FormalPara Key Points

  • Population-level cancer survival is an important measure of the impact of cancer among Indigenous and Tribal peoples.

  • Reporting survival statistics using a range of measures provides different perspectives and assists in appropriate interpretation.

  • Measures of survival require accurate information on ethnicity, which is lacking in many countries.

  • Absolute measures of survival, such as remaining life expectancy, crude probability of death, and avoidable cancer deaths, provide tangible estimates of the population-level impact of cancer diagnosis.

  • Population-level statistics may help to inform decision-making at the individual level, although the unique circumstances of each person also need to be considered.

In the context of generally poorer health among Indigenous and Tribal peoples [1], assessing the impact of a cancer diagnosis within this population is important. Cancer survival is an important measure of this and can enable informed decision-making to improve health outcomes and reduce the burden of cancer in these populations. However, to ensure that analyses are meaningful and robust, we need to ensure we are using the most appropriate methods possible. In this chapter, we briefly summarize the concept of cancer survival, as well as contemporary methods that measure this concept.

Concept of Cancer Survival

Cancer survival refers to the probability of individuals being alive after a certain period of time following a cancer diagnosis. Often expressed as a percentage, it is designed to reflect the success of cancer treatments and overall outcomes for people diagnosed with cancer. Survival rates are usually calculated over a specific timeframe post-diagnosis, such as five years, and can vary depending on the type and stage of cancer. Cancer survival statistics provide important information for patients, doctors, and researchers to understand the effectiveness of treatments, make informed decisions about care, and assess progress in cancer management. They can also be used to assess outcomes for specific population groups, such as Indigenous and Tribal peoples, and identify inequalities compared to other population groups.

Individual-Level Versus Population-Level Survival Statistics

Individual-level survival statistics provide personalized prognosis information based on an individual’s specific circumstances, including demographics (age, sex), cancer characteristics (type, how early it was diagnosed), treatment, and personal experiences. Population-level survival statistics examine groups of people to identify patterns and trends to help researchers and healthcare providers understand common risk factors, assess treatment effectiveness, and make decisions about public health strategies. Importantly, population survival statistics reflect the average over a specific population group, thus providing an important guide for planning; however, they are not necessarily relevant to an individual’s own cancer journey.

Cancer Survival Versus Cancer Mortality

“Cancer mortality” and “cancer survival” focus on different groups of people to understand deaths from cancer.

Cancer mortality is an examination of the entire population to determine how many people die from cancer within a specific time period, while cancer survival focuses on the outcomes of people who have been diagnosed with cancer during that time and how many are still alive after a certain number of years post-diagnosis.

Cancer mortality is calculated as the number of cancer deaths as a rate of the total population per year. Cancer survival is calculated as the number of cancer deaths as a percentage of those who have been diagnosed with cancer in a given period.

All-cause survival is another useful statistic, which is measured as the number of people who die from any cause during a specific period post-cancer diagnosis.

Sources of Cancer Survival Data

Population-based cancer registries play a crucial and ongoing role in collecting and reporting cancer diagnoses within specific populations. Cancer registries link these data with national death registration records to determine the number of individuals diagnosed with cancer who have died and the time between their diagnosis and death.

Understanding cancer survival for Indigenous and Tribal peoples requires the collection of robust ethnicity data. Obtaining the necessary information to accurately establish cancer survival rates is complex, and high-quality data collection for Indigenous people is lacking in many countries [2]. For example, population survival statistics for Australian Aboriginal and Torres Strait Islander people are typically reported at the semi-national level, restricting statistics to states and territories with cancer registries known to have reliable identification data. In Aotearoa New Zealand, ethnicity data are collected through routine healthcare records and can be linked to national cancer registry and mortality data collections, enabling national-level comparisons of cancer survival outcomes between ethnic groups [3]. However, such comparisons are likely to be conservative due to the underreporting of Indigenous Māori ethnicity within these collections [4]. In the United States, American Indian or Alaska Natives are often misclassified, although data linkages to patient registration data are used to improve accuracy [5].

