Mortality and Lithium-Protective Effects after First-Episode Mania Diagnosis in Bipolar Disorder: A Nationwide Retrospective Cohort Study in Taiwan

Abstract

Introduction

This study aimed to estimate all-cause mortality in patients after a first-episode mania (FEM) and examine whether six guideline-recommended medications can reduce mortality.

Methods

The cohort included population-based FEM samples and matched controls from Taiwan, spanning 2007 to 2018. The primary outcomes assessed were all-cause/suicide-related mortality, while the secondary outcome focused on mortality associated with pharmacological treatments. We compared mortality in post-FEM patients and age-/sex-matched controls without any diagnosed bipolar disorders and patients with and without psychopharmacological treatment using Cox regression analysis, respectively. Statistics were presented with time-to-event adjusted hazard ratios (AHRs) and 95% confidence intervals (CIs).

Results

The study included 54,092 post-FEM patients and 270,460 controls, totaling 2,467,417 person-years of follow-up. Post-FEM patients had higher risks of all-cause mortality (AHR 2.38, 95% CI: 2.31–2.45) and suicide death (10.80, 5.88–19.84) than controls. Lithium (0.62, 0.55–0.70), divalproex (0.89, 0.83–0.95), and aripiprazole (0.81, 0.66–1.00) were associated with reduced all-cause mortality compared to non-users. There were no significant all-cause mortality differences for quetiapine (0.95, 0.89–1.01), risperidone (0.92, 0.82–1.02), and paliperidone (1.24, 0.88–1.76) users. When accounting for drug action onset times in sensitivity analyses, only lithium significantly reduced all-cause mortality (AHR range 0.65–0.72). There were 35 and 16 suicide deaths in post-FEM patients and controls, respectively. No drug had a significant effect on suicide deaths (lithium: 6; divalproex: 7; aripiprazole: 0; quetiapine: 10; risperidone: 4; paliperidone: 1).

Conclusion

Post-FEM patients had a higher risk of all-cause/suicide-related mortality, and lithium treatment might reduce all-cause mortality.

Introduction

Individuals with bipolar disorder (BD) have a high mortality rate, which shortens their lifespan compared to general population counterparts [1]. This excess mortality rate is mainly driven by physical comorbidities in terms of absolute number of deaths, such as cardiovascular and metabolic diseases. Suicide also plays a major role and has the highest relative risk [2–4]. However, most of these previous findings do not derive from participants after a first-episode mania (FEM). This is a major confounder because studies often include participants with various age groups and statistical controlling does not necessarily eliminate the influence of age, for example, on the development of co-occurring somatic illnesses [5]. Additionally, the construct of early intervention posits that prompt initiation of evidence-based therapy should be associated with better clinical outcomes [6, 7]. Although several previous studies have focused on first-episode psychosis (FEP), evidence for effective interventions for FEM is extremely limited [8]. It is known that FEM and FEP have different prodromal symptoms and risk factors [9]. Furthermore, the prognosis of BD and schizophrenia differs [10]. Therefore, the results for FEP cannot be generalized to FEM.

Lithium is a first-line agent to treat bipolar mania and is well-known for its anti-suicidal properties [11, 12]. However, it remains unclear whether lithium decreases all-cause mortality in BD besides reducing suicide-related mortality, especially in BD after FEM. There is also no substantial evidence for the anti-suicidal effects of other drugs approved to treat bipolar mania, such as divalproex and certain antipsychotics recommended by the latest International Society for Bipolar Disorders (ISBD) guidelines [13]. Furthermore, an umbrella review noted a causal relationship between depression and suicide-related mortality but not depression and all-cause mortality [14], thereby demonstrating that a decrease in suicide-related mortality does not necessarily translate into a decrease in all-cause mortality. As for other pharmacological interventions, evidence from other populations has shown that antipsychotics can increase risk of life-threatening medical comorbidities and death in persons with dementia [15], but they are associated with decreased mortality in people with schizophrenia [16]. It is also unclear whether these associations would be similar in users with BD after a FEM.

Thus, the aim of this study was to estimate the risk of all-cause mortality and suicide-related mortality in the bipolar population after a FEM and compare it to an age- and sex-matched general population control group. We then also investigated the association between use of current treatment guidelines-recommended medications and these outcomes.

