Sleep quality in bipolar disorder: A comparative study of treatment with lithium and valproic acid

Abstract

BACKGROUND

Sleep disturbances are a prominent feature of bipolar disorder (BD) and often persist even in remission, thereby contributing to poor clinical outcomes. Despite the widespread use of lithium and valproic acid as mood stabilizers, their effects on sleep quality have not been examined in adequate detail.

AIM

To evaluate and compare the effects of lithium and valproic acid on sleep quality in BD patients under remission.

METHODS

A total of 130 patients meeting the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition criteria for BD in remission were included in this cross-sectional study. The participants were receiving either lithium (n = 78), or valproic acid (n = 52), for a minimum of six months either alone or in combination with antipsychotics. Sleep quality was measured using the Pittsburgh Sleep Quality Index (PSQI). Comparative analyses between the lithium and valproic acid groups were conducted using independent t-tests, χ² tests, and ANCOVA, adjusting for key variables such as age, sex, and body mass index.

RESULTS

Both groups demonstrated poor sleep quality, with the mean PSQI scores above the clinical threshold of 5. Patients in the lithium group exhibited significantly better habitual sleep efficiency (lithium: 0.47 ± 0.65, valproic acid: 0.78 ± 0.87, P = 0.009) and fewer sleep disturbances (lithium: 1.26 ± 0.57, valproic acid: 1.61 ± 0.84, P = 0.005). Other sleep parameters, including total sleep duration (P = 0.082) and sleep latency (P = 0.625), did not differ significantly.

CONCLUSION

Patients in the lithium group showed significantly better habitual sleep efficiency and fewer sleep disturbances compared to those receiving valproic acid, although other sleep parameters did not differ. These findings suggest a potential advantage of lithium in certain aspects of sleep quality in BD patients under remission. Future studies using objective sleep measures and longitudinal designs are warranted to confirm these findings.

Key Words: Sleep quality; Valproic acid; Bipolar disorder; Lithium; Mood stabilizers

Core Tip: This study explored how two commonly used mood stabilizers, lithium and valproic acid, affect sleep quality in patients with bipolar disorder (BD) during remission. Although overall sleep quality was poor in both groups, patients treated with lithium showed significantly better sleep efficiency and fewer sleep disturbances compared to those receiving valproic acid. These findings highlight the potential advantage of lithium in promoting more stable and restorative sleep in BD. Given the critical role of sleep in the long-term course and relapse risk of BD, our results suggest that sleep-related outcomes should be considered when selecting maintenance treatments. Further studies using objective sleep measures and longitudinal designs are needed to build on these findings.

INTRODUCTION

Bipolar disorder (BD) is a long-term and recurring mood condition that is marked by fluctuating episodes of depression, hypomania, and mania, which can profoundly affect an individual's quality of life and general well-being[1,2]. BD manifests through a range of symptoms, among which sleep disturbance stands out as a key feature that persists even during remission phases, contributing to both severity of illness and the risk of recurrence[3-6]. Numerous studies have highlighted a bidirectional relationship between BD and sleep disorders, where sleep disturbances have not only emerged as a symptom of mood episodes but can also act as potential triggers for their onset and relapse[3,4,7]. Disruptions in sleep patterns, including decreased sleep duration, fragmented sleep, and alterations in circadian rhythms have been associated with worse outcomes and increased risk of relapse[1,8,9]. Since sleep regulation plays a fundamental role in the pathophysiology of BD, an evaluation of how pharmacological treatments can influence sleep parameters is essential.

Mood stabilizers, particularly lithium and valproic acid, can effectively prevent recurrences in mood episodes and stabilize affective fluctuations; these drugs have therefore remained as the cornerstone of BD treatment[10,11]. Lithium is well-documented for its neuroprotective effects and mood-stabilizing capabilities, showing effectiveness in decreasing both the occurrence and intensity of depressive and manic episodes[12,13]. Moreover, lithium has been linked to enhanced sleep efficiency, increased slow-wave sleep, and the regulation of circadian rhythms, which tend to be dysregulated in BD[13,14]. Valproic acid is another widely used mood stabilizer that is primarily prescribed for its effectiveness in controlling manic symptoms and in preventing relapses[15]. However, its impact on sleep architecture remains less clear, with some studies suggesting potential disruptions in sleep efficiency and duration of rapid eye movement (REM) sleep[16,17]. Studies also suggest that valproic acid may prolong non-REM (NREM) sleep while negatively affecting sleep continuity[17].

