Fatigue after TBI: Association with neuroendocrine abnormalities

Brain Injury, November 2010; 24(12): 1379–1388

ORIGINAL ARTICLE

Fatigue after traumatic brain injury: Association withneuroendocrine, sleep, depression and other factors

JEFFREY ENGLANDER1, TAMARA BUSHNIK2, JEAN OGGINS3, &LAURENCE KATZNELSON4

1PM&R, 2Rehabilitation Research, Santa Clara Valley Medical Center, San Jose, CA, USA, 3Consulting and

Research/Evaluation Services, San Francisco, CA, USA, and 4Departments of Medicine and Neurosurgery, Stanford

University School of Medicine, Stanford, CA, USA

(Received 12 January 2010; revised 13 August 2010; accepted 8 September 2010)

AbstractObjective: Define associations between post-traumatic brain injury (TBI) fatigue and abnormalities in neuroendocrine axes,sleep, mood, cognition and physical functioning.Design: Survey.Setting: Large community hospital-based rehabilitation centre.Participants: Convenience sample of 119 individuals at least 1 year post-TBI.Outcome measures: Multidimensional Assessment of Fatigue (MAF); Fatigue Severity Scale (FSS); neuroendocrineassessments–growth hormone (GH) reserve, thyroid, cortisol and testosterone levels; visual analogue pain rating; PittsburghSleep Quality Index; Beck Depression Inventory–II; Disability Rating Scale; Craig Handicap Assessment and ReportingTechnique; Neurobehavioural Functioning Inventory.Results: Fifty-three per cent reported fatigue on the MAF and one-third on the FSS; 65% were found to have moderate/severe GH deficiency; 64% had adrenal insufficiency (low fasting cortisol); 12% had central hypothyroidism; and 15% ofmen had testosterone deficiency. Pituitary dysfunction did not correlate with fatigue or other symptoms. Predictors of MAFtotal scores were female gender, depression, pain and self-assessed memory deficits. Predictors of FSS scores weredepression, self-assessed motor deficits and anti-depressant usage.Conclusions: Robust correlates of fatigue were gender, depression, pain and memory and motor dysfunction. Investigation ofpost-TBI fatigue should include screening for depression, pain and sleep disturbance. There was no correlation betweenpituitary dysfunction and fatigue; however, the relatively high prevalence of hypothyroidism and adrenal dysfunctionsuggests screening for these hormone deficiencies.

Keywords: Brain injuries, rehabilitation, fatigue, neuroendocrine systems

Introduction

Although fatigue is a common experience in the

general population it is particularly bothersome to

those with traumatic brain injury (TBI) [1, 2].

Despite the fact that fatigue is such a common

phenomenon, it has been challenging to measure in a

quantitative manner [3, 4]. Depending on the scales

used to measure fatigue and time since injury,

its prevalence has been reported in 16–80% ofindividuals after injury [5, 6]. Associated symptomsinclude depression, sleep disturbance, pain,cognitive and motor disturbances [7].

Neuroendocrine abnormalities as manifestedby pituitary dysfunction can occur after TBI.One-to-several hypothalamic-pituitary axes may beaffected; posterior pituitary dysfunction typicallyresolves during the first several months post-injury,

Correspondence: Tamara Bushnik, PhD, FACRM, Director of Research, Rusk Institute for Rehabilitation Research/NYU Langone Medical Center, 400 East34th Street, RR115A, New York, NY 10010, USA. Tel: (212) 263-6547. Fax: (212) 263-2575. E-mail: [email protected]

ISSN 0269–9052 print/ISSN 1362–301X online � 2010 Informa UK Ltd.DOI: 10.3109/02699052.2010.523041

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

18/1

0Fo

r pe

rson

al u

se o

nly.

whereas anterior pituitary problems are more likelyto persist [8–11]. The correlation of fatigue withneuroendocrine abnormalities has been less thanrobust. Using the SF-36 Health Survey, Kelly et al.[12] did find associations between growth hormonedeficiency and the domains of physical health,energy and fatigue, emotional well-being and generalhealth after TBI. However, in an interim dataanalysis of the current study which looked only atthe association between neuroendocrine factors andfatigue in �50% of the participants, neuroendocrinedeficiencies alone were not found to be associatedwith fatigue severity as measured by the GlobalFatigue Index and Fatigue Severity Scale [13].

Given the challenges with measuring thedimensions of fatigue, the array of neuroendocrineabnormalities and additional contributing factors,the authors embarked on this exploratory study toattempt to better define those associations in acommunity sample of individuals with TBI. Theassessment measures were selected to representspecific symptoms or domains that are known tobe affected after TBI; for example, the BeckDepression Inventory–II for depression and theChart Handicap Assessment and ReportingTechnique (CHART) social integration sub-scaleto reflect participation in the social domain.In addition, all selected measures are in commonuse in the population with TBI. The hypotheseswere: (1) Neuroendocrine abnormalities would beassociated with fatigue in this larger sample; and (2)Sleep quality, depression, pain, memory and cogni-tive functioning would be independently associatedwith fatigue.

Methods

Participants

Individuals with TBI of at least 1 year duration wererecruited from the community through mailing listsand advertisements in hospital clinics and localcommunity events. The recruitment flyersasked for people with TBI who were interested inparticipating in a study of fatigue or tiredness afterTBI. Participants were excluded if they reported anyof the following conditions that can cause fatigueindependent of the TBI: cardiac/pulmonary disease,diabetes mellitus, rheumatoid arthritis, multiplesclerosis, post-polio, chronic fatigue syndrome,cancer, renal failure, anaemia, current pregnancyand known ongoing endocrine abnormalities. Thisstudy was approved by the Institutional ReviewBoard at Santa Clara Valley Medical Center andall participants were capable of giving informedconsent.

