Absolute and Relative Risks of Kidney Outcomes Associated With Lithium vs Valproate Use in Sweden

Abstract

Importance: Among patients with bipolar disorder, discordant findings have been published on the nephrotoxic effects of lithium therapy.

Objective: To quantify absolute and relative risks of chronic kidney disease (CKD) progression and acute kidney injury (AKI) in people who initiated lithium compared with valproate therapy and to investigate the association between cumulative use and elevated lithium levels and kidney outcomes.

Design, setting, and participants: This cohort study had a new-user active-comparator design and used inverse probability of treatment weights to minimize confounding. Included patients initiated therapy with lithium or valproate from January 1, 2007, to December 31, 2018, and had a median follow-up of 4.5 years (IQR, 1.9-8.0 years). Data analysis began in September 2021, using routine health care data from the period 2006 to 2019 from the Stockholm Creatinine Measurements project, a recurrent health care use cohort of all adult residents in Stockholm, Sweden.

Exposures: New use of lithium vs new use of valproate and high (>1.0 mmol/L) vs low serum lithium levels.

Main outcomes and measures: Progression of CKD (composite of >30% decrease relative to baseline estimated glomerular filtration rate [eGFR] and kidney failure), AKI (by diagnosis or transient creatinine elevations), new albuminuria, and annual eGFR decrease. Outcomes by attained lithium levels were also compared in lithium users.

Results: The study included 10 946 people (median [IQR] age, 45 [32-59] years; 6227 female [56.9%]), of whom 5308 initiated lithium therapy and 5638 valproate therapy. During follow-up, 421 CKD progression events and 770 AKI events were identified. Compared with patients who received valproate, those who received lithium did not have increased risk of CKD (hazard ratio [HR], 1.11 [95% CI, 0.86-1.45]) or AKI (HR, 0.88 [95% CI, 0.70-1.10]). Absolute 10-year CKD risks were low and similar: 8.4% in the lithium group and 8.2% in the valproate group. No difference in the risk of developing albuminuria or the annual rate of eGFR decrease was found between groups. Among more than 35 000 routine lithium tests, only 3% of results were in the toxic range (>1.0 mmol/L). Lithium values greater than 1.0 mmol/L, compared with lithium values of 1.0 mmol/L or less, were associated with increased risk of CKD progression (HR, 2.86; 95% CI, 0.97-8.45) and AKI (HR, 3.51; 95% CI, 1.41-8.76).

Conclusions and relevance: In this cohort study, compared with new use of valproate, new use of lithium was meaningfully associated with adverse kidney outcomes, with low absolute risks that did not differ between therapies. However, elevated serum lithium levels were associated with future kidney risks, particularly AKI, emphasizing the need for close monitoring and lithium dose adjustment.

Figure 1.. Weighted Cumulative Incidence Curves Showing the Risk of Chronic Kidney Disease (CKD) Progression, Acute Kidney Injury (AKI), and Albuminuria by Lithium or Valproate Initiation and Their Respective Absolute Risk Differences (Lithium vs Valproate) Throughout Follow-up

Curves were weighted for the following covariates: age, sex, attained education, baseline estimated glomerular filtration rate, number of hospitalizations during previous year and psychiatry related, number of outpatient contacts during previous year and psychiatry related, number of outpatient contacts during previous year, pregnancy in the 2 years prior, total number of medications in the previous year, comorbidities (bipolar disorder, depression, manic episode, anxiety disorder, mental disorders attributable to psychoactive substance use, schizophrenia spectrum disorders, hypertension, diabetes, acute coronary syndrome, other ischemic heart disease, heart failure, stroke, other cerebrovascular diseases, valve disorders, atrial fibrillation, other arrhythmia, hyperthyroidism, cancer, and liver disease) and ongoing medications (lamotrigine, carbamazepine, first- and second-generation antipsychotic drugs, other mood stabilizers, antidepressants, attention-deficit/hyperactivity disorder medication, drugs used in addictive disorders, opioids and pain medications, antiepileptic drugs, β-blockers, calcium channel blockers, diuretics, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, lipid-lowering drugs, and nonsteroidal anti-inflammatory drugs). Shading indicates 95% CIs.

Figure 2.. Hazard Ratios (95% CIs) for the Risk of Chronic Kidney Disease (CKD) Progression, Acute Kidney Injury (AKI), or Albuminuria Associated With the Cumulative Use of Lithium and Valproate

The x-axis is truncated at 3000 defined daily dosages (DDDs) because this is the maximum DDD in the valproate group. Cumulative dose was calculated as the time updated sum of all dispensed DDDs since therapy initiation and modeled as a time-dependent covariate. Models adjusted for the following variables: age, sex, attained education, baseline estimated glomerular filtration rate, number of hospitalizations during previous year and psychiatry related, number of outpatient contacts during previous year and psychiatry related, number of outpatient contacts during previous year, pregnancy in the 2 years prior, total number of medications in the previous year, comorbidities (bipolar disorder, depression, manic episode, anxiety disorder, mental disorders attributable to psychoactive substance use, schizophrenia spectrum disorders, hypertension, diabetes, acute coronary syndrome, other ischemic heart disease, heart failure, stroke, other cerebrovascular diseases, valve disorders, atrial fibrillation, other arrhythmia, hyperthyroidism, cancer, and liver disease), and ongoing medications (lamotrigine, carbamazepine, first- and second-generation antipsychotic drugs, other mood stabilizers, antidepressants, attention-deficit/hyperactivity disorder medication, drugs used in addictive disorders, opioids and pain medications, antiepileptic drugs, β-blockers, calcium channel blockers, diuretics, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, lipid-lowering drugs, and nonsteroidal anti-inflammatory drugs). Shaded areas indicate 95% CIs. HR indicates hazard ratio.