Despite their widespread use in BD treatment, the effects of lithium and valproic acid on sleep quality remain insufficiently studied[13,17-20] Given the significance of sleep disturbances in BD, an investigation into how these pharmacological agents can influence sleep parameters is crucial[16]. Although lithium and valproic acid are both mood stabilizers, their impact on sleep architecture may differ, necessitating further examination[15].

The current study set out to investigate how lithium and valproic acid, which are among the widely used mood stabilizers, influence sleep quality in patients with BD during remission. Rather than examining these agents in isolation, we assessed their effects within the framework of real-world treatment regimens, in which lithium or valproic acid was used either as monotherapy or in combination with adjunctive antipsychotics. This approach reflects the common clinical practice and enhances the overall validity of our findings. Based on prior literature, we hypothesized that patients treated with lithium-based regimens would demonstrate superior sleep quality compared to those receiving valproic acid-based regimens. By evaluating sleep-related parameters across these two treatment groups, we aimed to shed light on the distinct effects of these mood stabilizers on sleep architecture. Our data can help psychiatrists to make informed pharmacological decisions and strategies to address persistent sleep disturbances in BD.

MATERIALS AND METHODS

Study design

This cross-sectional study compared the sleep quality of individuals diagnosed with BD who were in remission and under maintenance treatment with either lithium or valproic acid. The study was conducted at the outpatient psychiatry clinic of Erzurum City Hospital. Approval for the study was granted by the Scientific Research Ethics Committee of the Faculty of Medicine at Erzurum University (IRB approval: 2025/03-88). All study procedures were designed in compliance with the ethical principles outlined in the Helsinki Declaration. Written informed consent was obtained from all individuals involved in the study prior to their participation.

Study participants

The current study included patients with a confirmed diagnosis of BD based on the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition. The diagnosis and remission status of the patients were validated by two experienced psychiatrists. Patients with a Hamilton Depression Rating Scale (HAM-D) score that remained below the threshold of 8 and a Young Mania Rating Scale (YMRS) score that did not exceed 7 were considered to be in remission. All study participants were aged between 18 and 65 years and had been receiving with either lithium or valproic acid at a stable dosage for at least six months before their enrollment to the current study. This criterion ensured that the pharmacological effects of the drugs remained stable during the assessment period. Patients with a history of alcohol or substance use disorder, diagnosis of neurological conditions that may independently affect sleep (e.g., epilepsy, stroke, multiple sclerosis, traumatic brain injury), significant medical comorbidities such as uncontrolled diabetes, cardiovascular disease, chronic respiratory disorders, or severe hypothyroidism, and those with primary sleep disorders (e.g., obstructive sleep apnea, chronic insomnia, restless legs syndrome) were excluded from the study. Additional exclusion criteria included other comorbid psychiatric diagnoses (e.g., schizophrenia, posttraumatic stress disorder, or anxiety disorders), chronic pain syndromes (e.g., fibromyalgia), current night-shift work, changes in psychotropic medication within the past three months, and the use of multiple mood stabilizers or high-dose benzodiazepines or hypnotic agents. These criteria were applied to minimize the potential confounding factors and to specifically isolate the effects of lithium and valproic acid on sleep quality. After applying these criteria, 130 patients met the eligibility requirements, with 78 receiving lithium and 52 receiving valproic acid, either alone or in combination with adjunctive antipsychotic medications.