Procedure

All study procedures were conducted at Santa ClaraValley Medical Center beginning between 8–10 a.m.Participants were required to fast from midnightthe night before with only water and morningmedications permitted. The protocol,including completion of blood tests and studyquestionnaires, required �4 hours. All question-naires were self-administered except forthe Disability Rating Scale which was assessed bythe research assistant through interview. Basicdemographic information was obtained, includingage, marital status, employment status, dateand aetiology of injury, self-reported length ofunconsciousness following injury and height andweight to calculate Body Mass Index (BMI). Resultswere shared with participants; those with abnormalblood tests and depression scores were advised toconsult with their healthcare practitioners.

Assessments

Blood tests. After an overnight fast, patientsunderwent laboratory testing including electrolytes,complete blood count (cbc) and neuroendocrineevaluation. An initial blood draw was processed forlevels of glucose, cortisol, free thyroid (T4), thyroidstimulating hormone (TSH), insulin growth factor–1(IGF-1) and a complete blood count series and basalgrowth hormone. Central hypothyroidism wasdefined as low serum-free T4 with low or low-normal serum TSH. Male testosterone deficiencywas defined as a serum testosterone below thediagnostic laboratory age-corrected level. Adrenalinsufficiency was defined as a fasting serum cortisolless than 15 mcg dl�1. Although provocative testingof adrenal function may be more accurate fordefining adrenal reserve, such as with a cortrosynstimulation test, basal cortisol levels have beenshown to correlate with stimulated values aswell [14].

Growth hormone reserve was assessed by aglucagon stimulation test, a provocative test forGH reserve that has been used in other studies ofneuroendocrine function after brain injury [15].An insulin tolerance test was not performed inthese patients with history of TBI because of theconcern of hypoglycaemia-induced seizures [16].GH was measured at 0, 90, 120, 150 and 180 min-utes following an intramuscular injection of gluca-gon, 0.03 mg kg�1, with a maximum dose of 1 mg.Severe GH deficiency (GHD) was defined as a peakGH less than 3 ng ml�1; moderate GHD as peakGH level between 3–9.9 ng ml�1; normal GHreserve as peak GH greater than or equal to10 ng ml�1 [11, 15]. Serum IGF-1 levels were alsoused as a marker of GH sufficiency, although IGF-1

1380 J. Englander et al.

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

18/1

0Fo

r pe

rson

al u

se o

nly.

levels in GHD individuals overlap with normal [17].Sub-normal IGF-1 values were based uponlaboratory standard values that were controlled forage and gender.

Multidimensional assessment of fatigue (MAF). TheMAF contains five fatigue sub-scales: distress,degree, severity, impact on ADLs and timing, aswell as a total score called the Global Fatigue Index(GFI) [18]. It has good concurrent and divergentvalidity [19], as well as demonstrated reliability ofscores over an 8-week period [18]. The GFIhas been used with individuals with rheumatoidarthritis [19], cancer [20], HIV [21] and multiplesclerosis [22]. Higher average GFI scores have beenreported for individuals with rheumatoid arthritis,28� 10, compared to controls, 16�11 [19].Increases in mean GFI scores were observed inindividuals with HIV after initiation of interleukin-2treatment, GFI¼ 27, compared to before treatment,GFI¼ 12 [21]. A score of 27 and above was usedas the cut-off for the presence of fatigue.

Fatigue severity scale (FSS). The FSS contains nineitems that assess self-reported severity offatigue related to physical functioning, exercise andsocializing [23]. The items are rated on a 7-pointLikert scale from a score of 1 (strongly disagree)to 7 (strongly agree). Internal consistency andtest–re-test reliability are good [24]. It has beenused in multiple populations: lupus and multiplesclerosis [23]; chronic fatigue syndrome [25]; andTBI [2]. A conservative score of 5.5 and above wasused as the cut-off for the presence of fatigue.

Drug/alcohol use. Alcohol use was assessed withfour questions used by the TBI Model Systems(based on the National Household Survey on DrugAbuse [26] and the Behavioural RiskFactor Surveillance Survey [27]) resulting in fourcategories: abstaining; infrequent/light; moderate;and heavy, as well as a dichotomous variableindicating binge drinking (five or more standarddrinks on one occasion). Participants were askedabout any use of illegal or non-prescription drugs inthe past year; answers were coded as ‘yes’ or ‘no’.

Pain. The level of pain over the past monthwas assessed using a visual analogue scale rangingfrom 1 (no pain at all) to 10 (most severe pain).

Pittsburgh sleep quality index (PSQI). The PSQIis designed to measure the quality and patterns ofsleep [28]. It differentiates poor from good sleep by

measuring seven areas: subjective sleep quality; sleeplatency; sleep duration; habitual sleep efficiency;sleep disturbances; use of sleeping medication; anddaytime dysfunction over the past month. Theindividual rates these areas on a scale of 0 (notduring the past month) to 3 (three or more times aweek). A global sum of 5 or greater indicates a ‘poor’sleeper. The PSQI has internal consistency anda Cronbach’s alpha value of 0.83 [28]. The PSQIhas concurrent validity with sleep diary data [28].The PSQI has been used as a screening tool forinsomnia among individuals with TBI [29].

Beck Depression Inventory–II (BDI-II). The BDI-IIis a 21-item self-report instrument widely used todetermine depressive symptomatology [30].Each item is rated on a scale of 0 (no symptomatol-ogy) to 3 (excessive symptomatology). Item scoresare totalled and range between 0–63. The followingcategories can be created: normal (scores 5–13);mild depression (14–19); moderate depression(20–28); and severe depression (29–63). Internalconsistency, concurrent validity and sensitivity tochange have been demonstrated [30, 31].

