Effect size of lithium, carbamazepine, and sodium valproate in child and adolescent bipolar 1 disorder during manic phase: A prospective open-label study.

AIM: The aim was to evaluate the "effect size (ES)," tolerability, and acceptability of lithium, carbamazepine, and sodium valproate in the acute phase treatment of pediatric Bipolar 1 disorder patients during manic phase.
MATERIALS AND METHODS: This hospital-based, prospective, open-label study included 67 patients in manic phase of bipolar I disorder, aged 6-17 years, after informed consent by the caregivers. The patients were randomly assigned to the lithium group (n = 30), carbamazepine group (n = 20), and sodium valproate group (n = 17). They were assessed with the Schedule for Affective Disorders for School Age Children's-Present and Life time version administered to the parent and child separately, Conner's Abbreviated Rating Scale, and Cassidy Scale for Manic States (CSMS). Lithium was started in the dose of 30 mg per kg of body weight, carbamazepine in the dose of 10-20 mg/kg/day, and sodium valproate in the dose of 10-20 mg/kg body weight. Antipsychotic (chlorpromazine [CPZ] 100-500 mg per day or haloperidol up to 750 mg of CPZ equivalent) was allowed in the study. Injection haloperidol 10 mg and injection promethazine 50 mg intramuscular were allowed for initial 3-5 days to combat acute agitation. Rescue medication such as injection lorazepam 2-4 mg intramuscular was allowed throughout the study duration. The patients were rated weekly on CSMS, Bipolar Clinical Global Impression, Udvalg for kliniske Undersogelser Side Effect Rating Scale, and side effect checklist for lithium, sodium valproate, and carbamazepine, respectively. The serum level of concerned drug was obtained at weekly intervals and dose hiked, if needed to get target serum level.
RESULTS: The response rate was 90% in lithium group, 70% in carbamazepine group, and 88% in sodium valproate group on the basis of ≥33% reduction from baseline CSMS. The effects of change of CSMS over the 6 weeks across the three treatment group were found to be highly statistically significant.
CONCLUSIONS: In the acute phase treatment of pediatric bipolar 1 disorder patients during manic phase, the ES for lithium was 0.85, for carbamazepine 0.71, and for sodium valproate 0.84. These agents are well tolerated in treating bipolar disorder in children. Copyright:

Adolescent bipolar disorder (BD) was described by Kraepelin who noted a significant emergence of BD at puberty.[1] Once thought to occur only rarely in youth, BD is now estimated to affect 1% of children and adolescents.[2] It has been reported that when patients recalled their first mood episode, approximately 65% of adults experienced onset of symptoms prior to the age of 18, while 27.7% experienced their first mood episode before the age of 13 years.[3] An estimated 30%–40% of the child and adolescent psychiatric hospitalization are due to BD.[4] There is growing evidence that early mood disorders are widespread, recurrent, and often chronic, increasing the risk of lifelong disability.[5] Children and adolescents with BD have significantly higher rates of morbidity and mortality as compared to healthy children. In addition, the disorder results in impaired social, family, and academic functioning, resulting in reduced quality of life.[6] Pediatric BD disrupts the normal development of children and adolescents.[7] The developmental variations in presentation, symptomatic overlap with other disorders, and lack of clinician awareness have all led to underdiagnosis or misdiagnosis in children and adolescents.[8]

Effect size (ES) is a name given to a family of indices that measure the magnitude of a treatment effect. Unlike significance tests, these indices are independent of sample size. EZ measures are the common currency of meta-analysis studies that summarize the findings from a specific area of research.[9] There is a wide array of formulas used to measure ES. In general, ES can be measured in two ways: as the standardized difference between two means, or as the correlation between the independent variable classification and the individual scores on the dependent variable. This correlation is called the “ES calculation.”[10] ESs can also be thought of as the average percentile standing of the average treated (or experimental) participant relative to the average untreated (or control) participant. An ES of 0.0 indicates that the mean of the treated group is at the 50th percentile of the untreated group. An ES of 0.8 indicates that the mean of the treated group is at the 79th percentile of the untreated group. An ES of 1.7 indicates that the mean of the treated group is at the 95.5th percentile of the untreated group. ESs can also be interpreted in terms of the percent of nonoverlap of the treated group’s score with those of the untreated group. An ES of 0.0 indicates that the distribution of score for the treated group overlaps completely with the distribution of scores for the untreated group; there is 0% of the nonoverlap. An Es of 0.8 indicates a nonoverlap of 47.4% in the two distributions. An ES of 1.7 indicates a nonoverlap of 75.4% in the two distributions.

Despite advances in the understanding of the symptomatology and phenomenology of BD in children and adolescents, there remains a dearth of information regarding pharmacotherapy for BD. Current practice parameters are based on limited evidence and/or studies using adult patients. General guidelines indicate that it is important for young patients with BD to continue treatment for an extended period of time in order to manage the frequent relapses of childhood-and adolescent-onset BD.[611] There is a paucity of systematic studies on the efficacy of mood stabilizers for prepubertal BD. This is a priority area given the morbidity and chronicity of this disorder. Studies of mood stabilizers for BDs are few, are small, and have used heterogeneous inclusion/exclusion criteria, making it difficult to know for which populations this approach is useful. Because such disorders are relatively intractable, this area deserves more study. In addition, comparative studies examining the efficacy of these agents, including time to response, have not been undertaken. Given data that valproate may have a quicker onset of action than lithium in bipolar adults and that it can be given in a rapid loading strategy, comparison of active treatments in adolescent BD might permit more rational treatment strategies.[1213] Systematic assessment of frequent and infrequent side effects of these compounds in children with psychiatric disorders is also needed. Existent data in the literature do not completely address side effects that may be more frequent in pediatric psychiatric populations. The present study is an attempt to give a better insight into the psychopharmacological treatment by evaluating the “ES” for lithium, carbamazepine, and sodium valproate in the acute phase treatment of bipolar 1 disorder patients during a manic phase and to ascertain the tolerability and acceptability of these three mood stabilizers in the child and adolescent population.