Methods of Reporting

Overall Survival

Arguably, overall survival is the most easily understood measure of survival. It reveals the proportion (or percentage) of people still alive after a certain number of years post-diagnosis. The example shown in Fig. 61.1 has an overall survival rate of 60%; however, this counts all deaths the same, irrespective of cause. An overall high survival rate could be due to fewer deaths from cancer, fewer deaths from other causes, or both. Overall survival, therefore, does not reveal the extent to which cancer diagnosis affects a person’s life or how it differs between population groups.

Fig. 61.1
A schematic depicts the overall 5-year survival rate, 5-year cause-specific survival rate, and 5-year survival rate of group A, cancer patients, and the 5-year relative survival rate of group B, the general population.

Hypothetical scenario showing the calculation of overall, cause-specific, and relative survival for a group of people diagnosed with cancer

Full size image

Net Survival

Net survival, which accounts for other causes of death, helps to discern the impact of cancer on survival. It reflects the impact of the diagnosed cancer only, excluding other causes of death. Net survival is typically measured in one of two ways—cause-specific survival and relative survival.

Cause-specific survival uses death from cancer as the main outcome, while censoring deaths from other causes. Censoring, in this context, means that information about death from other causes is used in survival calculations only until the point at which an individual dies from another cause. For example, an individual who died from an unrelated cause six years post-cancer diagnosis will be included in calculations of five-year survival, but not of 10-year survival. In the example shown in Fig. 61.1, cause-specific survival is 92%. Since 35 people have been censored, 65 are included in the calculation, 60 of whom are still alive. However, determining the precise cause of death can be difficult, due to the disease complexity, potential treatment-related complications, incomplete medical records, and multiple health factors or comorbidities that may have contributed to the death. This is a limitation for population-based cancer registry data [6].

The second method for measuring net survival is relative survival. This compares how long a group of cancer patients lives to a similar group without cancer. It assumes that any survival difference (shown as a ratio) is because of the cancer diagnosis. The comparison group is typically the general population, using official population life tables differentiated by age and sex. In example shown in Fig. 61.1, 60% of the cancer patients have survived for five years post-diagnosis, compared with 70% expected from the matched comparison group, resulting in a relative survival rate of 86%.

Owing to the challenges in determining a single cause of death in cancer registry data, relative survival has historically been the preferred method of reporting cancer survival in population-based cancer registries [6]. However, the interpretation of relative survival statistics has a number of potential limitations.

First, valid comparisons between the cancer cohort and the general population rely on accurate population life tables [7]. This can be a particular limiting factor for Indigenous peoples, for whom ethnic identification in population mortality or cancer registry data may not be robust. Second, relative survival assumes that the only difference between the group diagnosed with cancer and the general population is the cancer diagnosis. However, other factors may be more prevalent in the cancer cohort, such as smoking. For this reason and to ensure the impact of cancer on survival is accurately conveyed, comparisons of relative survival between Indigenous and non-Indigenous populations should use population life tables specific to the subpopulation in question [7]. Third, relative survival can be difficult to interpret, as it reflects a non-realistic scenario in which a person cannot die from a non-cancer—a scenario that is not reflected in real life. A relative survival estimate is better interpreted as a ratio [8]. For example, a relative survival estimate of 70% means cancer patients are 70% as likely, or 30% less likely, to survive five years than the general population. This lacks meaning if expectations for general population survival are not also reported.

Therefore, although relative survival is the most widely used framework for calculating and reporting cancer survival estimates among populations [6], there are substantial challenges to correctly understanding and communicating the statistics to a wide audience. In addition, because it is a relative rather than an absolute measure, interpreting comparisons between populations can be difficult as differences may be due to differences in observed survival for the cancer cohort or in expected survival for the specific population. For example, while overall life expectancy in Australia is high [9], the average life expectancy among Aboriginal and Torres Strait Islander people is approximately eight years shorter than other Australians [10]. Consequently, if the overall survival of the two populations is similar, relative survival among Aboriginal and Torres Strait Islander Australians may be higher.