Disability Rating Scale (DRS). The DRS wasdeveloped for use primarily with persons with TBIand with the continuum of recovery in mind [32].It consists of eight items that assess four categories:arousal; cognitive ability to handle self-carefunctions; dependence upon others for cognitive orphysical needs/reasons; and psychosocial adaptabil-ity for work, housework and school. The scoreranges from 0, denoting no disability, impairmentor handicap, to 29, signifying death. Inter-raterreliability, test–re-test reliability, concurrent validityand predictive validity have all been well established[32–34].

Craig Handicap Assessment and Reporting Technique

(CHART). The CHART was designed to providea simple, objective measure of the degree to whichimpairments and disabilities result in handicaps inthe years after initial rehabilitation [35]. The originalCHART included domains to assess five of theWHO dimensions of handicap and has been revisedto add an assessment of cognitive independence foruse in individuals with cognitive impairments;the revised scale has demonstrated utility andreliability [36]. Sub-scale scores range from 0–100with the maximum score corresponding to theexpected score of a person without a disability. Forthis study, the Cognitive Independence, SocialIntegration and Occupation sub-scales were used.

Factors associated with post-TBI fatigue 1381

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

18/1

0Fo

r pe

rson

al u

se o

nly.

Neurobehavioural Functioning Inventory (NFI). TheNFI is comprised of 70 items containing brief descrip-tions of neurobehavioural problems or symptoms, aswell as six ‘critical’ items which can be grouped into sixfunctional domain categories [37]. Individuals rate thefrequency of occurrence of each problem or symptomon a five-point scale: (1) never; (2) rarely; (3)sometimes; (4) often; or (5) always. Responses onitems within a scale may be summed to obtain sub-scalescores and converted to T-scores based on age andduration of post-TBI unconsciousness. For this study,the somatic, memory/attention and motor sub-scaleswere used.

Statistical analysis

Several measures were re-coded for use in analyses.Alcohol use was re-coded into three categories:none, light and moderate/heavy. Employmentstatus was re-coded into a three-category measureof Productivity: unemployed; engaged in productiveactivity but not employed (e.g. student, volunteer,homemaker, retired); and employed. The PSQI wasre-coded to indicate abnormal sleep (>4) comparedto normal sleep (0–4). The BDI II was re-coded toindicate normal (0–9), mild (10–18) or moderate–severe depression (19 and over). Abnormal endo-crine levels were coded: GH response (<10ngml�1

(abnormal)¼1, �10ngml�1 (normal)¼0), Cortisol(<15mcgdl�1 (abnormal)¼ 1,� 15¼ 0); IGF-1 (lownormed group (abnormal)¼ 1 normal¼ 0); T4 levels(<0.70 (abnormal)¼ 1,� 0.70¼ 0); and Testosterone(<260ngdl�1 (abnormal)¼ 1,� 260ngdl�1

¼ 0).Frequencies were run for demographic items;

measures of brain injury (time since injury, self-reported length of loss of consciousness); measuresof physical health (e.g. body mass index, pain,substance use, percentage taking medications,abnormal sleep and endocrine measures); DRS,CHART and NFI measures of cognitive, physical,social and occupational skills; BDI II Depressionand fatigue scales. Time in a coma and timesince brain injury were correlated with abnormalendocrine measures using Spearman’s two-tailedcorrelations and were correlated with CHART andDRS scores using Pearson’s two-tailed correlations.Female gender, depression, pain and reports of sleepproblems were also correlated with each other usingtwo-tailed Pearson’s correlations. To test for anynon-linear relation of duration of unconsciousnessand endocrine abnormalities, cross-tabulations withchi-squares were conducted.

Each fatigue sub-scale or total scale was alsocorrelated with independent variables, includingdemographic variables, endocrine levels, pain, sleepquality, NFI measures, depression, substance use,DRS and CHART occupational, social and

cognitive scores. Two-tailed Pearson’s correlationswere run for all items. Multiple regression analyseswere run using each fatigue sub-scale or scale as thedependent variable. Items that had independentlycorrelated at p< 0.05 with fatigue measures wereentered together into stepwise multiple regressionspredicting the fatigue measures, using a p<0.05criterion to include variables in the final model.Then, to increase the sample size for regressions,regressions were run again (but not in stepwisefashion), using only variables that had been signifi-cant in the previous regressions. For GFI and FSSdichotomous abnormal fatigue measures, the sameprocedure was conducted using logistic regressions.Data are presented as mean� standard deviationunless otherwise noted. A low alpha level of 0.01was selected due to the number of analyses thatwere conducted and the exploratory nature ofthis study.

Results

There were 119 subjects, including 80 (66%) males.A total of 15 potential participants were excludeddue to time constraints or unwillingness to undergothe blood tests. The mean age was 40� 12 years(range¼ 16–78 years). Patients were studied anaverage of 9� 7.6 years since injury. Over 60% ofparticipants had been injured due to a vehicleaccident (see Table I).

Over 50% reported loss of consciousness of morethan 1 week. One-half of participants were currentlyemployed; another 13% were engaged in productiveactivity, such as volunteer work or education;and 37% were unemployed. A little over one-quarterwere married. About one-fifth reported moderate-to-heavy alcohol use or illegal drug use sincethe TBI. This sample contained a relatively evendistribution of severities of injury between mild,moderate and severe TBI, which is not a typical TBIpopulation distribution that tends towards moremild cases.

