Tsest

Ha Gb Dc Scd Se D

International Journal of Nursing Studies 51 (2014) 892–899

A

Art

Re

Re

Ac

Ke

Co

Tra

Se

Sle

*

Te

00

htt

he effects of sleep on the relationship between brain injuryverity and recovery of cognitive function: A prospectiveudy

siao-Yean Chiu a, Wen-Cheng Lo b,c, Yung-Hsiao Chiang b,c, Pei-Shan Tsai a,d,e,*

raduate Institute of Nursing, College of Nursing, Taipei Medical University, Taipei, Taiwan

epartment of Neurosurgery, Taipei Medical University Hospital, Taipei, Taiwan

hool of Medicine, College of Medicine, Taipei Medical University, Taipei, Taiwan

leep Science Center, Taipei Medical University Hospital, Taipei, Taiwan

epartment of Nursing, Wan Fang Hospital, Taipei Medical University, Taipei, Taiwan

What is already known about the topic?

� Disturbed sleep is reported as a common symptom insubacute and chronic traumatic brain injury patients.� Sleep is related to maintenance of cognition functions.

R T I C L E I N F O

icle history:

ceived 17 May 2013

ceived in revised form 29 August 2013

cepted 19 October 2013

ywords:

gnitive function

umatic brain injury

verity of brain injury

ep

A B S T R A C T

Background: Disturbed sleep pattern is a common symptom after head trauma and its

prevalence in acute traumatic brain injury (TBI) is less discussed. Sleep has a profound

impact on cognitive function recovery and the mediating effect of disturbed sleep on

cognitive function recovery has not been examined after acute TBI.

Objectives: To identify the prevalence of disturbed sleep in mild, moderate, and severe

acute TBI patients, and to determine the mediating effects of sleep on the relationship

between brain injury severity and the recovery of cognitive function.

Design: A prospective study design.

Setting: Neurosurgical wards in a medical center in northern Taiwan.

Participants: Fifty-two acute TBI patients between the ages of 18 and 65 years who had

received a diagnosis of TBI for the first time, and were admitted to the neurosurgical ward.

Method: The severity of brain injury was initially determined using the Glasgow Coma

Scale. Each patient wore an actigraphy instrument on a non-paralytic or non-dominated

limb for 7 consecutive days. A 7-day sleep diary was used to facilitate data analysis.

Cognitive function was assessed on the first and seventh day after admission based on the

Rancho Los Amigos Levels of Cognitive Functioning.

Results: The mild (n = 35), moderate (n = 7) and severe (n = 10) TBI patients exhibited

poorer sleep efficiency, and longer total sleep time (TST) and waking time after sleep onset,

compared with the normative values for the sleep-related variables (P < .05 for all). The

severe and moderate TBI patients had longer daytime TST than the mild TBI patients

(P < .001), and the severe TBI patients had longer 24-h TST than the mild TBI patients

(P = .001). The relationship between the severity of brain injury and the recovery of

cognition function was mediated by daytime TST (t = �2.65, P = .004).

Conclusions: Poor sleep efficiency, prolonged periods of daytime sleep, and a high

prevalence of hypersomnia are common symptoms in acute TBI patients. The duration of

daytime sleep mediates the relationship between the severity of brain injury and the

recovery of cognition function.

� 2013 Elsevier Ltd. All rights reserved.

Corresponding author at: 250 Wu-Hsing St., Taipei 110, Taiwan.

l.: +886 2 27338813; fax: +886 2 23772842.

E-mail address: ptsai@tmu.edu.tw (P.-S. Tsai).

Contents lists available at ScienceDirect

International Journal of Nursing Studies

journal homepage: www.elsevier.com/ijns

20-7489/$ – see front matter � 2013 Elsevier Ltd. All rights reserved.

p://dx.doi.org/10.1016/j.ijnurstu.2013.10.020

�

�

1

trpvceimafaisthfudtoae(S

caHpthrofosArinArinc

pmrrhethain

2

2

a

H.-Y. Chiu et al. / International Journal of Nursing Studies 51 (2014) 892–899 893

What this paper adds

Poor sleep efficiency, hypersomnia, and prolonged day-time sleep duration occurred in acute TBI participants.

The relationship between severity of brain injury andcognition function recovery was mediated by daytimesleep duration.

. Introduction

Cognitive deficit is a common consequence afteraumatic brain injury (TBI). Deficits in attention, therocessing speed, learning and memory, verbal andisuospatial skills, and a range of executive functionsan occur within days to months following injury (Dikment al., 2009). Because cognitive function is related to

portant outcome indicators among TBI survivors, suchs time to return to work (Drake et al., 2000), exploring thectors associated with the recovery of cognitive function

of clinical importance. Previous studies have reportedat the severity of TBI predicts the initial level of cognitivenction (Lannoo et al., 2001; Novack et al., 2001). Other

emographic variables have also been shown to contribute poor cognitive outcome after TBI, including age (Chu et

l., 2007), male gender (Niemeier et al., 2007), lowducation level (Sherer et al., 2006), and minority statusherer et al., 2006).

