Psy 475_Week 2_Psychological Measure Paper

beck_depression_inventory-ii.pdf

Home >Psychology homework help >Psy 475_Week 2_Psychological Measure Paper

MILITARY MEDICINE, 179, 8:879, 2014

Psychometric Properties of the Beck Depression Inventory-II for OEF/OIF Veterans in a Polytrauma Sample

Glen A. Palmer, PhD, ABN; Maggie C. Happe, PsyD; Janine M. Paxson, PhD; Benjamin K. Jurek, PsyD; Joseph J. Graca, PhD; Stephen A. Olson, BA

ABSTRACT Objective: The Beck Depression Inventory-II (BDI-II) is widely used as a screening instrument for depressive symptomatology in clinical settings. The factor structure has been researched in a variety of settings with results ranging from a single factor to a five-factor structure. The purpose of this study was to examine several identified factor structures when applied to a mixed polytrauma sample. Method: A sample of 310 veterans was used for this study. All subjects were administered the BDI-II screening measure as part of an evaluation in an outpatient polytrauma clinic. Confirmatory factor analysis was used to determine the best model. Results: Confirmatory factor analysis revealed that a three-factor model provided a best fit. A model previously identified for individuals in residential treatment for substance abuse provided a best fit for this sample. Conclusions: The BDI-II may provide additional information for clinicians when examining the three-factor model with veterans in polytrauma settings.

INTRODUCTION The term “polytrauma” has been used by the Department of

Veterans Affairs to describe injuries to multiple body parts

and organs occurring as a result of blast-related wounds seen

in Operation Enduring Freedom (OEF) and Operation Iraqi

Freedom (OIF). 1 Traumatic brain injury (TBI) frequently

occurs in combination with other disabling conditions includ-

ing post-traumatic stress disorder (PTSD), depression, and

other mental health/medical conditions. Mild traumatic brain

injury (mTBI) as defined by the American Congress of Reha-

bilitation Medicine (ACRM) includes traumatically induced

changes in at least one or more areas including (1) loss of

consciousness for < 30 minutes and/or initial Glasgow Coma Scale score of 13–15 after 30 minutes, (2) loss of any mem-

ory of events that occurred within 24 hours before or after

the accident (i.e., post-traumatic amnesia), (3) alteration of

consciousness/mental state at the time of the accident (e.g.,

feeling confused, dazed, or disoriented), and/or (4) presence

of focal neurological deficits. 2,3

According to the ACRM,

several other symptoms may be present as evidence of mTBI

(e.g., nausea, headache, cognitive deficits).

Depression is a commonly reported problem for both out-

patient and inpatient populations following a TBI. 4 Research

on prevalence and incidence rates of depression and TBI has

been highly variable. Studies have reported frequency of

depression following TBI from anywhere between 6% and

77%. 4–8

Some of the variability between studies is likely

because of methodological differences and lack of uniformity

with the diagnosis of major depressive disorder. 5 Some stud-

ies that have incorporated more stringent criteria (e.g., use of

structured interview versus use of depression rating scales)

have found rates between 17% and 61%. 7

The Beck Depression Inventory (BDI) 9 and its latest revi-

sion, the Beck Depression Inventory-II (BDI-II), 10

have been

widely studied with a variety of different populations.

Christenson et al were perhaps the first researchers to explore

the factor structure of the BDI with individuals receiving

rehabilitation in a TBI sample. 11

They reported a five-factor

structure with the following descriptors: (1) symptoms of

major depression, (2) symptoms of TBI, (3) hopelessness/

anhedonia, (4) negative self-appraisal, and (5) cognitive dis-

tortions. Alternatively, Green et al reported a three-factor

structure labeled: (1) affective/performance complaints, (2)

negative attitudes toward self, and (3) somatic complaints. 12

Initial factor analysis of the BDI-II was conducted by

Beck et al. 10

They identified a two-factor solution based on a

sample of 500 psychiatric outpatients. The first factor consisted

of cognitive symptoms (items 4, 10, 11, 12, 13, 15, 16, 17,

18, 19, 20, and 21). The second factor consisted of somatic/

affective symptoms (items 1, 2, 3, 5, 6, 7, 8, 9, and 14).

