Reliability and validity data to support the clinical utility of the TBI-CareQOL

Abstract

Objective:

The TBI-CareQOL is a patient-reported outcome measurement system that is specific to caregivers of civilians and service members/veterans (SMVs) with traumatic brain injury (TBI). This measurement system includes 26 item banks that represent both generic (i.e., borrowed from existing measurement systems), and caregiver-specific components of health-related quality of life (HRQOL). This report provides reliability and validity data for measures within the TBI-CareQOL that have not previously been reported (i.e., four caregiver-specific and seven generic measures of HRQOL).

Design:

385 caregivers of persons with TBI completed caregiver-specific computer adaptive tests (CATs) for Feelings of Loss-Self, Caregiver Strain, Caregiver-Specific Anxiety, and Feeling Trapped, as well as generic measures of HRQOL from complementary measurement systems (i.e., Neuro-QoL Positive Affect and Well-Being; PROMIS Sleep-Related Impairment; NIH Toolbox Perceived Stress, General Life Satisfaction, and Self Efficacy; TBI-QOL Resilience and Grief/Loss). Caregivers also completed several additional measures to establish convergent and discriminant validity, as well as the Mayo Portland Adaptability Index-Fourth Edition.

Results:

Findings support the internal consistency reliability (all alphas > .85) and test-retest stability (all alphas > .73) of the TBI-CareQOL measures. Convergent validity was supported by moderate to high correlations between the TBI-CareQOL measures and related measures, while discriminant validity was supported by low correlations between the TBI-CareQOL measures and unrelated constructs. Known-groups validity was also supported.

Conclusions:

Findings support the reliability and validity of the item banks that comprise the TBI-CareQOL Measurement System. These measures should be considered for any standardized assessment of HRQOL in caregivers of civilians and SMVs with TBI.

Caring for an individual with traumatic brain injury (TBI) can have a profound impact on the caregiver’s health-related quality of life (HRQOL; Carlozzi et al., 2016; Carlozzi et al., 2015; Chronister, Chan, Sasson-Gelman, & Chiu, 2010; Griffin et al., 2017; Saban et al., 2016; Verhaeghe, Defloor, & Grypdonck, 2005), which is a multidimensional construct that reflects the impact a disease or disability has on physical, mental and social well-being (Cella, 1995; World Health Organization, 1946). Recently, a new patient-reported outcome system that is specific to caregivers of both civilians and service members/veterans (SMVs) with TBI was developed to capture the most important aspects of HRQOL for these individuals (Carlozzi et al., 2016; Carlozzi, Hanks, et al., 2018). This comprehensive measurement system, the TBI-CareQOL (Carlozzi, Kallen, Hanks, et al., 2018), includes 26 different measures of HRQOL that include both caregiver-specific and generic components of HRQOL (i.e., relevant for both caregivers and persons with chronic conditions; see Table 1 and Figure 1).

Table 1.

Summary descriptions of the measures that comprise the TBI-CareQoL

HRQOL Domain		Measure	Measure Description
Generic Domains (PROMIS, Neuro-QoL, NIH Toolbox, and TBI-QOL)	Negative Aspects of Mental Health	PROMIS Anger*	Angry mood, negative social cognition, and efforts to control anger
		PROMIS Anxiety*	Fear, hyperarousal, and somatic symptoms related to anxiety (e.g., racing heart)
		PROMIS Depression*	Negative mood, views of self, social cognition, and decreased engagement
		NIHTB Perceived Stress	Perceptions of one’s ability to cope with a situation given available resources
		TBI-QOL Grief/Loss	Sadness, and despair as a result of experiencing loss
	Positive Aspects of Mental Health	Neuro-QoL Positive Affect and Well-Being	Feelings of positive engagement, happiness, joy, and contentment
		NIHTB General Life Satisfaction	Evaluation of life experiences and whether one likes their life or not.
		NIHTB Self-Efficacy	Perceptions of one’s capacity to manage a stressor and have control over events
		TBI-QOL Resilience	Perceptions of one’s ability to adapt to life changes or challenging situations
	Physical Health	PROMIS Sleep Disturbance*	Perceived sleep quality, sleep depth, and feeling refreshed after sleeping
		PROMIS Fatigue*	Feelings of tiredness and exhaustion that interfere with completion of normal activities
		PROMIS Sleep-Related Impairment	Impairments during waking hours resulting from sleep problems; perceived alertness, sleepiness, and tiredness during usual waking hours
	Positive Aspects of Social Health	PROMIS Ability to Participate in Social Roles and Activities*	Perceptions of one’s ability to perform regular social roles and activities
		PROMIS Satisfaction with Social Roles and Activities*	Satisfaction with performing usual social roles and activities
		PROMIS Informational Support*	Availability of information or advice
		PROMIS Emotional Support*	Feelings of being cared for and having someone to confide in
	Negative Aspects of Social Health	PROMIS Social Isolation*	Feelings of being avoided or excluded by others
Caregiver-Specific Domains	Negative Aspects of Mental Health	TBI-CareQOL Caregiver-Specific Anxiety	Feelings of anxiety related to the caregiver role, and concern about the behavior and well-being of the person with TBI
		TBI-CareQOL Caregiver Strain	Feelings of being overwhelmed, stressed, self-defeated, downtrodden, or beat-down related to the caregiver role
		TBI-CareQOL Feelings of Loss - Self	Feelings of sorrow, mental suffering, or distress over changes that the caregiver has personally experienced related to the TBI
		TBI-CareQOL Feelings of Loss – Person with TBI	Caregiver feelings of sorrow, mental suffering, or distress over changes that the person with TBI has experienced, such as loss of functioning, potential future, or changes in behavior or personality
		TBI-CareQOL Emotional Suppression	Assesses hiding or suppressing negative emotions while acting in the caregiver role (e.g., to protect others or maintain functioning)
		TBI-CareQOL Caregiver Vigilance	Caregiver feelings related to anxiety, hyperarousal, or vigilance related to concerns regarding the emotional or behavioral status of the person with TBI
	Negative Aspects of Social Health	TBI-CareQOL Feeling Trapped	Feelings that one is unable to go places or do things due to caregiving responsibilities
		TBI-CareQOL Family Disruption	Feelings of disruption within the family as a result of the TBI
		TBI-CareQOL Military Health Care Frustration	Caregivers’ concern with health care services offered by the Department of Defense and Department of Veterans Affairs

^*= previous publications support the reliability and validity of these measures in caregivers of civilians and service members/veterans with TBI

Figure 1.

