“…The principal value of cohort designs lies in the opportunity to collect import ant potentially etiologic information on study subjects initially and then to provide an appropriate time for observational follow-up. Thus, typical studies (Miles, 1977) have begun with registers, for example, of patients (Babigian & Odoroff, 1969;Barner-Rasmussen, Dupont, & Billie, 1986;Black, Warrack, & Winokur, 1985;Kiloh, Andrews, & Neilson, 1988;Pokorny, 1966;Winokur & Black, 1987), including previous suicide attempters (Ettlinger, 1964;Paarregaard, 1975;Pederson, Teft, & Babigian, 1975); or of nonpatient cohorts formed by some other characteristic such as place of residence (Hagnell, Lanke, & Rorsman, 1981;Singer, Garfinkel, Cohen, & Srole, 1976), school attendance (Paffenbarger & Asnes, 1966;Paffenbarger, King, & Wing, 1969), or military service (Alleback, Allgulander, & Fisher, 1988;Datel & Johnson, 1979;Datel & Jones, 1981, 1982Eggersten & Goldstein, 1968;Keehn, Goldberg, & Beebe, 1974;Rothberg & Fagan, 1990;Rothberg & McDowell, 1988;Rothberg, Rock, & Jones, 1984;Rothberg, Rock, Shaw, et al, 1988;Selzer & Jablon, 1977).…”