Use of Lithium, Carbamazepine,

and Valproic Acid in a

State-Operated Psychiatric Hospital
LESLIE CJTROME

Objective: To examine the prescribing of mood stabilizers (lithium, carbamazepine, and valproic acid) in a
SOO-bed state-operated psychiatric hospital in New York.
Method: All 129 inpatients receiving mood stabilizers were identified and their medical records reviewed
using a standardized drug use evaluation form. Diagnosis, other indications, and prior experience with
mood stabilizers were examined, as well as outcome and adverse effects.
Results: Approximately one-quarter of the inpatient population received a mood stabilizer. The frequency of
carbamazepine use exceeded the use of lithium, with 72 patients receiving carbamazepine and only 62
receiving lithium. Twenty-eight patients received valproic acid. Indications found most frequently for
carbamazepine use included assaultive or aggressive behavior (70% for those receiving carbamazepine as the
only mood stabilizer). Of those patients with bipolar or schizoaffective disorder and receiving either lithium,
carbamazepine, or valproic acid, 36% were prescribed carbamazepine (10%as a first-line agent) and SO%
lithium (26% as a first-line agent). None of these indications for carbamazepine has been approved by the
Food and Drug Administration. In general, positive outcomes were documented but without supporting
objective measures. Significant adverse effects were documented in the medical record in one-quarter of the
patients.
Conclusions: There was widespread use of the three mood stabilizers examined, singly and in combination,
for a variety of indications. Lithium and valproic acid remain more frequently prescribed for the treatment of
bipolar and schizoaffective disorders. Monotherapy with carbamazepine or valproic acid results in
statistically significantly fewer adverse effects than lithium or combination therapy (p values between
p = 0.00038and p = 0.006).Current clinical practice has endorsed the use of carbamazepine for aggressive or
assaultive behavior, although there does not appear to be sufficient proof of effectiveness in the literature.
Formal studies of carbamazepine's antiagressive effects should be conducted.
J Pharm Tecllllol199S;11:SS-9.

The classic drug of choice for the treatment of mania has
been lithium.' In recent years two anticonvulsants, carba-
mazepine and valproic acid, have emerged as useful
treatment options.P Moreover the use of these three med-
ications has been expanded to cover a wide spectrum of
mental disorders. These mood stabilizers have been used
successfully in augmentation strategies with antidepres-
sants' and neuroleptics," Control of impulsive and ag-
gressive behavior also can be targeted as a treatment goal
for this group of agentsP
Because treatment with these medications requires
baseline laboratory measures as well as ongoing titration
and monitoring of blood concentrations, they are often a
target of a drug use evaluation (DUE), required as part of
a healthcare organization's quality assessment and im-
provement plan," Aspects explored in a DUE include ap-
propriateness of indication, management of adverse ef-
feets, and treatment outcome. We describe a DUE explor-
ing the prescribing of mood stabilizers in a Sao-bed state-
operated psychiatric hospital, Middletown Psychiatric
Center in Middletown, NY, whose primary mission is to
provide intermediate and long-term care to the seriously
mentally ill.
Methods
The pharmacy department identified all patients re-
ceiving lithium, carbamazepine, or valproic acid, either
alone or in combination, based on pharmacy computer
records in March 1993. One hundred twenty-nine pa-
tients were identified and their medical records reviewed
by one of three psychiatrists not directly involved in their
care.

LESLIE CITROME, MD, FRCP(C), Research Psychiatrist, Nathan S. Kline Institute for Psychiatric Research,Orangeburg, NY, and Clin-
ical Assistant Professorof Psychiatry, New York University, NY. Reprints: LeslieCitrome, MD, FRCP(C), Nathan S. KlineInstitute for Psy-
chiatric Research,Orangeburg, NY 10962, FAX 914/359-7029
This study was supported in part by a grant from Abbott Laboratories.

