years of the study groups (1982 for the 13-to 14-year-old children and 1989 for the6-to 7- year-old children). National data were ob- tained from the World Health Organization Expanded Programme up to 1996 (A. Bur- ton, personal communication, World Health Organization official estimates of immuniza- tion coverage 1980-1996, as of July 22, 1997. If data were. missing for any of the rel- evant years, we used the mean of i or 2 adja- cent years. Local immunization rates for the smallest geographic area containing the schools were obtained by writing to all the center collaborators (see "Acknowledgments" section). , with 1 follow-up letter. The associations between local and na- tional immunization rates and symptom prev- alence were tested with the Spearman rank correlation coefficient, and 95% confidence limits were calculated with the assumption ofa bivariate normal distribution for the ranks.