Standardized incidence ratio (SIR) uses a general community as the reference population, whereas incidence rate ratio (IRR) uses comparator control group in a cohort study as a reference population. How to make a comparison between them?
As in so many things in epidemiology, it depends on the design and power of the study. A community-based reference group has the advantage of large numbers and so statistical stability. However, occupational cohorts often vary in their characteristics from the profile of the local population, in ways that bias the risk estimate. (The healthy worker effect is the most important example.) On the other hand, a comparison population (not strictly a control group because exposure is unlikely but not controlled) is almost always relatively small and so power is more limited. If comparison population is not selected either randomly (which is usually not possible for occupational studies) or with care to match the cohort of interest, you can have considerable selection bias and not know it. Within a single employer, there is a reason some people work in jobs with exposure and others do not. That reason may confound the study.
Incidentally, SIR and SMR usually does imply community-based comparison "denominator" data, but not always. IRR is a technically correct term but often other terms, like RR, are used in perfectly good studies.
As in everything, be sure you know the methodology actually used and don't assume from the abbreviation which risk estimate was actually used.
- Badges
- Science topic
- Similar topics
- Medicine
- Public Health