Kidney function in patients with bipolar disorder with and without lithium treatment compared with the general population in northern Sweden: results from the LiSIE and MONICA cohorts.

Background: The clinical relevance of lithium nephropathy is subject to debate. Kidney function decreases with age and comorbidities, and this decline might lead to attribution bias when erroneously ascribed to lithium. We aimed to investigate whether patients with bipolar or schizoaffective disorder had faster decline in estimated glomerular filtration rate (eGFR) compared with the general population, whether observed differences in the steepness of the decline were attributable to lithium, and whether such changes depended on the length of lithium exposure.

Methods: In this cross-sectional cohort study, we used clinical data from the Lithium-Study into Effects and Side-effects (LiSIE) retrospective cohort study, which included patients with bipolar disorder or schizoaffective disorder whose medical records were reviewed up to Dec 31, 2017, and the WHO Monitoring of Trends and Determinants in Cardiovascular Disease (MONICA) study, covering a representative sample of the general population in northern Sweden aged 25-74 years. The primary outcome was the age-associated decline of creatinine-based eGFR, assessed using linear regression. We adjusted for sex and grouped for different lengths of lithium exposure (never or <1 year, 1-5 years, >5-10 years, and >10 years). For patients with moderate-to-severe kidney disease we identified the underlying nephropathy in the case records.

Findings: From LiSIE, we included 785 patients (498 [63%] female and 287 [37%] male), with a mean age of 49·8 years (SD 13·2; range 25-74). From MONICA, we included 1549 individuals (800 [52%] female and 749 [48%] male), with a mean age of 51·9 years (13·8; 25-74). No ethnicity data were collected. Adjusted for duration of lithium exposure, eGFR declined by 0·57 mL/min/1·73 m/year (95% CI 0·50-0·63) in patients with bipolar disorder or schizoaffective disorder and by 0·57 mL/min/1·73 m/year (0·53-0·61) in the reference population. Lithium added 0·54 mL/min/1·73 m (0·43-0·64) per year of treatment (p<0·0001). After more than 10 years on lithium, decline was significantly steeper than in all other groups including the reference population (p<0·0001). Lithium nephropathy was judged to be the commonest cause of moderate-to-severe chronic kidney disease, but comorbidities played a role. The effect of lithium on eGFR showed a high degree of inter-individual variation.

Interpretation: Steeper eGFR decline in patients with bipolar disorder or schizoaffective disorder can be attributed to lithium, but the trajectory of kidney function decline varies widely. Comorbidities affecting kidneys should be treated assertively as one possible means to affect the trajectory. In patients with a fast trajectory, a trade-off is required between continuing lithium to treat mental health problems and discontinuing lithium for the sake of renal health.

Funding: Norrbotten County Research and Learning Fund Sweden, Visare Norr (Northern County Councils Regional Federation Fund), Swedish Kidney Foundation (Njurfonden), Swedish Kidney Association (Njurförbundet), Norrbotten section.

Translation: For the Swedish translation of the Summary see Supplementary Materials section.