Components of the TBI-CareQOL Measurement System

Specifically, a call for research on caregivers (Verhaeghe et al., 2005), as well as several federal initiatives specifically focused on caregivers within the military system, including the Caregiver Curriculum (Bornstein, Baker, & Douglass, 1987), the Caregivers of Veterans report (Caregiving, 2010), and the 15-year Congressionally mandated study of caregivers of individuals with military-related TBI (substudy of section 721; 109th Congress) and the VA Caregiver Support Program prompted the development of a measurement system designed to include content that represents qualitative feedback about HRQOL from caregivers of civilians (Carlozzi et al., 2015) with TBI and caregivers of SMVs with TBI (Carlozzi et al., 2016). In order to ensure that both caregivers of civilians and caregivers of SMVs were adequately represented, the TBI-CareQOL Measurement System included multiple phases of development and validation testing. The first phase of TBI-CareQOL focused on the development of five caregiver-specific measures of HRQOL derived from caregiver focus groups and expert consensus: Caregiver Strain, Caregiver-Specific Anxiety, Feeling Trapped, Feelings of Loss-Self, and Feelings of Loss-Person with TBI. For each of these measures, development included a calibrated item bank that can be administered as a full item bank, a 6-item static short form (SF), or as a computerized adaptive test (CAT) or “smart test,” whereby item selection is based on the previous item response; the exact subset of items administered per individual is based on item response theory-based calibrated item parameters (van der Linden & Hambleton, 1997). CATs typically have the advantage of better precision and lower standard errors than associated static SFs, even when the number of items administered for each is identical (Lai et al., 2011).

The TBI-CareQOL measurement system offers several strengths over existing measures of HRQOL. This is the first measurement development effort that included input from key caregiver stakeholders in both the civilian and military communities, making it a sensitive tool for inclusion in both of these populations. As such, this measure bridges the gap between civilian and military efforts targeted towards caregivers by providing a measurement system that is based on input from both arenas. Second, this is the first time CAT methodology has been used to develop an HRQOL measure that is specific to caregivers; CATs are both efficient (given that only the most relevant items are administered to each respondent) and sensitive (they are programmed with predetermined stopping rules to ensure acceptable levels of error). In addition to CAT administration, other administration forms are available including a calibrated static short form (SF). Furthermore, given that item response theory (IRT) was used to develop these measures, scores can be generated, even if only a single item is administered; thus missing data is less problematic relative to measures that are developed using classical test theory methods. The measures themselves are both brief (taking less than a minute to administer) and comprehensive (including both caregiver-specific and generic content across multiple domains of functioning). This can be contrasted with most caregiver-specific measures that typically take longer to administer (they include more items), are less comprehensive (i.e., generally focus on only a single domain of function such as caregiver burden). Finally, this is the first measurement system that is specific to caregivers of persons with TBI; existing measures were developed for use in a different caregiving population (such as dementia of multiple sclerosis; e.g., the Caregiver Burden Inventory, Caregiver Burden Scale, Zarit Burden Interview, Caregiver Appraisal Scale; Lawton, Kleban, Moss, Rovine, & Glicksman, 1989; Montgomery, Gonyea, & Hooyman, 1985; Novak & Guest, 1989; Struchen, Atchison, Roebuck, Caroselli, & Sander, 2002; Zarit, Reever, & Bach-Peterson, 1980) and thus lack content that is specific for caring for an individual with TBI.

Previous publications from the first phase of data collection have highlighted preliminary reliability and validity data for both the SF and simulated CATs (i.e., CATs that were simulated using full item bank administration data) for these new measures using the data collected in the initial calibration sample of caregivers of civilians and SMVs with TBI (Carlozzi, Kallen, Hanks, et al., 2018; Carlozzi, Kallen, Ianni, et al., 2018; Carlozzi, Lange, et al., 2018). Those previous publications also highlighted the reliability and validity data for several generic measures of HRQOL within the TBI-CareQOL Measurement System; this included measures of mental HRQOL (PROMIS Anger, Anxiety, and Depression; Carlozzi, Hanks, et al., 2018), physical HRQOL (PROMIS Sleep Disturbance and Fatigue; Carlozzi, Ianni, Tulsky, et al., 2018), and social HRQOL (PROMIS Ability to Participate in Social Roles and Activities, Satisfaction with Social Roles and Activities, Social Support, Emotional Support, and Social Isolation; Carlozzi, Ianni, Lange, et al., 2018).

More recently, additional data from the second phase of data collection have highlighted the development and preliminary reliability and validity evidence for four additional caregiver-specific measures of HRQOL: Caregiver Vigilance (Carlozzi, Lange, Kallen, Boileau, Sander, Nakase-Richardson, Massengale, Hahn, et al., Under Review), Emotional Suppression (Carlozzi, Kallen, et al., Under Review), Family Disruption (Carlozzi, Lange, Boileau, et al., Under Review), and Health Care Frustration (Carlozzi, Lange, Kallen, Boileau, Sander, Nakase-Richardson, Massengale, Hanks, et al., Under Review). In addition to the development of these new measures, this second phase of data collection included the development and validation of several additional generic measures of HRQOL that are included in the TBI-CareQOL Measurement System. This included four additional measures of positive aspects of mental HRQOL (Neuro-QoL Positive Affect and Well-Being; NIH Toolbox General Life Satisfaction and Self Efficacy; TBI-QOL Resilience), two additional measures of negative aspects of mental HRQOL (NIH Toolbox Perceived Stress and TBI-QOL Grief/Loss), and one additional measure of physical HRQOL (PROMIS Sleep-Related Impairment).

The purpose of the current paper is to provide important reliability and validity data for the new measures of HRQOL that were developed in the first phase of this study and administered as CATs during the second phase of data collection (i.e., Caregiver Strain, Caregiver-Specific Anxiety, Feeling Trapped, and Feelings of Loss-Self), as well as the additional measures of mental and physical HRQOL (Neuro-QoL Positive Affect and Well-Being; NIH Toolbox Perceived Stress, General Life Satisfaction, and Self Efficacy; TBI-QOL Resilience and Grief/Loss; and PROMIS Sleep-Related Impairment). Specifically, we provide descriptive data for each of these TBI-CareQOL measures, as well as data regarding reliability (i.e., internal consistency reliability and test-retest stability) and validity (i.e., concurrent validity, discriminant validity, known-groups validity, and clinical impairment rates).