Sleep plays a crucial role in the maintenance ofognitive functions, such as learning, memory formation,nd neural plasticity (Ron et al., 1980; Stickgold andobson, 2000; Wagner et al., 2004). A study of 9 TBIatients in a subacute rehabilitation unit demonstratedat improved sleep efficiency correlated with memory

eturn (Makley et al., 2009), suggesting that the recoveryf cognitive function is associated with sleep qualityllowing TBI. In addition, disturbed sleep is a common

ymptom of TBI, with prevalence rates of 50% (Mathias andlvaro, 2012). However, the role of disturbed sleep in the

ecovery of cognitive function during the immediate post-jury period in TBI patients has not been investigated.ddressing this gap in the knowledge base regarding

ecovery from brain injury could facilitate improvements the management of sleep-related symptoms and

ognitive function in acute TBI survivors.The objectives of our study were to compare the

revalence of sleep disturbance among patients with mild,oderate, and severe TBI, and to determine whether the

elationship between the severity of brain injury and theecovery of cognitive function is mediated by sleep. Weypothesized that TBI survivors with more severe injuriesxperience a greater degree of sleep disturbance, and thate association between the severity of the brain injury

nd the recovery of cognitive function is mediated by post-jury sleep.

. Materials and methods

.1. Participants

The present study was a secondary analysis of data from prospective study that investigated the trajectories of

changes in sleep patterns in acute-care patients with TBI(Chiu et al., 2013). Participants were consecutive acute TBIpatients who were admitted to the neurosurgical wards ofthe study site. Briefly, 52 TBI inpatients between the agesof 18 and 65 years who had been diagnosed with acutebrain injury for the first time based on neuroradiologicalassessments were enrolled in the study. Patients whopresented with other traumatic injuries and those with ahistory of previous TBI, psychiatric disorders, alcoholabuse, or sleep disturbance, such as obstructive sleepapnea, were excluded from our study. Patients who wereadmitted to the intensive care unit (ICU) immediately afterreceiving a diagnosis of TBI were also excluded.

2.2. Measurements

2.2.1. Sleep disturbances

Sleep disturbance was actigraphically assessed usingthe ActiGraph GT1M instrument (ActiGraph, Pensacola, FL,USA), which is a watch-like accelerometer that recordsmovement data. This method of sleep assessment isuniquely suited to study the sleep patterns of patientswith TBI (Zollman et al., 2010). During 60-s intervals, theaccelerometer recorded data at a rate of 30 Hz. Data wereanalyzed using the ActiLife software, Version 5.3.0(Actigraph). Automatic scoring of the actigraphical datawas performed for each testing interval using the Cole–Kripke algorithm (Sadeh et al., 1994) to determine theminute-by-minute asleep/awake status, which correlateshighly with polysomnography (Littner et al., 2003). Todetermine sleep onset and wake times, a 7-day sleep diarywas used to facilitate the actigraphic data analysis.

The sleep disturbances examined in our study includedinsomnia, excessive daytime sleepiness (EDS), and post-traumatic hypersomnia (PTH). We defined PTH as anincrease in sleep duration of more than 2 h per 24-h period,compared with sleep duration before traumatic braininjury (Baumann et al., 2007). The EDS was defined as anincreased propensity for daytime sleep with difficultystaying awake during the waking hours (Overeem andReading, 2010). Total sleep time (TST) was calculated forboth day and night periods for the estimation of EDS andPTH. The TST was the duration of actual sleep time during aperiod of sleep. Nighttime TST was the duration of actualsleep that occurred during the nighttime sleeping periodbased on the actigraphy data, and the daytime TST was theduration of actual sleep that occurred in a daytime sleepingperiod based on the actigraphy data. The TST in healthypeople (aged 18–65 years) is 375–450 min (Ohayon et al.,2004).

Insomnia is generally defined as difficulty initiating ormaintaining sleep, early awakening, poor sleep quality,insufficient sleep time, or insufficient restorative proper-ties of sleep (Overeem and Reading, 2010). For ourevaluation of insomnia, various sleep parameters wereestimated during nighttime sleeping periods, includingsleep efficiency (SE), the latency of sleep onset (SOL), andthe waking time after sleep onset (WASO). The SE was theratio of the TST to the total time spent in bed. The SOL wasdefined as the period from bedtime to the beginning ofsleep. The WASO quantified as the time spent awake after

fashpe50

2.2

inavbrasJesenoGCFieoure

2.2

(RopcoHausdeafancocothobmwtraThespaqubacasc

2.2

inlevmhetiohythpagrcoasstu(Corad

H.-Y. Chiu et al. / International Journal of Nursing Studies 51 (2014) 892–899894

lling into a persistent sleep state. A previous studyowed that the SE, the SOL, and the WASO for healthyople aged 18–65 years was 83–95%, 16–18 min, and 17–

min, respectively (Ohayon et al., 2004).

