Five RCTs (n=1,072) were included in the meta-analysis.
Patients receiving lamotrigine were more likely to respond to treatment than those receiving placebo, using the HRSD (RR 1.27, 95% CI 1.09 to 1.47) and using the MADRS (RR 1.22, 95% CI 1.06 to 1.41). The NNT to achieve one more response than placebo was 11 (95% CI 7 to 25) using the HRSD and 13 (95% CI 7 to 33) using the MADRS. Remission rates for lamotrigine were significantly higher using the MADRS (RR 1.21, 95% CI 1.03 to 1.42). For continuous symptoms, there was a statistically significant difference in favour of lamotrigine using the MADRS (WMD -1.43, 95% CI -2.80 to -0.06; SMD -0.12, 95% CI -0.24 to -0.00).
There was no significant difference in discontinuation rates between lamotrigine and placebo, nor in interaction between diagnostic subgroup (bipolar I and II) and treatment effect. Lamotrigine was superior to placebo in patients with severe depressive symptoms (HRSD score more than 24) at randomisation (RR 1.47, 95% CI 1.16 to 1.87; NNT 7, 95% CI 4 to 17; SMD -0.24, 95% CI -0.42 to -0.06), and this interaction was also found using the continuous MADRS as the independent variable. The authors proposed that this interaction arose from a higher response rate in moderately ill placebo-treated patients, rather than a higher response rate from the severely ill treatment group.