MATERIALS AND METHODS

This hospital-based, prospective, open-label study was conducted at the Department of Child and Adolescent Psychiatry, Central Institute of Psychiatry, Kanke, Ranchi, a tertiary care postgraduate teaching institute. It has a total bed capacity of 643, out of which 18 beds are for the Department of Child and Adolescent Psychiatry. The study protocol was approved by the Institutional Ethical Committee.

Sample

Samples were collected using purposive sampling method from inpatients of the Child and Adolescent Psychiatry ward.

Inclusion criteria

Patients meeting Diagnostic and Statistical Manual of Mental Disorder-IV Text Revision (TR) diagnostic criteria for BD 1 during a manic episode[14]

Age between 6 and 17 years

Score >15 on Cassidy Scale for Manic States (CSMS)[15]

Informed consent by caregivers, i.e. parents or major first-degree relatives.

Exclusion criteria

Current medical or neurological conditions requiring treatment

Mental retardation judged by academic performance

Current or lifetime diagnosis of schizophrenia, autism, attention deficit hyperkinetic disorder (ADHD), obsessive compulsive disorder, depression and anxiety disorder

Substance abuse within the last 6 months

Use of psychotropic medicines within the last 2 weeks including narcoleptics, stimulants, and antidepressants

One month drug free if on depot antipsychotic.

Tools for assessment

Sociodemographic and clinical data sheet

It contained questions about age, sex, religion, informant, address, past history, family history, total number of episodes, and development history, which was filled up after a detailed interview with the patient and the caregiver.

Schedule for Affective Disorders for School Age Children’s-Present and Life time version

This was administered to the parent (usually the mother) and child separately. This tool was used for screening and to make the diagnosis. Apart from the current diagnosis, Schedule for Affective Disorders for School Age Children’s-Present and Life time versions (K-SADA-PL) also gives the lifetime diagnosis.[16]

Conner’s Abbreviated Rating Scale

The scale was developed by Conner to evaluate and diagnose ADHD. The items are rated on a 0–3 point scale. This study used the teacher version. It was used to exclude comorbid diagnosis of ADHD.[17]

Cassidy Scale for Manic States

The scale was used to assess the manic along with the mixed features on a 0–5 point scale. This consists of 20-point scale.[15]

Bipolar Clinical Global Impression-Bipolar versions

This is a 3-point scale which measures illness severity, global improvement, and therapeutic response. Each item on the Clinical Global Impression (CGI) is rated separately, and there is no overall score. The CGI is designed to be useful in situations in which changes over time have to be assessed.[18]

Side effect checklist for lithium, sodium valproate, and carbamazepine

This checklist was used to assess the side effect of lithium, sodium valproate, and carbamazepine separately.[19]

Udvalg for Kliniske Undersogelser Side Effect Rating Scale

This scale gives a comprehensive assessment of side effect related to antipsychotic use in four categories – psychic, neurologic, autonomic, and others. It also gives a global assessment of the interference by the existing side effects with the patient’s daily performance and its consequences on treatment regimen.[20]

Procedure

Patients were taken from the Department of Child and Adolescent Psychiatry with a diagnosis of BD 1 current episode-mania or mixed. Their parents were explained about the study, and written informed consent was obtained. They were allotted randomly to a naturalistic open-label study design of 5 weeks to lithium, carbamazepine, and sodium valproate group.

A detailed physical examination was done to rule out any physical illness. Laboratory testing for complete hemogram, blood urea, serum creatinine, liver function test, thyroid function test, and electrocardiogram was done. Dosage of the drug and their serum level was monitored at the 1st, 2nd, 3rd, 4th, and 5th weeks of the treatment. The primary efficacy measures were the weekly CGI-Bipolar version (BP) on improvement subscale and CSMS.

Drug schedule

Lithium was started at a dose of 30 mg per kg of body weight.[21] The target serum level was 0.8–1.2 mmol/L.[22] Carbamazepine was started at a dose of 10–20 mg/kg/day (200–600 mg/day). The target serum level was 8–12 μg per liter.[23] Sodium valproate was started at a dose of 10–20 mg per kg of body weight. The target serum level was 85–110 μg per liter.[24]

Daily dosages were usually divided into twice-daily or thrice-daily schedules.[25] The medication such as injection lorazepam 2–4 mg intramuscular was allowed throughout the study duration.

The treating team was allowed to use tablet chlorpromazine (CPZ) at a dose of 100–500 mg per day or tablet haloperidol at a dose of 5–15 mg per day. Injection haloperidol 10 mg and injection promethazine 50 mg intramuscular were allowed for initial 3–5 days to combat acute agitation at the time of admission.