Clinical and psychological assessments

Sociodemographic and clinical characteristics of the patients including age, sex, duration of illness, and history of episodes were recorded through structured clinical interviews. Depressive symptoms were assessed with the HAM-D, a 17-item clinician-administered tool that is widely used to measure depression severity, with remission indicated by a score below 8. Manic symptoms were evaluated using the YMRS, an 11-item instrument that quantifies the severity of manic episodes, with a remission threshold of 7 or lower. Sleep disturbances and overall sleep quality were assessed using the Pittsburgh Sleep Quality Index (PSQI). This well-established assessment tool evaluates the quality of sleep across seven components, with impaired sleep quality reflected by a total score of at least 5. The Turkish versions of these instruments have been previously validated for reliability in the relevant clinical population[21-23].

Procedure

After providing informed consent, the participants underwent a structured clinical evaluation that included standardized psychiatric assessments and collection of sociodemographic data. Assessment of sleep quality and functional status were completed within the same session. Anthropometric measurements, including height and weight, were recorded as part of the clinical examination. All assessments were conducted by trained psychiatrists to ensure consistency and reliability in data collection.

Statistical analysis

Data obtained in the current study was assessed using SPSS 23.0 software for statistical evaluation. Continuous variables were summarized as mean ± SD. The two study groups were compared with an independent samples t-test. Categorical data were presented as frequencies (n) and percentages (%), with group differences assessed with the χ² test. To evaluate variations in sleep quality between treatment groups, an analysis of covariance (ANCOVA) was conducted, adjusting for potential confounders such as age, sex, and body mass index (BMI). Additionally, relationships between sleep quality and clinical parameters were explored through Pearson correlation analyses. Statistical significance was set at P < 0.05 for all analyses.

A power analysis was conducted to assess the statistical adequacy of the study. With a medium effect size (Cohen’s d = 0.5), a significance level of 5% (α = 0.05), and a statistical power of 80% (1-β = 0.80), the obtained power with the current sample size (lithium = 52, valproic acid = 78) was calculated as 79.2%. This value was considered to be sufficient for the statistical reliability of the study, and the analyses were conducted accordingly.

RESULTS

The demographic and clinical characteristics of patients undergoing treatment with either lithium or valproic acid are shown in Table 1. The analysis indicated no significant difference in age in years (lithium: 46.46 ± 11.13, valproic acid: 44.26 ± 10.10, P = 0.256), years of formal education (lithium: 10.50 ± 4.49, valproic acid: 11.21 ± 4.25, P = 0.368), BMI (lithium: 29.27 ± 4.81, valproic acid: 29.52 ± 5.16, P = 0.780), duration of illness in years (lithium: 20.57 ± 8.94, valproic acid: 19.40 ± 10.12, P = 0.489), or age at first episode in years (lithium: 28.02 ± 20.30, valproic acid: 25.28 ± 8.56, P = 0.360) between the two groups. However, the total number of episodes was significantly higher in the valproic acid group (lithium: 6.46 ± 3.82, valproic acid: 8.80 ± 6.73, P = 0.013). Additionally, the number of depressive episodes was marginally higher in the valproic acid group (lithium: 3.00 ± 3.06, valproic acid: 4.28 ± 4.36, P = 0.050). The differences in sex distribution (41.0% males for lithium, 32.7% males for valproic acid, P = 0.337), marital status (33.3% not married in Lithium, 46.2% not married in valproic acid, P = 0.141), employment status (65.4% with no/irregular work in lithium, 71.2% with no/irregular work in valproic acid, P = 0.491), or smoking habits (53.8% smokers in lithium, 44.2% smokers in valproic acid, P = 0.283) between the two groups did not reach statistical significance. The presence of psychotic features during the episodes also did not differ significantly between the two groups (lithium: 76.9%, valproic acid: 73.1%, P = 0.618). Furthermore, there was no significant difference in the history of electroconvulsive therapy (lithium: 37.2%, valproic acid: 21.2%, P = 0.052), suicide attempts (lithium: 28.2%, valproic acid: 25.0%, P = 0.686), self-mutilation (lithium: 2.6%, valproic acid: 3.8%, P = 0.678), or family history of BD (lithium: 34.6%, valproic acid: 38.5%, P = 0.655).

Table 1 Demographic and clinical characteristics of bipolar disorder patients treated with lithium or valproic acid, mean ± SD or n (%).