.2. Severity of brain injury

The severity of brain injury was assessed using theitial Glasgow Coma Scale (initial GCS), which is the bestailable predictor of mortality and cognitive outcome inain injury patients (Campbell et al., 2004). The GCSsessment provides a composite score (Teasdale andnnett, 1974), ranging from 3 to 15, with 3 indicatingvere neurological deficit (severe coma) and 15 indicating

neurological deficit. The reliability and accuracy of theS were demonstrated in a previous study (Rowley andlding, 1991), and it is widely used in clinical practice. Inr study, initial GCS score was obtained from medicalcords in the emergency department.

.3. Cognitive function recovery

The Rancho Los Amigos Levels of Cognitive FunctioningLA) is a comprehensive behavioral rating scale devel-ed by an interdisciplinary team of clinicians to assessgnitive functioning in adult TBI patients (Hagen, 1998;gen et al., 1979; Malkmus et al., 1980). The RLA scaleed in our study consisted of 10 hierarchical levels thatscribe the patterns or stages of recovery typically seen

ter a brain injury, with Level 1 representing no responsed Level 10 representing purposeful and appropriategnitive function. The RLA is suitable for TBI patients withgnitive and memory deficits because it does not requiree patient’s cooperation. Patients are classified based onservations of their responses to fluctuating environ-ental or purposeful stimuli. The RLA has been usedidely by clinicians to classify patients for treatment andck their progress throughout recovery (Dowling, 1985).e reliability and validity of the RLA scale have beentablished in both acute- and chronic-care settings for TBItients (Flannery, 1998; Gouvier et al., 1987). Weantified each patient’s recovery of cognitive functionsed on the improvement in their RLA score, which waslculated by subtracting the initial RLA score from the RLAore on the seventh day after admission.

.4. Sociodemographic factors

The sociodemographic factors examined in our studycluded age, body mass index (BMI), gender, educationel, cause of injury, comorbidity, and the use of

edications. The BMI was calculated based on the patient’sight and body weight. Information on comorbid condi-ns including diabetes mellitus, hypertension andperlipidemia was obtained from medical records. Ofe 52 participants, 11 had one comorbid condition. Therticipants were subsequently categorized into twooups based whether they had any one of the comorbidnditions. Confounding variables that were found to besociated with the poor cognitive outcome in previousdies, including age, male gender, low education level

hu et al., 2007; Niemeier et al., 2007; Sherer et al., 2006) were significantly different among the TBI groups, werejusted during the statistical analysis.

2.3. Procedure

Our study was approved by the Research Ethics ReviewCommittee (No. 98-3969B) of the study site. All partici-pants provided informed consent before their participationin the study. An initial GCS score of 13–15, 9–12, and �8indicates mild, moderate, and severe brain injury, respec-tively (Sternbach, 2000). Each participant had an actigra-phy watch secured to his or her wrist within 24 h ofadmission to the neurosurgical ward, and was monitoredfor 7 consecutive days. A 7-day sleep diary was completedby the patients with an RLA Level of 9 or above, or by theprimary caregivers of the patients with lower RLA scores.The primary caregivers provided 24-h/d care for the TBIpatients. The RLA score was determined on the first and theseventh day after admission. A trained research assistantprovided verbal instructions regarding the use of theactigraph and the sleep diary. Bedtime was defined as thetime at which the patient retired to the bed to sleep.Waking time was defined as the time at which the patientrose from the bed to start his or her day. The diary recordedbedtime, waking time, and the various times at which anynaps occurred (daytime sleep periods) during the 7-daystudy observation period. These times were used to guidethe visualization of the sleep times, and the ActiLifesoftware provided automated scoring of the TST, WASO, SE,and SOL data. The actigraphy output was analyzed andverified by the researcher.