Assessment

At the time of admission, both patients and parents were interviewed. Information was gathered and sociodemographic datasheet was filled up. K-SADS-PL, CARS, and Baseline Ratings on CSMS were done. Patients were rated weekly on CSMS; CGI-BP; side effect checklist for lithium, sodium valproate, and carbamazepine; and Udvalg for Kliniske Undersogelser (UKU) Scale. Serum level of concerned drug was obtained and dose hiked, if needed to get target serum level.

Statistical analysis

Statistical analysis was done with Statistical Package of Social Sciences (SPSS) – version, 13.0 (SPSS13; IBM, Chicago, USA). To find out the group differences of the clinical variables having categorical data, the Chi-square test was used. Post hoc test was done using least significant fdifference (LSD) model because there was no significant difference with the use of other post hoc test models. Multivariate tests were done to show the multiple comparison across the three treatment groups using CSMS and CSMS versus lithium carbonate, carbamazepine, and sodium valproate group. Responders were classified as ≥33% reduction in CSMS or CGI-BP score of 1 or 2 in improvement subscale. ES was calculated by first calculating Cohen’s “d,” which is calculated as difference in sample mean divided by their common standard deviation.[2627] Tolerability and acceptability of lithium, carbamazepine, and sodium valproate were assessed using the side effect checklist to show the percentage of side effect across the three groups. P =0.05 was used throughout the analysis.

RESULTS

A total of 67 children and adolescent patients with Bipolar I disorder participated in this study after each caregiver gave informed consent. The patients were randomly assigned to lithium group (30 participants), carbamazepine group (20 participants), and sodium valproate group (17 participants). Out of a total of 67 patients, only 37 patients completed the 6 week study protocol, which included 20 in the lithium group, 9 in the carbamazepine group, and 8 in the sodium valproate group. The major reason for reduced level of participation during the initial week was the change of treatment regimen (e.g. use of electroconvulsive therapy or change of antipsychotic) due to patient being restless and aggressive, whereas the dropouts during later weeks were due to early discharge of the patients. Intent-to-treat analysis was done on the basis of Last Observation Carried Forward (LOCF) method for the patients who completed at least 1 week of the study (n = 67). LOCF method is an established method for substituting missing values in longitudinal studies.[28]

There was no statistically significant difference in the demographic and clinical variables in the three groups, indicating that the three groups were homogenous [Tables 1 and 2]. There was a statistically significant difference with the use of lithium in patients with a family history of affective disorders (P = 0.034). There was no significant difference with the use of carbamazepine in patients with any of the variables. The number of patients receiving sodium valproate reached statistically significant difference with the total number of episodes (P = 0.021) and past depressive episodes. There was also a statistically significant difference (P = 0.007) among the patients receiving sodium valproate with the number of past depressive episodes [Table 3].

Table 1

Sociodemographic details of the sample and comparison across the three treatment groups

Variables	Lithium (n=30), n (%)	Carbamazepine (n=20), n (%)	Sodium valproate (n=17), n (%)	df	χ2	P
Sex
Male	16 (53)	14 (70)	13 (76)	2	2.94	0.22
Female	14 (47)	6 (30)	4 (24)
Religion
Hindu	24 (80)	16 (80)	14 (82)	2	0.04	0.97
Muslim	6 (20)	4 (20)	3 (18)
Education
Illiterate	1 (03)	3 (15)	2 (12)	6	8.41	0.20
Primary	12 (40)	5 (25)	2 (12)
High school	17 (57)	11 (55)	13 (76)
Senior secondary	00	1 (05)	0
Number of past manic episodes
Nil	26 (87)	16 (80)	11 (66)	6	8.96	0.17
One	3 (10)	3 (15)	3 (17)
Two	1 (03)	0	3 (17)
Three	0	1 (05)	0
Number of past depressive episodes
Nil	22 (73)	16 (80)	15 (88)	2	1.47	0.47
One	8 (27)	4 (20)	2 (12)
Family history
Nil	14 (47)	13 (65)	10 (59)	4	5.96	0.20
Affective illness	15 (50)	4 (20)	6 (35)
Nonaffective illness	1 (03)	3 (15)	1 (06)
Type of onset of index episodes
Abrupt	9 (30)	8 (40)	12 (70)	4	8.84	0.06
Acute	17 (57)	11 (55)	5 (30)
Insidious	4 (13)	1 (05)	0
Total number of episodes (including index episodes)
One	20 (67)	14 (70)	10 (59)	6	7.65	0.26
Two	7 (23)	5 (25)	3 (18)
Three	3 (10)	0	4 (23)
Four	0	1 (05)	0
Antipsychotic drug
CPZ	21 (70)	15 (75)	16 (94)	2	3.74	0.15
Haloperidol	9 (30)	5 (25)	1 (06)

df – Degree of freedom; χ2 – Chi-square value; P<0.05; CPZ – Chlorpromazine

Table 2

Comparison of clinical variables in the three treatment groups

Clinical variables	Treatment group	Mean±SD	df	F	P
Age (years)	Lithium	15.33±1.72	2	0.28	0.756
	Carbamazepine	15.40±1.50
	Sodium valproate	15.70±1.75
Total number of past manic episode	Lithium	0.16±0.46	2	1.71	0.188
	Carbamazepine	0.30±0.732
	Sodium valproate	0.52±0.79
Past number of depressive episodes	Lithium	0.26±0.44	2	0.71	0.491
	Carbamazepine	0.20±0.41
	Sodium valproate	0.11±0.33
Duration of index episode (days)	Lithium	48.93±54.45	2	0.44	0.643
	Carbamazepine	34.90±36.19
	Sodium valproate	48.11±70.90
Total number of episodes (including index episode)	Lithium	1.43±0.67	2	0.59	0.556
	Carbamazepine	1.40±0.75
	Sodium valproate	1.64±0.86

df – Degree of freedom; P<0.05; F – ANOVA value. SD – Standard deviation

Table 3

Comparison of the clinical variables of the lithium, carbamazepine, and sodium valproate groups