JOURNAL OF PHARMACY TECHNOlOGY VOLUME 11 MARCH!APRIL 1995 55

 We abstracted data regarding the following four pa-
rameters: (1) indications for use (treatment/prophylaxis
for bipolar or schizoaffective disorders, impulse control
problem occurring within the context of any disorder, as-
saultive/aggressive behavior occurring within the con-
text of any disorder, adjunctive use for schizophrenia, ad-
junctive use for major depression); (2) reason for selection
(first-line, i.e., has not received any other mood stabilizer,
not tolerating other mood stabilizers, combination thera-
py with other mood stabilizers, treatment resistance to
other psychotropic medication); (3) outcome (improved,
unchanged, worse, or not documented); and (4) adverse
effects (none, present, or not documented).
Dosing, blood concentration monitoring, and the
monitoring of serum liver enzymes, electrolytes, or he-
matologic measures were not addressed in this DUE.
These issues have important risk management and cost
considerations and have been the subjects of other DUEs
completed at this institution.
The rate of adverse effects for the different regimens
was analyzed statistically with the use of multiple conti-
nuity-adjusted chi-square tests or two-tailed Fisher exact
tests. All results were interpreted using the Holm correc-
tion for multiplicity.
Results
Approximately one-quarter of our inpatient popula-
tion was receiving a mood stabilizer. Twenty-eight per-
cent of all patients were receiving lithium as the only
mood stabilizer, 39% only carbamazepine, and 9% only
valproic acid. Twenty-four percent received a combina-
tion of two or three mood stabilizers. All patients for
whom valproic acid was prescribed received the enteric-
coated divalproex preparation. Table 1 summarizes dem-
ographic and diagnostic information from the 129 rec-
ords reviewed,"
The most frequent diagnosis recorded for patients re-
ceiving a mood-stabilizing drug was schizoaffective dis-
order (44%). The most frequent diagnosis for patients re-
ceiving carbamazepine alone was schizophrenia.
Table 2 summarizes the symptoms/indications for
treatment, other than diagnosis, as documented in the
medical record. Problems with impulse control or as-
saultive/aggressive behavior were noted for almost half
of the patients. Other indications included seizure disor-
der (5 patients receiving carbamazepine alone and 2 re-
ceiving carbamazepine in combination with valproic acid
had a seizure disorder as well as a psychiatric disorder).

Table 1. Demographics and DSM-IIIR Diagnosis

DSM·IIIR DIAGNOSIS (%)
AVERAGE
AGE BIPOLAR SCHIZOAFFECTIVE
GROUP n (y) M:FRATIO DISORDER DISORDER SCHIZOPHRENIA OTHER
Lithium 36 40 25:11 6 (17) 22 (61) 7 (19) 1 (3)
Carbamazepine 50 49 28:22 8 (16) 12 (24) 23 (46) 7 (14)
Valproic acid 12 43 7:5 4 (33) 4 (33) 2 (17) 2 (17)
Lithium + carbamazepine 15 45 5:10 4 (27) 9 (60) 2 (13) 0
Lithium + valproic acid 9 54 5:4 3 (33) 6 (66) 0 0
Carbamazepine + valproic acid 5 44 3:2 0 3 (60) 1 (20) 1 (20)
Lithium + carbamazepine + 2 45 2:0 0 1 (50) 1 (50) 0
valproic acid
TOTAL 129 46 75:54 25 (19) 57 (44) 36 (28) 11 (9)

DSM·IIIR = Diagnostic and statistical manualof mentaldisorders. 3rd ed. Revised?

Table 2. Indications for Treatment, Other Than Diagnosis, as Documented in the Medical Record
INDICATION (%)
ASSAULTIVE DID NOT TOLERATE
IMPULSE OR AGGRESSIVE OR RESPOND TO
GROUP n CONTROL BEHAVIOR OTHER FIRST LINE" OTHER MOOD STABILIZER
Lithium 36 2 (6) 6 (17) 2 (6) 29 (81) 5 (14)
Carbamazepine 50 25 (50) 35 (70) 6 (12) 33 (66) 13 (26)
Valproic acid 12 6 (50) 4 (33) 0 0 10 (83)
Lithium + carbamazepine 15 13 (87) 7 (47) 0 0 5 (33)
Lithium + valproic acid 9 8 (89) 2 (22) 0 0 4 (44)
Carbamazepine + valproic acid 5 3 (60) 2 (40) 0 0 2 (40)
Lithium + carbamazepine + 2 2 (100) 1 (50) 2 (100) 0 1 (50)
valproic acid
TOTAL 129 59 (46) 57 (44) 10 (8) 62 (48) 40 (31)

"No other mood stabilizer ever used.

