differential pathways of psychological distress in spouses vs. parents of people with severe...

Brain Injury, November 2009; 23(12): 931–943

Differential pathways of psychological distress in spousesvs. parents of people with severe traumatic brain injury (TBI):Multi-group analysis

MALCOLM I. ANDERSON1, GRAHAME K. SIMPSON2, PETER J. MOREY3,MAGDALENA M. C. MOK4, TAMERA J. GOSLING1, & LAUREN E. GILLETT2

1Faculty of Nursing and Health, Avondale College (Sydney Campus), Wahroonga, NSW, Australia, 2Brain Injury

Unit, Liverpool Health Service, Liverpool, NSW, Australia, 3Faculty of Business and Information Technology,

Avondale College (Lake Macquarie Campus), Cooranbong, NSW, Australia, and 4Centre for Assessment Research and

Development, The Hong Kong Institute of Education, New Territories, Hong Kong

(Received 12 March 2009; revised 30 August 2009; accepted 1 September 2009)

AbstractPrimary objective: A contemporary model of psychological stress based on an amalgamation of Conservation of Resourcestheory and the McMaster Model of Family Functioning was devised to compare the effects of neurobehaviouralimpairments on family functioning and psychological distress in spouses and parents caring for relatives with TBI.Method: Participants were 64 spouses and 58 parents. They completed the Neurobehavioral Problem Checklist, FamilyAssessment Device and the Brief Symptom Inventory. Structural equation modelling (SEM) was used to test the modelfor the combined (spouses and parents) sample. Multi-group analysis was then employed for examining differences instructural weights for spouses and parents.Main results: SEM supported the model for the combined sample. Multi-group analysis showed for spouses cognitiveand behavioural impairments significantly disrupted family functioning, which in turn increased psychological distress.In contrast, cognitive and behavioural impairments did not significantly disrupt family functioning in parents. For parents,however, cognitive impairments increased psychological distress. Furthermore, parents who reported disrupted familyfunctioning also experienced higher levels of psychological distress. The effect of cognitive impairments was statisticallymore influential on the level of distress in parents when compared to spouses.Conclusions: Understanding these differences can assist in better targeting family support interventions.

Keywords: Family, neurobehavioral, psychological, stress, traumatic brain injury

Introduction

For the past three decades, substantial researchattention has documented the impact of TBI onfamily members. Many studies have examined familymembers as a single group, aggregating parents,spouses, siblings, adult children and other relativesinto one sample. However, the primary conduitsof family support have been the parents or spouse

of the person with TBI [1]. Researchers havetherefore been concerned to ascertain and comparethe impact of TBI on these family membersin particular. Since the early study of Panting andMerry [2], over 30 studies have examined orreported on this issue.

A predominant concern in this research has beento determine the level of ongoing psychological and

Correspondence: Malcolm Anderson, RN, PhD, Faculty of Nursing and Health, Avondale College (Sydney Campus), 185 Fox Valley Road, Wahroonga,NSW, Australia, 2076. Tel: þ61(2) 9487 9609. Fax: þ61(2) 9487 9625. E-mail: [email protected]

ISSN 0269–9052 print/ISSN 1362–301X online � 2009 Informa Healthcare Ltd.DOI: 10.3109/02699050903302336

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Ala

ska

Anc

hora

ge o

n 10

/29/

14Fo

r pe

rson

al u

se o

nly.

psychosocial adjustment parents and spouses havemade. Elevated levels of affective distress (depres-sion [3–7], anxiety [5, 8, 9], stress [10–12],perceived burden of care [3, 13, 14], unwelcomerole changes [10], instrumental difficulties [15, 16]and increased health-seeking behaviour [10, 15])have been documented among family members as awhole. Although early clinical descriptions suggestedthat spouses were more at risk than parents [2,17–19], subsequent research using standardizedmeasures have not reached a consensus on thisissue, with similar numbers of studies either support-ing these early findings or, alternatively, finding nodifference in the degree of spouse vs. parent distress.No studies have suggested that parents have elevatedlevels of distress in comparison to spouses.

Brooks et al. [13] and Linn et al. [20] suggestedthat aside from the quantitative dimension ofdistress, it was important to investigate whetherthere were qualitative differences between spousesand parents in the causes of distress. Based on theTBI family literature, qualitative differences couldexist in the pattern of initial stressors reportedby spouses vs. parents arising from the neuro-behavioural impairments displayed by the personwith TBI. Alternatively, the qualitative differencesmay be present in the factors that mediate the impactof those stressors on relative distress.

Few investigations have examined which aspectsof the neurobehavioural profile of people with TBI(cognition, behaviour, physical) impact moststrongly on the distress experienced by spouses vs.parents. In the broader family literature, studieshave found that behavioural impairments generallyhad a greater impact than cognitive or physicalimpairments on relative distress and that cog-nitive impairments had a greater impact than phys-ical impairments [13, 21–23]. Three studies that didinvestigate whether there was a differential impacton parents vs. spouses found no differences intheir subjective reports of neurobehavioural symp-toms displayed by the person with TBI [4, 7, 17].In contrast, Serio et al. [24] found that behaviouralimpairments had the strongest association to theunmet needs of spouses, whereas amotivation hadthe greater impact on parents.

Turning to mediating factors, various demo-graphic and psychological variables have beenidentified as mediators of the relationship betweenthe neurobehavioural stressors and the degree ofrelative distress. These have included pre-morbidsocial and emotional health [25], ethnicity [26],ineffective coping [4, 27, 28], the presence of unmetneeds [8, 24] and perceived adequacy of socialsupport [29–32]. The perception of social supporthas proved to be one of the key mediating factors.In the health literature more generally, social

support has been found to moderate the impactof stress across a broad range of life situations [33].In the TBI studies that compared parents andspouses on this variable, social support has beenmeasured by various instruments including theArizona Social Support interview [10], selecteditems from the Social Support Questionnaire [4]and the Interpersonal Support Evaluation list [34].In two of these studies, parents reported higher levelsof perceived social support than spouses [10, 34].

Interestingly, although these social support mea-sures tap into perceived support from other familymembers as well as friends and work colleagues, therole of the family system as a source of social supportand mediator of individual relative distress has rarelybeen examined. This reflects concerns [5, 35] thattoo much family research has focused on singlecaregivers in isolation of the interaction that existsbetween individual family members and the broaderfamily system. Two studies have incorporated familyfunctioning into a model of relative distress.Employing Structural Equation Modelling (SEM),Whinstanley et al. [8] found no significant path-ways between the health of family functioningand a latent distress variable. In contrast, Ponsfordet al. [7] conducted a series of regression analysesand found that Family Assessment Device–GeneralFunctioning was a significant correlate of elevateddepression and anxiety in caregivers. However,both these studies comprised an aggregated sampleof family members.