Materials and Methods

Samples

The retrospective cohort study followed the Reporting of Studies Conducted Using Observational Routinely-Collected Data (RECORD) guidelines (online suppl. eTable 1; for all online suppl. material, see https://doi.org/10.1159/000535777) [17]. All data were retrieved from the Taiwan National Health Insurance Research Institute Database (NHIRD), which was established for research purposes and has been audited by the Department of Health the Bureau of the National Health Insurance (NHI) program since 1996. The NHIRD contains comprehensive information on covered patients, such as demographics (sex, birthdate), medical visits (dates and diagnoses), pharmacy records (medication type, dosage, and duration of supply). To protect privacy, each patient is registered with a unique and anonymous identifier assigned by the NHI, which allows researchers to track their disease and outcomes through de-identified personal information. NHIRD covers almost the entire Taiwanese population at 99.99%. This study used the NHIRD data between 2007 and 2018 and determined the diagnosis using the International Classification of Diseases, ninth revision, clinical revision (ICD-9, from 2007 to 2015) or ICD-10 (from 2016 to 2018). Several previous observational studies extensively used the NHIRD [18–21]. The study protocol was approved by the Institutional Review Board of Chang Gung Memorial Hospital (No.: 202102426B0).

In this cohort, FEM patients were defined as participants with an initial diagnosis of bipolar I disorder with mania (ICD-9: 296.0 and 296.4; ICD-10: F30.1–F30.4, F30.9, F31.0–F31.2, F31.73, F31.74, and F31.89) and no previous diagnosis of BD (ICD-9: 296.1, 296.5, 296.6, 296.7, 296.8, and 296.9; ICD-10: F31.3–F31.6, F31.70–F31.72, F31.75–F31.78, and F31.9) between January 1, 2007, and December 31, 2018. The time of enrollment was defined as the diagnosis date of FEM diagnosis. A sex- and age-matched (1:5) control group was randomly selected after excluding study participants with a diagnosis of BD at any time in the database. The matching criteria were sex (the same) and birth year (the same).

Study Variables

We selected 23 physical comorbidities as confounding factors, including a history of stroke, migraine, epilepsy, Parkinson’s disease, dementia, low back pain, fibromyalgia, chronic fatigue syndrome, ischemic heart disease, heart failure, chronic obstructive pulmonary disease, cirrhosis, chronic kidney disease, polycystic ovarian syndrome, rheumatoid arthritis, systemic lupus erythematosus, diabetes, overweight/obesity, hypertension, dyslipidemia/hypercholesterolemia, hypothyroidism, hyperthyroidism, and cancer. They are considered comorbidities that are highly associated with individual disability and mortality according to the Global Burden of Disease (GBD) study [22]. The detailed ICD-9 and ICD-10 codes are shown in online supplementary eTable 2. Figure 1 depicts the flowchart of the selection process for this retrospective cohort study.

Fig. 1.

Flowchart of the selection process for the cohort study.

FEM participants were followed up from the index date until the outcome occurrence or at the end of 2018. The primary outcome measure for our cohort was all-cause mortality. Suicide (ICD-9: E950–E959; ICD-10: X71–X83), unintentional injury (ICD-9: E800–E949; V00–V99, W00–W99, X00–X58, X92–X99, Y00–Y09, and Y62–Y82), and death from natural causes were also extracted as variables of interest. Death from natural causes was defined as causes other than suicide and unintentional injury [23]. The secondary outcome was the effectiveness of medications for mania in BD recommended by ISBD guidelines, such as lithium, divalproex, quetiapine, aripiprazole, risperidone, and paliperidone [13]. Among these drugs, aripiprazole and paliperidone were analyzed in both their oral and long-acting injectable forms. We also investigated whether these drugs reduced mortality.

