Use of Lithium and Anticonvulsants and the Rate of Chronic Kidney Disease: A Nationwide Population-Based Study

Importance  Lithium is the main mood stabilizing drug for bipolar disorder. However, it is controversial whether long-term maintenance treatment with lithium or other drugs for bipolar disorder causes chronic kidney disease (CKD).

Objective  To compare rates of CKD and in particular rates of end-stage CKD among individuals exposed to successive prescriptions of lithium, anticonvulsants, or other drugs used for bipolar disorder.

Design, Setting, and Participants  This is a Danish nationwide population-based study of 2 cohorts. Cohort 1 comprised a randomly selected sample of 1.5 million individuals among all persons who were registered in Denmark on January 1, 1995, all patients with a diagnosis of a single manic episode or bipolar disorder between January 1, 1994, and December 31, 2012 (n =10 591), and all patients exposed to either lithium (n = 26 731) or anticonvulsants (n=420 959). Cohort 2 included the subgroup of 10 591 patients diagnosed as having bipolar disorder.

Main Outcomes and Measures  Possible CKD, definite CKD, and end-stage CKD (defined as long-term dialysis or renal transplantation).

Results  A total of 14 727 (0.8%), 18 762 (1.0%), and 3407 (0.2%) in cohort 1 and 278 (2.6%), 319 (3.0%), and 62 (0.6%) in cohort 2 were diagnosed as having possible, definite, or end-stage CKD, respectively. Based on the total sample and not considering diagnoses, use of lithium was associated with an increased rate of definite CKD (0 prescriptions: hazard ratio [HR] = 1.09, 95% CI, 0.81-1.45; ≥60 prescriptions: HR = 3.65, 95% CI, 2.64-5.05; P for trend < .001) and possible CKD (0 prescriptions: HR = 1.01, 95% CI, 0.79-1.30; ≥60 prescriptions: HR = 2.88, 95% CI, 2.17-3.81; P for trend < .001), whereas use of anticonvulsants, antipsychotics, or antidepressants was not. Neither use of lithium nor use of any other drug class was associated with increasing rates of end-stage CKD. In patients with bipolar disorder, use of lithium was associated with an increased rate of definite CKD (1-2 prescriptions: HR = 0.89, 95% CI, 0.39-2.06; ≥60 prescriptions: HR = 2.54, 95% CI, 1.81-3.57; P for trend < .001) or possible CKD (1-2 prescriptions: HR = 1.26, 95% CI, 0.65-2.43; ≥60 prescriptions, HR = 2.48, 95% CI, 1.80-3.42; P for trend < .001), as was use of anticonvulsants (definite CKD, 1-2 prescriptions: HR = 1.23, 95% CI, 0.76-1.99; ≥60 prescriptions, HR = 2.30, 95% CI, 1.53-3.44; P for trend < .001; possible CKD, 1-2 prescriptions: HR = 1.11, 95% CI, 0.70-1.76; ≥60 prescriptions: HR = 1.97, 95% CI, 1.34-2.90; P for trend < .001). There was no such association with antipsychotics or antidepressants. Also in patients with bipolar disorder, use of lithium was not significantly associated with an increased rate of end-stage CKD, whereas use of anticonvulsants was (1-2 prescriptions, HR = 0 [95% CI, 0.00-infinity]; 30-39 prescriptions: HR = 3.23, 95% CI, 1.26-8.27; ≥60 prescriptions: HR = 2.06, 95% CI, 0.82-5.16; P for trend = .002).

Conclusions and Relevance  Maintenance treatment with lithium or anticonvulsants as practiced in modern care is associated with an increased rate of CKD. However, use of lithium is not associated with an increased rate of end-stage CKD. The associations between use of medication and CKD may at least partly be attributed to bias.