Variables	Lithium group		Carbamazine group		Sodium valproate group
	n	χ2 (df), P	n	χ2 (df), P	n	χ2 (df), P
Sex
Male	16	2.71 (1), 0.100	14	2.14 (1), 0.143	13	1.12 (1), 0.290
Female	14		6		4
Family history
Nil	14	6.73 (2), 0.034	13	2.05 (2), 0.359	10	3.66 (2), 0.160
Affective	15		4		6
Nonaffective	1		3		1
Number of past manic episode
Nil	26	1.36 (2), 0.507	16	2.14 (2), 0.343	11	4.53 (2), 0.103
One	3		3		3
Two	1		1		3
Past number of depressive episodes
One	22	1.09 (2), 0.295	16	0.95 (2), 0.329	15	7.36 (2), 0.007
Two	8		4		2
Type of onset of index episode
Abrupt	9	2.40 (2), 0.300	8	2.84 (2), 0.241	12	0.04 (2), 0.825
Acute	17		11		5
Insidious	4		1		0
Total number of episodes (including index episode)
One	20	0.54 (2), 0.761	14	0.68 (2), 0.712	10	7.72 (2), 0.021
Two	7		5		3
Three	3		0		4
Four	0		1		0

df – Degree of freedom; χ2 – Chi-square value; P<0.05

Comparison of CSMS scores across the three treatment groups showed a marked reduction in the score of CSMS in the 2nd week in all the three groups with a worsening in the 4th week in lithium and carbamazepine group, but the difference was not statistically significant [Table 4]. Comparison of reduction in CSMS scores across the three treatment groups showed a marked reduction from the 1st to 3rd week in all the three groups. The reduction in CSMS scores across the three treatment groups was statistically significant at the 2nd week (P = 0.045), 4th week (P = 0.014), and 5th week (P = 0.043) [Table 5]. Table 6 shows the dose of mood stabilizer (mg/day) across the three treatment groups. Comparison of UKU total score week wise across the three treatment groups did not show statistically significant difference over the 6 weeks [Table 7]. Comparison of dose of trihexyphenidyl (mg/day) across the three treatment groups showed that it reached a statistically significant difference at the 1st week in the lithium group [Table 8]. The frequency of dystonia experienced by the patients across the three groups in the initial 3 weeks of the study is shown in Table 9. At the 1st week, lithium group experienced the highest number of dystonia.

Table 4

Comparison of Cassidy Scale for Manic States scores across the three treatment groups

CSMS	Treatment group	Mean±SD	df	F	P
Baseline	Lithium	45.03±7.26	2	1.814	0.171
	Carbamazepine	44.00±8.22
	Sodium valproate	48.47±6.746
First week	Lithium	24.00±8.97	2	0.045	0.956
	Carbamazepine	25.65±9.24
	Sodium valproate	24.82±12.99
Second week	Lithium	17.20±7.89	2	1.347	0.267
	Carbamazepine	22.00±12.98
	Sodium valproate	16.94±13.66
Third week	Lithium	15.97±11.11	2	0.684	0.508
	Carbamazepine	20.35±14.90
	Sodium valproate	16.52±15.44
Fourth week	Lithium	17.36±12.28	2	1.941	0.152
	Carbamazepine	21.70±15.04
	Sodium valproate	12.82±14.31
Fifth week	Lithium	13.30±11.45	2	1.492	0.233
	Carbamazepine	18.95±15.36
	Sodium valproate	12.00±14.52

df – Degree of freedom; P<0.05; F – ANOVA value; CSMS – Cassidy Scale for Manic States; SD – Standard deviation

Table 5

Comparison of reduction in Cassidy Scale for Manic States scores across the three treatment groups

CSMS	Treatment group	Mean±SD	df	F	P
Reduction in CSMS up to the 1st week	Lithium	20.52±9.19	2	1.58	0.213
	Carbamazepine	20.20±7.11
	Sodium valproate	18.35±6.82
Reduction in CSMS up to the 2nd week	Lithium	27.02±11.92	2	3.26	0.045
	Carbamazepine	27.83±9.09
	Sodium valproate	22.00±12.99
Reduction in CSMS up to the 3rd week	Lithium	28.11±13.33	2	1.95	0.150
	Carbamazepine	29.06±11.21
	Sodium valproate	23.65±15.02
Reduction in CSMS up to the 4th week	Lithium	28.08±14.10	2	4.51	0.014
	Carbamazepine	27.66±12.70
	Sodium valproate	22.30±14.00
Reduction in CSMS up to the 5th week	Lithium	31.73±12.68	2	3.31	0.043
	Carbamazepine	25.05±14.51
	Sodium valproate	36.47±14.20

CSMS – Cassidy scale for manic states; SD – Standard deviation; df – Degree of freedom; P<0.05; F – ANOVA value