Methods

Study Participants

A total of 385 caregivers of civilians with TBI (n = 218) or SMVs with TBI (n = 167) completed the study measures. Of these individuals, 267 caregivers (186 of civilians and 81 of SMVs) completed a re-test approximately three weeks after their initial study visit. Recruitment targeted existing medical record data capture systems (Hanauer, Mei, Law, Khanna, & Zheng, 2015) and TBI model system databases to identify individuals with TBI who were then contacted to obtain contact information for their caregiver. In addition, TBI caregiver-specific research registries were used to identify potential participants. Community-based recruitment engaged local support groups for caregivers of persons with TBI (including the Hearts of Valor). Caregivers of civilians were recruited by the University of Michigan, TIRR Memorial Hermann, and the Rehabilitation Institute of Michigan. Caregivers of SMVs were recruited by the James A. Haley Veterans’ Hospital in Tampa and by community outreach at the University of Michigan. Retest participants were recruited through the University of Michigan, TIRR Memorial Hermann and the Rehabilitation Institute of Michigan. Individuals with TBI were contacted by study coordinators by mail, e-mail, or telephone to determine if they would share the contact information of an individual that helped them with day-to day activities that might be interested in participating in a research study. These potential caregivers were then contacted (using the provided contact information) to determine if they were interested in participation in this study. The following inclusion criteria were employed for all participants: caregivers must be ≥ 18 years old and able to read and understand English. For the civilian sample, caregivers had to be caring for an individual with a medically documented TBI who met TBI Model Systems criteria for a complicated mild, moderate, or severe TBI (Corrigan et al., 2012) was ≥ 16 years of age at the time of injury, and was ≥ 1 year post-injury at the time of recruitment (as the majority of recovery occurs within the first 12-months post-TBI, we excluded individuals who were caring for individuals that were still in the acute recovery stages to maximize the homogeneity of the sample; Burns, Marino, Flanders, & Flett, 2012; Dikmen et al., 2009; Dikmen, Machamer, Winn, & Temkin, 1995; Dikmen, Mclean, & Temkin, 1986; Dikmen, Reitan, & Temkin, 1983; Ditunno, Stover, Freed, & Ahn, 1992; Jorgensen et al., 1995a, 1995b; Waters, Adkins, Yakura, & Sie, 1998; Waters, Adkins, Yakura, & Sie, 1993, 1994; Waters, Yakura, Adkins, & Sie, 1992). For the SMV sample, caregivers had to be caring for an individual with medical documentation of a TBI from a U.S. Department of Defense or U.S. Department of Veteran Affairs (DoD/VA) treatment facility, and who was ≥ 18 years of age at the time of injury, and ≥ 1-year post-injury at the time of recruitment. Given existing prevalence rates of TBI severity in the military which indicate that the vast majority of TBIs are of mild severity (DVBIC, 2015) we included caregivers of SMVs with mild TBI in the sample assuming that the caregiver also indicated that they provided at least a minimal level of physical assistance, financial assistance, or emotional support to an individual with a TBI. Although these selections would not exclude individuals caring for someone with a comorbid physical or mental health condition, it is possible that this criterion has increased the likelihood that caregivers were providing care for an individual who is reporting long-term symptoms following mild TBI. However, the contribution of these factors towards the reasons for caregiving was not evaluated here. Level of assistance was evaluated using the following question, “On a scale of 0-10, where 0 is “no assistance” and 10 is “assistance with all activities”, how much assistance does the person you care for require from you to complete activities of daily living due to problems resulting from his/her TBI? Activities could consist of personal hygiene, dressing and undressing, housework, taking medications, managing money, running errands, shopping for groceries or clothing, transportation, meal preparation and cleanup, remembering things, etc.?”. Caregivers of SMVs had to indicate a response ≥1 on this question in order to be eligible to participate.

Study Procedures

All surveys were administered via Assessment Center^SM, an online data collection platform. Participants completed surveys either independently at home (using a personal computer, phone or tablet with an internet connection) or in-person with study staff using a study-specific computer or tablet. All sites obtained local institutional review board approval for the study. Caregivers provided informed consent prior to participation in study activities or a waiver was granted due to the low-risk nature of this study.

Study Measures

TBI-CareQOL Measures of HRQOL.

Caregivers completed the TBI-CareQOL Caregiver-Specific Anxiety (27 item bank), Caregiver Strain (33 item bank), Feeling Trapped (15 item bank), and Feelings of Loss – Self (30 item bank) measures which were calibrated in a large sample of caregivers of persons with TBI; item responses are rated on a 5 point Likert-type scale ranging from 1 (Never) to 5 (Always). These measures were completed as computer adaptive tests (CATs), which administer items based on participants’ previous responses until a score can be estimated with minimal standard error.

We also administered CAT and SF versions of PROMIS Sleep-Related Impairment (16 item bank; 8 item short form) (Cella et al., 2010) which was calibrated in the general population; item responses are rated on a 5-point Likert-type scale ranging from 1 (Not at all) to 5 (Very much). Additionally, we administered Neuro-QoL Positive Affect and Well-Being CAT (23 item bank) and SF (9 items) (Cella, Lai, Nowinski, Victorson, Peterman, Miller, Bethoux, Heinemann, Rubin, & Cavazos, 2012; Gershon et al., 2012), which was calibrated in combined sample (general population and persons with neurological disorders; item responses are rated on a 5-point Likert-type scale ranging from 1 (Never) to 5 (Always). Caregivers also completed the TBI-QOL Resilience and Grief/Loss 8-item SF (Tulsky et al., 2016), which was calibrated in a large sample of persons with TBI; item responses are rated on a 5-point Likert-type scale ranging from 1 (Never) to 5 (Always).

In addition, three CATs from the NIH Toolbox Emotional battery (which were calibrated in the general population; Gershon et al., 2013), were administered: General Life Satisfaction (10 item bank), Self-Efficacy (10 item bank), and Perceived Stress (10 item bank); item responses for General Life Satisfaction are on a 7-point Likert-type scale ranging from 1 (Strongly disagree) to 7 (Strongly agree) and item responses for Self-Efficacy and Perceived Stress are on a 5-point Likert-type scale ranging from 1 (Never) to 5 (Very often).

All scores on these measures are on a T-score metric (M = 50, SD = 10); higher scores indicate more of the named construct (e.g., higher scores for positively named constructs indicate better HRQOL, whereas negatively worded constructs indicate worse HRQOL). All analyses were completed using T scores.

Other Measures Included for Convergent and Discriminant Validity.

As part of analyses designed to establish convergent and discriminant validity, caregivers completed a generic measure of HRQOL, the Rand-12 (Grandmougin et al., 1996), in addition to a well-established measure of caregiver appraisals, and the Caregiver Appraisal Scale (CAS; Lawton et al., 1989). The Rand-12 is a 12-item self-report measure that includes a Mental Health Composite score and a Physical Health Composite score. These scores are on a T-score metric, with higher scores indicating better HRQOL. The Caregiver Appraisal Scale (CAS; Lawton et al., 1989) is a 47-item measure developed to assess positive and negative perceptions of the caregiver role. The CAS was scored according to Struchen and colleagues’ (2002) recommendations (based on responses to 35 items), which includes four subdomain scores: Perceived Burden, Caregiver Satisfaction, Caregiver Ideology, and Caregiver Mastery. Sum scores are generated for each subdomain, as well as a total score. Higher scores indicate more positive perceptions of the caregiver role. Furthermore, we administered measures from the TBI-CareQOL that have been validated in previous studies for use in our current study to support convergent and discriminant validity of the measures mentioned above, including: PROMIS Anxiety, Fatigue, Social Isolation, and Ability to Participate in Social Roles and Activities (Carlozzi, Hanks, et al., 2018; Carlozzi, Ianni, Lange, et al., 2018; Carlozzi, Ianni, Tulsky, et al., 2018). These measures are on a T-score metric; again, higher scores indicate more of the named construct.

Functional Ability of the Person with TBI.