	All patients (n = 130)		t value	χ2	P value
	Lithium (n = 78)	Valproic acid (n = 52)
Age	46.46 ± 11.13	44.26 ± 10.10	1.140		0.256
Education level (years)	10.50 ± 4.49	11.21 ± 4.25	-0.904		0.368
Body mass index	29.27 ± 4.81	29.52 ± 5.16	-0.280		0.780
Duration of illness (years)	20.57 ± 8.94	19.40 ± 10.12	0.695		0.489
Total number of episodes	6.46 ± 3.82	8.80 ± 6.73	-2.526		0.013
Number of depressive episodes	3.00 ± 3.06	4.28 ± 4.36	-1.978		0.050
Age at the first episode	28.02 ± 20.30	25.28 ± 8.56	0.918		0.360
Sex				0.923	0.337
Male	32 (41.0)	17 (32.7)
Female	46 (59.0)	35 (67.3)
Marital status (not married)	26 (33.3)	24 (46.2)		2.167	0.141
Employment				0.475	0.491
No/irregular	51 (65.4)	37 (71.2)
Regular	27 (34.6)	15 (28.8)
Smoking				1.154	0.283
Yes	42 (53.8)	23 (44.2)
No	36 (46.2)	29 (55.8)
History of ECT (yes)	29 (37.2)	11 (21.2)		3.765	0.052
History of suicide attempt (yes)	22 (28.2)	13 (25.0)		0.163	0.686
History of self-mutilation (yes)	2 (2.6)	2 (3.8)		0.172	0.678
Family history of bipolar disease (yes)	27 (34.6)	20 (38.5)		0.200	0.655
Psychotic features during episodes (present)	60 (76.9)	38 (73.1)		0.249	0.618
Diagnosis				0.836	0.361
Bipolar I disorder	61 (78.2)	37 (71.2)
Bipolar II disorder	17 (21.8)	15 (28.8)

Student's t-test and χ² test were used, P < 0.05 was considered as statistically significant (bold values). ECT: Electroconvulsive therapy.

A comparison of the sleep quality between the lithium and valproic acid treated groups is shown in Table 2. The total PSQI score was higher in the valproic acid group (lithium: 6.91 ± 3.54, valproic acid: 8.13 ± 4.38), although the difference was not statistically significant (P = 0.114). Among the PSQI subcomponents, the valproic acid group had significantly worse habitual sleep efficiency (lithium: 0.47 ± 0.65, valproic acid: 0.78 ± 0.87, P = 0.009) and sleep disturbances (lithium: 1.26 ± 0.57, valproic acid: 1.61 ± 0.84, P = 0.005). However, no significant differences were observed in the subjective sleep quality (lithium: 0.84 ± 0.56, valproic acid: 1.07 ± 0.76, P = 0.169), sleep latency (lithium: 1.57 ± 1.21, valproic acid: 1.65 ± 1.18, P = 0.625), sleep duration (lithium: 0.33 ± 0.67, valproic acid: 0.53 ± 0.89, P = 0.082), use of sleep medication (lithium: 1.66 ± 1.48, valproic acid: 1.53 ± 1.48, P = 0.539), or daytime drowsiness (lithium: 0.74 ± 1.02, valproic acid: 0.92 ± 1.15, P = 0.344) between the two groups.

Table 2 Differences in sleep quality between lithium or valproic acid treated bipolar disorder patients, mean ± SD.

All Patients (n = 113)	Lithium (n = 78)	Valproic acid (n = 52)	F	P value
Pittsburgh sleep quality index total	6.91 ± 3.54	8.13 ± 4.38	2.553	0.114
Subjective sleep quality	0.84 ± 0.56	1.07 ± 0.76	1.921	0.169
Sleep latency	1.57 ± 1.21	1.65 ± 1.18	0.240	0.625
Sleep duration	0.33 ± 0.67	0.53 ± 0.89	3.073	0.082
Habitual sleep efficiency	0.47 ± 0.65	0.78 ± 0.87	7.035	0.009
Sleep disturbances	1.26 ± 0.57	1.61 ± 0.84	8.084	0.005
Use of sleep medication	1.66 ± 1.48	1.53 ± 1.48	0.379	0.539
Daytime drowsiness	0.74 ± 1.02	0.92 ± 1.15	0.903	0.344

ANCOVA analyses were adjusted for covariates, with age and sex, body mass index included for most variables. P < 0.05 statistically significant (bold values).

