Fig. 3From: Left-censored recurrent event analysis in epidemiological studies: a proposal for when the number of previous episodes is unknownAverage length of the 95% confidence interval according to population and maximum time at risk prior to the beginning of the cohortBack to article page