Introduction
Bipolar disorders (BD) are severe mental illnesses characterized by alternating episodes of (hypo)mania and depression, affecting approximately 1% of the global population [Reference Merikangas, Jin, He, Kessler, Lee and Sampson1]. These conditions are associated with significant functional and markedly elevated mortality risk, about 2.6 times higher than the general population, with an average reduction in life expectancy of 7 years [Reference Oliva, Fico, De Prisco, Gonda, Rosa and Vieta2, Reference Kessing, Ziersen, Andersen and Vinberg3]. Both unnatural causes (e.g., suicide) and natural causes (e.g., cardiovascular and metabolic comorbidities) contribute to this mortality gap, highlighting the need for effective long-term management strategies that address both mood symptoms and survival [Reference Yatham, Kennedy, Parikh, Schaffer, Bond and Frey4].
Lithium, a cornerstone in BD treatment, is well known for its mood-stabilizing and anti-suicidal properties [Reference Smith and Cipriani5]. Beyond these effects, emerging evidence suggests a broader protective role in reducing overall mortality. Meta-analyses of randomized controlled trials and longitudinal studies have identified lithium as the mood stabilizer associated with the lowest all-cause mortality risk [Reference Cipriani, Pretty, Hawton and Geddes6–Reference Chen, Tsai, Chen, Pan, Su and Chen8], including in first-episode and severe BD cases [Reference Carvalho, Hsu, Vieta, Solmi, Marx and Berk9, Reference Toffol, Hätönen, Tanskanen, Lönnqvist, Wahlbeck and Joffe10].
Some studies have also suggested that longer lithium treatment duration and higher cumulative exposure are associated with lower mortality risk [Reference Chen, Tsai, Chen, Pan, Su and Chen8]. However, most studies have classified patients simply as lithium-exposed or unexposed, without considering treatment continuity. Discontinuation is linked to increased relapse [Reference Baldessarini, Pinna, Contu, Vázquez and Tondo11] and suicidality [Reference Tondo and Baldessarini12], but its impact on mortality remains understudied. One large nationwide study found higher mortality among patients who discontinued lithium compared to those who discontinued valproate, suggesting that lithium discontinuation may be especially detrimental [Reference Smith, Austin, Kim, Eisen, Kilbourne and Miller13].
To address this gap, we examined the association between lithium exposure patterns and all-cause mortality in a large population-based cohort of individuals with BD in Catalonia, Spain. By distinguishing sustained, partial or intermittent, and no exposure, we aimed to clarify the role of treatment continuity in survival, extending our understanding of lithium’s protective effects beyond suicide prevention to long-term mortality outcomes.
Materials and methods
Study design and population
This prospective cohort study utilized data from the “Programa d’analítica de dades per a la recerca i la innovació en salut” (“Data Analytics Program for Health Research and Innovation” [PADRIS]), managed by the Agency for Health Quality and Assessment of Catalonia (AQuAS). PADRIS integrates anonymized sociodemographic, clinical, and healthcare data from the public health system in Catalonia (CatSalut), in line with legal, ethical, and transparency standards [14]. Since most of the Catalan population uses public healthcare, PADRIS provides a highly representative sample. All data are fully anonymized, with no possibility of patient identification. The study was approved by the Hospital Clínic of Barcelona Ethics Committee (HCB/2020/0735).
The PADRIS dataset included 473,812 individuals who accessed public specialized mental health services (outpatient, inpatient, or emergency) between January 1, 2015, and December 31, 2019. Retrospective follow-up extended back to January 1, 2010, using data from electronic health records for the entire period. Information was anonymized and de-identified in full compliance with legal and ethical standards. Additional details are reported elsewhere [Reference Mas, Clougher, Anmella, Valenzuela-Pascual, De Prisco and Oliva15–Reference Anmella, Primé-Tous, Segú, Solanes, Ruíz and Martín-Villalba17].
This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for reporting observational research [Reference Cuschieri18].
