Figure 2

Estimated biases resulting from imputing age at HCV infection as the age of first IDU. Bias is defined here as the reported age of first IDU minus the fitted mean from the multivariate models described by Figure 1 and Table 3. Circles below the horizontal line represent subjects who are likely to have been infected after their first IDU, while those above represent subjects likely to have been infected before their first IDU. We have added random numbers ranging from -0.4 to +0.4 to the integer ages in order to increase the visibility of distinct points. Included are HCV seropositive subjects with some history of IDU. a) WIHS data, n = 434; b) UHS data, n = 4047.