The distribution of medication usage among the participants receiving lithium and valproic acid is summarized in Table 3. The proportion of patients receiving lithium or valproic acid as a monotherapy was 41.0% and 34.6%, respectively. Among the adjunctive antipsychotic medications, the most frequently used were quetiapine (lithium: 19.2%, valproic acid: 21.2%), olanzapine (lithium: 20.5%, valproic acid: 7.7%), and aripiprazole (lithium: 14.1%, valproic acid: 13.5%). While paliperidone (9.6%) and risperidone (9.6%) were more frequently used in the valproic acid group, amisulpride was used only in the lithium group (1.3%). A comparative statistical analysis revealed no significant difference in the use of adjunctive antipsychotics between the two groups (P = 0.205).

Table 3 Characteristics of medication use in bipolar disorder patients, n (%).

Adjunctive antipsychotic medication	Lithium (n = 78)	Valproic acid (n = 52)	χ2	P value
None	30 (41.0)	18 (34.6)	9.721	0.205
Quetiapine	15 (19.2)	11 (21.2)
Olanzapine	16 (20.5)	4 (7.7)
Aripiprazole	11 (14.1)	7 (13.5)
Paliperidone	0 (0)	5 (9.6)
Risperidone	2 (2.6)	5 (9.6)
Clozapine	0 (0)	1 (1.9)
Amisulpride	1 (1.3)	0 (0)
Haloperidol	1 (1.3)	1 (1.9)

χ² test were used.

Table 4 presents the correlations between sleep quality and clinical variables. The PSQI total score correlated positively with BMI (r = 0.291, P = 0.001) and age (r = 0.275, P = 0.002). No meaningful associations were identified between the PSQI total score and variables such as illness duration (r = 0.137, P = 0.119), total episode count (r = 0.024, P = 0.784), number of depressive episodes (r = 0.100, P = 0.259), or level of education (r = -0.112, P = 0.205).

Table 4 Correlations between sleep quality and clinical variables in bipolar disorder patients.

	Correlations	Duration of illness	Total number of episodes	Number of depressive episodes	Body Mass Index	Education levels	Age
Pittsburgh sleep quality index total	Pearson correlation	0.137	0.024	0.100	0.2911	-0.112	0.2751
	Sig. (2-tailed)	0.119	0.784	0.259	0.001	0.205	0.002

¹Correlation is significant at the 001 Level (2-tailed).

DISCUSSION

The current cross-sectional study is among the very few that have investigated the differential effects of lithium and valproic acid on sleep quality in BD patients under remission. Although previous studies have explored the role of mood stabilizers in sleep regulation, data on the direct comparison of the effect of these two commonly prescribed agents on sleep quality is highly limited. Our findings indicate that lithium-treated patients exhibited better habitual sleep efficiency and experienced fewer sleep disturbances compared to those receiving valproic acid, suggesting that lithium may promote more stable sleep patterns. These differences were observed within a clinical context where both groups were maintained on either lithium or valproic acid as their primary mood stabilizer, reflecting typical maintenance treatment strategies in real-world practice. Furthermore, our results align well with a recent study by Ermis et al., which highlighted the continued efficacy of lithium and its relevance in the management of BD patients[24].

Sleep and circadian rhythm disruptions are commonly observed in BD and generally extend beyond acute mood episodes with a tendency to persist in remission, although with diminished intensity[7]. These disturbances may reflect the core neurobiological and genetic susceptibilities associated with BD and have been linked to increased recurrence risk despite pharmacological treatment[16,25]. In the current study, both the lithium and valproic acid groups demonstrated poor sleep quality, with the mean PSQI scores exceeding the clinical threshold of 5. This aligns well with previous literature highlighting the pervasiveness of sleep disruptions in BD[26,27]. Our findings further emphasize the need for targeted interventions to improve sleep quality as a fundamental component of BD management.