2.4. Statistical analysis

All statistical analyses were performed using the SPSS,Version 17.0, computer software, and a result with P < .05was considered significant for the 2-tailed analyses. Thedescriptive analyses, the Fisher exact test, and theKruskal–Wallis Test were used to examine the differencesin demographic and clinical characteristics and the sleep-related data among the mild, moderate, and severe TBIpatient groups. While controlling for variables that weresignificant predictors of cognitive outcomes, such as age,male gender, and education level (Chu et al., 2007;Niemeier et al., 2007; Sherer et al., 2006), the Baron andKenny criteria (1986) were used to assess whether therelationship between the initial GCS score (independentvariable) and cognitive function recovery score (depen-dent variable) was mediated by the sleep parameters(mediators). Based on Baron and Kenny criteria, 4statistical conditions were required to establish theexistence of a mediating effect, and 3 conditions wererequired to establish a partially mediating effect. The 4statistical conditions were evaluated as follows: (1) aregression model was used to estimate whether the initialGCS score was significantly associated with the recoveryof cognitive function, (2) we determined whether theinitial GCS score and sleep parameters were significantlyassociated, (3) we determined whether the sleep param-eters were associated with the recovery of cognitivefunction when both the initial GCS score and the sleepparameters were treated as predictors in a multivariateregression analysis, and (4) we examined whether theeffect of the initial GCS score on the recovery of cognitive

fuptwfin2sGfu(

3

3

inFEB

F

h

H.-Y. Chiu et al. / International Journal of Nursing Studies 51 (2014) 892–899 895

nction was absent while controlling for the sleeparameters to establish that the relationship betweenhe initial GCS score and the recovery of cognitive functionas completely mediated by the sleep parameters. If therst 3 conditions were met, but the fourth condition wasot, a partially mediating effect was indicated (Kenny,012). The Sobel test (MacKinnon et al., 2002) was used totatistically evaluate whether the indirect effect of initialCS (independent variable) on the recovery of cognitivenction (dependent variable) through sleep parameters

mediator variable) was significant.

. Results

.1. Demographic and clinical information of participants

The selection of TBI patients and the flow of participants the study is depicted in Fig. 1. From March 2009 to

ebruary 2010, 60 patients were enrolled in our study.ight patients subsequently dropped out of the study.

participants were categorized into the 3 TBI groups asfollows: the mild TBI group (n = 35), the moderate TBIgroup (n = 7), and the severe TBI group (n = 10). All thepatients underwent a cranial computed tomography scanin the emergency department. None of the TBI patientsused sedatives during the observational period. With theexception of age and the RLA score on the first day and theseventh day (P < .05 for all), no significant differences inthe baseline characteristics were observed among the 3patient groups (Table 1).

3.2. Sleep disturbances among patients with mild, moderate

and severe TBI

As shown in Fig. 2, the mild, moderate, and severe TBIpatient groups exhibited poor SE (69.9%, 72.6%, and 69.8%,respectively), longer WASO (160.2, 188.6, and 168.0 min,respectively), longer daytime TST (255.7, 378.6, and477.6 min, respectively), and longer 24-h TST (641.69,758.95, and 879.86 min, respectively), compared with the

ig. 1. Diagram depicting the selection and flow of participants in the study. TBI, traumatic brain injury; SAH, subarachnoid hemorrhage; SDH, subdural

emorrhage; EDH, epidural hemorrhage; ICH, intracerebral hemorrhage; DAI, diffuse axonal injury; ICU, intensive care unit.

ormative data (P < .05 for all). With the exceptions of the

ased on the initial GCS score, the remaining 52 n

dasigththdathTB

3.3

co

mes

Ta

Dis

C

G

C

E

C

U

A

B

SDy

z

Fig

(W

Th

H.-Y. Chiu et al. / International Journal of Nursing Studies 51 (2014) 892–899896

ytime TST and the 24-h TST (P < .001 and P = .001), nonificant difference was observed in the nighttime TST,

e SE, the SOL, or the WASO. Post hoc analyses indicatedat both moderate and severe TBI patients had longerytime TST than the mild TBI patients (both P < .05), ande severe TBI patients had longer 24-h TST than the mildI patients (P = .002).

. Relationship between TBI severity, sleep parameters, and

gnitive function recovery

After controlling for the covariates, only daytime TSTet the first 3 conditions of the Baron and Kenny criteria totablish a partially mediating relationship (Table 2). The

regression model revealed that the initial GCS score wassignificantly associated with the recovery of cognitivefunction (B = �0.34; P < .0001). The initial GCS score andthe daytime TST were significantly correlated (B = �20.63;P < .0001), and the daytime TST was associated with therecovery of cognitive function (B = 0.006; P = .02) whenboth the initial GCS score and the daytime TST wereconsidered as predictors in the multivariate regressionmodel. However, the fourth condition was not met. Ourresults showed that the association between the initial GCSscore and the recovery of cognitive function was reducedwhen controlling for daytime TST, but it was not absent.The Sobel test showed that the indirect effect of the initialGCS score (independent variable) on the recovery of

ble 1

tribution of demographic and clinical characteristics of the study sample based on initial Glasgow Coma Scale (n = 52).

haracteristics Mild

(N = 35)

Moderate

(N = 7)

