You are on page 1of 7

Journal of Critical Care 37 (2017) 119–125

Contents lists available at ScienceDirect

Journal of Critical Care


journal homepage: www.jccjournal.org

Valproate for agitation in critically ill patients: A retrospective study


David J. Gagnon, PharmD, BCCCP a,⁎, Gabriel V. Fontaine, PharmD, BCPS b, Kathryn E. Smith, PharmD, BCPS a,
Richard R. Riker, MD c,d, Russell R. Miller III, MD, MPH e, Patricia A. Lerwick, MD d, F.L. Lucas, PhD f,
John T. Dziodzio, BA d, Kristen C. Sihler, MS, MD g, Gilles L. Fraser, PharmD a,d
a
Department of Pharmacy, Maine Medical Center, 22 Bramhall St, Portland, ME 04102
b
Department of Pharmacy and Neurosciences Institute, Intermountain Medical Center, 5121 South Cottonwood St, Murray, UT 84107
c
Neuroscience Institute, Maine Medical Center, 22 Bramhall St, Portland, ME 04102
d
Department of Critical Care Medicine, Maine Medical Center, 22 Bramhall St, Portland, ME 04102
e
Department of Critical Care Medicine, Intermountain Medical Center, 5121 South Cottonwood St, Murray, UT 84107
f
Center for Outcomes Research & Evaluation, 509 Forest Ave, Suite 200, Portland, ME 04101
g
Department of Surgical/Trauma Critical Care, Maine Medical Center, 22 Bramhall St, Portland, ME 04102

a r t i c l e i n f o a b s t r a c t

Keywords: Purpose: The purpose was to describe the use of valproate therapy for agitation in critically ill patients, examine
Agitation its safety, and describe its relationship with agitation and delirium.
Delirium Materials and methods: This retrospective cohort study evaluated critically ill adults treated with valproate for
Valproate agitation from December 2012 through February 2015. Information on valproate prescribing practices and safety
Valproic acid
was collected. Incidence of agitation, delirium, and concomitant psychoactive medication use was compared between
Sedation
Critical care
valproate day 1 and valproate day 3. Concomitant psychoactive medication use was analyzed using mixed models.
Results: Fifty-three patients were evaluated. The median day of valproate therapy initiation was ICU day 7, and it
was continued for a median of 7 days. The median maintenance dose was 1500 mg/d (23 mg/kg/d). The incidence
of agitation (96% vs 61%, P b .0001) and delirium (68% vs 49%, P = .012) significantly decreased by valproate day 3.
Treatment with opioids (77% vs 65%, P = .02) and dexmedetomidine (47% vs 24%, P = .004) also decreased.
In mixed models analyses, valproate therapy was associated with reduced fentanyl equivalents (−185 μg/d,
P = .0003) and lorazepam equivalents (−2.1 mg/d, P = .0004). Hyperammonemia (19%) and thrombocytopenia
(13%) were the most commonly observed adverse effects.
Conclusions: Valproate therapy was associated with a reduction in agitation, delirium, and concomitant psycho-
active medication use within 48 hours of initiation.
© 2016 Elsevier Inc. All rights reserved.

1. Introduction for safe administration [5]. New therapies for treating agitation are
rarely introduced into practice, with dexmedetomidine being the most
Agitation occurs in up to 70% of critically ill patients and is a recent in 1999. Consequently, providers have increasingly repurposed
significant source of distress for patients, families, and providers [1]. older pharmacologic agents as ICU sedatives (eg, clonidine, phenobarbital,
Sedatives are administered to N50% of intensive care unit (ICU) patients and valproate) [6,7].
to alleviate agitation [2]. Choice of sedative is complex and largely Valproate is an antiepileptic and mood stabilizer approved for treat-
driven by patient context. No sedative has consistently been shown to ment of seizures, manic episodes associated with bipolar disorder, and
be superior to the rest, and alternative agents are greatly needed [3]. migraine prophylaxis [8]. Mechanistically, it blocks voltage-dependent
Most ICU patients, especially those requiring mechanical ventilation, sodium and calcium channels, increases γ-aminobutyric acid (GABA)
are treated with opioids, propofol, and/or benzodiazepines [2,4]. Use of synthesis, potentiates GABA activity at postsynaptic receptors,
these agents is limited by adverse effects (eg, hemodynamic derange- blocks GABA degradation, and attenuates the activity of glutamate
ment or respiratory depression) and need for a monitored environment upon N-methyl-D-aspartate receptors [9,10]. Recently, valproate has
been administered to critically ill patients to treat agitation and
⁎ Corresponding author at: Department of Pharmacy, Maine Medical Center, 22 delirium, but there are few published reports to support this practice
Bramhall St, Portland, ME 04102. Tel.: +1 207 662 2151; fax: +1 207 662 6273. [11-14]. Valproate is an emerging treatment for ICU agitation because
E-mail addresses: dgagnon@mmc.org (D.J. Gagnon), fontaine.gabe@gmail.com it allows patients to interact with their caregivers; can be administered
(G.V. Fontaine), smithk17@mmc.org (K.E. Smith), rriker@cmamaine.com (R.R. Riker),
russ.millerIII@imail.org (R.R. Miller), lerwip@mmc.org (P.A. Lerwick), flucas@mmc.org
outside of the ICU; has both an intravenous (IV) and enteral formulation;
(F.L. Lucas), dziodj@mmc.org (J.T. Dziodzio), sihlek@mmc.org (K.C. Sihler), has a low drug acquisition cost; and has not been associated with
fraseg@mmc.org (G.L. Fraser). respiratory depression, hemodynamic derangements, or delirium.

http://dx.doi.org/10.1016/j.jcrc.2016.09.006
0883-9441/© 2016 Elsevier Inc. All rights reserved.
120 D.J. Gagnon et al. / Journal of Critical Care 37 (2017) 119–125