Exposure and outcomes
We included patients diagnosed with BD according to the International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10). Based on lithium exposure, patients were categorized into three groups: sustained lithium exposure, partial or intermittent lithium exposure, and no lithium exposure. Lithium exposure was measured in defined daily doses (DDDs), as established by the WHO Collaborating Center for Drug Statistics Methodology, where one DDD for lithium corresponds to 24 mmol (approximately 900 mg) of lithium carbonate, the average maintenance dose per day for its main indication in adults [Reference Grandjean and Aubry19–Reference Ronning21]. In Spain, lithium is typically dispensed in 400 mg tablets, resulting in common daily doses of 800 mg. A patient taking 800 mg daily for a year would accumulate roughly 324 DDDs. Since DDDs in our dataset were recorded in increments of 50, we defined sustained use as ≥350 DDDs per year, allowing for minor interruptions while reflecting consistent maintenance therapy over ≥11.5 months annually. Patients with sustained lithium exposure consistently met this threshold from their first prescription until the end of follow-up (2019) or death. For individuals who died during follow-up, the final year was exempt from the 350-DDD requirement, provided they received some lithium in that year. Partial or intermittent lithium exposure referred to patients who received lithium at some point during the study period but did not consistently reach the annual threshold of 350 DDDs. This group included individuals with irregular treatment patterns, such as interruptions between years, early discontinuation, or annual prescriptions below the defined threshold, resulting in short-term, sporadic, or low cumulative exposure. Patients with no lithium exposure had no recorded lithium prescriptions during the study period.
This classification aligns with previous approaches, adapting WHO DDD standards to psychiatric indications while acknowledging variations in clinical practice [Reference Kendler, Ohlsson, Sundquist and Sundquist22]. Applying a consistent threshold helped distinguish distinct exposure trajectories and reduce misclassification due to brief or sporadic use.
The primary outcome was time to all-cause mortality, defined as the period from January 1, 2015, to death or censoring on December 31, 2019, based on linked electronic health records.
Statistical analysis
Descriptive statistics were calculated to summarize baseline characteristics, using means and standard deviations (SDs) for continuous variables and frequencies and percentages for categorical variables. Baseline characteristics included age, sex, BD type, socioeconomic level, somatic comorbidities, and combination with other psychotropic medications (mood stabilizers, antidepressants, and antipsychotics).
To compare age across lithium exposure groups (sustained, partial or intermittent, and none), we performed one-way ANOVA, with eta squared (η²) as the effect size. Categorical variables were compared using chi-square tests, with Cramér’s V to quantify effect sizes. When associations were significant (p < 0.05), post-hoc pairwise comparisons were conducted: Tukey’s HSD for continuous variables and chi-square tests with Bonferroni correction for categorical ones.
Kaplan–Meier survival curves were generated to compare survival across exposure groups. The log-rank test was used to assess differences, with Bonferroni-adjusted pairwise comparisons.
Cox proportional hazards regression models estimated hazard ratios (HRs) and 95% confidence intervals (CIs) for the association between lithium exposure and all-cause mortality. Three models were applied: an unadjusted model including only lithium exposure; a minimally adjusted model controlling for age and sex; and a fully adjusted model accounting for sociodemographic factors (age, sex, socioeconomic level), clinical characteristics (BD type, and somatic comorbidities), and treatment variables (combination with other mood stabilizers, antidepressants, and antipsychotics). The proportional hazards assumption was verified using Schoenfeld residuals. A subgroup analysis compared sustained versus partial or intermittent lithium exposure. Sensitivity analyses tested alternative DDD thresholds (≥150, 200, 250, 300, and 400 per year) to assess the robustness of exposure classification. For each threshold, we re-ran Kaplan–Meier and fully adjusted Cox models.
All analyses were performed using R version 4.4.1.
Results
The completed STROBE checklist is available in the Supplementary Appendix (p. 2).
The cohort included 15,384 individuals with a diagnosis of BD. Table 1 summarizes the distribution of the variables considered in the sample. During the study period, 8,815 (57.3%) patients were classified as lithium non-users, 4,331 (28.15%) as discontinuous lithium users, and 2,238 (14.55%) as continuous lithium users.