56 JOURNAL OF PHARMACY TECHNOlOGY VOlUME 11 MARCH/ApRIL 1995

 LITHIUM, CARUAMAZEPINE, AND VAlPROIC ACID

The greatest number of patients with documented as- Table 3 summarizes treatment outcome, as document-
saultive/aggressive behavior was found in the carba- ed in the physicians' progress notes. Improvement was
mazepine monotherapy subgroup (70%). Documented noted in 87% of patients, although there were many charts
nonresponse or intolerance to adverse effects of other that documented only slight improvement. No rating
mood stabilizers was seen in 31% of all patients. Forty- scales were used.
eight percent of all patients previously had not received a Table 4 summarizes adverse effects, as documented in
mood stabilizer. The agent of first choice as a first-line the progress notes. Thirty-four patients (26%) experi-
drug for all diagnoses was split almost evenly between enced significant adverse effects. Fifteen of 36 (42%) re-
lithium (29 patients) and carbamazepine (33 patients). ceiving lithium experienced an adverse effect, primarily
More patients received carbamazepine alone (50 pa- hypothyroidism, but also tremor, gastrointestinal distur-
tients) than lithium alone (36 patients). For patients with bance, and polydipsia. Only 6 of 50 patients (12%)receiv-
bipolar or schizoaffective disorder as defined by Diagnos- ing carbamazepine experienced adverse effects. Two of
ticandStatistical Manual ofMental Disorders, 3rdedition, these were hematologic disturbances requiring discontin-
Revised criteria.t lithium was favored over carbamaze- uation or dosage reduction. None of the patients receiv-
pine (50% vs. 36%, respectively), with lithium more com- ing valproic acid alone had any documented adverse ef-
monly used as a first-line agent than carbamazepine (26% fects, although the total number of patients in this cate-
vs. 10%, respectively). gory was low (12). Thirteen of the 31 patients (42%) re-
Of the 28 patients receiving valproic acid, 21 (75%) ceiving combination therapy experienced documented
had bipolar or schizoaffective disorder. Valproic acid was adverse effects, compared with 21 of 98 patients (21%) re-
used more commonly in combination therapy (12% of ceiving monotherapy.
the entire sample) compared with monotherapy (9%).
Only one case was found in which a mood stabilizer
was used as an adjunct to antidepressant therapy for the
treatment of major depression (a 57-year-old woman re- Table 4. Adverse Effects As Documented in
ceived lithium and phenelzine). A single case was found Physicians' Progress Notes
in which the mood stabilizer was used alone without a ADVERSE EFFECTS ('Yo)
neuroleptic, antidepressant, or benzodiazepine other NOT DOCUMENTED
GROUP n NONE PRESENT
than on an as needed basis (a 60-year-old man with de-
lithium 36 18 (50) 15 (42)' 3 (8)
mentia associated with alcoholism received carbamaze- 43 (86) 6 (12) 1 (2)
Carbamazepine 50
pine alone). Valproic acid 12 12 (100) 0 0
Patients identified as being mentally ill chemical abusers lithium + 15 9 (60) 6 (40)b 0
carbamazepine
comprised 12% of the total study population. Approxi- lithium + 9 4 (44) 5 (56) 0
mately equal numbers of these patients received lithium valproic acid
(9 patients) or carbamazepine (10 patients). Only 2 pa- Carbamazepine + 5 3 (60) 2 (40) 0
valproic acid
tients received valproic acid, both in combination with lithium + 2 2 (100) 0 0
other mood stabilizers. Seven patients had both a mental carbamazepine +
illness and mental retardation. All received carba- valproic acid
mazepine and two received valproic acid in combination TOTAL 129 91 (71) 34 (26) 4 (3)
with carbamazepine. None of these patients received 'Eleven of these 15 patients had documented hypothyroidism.
lithium. bThree of these 6 patients had documented hypothyroidism.

Table 3. Outcome As Documented in Physicians' Progress Notes

OUTCOME ('Yo)
GROUP n IMPROVED UNCHANGED WORSE NOT DOCUMENTED

lithium 36 33 (92) 2 (6) 0 1 (3)

Carbamazepine 50 43 (86) 6 (12) 0 1 (2)
Valproic acid 12 8 (67) 4 (33) 0 0
lithium + carbamazepine 15 14 (93) 1 (7) 0 0
lithium + valproic acid 9 8 (89) 0 1 (11) 0
Carbamazepine + valproic acid 5 4 (80) 1 (20) 0 0
lithium + carbamazepine + 2 2 (100) 0 0 0
valproic acid
TOTAL 129 112 (87) 14 (11) (1) 2 (2)

