Methods for continuous outcomes

1. If each study used the same scale or variable (i.e. all used serum creatinine or Mini-Mental State score):

mean_summary = å (weight_i x mean_i) / å weight_i

mean_i = mean_tx - mean_control

weight_i = 1 / variance_i = 1 / SD_i²

(use pooled variance)

95% CI = mean_s +/- (1.96 x (variance_s)^^0.5)

variance_s = 1 / å weight_i

Test of homogeneity: Q = å [weight_i x (mean_s - mean_i)²]

2. If each study used a similar but different scale (i.e. one used CAGE, one used MAST for diagnosis of alcoholism):

d_summary = å (weight_i x d_i) / å weight_i

d_summary = summary estimate of the difference in effect sizes

d_i = effect size = (mean_tx - mean_control) / SD_pooled

weight_i = 1 / variance_i = (2 x N_i) / (8 + d_i²)

(use pooled variance)

95% CI = d_s +/- (1.96 x (variance_s)^^0.5)

variance_s = 1 / å weight_i

Test of homogeneity: Q = å [weight_i x (d_s - d_i)²]