JAMES M. ROBINS. (1984). A statistical method to control for the healthy worker effect in intracohort comparisons