In this study, we sought to (1) describe the use of valproate for were unavailable, patients were assigned an agitation day if key words
agitation in critically ill patients, (2) examine the safety of valproate signifying agitation were identified in caregiver notes [1].
therapy, and (3) describe the relationship between valproate therapy The daily proportion of patients with delirium was also evaluated.
and agitation and delirium. The Confusion Assessment Method for the ICU (CAM-ICU) results
were assessed at both institutions. Because the recording of CAM-ICU
2. Material and methods results was infrequent, a validated chart review method was used in
conjunction (Electronic Supplement Fig. 2) [16]. The sensitivity and
2.1. Study design specificity of the chart review method are 64% and 85%, respectively.
The sensitivity and specificity of the CAM-ICU screening for ICU
We conducted this retrospective cohort study of critically ill patients delirium range between 47% and 100% and between 71% and 100%,
treated with valproate for agitation from December 2012 through respectively, suggesting that the chart review method is valid [17]. If
February 2015 at Maine Medical Center (MMC) (Portland, ME) and CAM-ICU scores were recorded, their accuracy was confirmed by
Intermountain Medical Center (IMC) (Murray, UT). Both institutions reviewing caregiver notes. If the 2 sources agreed, at least 1 positive
are tertiary, university-affiliated medical centers, with 42 and 84 ICU CAM-ICU assessment defined a delirium day. If the 2 sources disagreed
beds, respectively. The protocol was approved by the Institutional or scores were unavailable, a chart review was conducted, and patients
Review Boards at both medical centers (MMC IRB #4583X and IMC were assigned a delirium day if the abstractor answered yes to the fol-
IRB #1040472), and the need for patient consent was waived. lowing question: Is there any evidence from the chart of delirium (ie,
Patients were included in a convenience sample if they were ≥18 delirium, mental status change, inattention, disorientation, hallucina-
years of age and treated with valproate for agitation for ≥2 days while tions, inappropriate behavior, or other) [16]?
in the ICU. Patients were excluded if they were prescribed valproate The need for concomitant psychoactive medications was also
prior to hospitalization, if they received multiple courses of valproate examined and included opioids, benzodiazepines, antipsychotics,
during their admission, or if it was initiated for other indications clonidine, dexmedetomidine, phenobarbital, ketamine, and propofol.
(eg, seizures or migraines). Neither institution had a dosing or monitoring The number of doses and total daily dose of each agent were recorded,
protocol for valproate therapy when used for the treatment of except for propofol, dexmedetomidine, and ketamine, where mean
ICU agitation. hourly doses were recorded. Opioids were converted to fentanyl
equivalents (100 μg IV fentanyl =1.5 mg IV hydromorphone =10 mg
IV morphine =20 mg oral oxycodone), and benzodiazepines were
2.2. Demographics and patient characteristics
converted to lorazepam equivalents (1 mg IV lorazepam =2 mg IV
midazolam) [18,19].
Patient demographic data included age, sex, race, weight, body mass
index, history of psychiatric diagnosis, alcohol or substance abuse,
2.5. Safety outcomes
reason for ICU admission, need for mechanical ventilation, and Acute
Physiology and Chronic Health Evaluation (APACHE) III score within
Safety parameters were examined for the hospital duration
24 hours of ICU admission. Clinical outcomes were descriptive and
of valproate therapy. Records were specifically reviewed for
included hospital length of stay, ICU length of stay, and ICU mortality.
possible valproate-induced hepatotoxicity, hematologic toxicity,
hyperammonemia, pancreatitis, and Stevens-Johnson syndrome. The
2.3. Valproate prescribing practices number of patients who had valproate discontinued because of a
suspected adverse event was also recorded. Only patients who had
Drug administration data were collected from the electronic Medication baseline laboratory values prior to valproate initiation were assessed
Administration Record at both centers. Valproate data included total daily for laboratory-based adverse effects.
dose, route of administration, duration of therapy, time from ICU admission Hepatotoxicity was defined as a new alanine aminotransferase N3 times
to initiation, mechanical ventilation at the time of initiation, serum con- the upper limit of normal (ULN) (N120 U/L), alkaline phosphatase N2 times
centration monitoring, use of an initial loading dose, use of a taper to the ULN (N234 U/L), total bilirubin N2 times the ULN (N2 mg/dL), or a
discontinue therapy, and whether or not valproate was continued at doubling of the baseline value if it was already abnormal following
ICU and hospital discharge. Continuation at hospital discharge was valproate initiation [20]. Suspected cases of hepatotoxicity were
characterized as either possibly inadvertent or intentional with further assessed using the validated Roussel Uclaf Causality Assessment
rationale. Method (RUCAM) [21].
Hematologic toxicity was defined as a new leukocyte count b4200
2.4. Efficacy outcomes cells/mm 3, absolute neutrophil count b2400 cells/mm 3, or platelet
count b140 000 cells/mm 3 or platelet drop by N50% if platelets were
Efficacy data were collected starting 2 days before valproate initia- already b140 000 cells/mm 3 following valproate initiation. Thresholds
tion and continuing for 7 days or until discontinuation, whichever were set according to the lower limit of normal for the MMC laboratory
came first. The 7-day interval was selected to allow a reasonable time [22,23].
to observe efficacy or lack thereof. Hyperammonemia was defined as a new serum ammonia
The daily proportion of agitated patients was examined. Richmond level N60 μmol/L following valproate initiation. Suspected cases of
Agitation-Sedation Scale (RASS) scores were assessed at IMC, and hyperammonemia were further reviewed for treatment strategies. Pancreati-
Sedation-Agitation Scale (SAS) scores were assessed at MMC. Assess- tis was defined as a new serum lipase level N3 times the ULN (N189 IU/L) in
ments of sedation documented in the electronic medical record have re- the presence of clinical symptoms. A diagnosis of Stevens-Johnson syndrome
cently been shown to be reliable when dichotomized to oversedated or was assessed by reviewing caregiver notes.
not oversedated [15]. Because the recording of RASS and SAS scores was
inconsistent in this study and sometimes contradicted what was docu- 2.6. Statistical analysis
mented in caregiver notes, an algorithm was used to improve the accu-
racy of assessments (Electronic Supplement Fig. 1). If RASS or SAS scores Continuous variables were reported as median values with
were available, their accuracy was confirmed by reviewing caregiver interquartile range (IQR) and were compared between valproate day
notes. If the 2 sources agreed, at least 1 RASS score ≥1 or SAS score ≥5 1 and valproate day 3 with the Wilcoxon rank-sum test. Categorical
defined an agitation day. If the 2 sources disagreed or sedation scores and binary variables were reported as frequencies and percentages
D.J. Gagnon et al. / Journal of Critical Care 37 (2017) 119–125 121

