Standardized Incidence Ratio (SIR)

A standardized incidence ratio compares the number of new cancer cases actually seen in a group, such as a cohort of asbestos workers, with the number that would be expected if the group had the same cancer rates as the general population of the same age and sex. An SIR of 1.0 means no excess; 3.0 means three times as many cases as expected.

Origin

The SIR is the incidence counterpart of the standardized mortality ratio, built on the same method of indirect standardization developed for vital statistics. It became practical for cancer studies once population cancer registries could supply reliable expected rates.

Significance in Asbestos Context

Many asbestos cohort studies report their results as SIRs for mesothelioma, lung cancer, laryngeal cancer and other cancers. The SIR counts diagnoses, while the standardized mortality ratio counts deaths, so an SIR can capture cancers that were treated successfully or that killed people later than the study period. Each SIR comes with a confidence interval; a wide interval, common for rare outcomes, means the estimate is uncertain. Because expected numbers come from the general population, an SIR can be pulled down by the healthy worker effect, which tends to understate occupational risk. Experts on both sides cite SIRs when they argue about general causation, for example whether a particular fiber type or exposure raises the risk of a given cancer. A ratio above 2.0 is sometimes discussed in court in connection with the idea of a doubling of risk, but how, and whether, that is used depends on the court and the state.

Example Sentence

The cohort study reported an SIR of 2.1 for lung cancer among the plant's maintenance workers, with a confidence interval that did not include 1.0.

Related Terms

standardized mortality ratio (SMR), Incidence, Cohort Study, confidence interval, Relative Risk, Healthy Worker Effect, Cancer Registry, Doubling of Risk

Notes

• An SIR describes a group, not an individual; your own exposure history is what matters for your case.
• Look at the confidence interval as well as the ratio; small studies can give large but uncertain numbers.