Table 6

Dose of mood stabilizer (mg/day) across the three treatment groups

Dose of mood stabilizer (mg/day)	Treatment group	Mean±SD
Dose of mood stabilizer at baseline	Lithium	890.00±38.05
	Carbamazepine	400.00±0.00
	Sodium valproate	882.35±159.04
Dose of mood stabilizer at the 1st week	Lithium	1075.00±131.14
	Carbamazepine	555.00±82.55
	Sodium valproate	976.47±139.32
Dose of mood stabilizer at the 2nd week	Lithium	1165.00±140.28
	Carbamazepine	650.00±119.20
	Sodium valproate	1105.88±198.33
Dose of mood stabilizer at the 3rd week	Lithium	1220.00±161.13
	Carbamazepine	680.00±136.11
	Sodium valproate	1129.41±208.46
Dose of mood stabilizer at the 4th week	Lithium	1250.00±186.15
	Carbamazepine	710.00±165.11
	Sodium valproate	1141.17±193.83
Dose of mood stabilizer at the 5th week	Lithium	1250.00±186.15
	Carbamazepine	680.00±164.156
	Sodium valproate	1141.17±193.83

SD – Standard deviation

Table 7

Comparison of Udvalg for Kliniske Undersogelser-total score week wise across the three treatment groups

UKU rating score	Treatment group	Mean±SD	df	F	P
First week	Lithium	11.23±5.57	2	0.34	0.708
	Carbamazepine	11.20±3.23
	Sodium valproate	10.17±3.30
Second week	Lithium	8.17±3.94	2	1.98	0.146
	Carbamazepine	7.20±3.18
	Sodium valproate	6.05±2.97
Third week	Lithium	6.00±3.57	2	1.28	0.285
	Carbamazepine	6.50±4.92
	Sodium valproate	4.52±2.83
Fourth week	Lithium	5.20±4.05	2	1.35	0.265
	Carbamazepine	4.85±2.56
	Sodium valproate	3.64±1.41
Fifth week	Lithium	3.83±3.35	2	0.45	0.639
	Carbamazepine	4.05±2.18
	Sodium valproate	3.23±1.67

UKU – Udvalg for Kliniske Undersogelser Side Effect Rating Scale; df – Degree of freedom; P<0.05; F – ANOVA value; SD – Standard deviation

Table 8

Comparison of dose of trihexyphenidyl (mg/day) across the three treatment groups

Trihexyphenidyl (mg/day)	Treatment group	Mean±SD	df	F	P
Dose of trihexyphenidyl at 1st week	Lithium	2.66±1.51	2	24.62	0.000
	Carbamazepine	0.30±0.97
	Sodium valproate	0.47±1.32
Dose of trihexyphenidyl at 2nd week	Lithium	2.86±1.25	2	1.19	0.311
	Carbamazepine	2.50±1.43
	Sodium valproate	2.23±1.56
Dose of trihexyphenidyl at 3rd week	Lithium	2.86±1.13	2	0.30	0.738
	Carbamazepine	2.60±1.31
	Sodium valproate	2.82±1.23
Dose of trihexyphenidyl at 4th week	Lithium	2.86±1.00	2	0.60	0.551
	Carbamazepine	2.50±1.43
	Sodium valproate	2.82±1.23
Dose of trihexyphenidyl at 5th week	Lithium	2.46±1.00	2	0.52	0.597
	Carbamazepine	2.50±1.43
	Sodium valproate	2.82±1.23

df – Degree of freedom; P<0.05, F – ANOVA value. SD – Standard deviation

Table 9

Dystonia rates across the three treatment groups

Dystonia rates	Total	Lithium	Carbamazepine	Sodium valproate
First week	21	9	7	5
Second week	2	1	1	0
Third week	2	1	0	1

Table 10 shows the post hoc test using LSD model of requirement of mean antipsychotic (CPZ equivalent) dose and CSMS change from baseline to end of the study using multiple comparisons across the three treatment groups. LSD model was used because there was no significant difference across the three groups when other post hoc test models (such as post hoc Scheefe model) were used. There was a statistically significant difference between requirements of antipsychotic in carbamazepine group as compared to lithium group (P = 0.014). The table also showed a statistically significant difference in CSMS change from baseline to end of the study when multiple comparisons were made between carbamazepine and sodium valproate group (P = 0.059).

Table 10

Post hoc test showing requirement of mean antipsychotic (chlorpromazine equivalent) dose and Cassidy Scale for Manic States change from baseline to end of the study using multiple comparison across the three treatment groups

Dependent variable	Multiple comparison		Mean difference	P	95% CI
	Treatment group	Treatment group			Lower bound	Upper bound
Mean antipsychotic (CPZ equivalent)	Lithium	Carbamazepine	−87.50	0.014	−156.37	−18.62
		Sodium valproate	−75.41	0.042	−147.84	−2.99
	Carbamazepine	Lithium	87.50	0.014	18.62	156.37
		Sodium valproate	12.08	0.760	−66.61	90.78
	Sodium valproate	Lithium	75.41	0.042	2.99	147.84
		Carbamazepine	−12.08	0.760	−90.78	66.61
CSMS change from baseline to end of the study	Lithium	Carbamazepine	12.52	0.137	−4.11	29.16
		Sodium valproate	−5.80	0.510	−23.30	11.69
	Carbamazepine	Lithium	−12.52	0.137	−29.16	4.11
		Sodium valproate	−18.32	0.059	−37.34	0.68
	Sodium valproate	Lithium	5.80	0.510	−11.69	23.30
		Carbamazepine	18.32	0.059	−0.68	37.34

P<0.05. CI – Confidence interval; CPZ – Chlorpromazine; CSMS – Cassidy Scale for Manic States

It can be seen from Table 11 that the time required for getting therapeutic lithium level is 2nd week, 1st week in carbamazepine group and 3rd week in case of sodium valproate. The therapeutic serum lithium level was 0.8–1.2 mmols per liter. The therapeutic carbamazepine serum level was 8–12 μg per liter. The therapeutic sodium valproate serum level was 85–110 μg per liter.