Table 1. Baseline characteristics of the cohort of patients with bipolar disorders (n = 15,384)
Differences across the three groups are reported in Table 2. The one-way ANOVA indicated that there was no statistically significant difference in age between the groups. Significant differences between groups were found for sex distribution, BD type, socioeconomic level, presence of somatic comorbidities, and concomitant medications use. Pairwise comparisons between the three groups are reported in Table 3.
Table 2. Baseline characteristics of patients with bipolar disorders stratified by lithium exposure patterns
Table 3. Pairwise comparisons for categorical variables (chi-square tests)
Over the study period, 715 deaths were recorded in our cohort of patients with BD. Among these, 467 (65.3%) deaths occurred in patients with no lithium exposure, 199 (27.8%) in those with partial or intermittent exposure, and 49 (6.9%) in those with sustained lithium exposure. The overall mortality rate was 10.16 per 1,000 person-years (95% CI: 9.44–10.93). Mortality rates were 10.9 per 1,000 person-years (95% CI: 9.91–11.9) in the no-exposure group, 10.5 (95% CI: 9.11–12.0) in the partial or intermittent exposure group, and substantially lower at 5.87 (95% CI: 4.43–7.76) in the sustained exposure group.
Figure 1 presents Kaplan–Meier survival curves stratified by lithium use. The log-rank test demonstrated a significant difference among the three survival curves (χ2 = 14.8; p < 0.001). The pairwise log-rank test, adjusted for multiple comparisons using the Bonferroni method, revealed significant differences in survival curves. Specifically, patients with sustained lithium exposure showed significantly higher survival probabilities compared to those with no lithium exposure (p < 0.001) and those with partial or intermittent exposure (p = 0.001). No significant difference was observed between the partial or intermittent exposure group and the no-exposure group (p = 1).
Figure 1. Kaplan-Meier survival curves stratified by lithium exposure. Kaplan–Meier survival curves illustrating the probability of survival over time, stratified by patterns of lithium exposure: no exposure, partial or intermittent exposure, and sustained exposure. Survival probabilities are shown from January 1, 2015, with censoring events indicated by marks on the curves. Shaded areas represent 95% confidence intervals.
Table 4 presents the results of the Cox proportional hazards models examining the association between lithium use patterns and all-cause mortality. In the unadjusted model, sustained lithium use was associated with a significantly lower risk of mortality compared to no lithium exposure. This association remained significant after adjusting for age and sex, and persisted in the fully adjusted model accounting for sociodemographic, clinical, and pharmacological covariates. In contrast, discontinuous lithium use was not significantly associated with mortality in any model. Age was not a significant predictor of mortality in the adjusted models. Female sex was consistently associated with a lower mortality risk across both the minimally and fully adjusted models. Among covariates, the presence of somatic comorbidities emerged as the strongest predictor of increased mortality: individuals with one comorbidity had a more than threefold increased risk, while those with two or more comorbidities had over a 10-fold increase in risk. Antipsychotic use was also independently associated with a higher risk of death. Mid-level socioeconomic status was associated with a significantly lower mortality risk compared to the lowest-income group, whereas no significant difference was observed for the highest-income category. Other covariates, including BD type, antidepressant use, and mood stabilizer use, were not significantly associated with mortality. The model’s concordance index was 0.77, indicating good discriminative ability. The proportional hazards assumption was met (global test: χ² = 9.52, df = 13, p = 0.73). Full results of the Schoenfeld residual tests are available in the Supplementary Appendix (p. 4).
Table 4. Cox proportional hazards models for all-cause mortality according to lithium use patterns
Note: Significant results are reported in bold.