Lithium has for more than 60 years been the main mood stabilizing treatment for bipolar disorder, and the evidence for a prophylactic effect of lithium is strong^1,2 and still increasing.^3,4 Unfortunately, the potential adverse effects of lithium have always been an issue; among these, the possible effect of long-term lithium treatment on renal function has given rise to considerable concern.^5,6 With results from the 3 most recent and largest controlled studies,^7-9 it is still controversial whether the protracted use of lithium can cause progressive deterioration in renal function, culminating in renal failure. The first of these studies, based on Swedish data,⁷ concluded that modern lithium treatment within recommended serum levels may have eliminated the risk of lithium-induced end-stage renal disease,¹⁰ whereas 2 UK studies found that lithium was associated with an approximately 2-fold increased risk of renal failure.^8,9 However, detection bias is an important issue to consider as patients exposed to lithium may be offered intensive screening of their creatinine level in contrast to patients treated with other psychotropics. To our knowledge, no longitudinal study has investigated whether the risk of chronic kidney disease (CKD) is specifically confined to lithium or whether other drugs used for long-term maintenance treatment for bipolar disorder such as anticonvulsants and antipsychotics also bear an increased risk of renal failure, as in fact suggested by a few cross-sectional studies.^11,12 The aims of the present study were to compare rates of CKD and, to minimize the risk of detection bias, in particular rates of irreversible end-stage CKD (defined as long-term dialysis or renal transplantation) among individuals exposed to lithium, individuals exposed to anticonvulsants and other drugs used for bipolar disorder, and individuals unexposed to these drugs.

Data were obtained by linking Danish population-based registers using the unique personal identification number assigned to all 5.4 million persons living in Denmark.¹³ In this way, the Register of Medicinal Product Statistics¹⁴ was linked with the Danish Medical Register on Vital Statistics,¹⁵ the Danish National Hospital Register,¹⁶ the Danish Psychiatric Central Register,¹⁷ and the Danish National Register on Regular Dialysis and Transplantation.¹⁸ Details on data from the 4 registers are provided in eAppendix 1 in the Supplement.

The study was approved by the Danish Data Protection Agency. Informed consent was not required because the study used deidentified data from large databases.

The study included 2 cohorts. Cohort 1 (n = 1 800 591) consisted of the following: (1) a randomly selected sample of 1 500 000 individuals among all persons who were registered in Denmark on January 1, 1995; (2) all patients having their first psychiatric contact ever (as inpatients or outpatients) between January 1, 1994, and December 31, 2012, and receiving a main diagnosis of a single manic episode or bipolar disorder (n = 10 591) (International Statistical Classification of Diseases and Related Health Problems, Tenth Revision [ICD-10] codes DF30-31.9 plus 38.00) at that occasion; and (3) all persons exposed to either lithium (n = 26 731) or anticonvulsants (n = 420 959) identified on the date of their first prescription of either drug between January 1, 1995, and December 31, 2012. Cohort 2 is the subcohort of 10 591 patients diagnosed as having bipolar disorder. Thus, the analyses of cohort 1 investigated the effect of drugs regardless of the indication.

Patients were excluded if they had a hospital contact as inpatient or outpatient with a diagnosis of renal dysfunction of any kind (ICD-8 codes 580-591 and 593, ICD-10 codes N00-N19 and N25-N29) prior to inclusion in the study.

Detection bias may be more prevalent when CKD is the outcome measure but substantially less when end-stage CKD (dialysis or transplantation) is the outcome measure since almost all patients with end-stage illness will be diagnosed in Denmark and referred for dialysis or kidney transplantation. Consequently, we chose 2 main outcome measures: (1) first hospital contact with a discharge diagnosis of CKD defined in a narrow way, termed definite CKD (ICD-10 codes N18-N19.9 inclusive plus N14.1, N14.2, N16.8, N25.1, N26, and N27); and (2) end-stage CKD, defined as irreversible end-stage CKD with either dialysis or transplantation, extracted from the Danish National Register on Regular Dialysis and Transplantation as the first entry (if any) dated after January 1, 1995.

For sensitivity analysis, we further included a third and broad definition of CKD termed possible CKD (ICD-10 codes N18-N19.9 inclusive plus N00, N01, N03, N04, N05, N06, N8.8 plus N14.1, N14.2, N16.8, N17, N25.1, N26, and N27). For a detailed description of N diagnoses, see eAppendix 2 in the Supplement.

