Lithium prescription in mood disorders and other diagnosis groups

The present study demonstrates that utilization of Li in clinical practice far exceeds acute and prophylactic treatment in BD and UD. Almost one in three patients with Li prescription did not have diagnosis of BD or UD. More than one in five patients with Li prescription had a diagnosis of SAD or SCZ, even though evidence for effectiveness of Li use in monotherapy or as an adjunct to antipsychotics in SCZ is scarce (Leucht et al. 2015) and 2019 German S3 guidelines for SCZ, which no longer recommend Li for treatment of depressive symptoms in SCZ, even advise against its use as an augmentative in standard treatment for improvement of general or affective symptoms and aggression (DGPPN e.V. (Hrsg.) 2019). Clinicians might use Li in a symptom-oriented approach in patients with mood disorders that do not meet criteria for diagnosis of BD or UD. Another rationale, especially for administration of Li in patients with specific personality disorders or intellectual disability, could be the positive effects of Li on impulsive and violent behavior (Müller-Oerlinghausen and Lewitzka 2010).

Lithium in bipolar disorder

Internationally, prescription rates and trends in prescription differ widely (Singh et al. 2023). For example, in Sweden, Li prescription rates in outpatients with BD decreased from 51% in 2007 to 41% in 2013 (Karanti et al. 2016), while in the United States, data from the National Ambulatory Medical Care Surveys shows that Li prescription rates in BD decreased from 30.4% in 1997 to 17.6% in 2016 (Rhee et al. 2020).

Our study’s overall Li prescription rate of 31.3% in patients with BD seems to support previous findings about a steady decline in Li utilization in BD in Germany, as an analysis of data from the German Drug Safety Program in Psychiatry (AMSP) showed a decrease in Li prescription rates in selected German hospitals from 44.8% in 1994 to 34.4% in 2009 (Greil et al. 2012).

The AMSP Drug Safety project is designed similar to the Pharmako-EpiVig project and also collects data about drug use and sADRs on two reference days a year from more than 30 hospitals in Germany, Switzerland and Austria (Grohmann et al. 2004).

However, prescription rate in BD over the course of our study significantly increased from 28.8% in 2014 to 34.4% in 2019. Only after 2019 the data shows a decline in prescription rate to 30.8% in 2021.

This decline in prescription numbers after 2019 must be considered in light of the COVID-19 pandemic, as the pandemic and consequent policy responses influenced access to health care services. Because of the pandemic, in early 2020 there was a limited availability of inpatient treatment capacity and outpatient services in psychiatric hospitals in Germany (Adorjan et al. 2021). In Li treatment, unhindered access to health care services is a prerequisite, due to the need for regular blood level monitoring. Practitioners as well as patients might therefore have been reluctant to initiate or continue Li treatment because of restrictions associated with the COVID-19 pandemic.

Despite availability of alternative treatment options, like second generation antipsychotics and other anticonvulsants, Li should be strongly considered for every patient with BD in absence of contraindications. Li is still the only drug with a level A recommendation as a mood stabilizer in long-term treatment of BD by the 2019 German S3 guidelines for BD and evidence regarding its efficacy in treatment of bipolar disorder has been strengthening over the last decade ( DGBS e.V. und DGPPN e.V. 2019; Severus et al. 2014; Bschor et al. 2020). Additionally, its unique anti-suicidal effects can mitigate the risk for suicide in patients with major affective disorders (Cipriani et al. 2013; Lewitzka et al. 2015; Tondo et al. 2001).

Since this study includes patients at all time points of course of disease, a considerable number of patients that were not prescribed Li on the reference day, might have been treated with Li in the past and have discontinued treatment due to side effects, personal preferences or poor response. Depending on clinical, biological and genetic features, inter-individual response in Li varies substantially, full response is observed in about 30% of patients with BD (Hou et al. 2016; Tighe et al. 2011; Bauer and Gitlin 2016b). Therefore, comparisons to findings about prescription rate of Li in other studies must be done cautiously.

Lastly, changes in prescription rates over the observed period might also be influenced by differences in the study population at given reference days, i.e. proportion of patients with manic vs. depressive symptoms or type I vs. type II subtypes.

Lithium in UD

Overall, Li prescription rate in UD was 4.6% and remained stable over the observed time period. AMSP analysis showed similar prescription rates of about 4.9% in the years 2015 to 2017 (Seifert et al. 2021).