In the subgroup analysis restricted to patients with lithium exposure, those with sustained lithium exposure had a significantly lower risk of all-cause mortality compared to those with partial or intermittent exposure (HR = 0.70, 95% CI: 0.51–0.97, p = 0.03). Significant predictors of increased mortality included the presence of comorbidities, with a twofold increase in risk for patients with one comorbidity (HR = 2.46, 95% CI: 1.66–3.63, p < 0.001), and over a sixfold increase for those with two or more comorbidities (HR = 6.50, 95% CI: 4.67–9.04, p < 0.001). Female sex remained independently associated with a lower mortality risk (HR = 0.57, 95% CI: 0.44–0.73, p < 0.001), and individuals with mid-level socioeconomic status had a reduced risk compared to those in the lowest-income group (HR = 0.66, 95% CI: 0.47–0.94, p = 0.02). Other variables, including age (HR = 1.00, 95% CI: 0.99–1.00, p = 0.42), BD type (BD-II: HR = 0.67, 95% CI: 0.45–1.02, p = 0.06; BD-NOS: HR = 0.31, 95% CI: 0.04–2.24, p = 0.25), and the use of antidepressants (HR = 0.89, 95% CI: 0.67–1.18, p = 0.42), antipsychotics (HR = 1.42, 95% CI: 0.91–2.20, p = 0.12), or mood stabilizers (HR = 1.08, 95% CI: 0.78–1.48, p = 0.65), were not significantly associated with mortality.
The results of the sensitivity analyses across different DDD thresholds are presented in Table 5, reporting HRs and 95% CIs from fully adjusted Cox regression models. The protective association of sustained lithium exposure with reduced all-cause mortality was consistently confirmed at lower DDD thresholds, but this association lost significance as the DDD threshold increased. Corresponding Kaplan–Meier survival curves and regression outputs for each threshold are available in the Supplementary Appendix (pp. 5–6).
Table 5. Sensitivity analysis results using different thresholds of defined daily doses (DDDs) for categorizing lithium exposure patterns
Note: This table presents the sensitivity analysis results for alternative DDD thresholds used to classify lithium use patterns. Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for mortality risk, with lithium non-users as the reference category. Significant results are highlighted in bold.
Given the strong association between medical comorbidities and all-cause mortality observed in our primary analysis, and the potential for confounding by indication, we conducted a post-hoc Cox regression model including interaction terms between patterns of lithium exposure and comorbidity burden (none, one, or two or more comorbidities). This analysis aimed to explore whether the observed protective effect of sustained lithium exposure might reflect an underlying health selection effect – that is, whether healthier individuals are more likely to maintain long-term lithium treatment. The interaction between sustained lithium exposure and comorbidity status was not statistically significant. Among patients with one comorbidity, sustained lithium exposure was associated with a non-significant reduction in mortality risk (HR = 0.57, 95% CI: 0.24–1.34, p = 0.19) and similarly among those with two or more comorbidities (HR = 0.49, 95% CI: 0.23–1.03, p = 0.06). In contrast, a significant interaction was found for partial or intermittent lithium exposure. Among patients with no comorbidities, partial or intermittent lithium exposure was associated with a significantly increased risk of all-cause mortality compared to no lithium exposure (HR = 1.86, 95% CI: 1.20–2.90, p = 0.006). This elevated risk appeared to diminish in patients with one comorbidity (interaction HR = 0.55, 95% CI: 0.31–0.98, p = 0.04) and was further reduced in those with two or more comorbidities (interaction HR = 0.41, 95% CI: 0.25–0.67, p < 0.001). Full model results, including all interaction terms and stratified HRs, are reported in the Supplementary Appendix (p. 7).
Discussion
This study provides novel evidence that the pattern of lithium exposure – whether sustained or partial/intermittent – significantly affects mortality in individuals with BD. Using a large population-based cohort, we found that sustained lithium exposure was associated with a significantly lower risk of all-cause mortality compared to both no exposure and partial or intermittent exposure. Specifically, sustained users had a 31% lower mortality risk than non-users and a 30% lower risk than those with partial or intermittent exposure. To our knowledge, this is the first study to systematically evaluate the impact of lithium treatment continuity on all-cause mortality in a real-world population of individuals with BD. A post-hoc interaction analysis showed no significant interaction between sustained lithium exposure and comorbidity burden, suggesting that this protective effect is not merely due to healthier individuals being more likely to remain on treatment.