Poisson regression analyses used diagnosis of CKD (narrow and broad definition) and end-stage CKD as separate outcomes and with the number of prescriptions of lithium and the number of prescriptions of anticonvulsants as the main variables of interest. At any time in the study period from January 1, 1995, to December 31, 2012, the prescription histories of lithium (Anatomical Therapeutic Chemical Classification System codes N05AN, N05AN01), anticonvulsants (N03), antipsychotics (N05A), and antidepressants (N06A) were summarized as number of prescriptions and thereafter grouped (0, 1-2, 3-9, 10-19, 20-29, 30-39, 40-59, ≥60 prescriptions). History of any other prescription was summarized as 0 (no prescription since January 1, 1995) or 1 (≥1 other prescription since January 1, 1995).

We intended to include first-time users of lithium and anticonvulsants. Consequently, events and person-years from 1995 were not included to avoid misclassification of the first prescription period for lithium or anticonvulsants due to lack of information before the start of the registry. Person-years and diagnoses of CKD and end-stage CKD were counted in categories defined by calendar year (3-year categories), bipolar diagnosis or not, current age (5-year categories), sex, employment status (employed or student vs unemployed, disability or age pension, or other), use of all other kinds of medication (including somatic medication) (other medication was summarized as 0 or 1), and prescription history.

Poisson regression analyses were conducted both in cohort 1 and in the subcohort of patients diagnosed as having bipolar disorder (cohort 2).

Analyses were performed to estimate the rate of CKD in all individuals by the number of prescriptions of lithium, anticonvulsants, antipsychotics, and antidepressants. In these analyses, we defined 1 or 2 prescriptions as the reference to reduce potential confounding by indication as patients prescribed a medication have a disorder that may be associated with CKD itself.

In analyses for cohort 2, we defined the unexposed period (0 prescriptions) as the reference. Follow-up was terminated at the date of death or the end of the study period (December 31, 2012), whichever came first. Hazard ratios (HRs) and 95% confidence intervals are reported and trend tests of cumulative numbers of prescriptions were considered the primary analysis to reduce overinterpretation of a positive finding due to multiple testing.

Sensitivity analyses and calculations of cumulative incidences of definite CKD are described in eAppendix 3 in the Supplement.

Characteristics of cohort 1 (n = 1 800 591) are presented in eAppendix 4 in the Supplement. Table 1 presents HRs for definite CKD, possible CKD, and end-stage CKD related to the number of prescriptions of lithium, anticonvulsants, antipsychotics, and antidepressants and adjusted for age, sex, employment status, calendar year, a diagnosis of bipolar disorder, and use of all other kinds of medication. Patients with a main diagnosis of a single manic episode or bipolar disorder had increased rates of CKD compared with individuals without a diagnosis and regardless of outcome measure, but this was most pronounced in relation to end-stage CKD (HR = 2.47 [95% CI, 1.82-3.36]).

In cohort 1, 14 727 individuals (0.8%) were diagnosed as having definite CKD, 18 762 (1.0%) as having possible CKD, and 3407 (0.2%) as having end-stage CKD. Table 1 shows that rates of all 3 outcome measures increased with the number of prescriptions for lithium but not for anticonvulsants, antipsychotics, or antidepressants. Trend tests confirmed statistically significant associations between the number of prescriptions of lithium and the rate of definite CKD (0 prescriptions: HR = 1.09 [95% CI, 0.81-1.45]; ≥60 prescriptions: HR = 3.65 [95% CI, 2.64-5.05]; P for trend < .001) and possible CKD (0 prescriptions: HR = 1.01 [95% CI, 0.79-1.30]; ≥60 prescriptions: HR = 2.88 [95% CI, 2.17-3.81]; P for trend < .001) but not for end-stage CKD (P for trend = .88).

Prescription of all other kinds of medication was associated with substantially increased rates of all 3 outcome measures, and women and employed individuals/students had decreased rates (Table 1).

Effect modification analyses using likelihood ratio tests showed no significant changes of the association between lithium and CKD with respect to simultaneous use of anticonvulsants regardless of the outcome measures (all P > .20).