Severe

(N = 10)

P

ender, n (%)

Men 20 (57) 5 (71) 9 (90) .15y

Women 15 (43) 2 (29) 1 (10)

ause of injury, n (%)

Auto vehicle 26 (74) 6 (86) 9 (90) .55y

Fall 7 (20) 1 (14) 0 (0)

Hit by falling objects 2 (6) 0 (0) 1 (10)

ducational level

High school and below 22 (62.9) 6 (85.7) 7 (70.0) .49y

College and above 13 (37.1) 1 (14.3) 3 (30.0)

omorbidity, n (%)

Yes 3 (9) 3 (43) 2 (20) .07y

No 32 (91) 4 (57) 8 (80)

se of medication, n (%)

Analgesics 28 (80) 6 (85.7) 8 (80) .94y

Sedatives 0 (0) 0 (0) 0 (0) NS

ge, median (IQR) 40 (25–55) 21 (18–36) 27 (21–35) .04z

MI, median (IQR) 22.9 (19.2–25.1) 22.3 (19.5–24.3) 22.2 (20.7–27.9) .78z

, standard deviation; BMI, body mass index; NS, non-significant; IQR, interquartile range.

Calculated with chi-squared test and Fisher exact test.

Calculated with Kruskal–Wallis Test.

. 2. Distributions of sleep efficiency (SE), daytime total sleep time (TST), nighttime TST, 24-h TST, sleep onset latency (SOL), and waking after sleep onset

ASO) measurements derived from the actigraphy data for the mild, moderate, and severe traumatic brain injury patients during the observational period.

e SE as percentage (%). *P < .05 by post hoc test.

cins

4

obrp1pthim

ss(BeaedAhchpnaarathinthsnm

T

R

iG

e

H.-Y. Chiu et al. / International Journal of Nursing Studies 51 (2014) 892–899 897

ognitive function (dependent variable) was significantlyfluenced by the daytime TST (mediator variable; Sobel

tatistic = �2.65; P = .004).

. Discussion

Poor sleep efficiency and frequent awakening arebserved in TBI patients, regardless of their level of initialrain injury (Ohayon et al., 2004). Head trauma caused byapid acceleration and/or deceleration can disrupt sleepatterns, resulting in sleep disturbances (Graham et al.,988). Our findings concur with those of previous studiesatients with chronic TBI (Parcell et al., 2006), and foundat sleep disturbances can occur during the periodmediately following brain injury.Consistent with previous studies, our findings also

howed that more severe injuries led to greater daytimeleepiness and PTH (longer daytime and 24-h TST)

aumann et al., 2007; Guilleminault et al., 2000; Ponsfordt al., 2013). The prolonged sleep duration observedmong our TBI patients might have resulted from theffects of using medication for epilepsy or acute painuring hospitalization (Pandharipande and Ely, 2006).nother possible explanation is that damage to theypothalamus or reduced hypocretin signaling may haveontributed to prolonged sleep duration. Previous studiesave reported low or undetectable levels of the wake-romoting neurotransmitter hypocretin in the cerebrospi-al fluid of patients following head trauma (Baumann etl., 2007). Hypocretin deficiency is likely to reduce thectivation of the histaminergic system, which mighteduce the tonic inhibition of neurons secreting gamma-minobutyric acid (Huang et al., 2001; Jang et al., 2001),us causing severe sleepiness. It is therefore reasonable tofer that hypocretin deficiency could have contributed toe prolonged TST observed in our patients in the acute TBI

tage. Alternatively, brainstem injuries, such as damageear the midbrain tegmentum disrupting ascending

can cause sleepiness following TBI (Crompton, 1971).Therefore, multiple factors may contribute to sleepiness inacute TBI patients.

We found that the association between the initial GCSscore and the recovery of cognitive function was reducedwhen controlling for daytime TST. Thus, our resultsshowed that the daytime TST partially mediated the effectof the severity of brain injury on the recovery of cognitivefunction during the acute stage of TBI. These results mighthave been caused by reductions in hypocretin-mediatedneuron activity. Mounting evidence suggests that axonalsprouting is induced by neural lesions through a transientpattern of synchronous, low-frequency neuronal activitythroughout a network of cortical areas (Carmichael, 2003;Carmichael and Chesselet, 2002; Stickgold et al., 2001).Indirect evidence has also shown that the downregulationof hypocretinergic activity enhances plasticity after braininjury. Thus, longer duration of daytime sleep might exertbeneficial effects with regard to the recovery of cognitivefunction in TBI patients.