Scores on the assessment measures for the entiresample are reported in Table II. With regard to theDRS, only one person (with a score of 17) had ascore in the moderate-to-severe range; this outlierscore was removed from subsequent analyses usingthe DRS. There was some variability in the numberof responders to each questionnaire; some responseswere incomplete and could not be scored. In general,CHART social and cognitive independence scoreswere higher than occupational scores. NFI scores onmemory, somatic and motor skills were similar. Overtwo-thirds of participants noted abnormal sleep onthe PSQI. The average level of pain intensity was 4on a 1–10 visual analogue scale. One-third reportedmoderate-to-severe depression. Thirty-eight per cent

1382 J. Englander et al.

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

18/1

0Fo

r pe

rson

al u

se o

nly.

of the participants were not taking medications;about one-quarter reported taking at least one of thefollowing: anti-depressants (28%), pain medications(26%) or other neurologic medications (24%).Less than 10% were taking either anxiolytics orstimulants; about one-third were taking other typesof medications.

Table II also shows frequencies for the fatiguemeasures. More than half (53%) of the participantsscored in the abnormal range of the GFI (>27)compared to one-third scoring abnormal on the FSS(>5.5).

Based on response to glucagon stimulation, 34%(n¼ 39) of participants had severe GH deficiencyand 31% (n¼ 36) had moderate GH deficiency.Therefore, 65% of subjects had moderate-to-severeGHD. Only 19% (n¼ 22) had GHD defined bya sub-normal serum IGF-1 value. The correspon-dence between low IGF-1 and either moderate orsevere GHD was poor; only 21% (n¼ 8) had bothsevere GHD as indicated by glucagon stimulationand low IGF-1 and only 27% (n¼10) had bothmoderate GHD as indicated by glucagon stimulationand low IGF-1. Low serum cortisol as defined by afasting serum cortisol< 15 mcg dl�1 was noted in64% of subjects: none had values in the severe rangeof <5 mcg dl�1. Twelve per cent had central hypo-thyroidism. Testosterone deficiency was present in15% of men. Although menstrual history was

obtained in females, results were obscured by thevariable age of menopause onset.

Correlation of endocrine factors with time since injury

and duration of unconsciousness

Table III shows results from correlations of abnor-mal endocrine levels with time since injuryand duration of unconsciousness. There were nosignificant correlations between time since injury,duration of unconsciousness and endocrineabnormalities.

Correlation between neuroendocrine function, fatigue and

other symptoms

There were no significant correlations betweenany measure of abnormal endocrine function andfatigue, NFI, CHART or DRS scores. Additionally,NFI memory, somatic and motor sub-scale scorescorrelated positively with all fatigue scales. MAF and

Table II. Means and frequencies for assessment measures.

n M (SD) or % Range

Disability Rating Scale(DRS)

117 2.42 (1.99) 0–17

Craig Handicap Assessment and Reporting Technique (CHART)Social 111 82.42 (23.01) 20–100Cognitive Independence 118 76.39 (20.14) 20–100Occupation 114 62.46 (31.31) 3–100

Neurobehavioral Functioning Inventory (NFI)Memory 118 52.32 (9.97) 29–89Somatic 118 51.05 (10.23) 31–80Motor 117 49.21 (10.22) 26–74

Pittsburgh Sleep QualityIndex

108 7.61 (4.50) 0–20

Abnormal sleep (>4) 73 68%Pain intensity (1–10) 118 4.11 (2.84) 1–10Beck Depression Inventory

II (BDI-II)110 15.98 (10.87) 0–50

Normal (0–9) 33 30%Mild (10–18) 41 37%Moderate/severe (20–50) 36 33%

MedicationsNo medications 45 38%Anti-depressants 33 28%Pain medication 31 26%Other neurologics 29 24%Anxiolytics 9 8%Stimulant medication 4 3%Other medications 39 33%

Multidimensional Assessment of Fatigue (MAF)sub-scalesSeverity 118 10.76 (5.61) 2–20Distress 118 4.81 (3.34) 1–10ADLs 119 3.83 (2.49) 1–9.7Timing 119 2.53 (.81) 1–4GFI 118 26.12 (12.39) 6.5–49.7

Fatigue Severity Scale(FSS)

119 4.42 (1.85) 1–7

Table I. Frequencies for demographic and brain injurycharacteristics.

n %

Aetiology of traumatic brain injuryVehicle accident 71 63Fall 15 13Violence 12 11Other cause 12 11Sports injury 2 2

Duration of unconsciousnessLess than 1 day 30 261–7 days 24 21

8–14 days 17 15More than 14 days 44 38

ProductivityEmployed 59 50Unemployed 44 37Productive activity, not working 16 13

Marital statusSingle 53 45Divorced/separated/widowed 34 29Married 32 27

Alcohol useNone 56 48Light 37 31Moderate/heavy 25 21

Any illegal drug use 21 19Body-mass index (BMI) (M, SD) 119 27.64 (6.41)

Factors associated with post-TBI fatigue 1383

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

18/1

0Fo

r pe

rson

al u

se o

nly.

FSS scores correlated negatively with the CHARTCognitive score except for MAF ADLs and FSStotal score. Further, on all measures, greater fatiguewas reported by women and by both men andwomen, who reported more pain, sleep problems ordepression. Pain intensity correlated positively withdepression (r¼ 0.40, p< 0.0001) and sleep problems(r¼ 0.38, p< 0.0001). Depression and sleep prob-lems also correlated at r¼0.49, p< 0.0001. Womenwere especially likely to report pain, r¼ 0.31,p< 0.001, and depression, r¼ 0.29, p<0.002.

Participants taking anti-depressants scored higheron all fatigue scales except MAF ADLs and Timingand GFI Abnormal Fatigue. Those on pain medica-tions scored higher on MAF Distress and GFI scales.The variables that were not associated with any ofthe fatigue sub-scales or scores were CHART Socialand Occupational scores, DRS, BMI, age, engagingin productive activity (yes/no), severity of alcohol

use, any illegal drug use or taking stimulants,anxiolytic medications, other neurologics or othermedications.