In seeking to test complex models of familyfunctioning, SEM is a useful analytic technique.It enables the direct and indirect effects of variablesto be analysed simultaneously [36]. In the firststudy to use SEM to investigate family adaptation toTBI, Chwalisz [37] tested a theoretical model ofcaregiver stress derived from the broader stress-appraisal-coping theory [38]. In this study, mentaland physical health outcomes for spouses whosepartners had sustained a TBI were posited as beingassociated with their degree of self-reported stress.The level of perceived stress was hypothesizedas arising from the interaction between caregivercharacteristics (age, gender, prior mental healthhistory and appraisal of the degree of change intheir partner with TBI) as mediated by the spouses’coping style and degree of social support. The studyfound that coping and level of social support wereassociated with the degree of stress reported byspouses, but did not find a significant relationshipbetween the four caregiver characteristics and theother model variables. Chwalisz [37] observed thatfurther investigation into the possible contributionof the neurobehavioural impairments of the personwith TBI to caregiver distress was needed.Moreover, the study did not include a family systems

932 M. I. Anderson et al.

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Ala

ska

Anc

hora

ge o

n 10

/29/

14Fo

r pe

rson

al u

se o

nly.

dimension in seeking to understand the spouses’distress.

The current authors [39, 40] have previouslyproposed and tested a model of psychologicaldistress using SEM. The model draws uponHobfoll’s Conservation of Resources theory(COR) [41, 42], the McMaster Model of FamilyFunctioning [43] and Elliott and Eisdorfer’s [44]normative perspective of stress (see Figure 1).Within the model, the neurobehavioural sequelaeof TBI served as chronic stressors. Furthermore, themodel proposed that the family system acted as aresource, mediating the relationship between thesestressors and an individual relative’s psychologicalstatus. This was based on COR theory whichproposes that psychological distress occurs whenthere is a threat or actual loss of individual or familyresources [41]. Family resources were operationa-lized by use of the McMaster Model of FamilyFunctioning [43], which has characteristics thatare largely congruent with the conceptualizationsof COR theory [39, 40, 42]. The Family AssessmentDevice [45] has been the most frequently usedmeasure of family functioning in the TBI literature,employed in at least 20 studies [7, 8, 12, 21, 25 ,39,40, 46, 47].

The model was first tested on a sample of spouses[39] and subsequently on a combined sample ofparents and spouses [40]. The findings in bothstudies supported the model. The analyses foundsignificant pathways between the neurobehaviouralimpairments (as measured by the NeurobehavioralProblem Checklist (NPC) [48]) and relative distress.In addition, both studies found that the neurobeha-vioural impairments also had an indirect effect onrelative distress through an impact upon familyfunctioning.

Within SEM, multi-group analysis provides thecapability to test for differences in pathways betweensub-groups of a sample [36], namely spouses vs.parents in the current study. Multi-group analysisinvolved setting up an invariant model in whichspouses and parents were hypothesized to havethe same regression weights and a variant modelin which spouses and parents were hypothesized

to have different regression weights. The two modelswere then directly compared using a critical ratiostest as to their model fit. This analytic techniqueenabled one to investigate whether (i) the samepattern of neurobehavioural impairments acted asstressors on parents vs. spouses and (ii) whetherfamily functioning played a similar mediating socialsupport role for both groups.

Method

Participants

The current sample included the initial 64 spousesfrom the first study [39] combined with a subse-quent sample of 58 parents (29 couples). Ethicalapproval for the different components of the projectwas granted from the relevant New South WalesHealth Area Health Service Human Research EthicsCommittees. Participants were recruited from sixBrain Injury Rehabilitation Units across NSW.The inclusion criteria were that the family partici-pants (i) were living with the injured relative withTBI at the time of injury and after discharge fromhospital, (ii) had primary responsibility for support-ing the relative with TBI, (iii) had no pre-morbidpsychiatric history and (iv) that the relative’s TBIwas severe (post-traumatic amnesia (PTA) greaterthan 24 hours or a period of coma of a least 6 hoursif PTA was not known). Additional criteria wereemployed for the parent’s group, namely that (i) theywere a parental dyad, (ii) the biological parentsof the person with TBI and (iii) both partners agreedto participate in the study.

The total group of relatives numbered 122. Theparents of the adult child with TBI (n¼ 58) werea mean age of 52 years and had been married for amean of 27 years. Spouses (n¼ 64) who had partnerswith TBI had a mean age of 45 years and weremarried for a mean of 20 years. The occupationalstatus of the sample varied from professional andmanagerial positions to unskilled work, homemakersand retirees, which represents a broad range offamilies in terms of socio-economic background.

Chronic stressors Family resources Psychological distress

Psychological distress

Family functioning

Neurobehavioural impairments

Figure 1. Model of psychological distress in relatives of people with TBI.

Distress in spouses vs. parents of people with TBI 933

Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Ala

ska

Anc

hora

ge o

n 10

/29/

14Fo

r pe

rson

al u

se o

nly.

The relatives were the carers of 93 people withTBI, who were an average age of 34 years anda mean of 41 months post-injury (range 4–183months). All these individuals had severe-to-extremely severe brain injuries based on durationof PTA or length of coma as reported by thecaregivers if PTA was not known. Reports frommedical records revealed 84 injured relatives hadPTA scores ranging from 2–224 days (mean¼ 57,SD¼ 40). The remaining nine relatives with TBIwere in coma for more than 6 hours (range¼4 daysto 4 months), indicating that their injuries were alsosevere. The majority of injuries resulted from motortraffic accidents (70%), falls (9%), assaults (3%),sports (6%) and other (12%). Comparing thepeople with TBI from the two groups, there wereno significant differences in gender and time post-injury between the two groups based on chi-squareand t-test analyses. However, results from t-testanalysis (t¼ 3.023; p¼ 0.03) found that the durationof PTA was significantly longer in adult children(M¼ 73 days, SD¼ 45) compared to injuredspouses (M¼42 days, SD¼ 42). Although thedifference was statistically significant, the meanPTA for both groups still fell into the extremelysevere range.