Characteristics of cohort 2 (n = 10 591) are presented in eAppendix 5 in the Supplement. As can be seen from Table 2, findings in cohort 2 differed from findings in cohort 1. A total of 278 (2.6%) were diagnosed as having definite CKD, 319 (3.0%) as having possible CKD, and 62 (0.6%) as having end-stage CKD. Rates of definite CKD and possible CKD did increase not only with the number of prescriptions of lithium (definite CKD, 1-2 prescriptions: HR = 0.89 [95% CI, 0.39-2.06]; ≥60 prescriptions: HR = 2.54 [95% CI, 1.81-3.57]; P for trend < .001; possible CKD, 1-2 prescriptions: HR = 1.26 [95% CI, 0.65-2.43]; ≥60 prescriptions, HR = 2.48 [95% CI, 1.80-3.42]; P for trend < .001) but also for the number of prescriptions of anticonvulsants (definite CKD, 1-2 prescriptions: HR = 1.23 [95% CI, 0.76-1.99]; ≥60 prescriptions: HR = 2.30 [95% CI, 1.53-3.44]; P for trend < .001; possible CKD, 1-2 prescriptions: HR = 1.11 [95% CI, 0.70-1.76]; ≥60 prescriptions: HR = 1.97 [95% CI, 1.34-2.90]; P for trend < .001) and with rather similar effect sizes. All trend tests showed statistically significant associations. The rates of end-stage CKD did not increase with the number of lithium prescriptions (P for trend = .30), whereas they significantly increased with the number of prescriptions of anticonvulsants (1-2 prescriptions: HR = 0 [95% CI, 0.00-infinity]; 30-39 prescriptions: HR = 3.23 [95% CI, 1.26-8.27]; ≥60 prescriptions: HR = 2.06 [95% CI, 0.82-5.16]; P for trend = .002).

There were no statistically significant associations between the number of prescriptions of antipsychotics or antidepressants and CKD for any of the 3 outcome measures. As in cohort 1, women and employed individuals/students systematically had decreased rates (adjusted for all other variables in Table 2).

Also, the effect of cumulative lithium treatment on the rate of definite CKD was not significantly modified by treatment with anticonvulsants (P > .05). The linear trend associated with cumulative lithium exposure and rate of definite CKD was significantly different in men with bipolar disorder (HR = 1.31 [95% CI, 1.16-1.48]) and women with bipolar disorder (HR = 1.08 [95% CI, 0.98-1.19]) (P = .01). There were no significant modifications of linear trends between cumulative lithium exposure and any of the 3 outcome measures by age (<50 vs ≥50 years of age) or employment status (employed/student vs all others) (all P > .05).

Table 3 shows that rates of definite CKD and possible CKD increased with the number of prescriptions of carbamazepine and the group all other anticonvulsants but not clearly with the number of prescriptions of the other anticonvulsants (lamotrigine, clonazepam, or valproate sodium).

Figure 1 and Figure 2 show the cumulative incidences of definite CKD starting at the 30th, 45th, 60th, and 75th birthday according to ever previous treatment with lithium and anticonvulsants, respectively.

To our knowledge, this is the first nationwide population-based longitudinal study including data on all individuals who, during a period of up to 17 years, were prescribed lithium, anticonvulsants, or other drugs for bipolar disorder as well as all data on CKD and death. We found that patients diagnosed as having mania or bipolar disorder had increased rates of definite, possible, and end-stage CKD that persisted after adjustment for treatment with lithium, anticonvulsants, antipsychotics, antidepressants, and all other types of drugs. Given unknown indications for treatment, use of lithium was associated with an increased rate of definite and possible CKD, whereas use of anticonvulsants, antipsychotics, or antidepressants was not; neither use of lithium nor use of any other drug class was associated with increasing rates of end-stage CKD. In the subcohort of patients diagnosed as having bipolar disorder, treatment with lithium and treatment with anticonvulsants were associated with increased rates of definite and possible CKD and at a similar magnitude (up to 2-fold), whereas there was no such association regarding treatment with antipsychotics or antidepressants. In this subcohort, rates of end-stage CKD did not increase with the number of lithium prescriptions but did increase with the number of prescriptions of anticonvulsants. The association between lithium and CKD was not modified by concomitant use of anticonvulsants (ie, combination therapy), regardless of outcome measure. Continued treatment with carbamazepine and the group all other anticonvulsants was associated with increased rates of definite and possible CKD.