Next multiple regressions were run predictingMAF sub-scales and GFI and FSS scores fromvariables that had correlated significantly with thesescales in stepwise regressions. Findings are shown inTable IV. MAF Severity was positively associatedwith reported frequency of memory deficits andpain. The MAF ADL fatigue sub-scale significantlycorrelated with female gender, depression and lowerCHART social scores. Higher MAF Distress scoreswere associated with reports of greater depressionand pain. Reports of more frequent fatigue on theMAF Timing scale were positively associated withfemale gender, depression and self-assessment ofmore frequent memory deficits. The GFI score wasindependently associated with depression, painintensity and self-assessment of more frequentmemory deficits. The FSS correlated positivelywith reports of depression and NFI motor deficits.

In a logistic regression predicting whether anindividual would score in the ‘fatigued’ level on theGFI, based on findings from stepwise regressions,significant predictors were BDI Depression(r¼ 0.23, p<0.002), Pain intensity (r¼ 0.20,p< 0.006) and NFI motor deficits (r¼ 0.19,p< 0.007), �2(3)¼35.62, p< 0.0001. In a logisticregression predicting FSS abnormal fatigue based onfindings from stepwise regressions, significant pre-dictors were BDI Depression (r¼ 0.23, p<0.003)and NFI motor deficits (r¼ 0.22, p< 0.003),�2(2)¼ 57.37, p< 0.0001).

Discussion

This study reports the findings of post-TBI fatigueand potential correlative factors in one of the largestsamples of individuals with TBI that have been

Table IV. Correlation of fatigue measures with significant predictors in stepwise multipleregressions.

MAFSeverity

Beta(n¼115)

MAFADLSBeta

(n¼ 96)

MAFDistress

Beta(n¼ 106)

MAFTiming

Beta(n¼ 108)

GFIBeta

(n¼ 106)

FSSBeta

(n¼ 105)

Female 0.18 0.30* – 0.25* 0.21 –BDI Depression – 0.25* 0.46* 0.27* 0.28* 0.35*NFI Memory 0.40* – – 0.36* 0.30* –Pain intensity 0.39* – 0.22* – 0.30* –NFI Motor – 0.20 0.16 – – 0.33*CHART Social – �0.21* – – – –Pittsburgh Sleep – 0.20 – – – –Use anti-depressant – – – – – 0.19F 38.24* 17.55* 22.01* 26.51* 41.31* 25.28*Adjusted R2 0.49 0.46 0.49 0.42 0.60 0.43

*p< 0.01.

Table III. Spearman’s one-tailed correlations of endocrinemeasures with time since injury and duration of unconsciousness.

n Spearman’s r

Time since injury

Growth Hormone Deficient (glucagon test) 111 0.00Growth Hormone Deficient (basal IGF-1) 117 0.20Growth Hormone (abnormal) 114 0.06Adrenal Insufficiency 118 0.00Hypothyroid 117 0.00Hypogonadal (male) 77 0.11

Duration of unconsciousness

Growth Hormone Deficient (glucagon test) 111 0.00Growth Hormone Deficient (basal IGF-1) 117 0.20Growth Hormone (abnormal) 111 0.11Adrenal Insufficiency 118 0.00Hypothyroid 117 0.00Hypogonadal (male) 77 0.11

1384 J. Englander et al.

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

18/1

0Fo

r pe

rson

al u

se o

nly.

reported to date [5–7, 12, 13]. In addition, thisis a study that includes not only an examinationof neuroendocrine findings after TBI, of which therehave been many, but adds the quantificationand description of fatigue, sleep, depression,community participation and self-assessments ofcognition and motor activity. This study did notdemonstrate a correlation between neuroendocrinedeficiencies and post-TBI fatigue; however,fatigue was associated with a number of otherproblems.

In this community sample of individuals with TBI,one-third to one-half of this sample scored abnormalon the fatigue scales used. This proportion is lessthan that reported by Olver et al. [6], who simplyasked their sample of individuals 2 and 5 years postinjury, ‘Do you experience fatigue?’ In this studypopulation, there was a relatively high levelof employment and productive activity and thesubjects included individuals who were many yearspost-injury with a slightly higher proportion ofwomen than in other studies of individuals afterTBI. These subjects also volunteered for a study tolearn more about fatigue. The bias of this sampletherefore includes people who have lived longenough after their injury to at least be curiousabout their degree of fatigue symptoms and tospend the half day to undergo questioning andmultiple blood tests. The discrepancies between theprevalence of fatigue in this study to that foundin other studies point to the challenges of studies inspecific populations and the application of fatiguescales developed for other populations to the TBIpopulation.

The prevalence of endocrine dysfunction in thispopulation is no longer surprising, but still worthmentioning. The most common abnormalities weredeficiencies in basal cortisol levels and in growthhormone response to glucagon stimulation.Standard cortrosyn stimulation testing isvalidated for assessment of adrenal insufficiency inthis population and further study in these subjectsusing provocative testing may be of use [38]. Growthhormone replacement in adults is still controversialand very expensive. Kelly et al. [12] described alower incidence of GH insufficiency in a prospectivesub-sample of individuals with TBI of 6–9 monthsduration and advocated for randomized clinical trialsof GH replacement. In this same sample, theseauthors found an association of GH abnormalitieswith worse quality-of-life including depression,energy/fatigue, emotional well-being, pain andlimitations to physical health at 6–9 monthspost-TBI compared with controls. However, fatiguewas assessed with one question rated on a 7-pointLikert Scale. In a study of 34 subjects who were 5–12months post-TBI or subarachnoid haemorrhage,

hypothyroidism and GHD were associatedwith diminished life satisfaction and performancefunction on multiple assessments [39].A TBI-specific assessment of fatigue was not usedin the study. In the current study, fatigue wasassessed using both a unidimensional and multi-dimensional scale with multiple items. The studyparticipants were also on average �10 years post-TBI which is a sample that has not been examinedfor neuroendocrine deficiencies and associatedsequelae. These two differences may explain whythe presence of GHD did not correlate with fatiguein this study. This finding is similar to the authors’study in a smaller cohort of participants [13].Although GHD has been shown to correlate withother markers of quality-of-life, GHD does notappear to correlate with fatigue following TBI oflong-lasting duration.