Measures

Neurobehavioral Problem Checklist. TheNeurobehavioral Problem Checklist (NPC) of theGeneral Health and History Questionnaire [48]was utilized to measure the range of domains ofneurobehavioural impairments that arise as sequelaeof a TBI. The NPC is a self-report questionnaireconsisting of 105 items on a 4-point Likert scale thatis rated from ‘Never’ to ‘Always’ by the participant.‘Does Not Apply’ may be chosen if the item is notrelevant to the current circumstance. The itemsare grouped to make up five sub-scales including:Somatic/physical, Thinking/cognition, Behaviour,Communication/language and Social. The instru-ment has sound psychometric properties [40, 49].In this study the NPC scales had reliabilitycoefficients of 0.87, 0.86, 0.94, 0.79 and 0.69,respectively, which are quite acceptable.

Family Assessment Device. The family functioningconstruct was operationalized using the FamilyAssessment Device (FAD) [45] that is based onthe respected McMaster Model of FamilyFunctioning [43] and has acceptable psychometricproperties [50, 51]. The FAD is a self-reportquestionnaire consisting of 60 items on a 4-pointLikert scale, which is rated from Strongly Agree (1)to Strongly Disagree (4). The FAD is a screeningmeasure that consists of six scales that represent the

hypothesis of the MMFF including: problem-solving, communication, roles, affective responsive-ness, affective involvement and behaviour control.Additionally, a general functioning scale made upof 12 items provides a global rating of overallfamily functioning. Scores for unhealthy or ineffec-tive family functioning for each scale range from 1–4,with higher scores representing ineffective familyfunctioning. Cut-off scores for unhealthy familyfunctioning for each scale range from 1.9–2.3.A score greater than 2 for the general functioningscale and a score of more than 2.3 for roles areconsidered to be in the unhealthy or ineffectiverange. Only scores from the general functioningand roles scales were reported in this study, whichis in line with other research that has used the FADgeneral functioning scale as a global indicator [7, 8]or studied roles as a specific construct [52].

Brief Symptom Inventory. The psychological dis-tress construct was operationalized by means of theBrief Symptom Inventory (BSI) [53], which is a self-report questionnaire consisting of 53 items on a5-point scale which is rated from (0) ‘not at all’ to(4) ‘extremely’. It is designed to reflect the psycho-logical symptoms of psychiatric, medical and non-clinical individuals from nine primary symptomdimensions and three global indices: Somatization,depression, obsessive-compulsive symptoms, anxi-ety, interpersonal sensitivity, hostility, phobic anxi-ety, paranoia and psychotic symptoms, globalseverity index, positive symptom total and positivedistress index. The BSI has been well received byinvestigators in psychiatric, medical, non-clinical[53] and TBI populations [11, 21, 39, 40, 46, 49].The reliability and validity of the BSI is very goodand has been extensively reported [49]. There aretwo criteria for making a diagnosis for casenessincluding (a) a global severity index T-score(non-patient norms) of 63 or more or (b) a T-score(non-patient norms) of any two dimensions equalto or greater than 63. This study only assessedthe depression, anxiety and hostility sub-scales inthis study as BSI sub-scales such as psychoticism,obsessive compulsive and paranoid ideation maynot be direct indicators of psychopathology in thispopulation [11, 46].

Procedures

Health professionals from the six participating unitsreviewed their active client lists to identify partici-pants who met the inclusion criteria and thencontacted potential participants to obtain theirpermission to be interviewed. Each participantreceived an information letter, explaining the project


Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Ala

ska

Anc

hora

ge o

n 10

/29/

14Fo

r pe

rson

al u

se o

nly.

and a consent form. Informed consent was grantedby a total of 122 participants, with only 16 people(11.6%; 16/138) refusing to take part. An investiga-tor from the research team then made contact to seta date for the interviews with the uninjured spousesand parents. The interviews took place either at theparticipant’s home or at the respective rehabilitationfacility of the person with TBI. For the parentaldyads, each parent completed the NPC, FAD andBSI independently.

Statistical analysis

Scores from spouses and both parents were enteredinto SPSS version 15.0. Descriptive statistics weregenerated for all key variables. Inspection found thatthe variables were normally distributed and, there-fore, parametric statistical procedures were used.t-tests were conducted to ascertain whether therewere significant between-groups (parents vs.spouses) differences on the variables generatedfrom the NPC, FAD and BSI. For these analyses,a Bonferroni correction was applied to the signifi-cance level (0.05/10, �¼ 0.005) to control for Type1 error arising from multiple comparisons.

SEM was used as a technique that allowedthe analysis of relationships between exogenous(explanatory) and endogenous (dependent) variables

simultaneously [36]. Furthermore, it offered theability to incorporate latent (unobserved) variablesin the model, which were approximated by observedor measured variables [36]. To achieve this a four-step approach was adopted to examine the relation-ships between the constructs neurobehaviouralimpairments, family functioning and psychologicaldistress. First, using AMOS 7 software [54], a partialhybrid model (Figure 2) was constructed with acombination of latent (i.e. unobserved) and observed(i.e. measured) variables. The model comprised fiveobserved explanatory variables including somatic,thinking (cognition), behaviour, communication andsocial impairments. Family functioning and psycho-logical distress were represented as latent variablesthat were measured by multiple observed variables.That is, the two observed variables general function-ing and roles measured the latent mediatingvariable family functioning. The latent variablepsychological distress was measured by threeobserved variables including depression, anxietyand hostility. Psychological distress was also theendogenous variable in the model. Notably, thismodel was more extensive than the initial research[39], which was composed of a path model thatspecified the structural relations amongst observedvariables only, with single global scales for measuringfamily functioning and psychological distress.

Thinking

Behaviour

Social

Familyfunctioning

Psychologicaldistress

Generalfunctioning

e3

Roles

e4

Anxiety

e6

1

Depression

e5

1

e1

e21

1

Hostility

e7

1

1

Communication

Somatic1 1

1

Figure 2. Initial path model of psychological distress in relatives of family members with TBI.


Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Ala

ska

Anc

hora

ge o

n 10

/29/

14Fo

r pe

rson

al u

se o

nly.

In a second step maximum likelihood estimates forthe model parameters using the covariance matricesfor the total sample (n¼ 122) were calculated [54].The sample size was acceptable as it was within thesuggested minimum range of 100–200 subjectsfor SEM [55–57]. Multiple measures were used toassess the fit of the data to the model namely;chi square (�2) statistic, RMSEA and base linecomparisons fit indices of NFI, RFI, IFI, TLI andCFI. Once a model with adequate fit was obtained,it was used as a foundation to apply multi-groupanalysis (step 3) to test for differences betweenspouses and parents. To do this, an invariant andvariant model were set up, which were directlycompared as to their model fit. The AIC measurewas employed to determine which competing modelwas more parsimonious and better fitting. In thefourth step the critical ratio test was engaged to testfor kinship differences in the regression weights [36].