and were compared between valproate day 1 and valproate day 3 with and 22 (13-30) days, respectively. Seven (13%) patients died in the
McNemar test for paired proportions. Valproate day 1 (day of valproate ICU, and 9 (17%) died in the hospital.
initiation) was selected as the baseline because it had the highest inci-
dence of agitation, representing maximum agitation and the decision
3.2. Valproate prescribing practices
to prescribe valproate. Valproate day 3 was chosen for comparison
based on the premise that failure to respond in 2 days would lead to a
The median time from ICU admission to valproate initiation
change in therapy. Analyses were conducted using Analyze-it v2.25,
was 7 (4-10) days (Table 2). Loading doses were administered to
and P b .05 was considered statistically significant.
22 (42%) patients at a median dose of 1800 (1500-2000) mg or 28
Repeated-measures models were used to assess concomitant psy-
(19-31) mg/kg. The median maintenance dose on valproate day 3 was
choactive medication use over time. Linear mixed models were devel-
1500 (1000-2275) mg/d or 23 (15-31) mg/kg/d administered in 1 to 4
oped using a between-within degrees-of-freedom method, identifying
divided doses. Valproate was tapered in 14 (26%) patients over a
subjects as repeated effects. Fixed effects were related to day of treat-
median of 3 (2-7) days by reducing the dose and/or increasing the
ment, and a Toeplitz-structured covariance matrix was selected from a
dosing interval. Eighteen valproate total serum concentrations were
range of alternatives by minimization of the Akaike information criteri-
obtained (n = 9 trough, n = 8 random, n = 1 unclear) in 8 (15%)
on. The structure was fit to the data via the restricted maximum likeli-
patients and were a median of 49 (34-63) mg/L. Seventeen (32%)
hood method, similarly selected to minimize the fit statistics. To avoid
patients were continued on valproate at hospital discharge; 3 (18%)
inappropriate deflation of the mean, each medication was assessed for
patients did not have clear rationale.
response only for patients who actually received that agent. Linear
models were obtained for each agent, estimating a slope and intercept
relating day of treatment to daily dose. Analyses were conducted 3.3. Efficacy outcomes
using SAS version 9.3, and P b .05 was considered statistically significant.
The incidence of agitation increased from 73% on valproate day −2
3. Results to 80% on valproate day −1 and to 96% on valproate day 1 (Fig. 2).
The incidence of agitation significantly decreased following the initia-
3.1. Demographics and patient characteristics tion of valproate to 61% on valproate day 3 (P b .0001). The incidence
of delirium exhibited a similar trend, occurring in 41% of patients on
Three-hundred fifty-one adult patients were treated with valproate valproate day −2, 51% on valproate day −1, and 68% on valproate day
during the study period. Two-hundred ninety-eight were excluded, 1 (Fig. 2). The incidence of delirium also decreased by valproate day 3
leaving 53 evaluable patients representing 522 patient-days of (49%, P = .012). The source of agitation and delirium assessments
valproate therapy (Fig. 1). The median (IQR) patient age was 53 (40- (SAS/RASS or CAM-ICU vs chart review) is described in Electronic
70) years, and most patients were male (64%) (Table 1). Respiratory Supplement Table 1. The impact of a loading dose on the incidence of
failure (26%), trauma (22%), and acute stroke (15%) were the most com- agitation and delirium was also assessed, but statistical testing was
mon reasons for ICU admission. The median APACHE III score was 59 not conducted because of the small number of patients in each group
(40-73). The median ICU and hospital lengths of stay were 14 (9-20) (Electronic Supplement Table 2).

Patients Screened (n = 12,776)


MMC (n = 5,404)
IMC (n = 7,372)
Received Valproate (n = 351)
MMC (n = 184)
IMC (n =167)

Excluded (n = 298) Excluded IMC (n = 154)


Excluded MMC (n = 144) Home drug (n = 171) Home drug (n = 104)
Home drug (n = 67) Seizures (n = 79) Seizures (n = 45)
Seizures (n = 34) Less than 2 days (n = 18) Less than 2 days (n = 2)
Less than 2 days (n = 16) Not evaluated (n = 14) Seizure prophylaxis (n = 2)
Not evaluated (n = 14) Multiple courses (n = 6) Migraines (n = 1)
Multiple courses (n = 6) Other indication (n = 6)
Bipolar disorder (n = 3) Seizure prophylaxis (n = 4)
Seizure prophylaxis (n = 2)
Lip biting (n = 1)
Headache (n = 1)

Evaluable Patients (n = 53)


MMC (n = 40)
IMC (n = 13)

Fig. 1. Study cohort and application of exclusion criteria. Patients admitted to the ICU at both medical centers were screened for the receipt of valproate during the study period. Patients
were included if they were ≥18 years of age and treated with valproate for agitation for ≥2 days while in the ICU. Patients were excluded if they were prescribed valproate prior to
admission, if they received multiple courses of valproate during their admission, or if it was initiated for other indications (eg, seizures or migraines).
122 D.J. Gagnon et al. / Journal of Critical Care 37 (2017) 119–125

Table 1 Table 2
Demographics and patient characteristics Valproate prescribing patterns

All patients (N = 53) All patients (N = 53)

Age, y 53 (40-70) ICU admission to valproate initiation, d 7 (4-10)


