The analysis comprised 385 studies: 22 RCTs; 197 case-control, uncontrolled cohort or cross-sectional studies; and 166 case reports. The authors reported that there was little high-quality evidence. Sample sizes of observational studies and RCTs were generally small. Drop-out rates were high. Follow-up was poorly-defined. Confounding could not be ruled out. Publication bias was a possibility.
Renal function: Thirty studies (nine case-control studies) were included in the quantitative analysis. Lithium therapy resulted in a statistically significant reduction (15% of normal maximum) in urinary concentrating ability (WMD -158.43 milliosmoles (one thousandth of an osmole, mOsm)/kg, 95% CI -229.78 to -87.07 I2=81.3%; four case control studies) compared to control. There was no statistically significant difference for glomerular filtration rate (six case-control studies). The absolute risk of renal failure was small (0.5% of patients received renal replacement therapy; one study).
Thyroid function: Seventy-seven studies (16 case-control studies) were included in the quantitative analysis. Clinical hypothyroidism increased significantly with lithium therapy compared to control (OR 5.78, 95% CI 2.00 to 16.67, I2=52.5%; eight studies). Thyroid stimulating hormone increased significantly, on average by 4.00 IU/mL (95% CI 3.90 to 4.10; 16 studies). Thyroid function did not increase significantly (four case-control studies).
Parathyroid function: Sixty studies (four cohort, 14 case-control and six cross-sectional) were included in the quantitative analysis. Statistically significant differences for increased blood calcium were noted for lithium (WMD 0.09mmol/L, 95% CI 0.02 to 0.17, I2=99.5%; 13 case-control studies) and parathyroid hormone (7.32pg/mL, 95% CI 3.42 to 11.23, I2=89.2%; 10 case-control studies).
Bodyweight: Fifty-five studies (14 RCTs and 23 cohort studies) were included in the quantitative analysis. Weight gain was significantly greater following lithium therapy compared to placebo (RR 1.89, 95% CI 1.27 to 2.82, I2=0%; five RCTs) and significantly lower when compared to olanzapine (RR 0.32, 95% CI 0.21 to 0.49, I2=0%; two RCTs).
Skin disorders: Seventy-seven studies (two RCTs) were included in the quantitative analysis. There was no statistically significant difference in skin disorders between lithium and placebo in a pooled analysis of two RCTs.
None of the studies that assessed hair disorders (24 studies) were included in a quantitative synthesis. There were no significant differences between groups in terms of teratogenic potential (risk of congenital malformations).
Results of meta-regression and sensitivity analyses were not reported.