Results

Univariate analyses

Results of t-tests conducted to ascertain whetherthere were any between-groups differences (spousesvs. parents) on the three measures are displayedin Table I. The Thinking/cognitive impairmentssub-scale of the NPC was reported to be the mostproblematic by the relatives as a whole, scoringabove the mid-point (2) of the scale. The between-groups analysis found that spouses reported signifi-cantly more cognitive and social impairments in therelative with TBI than parents.

The FAD scores for the group as a whole fellbetween psychiatric and medical norms [51], whichare similar to findings on TBI groups from NorthAmerica [46]. Upon closer inspection spousesexceeded the mean cut-off scores for unhealthy

role functioning and general functioning. Nearly twothirds of the spouses had scores in the unhealthyrange for general functioning (64%) and roles(64%), which were remarkably similar to thefindings reported by Camplair [49]. In contrast,general functioning and role performance for parentswere in the effective range, which was consistentwith non-patient norms [57] and other TBI families[7]. Nevertheless, half of the parents reporteddifficulties with role functioning and 42% reportedineffective family functioning on the generalfunctioning scale. The t-tests found that spousesexperienced significantly more difficulties withgeneral functioning than parents, unlike previousresearch [7, 46].

The group mean BSI scores for relatives exceededthe means for non-patients (T¼ 50) for depression,anxiety and hostility, which indicated that thecaregivers experienced a broad range of symptomsof psychological distress. Higher percentages ofspouses reported psychological distress at casenesslevel than parents for depression (50% vs. 35%)anxiety (36% vs. 29%) and hostility (39% vs. 26%),respectively; however, t-tests indicated that thedifferences on the sub-scale scores were notsignificant.

Model of psychological distress: Total group

First, the data for the aggregated group was screenedusing descriptive and correlational techniques todetermine the most suitable variables to be includedin the model for testing. Correlations among the10 variables of the matrix were positive and rangedfrom 0.23–0.75, with 45% of the correlations above0.50. Scores obtained on the FAD general function-ing scale had correlations with the NPC variablessomatic, communication and social impairmentsthat were less than 0.30 and the FAD roles scale

Table I. NPC, FAD and BSI scores for relatives of family members with TBI.

Total groupM (SD)(n¼ 122)

SpousesM (SD)(n¼ 64)

ParentsM (SD)(n¼ 58) t statistic*

NPC Somatic 1.82 (0.44) 1.91 (0.042) 1.71 (0.44) nsThinking 2.05 (0.56) 2.19 (0.55) 1.89 (0.58) � 2.860*Behaviour 1.93 (0.54) 2.04 (0.53) 1.81 (0.53) nsCommunication 1.80 (0.58) 1.85 (0.62) 1.74 (0.53) nsSocial 1.85 (0.58) 2.01 (0.58) 1.67 (0.54) � 3.280*

FAD Roles 2.28 (0.41) 2.34 (0.42) 2.21 (0.38) nsGeneral functioning 2.01 (0.51) 2.16 (0.47) 1.83 (0.51) � 3.726*

BSI Depression 58.72 (10.80) 60.73 (10.1) 56.50 (11.2) nsAnxiety 56.19 (11.70) 57.39 (11.2) 54.86 (12.18) nsHostility 57.02 (11.29) 59.58 (10.6) 54.21 (11.39) ns

NPC¼Neurobehavioral Problem Checklist; FAD¼Family Assessment Device; BSI¼Brief Symptom Inventory.*Significant at Bonferroni adjusted p¼ 0.005.


Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Ala

ska

Anc

hora

ge o

n 10

/29/

14Fo

r pe

rson

al u

se o

nly.

had a correlation with the NPC variable social,which was less than 0.3 as well. Subsequently, theseweaker correlations were not included in the pathmodel for analysis.

Then the model was submitted for analysis [54],making use of the techniques developed by Joreskogand Sorbom [58], which was theory driven andempirically validated. An optimal model was derivedusing this approach by an iterative process ofinspection between statistical significance of pathcoefficients and theoretical relevance of constructsin the model. The plausible explanatory factorssomatic and communication were not statisticallysignificant (at the 0.05 level) to be retained in themodel. The model (Figure 3) for the aggregatedsample comprised two explanatory observed vari-ables, namely thinking and behaviour, whichrevealed significant, direct effects on the latentvariable family functioning. Further, the threeexplanatory variables thinking and social and thelatent variable Family functioning had significantdirect effects on the latent variable Psychological

distress. Together, these variables explained 62%of the variance of psychological distress in relatives.

The model as a whole fitted the data very well,as indicated by the goodness-of-fit indices(�2¼ 19.252; p¼ 0.203; NFI¼ 0.966, RFI¼ 0.937;

IFI¼ 0.992; TLI¼0.985, CFI¼ 0.992 andRMSEA¼0.048). The base line comparisons fitindices were all above 0.9, which constituted a goodfit [59, 60]. Moreover, the RMSEA value was lessthan 0.05, which indicated a close fit between thedata and the model [61]. Using these criteria,there was strong empirical evidence to support thehypothesized theoretical model.

Multi-group analysis was conducted comparingthe fit of the group-variant model and the group-invariant model and assess differences betweenspouses and parents in the strength of the pathsbetween neurobehavioural impairments, familyfunctioning and psychological distress. For bothgroup-variant and group-invariant models a good-fitting model of the relationship between neurobe-havioural impairment of those with TBI and familyfunctioning and psychological distress in caregiverswho were relatives was achieved (group-variant�2¼ 35.85; df¼ 33; p¼ 0.34; group-invariant

�2¼ 46.31; df¼38; p¼ 0.167). The baseline

Thinking

Behaviour

Social

0.29

Familyfunctioning

0.62


0.67

Generalfunctioning

e3

0.61

Roles

e4

0.85

Depression

e5

0.56 0.63 0.45

e1

e2

0.75

0.27

0.63

Anxiety

e6

0.82

0.29

0.29

0.67

Hostility

e7

0.82

0.25

0.92 0.80

0.78

Figure 3. Path model of psychological distress in relatives of family members with TBI.


Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Ala

ska

Anc

hora

ge o

n 10

/29/

14Fo

r pe

rson

al u

se o

nly.

comparison fit indices NFI, RFI, IFI, TLI and CFIfor both models were close to or are above 0.90(range: 0.886–0.995), which also indicated the fitof the models was sufficient. Furthermore, theRMSEA yielded values for the group-variant andgroup-invariant path models of 0.027 and 0.043,respectively, which also demonstrated the fit of thetwo models was very good [62].