The findings should be evaluated in the context of the observational design of the study. Even detection bias might contribute to the result that patients diagnosed as having bipolar disorder had increased rates of definite, possible, and end-stage CKD since patients with bipolar disorder regardless of treatment will often undergo screening for somatic diseases, and there likely may be more profound links between bipolar disorder and CKD. To our knowledge, the association between bipolar disorder and CKD has not been studied before except in 1 other study that also found a similar 2-fold increased risk of CKD in patients with bipolar disorder¹⁹ (odds ratio = 2.42 [95% CI, 2.04-2.86]; P < .001), although these results, in contrast to ours, were not adjusted for other pharmacological treatment. Potential pathogenetic mechanisms involved in the association between bipolar disorder and CKD are discussed in eAppendix 6 in the Supplement. In relation to the analyses leading to the result that lithium was associated with an increased rate of definite and possible CKD but use of anticonvulsants, antipsychotics, or antidepressants was not, we aimed to reduce confounding by indication by defining 1 or 2 prescriptions as the reference because lithium is prescribed mainly for bipolar disorder in contrast to the other drugs; the increasing risk of CKD associated with an increasing number of lithium prescriptions indicates detection bias more than confounding by indication, if not indicating a causal link. Likewise, regarding the result that lithium and anticonvulsants, but not antipsychotics or antidepressants, were associated with increased rates of definite and possible CKD in the subcohort of patients with bipolar disorder, the increased rate of CKD associated with an increasing number of lithium prescriptions in patients with bipolar disorder may in part be explained by detection bias, whereas this is less likely to explain the increased rate associated with the use of anticonvulsants. This latter finding together with the result in this subcohort that rates of end-stage CKD did not increase with the number of lithium prescriptions but did increase with the number of anticonvulsant prescriptions could be due to another kind of selection bias, for example, patients with decreased renal function will less likely begin treatment with lithium and will more likely begin treatment with anticonvulsants. However, the observed increase in the rate of renal failure with the number of prescriptions of anticonvulsants might indicate a causal relationship. This is supported by the result that the association between lithium and CKD was not modified by concomitant use of anticonvulsants, indicating that the rate of definite CKD associated with anticonvulsants was independent of lithium treatment. On the other hand, it can be argued that a kidney disease detected early in a mild phase simply may progress over time parallel to the accumulating treatment with anticonvulsants over time.

Although we found no statistically significant association between the number of lithium prescriptions and end-stage CKD (trend test), within the limitations of our study and the relatively low number of patients with end-stage CKD, this may be a false-negative finding. To this end, note that nearly all HRs for more than 1 or 2 lithium prescriptions were greater than 1, possibly suggesting increased risk of end-stage CKD (Table 1). However, in our interpretation we emphasize the results of the trend test owing to the multiple testing.

To our knowledge, the association between anticonvulsants and CKD has not been systematically investigated previously in human or animal studies; however, a recent cross-sectional study found that among 25 potential correlates, treatment with valproate and other unspecified anticonvulsants was more prevalent among patients with CKD than among patients without CKD.¹² Biologically, valproate seems to affect nitric oxide production from endothelial cells,²⁰ and endothelial dysfunction may play a central role in development of CKD.²¹ We know of no reports on chronic renal failure associated with lamotrigine, carbamazepine, or any other anticonvulsant. It should be noted that our analyses on selected anticonvulsants were exploratory, decided post hoc, and without corrections for multiple testing—increasing the risk of chance findings—while all other analyses were planned a priori.

Potential pathogenetic mechanisms involved in the association between anticonvulsants and CKD are discussed in eAppendix 7 in the Supplement.