Testosterone deficiency in men was the next mostcommon pituitary dysfunction in the study, althoughthis study did not demonstrate correlation of testos-terone deficiency and fatigue. Replacement of thishormone is not without risk, as it may exacerbateimpulsive or disinhibited behaviour in individualswith brain injury. Therapeutic replacement deservesa randomized controlled trial in this population.Central hypothyroidism (present in 12% of theindividuals in this study) is perhaps the easiest totreat; however, given the concomitant occurrence oflow cortisol in subjects with hypopituitarism, thyroidreplacement should be evaluated concomitantly withcortisol replacement by an endocrinologist.Although it is far from clear that hormone replace-ment will affect the level of fatigue or associatedsymptoms after TBI, it is prudent to screen forpituitary dysfunction after brain injury (at least withfree T4, TSH and basal cortisol levels or cortrosynstimulation test), with referral to an endocrinologistto ascertain the best replacement strategy [40].

Fatigue and associated symptoms

The most robust correlates of fatigue in this samplewere gender, depression, pain and self-assessment ofmemory or motor dysfunction. These predictedbetween 42–60% of the variance on the variousfatigue scales or sub-scales. These factors point tothe multidimensionality of post-TBI fatigue:it encompasses both physical and mental function-ing, insofar as these can be assessed using self-reportmeasures. While gender is not conveniently changed,other associated symptoms of fatigue can be treatedthrough various modalities. Pain can be managedwith therapeutic interventions such as exercise,physical therapy and medication management.Depression is treatable with psychotherapy, groupor peer support and anti-depressant medications.

Factors associated with post-TBI fatigue 1385

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

18/1

0Fo

r pe

rson

al u

se o

nly.

Although it appears that in this sample those takinganti-depressant medications had higher fatiguescores, suggestive of a medication side-effect, it isjust as likely that underlying depression contributesto fatigue. While memory compensation techniquesmay not be easily incorporated into one’s dailyroutine, cognitive remediation may still be helpfuleven years post-injury [41]. Further, although anearly study of modafinil did not improve fatigue[42], other studies are underway to ascertain theimpact of regular exercise on fatigue and relatedsymptoms after TBI.

Sleep disturbances were also highly associatedwith all of the fatigue measurements in this study,although they did not explain any of the variance inthe fatigue sub-scales and total scores when otherfactors were taken into consideration. This study didnot characterize the various types of insomnia, suchas neurogenic or obstructive sleep apnea, initial vsterminal insomnia or restless legs syndrome.Interventions for sleep disorders include improvingsleep hygiene and regulating awake routines, includ-ing following a regular schedule of activities, exerciseand avoiding beverages with alcohol or caffeine aswell as medication management. Given that over20% of this sample reported moderate-to-heavyalcohol use, addressing and lowering substance usemay impact sleep quality and fatigue.

MAF scales also helped in understanding nuancesin fatigue in different domains. For example,memory deficits were associated with greater fatiguein the mental domain; depression and pain intensitywere related to greater distress caused byfatigue; and depression, motor problems and takinganti-depressant medication correlated positively withgreater fatigue on the unidimensional FSS. Furtherinvestigation and validation of these fatigue domainsmay lead to improvements in the treatment of fatigueby suggesting to clinicians which interventionsshould be attempted first.

Limitations and directions for future research

This study has a number of limitations that supportthe exploratory nature of these findings. The studypopulation was recruited from the larger communityof individuals with TBI living in the San Jose area.As such, it was not possible to obtain independentverification of the severity of injury and other injurycharacteristics through medical record review; theauthors were required to rely primarily on partici-pant self-report or, in some cases, information froma significant other. This study was also conducted ata single site; future studies should endeavour toconduct collaborative multi-site trials in order toimprove the generalizability of the findings. As withthe verification of injury severity, the assessment

measures, with the exception of the blood tests andthe Disability Rating Scale, were self-administeredand based upon self-report. While this is a limitationof the study, it can also be viewed as a strength; theindividual with TBI is experiencing these symptomsand the degree to which his/her life is impacted is animportant fact to take into account when investigat-ing these relationships.

Accurate measurement of fatigue in the TBIpopulation remains a challenge. Studies are cur-rently underway to develop and validate a morespecific measurement of fatigue in individuals withTBI. Prospective study of individuals post-injuryshould give a better idea of the natural history offatigue; it would also better capture injury severitywith more precise measures of length of coma orpost-traumatic amnesia. With regard to possiblepredictors of fatigue, it would also be useful toanalyse data collected by clinicians (e.g. medicalrecord data or clinician assessment of deficits) inorder to assess the relative efficacy of these measuresand self-report measures in predicting fatigue.Longitudinal research could also help clarify towhat extent perceptions of deficits change overtime and are associated with changes in fatigue,including specific types of fatigue.