When the models were compared using the AICmeasure [63], which takes into account both modelparsimony and model fit, the score for the group-variant model (113.854) was lower than the group-invariant model (114.308). This indicated thegroup-variant model was both more parsimoniousand better fitting than the group-invariant model.Subsequently, the group-variant model’s estimateswere used rather than the group-invariant model’sestimates [36]. Based on this recommendationthe group variant model, which indicated spousesand parents have one or more path coefficientsthat were different, was adopted for this analysis.

Model of psychological distress experienced

by caregivers: Spouses

The model for psychological distress in spousesof partners with TBI, including standardized path

coefficients, is presented in Figure 4. Parametersreaching statistical significance included the twoexplanatory variables thinking ( p< 0.01) and beha-viour ( p< 0.01), which had significant direct effectson family functioning, while two explanatory vari-ables had significant effects on psychological dis-tress, namely social ( p< 0.01) and the mediatingvariable family functioning ( p<0.01). Together, thevariables thinking, behaviour, social and familyfunctioning explained 75% of the variance of psy-chological distress in spouses, which indicatedthe model was a very good approximation of thosefactors that contribute to psychological distressin spouses of partners with TBI.

Direct and indirect effects: Spouses

Based on standardized estimates (�’s) the variablethinking had a considerable direct effect on familyfunctioning (�¼ 0.397), which suggested spouseswho had partners with many cognitive impairmentswere likely to report high levels of ineffective orunhealthy family functioning. Furthermore, as beha-vioural impairments increased spouses reportedmore ineffective family functioning (�¼ 0.461),which was expected. As hypothesized, spouses whoreported high levels of ineffective family functioning

Thinking

Behaviour

Social

0.59

Familyfunctioning

0.75


0.53

Generalfunctioning

e3

0.73

0.45

Roles

e4

0.67

0.59

Hostility

e7

0.61

Anxiety

e6

0.78

0.86

Depression

e5

e1

e2

0.46

0.76

0.28

0.420.48

0.61

–0.06

0.40

0.770.93

Figure 4. Model of psychological distress in spouses of partners with TBI.


Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Ala

ska

Anc

hora

ge o

n 10

/29/

14Fo

r pe

rson

al u

se o

nly.

also reported many symptoms of psychologicaldistress (�¼ 0.758); which was the strongest directpath coefficient in the model. Additionally, socialimpairments had a direct effect on psychologicaldistress (�¼ 0.284), which suggested spouses whohad partners with many social impairments experi-enced high levels of psychological distress.

Indirect effects were also examined to determinethe influence of neurobehavioural impairmentsthrough family functioning on psychological distress.Both thinking impairments (�¼ 0.301) and beha-vioural impairments (�¼0.350) had a considerableindirect influence on psychological distress. Thesefindings suggested cognitive and behavioural impair-ments disrupted family functioning and subse-quently increased the level of psychological distressin spouses of partners with TBI, which wasexpected.

Psychological distress accounts for 86%, 61% and59% of the variance of depression, anxiety andhostility, respectively; indicating respective reliabil-ities of at least 0.86, 0.61 and 0.59. Further, familyfunctioning accounts for 53% and 45% of thevariance in general functioning and roles; indicatingrespective reliabilities of 0.53 and 0.45.

Model of psychological distress experienced

by caregivers: Parents

In the parent model (Figure 5) parameters reachingstatistical significance comprised two plausibleexplanatory variables, namely thinking for the levelof psychological distress ( p< 0.01) and familyfunctioning for the level of psychological distress( p< 0.01). Jointly, thinking and family functioningexplained 57% of the variance of psychologicaldistress in parents, which indicated the model wasa sound approximation of those factors that contrib-ute to psychological distress. Nevertheless, thinkingand behaviour had no significant influence on familyfunctioning, with only 8% of the variance in familyfunctioning explained by these variables, whichsuggested other factors might be influencing fam-ily dynamics where parents were caring for adultchildren with TBI.

Direct and indirect effects: Parents

The variable thinking had a considerable directeffect on psychological distress (�¼ 0.436), whichsuggested parents of adult children with manycognitive impairments were likely to report high

Thinking

Behaviour

Social

0.08

Familyfunctioning

0.57


0.74

Generalfunctioning

e3

0.86

0.73

Roles

e4

0.86

0.72

Hostility

e7

0.66

Anxiety

e6

0.81

0.84

Depression

e5

e1

e2

0.08

0.37

0.17

0.650.76

0.87

0.44

0.22

0.850.91

Figure 5. Model of psychological distress in parents of adult children with TBI.


Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Ala

ska

Anc

hora

ge o

n 10

/29/

14Fo

r pe

rson

al u

se o

nly.

level of psychological distress. Furthermore, parentswho reported high levels of ineffective familyfunctioning also reported many symptoms of psy-chological distress (�¼ 0.372), which was expected.On the other hand, the neurobehavioural impair-ment constructs had no significant direct effecton family functioning or subsequent indirect effecton psychological distress in the model, which wasunexpected.

Psychological distress accounts for 84%, 66% and85% of the variance of depression, anxiety andhostility, respectively; indicating respective reliabil-ities of at least 0.84, 0.66 and 0.85. Furthermore,family functioning accounts for 74% and 73% of thevariance in general functioning and roles; indicatingrespective reliabilities of 0.74 and 0.73.

Spouse vs. parent

For both spouses and parents, family functioninghad a direct influence on psychological distress.For parents only, more cognitive impairments ledto more symptoms of psychological distress. On theother hand, more cognitive and behavioural impair-ments reported by spouses resulted in high levelsof ineffective family functioning and subsequentpsychological distress. For spouses only, moresocial impairments led to an increase in symptomsof psychological distress.

In order to determine the differences between pathcoefficients for spouses and parents the critical ratiosfor differences test was applied (Step 4). The pathcoefficient between thinking and psychologicaldistress was the only path coefficient that wassignificantly different (critical ratio difference¼�2.288) between the groups. For parents, theassociation between thinking and psychologicaldistress was significant but not for spouses. Thismeans thinking or cognitive impairments in theadult child with TBI were more influential on thelevel of psychological distress in parents whencompared to spouses of partners with TBI.