Male, n (%) 34 (64) Mechanical ventilation at valproate initiation, n (%) 24 (45)
Weight, kg 76 (64-94) Loading dose of valproate, n (%) 22 (42)
Body mass index, kg/m2 26 (22-31) mg 1800 (1500-2000)
White, n (%) 46 (87) mg/kg 28 (19-31)
Psychiatric diagnosis, n (%) Maintenance dose on valproate day 3a
None 22 (42) mg/d 1500 (1000-2275)
Depression 21 (40) mg/kg/d 23 (15-31)
Anxiety 16 (30) Route of administration on valproate day 3a
Bipolar 6 (11) IV 31 (61)
Dementia 3 (6) Enteral 18 (35)
Multiple personality 1 (2) IV and enteral 2 (4)
Othera 5 (9) Valproate total serum concentration obtained, n (%) 8 (15)
Substance abuse history, n (%) Total valproate duration, d 7 (4-10)
None 33 (62) Valproate taper, n (%) 14 (26)
Alcohol 7 (13) Reduced dose 2 (14)
Substance 4 (8) Increased interval 1 (7)
Both 9 (17) Combination of both 11 (79)
ICU admission reason, n (%) Duration of valproate taper, d 3 (2-7)
Respiratory failure 14 (26) Valproate continued at ICU discharge, n (%) 23 (43)
Stroke 8 (15) Valproate continued at hospital discharge, n (%) 17 (32)
Trauma, not TBI 7 (13) Without clear rationale 3 (18)
Trauma, TBI 5 (9) Taper instructions present 6 (35)
Sepsis 5 (9) Goal of chronic therapyb 6 (35)
Alcohol withdrawal 1 (2) Follow-up instructions present 2 (12)
Surgery 1 (2)
Continuous variables reported as median (IQR) and categorical or binary variables as
Otherb 12 (23)
frequency (%).
ICU LOS, d 14 (9-20) a
There were 51 evaluable patients on valproate day 3.
Hospital LOS, d 22 (13-30) b
Ongoing agitation and mood stabilization (n = 3), agitation associated with dementia
ICU mortality, n (%) 7 (13)
(n = 2), and posttraumatic seizures developing 8 days after the initiation of valproate for
Hospital mortality, n (%) 9 (17)
agitation (n = 1).
Mechanical ventilation, n (%) 43 (81)
APACHE III score 59 (40-73)
Predicted mortality (%) 21 (9-32)
Predicted ICU LOS, days 7 (4-8) 3.4. Safety outcomes
Continuous variables reported as median (IQR) and categorical or binary variables as
frequency (%). LOS indicates length of stay; TBI, traumatic brain injury. Most patients (79%) had liver function tests monitored before and
a
Attention deficit hyperactivity disorder (n = 2), posttraumatic stress disorder (n = 2), during valproate therapy (Table 4). One patient developed an isolated
and other disorder not specified (n = 1). elevated alkaline phosphatase (249 U/mL), with a RUCAM score of 4,
b
Drug/substance overdose (n = 3), encephalopathy (n = 2), cardiac arrest (n = 1), suggesting a possible association with valproate. One patient had an
seizures (n = 1), shock (n = 1), endocarditis (n = 1), bifrontal meningioma resection
(n = 1), traumatic subarachnoid hemorrhage (n = 1), and subarachnoid hemorrhage
elevated total bilirubin (2.4 mg/dL), but the RUCAM score was 1,
with hypoxic-ischemic encephalopathy (n = 1). suggesting that it was unlikely related to valproate. Valproate was not
discontinued in either case.
All patients had a complete blood count obtained before and during
valproate therapy (Table 4). Thrombocytopenia was the most common
potential hematologic toxicity observed, occurring in 7 (13%) patients,
The proportion of patients receiving an opioid decreased from with 1 having valproate discontinued. Two (5%) patients developed
valproate day 1 to valproate day 3 (77% vs 65%, P = .02) (Table 3). neutropenia, but neither had valproate discontinued.
Median daily fentanyl equivalents administered also decreased Thirty-two (60%) patients had an ammonia level measured during
(1347 vs 800 μg/d, P = .04). The number of patients receiving valproate therapy (Table 4). Six (19%) patients had an elevated ammo-
dexmedetomidine (47% vs 24%, P = .004) and quetiapine (49% vs 35%, nia level N60 μmol/L. Valproate was discontinued in 4 of these patients.
P = .04) decreased by valproate day 3. Neither the number of patients No relationship between dose and hyperammonemia was observed
receiving a benzodiazepine (53% vs 47%, P = .2) nor the median loraze- (data not shown).
pam equivalents administered (10.5 vs 10 mg/d, P = .7) were reduced.
The impact of a loading dose on concomitant psychoactive medication 4. Discussion
use was also examined (Electronic Supplement Table 3). Fentanyl
equivalents and lorazepam equivalents paradoxically increased in the Our study provides the most comprehensive information to date on
loading dose subgroup, which might have been due to 3 patients who valproate prescribing practices for ICU agitation. In addition, we de-
had excessively high sedative requirements. scribe the safety and efficacy of valproate therapy in greater detail
Mixed models were created to assess the impact of valproate on than prior studies and more than double the number of reported pa-
daily concomitant psychoactive medication use (Electronic Supplement tients treated with valproate for ICU agitation. In our study, valproate
Figs. 3A-E). Fentanyl equivalents (−185 μg/d, P = .0003), was initiated 1 week after ICU admission to treat escalating agitation.
lorazepam equivalents (−2.1 mg/d, P = .0004), dexmedetomidine Loading doses were administered to 42% of patients, and the mainte-
dose (−0.15 μg/kg per hour per day, P b .0001), and propofol dose nance dose was 1500 mg/d (23 mg/kg/d) administered in 1 to 4 divided
(−5 μg/kg per minute per day, P = .003) decreased, whereas doses. This therapy was associated with a reduction in the incidence of
quetiapine (−1 mg/d, P = .9) and clonidine (−0.02 mg/d, P = .8) agitation and delirium within 48 hours of initiation and a reduction in
dose did not. There were too few observations to create models for daily fentanyl and lorazepam equivalents administered in mixed
the remaining concomitant medications. Between-centers comparisons models analyses. Observed adverse effects were infrequent, with
were not performed because of the small number of patients studied. thrombocytopenia and hyperammonemia being the most common.
D.J. Gagnon et al. / Journal of Critical Care 37 (2017) 119–125 123

Fig. 2. Incidence of agitation and delirium by day of valproate therapy. Incidence of agitation and delirium steadily increased on valproate day −2 and valproate day −1 and ultimately
peaked on the day valproate was initiated (valproate day 1). The incidence of both agitation (P b .0001) and delirium (P = .012) significantly decreased by valproate day 3.