Statistical Analysis

In this study, continuous variables were presented as means with standard deviations and categorical variables as numbers and proportions. Kaplan-Meier curves were also used for survival analysis between FEM patients and matched controls. We performed a Cox proportional hazard model to generate adjusted hazard ratios (AHRs) with 95% confidence intervals (CIs) to compare the risk of all-cause mortality and suicide death between FEM patients and healthy controls, adjusting for all baseline covariates listed in Table 1. Healthy controls were considered the reference group. We also used extended Kaplan-Meier estimator (all-cause mortality and suicide-related mortality) for time-varying medications in post-FEM patients [24, 25]. Illustrative examples are shown in online supplementary eFigure 1. To account for follow-up intervals defined by different drugs (user) or no drug (non-user) for the same individual, we fit a Cox proportional hazards discrete time model. The model defined time as the duration of use or non-use of the selected drug and adjusted for confounding factors of sex and age when the selected drug was initially used. Several studies have demonstrated a reduction in manic symptoms within three weeks of drug treatment [26], so we defined deaths less than three weeks after selected drug use as drug-independent (non-user) deaths. Thus, as a sensitivity analysis, we also removed FEM patients who died within less than three weeks of selected drug use (model 1). We also performed other sensitivity analyses using two weeks (models 2 and 3) and 1 week (models 4 and 5) as a buffer period (online suppl. eFig. 2). Moreover, all analyses regarding the effect of selected drugs were repeated with lithium, divalproex, quetiapine, aripiprazole, risperidone, and paliperidone, respectively. Each selected drug group was not mutually exclusive, as FEM patients were often treated with 2 or more of these drugs at the same time [27–30]. All analyses were performed with SAS 9.4 Software (SAS Institute Inc., Cary, NC, USA). A two-tailed p value <0.05 was regarded as statistically significant.

Table 1.

Characteristics of all included participants from 2007 to 2018

Characteristics	Controls (N = 270,460)	FEM (N = 54,092)	t or χ2	p value
Age, year	46.7 (46.6–46.8)	46.7 (46.5–46.8)	0	1.000
Sex, female	153,200 (56.6)	30,640 (56.6)	0	1.000
Comorbidities
Stroke	11,168 (4.1)	5,421 (10.0)	3,227	<0.001
Migraine	6,151 (2.3)	3,365 (6.2)	2,467	<0.001
Epilepsy	1,642 (0.6)	1,589 (2.9)	2,484	<0.001
Parkinson’s disease	1,543 (0.6)	1,744 (3.2)	3,166	<0.001
Dementia	2,091 (0.8)	2,529 (4.7)	4,892	<0.001
Low back pain	30,067 (11.1)	9,208 (17.0)	1,478	<0.001
Fibromyalgia	37,287 (13.8)	12,275 (22.7)	2,763	<0.001
Chronic fatigue syndrome	2,042 (0.8)	995 (1.8)	572	<0.001
Ischemic heart disease	18,292 (6.8)	6,784 (12.5)	2,111	<0.001
Heart failure	4,250 (1.6)	1,886 (3.5)	891	<0.001
Chronic obstructive pulmonary disease	30,295 (11.2)	10,944 (20.2)	3,314	<0.001
Cirrhosis	22,237 (8.2)	8,107 (15.0)	2,434	<0.001
Chronic kidney disease	3,770 (1.4)	1,662 (3.1)	772	<0.001
Polycystic ovarian syndrome	1,760 (0.7)	470 (0.9)	31	<0.001
Rheumatoid arthritis	2,493 (0.9)	912 (1.7)	254	<0.001
Systemic lupus erythematosus	520 (0.2)	203 (0.4)	68	<0.001
Diabetes	24,182 (8.9)	8,381 (15.5)	2,144	<0.001
Overweight/obesity	1,799 (0.7)	744 (1.4)	293	<0.001
Hypertension	45,786 (16.9)	14,084 (26.0)	2,486	<0.001
Dyslipidemia/hypercholesterolemia	7,955 (2.9)	2,339 (4.3)	281	<0.001
Hypothyroidism	2,090 (0.8)	957 (1.8)	481	<0.001
Hyperthyroidism	4,552 (1.7)	1,935 (3.6)	826	<0.001
Cancer	8,113 (3.0)	2,656 (4.9)	513	<0.001
Mortality	15,357 (5.7)	8,073 (14.9)	5,754	<0.001
Suicide	16 (0.01)	35 (0.1)
Natural causes + unintentional injurya	15,231 (5.6)	7,983 (14.8)

Data were expressed as mean (95% CIs) or N (percentage).