Since the development of BDI-II, the factor structure of

the BDI-II has been examined with a variety of different

samples with different results. Findings suggest the factor

structure of the BDI-II varies among, and has been inconsis-

tent within, clinical populations. Rowland et al evaluated the

factor structure of the BDI-II for individuals with TBI and

found a three-factor structure labeled: (1) Negative Self Eval-

uation (items 3, 4, 5, 6, 7, 8, and 12), (2) Symptoms of Depres-

sion (items 1, 2, 9, 10, 13, 14, 15, 17, 19, and 20), and

(3) Vegetative symptoms (items 11, 16, 18, and 21). 13

The

study was one of the first to explore the factor structure of

BDI-II with a TBI sample, although the sample size was very

small (N = 51). Injury severity of the sample was reported to be roughly evenly divided between mild/moderate (49%) and

severe (51%). About 26% of this sample also had a substance

abuse history. Findings of the study revealed that Vegetative

symptoms of depression were unique to the TBI sample.

Department of Psychology, St. Cloud VA Health Care System, 4801

Veterans Drive, EC-117; St. Cloud, MN 56303.

The contents of this manuscript do not represent the views of the Depart-

ment of Veterans Affairs or the United States Government.

doi: 10.7205/MILMED-D-14-00048

MILITARY MEDICINE, Vol. 179, August 2014 879

Downloaded from publications.amsus.org: AMSUS - Association of Military Surgeons of the U.S. IP: 203.056.241.002 on Aug 10, 2014.

Siegert et al examined the factor structure of the inventory

with an inpatient neurorehabilitation sample. 14

Confirmatory

factor analysis (CFA) was conducted that revealed, in addi-

tion to a general depression factor, two underlying specific

factors labeled: (1) Cognitive/Affective (items 1, 2, 3, 5, 6, 7,

8, 9, 10, 11, and 14) and (2) Somatic (items 4, 12, 13, 15, 16,

17, 18, 19, and 20). Two items for the two-factor model did

not correlate well with either factor (i.e., item 17-Irritability;

and item 21-Loss of Interest in Sex), but item 17 was

included in the Somatic factor structure as listed above.

The authors combined items into parcels for portions of

their analyses.

Buckley et al examined the factor structure of the BDI-II

in a sample of veterans (N = 416) admitted to a 28-day residential treatment program for chemical dependency in

the southern United States. 15

The authors conducted a series

of CFAs (i.e., one-, two-, three-factor models) and concluded

that a three-factor model provided the best fit for the data.

The model consisted of (1) Cognitive (items 1, 2, 3, 5, 6, 7, 8,

9, and 14), (2) Affective (items 4, 10, 12, and 13); and

FIGURE 1. Three-factor substance abuse model (as identified in Buckley et al15).

MILITARY MEDICINE, Vol. 179, August 2014880

Psychometric Properties of the BDI-II for Veterans in a Polytrauma Sample

Downloaded from publications.amsus.org: AMSUS - Association of Military Surgeons of the U.S. IP: 203.056.241.002 on Aug 10, 2014.

(3) Somatic (items 11, 15, 16, 17, 18, 19, 20, and 21) factors.

Figure 1 describes the three-factor model.

Given the various findings regarding factor structure of the

BDI-II, research is needed to determine what factor structure

is most appropriate for polytrauma and dual diagnosis samples.

To date, little research exists regarding the factor structure of

this instrument with veterans in a polytrauma sample. The

purpose of this study was to identify the best factor structure

for the BDI-II in a mixed polytrauma sample of veterans by

use of CFA. It was hypothesized that the factor structure

would be similar to findings of a TBI sample, 13

given the

majority of subjects were referred because of subjective report

of head trauma. Therefore, theoretically one might expect

that vegetative symptoms of depression as outlined in the

three-factor TBI model would provide a unique contribution

to veterans with mTBI when compared with other models

(e.g., two-factor model for psychiatric samples).