JOURNAL OF PHARMACY TECHNOLOGY VOLUME 11 MARcHIAPRIL 1995 57

Discussion
The use of carbamazepine as a mood stabilizer has been
widely accepted at this state-operated psychiatric hospi-
tal. It has surpassed lithium as the most popular agent of
its class. Valproic acid is seen by the practitioners as a sec-
ond-line drug and is used less frequently. This may be
because it has only recently become available on the hos-
pital formulary in the enteric-coated divalproex form.
Valproic acid is becoming more popular as a standard
treatment for mania, and support for this medication can
now be found in the literature," At the time this study
was completed, the Food and Drug Administration had
not yet approved the drug for this new indication.
Outcome, as documented in the
physicians' progress notes,
appears positive, but in many
cases only minimal
improvement was noted.
were grouped together, as the number of patients was
too low to permit comparisons of various drug combina-
tions within the group. Monotherapy with carbamaze-
pine or valproic acid appears to be statistically signifi-
cantly better tolerated when compared with lithium or
combination therapy (adverse effect rates of 12% and 0%
compared with 42% and 42%, respectively; see Table 5
for p values). There was no statistically significant differ-
ence in adverse effect rate between lithium monotherapy
and combination therapy, nor between monotherapy
with carbamazepine or valproic acid. The difference ob-
served in adverse effect rate between monotherapy and
combination therapy was marginal using the conserva-
tive continuity adjusted chi-square test (chi-square = 3.59;
df = 1; p =0.058). This can be explained by the inclusion
of patients receiving lithium monotherapy in the mono-
therapy group, even though we have noted that lithium
causes a relatively high rate of adverse effects.
The extent of the popularity of carbamazepine was
unexpected, given the lack of Food and Drug Adminis-
tration approval for its use in mania or for the control of
aggressive and impulsive behavior. Its effectiveness for
antiaggressive effects has not been well studied-': the
studies published to date have not been blind and have
not controlled for placebo effect. The plasma concentra-
tions of concomitant neuroleptic medications were not
reported in those studies, but may have been affected by
the carbamazepine." The total number of patients in all
of these studies combined is less than 100,n making fur-
ther research in this area desirable. ~

Outcome, as documented in the physicians' progress

notes, appears positive, but in many cases only minimal
improvement was noted. Rather than relying on impres- Table5. Statistical Analysis of the Rate of
sions gleaned from the progress notes, the use of a rating Adverse Effects
scale such as the Brief Psychiatric Rating Scale would be
CONTINUITY·ADJUSTED TWO·TAILED
a more valuable measure when assessing outcome. In CHI-SQUARE (df = 1) FISHER
addition, a prospective approach, with the inclusion of EXACT TEST
COMPARISON STATISTIC PROBABILITY PROBABILITY"
blood concentration data to aid analysis, would provide
a better means of accurately assessing therapeutic out- lithium vs. 9.7 p = 0.002 b NA
carbamazepine
comes. The higher rate of adverse effects documented in lithium vs. NA NA P = 0.00038 b
combination therapy (42%)compared with monotherapy valproic acid
(21%), along with the higher material and laboratory lithium vs. 0.001 p = 0.975< NA
combination therapy
costs, emphasizes the need to demonstrate improved Carbamazepine vs, NA NA P = 0.588<
outcome to justify usage. valproic acid
The number of adverse effects documented in the pa- Carbamazepine vs. 7.68 p = O.006b NA
combination therapy
tients in this study may have been underreported; never- Valproic acid vs. NA NA P = 0.00081 b
theless, the number of adverse effects with lithium is no- combination therapy
table. Twenty-three percent of all patients receiving Monotherapy vs. 3.59 p = 0.058 d NA
combination therapy
lithium, alone or in combination with other mood stabi-
lizers, were found to have hypothyroidism on routine NA = not applicable.
'The Fisher exact test was used when one of the cells in the 2 x 2 chi-
laboratory testing. This is somewhat higher than the square was less than 5.
5-15% range reported in the literature." bStatislically significant even after applying the Holm correction for
Table 5 summarizes the statistical analysis comparing multiplicity.
<Clearlystatistically not significant.
the rate of adverse effects for the different treatment regi- dMarginally significant. Holm correction cut-off for significance is
mens. Patients receiving any combination of medications p=0.0167.