FEM, first-episode mania.

^aThe FEM and control groups each had less than three cases of unintentional injury, but specific numbers cannot be disclosed due to Taiwan database usage rules.

Results

From our study database, we identified 54,092 FEM cases and 270,460 matched controls between 2007 and 2018 (Table 1). The mean age of cases and controls was 47 years and 57% of patients were female. A higher proportion of FEM cases suffered from the 23 death-related comorbidities investigated in this study (range from 0.4% [systemic lupus erythematosus] to 26.0% [hypertension]), and there was higher mortality (14.8%) than in matched controls (5.6%). Crude mortality rates for matched controls and FEM patients were 0.74 and 2.06 all-cause deaths per 100 person-years, respectively. In the multivariable Cox proportional hazards model, FEM patients had higher all-cause mortality (AHR, 95% CI: 2.38, 2.31–2.45) as well as mortality due to suicide (10.80, 5.88–19.84) than matched controls (Table 2). Online supplementary eFigure 3 reveals the Kaplan-Meier curves during the overall study period.

Table 2.

Comparing the risk of all-cause mortality and suicide between healthy controls and FEM patients through the Cox proportional hazard model

Group	Death cases	Total cases	HR	AHRa
All-cause mortality
Control	15,357	270,460	1.00 [reference]	1.00 [reference]
FEM	8,073	54,092	2.80 (2.72–2.87)*	2.38 (2.31–2.45)*
Suicide death
Control	16	270,460	1.00 [reference]	1.00 [reference]
FEM	35	54,092	11.46 (6.34–20.70)*	10.80 (5.88–19.84)*

AHR, adjusted hazard ratio; HR, hazard ratio.

^aHazard ratio was adjusted for sex and age; *indicated p < 0.05.

Table 3 presents analysis of extended Kaplan-Meier estimator for time-varying exposure to selected drug treatment (user), with no exposure to selected drug treatment (non-user) as the reference group. Lithium significantly reduced all-cause mortality of FEM patients (AHR, 95% CI: 0.62, 0.55–0.70). This significant reduction was consistent across five additional sensitivity analyses (model 1: 0.65, 0.58–0.74; model 2: 0.67, 0.60–0.75; model 3: 0.69, 0.62–0.77; model 4: 0.71, 0.64–0.80; model 5: 0.72, 0.64–0.80). While divalproex (0.89, 0.83–0.95) and aripiprazole (0.81, 0.66–0.996) also exhibited a reduction in all-cause mortality in FEM patients in the primary analysis, subsequent sensitivity analyses did not establish a significant association with mortality reduction for these two drugs (divalproex: AHR range 0.99–1.16; aripiprazole: AHR range 0.86–0.95). The remaining three drugs did not show a significant association with reduced all-cause mortality in patients with FEM in the primary analysis (quetiapine: 0.95, 0.89–1.01; risperidone: 0.92, 0.82–1.02; paliperidone: 1.24, 0.88–1.76) and five other sensitivity analyses (quetiapine: AHR range 1.10–1.30; risperidone: AHR range 1.02–1.20; paliperidone: AHR range 1.29–1.69). Detailed information is shown in online supplementary eTables 3–5.

Table 3.

Comparing the risk of all-cause mortality of FEM patients between drug non-user and drug user through the Cox proportional hazards discrete time model, with a 3-week buffer interval

Group	Death cases	Total cases	HR	AHRa
Lithium
Non-user	7,750	148,028	1.00 [reference]	1.00 [reference]
User	323	105,333	0.45 (0.40–0.51)*	0.62 (0.55–0.70)*
Divalproex
Non-user	6,984	221,433	1.00 [reference]	1.00 [reference]
User	1,089	193,067	0.74 (0.69–0.80)*	0.89 (0.83–0.95)*
Quetiapine
Non-user	6,571	213,428	1.00 [reference]	1.00 [reference]
User	1,502	181,594	1.01 (0.95–1.07)	0.95 (0.89–1.01)
Aripiprazole
Non-user	7,976	88,045	1.00 [reference]	1.00 [reference]
User	97	37,915	0.52 (0.43–0.64)*	0.81 (0.66–0.996)*
Risperidone
Non-user	7,658	120,210	1.00 [reference]	1.00 [reference]
User	415	76,034	0.78 (0.70–0.87)*	0.92 (0.82–1.02)
Paliperidone
Non-user	8,040	61,918	1.00 [reference]	1.00 [reference]
User	33	8,641	0.72 (0.51–1.01)	1.24 (0.88–1.76)

AHR, adjusted hazard ratio; HR, hazard ratio.