Absolute Measures of Survival

Absolute measures may provide more tangible estimates of the population-level impact of a cancer diagnosis on a person’s life [11]. Three absolute measures are remaining life expectancy, crude probability of death, and avoidable deaths. Each is calculated within the relative survival framework, meaning that specific information about the cause of death is not required, but population-specific life tables comparable to the specific cancer cohort are required.

Remaining life expectancy, or average lifespan, post-cancer diagnosis provides another perspective on cancer survival. Remaining life expectancy is the number of years, on average, an individual can expect to live post-cancer. It quantifies the long-term impact of cancer throughout a person’s remaining lifetime, rather than focusing on a specific time period following diagnosis. While this information provides an important perspective for people diagnosed with cancer, comparisons of remaining life expectancy between population groups need to first consider the differences in overall life expectancy among people not diagnosed with cancer.

Crude probability of death estimates consider other causes of death as well, making them more suitable for risk communication and clinical decision-making [12]. They describe the number of people per 100 diagnosed with a specific cancer who die from that cancer, die from other causes, or remain alive after a certain time. Cause-of-death information is not required, as the general population mortality is used as a proxy for non-cancer causes of death. The difference between net and crude survival is small for younger people, who are less likely to die of other causes. However, for older individuals, crude probability estimates are important for accurately communicating the prognosis and the implications of a cancer diagnosis. Crude probability of death estimates must be expressed in terms of an “at-risk” period, such as the number of cancer deaths within five years of diagnosis. Crude probability of death can be very useful in understanding inequalities in cancer survival between different groups, such as Indigenous people, because it describes both the risk of death from cancer and the risk of death from other causes.

The concept of avoidable deaths conveys the impact of survival differences between two population groups by estimating the number of deaths that could be avoided if both populations have the same cancer survival probabilities [13]. Since this estimate is based on crude probability of death estimates, it must also be expressed in terms of an “at-risk” period.

Case Study

Data

To demonstrate different population-level cancer survival statistics in an Indigenous population, in collaboration with Aboriginal leader, Professor Gail Garvey, and other colleagues, we obtained cancer registry data from Australian cancer registries with sufficiently high levels of identification over the study period: Queensland, Western Australia, Northern Territory, and New South Wales [14]. These account for approximately 84% of Australian’s Aboriginal and Torres Strait Islander population [15]. The study cohort consisted of over 700,000 Australians diagnosed with cancer between 2005 and 2016, including nearly 13,000 Aboriginal and Torres Strait Islander people.

Ethics approval for this study was obtained from the Aboriginal Health and Medical Research Council Ethics Committee (1256/17), the Northern Territory (NT) Department of Health, the Menzies School of Health Human Research Ethics Committee (2016–2689), and the New South Wales (NSW) Population and Health Services Research Ethics Committee (2017/HRE0204).

Relative Survival

Between 2005 and 2016, five-year relative survival for people diagnosed with cancer was 49.1% [16]. This was lower than for the rest of the Australian population over the same period (59.6%). The five-year relative survival rate varied substantially by cancer type, being substantially lower (<15%) for liver, pancreatic, and lung cancer, and higher (>70%) for prostate, uterine, and breast cancers and melanoma.

Remaining Life Expectancy

On average, Aboriginal and Torres Strait Islander people diagnosed with cancer survived for 12 years post-diagnosis. This was lower than for other Australian patients who lived, on average, for 20 years post-diagnosis (Fig. 61.2) [16]. This is after cancer type and age at diagnosis were taken into account. The disparity was evident across all cancer types (Fig. 61.2). On average and across all cancer types, approximately one-quarter of this gap can be attributed to higher cancer-related mortality, and three-quarters to other causes of death (although it varied substantially by cancer type) (Fig. 61.3). Typically, for cancer types with high survival rates, non-cancer causes contributed more to the life expectancy gap.

