Five RCTs (n=770) were included in the review.
Quality.
Concealment of allocation was unclear in all 5 RCTs. The blinding of patients and outcome assessors was reported in 3 RCTs, with clinicians also blinded in two of these.
Relapse.
Lithium showed a statistically significant benefit over placebo for preventing new episodes of mood disturbance, with the average risk of relapse being 40% and 60%, respectively. The overall random-effects relative risk (RR) was 0.65 (95% CI: 0.50, 0.84, P=0.001; 5 RCTs) and the NNT was 5. There was statistically significant heterogeneity between studies (P=0.04).
Lithium showed a statistically significant benefit over placebo for preventing manic episodes, with the average risk of a manic episode being 14% and 24%, respectively. The overall fixed-effect RR was 0.62 (95% CI: 0.43, 0.88, P=0.008; 4 RCTs) and the NNT was 10. There was no statistically significant heterogeneity between studies (P=0.27).
Lithium showed no statistically significant benefit over placebo for preventing depressive episodes, with the average risk of a depressive episode being 25% and 32%, respectively. The overall fixed-effect RR was 0.78 (95% CI: 0.60, 1.01, P=0.06; 4 RCTs) and the NNT was 14. There was no statistically significant heterogeneity between studies (P=0.20).
Acceptance.
There were statistically significantly fewer withdrawals in patients taking lithium compared with those taking placebo; the fixed-effect RR was 0.86 (95% CI: 0.80, 0.93).
Adverse events.
There was statistically significantly more cases of somnolence (RR 1.98, 95% CI: 1.02, 3.84), nausea (RR 1.76, 95% CI: 1.07, 2.92) and diarrhoea (RR 2.35, 95% CI: 1.35, 4.10) with lithium compared with placebo. The results of the statistical test for heterogeneity were not reported, and it was unclear whether the results were from a fixed-effect or random-effects meta-analysis. There was no significant difference in the occurrence of hypothyroidism or suicide between groups.