The Mayo-Portland Adaptability Inventory-Fourth Edition (MPAI-4; Malec, 2005) was used to provide an estimate of the functional ability of individuals with TBI (i.e., their post- injury functional status). This 35-item measure captures the caregiver’s perceptions of the functional ability of the person with TBI. Resulting scores are on a T-score metric (M = 50, SD = 10). Lower scores (< 60) on the MPAI-4 indicate higher functioning, and scores ≥ 60 indicate significant functional impairment (Malec, 2005). This measure was used to examine known-groups validity.

Data Analysis

All statistical analyses were conducted using SAS 9.4 (SAS Institute, 2013). To assess normality, the distribution of the response data for each of the assessment measures was examined for skewness and kurtosis prior to conducting statistical analyses. Using Bulmer’s criteria (Bulmer, 1979), we assessed whether the data were highly skewed or had a large kurtosis. Skewness ranged from ∣0.02∣ (for Feeling Trapped and Perceived Stress) to ∣0.60∣ (for Caregiver-Specific Anxiety, and kurtosis ranged from ∣0.08∣ (for Caregiver Strain) to ∣0.78∣ (for Feeling Trapped) for the overall sample. In addition, when stratified by military status, all HRQOL measures met Bulmer’s criteria. As each measure had a normal distribution (skewness < 1 and kurtosis < 3), parametric analyses were used in the data analyses, described below.

Reliability.

In order to establish evidence of internal consistency reliability, Cronbach’s alphas were calculated for all SFs, and a comparable item response theory-based internal consistency reliability (i.e., average standard error-based reliability) was calculated for the CATs. Specifically, the item response theory-based reliability estimates are unique, score-level-based estimates of precision. These estimates are determined by using the individual contribution of each item responded to, per individual respondent, in the course of a CAT administration of a measure. Each item contributes a specific quantity of mathematical “information,” dependent on how much information is actually available from the item at the respondent’s estimated score level. Item information is then summed across all administered items and becomes the key component for estimating each individual respondent’s score-level standard error and score-level reliability. This is then used to obtain “mean” score-level estimates for the CAT administrations, averaging across individual respondents score-level reliability estimates. We specified minimal acceptable reliability as ≥ 0.70 (Cohen, 1988; DeVellis, 2017). We also examined 3-week test-retest stability for the participants that completed the retest. Test-retest reliability was assessed using the intraclass correlation coefficient (ICC; random two-way consistency). Within the 3-week time frame, we specified our minimum test-retest stability reliability coefficient to be ≥ 0.70.

Floor and Ceiling Effects.

We calculated floor and ceiling effects for all study measures. Floor effects were defined as the percentage of participants who scored the best score possible (lowest score for negative constructs; highest score for positive constructs); ceiling effects were defined as the percentage of participants who scored the worst scores possible. We specified acceptable floor and ceiling rates as ≤ 20% (Andresen, 2000; Cramer & Howitt, 2004).

Administration time.

Administration times were captured using the online data collection platform (Assessment Center^SM) .

Convergent and Discriminant Validity.

Evidence supporting convergent and discriminant validity was established by examining Pearson correlations between similar and dissimilar traits (Campbell & Fiske, 1959). Specifically, we looked at how positive affect, negative affect, and physical measures correlated to measures within the same construct and with more generic measures of HRQOL. Evidence for convergent validity can be established when correlations between scores from measures of the same trait are strong (e.g., r > 0.60; Campbell & Fiske, 1959). Evidence for discriminant validity can be established when correlations between scores of different traits are poor (e.g., r < 0.30; Campbell & Fiske, 1959).

Known-groups validity.

Caregivers were split into two groups based on the MPAI-4 scores for the individual with TBI: Scores < 60 were considered “high functioning,” and scores ≥ 60 were considered “low functioning.” This cutoff was chosen because scores of 60 or greater indicate severe limitations in functioning (Malec, 2005). T-tests were used to compare low vs. high-functioning groups. The Folded F test was used to determine whether variances were equal between the two groups; pooled degrees of freedom method was used when variances were equal, and Satterthwaite degrees of freedom method was used when variances were unequal. We hypothesized that caregivers of high-functioning individuals would report better HRQOL compared to those who cared for low-functioning individuals for all measures except Resilience and Self-Efficacy (where previous research indicate sthat these factors are stable absent intervention and unrelated to disease symptoms in the care recipient; Beck & Lund, 1981; Dias et al., 2016).

Base Rates for Severe Symptoms/Impairment. Severe symptoms (i.e., participants whose scores were > 1 SD worse than the mean of the normative sample of each measurement system ([TBI-CareQOL N = 560; M = 50, SD = 10; (Carlozzi, Kallen, Hanks, et al., 2018), [TBI-QOL N = 675, M = 50, SD = 10]; (Tulsky et al., 2016), [NIH Toolbox N = 4,859, M = 50, SD = 10]; (Gershon et al., 2013), [Neuro-QoL N = 3,247, M = 50, SD = 10; (Cella, Lai, Nowinski, Victorson, Peterman, Miller, Bethoux, Heinemann, Rubin, Cavasos, et al., 2012; Gershon et al., 2012), and [PROMIS N = 21,133, M = 50, SD = 10]; (Cella et al., 2010)) were evaluated to determine if caregivers of individuals with TBI were at greater risk than the United States general population for severe symptoms/impairment. We expect that, based on the mean (i.e., 16% of the US general population have severe symptoms/impairment); thus, rates greater than 16% would indicate greater symptom severity than expected.

Effect sizes.

Effect sizes for TBI-CareQOL scores were computed for the high-functioning and low-functioning groups using Cohen’s d effect size estimation (Cohen, 1988). Values of d between 0.20 and 0.39 were considered “small”, values between 0.40 and 0.59 were considered “moderate,” and values greater than or equal 0.60 were considered “large”. Except for Resilience and Self-Efficacy, we expected effect sizes to be larger for caregivers caring for individuals with greater functional impairment than those caring for individuals with less functional impairment.

Results

Study Participants

Three hundred and eighty-four caregivers of civilians with TBI (n = 218) or SMVs with TBI (n = 167) participated in this study. Caregivers of civilians were primarily female (78%), white (70%), and non-Hispanic (90.8%). Approximately 45.4% had a college degree, 29.4% had some college, 20.2% had a high school degree, and 5% had less than a high school degree. The mean age was 51 years, and 32% of civilian caregivers were spouses or partners, 38% were parents, 19% were other family members, and 10% were friends or some other relationship to the person with TBI. Caregivers of civilians were in the caregiver role for an average of seven years (range 4 months to 26 years) and were caring for an individual who was, on average, was 43 years of age. Care recipients were 72% male and an average of 10 years post-injury. Overall, 57.3% were classified as severe TBI, 17.4% were classified as moderate TBI and 25.2% were complicated mild TBI.