We are aware of only 4 published reports of valproate to treat ICU ICU agitation. Daily maintenance doses were slightly higher than those
agitation or delirium. A case series of 6 patients reported behavioral im- previously reported [11-13]. The optimal dosing regimen for valproate
provement with the addition of valproate (500-2500 mg in 2-4 divided in the ICU—and whether loading doses are beneficial—is not answered
daily doses) to more conventional treatments [11]. A second case series by our research and should be addressed in future studies.
of 15 patients with hyperactive delirium demonstrated that valproate Valproate therapy was continued for a median of 7 days, and 43% of
therapy (1133-1258 mg in 2-3 divided daily doses) resulted in delirium patients were transferred out of the ICU on it, which suggests that many
resolution in 13 of 16 episodes within 6.2 days [12]. A third case series of patients were responding favorably and providers were comfortable
2 patients with agitation and delirium reported that valproate therapy transitioning patients to a non-ICU environment on this agent. On the
(500 mg in 2 divided daily doses) resulted in reduced agitation within other hand, it is possible that valproate was inadvertently continued
24 hours [13]. Lastly, an abstract-only publication reported that deliri- upon ICU discharge. Seventeen patients (32%) were discharged
um resolution occurred by day 7 in 18 patients receiving valproate from the hospital on valproate, which is consistent with published
[14]. These reports provided important data but are limited by small reports on the rate of continuation of psychoactive medications
sample sizes and limited descriptions of valproate prescribing patterns, upon hospital discharge [6,24,25]. Strategies to prevent inadvertent
safety, and efficacy. psychoactive medication continuation are needed and are under way
Our study complements these reports in several ways. We collected at our medical centers.
detailed information on valproate prescribing practices. Valproate ther- Within 48 hours of initiating valproate, the incidence of agitation
apy was initiated a median of 1 week after ICU admission, which sug- and delirium decreased by 36% and 28%, respectively, relative to
gests that it was used in patients who were refractory to or intolerant valproate day 1. Although we did not assess for the resolution of agita-
of more traditional agents. Loading doses were used in 42% of our pa- tion and delirium, we found that symptom control occurred more
tients. Loading doses may achieve therapeutic serum concentrations quickly in our study than others, perhaps a function of slightly higher
more rapidly, but this practice has not been previously described for daily maintenance doses [12-14]. The anxiolytic effects of valproate

Table 3
Concomitant psychoactive medication use by day of valproate therapy

Medicationa Valproate day −1 Valproate day 1 Valproate day 2 Valproate day 3 Valproate day 4

Opioid, n (%) 41 (80) 41 (77) 41 (77) 33 (65)* 30 (64)


Fentanyl equivalents, μg/d 1707 1347 629 800* 467
Benzodiazepine, n (%) 31 (61) 28 (53) 25 (47) 24 (47) 24 (51)
Lorazepam equivalents, mg/d 6 10.5 12 10 4
Dexmedetomidine, n (%) 27 (53) 25 (47) 19 (36) 12 (24)** 12 (26)
Dexmedetomidine, μg/kg/h 1 1.2 0.9 1 0.9
Dexmedetomidine or clonidine, n (%) 34 (67) 35 (66) 31 (59) 28 (55) 22 (47)
Quetiapine, n (%) 21 (41) 26 (49) 21 (40) 18 (35)* 17 (36)
Quetiapine, mg/d 200 100 150 225 225
Propofol, n (%) 10 (20) 9 (17) 8 (15) 9 (18) 8 (17)
Propofol, μg/kg/min 41 45 40 37 30
Haloperidol, n (%) 8 (16) 10 (19) 9 (17) 8 (16) 9 (19)
Haloperidol, mg/d 8.5 14 10 10 10
Olanzapine, n (%) 9 (18) 11 (21) 9 (17) 7 (14) 7 (15)
Olanzapine, mg/d 10 10 5 10 10
Risperidone, n (%) 3 (6) 2 (4) 0 (0) 0 (0) 0(0)
Risperidone, mg/d 1 1 0 0 0
Phenobarbital, n (%) 2 (4) 1 (2) 2 (4) 3 (6) 6 (13)
Phenobarbital, mg/d 130 325 780 812 655

Continuous variables reported as median (IQR) and categorical or binary variables as frequency (%).
*P b 0.05 and **P b .01 for comparisons between valproate day 1 and valproate day 3.
a
One patient received ketamine on valproate days 2 to 5 at a rate of 0.16 to 0.2 mg/kg/h.
124 D.J. Gagnon et al. / Journal of Critical Care 37 (2017) 119–125