^aHazard ratio was adjusted for sex and age.

*Indicated p < 0.05.

Table 4 focuses specifically on suicide death data (35 cases). No suicide events occurred with patients treated with aripiprazole; one, four, six, seven, and ten events occurred in those treated with paliperidone, risperidone, lithium, divalproex, or quetiapine, respectively. However, the 95% CIs for all drugs were large and not statistically significant for occurrence of suicide-related mortality.

Table 4.

Comparing the risk of suicide death of FEM patients between drug non-user and drug user through the Cox proportional hazards discrete time model

Group	Death cases	Total cases	HR
Lithium
Non-user	29	148,028	1.00 [reference]
User	6	105,333	1.49 (0.58–3.80)
Divalproex
Non-user	28	221,433	1.00 [reference]
User	7	193,067	0.96 (0.41–2.29)
Quetiapine
Non-user	25	213,428	1.00 [reference]
User	10	181,594	1.12 (0.52–2.43)
Aripiprazole
Non-user	35	88,045	1.00 [reference]
User	0	37,915	0.00 (0.00–N/A)
Risperidone
Non-user	31	120,210	1.00 [reference]
User	4	76,034	1.08 (0.35–3.30)
Paliperidone
Non-user	34	61,918	1.00 [reference]
User	1	8,641	2.34 (0.29–19.19)

HR, hazard ratio; N/A, not applicable.

Discussion

In this nationwide retrospective cohort study of 54,092 FEM participants with BD and 270,460 age- and sex-matched controls, a FEM was associated with a 2-fold increase in the risk of all-cause mortality and 11-fold increased risk of mortality due to suicide. Furthermore, this higher risk of mortality was significantly reduced during lithium treatment in post-FEM patients, which was consistently observed in both the primary analysis and various sensitivity analyses conducted with different treatment durations. In contrast, treatment with divalproex, quetiapine, aripiprazole, risperidone, or paliperidone did not show a similarly consistent reduction in mortality risk. This major finding suggests that lithium may also decrease all-cause mortality in BD, even after a FEM. This study provides support for the notion of early intervention with optimal therapy, in this case lithium, being associated with enduring benefits, and provides the evidence base for guidelines and expert advice suggesting lithium as first-line treatment of BD since the first episode [31].

Previous research on BD looked at overall mortality and suicide risk but did not specifically mention the number of mood episodes [2, 32, 33]. Our study focused on BD after a FEM and demonstrated an increased risk of all-cause mortality (AHR: 2.38) and mortality due to suicide (AHR: 10.80) compared to healthy controls. These findings were broadly consistent with the literature, which reported an overall risk of mortality and suicide that were 2-fold [32] and 20-fold [34] higher, respectively, than the general population. Premature mortality may be associated with an increased prevalence of common physical comorbidities in BD patients, such as vascular diseases, heart diseases, inflammatory diseases, and metabolic diseases [3, 35, 36], which can lead to disability and a decrease in lifespan [1], and high societal costs [37]. Several clinical implications may be suggested based on our results for all-cause mortality risk in BD after FEM. Clinicians should develop physical and mental health interventions at an early stage, including health promotion programs such as lifestyle changes and early detection and intensive management of physical conditions such as the metabolic syndrome [38].