Caregivers of SMVs with TBI were primarily female (94%) white (85%), and non-Hispanic (89.8%). Approximately 50.9% of military caregivers had a college degree, 33.5% had some college education, 10.8% had a high school degree, and 4.8% had less than a high school degree. The mean age was 44 years, and 70% of military caregivers were spouses, followed by parents (20%), other family members (8%), and some other relationship to the person with TBI (e.g., friends; 2%). The average length of providing care was seven years (range = 1 year to 26 years). The SMV was on average 39 years of age and 8 years post-injury; 96% of SMVs were male. TBI severity data were not available for the military participants recruited through community outreach at the University of Michigan (57% of the military sample). For the remaining participants, medical documentation indicated that 1.2% were uncomplicated mild, 2.4% were complicated mild, 5.4% were moderate, 25.8% were severe TBI. In addition, 8.4% were characterized as penetrating TBI. Given that the University of Michigan sample was recruited primarily from the community, we assume that most of these individuals (i.e., >80%) would have mild TBI (given established prevalence rates(DVBIC, 2015)).

Reliability (Table 2).

Table 2.

Descriptive Information and Reliability Data for TBI-CareQOL Measures

Combined Sample (N = 385)
	Internal consistency*	Test-retest stability**	Floor %	Ceiling %	Administration time (seconds) M (SD)	Average Administration time per item (seconds)
Caregiver-Specific HRQOL
TBI-CareQOL Caregiver-Specific Anxiety CAT	0.93	0.93	4.18	0.26	40.08 (26.72)	8.16
TBI-CareQOL Caregiver Strain CAT	0.92	0.93	5.21	0.26	32.26 (22.70)	6.35
TBI-CareQOL Feeling Trapped CAT	0.92	0.94	14.32	0.26	41.53 (27.41)	7.38
TBI-CareQOL Feelings of Loss-Self CAT	0.92	0.92	10.42	0.52	40.59 (28.04)	6.96
Generic HRQOL Concepts
TBI-QOL Resilience SF	0.91	N/A***	7.12	0.00	42.93 (25.05)	5.37
TBI-QOL Grief/Loss SF	0.94	0.88	20.16	1.05	34.71 (19.21)	4.96
Neuro-QoL Positive Affect and Well Being CAT	0.96	0.83	4.22	0.00	21.81 (14.82)	5.04
Neuro-QoL Positive Affect and Well Being SF	0.95	0.87	6.60	0.00	38.70 (21.42)	4.84
NIH Toolbox General Life Satisfaction CAT	0.93	0.88	4.51	0.27	30.15 (19.21)	6.88
NIH Toolbox Perceived Stress CAT	0.87	0.92	1.84	0.00	90.14 (45.25)	9.01
NIH Toolbox Self-Efficacy CAT	0.90	0.78	0.00	0.26	42.30 (25.99)	6.99
PROMIS Sleep Related Impairment CAT	0.93	0.90	1.32	0.26	31.95 (26.54)	5.67
PROMIS Sleep Related Impairment SF	0.95	0.91	3.41	0.52	41.88 (26.67)	5.23
Caregivers of Civilians with TBI (n = 218)
	Internal consistency*	Test-retest stability**	Floor %	Ceiling %	Administration time (seconds) M (SD)	Average Administration time per item (seconds)
Caregiver-Specific HRQOL
TBI-CareQOL Caregiver-Specific Anxiety CAT	0.92	0.91	6.91	0.00	46.57 (30.13)	8.99
TBI-CareQOL Caregiver Strain CAT	0.92	0.90	8.26	0.00	37.55 (24.48)	6.82
TBI-CareQOL Feeling Trapped CAT	0.90	0.90	22.48	0.00	48.08 (29.36)	7.58
TBI-CareQOL Feelings of Loss-Self CAT	0.90	0.92	16.06	0.46	46.03 (28.26)	7.30
Generic HRQOL Concepts
TBI-QOL Resilience SF	0.91	N/A***	8.80	0.00	44.60 (23.37)	5.58
TBI-QOL Grief/Loss SF	0.93	0.86	24.77	0.00	37.57 (20.81)	5.37
Neuro-QoL Positive Affect and Well Being CAT	0.96	0.84	6.48	0.00	23.94 (15.18)	5.41
Neuro-QoL Positive Affect and Well Being SF	0.95	0.86	9.72	0.00	42.42 (23.56)	5.30
NIH Toolbox General Life Satisfaction CAT	0.93	0.86	3.26	0.47	33.50 (17.51)	7.62
NIH Toolbox Perceived Stress CAT	0.86	0.91	1.38	0.00	97.27 (49.06)	9.73
NIH Toolbox Self-Efficacy CAT	0.90	0.77	6.94	0.46	47.64 (28.8)	7.82
PROMIS Sleep Related Impairment CAT	0.92	0.85	1.86	0.00	36.04 (29.02)	6.00
PROMIS Sleep Related Impairment SF	0.95	0.85	3.69	0.00	44.24 (26.95)	5.53
Caregivers of Service Members/Veterans with TBI (n = 167)
	Internal consistency*	Test-retest stability**	Floor %	Ceiling %	Administration time (seconds) M (SD)	Average Administration time per item (seconds)
Caregiver-Specific HRQOL
TBI-CareQOL Caregiver-Specific Anxiety CAT	0.94	0.92	0.60	0.60	31.58 (18.32)	7.07
TBI-CareQOL Caregiver Strain CAT	0.93	0.92	1.20	0.60	25.29 (17.92)	5.73
TBI-CareQOL Feeling Trapped CAT	0.94	0.88	3.61	0.60	32.91 (21.89)	7.11
TBI-CareQOL Feelings of Loss-Self CAT	0.93	0.87	3.01	0.60	33.53 (26.18)	6.51
Generic HRQOL Concepts
TBI-QOL Resilience SF	0.92	N/A***	4.91	0.00	40.69 (27.05)	5.09
TBI-QOL Grief/Loss SF	0.95	0.85	14.02	2.44	30.89 (16.14)	4.41
Neuro-QoL Positive Affect and Well Being CAT	0.96	0.73	1.23	0.00	19.03 (13.90)	4.55
Neuro-QoL Positive Affect and Well Being SF	0.95	0.84	2.45	0.00	33.70 (16.98)	4.21
NIH Toolbox General Life Satisfaction CAT	0.93	0.91	6.17	0.00	25.73 (12.78)	5.92
NIH Toolbox Perceived Stress CAT	0.87	0.88	2.45	0.00	80.63 (37.70)	8.06
NIH Toolbox Self-Efficacy CAT	0.90	0.76	8.54	0.00	35.19 (19.59)	5.89
PROMIS Sleep Related Impairment CAT	0.93	0.90	0.61	0.61	26.53 (21.75)	5.24
PROMIS Sleep Related Impairment SF	0.95	0.92	3.05	1.22	38.72 (26.05)	4.84

Note. Mean retest interval is 19.7 days (SD = 6.5)

^*For SFs = Cronbach’s alpha; for CATs = mean standard error-based reliability

^**Test-retest stability assessed using ICC’s (criterion ≥0.70)

^***Resilience was not administered as part of the retest

For both caregiver groups, all of the TBI-CareQOL measures exceeded a priori standards for internal consistency reliability (≥ 0.70). Internal consistency reliabilities for the TBI-CareQOL measures were ≥ 0.86 for all measures. Test-retest stability was ≥ 0.73 for all measures.