Discussion

This study tested a contemporary model of stressbased on COR theory [41, 42] and the McMasterModel of Family Functioning [43] employing multi-group analysis. Overall, the model performedstrongly, accounting for a substantial proportion ofthe variance in relative distress for the aggregatedsample as a whole (step 2). Moreover, the initialanalysis indicated that different interactional pat-terns for spouses vs. parents existed among the threesets of variables (i.e. a variant model). The nextstep of analysis generated these unique modelsfor parents and spouses (step 3, displayed in

Figures 4 and 5). Finally, employing a criticalratios test, the pathways in the two models weretested against each other, identifying that the asso-ciation between cognition and relative distressfor parents distinguished between the two models.

The association between cognitive impairmentsand distress for the parents identified a differentialimpact in the pattern of neurobehavioural impair-ments upon parents vs. spouses. Previous researchhas reported the contribution of cognitive impair-ments to relatives including parental distress [7, 21].Interestingly, the descriptive statistics indicatedthat the parents did not identify as high a level ofcognitive impairment as the spouses. Despite this,the strong direct pathway between cognition anddistress was not found in the spouses group. Thereare at least two possible explanations for thisassociation among parents. First, the distress couldbe associated with the demands of managing aperson with memory, attentional or executiveimpairments [21]. Another possibility arises fromfindings in previous studies that parents are moreconcerned about the future care needs of theperson with TBI than spouses [4, 20]. The cognitiveimpairments displayed by their adult child may bean ongoing reminder of the ever-present problemof ‘Who will care for my child when I am no longerable to do so?’.

The models generated at step 3 accounted fora significant proportion of the variance in distress forthe spouses and the parents. In terms of the impactof neurobehavioural impairments, the spouse’smodel displayed an association between cognitive,behavioural and social impairments and familyfunctioning or psychological distress. However, asreported in previous studies [7, 21], neither physical(somatic) nor communication impairments had anegative association with disrupted family function-ing or relative distress.

Furthermore, in the spouses group, the familyunit was vulnerable to the impact of high levelsof behavioural and cognitive impairments (a directeffect). Marital dyads typically form the primarystructure within family units and, therefore, when apartner is injured there is significant potential fordisruption to family functioning overall [17, 64].This was reflected in the descriptive statistics, withthe spouses’ mean scores for general functioning androles above the cut-off scores for disrupted familyfunctioning and nearly two thirds of the samplefalling in the unhealthy range. It was this disruption,a diminishing of social support, which in turn linkedthe neurobehavioural impairments to increasedlevels of psychological distress (indirect effect).This reflects findings from a previous study usingSEM on an aggregated sample of parents andspouses, which also found that neurobehavioural


Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Ala

ska

Anc

hora

ge o

n 10

/29/

14Fo

r pe

rson

al u

se o

nly.

impairments did not have a direct effect on relativedistress, but an indirect effect mediated through thedisruption to family functioning [8].

In the model for the parents, neurobehaviouralimpairments were not associated with disruptedfamily functioning. One reason could be that thepreserved parental dyads mitigated the extent of thenegative impact of the TBI on the family systemas a whole.

However, it may also be that the presence ofanother uninjured adult within the family setting(viz. the second parent) helped to balance generalfamily functioning. Moreover, the recruitment of thisparticular sub-sample of parents may account for thedifference between the two-groups of relatives inthe univariate analyses. The finding contrasts withall previous studies, none of which found a signifi-cant difference between spouses and parents on thefull FAD [7, 21] or general functioning sub-scale[36, 65]. If a broader range of parents had beenrecruited, including parents who were single, sepa-rated, divorced and widowed or a parent who hadremarried so that a step-parent was also involvedin providing care, the results may have beendifferent. Nevertheless, despite their relationshipsremaining intact, parents were not emotionallyimmune to the consequences of TBI, with abouta third of parents presenting with clinically elevatedscores for anxiety, depression and hostility.

Clinical implications

A number of clinical implications arise from thesefindings. It highlights again the potential role thatspouse support groups could play [66]. Moreover;interventions focused on restoring the family rela-tionships may alleviate individual relative distress.Programmes that provide family/relationship coun-selling [67] or training for spouses in the knowledgeand skills to manage and minimize the impactof behavioural and cognitive impairments [68] maybe a useful adjunct to the provision of counsellingfor depression or anxiety.

Limitations and directions for future studies

There are some limitations to be considered whentaking into account the findings of this investigation.The findings in this study are based on familieswho remained together following rehabilitation andwell into the future, which suggests these familiesmay have made some degree of adjustment, com-pared to those families that have chosen to disen-gage. Moreover, a control group is needed to identifythe extent caregivers of relatives with TBI differ fromfamilies supporting people who sustained other typesof trauma, chronic illness or disability. The modelwas not exhaustive in testing all possible types

of mediating variables. Other variables that mightalso influence the link between neurobehaviouralimpairments and psychological distress includecoping skills [37], physical health status [69], thedegree of psychosocial integration of the person withTBI [8], the level of family empowerment [70, 71]and the degree of perceived social support froma broader range of sources. The current findingsshould be replicated taking into account theseimportant additional variables. The generalizabilityof the findings was limited by certain characteristicsof the sample. The mean age of the non-injuredspouses (45 years) and duration of the marriages(average 20 years) were greater than in many studiesand therefore the findings may not apply to youngercouples married over shorter time periods. Finally,the caregivers in this study were Caucasian fromEnglish speaking backgrounds. Sander et al. [26]have reported that family support arrangements dovary depending on culture. Therefore, additionalresearch is needed to replicate the current findingswith people from various cultural backgrounds.

Conclusion

Given the complex interactions between multiplevariables that characterize the experience of families,the findings from this study can be considered to beamong the first in the TBI literature to employ SEMto identify critical pathways and relationships infamily stress and adaptation. Results from thisresearch confirm the hypothesized model, namelythat neurobehavioural impairments had both directand indirect (mediated by the disruption caused byfamily functioning) effects on relative distress. Moreparticularly, employing a more sophisticated SEM,namely multi-group analysis, the results confirmedthat there were different pathways associated withthe distress experienced by spouses in contrast to theparents, which can assist in better targeting of familysupport interventions.

Acknowledgements

This work was supported by the Avondale CollegeFoundation and the Australasian Research Institute(Grant 050301). The researchers acknowledge thecontributions made by the staff from the Brain InjuryRehabilitation Programmes in NSW in facilitatingthis project.

Declaration of interest: The authors report noconflicts of interest. The authors alone are respon-sible for the content and writing of the paper.


Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Ala

ska

Anc

hora

ge o

n 10

/29/

14Fo

r pe

rson

al u

se o

nly.

References

1. Perlez A, Kinsella G, Crowe S. Impact of traumatic braininjury on the family: A critical review. RehabilitationPsychology 1999;44:6–35.