Conclusions

Post-TBI fatigue is a commonly acknowledgedsymptom even many years after injury. It appearsto be more severe in women or those with associatedsymptoms of depression, pain or sleep disturbance.Therefore, clinical and research investigations offatigue post-TBI should include concomitantscreening for treatable depressive symptoms, painand sleep disorders potential interventions includemedications, behavioural therapy to assist with sleephygiene and psychological symptoms. While thisstudy did not find a significant association betweenpituitary dysfunction and fatigue, it appears thatneuroendocrine evaluation, at least for central hypo-thyroidism and adrenal function, is warranted infatigued individuals given the high prevalence rate inthis population and relative ease and cost efficacy ofreplacement.

Acknowledgements

We thank Susan Crawford for performance ofneuroendocrine testing and Kimberly Bellon, LauraJamison and Ketra Toda for their assistance withdata collection.

This research was funded in part by a grant from theU.S. Department of Education, Office of SpecialEducation and Rehabilitative Services, National

1386 J. Englander et al.

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

18/1

0Fo

r pe

rson

al u

se o

nly.

Institute on Disability and Rehabilitation Research(Grant # H133A020524) and by a supplement grantfrom Pfizer Pharmaceuticals Inc. for neuroendocrinetesting.

Declaration of interest: We certify that no partyhaving a direct interest in the results of the researchsupporting this article has or will confer a benefit onus or on any organization with which we areassociated and, if applicable, we certify that allfinancial and material support for this research andwork are clearly identified.

References

1. Cantor JB, Ashman T, Gordon W, Ginsberg A, Engmann C,Egan M, Spielman L, Dijkers M, Flanagan S. Fatigue aftertraumatic brain injury and its impact on participation andquality of life. The Journal of Head Trauma Rehabilitation2008;23:41–51.

2. La Chappelle DL, Finlayson MA. An evaluation of subjectiveand objective measures of fatigue in patients with brain injuryand healthy controls. Brain Injury 1998;12:649–659.

3. Dijkers M, Bushnik T. Assessing fatigue after traumatic braininjury: An evaluation of the Barroso Fatigue Scale. TheJournal of Head Trauma Rehabilitation 2008;23:3–16.

4. Bogaro SR, Baker J, Wethe JV, Prigitano GP, Kwasnica C.Subjective reports of fatigue during early recovery fromtraumatic brain injury. The Journal of Head TraumaRehabilitation 2005;5:416–415.

5. Bushnik T, Englander J, Wright J. The experience of fatiguein the first 2 years after moderate to severe traumatic braininjury: A preliminary report. The Journal of Head TraumaRehabilitation 2008;23:17–24.

6. Olver JH, Ponsford JL, Curran CA. Outcome following TBI:A comparison between 2 and 5 years post injury. Brain Injury1996;10:841–848.

7. Bushnik T, Englander J, Wright J. Patterns of fatigue and itscorrelates over the first 2 years after traumatic brain injury.The Journal of Head Trauma Rehabilitation 2008;23:25–32.

8. Bondanelli M, Ambrosio MR, Zatelli MC, De Marinis L,degli Uberti EC. Hypopituitarism after traumatic braininjury. European Journal of Endocrinology 2005;152:679–691.

9. Rothman MS, Arciegas DB, Filley CM, Wierman ME. Theneuroendocrine effects of traumatic brain injury. The Journalof Neuropsychiatry and Clinical Neurosciences 2007;19:363–372.

10. Behan LA, Phillips J, Thompson CJ, Agha A.Neuroendocrine disorders after traumatic brain injury.Journal of Neurology, Neurosurgery and Psychiatry2008;79:753–759.

11. Leal-Cerro A, Giorest JM, Rincon M, Murillo F, Pujol M,Garcia-Pesqueras F, Dieguez C, Casanueva FF. Prevalenceof hypopituitarism and growth hormone deficiency in adultslong-term after severe traumatic brain injury. ClinicalEndocrinology 2005;62:525–532.

12. Kelly DF, McArthur DL, Levin H, Swimmer S, Dusik JR,Cohan P, Wang C, Swerdloff R. Neurobehavioral and qualityof life changes associated with growth hormone insufficiencyafter complicated mild, moderate or severe traumatic braininjury. Journal of Neurotrauma 2006;23:928–942.

13. Bushnik T, Englander J, Katznelson L. Fatigue after TBI:Association with neuroendocrine abnormalities. Brain Injury2007;21:559–566.

14. Widmer IE, Puder JJ, Konig C, Pargger H, Zerkowski HR,Girard J, Muller B. Cortisol response in relation to theseverity of stress and illness. Journal of ClinicalEndocrinology and Metabolism 2005;90:4579–4586.

15. Agha A, Rogers B, Sherlock M, O’Kelly P, Tormey W,Phillips J, Thompson CJ. Anterior pituitary dysfunction insurvivors of traumatic brain injury. Journal of ClinicalEndocrinology and Metabolism 2004;89:4929–4936.

16. Jones SL, Trainer PJ, Perry L, Wass JA, Besser GM,Grossman A. An audit of the insulin tolerance test in adultsubjects in an acute investigation unit over one year. ClinicalEndrocrinology 1994;41:123–128.

17. Baum HB, Biller BM, Katznelson L, Oppenheim DS,Clemmons DR, Cannistraro KB, Schoenfeld DA,Best SA, Klibanski A. Assessment of growth hormone(GH) secretion in men with adult-onset GH deficiencycompared with that in normal men—a clinical researchcenter study. Journal of Clinical Endocrinology andMetabolism 1996;81:84–92.

18. Bormann J, Shively M, Smith TL, Gifford AL. Measurementof fatigue in HIV-positive adults: Reliability and validity ofthe Global Fatigue Index. Journal of the Association ofNurses in AIDS Care 2001;12:75–83.

19. Belza BL. Comparison of self-reported fatigue in rheumatoidarthritis and controls. The Journal of Rheumatology 1995;22:639–643.