Lithium is well-documented for its impact on sleep architecture and its ability to stabilize circadian rhythms in individuals with BD. It has been shown to improve sleep efficiency, increase slow-wave sleep, and reduce the duration of REM sleep and prolong REM latency, all of which may contribute to its stabilizing effects on mood[13]. Additionally, lithium has been linked to phase-shifting circadian rhythms and delayed sleep-wake cycles, reinforcing its regulatory influence on sleep patterns[7]. At the molecular level, lithium appears to regulate circadian function through the inhibition of GSK-3β, a key enzyme involved in the modulation of core clock genes such as PER2, thereby contributing to its chronobiotic properties and potential to enhance sleep stability[28]. In contrast, research on the effects of valproic acid on sleep remains inconclusive. Some studies indicate that valproic acid increases the NREM sleep duration[17]. The use of valproic acid, lamotrigine, and carbamazepine was reported to be associated with increased total sleep duration and improved sleep onset latency[17,29]. Carbamazepine was also reported to enhance sleep intensity by favoring NREM sleep[30]. Other studies indicate that the sedative properties of valproic acid do not necessarily improve sleep efficiency[31]. Our findings are largely aligned with these observations, demonstrating that lithium-treated patients exhibited comparatively superior habitual sleep efficiency and fewer sleep disturbances compared to those treated with valproic acid. These results suggest that the regulatory influence of lithium on sleep stability may be a key factor that contributes to its broader therapeutic benefits in BD management.

Pharmacological treatments used in BD such as mood stabilizers, antidepressants and antipsychotics, are known to influence sleep quality[32]. Among these, antipsychotics have been especially associated with negative effects on sleep[33]. In the current study, a subset of patients in both lithium and valproic acid groups were receiving adjunctive antipsychotic treatment, although there were no significant differences in the use of these drugs between the two groups. This suggests a comparable distribution of the use of additional antipsychotics between the two groups, which also minimizes any potential confounding effects of the antipsychotics on sleep outcomes. However, the potential adverse effects of such drugs on sleep cannot be entirely ruled out.

Sleep quality can be influenced by multiple factors beyond the effects of medications. These factors include age, BMI, and medical conditions, all of which have been linked to various health complications[34]. Our findings are aligned with the published literature, since we observed that older age and higher BMI were associated with worse sleep quality in BD patients. These results highlight the complex interplay between biological factors and sleep regulation, and emphasize the need for a multifaceted approach when addressing sleep disturbances in BD management.

The limitations of the current study should be acknowledged. First, the cross-sectional nature of the study limits the establishment of a clear causal link between the use of mood stabilizers and sleep quality. Second, although the number of patients receiving adjunctive antipsychotics in the two groups were comparable, these drugs may still have influenced sleep parameters that could not be detected with the tools used in the current study. Third, serum levels of lithium and valproic acid were not monitored during the study, which limits the evaluation of any dose-response relationships. Fourth, the relatively modest sample size may have decreased the statistical power, particularly for subgroup analyses, and limits the generalizability of the findings. Lastly, sleep quality was assessed using self-reported measures rather than objective tools such as polysomnography; the latter could have provided a more precise evaluation of sleep architecture. Future research should utilize longitudinal designs and incorporate objective sleep assessments such as actigraphy or polysomnography, along with neurophysiological measures of sleep architecture (e.g., REM, NREM, and deep sleep stages), to further clarify the impact of mood stabilizers on sleep in BD patients.

CONCLUSION

The current study provides evidence for the differential effects of lithium and valproic acid on sleep quality in remitted BD patients. Our findings suggest that lithium treatment was associated with better habitual sleep efficiency and fewer sleep disturbances compared to valproic acid, highlighting a potential role of the former in stabilizing sleep patterns. Given the strong link between sleep disturbances and BD recurrence, our findings emphasize the importance of considering sleep quality in treatment strategies. However, due to the cross-sectional design and reliance on subjective sleep assessments, further longitudinal studies with objective sleep measures are needed to clarify the long-term effects of mood stabilizers on sleep architecture in BD.

ACKNOWLEDGEMENTS

The authors would like to thank the volunteers who participated in the study.