These results align with prior studies showing reduced mortality with lithium compared to placebo or other treatments [Reference Cipriani, Pretty, Hawton and Geddes6, Reference Chen, Tsai, Chen, Pan, Su and Chen8, Reference Smith, Austin, Kim, Eisen, Kilbourne and Miller13]. Several mechanisms may explain this beneficial effect. Clinical research has demonstrated that lithium treatment is associated with a reduced risk of cardiovascular disease in individuals with BD [Reference Tsai, Shen, Kuo, Chen, Chung and Hsiao23]. Additionally, preclinical studies using cellular and animal models of myocardial infarction suggest that lithium, when administered at therapeutic levels, may mitigate pathological hypertrophy and cardiac fibrosis [Reference Lee, Lin and Chang24]. Its cardioprotective effects are thought to be mediated through the modulation of inflammatory pathways, oxidative stress reduction, and the preservation of endothelial function, involving various molecular signaling mechanisms [Reference Lee, Lin and Chang25]. Furthermore, beneficial effects on neuronal plasticity and neuroinflammatory modulation may contribute to improved overall health and longevity [Reference Bortolozzi, Fico, Berk, Solmi, Fornaro and Quevedo26]. These biological effects of lithium, combined with its well-established efficacy in preventing relapse of any mood episode [Reference Kishi, Ikuta, Matsuda, Sakuma, Okuya and Mishima27], treating manic episodes [Reference Kishi, Ikuta, Matsuda, Sakuma, Okuya and Nomura28], and its anti-suicidal properties [Reference Smith and Cipriani5], continue to make it a first-line treatment in the management of BD [Reference Yatham, Kennedy, Parikh, Schaffer, Bond and Frey4].
Sensitivity analyses confirmed the robustness of our findings across varying DDD thresholds. Lowering the threshold expanded the group of sustained users and restricted the partial/intermittent group to those with truly irregular exposure. While the protective effect on mortality remained consistent for sustained lithium users, individuals with partial or intermittent exposure exhibited an increased mortality risk when the lowest DDD threshold was applied. These findings suggest that treatment continuity may play a critical role in maximizing lithium’s protective effect on survival. This is consistent with prior research showing that discontinuing lithium is associated with a higher mortality risk than discontinuing valproate [Reference Smith, Austin, Kim, Eisen, Kilbourne and Miller13], supporting the notion that lithium discontinuation may have more detrimental effects on survival than non-use. While the precise mechanisms underlying this association remain unclear, they may extend beyond lithium’s established anti-suicidal properties to broader benefits for physical health when used consistently over time. Conversely, at higher thresholds (≥400 DDD/year), the protective effect of sustained exposure diminished, possibly due to dose-dependent toxicity offsetting benefits. Higher lithium doses are associated with increased risk of renal and thyroid dysfunction, cardiovascular burden, and other adverse effects, particularly in older or medically vulnerable patients. These risks may attenuate the survival benefits observed at moderate doses, suggesting the existence of a therapeutic window in which lithium confers maximal protection. This finding highlights the importance of individualized dosing and regular monitoring to balance efficacy and tolerability in long-term treatment strategies [Reference Bortolozzi, Fico, Berk, Solmi, Fornaro and Quevedo26].
Our post-hoc analyses also revealed that the excess mortality associated with partial or intermittent lithium exposure was particularly marked among individuals without comorbidities. This suggests that abrupt or inconsistent use may be especially detrimental in otherwise physically healthy individuals, while the relative impact is less visible in those with high baseline mortality risk due to medical burden.
Taken together, these findings underscore the clinical importance of maintaining lithium continuity. Once initiated, lithium treatment should be actively monitored and, when clinically appropriate, maintained over time. Ensuring long-term adherence is key to maximize both psychiatric and physical health outcomes [Reference Berk, Hallam, Colom, Vieta, Hasty and Macneil29].
