Selection effects on an estimation of long-term changes in pulmonary function.

To reveal the association of initial pulmonary function level with subsequent mortality and participation in a follow-up reexamination, a prospective cohort study was performed. Female residents in a volcanic area of southern Kyushu, Japan, were followed up for their vital status and the pulmonary function 15 years after they received the first pulmonary function test. A cohort of 512 Japanese female residents who were examined for pulmonary function as indicated by forced expiratory volume and forced expiratory volume in one second was measured in a baseline examination in 1980. After 15 years, 35 females were lost to follow-up. Of the remaining 477 females, 340 and 137 females provided good and poor levels of pulmonary function tests (PFT) at baseline, respectively. Mortality by 1995 in the poor PFT group was significantly higher than that in the good PFT group (33.6% vs 9.4%). The mortality differences were still highly significant when the 35 lost cases were included as all alive. Among the 399 survivors, the nonparticipation rate in the reexamination in 1995 was significantly higher in the poor PFT group than that in the good PFT group (80.2% vs 69.5%). The results of the present study, a longitudinal study of pulmonary function, provide evidence of selection effects due to death or failure to participate in a subsequent reexamination.