In human studies, antipsychotics have been associated with higher rates of tubular dysfunction¹¹ and CKD.¹² A recent animal study found structural changes of the glomerulus associated with haloperidol.²² Nevertheless, our findings from this first-ever longitudinal study consistently refute an association between use of antipsychotics and CKD, both in general and in patients with bipolar disorder. This result is in accordance with the finding in the first study in schizophrenia showing no increased risk of CKD with the use of atypical or typical antipsychotics.²³

The conclusion from our study is in accordance with the Swedish case-control study concluding that modern lithium treatment within recommended serum levels between 0.6 and 0.8 mEq/L (to convert to millimoles per liter, multiply by 1.0) has eliminated the risk of lithium-induced end-stage renal disease, supporting the continued use of lithium as a safe drug for the long-term treatment of mood disorders.¹⁰ The UK study⁸ from 2014 is, besides our study, the only one that adjusts the analyses of the effect of lithium for the effect of concomitant use of other medication, including other psychotropics. As shown in our study, the variable of all other medication including all medication for all physical illnesses was substantially associated with increased rates of definite, possible, and end-stage CKD (with 4 to 6 times increased rates; Table 1). Accordingly, medication such as loop diuretics and hydrochlorothiazide have been found to be independently associated with CKD¹² and it is well known that treatment with nonsteroidal anti-inflammatory drugs, COX-2 inhibitors, and hydrochlorothiazide may cause an increase in serum lithium levels, potentially leading to long-term adverse effects in the kidney.²⁴ Nevertheless, taking the high HRs of CKD found in our study into account, it is likely that a number of unidentified drugs may cause CKD. The second UK study⁹ found higher serum lithium concentrations associated with stage 3 CKD and it is not clear whether the overall finding of a nearly 2 times increased rate of kidney disease associated with lithium use was driven by higher lithium dosing as the study included a study period back to 1982. Additionally, because patients receiving lithium most likely had their serum creatinine level measured more frequently and during longer periods than those not receiving lithium, detection bias may also be a major explanation. Further, the study included no data on psychiatric diagnosis and did not take into account the effect of somatic illnesses and medication in general or the effect of psychotropics other than lithium as such data were not available.⁹ Other limitations in prior studies are discussed in eAppendix 8 in the Supplement.

As with all other registers including nationwide medication data, the Danish Register of Medicinal Product Statistics includes no information on dose of lithium¹⁴ and we had no information on serum levels of lithium or creatinine or other measures of kidney function. Further, although this was, to our knowledge, the largest sample ever, it included few events of end-stage CKD as illustrated by the wide confidence intervals in these analyses, potentially resulting in type II errors in these analyses.

Our results indicate that bipolar disorder is associated with CKD independent of drug treatment. Additionally, in patients with bipolar disorder, CKD is associated not only with lithium but also with anticonvulsants, with the latter in fact being associated with increased end-stage CKD. In contrast, long-term maintenance treatment with lithium as practiced in Denmark during 2 recent decades with initial and regular monitoring of the serum creatinine level every 3 to 6 months and aiming for a serum lithium level of 0.6 to 0.8 mEq/L is not associated with end-stage CKD. It cannot be excluded that at least part of the associations between medication and CKD is a result of bias.

Corresponding Author: Lars Vedel Kessing, MD, DMSc, Psychiatric Center Copenhagen, Rigshospitalet, 6233, Blegdamsvej 9, 2100 Copenhagen, Denmark (lars.vedel.kessing@regionh.dk).

Submitted for Publication: June 18, 2015; final revision received August 10, 2015; accepted August 11, 2015.

Correction: This article was corrected on February 3, 2016, to fix an error in the author affiliations.

Published Online: November 4, 2015. doi:10.1001/jamapsychiatry.2015.1834.

Author Contributions: Dr Kessing had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: All authors.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Kessing, Andersen.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Kessing, Gerds, Feldt-Rasmussen, Andersen.

Obtained funding: Licht.

Administrative, technical, or material support: Kessing, Feldt-Rasmussen.

Study supervision: Kessing, Andersen.

Conflict of Interest Disclosure: Dr Kessing has within the preceding 3 years been a consultant for Lundbeck and AstraZeneca. Dr Feldt-Rasmussen has within the past 3 years been a consultant for Novo Nordisk. Dr Licht has within the preceding 3 years served on an advisory board of Janssen-Cilag and served as a consultant for AstraZeneca, Janssen-Cilag, Servier, Bristol-Myers Squibb, Lundbeck, and Otsuka. No other disclosures were reported.

Funding/Support: This work was supported by the Department of Psychiatry, Aalborg University Hospital.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.