Beyond lithium use patterns, our Cox regression models identified additional factors influencing mortality risk. The presence of somatic comorbidities was the strongest predictor of increased mortality, with an HR of 3.36 for one comorbidity and 10.32 for two or more comorbidities. These findings emphasize the substantial impact of medical burden on survival in BD [Reference Biazus, Beraldi, Tokeshi, LdS, Dragioti and Carvalho30], reinforcing the need for integrated medical and psychiatric care [Reference Liu, Daumit, Dua, Aquila, Charlson and Cuijpers31]. Antipsychotic use was also associated with increased mortality, consistent with prior evidence linking these medications to physical health risks [Reference Correll, Detraux, De Lepeleire and De Hert32]. While essential for managing acute symptoms, antipsychotics require close monitoring, particularly in high-risk patients [Reference Yatham, Kennedy, Parikh, Schaffer, Bond and Frey4]. Consistent with prior studies, female sex was associated with a lower risk of mortality [Reference Schumacher, Kyu, Aali, Abbafati, Abbas and Abbasgholizadeh33], likely reflecting lower suicide rates [Reference Feigin, Vos, Nair, Hay, Abate and Abd Al Magied34], better healthcare-seeking behaviors [Reference Ballering, Olde Hartman, Verheij and Rosmalen35], and biological differences in disease progression [Reference Serra-Navarro, Clougher, Oliva, Valenzuela-Pascual, De Prisco and Forte36, Reference Menculini, Steardo, Sciarma, D’Angelo, Lanza and Cinesi37]. Furthermore, patients with mid-level socioeconomic status had a lower mortality risk compared to those in the lowest socioeconomic group, suggesting that socioeconomic factors play a crucial role in treatment access, adherence, and overall health outcomes [Reference Stringhini, Carmeli, Jokela, Avendaño, Muennig and Guida38].
Although significant differences in baseline characteristics were observed across lithium groups, effect sizes were generally small (Cramér’s V < 0.2), consistent with large sample studies where minor differences often reach statistical significance [Reference Lee39, Reference Sullivan and Feinn40]. The most notable group differences involved antipsychotic use and comorbidities, though effect sizes were weak. For antipsychotic use, the overall Cramér’s V was 0.155, with pairwise values of 0.139 (no exposure versus partial/intermittent) and 0.128 (no exposure versus sustained); no differences emerged between lithium-exposed groups. For comorbidities, weak associations were found between no exposure and sustained exposure (Cramér’s V = 0.134) and between partial/intermittent and sustained exposure (Cramér’s V = 0.159).
A major strength of this study is the use of a large, population-based cohort with longitudinal prescription and mortality data, enabling a robust evaluation of lithium exposure patterns and their association with all-cause mortality over a 10-year period. By leveraging real-world data from all levels of psychiatric care within a universal healthcare system, the study enhances the external validity of its findings and supports their generalisability to routine clinical practice. The detailed categorization of lithium exposure, combined with rigorous sensitivity analyses, further strengthens the reliability and interpretability of our results. Finally, the interaction analysis with comorbidity burden provides novel insight into how the association between lithium use and mortality may vary across different clinical profiles, thereby reducing residual confounding and supporting a more nuanced causal interpretation. However, certain limitations should be acknowledged. First, lithium exposure was inferred from prescription data, which does not account for actual medication intake, adherence, or potential dose adjustments [Reference De Prisco and Oliva41]. While DDD provide a standardized measure of drug exposure, they may not fully capture individual variations in prescribing practices, such as dose titration, temporary discontinuations, or therapeutic drug monitoring adjustments. Indeed, DDDs reflect the amount of drug dispensed, not necessarily consumed, and do not account for whether patients actually took the medication as prescribed. Therefore, they approximate adherence but cannot measure it directly. However, they offer a useful proxy for long-term treatment continuity when aggregated over time. Using prescription records offers an objective and reproducible method to assess long-term lithium exposure in a large, real-world population, reducing recall bias and providing valuable insights into treatment patterns and outcomes. This approach has been widely used in previous pharmacoepidemiologic studies on psychopharmacology [Reference Kendler, Ohlsson, Sundquist and Sundquist22]. Second, unmeasured confounders may influence both lithium adherence and mortality risk. While our study adjusted for possible key demographic and clinical covariates, residual confounding remains possible. Third, this study should be considered a post-hoc analysis, as the data were collected without a predefined hypothesis [Reference Oliva and Vieta42]. Consequently, its conclusions should be interpreted as preliminary and require confirmation in longitudinal studies specifically designed to evaluate the association between lithium use patterns and mortality [Reference Scott, Hidalgo-Mazzei, Strawbridge, Young, Resche-Rigon and Etain43]. Fourth, potential heterogeneity within the study population, including differences in illness severity, treatment history, and comorbidities, may have influenced the observed associations despite statistical adjustments [Reference Oliva and De Prisco44]. Fifth, specific causes of death were not available in our dataset. This precluded analyses of whether lithium’s protective effects vary by cause of death – such as cardiovascular, metabolic, or suicide-related mortality – which would be important to clarify the underlying mechanisms. Finally, as the study was conducted within a single public healthcare system in Catalonia, Spain, replication in other settings is needed to confirm the external validity and generalizability of our findings.
