Figure 2. Seroepidemiological investigation of parvovirus B19 infection in Japan.

A. Comparison between observed and predicted proportion of seropositive from 2004–07 without accounting for time-dependency. The mean of observed data are shown as filled circles with the 95% confidence intervals by whiskers. Thick solid line represents the prediction based on a constant force of infection model, dashed line a gamma-type age-dependent model, and thin solid line a piecewise constant age-dependent model. B. Estimated force of infection with parvovirus B19 in Japan. The differential styles of lines correspond to those in panel A. C. Comparison of observed seroprevalence data between 1993 and 2004–07. Filled circles represent the observed data from 2004–07, while unfilled diamonds represent the data from 1993. Whiskers extend to both ends of the 95% confidence intervals. D. Comparison of seroprevalence data between 1993 and 2004 by birth year. Dashed lines represent the 95% confidence intervals. The horizontal line shows the birth year of 2004 participants, while the birth year of 1993 participants is calculated as the birth year minus 1 (i.e. those born from 1955–64 for 2004 participants are compared against those born from 1954–63 for 1993 participants). E. Comparison of predicted data from 2004–07 using age-dependent model and age- and time-dependent model. Filled circles represent the observed data, and whiskers extend to both ends of the 95% confidence intervals. Thick line shows the prediction based on time- and age-dependent assumption. F. The comparison between the time-dependent element of the force of infection (dashed line; left vertical axis) and the annual notification rate of erythema infectiosum from sentinel medical institutions in Fukuoka and Saga prefectures, Japan, from 1982 onwards (circles and solid line; right vertical axis).