2. Panting A, Merry PH. The long-term rehabilitation of severehead injuries with particular reference to the need of socialand medical support for the patients family. Rehabilitation1972;28:33–37.

3. Rivera P, Elliott TR, Berry JW, Grant JS, Oswald K.Predictors of caregiver depression among community-residing families living with traumatic brain injury.NeuroRehabilitation 2007;22:3–8.

4. Knight RG, Devereux RT, Godfrey HPD. Caring for a familymember with a traumatic brain injury. Brain Injury 1998;12:467–481.

5. Perlesz A, Kinsella G, Crowe S. Psychological distress andfamily satisfaction following traumatic brain injury: Injuredindividuals and their primary, secondary, and tertiary carers.Journal of Head Trauma Rehabilitation 2000;15:909–929.

6. Gillen R, Tennen H, Affleck G, Steinpreis R. Distress,depressive symptoms, and depressive disorder among caregivers of patients with brain injury. Journal of Head TraumaRehabilitation 1998;13:31–43.

7. Ponsford J, Olver J, Ponsford M, Nelms R. Long termadjustment of families following traumatic brain injury wherecomprehensive rehabilitation has been provided. Brain Injury2003;17:453–468.

8. Winstanly J, Simpson GK, Tate RL, Miles B. Early indicatorsand causal factors of psychological distress in relatives duringrehabilitation following severe TBI: Findings from theBrain Injury Outcomes Study. Journal of Head TraumaRehabilitation 2006;21:453–466.

9. Livingston MG, Brooks DN, Bond MR. Patient outcome inthe year following severe head injury and relatives’ psychiatricand social functioning. Journal of Neurology, Neurosurgery,and Psychiatry 1985a;48:876–881.

10. Leatham J, Heath E, Woolley C. Relatives perceptions of rolechange, social support and stress after traumatic brain injury.Brain Injury 1996;10:543–546.

11. Gervasio AH, Kreutzer JS. Kinship and family members’psychological distress after traumatic brain injury: A largesample study. Journal of Head Trauma Rehabilitation1997;12:14–26.

12. Groom KM, Shaw TG, O’Connor ME, Howard NI,Pickens A. Neurobehavioural symptoms and family function-ing in traumatically brain injured adults. Archives of ClinicalNeuropsychology 1998;13:695–711.

13. Brooks N, Campsie L, Symington C, Beatie A, McKinlay W.The effects of severe head injury on patient and relativewithin seven years of injury. Journal of Head TraumaRehabilitation 1987;2:1–13.

14. Hoofien D, Gilboa A, Vakil E, Donovik P. Traumatic braininjury (TBI) 10–20 years later: A comprehensive outcomestudy of psychiatric symptomatology, cognitive abilities andpsychosocial functioning. Brain Injury 2001;15:189–209.

15. Hall KM, Karzmark P, Stevens M, Englander J, O’Hare P,Wright J. Family stressors in traumatic brain injury: A two-year follow-up. Archives of Physical Medicine andRehabilitation 1994;75:876–884.

16. Wells R, Dywan J, Dumas J. Life satisfaction and distressin family caregivers as related to specific behaviouralchanges after traumatic brain injury. Brain Injury 2005;19:1105–1115.

17. Lezak M. Brain damage is a family affair. Journal of Clinicaland Experimental Neuropsychology 1988;10:111–123.

18. Thomsen IV. The patient with severe head injury and hisfamily. Scandinavian Journal of Rehabilitation Medicine1974;6:180–183.

19. Thomsen IV. Late outcome of very severe blunt head trauma:A 10–15 year follow-up. Journal of Neurology, Neurosurgery,and Psychiatry 1984;47:260–268.

20. Linn RT, Allen K, Willer BS. Affective symptoms in thechronic stage of traumatic brain injury: A study of marriedcouples. Brain Injury 1993;8:135–147.

21. Kreutzer JS, Gervasio AH, Camplair PS. Patient correlatesof caregivers’ distress and family functioning after traumaticbrain injury. Brain Injury 1994;8:211–230.

22. Marsh NV, Kersel DA, Havill JH, Sleigh JW. Caregiverburden at 6 months following severe traumatic brain injury.Brain Injury 1998;12:225–238.

23. Marsh NV, Kersel DA, Havill JH, Sleigh JW. Caregiverburden during the year following severe traumatic braininjury. Journal of Clinical and ExperimentalNeuropsychology 2002;24:434–447.

24. Serio CD, Kreutzer JS, Gervasio AH. Predicting family needsafter brain injury: Implications for intervention. Journal ofHead Trauma Rehabilitation 1995;10:32–45.

25. Sander AM, Sher M, Malec JF, High Jr WM,Thompson RN, Moessner AM, Josey J. Preinjury emotionaland family functioning in caregivers of persons with traumaticbrain injury. Archives of Physical Medicine andRehabilitation 2003;84:197–203.

26. Sander AM, Davis LC, Struchen MA, Atchison T, Sherer M,Malec JF, Nakase-Richardson R. Relationship of race/ethnicity to caregivers’ coping, appraisals, and distress aftertraumatic brain injury. NeuroRehabilitation 2007;22:9–17.

27. Flanagan DAJ. A retrospective analysis of expressed emotion(EE) and affective distress in a sample of relatives caring fortraumatically brain-injured (TBI) family members. BritishJournal of Clinical Psychology 1998;37:431–439.

28. Stebbins P, Pakenham KI. Irrational schematic beliefs andpsychological distress in caregivers of people with severetraumatic brain injury. Rehabilitation Psychology 2001;46:178–194.

29. Sander AM, High Jr WM, Hanny HJ, Sherer M. Predictors ofpsychological health in caregivers of patients with closed headinjury. Brain Injury 1997;11:235–249.

30. Harris JK, Godfrey HPD, Partridge FM, Knight RG.Caregiver depression following traumatic brain injury(TBI): A consequence of adverse effects on family members?Brain Injury 2001;15:223–238.

31. Ergh TC, Rapport LJ, Coleman RD, Hanks RA. Predictorsof caregiver and family functioning following traumatic braininjury: Social support moderates caregiver distress. Journal ofHead Trauma Rehabilitation 2002;17:155–175.

32. Ergh TC, Hanks RA, Rapport LJ, Coleman RD. Socialsupport moderates caregiver life satisfaction following trau-matic brain injury. Journal of Clinical and ExperimentalNeuropsychology 2003;25:1090–1101.

33. Rice VH. Handbook of stress and coping: Implications fornursing research, theory and practice. Thousand Oaks, CA:Sage; 2000.