Further research should aim to better characterize individuals with partial or intermittent lithium exposure and identify strategies to improve treatment continuity. Prospective studies incorporating biomarkers, genetic profiles, and real-time adherence monitoring could help clarify the mechanisms behind the observed mortality differences. Future work should also explore the relationship between lithium exposure patterns and specific causes of death, especially cardiovascular and metabolic conditions.
Conclusions
This study reinforces the critical role of sustained lithium treatment in improving survival in individuals with BD. Sustained lithium exposure was consistently associated with reduced all-cause mortality, whereas partial or intermittent exposure conferred no benefit and may entail additional risks. These findings support clinical strategies that prioritize treatment continuity and long-term adherence to optimize both psychiatric and physical health outcomes in BD.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1192/j.eurpsy.2025.10090.
Data availability statement
The dataset used and analyzed in this study will be available upon reasonable request from the corresponding author, in accordance with the agreement between PADRIS-AQuAS and Hospital Clínic of Barcelona.
Financial support
V.O. is supported by a Rio Hortega 2024 grant (CM24/00143) from the Spanish Ministry of Science, Innovation and Universities financed by the Instituto de Salud Carlos III (ISCIII) and co-financed by the Fondo Social Europeo Plus (FSE+). M.D.P. is supported by the Translational Research Program for Brain Disorders, IDIBAPS. G.A. thanks the support of the Spanish Ministry of Science, Innovation and Universities financed by the Instituto de Salud Carlos III (ISCIII) and co-financed by the European Social Fund+ (ESF+) (JR23/00050, MV22/00058, CM21/00017); the ISCIII (PI24/00584, PI24/01051, PI21/00340, PI21/00169); the Milken Family Foundation (PI046998); the Fundació Clínic per a la Recerca Biomèdica – Pons Bartan 2020 grant (PI04/6549), the Sociedad Española de Psiquiatría y Salud Mental; the Fundació Vila Saborit; the Societat Catalana de Psiquiatria i Salut Mental; and the Translational Research Programme for Brain Disorders, IDIBAPS. C.V.-P. is supported by an FPU grant for University Teaching Training (FPU23/01555) from the Spanish Ministry of Science, Innovation, and Universities. A.M. is supported by a contract funded by the “European Union Next Generation EU/PRTR.” A.Mu. thanks the support of the Spanish Ministry of Science and Innovation (PI19/00672, PI22/00840) integrated into the Plan Nacional de I+D+I and co-financed by the ISCIII-Subdireccion General de Evaluación and the Fondo Europeo de Desarrollo Regional (FEDER). J.R. thanks the support of the Spanish Ministry of Science, Innovation and Universities (CPII19/00009, PI22/00261) integrated into the Plan Nacional de I+D+I and co-financed by the ISCIII-Subdirección General de Evaluación and the FEDER and the Instituto de Salud Carlos III; and the Secretaria d’Universitats i Recerca del Departament d’Economia i Coneixement (2021-SGR-01128). E.V. thanks the support of the Spanish Ministry of Science, Innovation and Universities (PI18/ 00805, PI21/00787) integrated into the Plan Nacional de I+D+I and cofinanced by the ISCIII-Subdireccio ́n General de Evaluacio ́n and the Fondo Europeo de Desarrollo Regional (FEDER); the Instituto de Salud Carlos III; the CIBER of Mental Health (CIBERSAM); the Secretaria d’Universitats i Recerca del Departament d’Economia i Coneixement (2017 SGR 1365), the CERCA Program, and the Departament de Salut de la Generalitat de Catalunya for the PERIS grant SLT006/17/00357. Thanks to the support of the European Union Horizon 2020 research and innovation program (EU.3.1.1. Understanding health, wellbeing and disease: grant no. 754907 and EU.3.1.3. Treating and managing disease: grant no. 945151). D.H.