Floor and Ceiling Effects (Table 2).

All measures were free of ceiling effects for both samples. In addition, all measures except TBI-CareQOL Feeling Trapped CAT and TBI-QOL Grief/Loss SF were free of floor effects.

Administration Time (Table 2).

Completion time for all measures was less than one minute, with the exception of the NIH Perceived Stress CAT, which took approximately 90 seconds on average for participants to finish.

Convergent and Discriminant Validity (Table 3).

Table 3.

Convergent and Discriminant Validity

Combined Sample (N = 385)
		Discriminant Validity
TBI-CareQOL Measure	Convergent Validity	RAND-12 Physical Component	CAS Satisfaction	CAS Ideology	CAS Mastery
TBI-CareQOL Specific Anxiety CAT	PROMIS Anxiety 0.63	−0.21	−0.40	0.07*	−0.31
TBI-CareQOL Caregiver Strain CAT	CAS Burden −0.79	−0.24	−0.24	0.07*	−0.37
TBI-CareQOL Feeling Trapped CAT	PROMIS Social Isolation 0.68 PROMIS Ability to Participate in SRA −0.75	−0.21	−0.14	0.15	−0.31
TBI-CareQOL Feelings of Loss - Self CAT	TBI-QOL Grief/Loss 0.75	−0.28	−0.30	0.11	−0.33
TBI-QOL Resilience SF	NIH Self-Efficacy 0.62 Neuro-QoL Positive Affect and Well-Being 0.66	0.25	0.32	−0.04*	0.35
Neuro-QoL Positive Affect and Well Being CAT	NIH General Life Satisfaction 0.75	0.25	0.31	0.01*	0.31
Neuro-QoL Positive Affect and Well-Being SF	NIH General Life Satisfaction 0.74	0.25	0.35	0.04*	0.34
NIH General Life Satisfaction CAT	Neuro-QoL Positive Affect and Well-Being 0.75	0.29	0.34	0.04*	0.36
NIH Perceived Stress CAT	TBI-CareQOL Caregiver Strain 0.68 CAS Burden −0.70	−0.23	−0.35	0.07*	−0.45
NIH Self-Efficacy CAT	TBI-QOL Resilience 0.62	0.22	0.30	−0.06*	0.39
PROMIS Sleep Related Impairment CAT	PROMIS Fatigue 0.84	RAND-12 Mental Health −0.70	−0.26	0.06*	−0.38
PROMIS Sleep Related Impairment SF	PROMIS Fatigue 0.86	RAND-12 Mental Health −0.70	−0.25	0.06*	−0.39
Caregivers of Civilians with TBI (n = 218)
		Discriminant Validity
TBI-CareQOL Measure	Convergent Validity	RAND-12 Physical Component	CAS Satisfaction	CAS Ideology	CAS Mastery
TBI-CareQOL Specific Anxiety CAT	PROMIS Anxiety 0.59	−0.16	−0.43	0.03*	−0.31
TBI-CareQOL Caregiver Strain CAT	CAS Burden −0.76	−0.21	−0.28	0.01*	−0.41
TBI-CareQOL Feeling Trapped CAT	PROMIS Social Isolation 0.58 PROMIS Ability to Participate in SRA −0.71	−0.21	−0.19	0.10*	−0.35
TBI-CareQOL Feelings of Loss - Self CAT	TBI-QOL Grief/Loss 0.72	−0.30	−0.34	0.07*	−0.38
TBI-QOL Resilience SF	NIH Self-Efficacy 0.62 Neuro-QoL Positive Affect and Well-Being 0.68	0.26	0.26	−0.14	0.35
Neuro-QoL Positive Affect and Well Being CAT	NIH General Life Satisfaction 0.71	0.26	0.27	−0.04*	0.34
Neuro-QoL Positive Affect and Well Being SF	NIH General Life Satisfaction 0.71	0.24	0.29	0.01*	0.35
NIH General Life Satisfaction CAT	Positive Affect and Well-Being 0.71	0.31	0.50	0.00*	0.40
NIH Perceived Stress CAT	TBI-CareQOL Caregiver Strain 0.65 CAS Burden −0.63	−0.14	−0.32	0.04*	−0.46
NIH Self-Efficacy CAT	TBI-QOL Resilience 0.62	0.22	0.26	−0.09*	0.39
PROMIS Sleep Related Impairment CAT	PROMIS Fatigue 0.78	RAND-12 Mental Health −0.64	−0.21	0.05*	−0.37
PROMIS Sleep Related Impairment Short Form	PROMIS Fatigue 0.80	RAND-12 Mental Health −0.66	−0.22	0.04*	−0.38
Caregivers of Service Members/Veterans with TBI (n = 167)
		Discriminant Validity
TBI-CareQOL measure	Convergent Validity	RAND-12 Physical CAS	CAS Satisfaction	CAS Ideology	CAS Mastery
TBI-CareQOL Specific Anxiety CAT	PROMIS Anxiety 0.62	−0.36	−0.54	−0.02*	−0.30
TBI-CareQOL Caregiver Strain CAT	CAS Burden −0.76	−0.33	−0.34	0.00*	−0.30
TBI-CareQOL Feeling Trapped CAT	Social Isolation 0.69 Ability to Participate in SRA −0.72	−0.20*	−0.27	0.04*	−0.26
TBI-CareQOL Feelings of Loss – Self CAT	TBI-QOL Grief/Loss 0.77	−0.20*	−0.37	0.04*	−0.22
TBI-Qol Caregiver Resilience SF	NIH Self-Efficacy 0.64 Neuro-QoL Positive Affect and Well-Being 0.63	0.22	0.42	0.14*	0.35
Neuro-QoL Positive Affect and Well Being CAT	NIH General Life Satisfaction 0.81	0.21	0.43	0.14*	0.25
Neuro-QoL Positive Affect and Well Being Short Form	NIHTB General Life Satisfaction 0.80	0.26	0.49	0.14*	0.31
NIH General Life Satisfaction CAT	Positive Affect and Well-Being 0.81	0.23	0.49	0.15*	0.29
NIH Perceived Stress CAT	TBI-CareQOL Caregiver Strain 0.70 CAS Burden −0.76	−0.43	−0.45	0.03*	−0.43
NIH Self-Efficacy CAT	TBI-QOL Resilience 0.64	0.24	0.35	−0.01*	0.40
PROMIS Sleep Related Impairment CAT	PROMIS Fatigue 0.87	RAND-12 Mental Health −0.61	−0.41	−0.06*	−0.38
PROMIS Sleep Related Impairment Short Form	PROMIS Fatigue 0.90	RAND-12 Mental Health −0.61	−0.38	−0.03*	−0.39

Note: All p-values < .05 unless otherwise indicated. PROMIS = Patient Reported Outcomes Measurement Information System; Neuro-QoL = Quality of Life in Neurological Disorders; TBI-QOL = Traumatic Brain Injury - Quality of Life; TBI-CareQOL = Traumatic Brain Injury Caregiver Quality of Life; NIH = National Institutes of Health Toolbox; CAS = Caregiver Appraisal Scale; CAT = Computer Adaptive Test; SF = Short Form

^*p>.05

Convergent and discriminant validity were supported for the TBI-CareQOL measures. Specifically, measures demonstrated the strongest correlations (r > 0.60) with item banks of similar constructs, supporting convergent validity, whereas correlations with measures of different constructs were smaller in magnitude and generally negligible to small. Exceptions to the proposed hypotheses were the unexpected strong correlations between Sleep-Related Impairment and Rand-12 Mental Health.