20. Winstead-Fry P. Psychometric assessment of four fatiguescales with a sample of rural cancer patients. Journal ofNursing Measurement 1998;6:111–122.

21. Grady C, Anderson R, Chase GA. Fatigue in HIV-infectedmen receiving investigational interleukin-2. Nursing Research1998;47:227–234.

22. Schwartz CE, Coulthard-Morris L, Zeng Q. Psychosocialcorrelates of fatigue in multiple sclerosis. Archives of PhysicalMedicine and Rehabilitation 1996;7:165–170.

23. Krupp LB, LaRocca NG, Muir-Nash J, Steinberg AD. TheFatigue Severity Scale. Application to patients with multiplesclerosis and systematic lupus erythematosus. Archives ofNeurology 1989;46:1121–1123.

24. Kleinman L, Zodet MW, Hakim Z, Aledort J, Barker C,Chan K, Krupp L, Revicki D. Psychometric evaluation of thefatigue severity scale for use in chronic hepatitis C. Quality ofLife Research 2000;9:499–508.

25. Packer TL, Sauriol A, Brouwer B. Fatigue secondary tochronic illness: Postpolio syndrome, chronic fatigue syn-drome, and multiple sclerosis. Archives of Physical Medicineand Rehabilitation 1994;75:1122–1126.

26. Substance Abuse and Mental Health ServicesAdministration, Office of Applied Studies. NationalHousehold Survey on Drug Abuse. Available online at:www.drugabusestatistics.samhasa.gov/nhsda.htm, accessed 6July 2009.

27. Centers for Disease Control and Prevention. Behavioral RiskFactor Surveillance System User’s Guide. Atlanta, GA: USDepartment of Health and Human Services; 1998.

28. Buysse DJ, Reynolds 3rd CF, Monk TH, Berman SR,Kupfer DJ. The Pittsburgh Sleep Quality Index: A newinstrument for psychiatric practice and research. PsychiatryResearch 1989;28:193–213.

29. Fichtenberg NL, Zafonte RD, Putnam S, Mann NR,Millard AE. Insomnia in a post-acute brain injury sample.Brain Injury 2002;16:197–206.

30. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. Aninventory for measuring depression. Archives of GeneralPsychiatry 1961;4:561–571.

31. Beck AT, Steer RA, Garbin MG. Psychometric properties ofthe Beck Depression Inventory: Twenty-five years of evalu-ation. Clinical Psychology Review 1988;8:77–100.

Factors associated with post-TBI fatigue 1387

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

18/1

0Fo

r pe

rson

al u

se o

nly.

32. Rappaport M, Hall KM, Hopkins K, Belleza T, Cope DN.Disability rating scale for severe head trauma: Coma tocommunity. Archives of Physical Medicine andRehabilitation 1982;63:118–123.

33. Fryer L, Haffey W. Cognitive rehabilitation and communityreadaptation: Outcomes from two program models. TheJournal of Head Trauma Rehabilitation 1987;2:51–63.

34. Gouvier WD, Blanton PD, LaPorte KK, Nepomuceno C.Reliability and validity of the Disability Rating Scale and theLevels of Cognitive Functioning Scale in monitoring recoveryfrom severe head injury. Arcives of Physical Medicine andRehabilitation 1987;68:94–97.

35. Whiteneck GG, Charlifue SW, Gerhart KA, Overholser JD,Richardson GN. Quantifying handicap: A new measure oflong-term rehabilitation outcomes. Archives of PhysicalMedicine and Rehabilitation 1992;73:519–526.

36. Mellick D, Walker N, Brooks CA, Whiteneck G.Incorporating the cognitive independence domain intoCHART. Journal of Rehabilitation OutcomesMeasurements 1999;3:12–21.

37. Kreutzer J, Marwitz J, Seel R, Serio C. Validation of aneurobehavioral functioning inventory for adults with trau-matic brain injury. Archives of Physical Medicine andRehabilitation 1996;77:116–124.

38. Aimaretti G, Ambrosio MR, Di Somma C, Gasperi M,Cannavo S, Scaroni C, Fusco A, Del Monte P, De Menis E,

Faustini-Fustini M, et al. Residual pituitary function afterbrain injury-induced hypopituitarism: A prospective 12-month study. Journal of Clinical Endocrinology andMetabolism 2005;90:6085–6092.

39. Srinivasan L, Roberts B, Bushnik T, Englander J, Spain DA,Steinberg GK, Ren L, Sandel ME, Al-Lawati Z, Teraoka J,et al. The impact of hypopituitarism on function andperformance in subjects with recent history of traumaticbrain injury and aneurismal subarachnoid hemorrhage. BrainInjury 2009;23:639–648.

40. Schneider HJ, Kreitschmann-Andermahr I, Ghigo E,Stalla GK, Agha A. Hypothalamopituitary dysfunction fol-lowing traumatic brain injury and aneuysmal subarachnoidhemorrhage. JAMA: Journal of the American MedicalAssociation 2007;298:1429–1438.

41. Cicerone KD, Mott T, Azulay J, Sharlow-Galella MA,Ellmo WJ, Paradise S, Friel JC. A randomized controlledtrial of holistic neuropsychologic rehabilitation after trau-matic brain injury. Archives of Physical Medicine andRehabilitation 2008;89:2239–2249.

42. Jha A, Weintraub A, Allshouse A, Morey C, Cusick C,Kittleson J, Harrison-Felix C, Whiteneck G, Gerber D.A randomized trial of modafinil for the treatment of fatigueand excessive daytime sleepiness in individuals with chronictraumatic brain injury. The Journal of Head TraumaRehabilitation 2008;23:52–63.

1388 J. Englander et al.

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y N

yu M

edic

al C

ente

r on

11/

18/1

0Fo

r pe

rson

al u

se o

nly.