Certain limitations to our findings should be consid-ered. The small size of our sample limits the generalizationof our findings, and the limited observation period of 7days following TBI may restrict inferences of the impact ofsleep disturbances on cognitive recovery in TBI patients tothe acute phase. In addition, factors associated withhospitalization that may interfere with sleep, such asfrequent neurologic checks, sleeping in a well-lit room, orreceiving therapy during the day were not controlled, andthe data for certain predictors of sleep disturbancesidentified in previous studies, such as pain at night,anxiety, and depression, were not available for analysis inour study. Furthermore, because a ceiling effect wasobserved for the RLA score in the mild TBI group, theresults of our analysis should be interpreted with caution.

In conclusion, hypersomnia and longer daytime sleepduration are universal symptoms of TBI, especially insurvivors with severe TBI. The duration of daytime sleep

able 2

esults from analyses examining the mediating effect of daytime TST on relationship between initial GCS and cognitive function recovery (n = 52).

B SE b P 95% CI

Step 1

GCS predicts CFR

iGCS = �0.34

Age = �0.03

Education = 0.16

Male = 0.20

iGCS = 0.06

Age = 0.01

Education = 0.40

Male = 0.42

iGCS = �0.61

Age = �0.19

Education = 0.04

Male = 0.05

<.0001

.09

.70

.63

(�0.46 to �0.21)

(�0.05 to 0.04)

(�0.65 to 0.97)

(�1.05 to 0.64)

.

Step 2

GCS predicts D-TST

iGCS = �20.63

Age = �1.71

Education = 1.06

Male = 0.38

iGCS = 5.38

Age = 1.17

Education = 34.62

Male = 36.13

iGCS = �0.50

Age = �0.18

Education = 0.004

Male = 0.01

<.0001

.15

.98

.99

(�31.45 to �9.82)

(�4.05 to 0.63)

(�68.58 to 70.70)

(�73.03 to 72.30)

Step 3

D-TST predicts CFR

iGCS = �0.27

D-TST = 0.006

Age = �0.02

Education = 0.15

Male = 0.20

iGCS = 0.07

D-TST = 0.002

Age = 0.01

Education = 0.39

Male = 0.41

iGCS = �0.49

D-TST = 0.24

Age = �0.14

Education = 0.04

Male = 0.05

<.0001

.02

.20

.69

.62

(�0.36 to �0.05)

(0.001 to 0.007)

(�0.04 to 0.009)

(�0.63 to 0.94)

(�1.02 to 0.62)

Mediation (Sobel test) �2.65 .004

CS, initial GCS; CFR, cognitive function recovery; D-TST, daytime total sleep time; education, high school and below; B, nonstandardized beta; SE, standard

rror; b, standardized beta; CI, confidence interval.

as a mediating effect on the relationship between the

onoaminergic or cholinergic wake-promoting pathways, h

sefuththusanundaduinunTocodesuin

thdaWin

copu

au

Reth

Ac

26TafeNa

Re

Ba

Ba

Ca

Ca

Ca

Ch

Ch

Cro

H.-Y. Chiu et al. / International Journal of Nursing Studies 51 (2014) 892–899898

verity of brain injury and the recovery of cognitivenction in acute TBI patients. The underlying mechanismrough which the duration of daytime sleep influencese recovery of cognitive function could not be identifieding our methods. Future studies with larger sample sizesd control patients without head injury are warranted toderstand better the association between the duration ofytime sleep and the recovery of cognitive functionring the acute phase of TBI. In clinical settings, nursing

terventions should address the need to maximizeinterrupted periods of rest and sleep for TBI patents.

improve patients’ sleep duration, we suggest providingncentrated nursing care, limiting the number of visitors,creasing noise and light in the patients’ environment,pplying high-protein food, and encouraging short naps

the afternoon.Author contributions: P.S. Tsai and H.Y. Chiu provided

e idea and contributed to the study design, data analysis,ta interpretation, revision and manuscript preparation..C. Lo and Y.H. Chiang were involved in the dataterpretation, revision and manuscript preparation.

Conflict of interest: The authors declared no potentialnflicts of interest with respect to the authorship and/orblication of this article.Funding: No financial disclosures were reported by the

thors of this paper.Ethical approval: This study was approved by the

search Ethics Review Committee (No. 98-3969B) ofe study site.

knowledgements

P.S. Tsai was supported, in part, by a grant (NSC 102-28-B-038-004-MY3) from National Science Council,iwan. H.Y. Chiu was supported by the postdoctoralllowship funding (NSC 102-2811-B-038-028) from thetional Science Council, Taiwan.

ferences

ron, R.M., Kenny, D.A., 1986. The moderator–mediator variable distinc-tion in social psychological research: conceptual, strategic, and sta-tistical considerations. Journal of Personality and Social Psychology51 (6) 1173–1182.

umann, C.R., Werth, E., Stocker, R., Ludwig, S., Bassetti, C.L., 2007. Sleep-wake disturbances 6 months after traumatic brain injury: a prospec-tive study. Brain 130 (Pt 7) 1873–1883.