Known-Groups Validity (Table 4).

Table 4.

Known Groups Validity for TBI-CareQOL Measures

	Caregiver of a High Functioning Individual		Caregiver of a Low Functioning Individual
	(MPAI-4 <60)		(MPAI-4 ≥60)
	Mean (SD)	% Severe Symptoms/ Impairment*	Mean (SD)	% Severe Symptoms/ Impairment*	Cohen's d Effect Size	t	p
Civilian sample	n = 163		n = 26
Caregiver-Specific Anxiety CAT	49.59 (10.25)	14.20	55.08 (6.92)	23.08	−0.63	−3.48	0.0011
Caregiver Strain CAT	45.65 (9.92)	6.13	55.72 (9.79)	34.62	−1.02	−4.81	<.0001
Feeling Trapped CAT	44.21 (7.70)	3.07	55.06 (5.81)	19.23	−1.59	−6.87	<.0001
Feelings of Loss – Self CAT	44.96 (10.05)	5.52	55.79 (8.71)	42.31	−1.15	−5.19	<.0001
Resilience SF^	49.32 (8.88)	16.15	46.18 (9.05)	19.23	0.35	1.67	0.0962
Grief/Loss SF	42.53 (8.98)	2.45	50.95 (9.24)	11.54	−0.92	−4.43	<.0001
Positive Affect and Well-Being CAT^	53.08 (7.78)	3.73	49.42 (4.68)	3.85	0.57	3.32	0.0017
Positive Affect and Well-Being SF^	53.51 (7.39)	2.48	50.08 (5.52)	3.85	0.52	2.26	0.0250
General Life Satisfaction CAT^	49.62 (9.42)	13.13	41.03 (7.93)	50.00	0.99	4.40	<.0001
Perceived Stress CAT	46.91 (11.33)	12.27	54.58 (10.06)	23.08	−0.72	−3.25	0.0014
Self-Efficacy CAT^	49.75 (10.46)	16.77	49.62 (9.73)	15.38	0.01	0.06	0.9507
Sleep-Related Impairment CAT	51.00 (10.67)	22.50	58.60 (9.26)	34.62	−0.76	−3.42	0.0008
Sleep-Related Impairment SF	49.62 (10.66)	17.90	57.37 (9.08)	30.77	−0.78	−3.50	0.0006
SMV sample	n = 72		n = 85
Caregiver-Specific Anxiety CAT	52.60 (7.94)	20.55	59.34 (7.25)	46.43	−0.75	−4.73	<.0001
Caregiver Strain CAT	51.46 (9.15)	17.81	57.81 (7.51)	38.10	−0.76	−4.77	<.0001
Feeling Trapped CAT	50.68 (7.85)	13.89	58.15 (5.97)	47.06	−1.07	−6.62	<.0001
Feelings of Loss – Self CAT	49.58 (9.86)	12.50	55.71 (8.20)	30.59	−0.68	−4.25	<.0001
Emotional Suppression CAT	49.51 (8.95)	11.24	54.15 (8.78)	20.74	−0.52	−4.53	<.0001
Resilience SF^	49.56 (7.67)	9.72	47.79 (9.11)	21.95	0.21	1.29	0.1979
Grief/Loss SF	45.44 (9.26)	5.56	51.14 (10.13)	18.07	−0.59	−3.63	0.0004
Positive Affect and Well-Being CA^	53.21 (6.78)	4.17	49.03 (6.70)	8.54	0.62	3.84	0.0002
Positive Affect and Well-Being SF^	53.33 (6.38)	1.39	49.22 (6.44)	4.88	0.64	3.97	0.0001
General Life Satisfaction CA^	49.45 (9.17)	11.27	45.18 (9.42)	23.17	0.46	2.83	0.0053
Perceived Stress CAT	49.61 (11.53)	18.31	56.19 (11.52)	38.55	−0.57	−3.53	0.0005
Self-Efficacy CA^	50.50 (9.18)	16.67	51.64 (10.73)	14.46	−0.11	−0.71	0.4777
Sleep-Related Impairment CAT	54.97 (10.99)	43.06	62.09 (9.62)	65.06	−0.69	−4.30	<.0001
Sleep-Related Impairment SF	55.79 (10.56)	41.18	61.31 (9.68)	61.82	−0.54	−4.11	<.0001

^{^}Higher scores indicate better outcomes

^*Severe symptoms/impairment is defined as scores 1 standard deviation greater or less than the mean. In a normal distribution we would expect percentages to be approximately 16%

Findings for known-groups validity were consistent with our proposed hypotheses. Specifically, caregivers of individuals of lower functioning had worsen HRQOL relative to those caring for individuals that were higher functioning; this was true for both samples.

Base Rates for Severe Symptoms/Impairment (Table 4).

As hypothesized, caregivers of low-functioning persons with TBI consistently had higher symptom severity/impairment rates than those that were caring for a high-functioning individual with TBI. As hypothesized, TBI-CareQOL measures generally exhibited moderate to strong Cohen’s d effect sizes. Also as hypothesized, there were negligible and small effect sizes for NIHTB Self-Efficacy and TBI-QOL Resilience, respectively.

Discussion

The purpose of this paper is to provide important reliability and validity data for measures within TBI-CareQOL Measurement System. This includes data on the Caregiver Strain, Caregiver-Specific Anxiety, Feeling Trapped, and Feelings of Loss-Self CATs, as well as additional measures of mental and physical HRQOL (Neuro-QoL Positive Affect and Well-Being; NIH Toolbox Perceived Stress, General Life Satisfaction, and Self Efficacy; TBI-QOL Resilience and Grief/Loss; and PROMIS Sleep-Related Impairment).

Findings provide support for the internal consistency reliability of these measures, which was excellent for all TBI-CareQOL Measurement System measures (all alpha ≥ 0.86). In addition, with the exception of a small floor effect for the TBI-QOL Grief/Loss measure (20.16% were at the floor, for the full sample and 24.77% for the civilian sample whereas a priori standards were specified as ≤ 20%) and a small floor effect for TBI-CareQOL Feeling Trapped (22.48% for the civilian sample only) all measures were also devoid of floor or ceiling effects. Administration times were brief: All measures except the NIH Toolbox Perceived Stress CAT took less than 60 seconds to complete. Test-retest stability met a priori standards for all measures: All measures exhibited very good (i.e., ≥ .80) or excellent (i.e., ≥ .90) test-retest reliability except for the NIH Toolbox Self-Efficacy CAT, which demonstrated adequate stability (i.e., r = 0.78 for the full sample).

