Fig. 3. Method of randomization was poorly reported in all four journals during the three periods, although the reporting increased from 0% in 1981–1985 to 35% in 1991–1995 in one journal (P < 0.0083). When the rate of reporting the method of randomization for the journals was pooled within each period, the rate increased significantly from 1981–1985 to 1991–1995 (P < 0.008). The symbols represent the mean percentages of trials that reported the method of randomization for the four journals evaluated in each period.