Table 11

Comparison of serum level week wise across the three treatment groups

Serum level	Treatment group	Mean±SD
First week	Lithium	0.58±0.23
	Carbamazepine	9.53±4.34
	Sodium valproate	49.21±18.21
Second week	Lithium	0.84±0.26
	Carbamazepine	10.94±2.54
	Sodium valproate	59.22±21.70
Third week	Lithium	0.90±0.26
	Carbamazepine	10.69±3.00
	Sodium valproate	90.99±29.66
Fourth week	Lithium	0.88±0.23
	Carbamazepine	10.90±3.36
	Sodium valproate	101.22±17.93
Fifth week	Lithium	1.08±1.10
	Carbamazepine	10.60±2.29
	Sodium valproate	114.41±41.34

SD – Standard deviation

The effect of change of CSMS scores over the 6 week across the three treatment groups was found to be significant at 0.000 level. The effect of the change of CSMS scores was compared over the 6 weeks with lithium carbonate, carbamazepine, and sodium valproate groups using repeated-measures ANOVA. The results were not found to be statistically significant (P = 0.296). Partial Eta square value is an indirect evidence of the ES of the group [Table 12]. Table 13 shows the percentage of responders on the basis of CGI change score of 1 or 2 or a ≥33% reduction in baseline CSMS. Sodium valproate group had the highest responders on the basis of CGI change Score of 1 or 2 and lithium group had the highest responders on the basis of ≥33% reduction in baseline CSMS. When the mean of the two response variable was taken into consideration, sodium valproate group showed the highest response rate of 85%. Lithium had the highest ES of 0.85 followed by sodium valproate with an ES of 0.84 and carbamazepine having ES of 0.71. Thus, lithium and sodium valproate were said to have a large ES (>0.8%) whereas carbamazepine had medium ES (>0.7%) [Table 14].

Table 12

Multivariate tests showing comparison across the three treatment groups using Cassidy Scale for Manic States and Cassidy Scale for Manic States versus lithium carbonate, carbamazepine, and sodium valproate groups using repeated-measures ANOVA

Variables	Effect	Value	df	F	P	Partial η2
CSMS	Wilks’ Lambda	0.115	5.00	92.15	0.000	0.885
CSMS versus lithium carbonate, carbamazepine, and sodium valproate groups	Wilks’ Lambda	0.826	10.00	1.20	0.296	0.091

df – Degree of freedom; P<0.05. F – ANOVA value. CSMS – Cassidy Scale for Manic States

Table 13

Intent-to-treat sample: Percentage of responders in each treatment group by response variable

Drug	Response variables
	n/all (%)		Mean of CGI-BP and CSMS response (%)
	CGI-BP change score of 1 or 2	≥33% reduction in baseline CSMS
Lithium carbonate	21/30 (70)	27/30 (90)	80
Carbamazepine	13/20 (65)	14/20 (70)	67
Sodium valproate	14/17 (82)	15/17 (88)	85

CGI-BP – Clinical Global Impression-Bipolar Version; CSMS – Cassidy Scale for Manic States

Table 14

Comparison of the effect size across the three treatment groups

	Drug
	Lithium	Carbamazepine	Sodium valproate
Cohen’s D	3.30	2.03	3.22
Effect size	0.85	0.71	0.84

The most common side effect on lithium carbonate was tremor (38.76%) followed by thirst (28.90%), dry mouth (25.22%), and headache (24.18%). On carbamazepine, tremor (24.18%) was the most common side effect followed by drowsiness (22.74%) and constipation (15.24%). One patient was dropped due to development of rash. On sodium valproate, tremor (14.42%) was the most common side effect followed by diarrhea/constipation (13.78%) and headache (8.36%) [Table 15].

Table 15

Side effect profile of patients on lithium carbonate, carbamazepine, and sodium valproate

Lithium (n=30)	Percentage	Carbamazepine (n=20)	Percentage	Sodium valproate (n=17)	Percentage
Tremor	38.76	Tremor	24.18	Tremor	14.42
Headache	24.18	Headache	14.94	Headache	8.36
Anorexia	10.46	Decreased appetite	6.88	Decreased appetite/heart burn	7.76
Nausea	6.56	Nausea	5.98	Nausea	6.88
Vomiting	1.20	Vomiting	0.30	Vomiting	0.30
Diarrhea	1.80	Diarrhea	1.20	Diarrhea/constipation	13.78
Stomach pain	6.90	Stomach pains	4.78	Bloated abdomen	7.48
Dry mouth	25.22	Dry mouth	14.04
Urinary frequency	11.06	Drowsiness	22.74	Drowsiness	6.58
Thirst	28.90	Feeling dizzy	5.98	Dizziness	4.48
Blurry vision	0.30	Blurred vision	2.00	Tingling of hands/feet	0.30
Fatigue	24.48	Double vision	0.90	Muscle weakness	3.58
Metallic taste	3.00	Constipation	15.24	Weight gain	4.20
Feeling hot/cold	0.90	Unsteady gait	0.90	Unsteady gait	0.30
Weight gain	5.10	Muscle ache	0.90
Confusion	3.90	Tinnitus	0.30
Acne	2.40	Skin rash	1.20
Itchiness	0.30
Folliculitis	0.90
Seborrhea	0.30