mpbell, C.G., Kuehn, S.M., Richards, P.M., Ventureyra, E., Hutchison, J.S.,2004. Medical and cognitive outcome in children with traumaticbrain injury. The Canadian Journal of Neurological Sciences 31 (2)213–219.

rmichael, S.T., 2003. Plasticity of cortical projections after stroke.Neuroscientist 9 (1) 64–75.

rmichael, S.T., Chesselet, M.F., 2002. Synchronous neuronal activity is asignal for axonal sprouting after cortical lesions in the adult. Journal ofNeuroscience 22 (14) 6062–6070.

iu, H.Y., Chen, P.Y., Chen, N.H., Chuang, L.P., Tsai, P.S., 2013. Trajectoriesof sleep changes during the acute phase of mild traumatic braininjury: a 7-day actigraphy study. Journal of the Formosan MedicalAssociation 112 (9) 545–553.

u, B.C., Millis, S., Arango-Lasprilla, J.C., Hanks, R., Novack, T., Hart, T.,2007. Measuring recovery in new learning and memory followingtraumatic brain injury: a mixed-effects modeling approach. Journal ofClinical and Experimental Neuropsychology 29 (6) 617–625.

Dikmen, S.S., Corrigan, J.D., Levin, H.S., Machamer, J., Stiers, W., Weisskopf,M.G., 2009. Cognitive outcome following traumatic brain injury. TheJournal of Head Trauma Rehabilitation 24 (6) 430–438.

Dowling, G.A., 1985. Levels of cognitive functioning: evaluation of inter-rater reliability. Journal of Neurosurgicl Nursing 17 (2) 129–134.

Drake, A.I., Gray, N., Yoder, S., Pramuka, M., Llewellyn, M., 2000. Factorspredicting return to work following mild traumatic brain injury: adiscriminant analysis. The Journal of Head Trauma Rehabilitation 15(5) 1103–1112.

Flannery, J., 1998. Using the levels of cognitive functioning assessmentscale with patients with traumatic brain injury in an acute caresetting. Rehabilitation Nursing 23 (2) 88–94.

Gouvier, W.D., Blanton, P.D., LaPorte, K.K., Nepomuceno, C., 1987. Reli-ability and validity of the disability rating scale and the levels ofcognitive functioning scale in monitoring recovery from severe headinjury. Archives of Physical and Medicine Rehabilitation 68 (2)94–97.

Graham, D.I., Lawrence, A.E., Adams, J.H., Doyle, D., McLellan, D.R., 1988.Brain damage in fatal non-missile head injury without high intracra-nial pressure. Journal of Clinical Pathology 41 (1) 34–37.

Guilleminault, C., Yuen, K.M., Gulevich, M.G., Karadeniz, D., Leger, D.,Philip, P., 2000. Hypersomnia after head-neck trauma: a medicolegaldilemma. Neurology 54 (3) 653–659.

Hagen, C. (Ed.), 1998. Levels of Cognitive Functioning. Rehabilitation ofthe Head Injuried Adult: Comprehensive Physical Management. thirded. Professional Staff Association of the Rancho Los Amigos Hospital,Inc., Dowey, CA.

Hagen, C., Malkmus, D., Durham, P. (Eds.), 1979. Levels of CognitiveFunctioning. Rehabilitation of the Head Injuried Adult: Comprehen-sive Physical Management. Professional Staff Association of the Ran-cho Los Amigos Hospital, Inc., Dowey, CA.

Huang, Z.L., Qu, W.M., Li, W.D., Mochizuki, T., Eguchi, N., Watanabe, T.,Hayaishi, O., 2001. Arousal effect of orexin A depends on activation ofthe histaminergic system. Proceedings of the National Academy ofSciences of the United States of America 98 (17) 9965–9970.

Jang, I.S., Rhee, J.S., Watanabe, T., Akaike, N., 2001. Histaminergic modu-lation of GABAergic transmission in rat ventromedial hypothalamicneurones. Journal of Physiology 534 (Pt 3) 791L 803.

Kenny, D.A., 2012. Mediation. Retrieved April 8, 2013. From http://davi-dakenny.net/cm/mediate.htm#BK.

Lannoo, E., Colardyn, F., Jannes, C., de Soete, G., 2001. Course of neuro-psychological recovery from moderate-to-severe head injury: a 2-year follow-up. Brain Injury 15 (1) 1–13.

Littner, M., Kushida, C.A., Anderson, W.M., Bailey, D., Berry, R.B., Davila,D.G., Johnson, S.F., 2003. Practice parameters for the role of actigraphyin the study of sleep and circadian rhythms: an update for 2002. Sleep26 (3) 337–341.

