Relative Survival

A survival measure comparing observed survival in a population to expected survival in a comparable general population, without needing cause-of-death data.

Source: Ederer, Axtell & Cutler 1961

Cancer registries follow large populations whose individual causes of death are often unrecorded or unreliable, yet they need to measure how much a cancer shortens life. Relative survival meets this by dividing the survival actually observed in the patient group by the survival expected in a matched general population, so the ratio reflects the excess mortality from the cancer without any death needing to be classified by cause. Its independence from cause-of-death coding is what makes it the registry standard. Rutherford and colleagues (2015) describe this use.

Source: Rutherford et al. 2015

Relative survival is calculated as the ratio of the observed survival in the diseased population to the expected survival of a comparable group from the general population, matched for factors such as age, sex, and calendar period, with expected survival taken from population life tables. Dividing observed by expected survival gives the proportion surviving beyond what background mortality alone would produce, attributing the shortfall to the disease. This yields relative survival without requiring the cause of each death to be known.

Source: Ederer, Axtell & Cutler 1961

Relative survival is useful because it isolates disease-related survival without needing cause-of-death information, which is often missing or unreliable in population data, and because it removes the effect of differing background mortality, making survival comparable across populations, periods, and age groups. This is valuable for comparing cancer survival between countries or over time, where general mortality differs. By contrasting observed with expected survival, relative survival provides a measure of disease-specific outcome not distorted by other-cause mortality or dependent on cause coding.

Source: Latimer 2013

Relative survival avoids needing cause-of-death data by comparing observed survival in the diseased population with the survival expected from general population mortality, rather than counting deaths by cause. The difference between observed and expected survival is attributed to the disease, so disease-specific mortality is inferred from the comparison with background rates rather than from classifying each death. This is advantageous where cause of death is unavailable or unreliable, allowing disease survival to be estimated from observed all-cause survival and population life tables.

Source: Ederer, Axtell & Cutler 1961

Relative survival relies on the accuracy of the general population life tables used for expected survival and on the assumption that the diseased population's other-cause mortality matches the general population's, which may not hold if patients differ in other risks. It estimates a net, somewhat hypothetical measure of disease survival rather than the actual probability of dying from the disease amid competing causes. Its interpretation and estimation can be sensitive to method. These limitations mean relative survival is used as a comparative measure with its assumptions in mind.

Source: Latimer 2013