34. Chronister J, Chan F. A stress process model of caregivingfor individuals with traumatic brain injury. RehabilitationPsychology 2006;51:190–201.

35. Gan C, Schuller R. Family system outcome followingacquired brain injury: Clinical and research perspectives.Brain Injury 2002;16:311–322.

36. Ho R. Handbook of univariate and multivariate dataanalysis and interpretation with SPSS. Boca Raton:Chapman & Hall/CRC; 2006.


Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Ala

ska

Anc

hora

ge o

n 10

/29/

14Fo

r pe

rson

al u

se o

nly.

37. Chwalisz K. The perceived stress model of caregiver burden:Evidence from spouses of persons with brain injuries.Rehabilitation Psychology 1996;41:91–113.

38. Lazarus R, Folkman S. Transactional theory and research onemotions and coping. European Journal of Personality1987;1:141–169.

39. Anderson MI, Parmenter TR, Mok M. The relationshipbetween neurobehavioural problems of severe traumaticbrain injury (TBI), family functioning and the psychologicalwell-being of the spouse/caregiver: Path model analysis. BrainInjury 2002;6:743–757.

40. Anderson MI, Simpson G, Mok MC, Parmenter TR.A contemporary model of stress for understanding familyfunctioning and the psychological distress in relatives ofpeople with severe traumatic brain injury (TBI). In: Johns D,editor. Stress and its impact on society. New York: NovaScience Publishers; 2006. p 23–56.

41. Hobfoll SE. Conservation of resources: A new attemptat conceptualizing stress. American Psychologist 1989;44:513–524.

42. Hobfoll SE, Spielberger CD. Family stress: Integratingtheory and measurement. Journal of Family Psychology1992;6:99–112.

43. Epstein NB, Bishop DS, Levin S. The McMaster modelof family functioning. Journal of Marriage and FamilyCounseling 1978;4:19–31.

44. Elliott GR, Eisdorfer C. Stress and human health. New York:Springer Publishing Company; 1982.

45. Epstein NB, Baldwin LM, Bishop DS. The McMaster familyassessment device. Journal of Marital and Family Therapy1983;9:171–180.

46. Kreutzer JS, Gervasio AH, Camplair PS. Primary caregivers’psychological status and family functioning after traumaticbrain injury. Brain Injury 1994;8:197–210.

47. Kosciulek JF. Relationship between family schema to familyadaptation to brain injury. Brain Injury 1997;11:821–830.

48. Kreutzer J, Doherty K, Turner, H, Walland P, Leininger B.The General Health and History Questionnaire. Richmond,VA: Rehabilitation Research and Training Center on SevereTraumatic Brain Injury, Medical College of Virginia; 1987.

49. Camplair PS. The impact of disability following head injuryon caregivers’ psychological status and family functioning[dissertation]. Richmond, VA: Virginia CommonwealthUniversity; 1989. 153p. Available from: UMI DissertationServices, Ann Arbor MI; 9019593.

50. Bishop DS, Miller IW. Traumatic brain injury: Empiricalfamily assessment techniques. Journal of Head TraumaRehabilitation 1988;3:6–30.

51. Kabacoff RI, Miller IW, Bishop DS, Epstein NB, Keitner GI.A psychometric study of the McMaster family assessmentdevice in psychiatric, medical, and nonclinical samples.Journal of Family Psychology 1990;3:431–439.

52. Frosch S, Gruber A, Jones C, et al. The long-term effects oftraumatic brain injury on the roles of caregivers. Brain Injury1997;11:891–906.

53. Derogatis LR. Brief Symptom Inventory: Administration,scoring and procedures manual. Minneapolis: NationalComputer Systems, Inc; 1993.

54. Arbuckle JL, Wothke W. AMOS 7 [Computer software],Chicago: SPPS Inc; 2006.

55. Ding L, Velicer WF, Harlow LL. Effects of estimationmethods, number of indicators per factor, and impropersolutions on structural equation modeling fit indices.Structural Equation Modeling 1995;2:119–143.

56. Kline RB. Principles and practice of structural equationmodelling. 2nd ed. New York: Guilford Press; 2005.

57. Hoyle RH. Structural equation modeling: Concepts, issues,and applications. Thousand Oaks, CA: Sage Publications;1995.

58. Joreskog KG, Sorbom D. LISREL 8: Users reference guide.Chicago: Scientific Software International; 1993.

59. Bentler PM. Comparative indices in structural models.Psychological Bulletin 1980;107:238–246.

60. Bentler PM, Bonnett DG. Significance tests for goodness offit in the analysis of covariant structures. PsychologicalBulletin 1980;88:588–606.

61. MacCullam RC, Hong S. Power analysis in covariancestructure modeling using GFA and AGFI. MultivariateBehavioural Research 1997;32:193–210.

62. Browne MW, Cudeck R. Alternatives ways of assessingmodel fit. In: Bollen KA, Long JS, editors. Testing structuralequation models. Newbury Park, CA: Sage; 1993. p 136–162.

63. Akaike H. Factor analysis and AIC. Psychometrica 1987;52:317–332.

64. Verhaege S, Defloor T, Grypdonck M. Stress and copingamong families of patients with traumatic brain injury:A review of the literature. Journal of Clinical Nursing2005;14:1004–1012.

65. Tate R, Cameron ID, Whinstanley J, Myles B, Harris R. Thebrain injury outcomes study-Final report 2004 June.Rehabilitation Studies Unit, University of Sydney.

66. Zeigler EA. Reflections of a spouses’ group. CognitiveRehabilitation 1990;8:14–19.

67. Larøi M. The family systems approach to treating families ofpersons with brain injury: A potential collaboration betweenfamily therapist and brain injury professional. Brain Injury2003;17:175–187.

68. Carnevale GJ, Anselmi V, Busichio K, Millis SR. Changes inratings of caregiver burden following a community-basedbehavior management program for persons with traumaticbrain injury. Journal of Head Trauma Rehabilitation2002;17:83–95.

69. McPherson KM, McNaughton HK. Brain injury-the per-ceived health of carers. Disability and Rehabilitation2000;22:683–689.

70. Man DKW. Family caregivers’ reactions and coping forpersons with brain injury. Brain Injury 2002;16:1025–1038.

71. Man DKW. The empowering of Hong Kong Chinesefamilies with a brain damaged member: Its investigationand measurement. Brain Injury 1998;12:245–254.


Bra

in I

nj D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Ala

ska

Anc

hora

ge o

n 10

/29/

14Fo

r pe

rson

al u

se o

nly.

differential pathways of psychological distress in spouses vs. parents of people with severe...

Documents