While our study has limited power to draw definitive conclusions due to the small number of suicide death cases (35 cases), lithium has been well documented in many sources for its anti-suicidal properties [39]. Our study further showed that the risk of all-cause mortality was significantly reduced during lithium treatment. This result may expand the protective effect of lithium against premature death or aging in selected populations [40, 41], at least for BD after FEM. Two possible mechanisms could explain this effect on suicide and all-cause mortality. First, previous meta-analyses results suggested that higher C-reactive protein (CRP) levels were associated with higher suicide rates in individuals with mood disorders and that peripheral interleukin-6 (IL-6), tumor necrosis factor α (TNF-α), and CRP levels were elevated in patients with BD [42, 43]. Lithium is anti-inflammatory and acts as a glycogen synthase kinase 3 inhibitor [44, 45]. In preclinical models, lithium enhances the adaptive response to lipopolysaccharide by down-regulating pro-inflammatory genes such as IL-6, TNF-α, and CRP and up-regulating genes associated with anti-inflammatory functions such as interleukin-10 [46]. Taken together, lithium may reduce neuroinflammation-related suicidal behavior by inhibiting inflammatory pathways. Second, lithium may also prevent some of the physical comorbidities that lead to premature death [47]. For example, lithium has potential protective effects against certain cancers [48], is associated with reduced risk of dementia [49, 50], maintains vascular elasticity by promoting elastin biosynthesis to repress abdominal aortic aneurysm [51], and reduces osteoporosis risk through potential bone-protective properties [52]. These protective effects of lithium may lead to increased health and reduced disability during aging [53]. It is also conceivable that lithium may have an anti-senescent effect. People with BD manifest accelerated brain age which appears to be normalized by lithium. There is only a single randomized controlled trial on the use of lithium compared to quetiapine after a FEM. Lithium manifested overtly superior clinical outcomes [54], neurocognitive advantages [55], and structural brain protection compared to quetiapine [54, 56]. Aggregation of the extant clinical evidence supports a unique effect of lithium in preventing both neuroprogression and somatoprogression of BD [57].

The other drugs in this study, divalproex, quetiapine, aripiprazole, and risperidone, failed to provide conclusive evidence of reduced mortality in BD after FEM based on sensitivity analyses with different treatment durations. This may also be related to physical comorbidities (cardiovascular or cardiometabolic systems). One study found that quetiapine users, but not lithium users, manifested significantly worsened cardiometabolic risks (higher cholesterol) during a 24-week treatment period, which was associated with suicidality [58]. The issue of side effects on the cardiovascular system (cardiovascular events) also extends to most antipsychotics [59]. Moreover, divalproex in patients with cardiovascular comorbidities was associated with a higher rate of concomitant heart failure and all-cause mortality. Regarding suicide risk, our results are consistent with most previous studies showing no or controversial protective effects of valproate and most antipsychotics [60, 61].

This cohort study enrolled participants with BD since their FEM, and we comprehensively followed their visits and mortality from 2007 to 2018. In contrast to the previous study in Taiwan [62], we focused specifically on FEM patients (not all BD) and analyzed medications recommended by ISBD guidelines (several antipsychotics in addition to the mood stabilizers lithium and divalproex). However, some limitations of this study should be acknowledged. First, with database-based retrospective cohort studies, it is difficult to avoid possible indication bias [63], such as the use of specific treatment agents, and unmeasured confounders, such as family and child development history or access to psychotherapy. To minimize these potential limitations, we extracted observable differences in baseline characteristics, applied an AHR model, and performed further sensitivity analyses. Second, current NHIRD medical records do not provide scores from validated manic symptom rating scales, such as the Hypomanic Rating Scale [64]. Therefore, we could not adjust for covariates such as initial symptom severity of FEM participants [9]. Furthermore, psychiatrists in Taiwan are well trained by the Taiwan Psychiatric Society to develop standard and consistent coding behaviors; however, the diagnosis of BD as we define it may be relatively narrow and specific and focused on classic BD. Third, we used extended Kaplan-Meier estimator for time-varying exposure to selected drug treatments and classified participants into users or non-users of the selected drug. In fact, participants in the non-user groups may have used other drugs. For example, the lithium non-user group may have used divalproex or other antiepileptics, quetiapine or other antipsychotics, or may have been truly drug-free. Therefore, the related result shall be explained as a lower mortality rate for lithium users compared to non-lithium users. Conversely, drug exposure in this study was determined using prescription records, which introduced a confounding factor due to possible patient non-adherence [65]. For example, non-adherent patients might be mistakenly classified as lithium users. However, this potential misclassification may still suggest lithium’s mortality-reducing effect, as it leads to underestimation of its true protective role. Fourth, data for this cohort were obtained from NHIRD Taiwan, representing a predominantly East Asian population. Therefore, these findings warrant replication in population groups of different ethnic backgrounds, especially groups other than the East Asian population. Fifth, the number of suicide deaths in our study was small, resulting in low statistical power. This may be due to two factors. For one, our study focused only on FEM patients, with a follow-up period of approximately 7 years. Previous literature suggested that the risk of suicide death was more strongly associated with longer disease duration [66, 67]. Thus, our FEM cohort might not have reached the peak risk of suicide death. The other factor is that our study’s definition for suicide-related mortality includes only deaths directly caused by suicide. Patients with attempted suicide who later succumbed to complications or secondary illnesses, such as infection, may not have been formally coded as suicide death.