DISCUSSION

The sample size of our study group consists of 67 patients (30 patients were allotted to lithium group, 20 patients to carbamazepine group, and 17 patients to the sodium valproate group). Though the sample size was modest, it is larger than previous studies done on mood stabilizers in child and adolescent population.[293031] It has been a truism in child psychiatry that “Comorbidities are a rule rather than exception.” In contrast to this statement, the sample size of this study consisted of only patients with BD I which is considered to be statistically sound as compared to previous studies, which included various comorbidities such as ADHD and substance abuse.[3233]

The three groups in this study, i.e. lithium (n = 30), carbamazepine (n = 20), and sodium valproate (n = 17) included more patients, than previous studies comparing these three mood stabilizers.[29] Previous studies had not assessed patients on a weekly basis, which is an improvement in methodology. Assessment of patients on a weekly basis could identify minor change in the psychopathology throughout the treatment trial as fluctuations in the psychopathology are very common in the child and adolescent psychiatry. Regarding dose and serum level, dose was adjusted only after getting the serum level on a weekly basis, which gives almost timely attainment of an adequate serum level. This is an improvement over a previous study which assessed serum level at only 2, 4, and 6 weeks.[29] No other study till now has titrated the dose on the basis of serum level on a weekly basis.

Comparing the patients at the time of entry and the patients at completion of the study shows that this study has less dropouts than that of previous studies. In this study, out of 67 patients, 37 (55.22%) completed the 6-week study protocol as compared to three (7.14%) out of 42 patients completed an 8-week study.[29]

In this study, the tool used for diagnosis was K-SADS-PL version, which is at par with earlier studies, and it is considered a good diagnostic tool as it assesses the current condition as well the lifetime diagnosis.[29]

For assessing psychopathology, CSMS was used which could assess mixed features which is typical of pediatric mania. Most of the previous studies had used Young Mania Rating Scale (YMRS), which assesses more of euphoric mania symptoms.[29303134] In this study, UKU side effect checklist was used along with the side effect checklist of lithium, carbamazepine, and sodium valproate. Due to this, side effects related to antipsychotic and mood stabilizers could be assessed separately, which strengthens the study. Most of the earlier studies done on this area used only the side effect checklist of mood stabilizers.[29] Regarding concurrent medications used along with mood stabilizer trial, we used only typical antipsychotic (CPZ or haloperidol). Atypical antipsychotics were not used as it is said to have mood stabilizing properties.[35] Earlier studies used stimulants, clonazepam, and clonidine, etc., along with the mood stabilizer trial.[31]

Demographic variables

The sample of this study was predominantly male (64.20%) compared to females (35.80%).

The mean age of the patients in our study was 15.33 ± 1.72 years in lithium group, 15.40 ± 1.50 years in the carbamazepine group, and 15.70 ± 1.75 years in the sodium valproate group. In an earlier study,[29] the mean age was 11.40 years across the lithium, carbamazepine, and the sodium valproate groups. In two other studies,[3031] the mean age was 12.1 ± 3.62 years and 12.3 ± 3.7 years respectively. The mean age in our study is more than the mean age of patients in earlier studies.. The mean age in our study is more than that of earlier studies. This disparity can be due to the characteristic of the Indian society. In our society, females and very young children only infrequently opt for inpatient management due to cultural reasons and fear of being stigmatized. Majority of the study sample were Hindu (80.60%) compared to Muslims (19.40%). This finding could be due to the predominant Hindu population in the catchment area of this tertiary care hospital.

Comparison of the mood stabilizers

In this study, the mean duration of the index episode was 48.93 ± 54.45 days in the lithium group, 34.90 ± 36.19 days in the carbamazepine group, and 48.11 ± 70.90 days in the sodium valproate group. This result shows the typical pattern of childhood BD, which is known for its chronicity and long episodes.[36] Chi-square test was done to show the association of clinical variables across the lithium, carbamazepine, and sodium valproate groups. The result showed a significant association of use of lithium with the presence of a family history of BD. There is also a significant association of use of sodium valproate with the number of past depressive episodes and total number of episodes. This result supports earlier finding of the research done on predictors of mood stabilizer response.[37] This can be of important bearing in the selection of mood stabilizer for a particular case. It can be seen from the result that there is an abrupt fall of an average of 20.52 ± 9.19 points in the lithium group, 20.20 ± 7.11 points in the carbamazepine group, and 18.35 ± 6.82 points in the sodium valproate group in the CSMS at the 1st week. This finding can be explained by the use of high dose of antipsychotic in initial few days to control aggression. This finding is in agreement with the results of an earlier study, which showed similar reduction on CSMS at the end of the 1st week.[29]

The result of this study showed that in some patients in lithium and carbamazepine groups, there is a slight worsening at the 4th week, which is in contrast with an earlier study in which there was worsening in some patients of divalproex sodium group at the 3rd week.[29] This phenomenon is not reported in mood stabilizer trial of adult population. This finding suggests a transitory difference between adult and children in their neurochemical response to mood stabilizer. This finding also suggests that clinicians should not discontinue treatment with lithium, carbamazepine, or sodium valproate if there is worsening at 4–6 weeks of treatment, but should continue treatment for 1 or 2 weeks before deciding to discontinue the treatment.

