Seven RCTs were included in the review (2,413 patients).
There was a small but statistically significant effect size (hedges' g) of pregabalin on HARS total score (0.364, 95% CI 0.256 to 0.471; seven RCTs, 1,352 patients). The value of the Q statistic indicated that the sample was homogeneous. Duval and Tweedie's trim-and-fit calculation suggested a small reduction in the overall mean effect size (0.325, 95% CI 0.226 to 0.424).
Effect sizes for HARS subscales were 0.349 (95 CI 0.256 to 0.471; four RCTs) for psychic anxiety symptoms and 0.239 (95% CI 0.107 to 0.370; four RCTs) for somatic anxiety symptoms.
The fail-safe n calculation using the effect size for HARS total scores found that 77 new or unpublished studies with negative findings would be needed to reverse the present results.