Augmentation treatment, as an add-on to an antidepressant in treatment-resistant depression is Li’s main indication in UD. German 2022 S3 guidelines on UD and several international guidelines, recommend Li as first-line therapy in treatment-resistant depression (Bundesärztekammer (BÄK) et al. 2022; Taylor et al. 2020). However, recent studies and metanalyses confirm profound efficacy and safety of several second-generation antipsychotics as an alternative in adjunctive therapy, on a similar or even superior evidence base when compared to Li, contributing to a decline in Li prescription (Marcus et al. 2008; Nuñez et al. 2022).

In contrast to numerous publications about Li prescription practice in BD, studies about prescription practice of Li in UD are hardly available. This restricts interpretation of our findings and demonstrates further need of research in this area.

Lithium in older patients

Patients older than 65 years had significantly lower probability to be treated with Li in BD, SCZ and SAD. Lower utilization of Li in older patients in BD has been reported before (Rej et al. 2017).

In old age Li is still considered first choice for maintenance-treatment in BD (Volkmann et al. 2020). Li not only mitigates the manifold increased risk of suicide in patients with BD, but also reduces excess cardiovascular mortality (Lewitzka et al. 2015; Ahrens et al. 1995). Long-term treatment reduces frequency of psychiatric as well as somatic hospitalization (Lähteenvuo et al. 2023).

Patients with BD suffer from extensive medical comorbidity, standardized mortality is about twice as high as in the general population (Westman et al. 2013; Walker et al. 2015). Therefore, treatment options in older patients can be limited by somatic comorbidity (e.g. kidney failure) due to reduced drug tolerability and altered drug-metabolism. Also there might be a reluctance to prescribe Li in elderly patients due to increased vulnerability to Li intoxication, even though this risk can be mitigated by lowering dosage and intensifying serum level controls (Gitlin 2016b). Lastly, since onset of BD, SCZ and SAD is typically in young age, older patients might simply have been treated with Li in the past and discontinued it due to poor response or side effects. Studies suggest that in long-term treatment up to more than half of patients with BD ore SAD decide to discontinue Li at some point (Öhlund et al. 2018).

It is notable, that in UD, old age did not have an influence on prescription rate.

Patients with comorbid substance use disorders

Concomitant substance use disorders are common in severe psychiatric disorders (Singh et al. 2023; Davis et al. 2008). To our best knowledge, there are no previous studies about Li prescription rate in patients with comorbid substance use disorders. Knowledge about Li therapy in patients with comorbid substance use is limited, since substance use disorders are common exclusion criteria in randomized trials. While some studies associated alcohol use disorders with poor response to Li (Sportiche et al. 2017; Grillault Laroche et al. 2020), a recent systematic review concludes that valproate and lamotrigine should preferably be used in BD with concomitant substance abuse disorder but emphasizes poor quality of evidence and need for further research (Coles et al. 2019).

Even though substance use disorders are linked to poor adherence in patients with severe mental illness (García et al. 2016), and therefore this subgroup of patients might be more exposed to the specific risks of poor adherence such as Li toxicity and rebound suicidality after abrupt Li cessation, prescription rate of Li in patients with comorbid substance use disorder in our study was only significantly lower in UD.

Comorbidities that constitute relative or absolute contraindication for lithium prescription

Prevalence of somatic comorbidities that constitute relative or absolute contraindications for Li prescription was 1.5% in patients with Li prescription and 2.2% in the study population. Due to differences in psychiatric morbidity and consequent sociodemographic heterogeneity, comparison between those groups is only reasonable to a very limited extend. Reported absolute contraindications (two cases of acute myocardial infarction and four cases of acute renal failure), were not reported as sADRs, so it is probable that time of diagnosis preceded the day of the survey by at least 2 weeks. Most prevalent contraindications were acute or chronic renal failure and psoriasis.

Renal failure and psoriasis can be caused and aggravated by Li treatment, but neither is a determinant reason to discontinue Li treatment. An option for managing psoriasis can be lowering dosage of Li. In most cases Li-associated renal effects are relatively mild and progressive renal impairment due to long-term Li use can be monitored by regular blood tests. Psychiatric disorders are no less debilitating in terms of quality of life and mortality compared to numerous other chronic medical conditions. A prime example is rheumatic diseases. In this case as well, within the framework of guideline-compliant treatment, the risks and benefits (e.g., with chemotherapy) need to be carefully weighed against the potential side effects. Li treatment also requires a constant monitoring of response and side effects. In absence of response Li should be discontinued. In patients that benefit from Li therapy, it is not a trivial task to weigh the relief of highly debilitating, sometimes life-threatening affective symptoms against the risks of aggravation of already perceivable long-term side effects, that might lead to severe disability like need for lifelong renal replacement therapy (Gitlin 2016b; Jafferany 2008; Tondo et al. 2017).