Table 4 valproate-treated patients, might be due to urea cycle interference by


Safety outcomes valproate and its reactive metabolite, valproyl-CoA [33,34]. A consistent
All patients (N = 53) correlation between hyperammonemia and valproate dose, serum con-
Hepatotoxicity
centration, or duration of therapy has yet to be described, and no such
Liver function test obtained, n (%) 42 (79) relationship was identified in our study [35]. In addition, ammonia
ALT N120 U/L, n (%) 0 (0) levels may not correlate with the level of encephalopathy and are
Alkaline phosphatase N234 U/L, n (%)a 1 (2) often elevated in patients without encephalopathy [35].
Total bilirubin N2 mg/dL, n (%)b 1 (2)
Thrombocytopenia was the most common possible hematologic ad-
Discontinued because of hepatotoxicity, n (%) 0 (0)
Hematologic toxicity verse effect observed, occurring in 13% of patients. Only 1 patient had a
Complete blood count obtained, n (%) 53 (100) platelet count decrease of N50% following valproate initiation. Throm-
WBC b4200 cells/mm3, n (%) 1 (2) bocytopenia may occur in 5% to 60% of valproate-treated patients and
Platelets b140 000 cells/mm3 or 50% drop, n (%) 7 (13) in 5% to 28% of valproate-treated psychiatric inpatients [36-39].
Neutrophil count obtained, n (%) 44 (83)
Valproate-induced thrombocytopenia may be due to dose-dependent
Neutrophils b2400 cells/mm3, n (%)c 2 (5)
Discontinued because of hematologic toxicity, n (%)d 1 (2) bone marrow suppression and/or autoantibody formation [36]. We
Pancreatitis did not evaluate for valproate-associated disorders of hemostasis, but
Lipase obtained, n (%) 4 (8) thrombocytopathy, hypofibrinogenemia, acquired von Willebrand dis-
Lipase N189 IU/L, n (%) 0 (0)
ease type 1, and clotting factor deficiencies have been described [36].
Discontinued because of pancreatitis, n (%) 0 (0)
Rash Thrombocytopenia and coagulopathy are common in critically ill pa-
Rash, n (%)e 1 (2) tients, and etiologies were difficult to assess given the retrospective na-
Discontinued because of rash, n (%) 1 (2) ture of our study [40].
Hyperammonemia Total serum valproate concentrations were measured in 15% of pa-
Ammonia obtained, n (%) 32 (60)
tients, but no consistent approach to therapeutic drug monitoring was
Ammonia N60 μmol/L, n (%) 6 (19)
Valproate initiation to ammonia N60 μmol/L, d 10 (6-13) evident, as half of serum concentrations were trough determinations
Peak ammonia level, μmol/L 75 (70-222) and half were randomly obtained. A total valproate serum concentra-
Management of hyperammonemia tion of 50 to 125 mg/L is recommended by the American Psychiatric
Valproate discontinued, n (%) 4 (67)
Association's guidelines for the treatment of patients with bipolar disor-
No action taken, n (%) 2 (33)
Lactulose, n (%) 1 (17)
der, but this reference range has not been validated for ICU agitation
[41]. Because valproate is highly protein bound to albumin (≥90%), its
Continuous variables reported as median (IQR) and categorical or binary variables as
frequency (%). ALT indicates alanine aminotransferase; WBC, white blood cell.
free fraction can vary as a function of altered plasma protein binding,
a
Alkaline phosphatase of 249 U/mL. RUCAM score = 4. which is common in critically ill patients [42]. Future research should at-
b
Total bilirubin of 2.4 mg/dL. RUCAM score = 1. tempt to correlate clinical effect and toxicity with free and total serum
c
Neutrophils were 1800 cells/mm3 in one patient and 2300 cells/mm3 in another. valproate concentrations [43].
d
Discontinued because of thrombocytopenia (106 000 cells/mm3) on valproate day 8
Our study has limitations that warrant discussion. The small number
per provider notes.
e
Rash not consistent with Stevens-Johnson syndrome. of patients evaluated means that we were unable to confidently quanti-
fy adverse effects or estimate event rates. The retrospective design did
not allow for direct patient observations; SAS or RASS scores and
might be due to GABA modulation or antagonism of glutamate at N- CAM-ICU results were sometimes unavailable. Our chart review method
methyl-D-aspartate receptors [9,10]. Valproate is a potential option for made up for some of this shortcoming, but prospective and routine eval-
the management of ICU delirium due to its proposed biochemical, anti- uations of agitation and delirium would improve the reliability of our
inflammatory, antioxidant, and transcriptional/neurotrophic effects [26]. results. Other retrospective studies of ICU agitation and delirium have
A randomized controlled trial to assess valproate therapy for hyperactive used a chart review similar to ours [44-46]. In addition, agitation is a
or mixed ICU delirium is currently under way (NCT02343575). continuum from mild restlessness to dangerous agitation; in this retro-
Valproate has been used to treat agitation in other patient spective study, we could not define the severity of agitation that should
populations with mixed results. It has been used to treat behavior disor- trigger valproate therapy, but future prospective studies should address
ders in patients recovering from traumatic brain injury, but the overall this issue. Because of these limitations, the effectiveness of valproate for
quality of data is low [27,28]. Valproate has been associated with a both agitation and delirium in our study must be considered hypothesis
trend toward increased mortality when used for seizure prophylaxis generating and interpreted within the context of our methodology. The
following traumatic brain injury (13.4% in the valproate group vs 7.2% etiology of agitation was not defined, and we cannot recommend
in the phenytoin group; P = .07; relative risk, 2.0; 95% confidence inter- valproate therapy for a specific ICU patient population. Several patients
val, 0.9-4.1) [29]. This finding must be interpreted with caution because had a history of a psychiatric diagnosis at baseline, but this was expect-
the study was not powered to assess mortality, the mortality rate in ed, as it has been associated with up to a 5-fold increase in the risk of ICU
both groups was lower than the rate observed in a previous study in agitation [45,47]. Although our mixed models analyses reduced the bias
the same patient population at the same medical center, and the obser- from repeated measures, our sample size precluded their use in infre-
vation has not been confirmed in subsequent studies. A reassuring case- quently used medications. Practice variations between centers were
control study of more than 90 000 patients with dementia, a patient not controlled and could have confounded our results. Lack of a control
population that may be more similar to ours, found that valproate was group prevented us from concluding that valproate was responsible for
not associated with mortality [30]. A systematic review and meta- the behavioral improvement and not the result of critical illness resolu-
analysis of valproate to treat agitation in dementia found that there tion. Adverse effect monitoring was not standardized and could be under-
were limited data to support that practice [31]. These findings have represented. Because of these limitations, our findings must be
been corroborated by a more recent evidence summary published by interpreted with caution and require confirmatory testing.
the National Institute for Health and Care Excellence [32]. It should be
emphasized that data describing valproate use in these patient popula- 5. Conclusions
tions cannot be generalized to ICU patients.
Hyperammonemia was the most common possible adverse Valproate therapy is an emerging treatment option for ICU agitation
effect observed, occurring in 19% of patients who had an ammonia not responding to more traditional pharmacologic agents. In our study,
level evaluated. Hyperammonemia, which may occur in 6% to 58% of valproate therapy at a dose of 23 mg/kg/d was associated with a
D.J. Gagnon et al. / Journal of Critical Care 37 (2017) 119–125 125