The result also shows the time required across the treatment group for getting therapeutic level of each mood stabilizer. The time required for therapeutic level is in the 2nd week for the lithium group, 1st week in carbamazepine group, and 3rd week in case of sodium valproate. The “therapeutic serum level” is defined as 0.8–1.2 mmols per liter for lithium, 8–12 μg per liter for carbamazepine, and 85–110 μg per liter for sodium valproate. This result is comparable with that of an earlier study.[29] The reason for attainment of adequate serum level in the 3rd week in case of the sodium valproate group is the use of a very narrow range (85–110 μg per liter) as defining therapeutic range in our study.

The results of our study show a gradual decline in the UKU-side effect rating scale, but the requirement of trihexyphenidyl remains almost same across the three treatment groups throughout the study period. This can be explained by high rates of dystonia in the three groups. The occurrence of dystonia requires the need of prophylactic trihexyphenidyl in the risk-prone group of child population.[38]

The result of post hoc test using LSD model of requirement of mean antipsychotic (CPZ equivalent) dose and CSMS score for manic state change from baseline to end of the study using multiple comparisons across the three treatment group revealed a statistically significant difference between the requirements of antipsychotic in carbamazepine group as compared to lithium group (P = 0.014). LSD model was used because there was no significant difference across the three groups when other post hoc test models (such as post hoc Scheefe model) were used.

The results also show a statistically significant difference in CSMS scores from baseline to end of the study when multiple comparisons were made between carbamazepine and sodium valproate groups (P = 0.059). This result shows that the requirement of antipsychotic is same at the baseline for all the three treatment groups, but the requirement of antipsychotic was greater in the carbamazepine group across the 6 weeks.

In this study, we also calculated the response rate of lithium, carbamazepine, and sodium valproate on the basis of ≥33% reduction from baseline and CSMS and CGI-BP improvement Score of l or 2. We got a response rate 90% in lithium group, 70% in carbamazepine group, and 88% in sodium valproate group on the basis of ≥−33% reduction from baseline CSMS. The result also shows a response of 70% in lithium group, 65% in carbamazepine group, and 82% in sodium valproate group on the basis of CGI-BP improvement Score of l or 2. Previous study by Kowatch et al. showed a response rate of 46%–50% for divalproex sodium, 42%–45% for lithium, and 34%–44% for carbamazepine on the basis of ≥ 50% reduction from baseline YMRS and CGI-BP improvement Score of 1 or 2.[29] The discrepancy regarding response rate can be explained by the fact that our study used a cutoff of 33% reduction rather than 50%, along with use of antipsychotic in the initial week to control aggression.

The result of multivariate tests using repeated-measures ANOVA shows the comparison across the three treatment groups using CSMS and CSMS versus lithium carbonate, carbamazepine, and sodium valproate. First, the effect of change of CSMS over the 6 weeks across the three treatment group was determined, which was found to be highly statistically significant (at 0.000 level). Then, the effect of the change of CSMS was compared with lithium carbonate, carbamazepine, and sodium valproate. All the three groups were compared using repeated-measures ANOVA. The results were not found to be statistically significant. This result can be better explained on the basis of the small and uneven sample in the three groups (30 in lithium, 20 in carbamazepine, and 17 in sodium valproate groups). For getting a large difference between these three mood stabilizers, we may need a large sample at least 10–20 times more than our sample. Such study is not possible because of high cost and the need of large number of pediatric bipolar patients in a short period of time.

Finally, our results show an ES of 0.85 for lithium, 0.71 for carbamazepine, and 0.84 for sodium valproate. The ES in our study is less than tha of the earlier study by Kowatch et al., which showed an ES of 1.06 for lithium, 1.00 for carbamazepine, and 1.63 for divalproex sodium.[29] Again, this discrepancy could be due to the small and uneven sample in our three groups. The use of high-dose antipsychotic further adds to these ES results.

As regards the tolerability of these three mood stabilizers in pediatric mania, our results show a side effect profile of the patients on lithium carbonate, in which tremor (38.76%) was the most common side effect followed by thirst (28.90%), dry mouth (25.22), and headache (24.18%).

The side effect profile of the patients on carbamazepine showed tremor (24.18%) as the most common side effect followed by drowsiness (22.74%) and constipation (15.24%). Only one out of twenty patients developed rashes in the 2nd week for which carbamazepine was stopped. The side effect profile of the patients on sodium valproate was tremor (14.42%) followed by diarrhea/constipation (13.78%) and headache (8.36%). Our results show the tolerability of these agents in treating BD in children, which supports earlier studies done in this group.[293031] To summarize, the finding of this study is comparable to that of the earlier studies. It can be said that the cumulative time for response data suggests that adequate trial in childhood BD should be at least 6–8 weeks, but this finding can only be generalized after studying a large group of pediatric population for a long duration of time.

Limitations

The sample size in this study was modest and their distribution across the treatment groups was uneven. Antipsychotics were used to control aggression and the follow-up period was limited.

CONCLUSIONS

Adolescent Bipolar 1 disorder patients during Manic phase showed an ES of 0.85 for lithium, 0.71 for carbamazepine, and 0.84 for sodium valproate. These agents are well tolerated in treating BD in children.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.