Comedication with risk of drug–drug interactions

Patients with Li prescription were prescribed significantly more drugs than patients without Li prescription. As mentioned above, due to differences in psychiatric morbidity and consequent sociodemographic heterogeneity comparison between those groups is limited. Almost two out of three patients with Li prescription were prescribed five drugs or more simultaneously.

For classification of drug–drug interactions with Li, we used the mediQ database which in a recent study has been evaluated as the most suitable interaction database for psychopharmacotherapy (Hahn and Roll 2018).

Prescribed drugs with highly relevant drug–drug interactions (n = 178) were the diuretics hydrochlorothiazide (n = 157), indapamide (n = 4) and chlortalidone (n = 1), and the antihypertensive olmesartan (n = 16). Since these diuretics as well as olmesartan increase Li blood levels by increased reuptake in the kidneys, acute and chronic toxicity of Li can be increased. However, with reduction of dosage and regular monitoring of serum concentration, this interaction can be well managed (Malhi et al. 2020).

Severe adverse drug reactions with lithium as the reported probable causative agent

Even though underreporting cannot be ruled out, considering the total number of 4543 patients treated with Li, a figure of 19 patients with sADRs seems comparatively low. This suggests a high level of safety for Li therapy in inpatient settings.

Almost one in three patients with reported sADRs had diagnosis of comorbid substance use disorder. This suggests that patients with comorbid substance use disorder are more vulnerable to adverse drug reactions. Almost three in four patients were prescribed at least two comedications with intermediate- or high-priority drug–drug interaction with Li. Four patients were reported to have lithium intoxications with levels ranging from 1.3 up to 2.2 mmol/l. Of these, one patient, who had received hydrochlorothiazide as comedication, had to be transferred to the intensive care unit. The other patients received blood pressure drugs such as ACE inhibitors or sartans. One patient in which Li had to be discontinued due to symptoms of Li intoxication (restlessness/agitation) was prescribed Flupentixol. Some studies suggest that phenothiazines might increase intracellular Li concentration (Pandey et al. 1979).

Tremor is the most common reason for Li discontinuation, therefore it is important to discuss this and other common side effects, like weight gain before the start of treatment, so counter measures can be taken, and unnecessary discontinuation prevented (Öhlund et al. 2018; McCreadie et al. 1985).

On the other hand, it should also be mentioned that many alternatives to Li in treatment of affective disorders have their own inherent side effect profile. Alternatives for maintenance-treatment like valproate, olanzapine or quetiapine, for example, have much more unfavorable metabolic side effects than Li (Greil et al. 2023).

While it is known that multiple drugs have additive effect on side effect rate in Li therapy (Gitlin 2016b), it is still remarkable that in all cases of the most severe sADRs, comedications with drugs that increase Li serum levels, namely hydrochlorothiazide, ramipril, enalapril, valsartan and candesartan were prescribed. These findings emphasize the significance of attention to drug–drug interaction in psychopharmacology in general and Li therapy in particular.

Strengths and limitation

Data for this study was collected in a repeated cross-sectional approach, therefore the study design does not allow to draw conclusions about causal relationship of findings. No information was available about patient’s treatment history or course of diseases restricting interpretation of findings. Due to the use of ICD-10 coding, a distinction between bipolar 1 and bipolar 2 disorders was not made. Consequently, separate analyses for both conditions regarding prescription frequency were not conducted. Furthermore, the diagnosis F30 was excluded from definition of BD. Since Li is also approved for the treatment of mania in Germany, it is possible that this may have led to an overall underestimation of prescription rates. Because of to the design of this deadline survey, which only captures severe adverse events appearing within the two weeks before the deadline the incidence of severe adverse events may be underestimated. Since the source of this study is an observational database, there is possibility of underreporting, missing or incorrect information about diagnoses, drug prescriptions or sADRs. Due to the naturalistic nature of the data and the exploratory approach of the analysis, we refrained from using more elaborate statistical methods, which also includes correction for multiple testing.

However, the quasi-naturalistic setting and inclusion of all patients present at the included hospitals, made it possible to demonstrate the full picture of current Li prescription practice in inpatient care in Bavaria.