“…For trials of sodium restriction, the statistically significant relationship between size of study and effect of treatment on DBP (size, P = 0.008) was because smaller trials showed larger reductions in blood pressure [20,92,95,97] than larger trials [93,98]; however, the two larger trials both showed significant reductions, suggesting that the treatment Galloe, 1993Grobbee, 1986Lyle, 1992McCarron, 1985‡ Nowson, 1989a‡ Nowson, 1989b‡ Strazullo, 1986Takagi, 1991Tanji, 1991Weinberger, 1993‡ Zoccali, 1988 Overall effect x x x effect was genuine. For fish oil trials, the relationship between size of study and treatment effect (size: P ¼ 0.10 and 0.03 for SBP and DBP, respectively) occurred because larger trials [138,141,142,144] showed greater reductions in blood pressure than smaller trials [139,140,143,145], and cannot be ascribed to publication bias.…”