Fig. 61.2
A horizontal double bar graph plots the remaining life expectancy versus various cancer affected organs. The bars represent aboriginal and Torres Strait Islander people and other Australians, respectively. The life expectancy of other Australians is higher for all cancer affected organs.

Remaining life expectancy following a diagnosis of selected cancer types for Aboriginal and Torres Strait Islander people and other Australians, Australia, 2005–2016. (Source: Adapted from Ref. Dasgupta et al. [16])

Full size image
Fig. 61.3
A stacked bar chart compares the contribution to disparity in remaining life expectancy versus various cancer-affected organs. It depicts that cancer mortality is highest in liver and lung cancer, and non-cancer mortality is highest in prostate cancer.

Contributions to differences in remaining life expectancy following a diagnosis of selected cancer types between Aboriginal and Torres Strait Islander people and other Australians, Australia, 2005–2016. (Source: Adapted from Ref. Dasgupta et al. [16])

Full size image

Crude Probability of Death

Using the semi-national Australian cohort, out of every 100 Aboriginal and Torres Strait Islander people diagnosed with any invasive cancer between 2011 and 2016, approximately 43 would have died from their cancer within five years, and seven would have died from other causes (Fig. 61.4) [13], compared to 35 deaths from cancer and four from other causes for non-Indigenous Australians.

Fig. 61.4
A schematic compares cancer mortality rates between Aboriginal and Torres Strait Islander people and other Australians. It exhibits a higher mortality rate for Aboriginal and Torres Strait Islander people, with 43 deaths compared to 35 deaths for other Australians.

Crude probability of death within 5 years of diagnosis for Aboriginal and Torres Strait Islander people and other Australians diagnosed with cancer, Australia, 2011–2016. (Source: Adapted from Ref. Dasgupta et al. [13])

Full size image

Avoidable Deaths

In this study, approximately 1270 Aboriginal and Torres Strait Islander people were diagnosed with cancer annually between 2012 and 2016 [13]. Of these, 646 died within five years post-diagnosis. If, however, this population group had the same survival as other Australians, 133 deaths (three-quarters of which were caused by cancer) within five years of diagnosis were potentially avoidable. In other words, for every 100 Aboriginal and Torres Strait Islander people diagnosed with cancer, approximately 11 of 50 deaths (seven due to cancer) within five years of diagnosis could be avoided if they had the same overall survival as other Australians with cancer (Fig. 61.5) [13].

Fig. 61.5
A schematic highlights the number of cancer deaths among Aboriginal and Torres Strait Islander people within 5 years of diagnosis. It depicts the number of deaths due to cancer is 50, of which 11 are avoidable deaths.

Deaths among Aboriginal and Torres Strait Islander people that could have been avoided within 5 years of their cancer diagnosis if they had the same overall survival as other Australians diagnosed with cancer, Australia, 2012–2016. (Source: Adapted from Ref. Dasgupta et al. [13])

Full size image

Conclusion

Cancer survival is multifaceted and a variety of statistics provide different perspectives, particularly regarding inequalities in outcomes for specific population groups. While recognizing their limitations, these statistics can help decision-makers and planners to appropriately focus efforts on improving outcomes and reducing avoidable deaths. The statistics can also be used to highlight where gains have been made; for example, survival among Aboriginal and Torres Strait Islander people diagnosed with cancer has improved substantially over time. However, because survival among other Australians has improved as well, the gap hasn’t decreased. Effective communication of these statistics empowers people diagnosed with cancer, and those providing support, to be actively involved in the treatment decision-making process. Using different methods to explain survival rates at the population level allows for different perspectives, particularly when reporting absolute measures. Ultimately, statistics act as a guide to inform the decisions that individuals and their health professionals need to consider when taking into account each individual’s unique circumstances.