Conclusion

The results of this study suggest that FEM patients have an elevated risk of all-cause mortality as well as mortality due to suicide, and that lithium treatment was associated with a reduced risk of all-cause mortality. These findings suggest that not only suicide-related mortality but also all-cause mortality should be prioritized in the clinical management of post-FEM BD. These data reinforce clinical guidelines which recommend the use of lithium from the first episode. Furthermore, the potential mortality-protective properties of lithium should be further investigated in the context of BD.

Acknowledgments

The authors would like to thank Ms. Pei-Ying Yang, Mr. Chien-An Hu, Mr. Yang-Chieh Brian Chen, and Mr. Yao Hsu Yang for the technical support, and also sincerely thank Health Information and Epidemiology Laboratory at the Chiayi Chang Gung Memorial Hospital for statistics work.

Statement of Ethics

The research was conducted ethically in accordance with the World Medical Association Declaration of Helsinki. The Institutional Review Board of Chang Gung Memorial Hospital approved the study protocol and waived the need for informed consent (No.: 202102426B0).

Conflict of Interest Statement

Prof. Eduard Vieta has received grants and served as a consultant, advisor, or CME speaker for the following entities: AB-Biotics, Abbott, AbbVie, Angelini, Biogen, Biohaven, Bristol-Myers Squibb, Celon, Compass, Dainippon Sumitomo Pharma, Farmindustria, Ferrer, Gedeon Richter, GH Research, Glaxo-Smith-Kline, HMNC, Idorsia, Janssen, Lundbeck, Medincell, Merck, Novartis, Orion, Otsuka, Pfizer, Roche, Rovi, Sanofi-Aventis, Sunovion, Takeda, Viatris, the Brain and Behaviour Foundation, the Spanish Ministry of Science and Innovation (CIBERSAM), the EU Horizon 2020, and the Stanley Medical Research Institute. Other authors declare no financial interests or potential conflicts of interest regarding the authorship and publication of this article.

Funding Sources

This study is supported by grants from the National Science and Technology Council, Taiwan (NSTC 109-2314-B-182A-009-MY2, 111-2314-B-182A-027-, and 112-2314-B-182A-036-MY3), and the funding sources had no role in the design of the study. M.B. is supported by a NHMRC Senior Principal Research Fellowship and Leadership 3 Investigator Grant (1156072 and 2017131).

Author Contributions

A.F.C. and L.-J.W. conceived the research idea for the study. A.F.C. and L.-J.W. led the study design, with C.-W.H. C.-W.H. and L.-J.W. contributed to data acquisition and extraction. C.-W.H. performed statistical analysis. A.F.C. and C.-W.H. drafted the manuscript first, then E.V., M.S., W.M., M.B., C.-S.L., P.-T.T., and L.-J.W. revised the manuscript. All authors contributed important intellectual content during manuscript revision, had full access to all the data in the study, and accepted responsibility to submit for publication.

Funding Statement

This study is supported by grants from the National Science and Technology Council, Taiwan (NSTC 109-2314-B-182A-009-MY2, 111-2314-B-182A-027-, and 112-2314-B-182A-036-MY3), and the funding sources had no role in the design of the study. M.B. is supported by a NHMRC Senior Principal Research Fellowship and Leadership 3 Investigator Grant (1156072 and 2017131).

Data Availability Statement

The data that support the findings of this study are not publicly available due to privacy reasons but are available from corresponding author upon reasonable request and with Institutional Review Board approval.