reduction in agitation, delirium, and concomitant psychoactive [17] Gusmao-Flores D, Salluh JI, Chalhub RÁ, Quarantini LC. The Confusion Assessment
Method for the Intensive Care Unit (CAM-ICU) and Intensive Care Delirium Screen-
medication use within 48 hours of initiation. Observed adverse effects ing Checklist (ICDSC) for the diagnosis of delirium: a systematic review and meta-
were infrequent; providers should regularly monitor complete blood analysis of clinical studies. Crit Care 2012;16(4):R115.
counts, liver function test results, and ammonia levels during valproate [18] McPherson ML. Demystifying opioid conversion calculations: a guide for
effective dosing. Bethesda, MD: American Society of Health-System Pharmacists;
therapy. These promising results should be validated in prospective 2009.
controlled studies. [19] Mehta S, Burry L, Cook D, Fergusson D, Steinberg M, Granton J, et al. Daily
sedation interruption in mechanically ventilated critically ill patients cared for
with a sedation protocol: a randomized controlled trial. JAMA 2012;308(19):
Conflicts of interest 1985–92.
[20] Navarro VJ, Senior JR. Drug-related hepatotoxicity. N Engl J Med 2006;354(7):731–9.
[21] Danan G, Benichou C. Causality assessment of adverse reactions to drugs—I. A novel
None.
method based on the conclusions of international consensus meetings: application
to drug-induced liver injuries. J Clin Epidemiol 1993;46(11):1323–30.
Financial disclosure [22] National Cancer Institute. Common terminology criteria for adverse events (CTCAE)
v4.0. http://ctep.cancer.gov/protocolDevelopment/electronic_applications/ctc.htm.
[Accessed April 13, 2015].
This research did not receive any specific grant from funding [23] Wazny LD, Ariano RE. Evaluation and management of drug-induced thrombocyto-
agencies in the public, commercial, or not-for-profit sectors. penia in the acutely ill patient. Pharmacotherapy 2000;20(3):292–307.
[24] Marshall J, Herzig SJ, Howell MD, Le SH, Mathew C, Kats JS, et al. Antipsychotic uti-
lization in the intensive care unit and in transitions of care. J Crit Care 2016;33:
Prior presentation of data 119–24.
[25] Kram BL, Kram SJ, Brooks KR. Implications of atypical antipsychotic prescribing in
the intensive care unit. J Crit Care 2015;30(4):814–8.
Preliminary data were presented at the Society of Critical Care [26] Sher Y, Miller Cramer AC, Ament A, Lolak S, Maldonado JR. Valproic acid for treat-
Medicine 45th Annual Critical Care Congress; February 20-24, 2016; ment of hyperactive or mixed delirium: rationale and literature review. Psychoso-
matics 2015;56(6):615–25.
Orlando, FL.
[27] Plantier D. Luauté J; SOFMER group. Drugs for behavior disorders after traumatic
brain injury: systematic review and expert consensus leading to French recommen-
Acknowledgments dations for good practice. Ann Phys Rehabil Med 2016;59(1):42–57.
[28] Luauté J, Plantier D, Wiart L, Tell L, SOFMER Group. Care management of
the agitation or aggressiveness crisis in patients with TBI. Systematic review
The authors would like to acknowledge Casey Hilfrank, PharmD, and of the literature and practice recommendations. Ann Phys Rehabil Med 2016;
Jonathon Ashton, PharmD, for their assistance with data development. 59(1):58–67.
[29] Temkin NR, Dikmen SS, Anderson GD, Wilensky AJ, Holmes MD, Cohen W, et al.
Valproate therapy for prevention of posttraumatic seizures: a randomized trial. J
Appendix A. Supplementary data Neurosurg 1999;91(4):593–600.
[30] Maust DT, Kim HM, Seyfried LS, Chiang C, Kavanagh J, Schneider LS, et al. Antipsy-
chotics, other psychotropics, and the risk of death in patients with dementia: num-
Supplementary data to this article can be found online at http://dx. ber needed to harm. JAMA Psychiatry 2015;72(5):438–45.
doi.org/10.1016/j.jcrc.2016.09.006. [31] Lonergan E, Luxenberg J. Valproate preparations for agitation in dementia. Cochrane
Database Syst Rev 2009;3:CD003945.
[32] Management of aggression, agitation and behavioural disturbances in dementia:
References valproate preparations. National Institute for Health and Care Excellence (NICE);
2015 [https://www.nice.org.uk/advice/esuom41/chapter/Full-evidence-summary,
[1] Fraser GL, Prato BS, Riker RR, Berthiaume D, Wilkins ML. Frequency, severity, and Accessed 1 Jun 2016].
treatment of agitation in young versus elderly patients in the ICU. Pharmacotherapy [33] Carr RB, Shrewsbury K. Hyperammonemia due to valproic acid in the psychiatric set-
2000;20(1):75–82. ting. Am J Psychiatry 2007;164(7):1020–7.
[2] Wunsch H, Kahn JM, Kramer AA, Rubenfeld GD. Use of intravenous infusion sedation [34] Aires CC, van Cruchten A, Ijlst L, de Almeida IT, Duran M, Wanders RJ, et al. New in-
among mechanically ventilated patients in the United States. Crit Care Med 2009; sights on the mechanisms of valproate-induced hyperammonemia: inhibition of he-