MacKinnon, D.P., Lockwood, C.M., Hoffman, J.M., West, S.G., Sheets, V.,2002. A comparison of methods to test mediation and other inter-vening variable effects. Psychological Methods 7 (1) 83–104.

Makley, M.J., Johnson-Greene, L., Tarwater, P.M., Kreuz, A.J., Spiro, J., Rao,V., Celnik, P.A., 2009. Return of memory and sleep efficiency followingmoderate to severe closed head injury. Neurorehabilitation andNeuralrepair 23 (4) 320–326.

Malkmus, D., Booth, B., Kodimer, C. (Eds.), 1980. Rehabilitation of theHead Injuried Adult: Comprehansive Cognitive Management. Profes-sional Staff Association of the Rancho Los Amigos Hospital Inc.,Downey, CA.

Mathias, J.L., Alvaro, P.K., 2012. Prevalence of sleep disturbances, disor-ders, and problems following traumatic brain injury: a meta-analysis.Sleep Medicine 13 (7) 898–905.

Niemeier, J.P., Marwitz, J.H., Lesher, K., Walker, W.C., Bushnik, T., 2007.Gender differences in executive functions following traumatic braininjury. Neuropsychological Rehabilitation 17 (3) 293–313.

Novack, T.A., Bush, B.A., Meythaler, J.M., Canupp, K., 2001. Outcome aftertraumatic brain injury: pathway analysis of contributions from pre-morbid, injury severity, and recovery variables. Archives of Physicaland Medicine Rehabilitation 82 (3) 300–305.

Ohayon, M.M., Carskadon, M.A., Guilleminault, C., Vitiello, M.V., 2004.Meta-analysis of quantitative sleep parameters from childhood to oldage in healthy individuals: developing normative sleep values acrossthe human lifespan. Sleep 27 (7) 1255–1273.

Overeem, S., Reading, P., 2010. Sleep Disorder in Neurology: A PracticalApproach. Wiley-Blackwell, West Susgender.

Pandharipande, P., Ely, E.W., 2006. Sedative and analgesic medications:risk factors for delirium and sleep disturbances in the critically ill.Critical Care Clinics 22 (2) 313–327.

Parcell, D.L., Ponsford, J.L., Rajaratnam, S.M., Redman, J.R., 2006. Self-

reported changes to nighttime sleep after traumatic brain injury.Archives of Physical and Medicine Rehabilitation 87 (2) 278–285.

mpton, M.R., 1971. Hypothalamic lesions following closed head injury.Brain 94 (1) 165–172.

P

R

R

S

S

H.-Y. Chiu et al. / International Journal of Nursing Studies 51 (2014) 892–899 899

onsford, J.L., Parcell, D.L., Sinclair, K.L., Roper, M., Rajaratnam, S.M.,2013. Changes in sleep patterns following traumatic brain injury: acontrolled study. Neurorehabilitation and Neural Repair 27 (7)613–621.

on, S., Algom, D., Hary, D., Cohen, M., 1980. Time-related changes in thedistribution of sleep stages in brain injured patients. Electroenceph-alographic Clinical Neurophysiology 48 (4) 432–441.

owley, G., Fielding, K., 1991. Reliability and accuracy of the GlasgowComa Scale with experienced and inexperienced users. Lancet 337(8740) 535–538.

adeh, A., Sharkey, K.M., Carskadon, M.A., 1994. Activity-based sleep-wake identification: an empirical test of methodological issues. Sleep17 (3) 201–207.

herer, M., Stouter, J., Hart, T., Nakase-Richardson, R., Olivier, J., Manning,E., Yablon, S.A., 2006. Computed tomography findings and early

cognitive outcome after traumatic brain injury. Brain Injury 20(10) 997–1005.

Sternbach, G.L., 2000. The Glasgow coma scale. The Journal of EmergencyMedicine 19 (1) 67–71.

Stickgold, R., Hobson, J.A., Fosse, R., Fosse, M., 2001. Sleep, learning, anddreams: off-line memory reprocessing. Science 294 (5544) 1052–1057.

Stickgold, R.J.L., Hobson, J.A., 2000. Visual discrimination learningrequires sleep after training. Nature Neuroscience 3, 1237–1238.

Teasdale, G., Jennett, B., 1974. Assessment of coma and impaired con-sciousness. A practical scale. Lancet 13 (7872) 81–84.

Wagner, U., Gais, S., Haider, H., Verleger, R., Born, J., 2004. Sleep inspiresinsight. Nature 427 (6972) 352–355.

Zollman, F.S., Cyborski, C., Duraski, S.A., 2010. Actigraphy for assessmentof sleep in traumatic brain injury: case series, review of the literatureand proposed criteria for use. Brain Injury 24 (5) 748–754.