[Using a method--'floating absolute risks' to estimate the confidential intervals of relative hazards].

Compare the risks between multiple groups and to avoid bias, using arbitrary reference group. Floating absolute risks (FAR) is used to estimate the relative hazard ratios and to get the confidence intervals. From the example, FAR can describe the construction of standard error for all groups. It gets the relative hazard ratios and 95% confidence intervals between any two groups; even if neither is used as baseline data. FAR can eliminate the bias in comparison of any two groups by selecting a baseline group. It is useful in many epidemiological studies.