37(12):3031–9. patic N-acetylglutamate synthase activity by valproyl-CoA. J Hepatol 2011;55(2):
[3] Roberts DJ, Haroon B, Hall RI. Sedation for critically ill or injured adults in the inten- 426–34.
sive care unit: a shifting paradigm. Drugs 2012;72(14):1881–916. [35] Chopra A, Kolla BP, Mansukhani MP, Netzel P, Frye MA. Valproate-induced
[4] Weinert CR, Calvin AD. Epidemiology of sedation and sedation adequacy for me- hyperammonemic encephalopathy: an update on risk factors, clinical correlates
chanically ventilated patients in a medical and surgical intensive care unit. Crit and management. Gen Hosp Psychiatry 2012;34(3):290–8.
Care Med 2007;35(2):393–401. [36] Verrotti A, Scaparrotta A, Grosso S, Chiarelli F, Coppola G. Anticonvulsant drugs and
[5] Riker RR, Fraser GL. Adverse events associated with sedatives, analgesics, and other hematological disease. Neurol Sci 2014;35(7):983–93.
drugs that provide patient comfort in the intensive care unit. Pharmacotherapy [37] Vasudev K, Keown P, Gibb I, McAllister-Williams RH. Hematological effects of
2005;25(5 Pt 2):8S–18S. valproate in psychiatric patients: what are the risk factors? J Clin Psychopharmacol
[6] Gagnon DJ, Riker RR, Glisic EK, Kelner A, Perrey HM, Fraser GL. Transition from 2010;30(3):282–5.
dexmedetomidine to enteral clonidine for ICU sedation: an observational pilot [38] Conley EL, Coley KC, Pollock BG, Dapos SV, Maxwell R, Branch RA. Prevalence and
study. Pharmacotherapy 2015;35(3):251–9. risk of thrombocytopenia with valproic acid: experience at a psychiatric teaching
[7] Fraser GL, Riker RR. Phenobarbital provides effective sedation for a select cohort of hospital. Pharmacotherapy 2001;21(11):1325–30.
adult ICU patients intolerant of standard treatment: a brief report. Hosp Pharm [39] Trannel TJ, Ahmed I, Goebert D. Occurrence of thrombocytopenia in psychiatric pa-
2006;41:17–23. tients taking valproate. Am J Psychiatry 2001;158(1):128–30.
[8] Divalproex [package insert]. North Chicago, IL: AbbVie Inc.; 2014. [40] Hunt BJ. Bleeding and coagulopathies in critical care. N Engl J Med 2014;370(9):
[9] Perucca E. Pharmacological and therapeutic properties of valproate: a summary after 847–59.
35 years of clinical experience. CNS Drugs 2002;16(10):695–714. [41] American Psychiatric Association. Practice guideline for the treatment of patients
[10] Rosenberg G. The mechanisms of action of valproate in neuropsychiatric disorders: with bipolar disorder (revision). Am J Psychiatry 2002;159:1–50.
can we see the forest for the trees? Cell Mol Life Sci 2007;64(16):2090–103. [42] Levy RH. Monitoring of free valproic acid levels? Ther Drug Monit 1980;2(2):
[11] Bourgeois JA, Koike AK, Simmons JE, Telles S, Eggleston C. Adjunctive valproic acid 199–201.
for delirium and/or agitation on a consultation-liaison service: a report of six [43] Hatton C, Riker RR, Gagnon DJ, May T, Seder DB, Fraser GL. Free serum valproate con-
cases. J Neuropsychiatry Clin Neurosci 2005;17(2):232–8. centration more reliable than total concentration in critically ill patients. Resuscita-
[12] Sher Y, Miller AC, Lolak S, Ament A, Maldonado JR. Adjunctive valproic acid in tion 2016;105:e15–6.
management-refractory hyperactive delirium: a case series and rationale. J Neuro- [44] Burk RS, Grap MJ, Munro CL, Schubert CM, Sessler CN. Agitation onset, frequency,
psychiatry Clin Neurosci 2015;27(4):365–70. and associated temporal factors in critically ill adults. Am J Crit Care 2014;23(4):
[13] Alam A, Puri NV. Inefficacy of antipsychotics in treatment of delirium and agitation 296–304.
in two cases of Bickerstaff brainstem encephalitis. J Neuropsychiatry Clin Neurosci [45] Burk RS, Grap MJ, Munro CL, Schubert CM, Sessler CN. Predictors of agitation in crit-
2014;26(2):176–8. ically ill adults. Am J Crit Care 2014;23(5):414–23.
[14] Fitz K, Harding A. Safety and efficacy of valproic acid for treatment of delirium in [46] O'Connor H, Al-Qadheeb NS, White AC, Thaker V, Devlin JW. Agitation during
critically ill patients [abstract]. Crit Care Med 2011;39(12 Suppl.):239. prolonged mechanical ventilation at a long-term acute care hospital: risk factors,
[15] Shekar RD, Singh TD, Guru PK, Sakusic A, Gajic O, O'Horo JC, et al. Identification of treatments, and outcomes. J Intensive Care Med 2014;29(4):218–24.
acute brain failure using electronic medical records. J Crit Care 2016;34:12–6. [47] Jaber S, Chanques G, Altairac C, Sebbane M, Vergne C, Perrigault PF, et al. A prospec-
[16] Pisani MA, Araujo KL, Van Ness PH, Zhang Y, Ely EW, Inouye SK. A research algorithm to tive study of agitation in a medical-surgical ICU: incidence, risk factors, and out-
improve detection of delirium in the intensive care unit. Crit Care 2006;10(4):R121. comes. Chest 2005;128(4):2749–57.

You might also like