 Study". IARC
Berrino et al. (2007) "Survival for eight major cancers and all cancers combined for European adults diagnosed in 1995–99: results of the EUROCARE-4 study" Lancet Oncology. DOI:10.1016/S1470-2045(07)70245-0
Coleman M.P., Gatta G. et al. (2003). "EUROCARE-3 summary: cancer survival in Europe at the end of the 20th century". Ann Oncol 14 Suppl 5: V128-V149
Department of Health (2000). "The NHS Cancer Plan". London, Department of Health
Ugnat et al. (2004). "Survival of women with breast cancer in Ottawa, Canada: variation with age, stage, histology, grade and treatment". British Journal of Cancer 90 (6) 1138-43
Verdecchia et al. (2007) "Recent cancer survival in Europe: a 2000–02 period analysis of EUROCARE-4 data" Lancet Oncology. DOI:10.1016/S1470-2045(07)70246-2