Relative risk represents the strength of association between an exposure of interest and certain disease outcome.
There are three ways that relative risk is typically estimated: a risk ratio (the ratio of risk in exposed and unexposed groups; see example), a rate ratio (the ratio of disease rates in exposed and unexposed groups) and an odds ratio (an estimate of relative risk based on a comparison of odds of disease in exposed and unexposed groups). Odds ratios approximate risk ratios for diseases with incidence or prevalence of less than 10 percent of the study population. For all measures of relative risk, if the relative risk is larger than one, there is a positive association between the exposure and the disease, and the exposure may be a risk factor for the disease. If the relative risk is less than one, the exposure may have a protective effect. If the relative risk is equal to one, no association exists between the exposure and the disease.
Consider the following hypothetical cohort study of the association between cigarette smoking among pregnant women and the birth weights of their newborn infants:
|Exposure: Smoking during pregnancy?
||Outcome: Birth weight of newborn
For smokers, the risk of having babies with low birth weights = 10/(10 + 90) = 0.1
For non-smokers, the risk of having babies with low birth weights = 5/(5 + 95) = 0.05
The risk ratio of disease comparing smokers to non-smokers = 0.1/0.05 = 2.00
The odds ratio of disease comparing smokers to non-smokers = 10/5/90/95 = 10x95/5x90 = 2.11
This indicates that women who smoke during pregnancy have approximately two times the risk of delivering babies with low birth weights than those who do not smoke during pregnancy. Of note, in a cohort study, it is unnecessary to calculate the odds ratio since we can obtain the risk ratio directly from the study. The above-mentioned odds ratio calculation is only intended to show readers how an odds ratio is derived.
Relative risk is a measure of association, not causation. To conclude that a factor "causes" a disease to occur either requires experiment or observational data that meet several criteria for causal associations (see the following web source).