Convergent and discriminant validity were also supported for the TBI-CareQOL measures. In all cases, measures demonstrated the strongest correlations (r > 0.60) with item banks of similar constructs, supporting convergent validity, whereas correlations with measures of different constructs were smaller in magnitude (supporting discriminant validity). The one unanticipated finding was the stronger than anticipated relationship between Sleep-Related Impairment and the Rand-12 MHC (correlations were −0.70 for both the CAT and SF), which would suggest that this measure is tapping both fatigue (as evident by the correlations with PROMIS Fatigue) as well as general mental health. Given the strong relationship between sleep quality and mental health including caregivers in other disease groups, this finding may not be surprising (Hertenstein et al., 2019; Ornstein & Gaugler, 2012).

Findings also supported known-groups validity for all of the TBI-CareQOL measures. As hypothesized, caregivers of high-functioning individuals reported better HRQOL compared to those who cared for low-functioning individuals for all measures except Resilience and Self-Efficacy (where previous research indicate sthat these factors are stable absent intervention and unrelated to disease symptoms in the care recipient; Beck & Lund, 1981; Dias et al., 2016). And finally, caregivers of low-functioning persons with TBI consistently had higher base rates for severe symptoms/impairment than those that were caring for a high-functioning individual with TBI. Most measures exhibited moderate to strong Cohen’s d effect sizes (except for the negligible and small effect sizes for NIH Self-Efficacy and TBI-QOL resilience, which were also as hypothesized).

The TBI-CareQOL provides clinicians and researchers with a well validated set of measures that can be used to evaluate the impact of interventions focused on persons with TBI and their families. In fact, these measures are well matched to evaluate educational interventions that prioritize caregivers’ needs including the physical, cognitive, and emotional/behavioral changes that they encounter caring for an individual with TBI (J. Kreutzer, Serio, & Bergquist, 1994; Serio, Kreutzer, & Gervasio, 1995; Witol, Sander, & Kreutzer, 1996).

In addition, these measures could be used in larger trials of interventions that have demonstrated some preliminary effectiveness in early caregiver trails; this includes cognitive behavioral therapy (Sander, 2005), holistic family interventions (J. S. Kreutzer, Marwitz, Sima, & Godwin, 2015; J. S. Kreutzer et al., 2009; Niemeier, Kreutzer, Marwitz, & Sima, 2018), and problem-solving training (Powell, Fraser, Brockway, Temkin, & Bell, 2016; Rivera, Elliott, Berry, & Grant, 2008). Furthermore, these measures reflect HRQOL concepts that have been identified by caregivers as being important (Carlozzi et al., 2016) that might lend themselves to future interventions targeted at concepts such as hypervigilance and emotional suppression (Sander, Boileau, Hanks, Tulsky, & Carlozzi, 2019).

While this study exhibits several strengths, including the large sample size and the very comprehensive evaluation of HRQOL in caregivers, it is also important to recognize several study limitations. First, caregivers in the present study were primarily female. Thus, future work is needed to confirm that findings are generalizable to male caregivers. In addition, caregivers had to be providing care for individuals who were at least 1 year post-injury. Thus, future work is also needed to determine if findings are relevant to persons providing care in the more acute stages of TBI recovery. Furthermore, documentation of TBI severity was not available for a large portion of the study sample, and, thus, future work should focus on the examination of TBI severity in relationship to caregiver HRQOL. Furthermore, there are important differences between the civilian and SMV samples (i.e., the inclusion of mild TBI in the SMV cohort, as well as differences in demographic characteristic) preclude a direct comparison of these two groups. Future work is needed in comparable samples in order to better understand how these groups compare to one another in terms of their HRQOL experience. Future work is also needed to examine the longitudinal validity of these measures to support their clinical utility for repeated-assessment study designs. Finally, the majority of caregivers in our sample were white, particularly for the military sample. Racial/ethnic differences in the validity of the TBI-CareQOL measures, as well as the generic HRQOL measures, should be investigated.

These findings, in conjunction with previously published findings for other measures within the TBI-CareQOL measurement system (Carlozzi, Hanks, et al., 2018; Carlozzi, Ianni, Lange, et al., 2018; Carlozzi, Ianni, Lange, et al., 2018; Carlozzi, Ianni, Tulsky, et al., 2018; Carlozzi, Kallen, et al., Under Review; Carlozzi, Lange, Boileau, et al., Under Review; Carlozzi, Lange, et al., 2018; Carlozzi, Lange, Kallen, Boileau, Sander, Nakase-Richardson, Massengale, Hahn, et al., Under Review; Carlozzi, Lange, Kallen, Boileau, Sander, Nakase-Richardson, Massengale, Hanks, et al., Under Review) provide strong support for the clinical utility of this system’s measures. As such, clinicians and researchers should feel comfortable selecting appropriate TBI-CareQOL measures for use as meaningful endpoints in trials focused on improving outcomes in persons with TBI and their informal family caregivers. In addition, these measures have recently been adopted for inclusion as part of the 15-year Congressionally mandated study of caregivers of service members/veterans with TBI (substudy of section 721; 109th Congress), and are publically available through healthmeasures.net, as well as through a study-specific website for broad use in both civilian and military populations. It is our hope that TBI-CareQOL Measurement System can provide the foundation for a unified set of HRQOL measures that can be used to comprehensively assess the most important aspects of HRQOL for caregivers of both civilians and SMVs with TBI. This measurement system should also be helpful in planning long-term rehabilitation needs and support services for friends and family of those with TBI. Future work examining the sensitivity of these measures to change over time, as well as to treatment, and establishing minimal important differences will further foster the clinical utility of this measurement system.

Implications:

• Understanding the impact that caring for an individual with traumatic brain injury has on the caregiver can help inform the treatment needs and identify appropriate support services for these individuals and their families.

• Reliable and valid assessments of health-related quality of life can be used to better understand the needs of caregivers of persons with TBI.

List of Abbreviations:

CAS

Caregiver Appraisal Scale

CAT

Computer Adaptive Test

HRQOL

Health-Related Quality of Life

Neuro-QoL

Quality of Life in Neurological Disorders

NIH

National Institutes of Health

NIHTB

National Institutes of Health Toolbox

PROMIS

Patient-Reported Outcomes Measurement Information System

SF

Short Form

SMVs

Service Members/Veterans

SRA

Social Roles and Activities

TBI

Traumatic Brain Injury

TBI-CareQOL

Traumatic Brain Injury Caregiver Quality of Life

TBI-QOL

Traumatic Brain Injury Quality of Life