-M. thanks the support of the Spanish Ministry of Health, Instituto de Salud Carlos III (PI049759) and the Pons-Bartran Legacy grant 2024 (FCRB_IPB1_2024). The PRESTO project was supported by the Fundació Clínic per a la Recerca Biomèdica through the Pons Bartran 2020 grant (PI046549); the Spanish Foundation for Psychiatry and Mental Health, the Spanish Psychiatric Society, and the Spanish Society of Biological Psychiatry (PI046813); the Agència de Gestió d’Ajuts Universitaris i de Recerca (AGAUR) – PANDÈMIES 2020 grant (2020PANDE00081), Generalitat de Catalunya; the Ministerio de Sanidad through the Pla Director de Salut Mental i Addiccions and the Direcció General de Planificació i Recerca en Salut, Departament de Salut, Generalitat de Catalunya; and by the Ministerio de Ciencia, Innovación y Universidades (MCIN) and the Agencia Estatal de Investigación (AEI, project TED2021-131999BI00, Strategic Projects Oriented to the Ecological Transition and the Digital Transition 2021), with funding from the European Union NextGenerationEU/PRTR.
Authors’ contribution
V.O. and M.D.P., with input from D.H.-M., conceptualized and planned the study. V.O., M.D.P., and D.H.-M. managed and coordinated the research activity planning and execution. V.O., M.D.P., and D.H.-M. conducted the statistical analyses, with important contributions from G.A. and J.R. V.O. prepared the first draft of the manuscript. All authors critically reviewed and commented on the manuscript and approved the final version. D.H.-M. had supervisory and leadership responsibility for the research activity planning and execution. All authors had full access to all the materials of the study and had final responsibility for the decision to submit for publication.
Competing interests
G.A. has received CME-related honoraria or consulting fees from Abartis Pharma, Adamed, Angelini, Casen Recordati, Johnson & Johnson, Lundbeck, Lundbeck/Otsuka, Rovi, and Viatris, with no financial or other relationship relevant to the subject of this article. G.F. has received CME-related honoraria or consulting fees from Angelini, Janssen-Cilag. and Lundbeck. A.Mu. has received grants and served as a consultant, advisor or CME speaker for the following entities: Angelini, Lundbeck, Pfizer, Takeda, outside of the submitted work. M.V. has received research grants from Eli Lilly & Company and has served as a speaker for Abbott, Bristol–Myers Squibb, GlaxoSmithKline, Janssen–Cilag, and Lundbeck. J.R. has received CME honoraria from Inspira Networks for a machine learning course promoted by Adamed, outside the submitted work. A.H.Y. has received honoraria for lectures and advisory boards for all major pharmaceutical companies with drugs used in affective and related disorders, with no financial or other relationship relevant to the subject of this article. E.V. has received grants and served as a consultant, advisor, or CME speaker for the following entities: AB-Biotics, AbbVie, Angelini, Biogen, Biohaven, Boehringer-Ingelheim, Celon Pharma, Dainippon Sumitomo Pharma, Ferrer, Gedeon Richter, GH Research, Glaxo-Smith Kline, Idorsia, Janssen, Lundbeck, Novartis, Orion Corporation, Organon, Otsuka, Sage, Sanofi-Aventis, Sunovion, and Takeda, outside the submitted work. D.H.-M. has received CME-related honoraria and served as consultant for Abbott, Angelini, Neuraxpharm, Ethypharm Digital Therapy, Lundbeck and Viatris. All the other authors have no conflicts to declare.
Open access
Comments
No Comments have been published for this article.