58 JOURNAL OF PHARMACY TECHNOLOGY VOLUME 11 MARCH/ApRIL 1995

 LITHIUM, CARBAMAZEPINE, AND VALPROIC ACID

The author acknowledges Ba Han, MD, and Quazi AI-Tariq, MD, Mid- 5. Wassef A, Watson DJ, Morrison P, Bryant S, Flack J. Neuroleptic-
dletown Psychiatric Center, Middletown, NY, for assisting in the data col- valproic acid combination in treatment of psychotic symptoms:
lection.The author also thanks Pal Czobor,PhD, ResearchScientist,Nathan a three-case report. J Clin PsychopharmacoI1989;9:45-8.
Kline Institute, for statistical analysis and Jan Volavka,MD, PhD, Director 6. Sheard MH, Marini JL, Bridges CI, Wagner E. The effect of lithi-
of Clinical Research, Nathan Kline Institute, for critical review of the um on impulsive aggressive behavior in man. Am J Psychiatry
manuscript. 1976;133:1409-13.
7. Luchins DL. Carbamazepine in violent non-epileptic schizophren-
ics. Psychopharmacol Bull 1984;20:569-71.
8. Joint Commission on Accreditation of Healthcare Organizations.
References Accreditation manual for hospitals. Vol 1. Standards. MS.5.1.3.
Drug usage evaluation. Chicago: Joint Commission on Accredi-
1. Chou JCy. Recent advances in treatment of acute mania. J Clin Psy- tation of Healthcare Organizations, 1994.
chopharmacoI1991;11:3-21.
9. Diagnostic and statistical manual of mental disorders. 3rd ed. Re-
2. Gerner RH, Stanton A. Algorithm for patient management of acute vised. Washington, DC: American Psychiatric Association, 1987.
manic states: lithium, valproate or carbamazepine? J Clin Psy-
10. Vestergaard P.Clinically important side effects of long term lithi-
chopharmacoI1992;12(suppl):575-63S.
um treatment: a review. Acta Psychiatr Scand 1983;67(suppl):11-36.
3. Bowden CL, Brugger AM, Swann AC, Calabrese JR, [anicak PG,
11. Volavka J. Pharmacological treatment of violence. In: Neurobiol-
Petty F,et a\. Efficacy of divalproex vs lithium and placebo in the
ogy of violence. Washington, DC: American Psychiatric Press, 1995.
treatment of mania. JAMA 1994;271:918-24.
12. Levy RH, Kerr BM. Clinical pharmacokinetics of carbamazcpinc.
4. Heninger GR, Charney DS, Sternberg DE. Lithium carbonate aug-
J Clin Psychiatry 1988;49(suppl):58-61.
mentation of antidepressant treatment. Arch Gen Psychiatry 1983;
40:1335-42.

Historical Quote
Cost and Contamination Study with Technicians. Hospital pharmacists who are planning to develop admixture
services are faced with the problem of evaluating and selecting the most appropriate system. Selection may be
based on which of the systems offers maximum safety, economy and sterility. There have been no comparative
studies of these factors on the available admixture systems....The purpose of this study was to make a comparative
evaluation of intravenous admixture systems. More specifically, the objectives were: 1. To determine which system
is the most economic for use in a parenteral admixture program. 2. To determine which system will assure mini-
mal contamination assuming that the manufacturers' recommended aseptic techniques and procedures are ob-
served in the compounding process. 3. To determine the most suitable techniques for transferring additives from
ampuls or vials aseptically into each admixture system....Three technicians were employed to prepare the various
admixtures. They had no previous experience with intravenous solutions or parenteral admixtures. They were si-
multaneously trained to prepare admixtures with all of the systems. Data were not collected until the technicians
were proficient at preparing parenteral admixtures. Learning curves were developed during the initial phases of the
study in order to determine when technicians' performance became reasonably stable....The compounding or
preparation cost consisted of the materials used (exclusive of the cost of the basic solution and admixed drugs) and
personnel costs required to prepare an admixture. The personnel costs were calculated based on average total com-
pounding time. The standard time data were used to calculate compounding time and cost for each drug admix-
ture using the different systems. Every admixture was prepared by each individual technician approximately 50
times....The results of the cost analysis studies for the seven types of parenteral additives indicate that the open ad-
mixture system is the most economical system. The open system had the lowest compounding cost for all seven
drug additives except a drug additive which represents the entire contents of an ampul in which the plastic bag sys-
tem was the more economical. Recently major modifications were made with the investigational plastic bag used
during the study. It is anticipated that these modifications will greatly lower the cost data for the use of a plastic bag
system in a centralized admixture program. Compounding cost was influenced more by the cost of materials than
personnel. ... The results of the sterility studies have indicated that contamination most likely results from poor asep-
tic technique by the technician. Further sterility studies should be conducted to determine the specific causes of
contamination. Once the causes of contamination are known the compounding procedures and techniques can
be altered to assure minimal contamination rates with each of the admixture systems. Miller WA, Smith GL, Latio-
lais C]. A comparative evaluation of compounding costs and contamination rates of intravenous admixture systems.
Drug Intell Clin Pharm 1971 ;5:51-5,60.

JOURNAL OF PHARMACY TECHNOLOGY VOLUME 11 MARCH/APRIL 1995 59