Trends and inequalities in cancer survival
Our current research projects document trends in cancer survival, and aim to advance knowledge on the extent, the causes of, and possible solutions for inequalities in cancer survival.
The data for these research projects are provided by the national cancer registries of England, Scotland, Wales and Northern Ireland, under the relevant statutory and ethical approvals. For international projects, we have the relevant institutional, ethical and/or statutory approvals and data supply agreements with the cancer registries concerned.
Cancer Survival in the United Kingdom
The Cancer Survival Group produces UK-wide cancer survival figures for the Department of Health, and has done so for the last six years. We update these figures annually for all four UK nations, and every two years for the biennial report “Health at a Glance” published by the Organisation for Economic Co-operation and Development (OECD). This submission is part of the UK’s responsibility as an OECD Member State to contribute official data on health and a range of other topics to the regular statistical publications of OECD. We estimate net survival up to five years after diagnosis for all patients diagnosed with a first, primary, invasive cancer in the UK, and the pattern of survival for each of the main types of cancer by age, sex and calendar period.
We routinely produce yearly reports which include the production of cancer survival estimates in England for the most recent cohort of patients, as well as trends in survival since the mid-1990s.
We are investigating demographic factors that are known to impact inequalities in survival. These include age, by looking at survival in a cohort of elderly patients; and ethnicity, by comparing survival of South Asian migrants to England to non-South Asians. We are also investigating structural factors such as the quality and volume of hospital care, and screening practices for breast cancer.
We are going to estimate trends and inequalities in survival for patients diagnosed with one of the twenty one most common cancers until 2011 and followed up until 31/12/2012 in England, research which will update the early evaluation that we produced of the NHS Cancer Plan. One project specifically looks at trends and inequalities in survival from laryngeal cancer. We are also examining the various routes through which deprivation affects survival.
The group is involved in a number of international collaborations. These focus on comparing survival in England to that of other countries, and exploring the reasons for differences in survival.
We routinely produce estimates of survival for patients diagnosed with a cancer of the colorectum, breast or cervix in the UK; these estimates are included in the Organisation for Economic Co-operation and Development biennial Health at a Glance publication for comparisons of the levels of survival between the OECD state members.
The CONCORD-2 project, run at the London School of Hygiene and Tropical Medicine, aims at comparing cancer survival worldwide for 9 different types of cancer.
Additionally, the next phase of the International Cancer Benchmarking Partnership will examine the impact of stage at diagnosis and stage-specific treatments on the survival differences between six high-income countries.
We are also involved in a collaboration which compares survival between England and Japan. Specifically, we are exploring the trends in sex ratios in gastric cancer survival, and separately the varying deprivation gaps in survival between the two countries. A further project being conducted by Laura Woods examines the differences between England and Australia for breast cancer, by comparing survival from breast cancer for screen-detected women compared to those whose cancer was not diagnosed via screening, corrected for the effects of lead time bias and over-diagnosis.
A collaboration with the Geneva Cancer Registry aims at estimating long-term net survival among women with breast cancer as well as measuring the effect of predictive factors. High-quality data on follow-up and causes of death, as well as life tables of the general population, enable us to evaluate the best approach estimating net survival using both cause-specific and relative survival settings. By using the appropriate setting and flexible models, we will able to identify the main factors that predict long-term net survival and take into account their evolution in time providing a more comprehensive picture of the evolution of the disease.
We also look at several cancers in more detail to explore the reasons behind the survival differences.
In our work on breast cancer survival in the West Midlands, discrepancies in survival by screening and by deprivation category are significant. The underlying reasons for this are still under investigation.
We are also applying causal mediation methods to investigate the reasons behind inequalities in survival from breast cancer in the North England; colorectal cancer in Calvados, France, and lung cancer in Osaka, Japan. This is in collaboration with the Centre for Statistical Methodology.
Our previous research has shown that there are inequalities in survival for many cancers between sub-groups in the United Kingdom and also between the UK and other countries.
We conducted an extensive study of socio-economic differences in cancer survival in England and Wales in 2001 which showed substantial differences in survival between rich and poor patients diagnosed up to 1990. In 2004 we updated these analyses for patients diagnosed up to 1999 with one of the most common twenty malignancies. Our analyses demonstrated the persistence of inequality and of how these increased over time for cancers with improving prognosis. We have subsequently published a more detailed description of trends in inequalities complemented by expert clinical commentaries which address the patterns and suggest possible reasons for differentials in survival. More recently we have shown how the NHS cancer plan has influenced these differentials, how they have changed over time, and the extent of socio-economic differential in survival in Scotland.
We have summarised the evidence as to why socio-economic differentials in survival exist, and have also investigated the role of treatment, screening, co-morbidity, and health insurance in explaining socio-economic differentials. We have also shown that there are minimal socio-economic differentials amongst patients taking part in clinical trials: suggesting that many of the differences observed relate to differential cancer treatment between socio-economic groups.
In terms of geographical differences, we have documented how survival varies across the nations of the UK and Ireland. We have shown that survival in the UK is lower than similarly developed nations across the globe by collaboration in the EUROCARE studies, and by conducting the CONCORD studies , the International Cancer Benchmarking Study, as well as examining data from the USA and Australia. In the course of these studies we have examined the role of stage, treatment, subsite and morphology in explaining these differentials. We have routinely published survival estimates for small-areas of England and Wales, including . We have formulated means of quantifying how significant these inequalities are for public health in terms of the number of avoidable deaths.
We have also examined how breast cancer incidence, survival and mortality varies between South Asian and non-South Asian women within England. We have developed new ethnic life tables for use in net survival analysis which more accurately correct for the background mortality experienced by Asian, Black and White ethnic groups. These highlight the mortality disadvantage experienced by Black groups in particular.
We have used these life tables in analyses of net survival from breast cancer in the West Midlands, to see if there are differences in survival by ethnic group, deprivation and screening history. Early results show that there is little difference to be seen between ethnic groups in their survival from breast cancer within in each screening group (screen-detected vs non-screen detected). Discrepancies in survival by screening and by deprivation category are significant however.