To the authors’ knowledge, CFA with the BDI-II has not

been conducted specifically with veterans in a polytrauma

sample. In general, little research has been published regard-

ing subjects served specifically in level 3 polytrauma settings.

Findings in these particular settings are relevant to a variety

of other outpatient (e.g., rehabilitation, mental health, or

other clinical settings) settings because of the increasing

number of veterans in the clinic settings who have suffered

mTBI and/or other comorbid conditions. In the VA system,

level 3 polytrauma settings usually have outpatient teams

with rehabilitation expertise that deliver follow-up services

with supportive consultation from regional and/or network

facilities (e.g., level 1 or level 2 facilities) as needed. 16

Eval-

uation of the factor structure of the BDI-II with polytrauma

samples is extremely important as the instrument is commonly

used for assessment of depressive symptoms in a variety of

settings. If a three-factor model exists, elevated scores on

certain factors (e.g., Vegetative symptoms) might be attrib-

uted to mTBI versus true depressive symptomatology, which

could affect treatment recommendations and outcomes.

Further, post hoc analyses were planned to see if select

models could be improved by adding a higher order single

factors (i.e., total depression score) to existing two- and

three-factor models.

METHOD

Participants

Retrospective data collection was used for this study with a

sample of 310 veterans. The sample of veterans used in

this particular study was referred to an outpatient level 3

polytrauma clinic for evaluation. All of the patients for this

particular research investigation were administered the

BDI-II as part of the evaluation process. All of the BDI-II

screening instruments were administered, scored, and

interpreted by licensed psychologists.

This research study was approved by the facility’s affiliate

Institutional Review Board and local Research and Develop-

ment Committee. A thorough review and consideration of

human subjects’ protections for vulnerable populations was

conducted as part of the process. Waiver of informed consent

for this minimal risk study was obtained by the affiliate

Institutional Review Board and all data containing personal

health information were deidentified before analysis.

Demographic characteristics of the sample are presented

in Table I. The sample was predominantly male (299 males

and 11 females). Mean age of the sample was 30.07 years

(SD = 7.56). Ethnicity of the sample was predominantly white (n = 289; 93.2%). Based on self-report and record review of the patient’s TBI evaluation (conducted by a licensed

psychologist and physiatrist), the majority of the veterans met

criteria for having suffered from mTBI (n = 239; 77.1%). A small percentage of those in the sample met criteria for

moderate to severe TBI (n = 18; 5.8%). Various branches of the military were represented, with the majority of subjects

having served in the Army (n = 85; 27.4%) or Army National Guard (n = 139; 43.5%) at the time of deployment and/ or injury.

Measures

The BDI-II 10

was used as a screening measure to assess for

depressive symptoms in the polytrauma clinic. The screening

measure consists of 21 items assessing affective, cognitive,

and physiological symptoms associated with depression. The

TABLE I. Characteristics of the Sample

Characteristic n (%) or M (SD) Range

N 310

Male 299 (96.5%)

Female 11 (3.5%)

Ethnicity

White 289 (93.2%)

African–American 9 (2.9%)

Hispanic 4 (1.3%)

Native American 4 (1.3%)

Other or Unknown 4 (1.3%)

Age 30.07 (7.56) 20–57

Branch of the Military

Army 85 (27.4%)

Army National Guard 135 (43.5%)

Navy 11 (3.5%)

Marines 37 (11.9%)

Army Reserves 25 (8.1%)

Airforce 13 (4.2%)

Other 4 (1.3%)

Concussions 257 (82.9%)

Psychiatric Diagnosis

PTSD 148 (47.7%)

Depression 150 (48.4%)

Anxiety 84 (27.1%)

Adjustment Disorders 48 (15.5%)

Other Mental Health 24 (7.7%)

Alcohol and/or 121 (39.0%)

Other Substance 31 (10.0%)

No Diagnosis 23 (7.4%)

Comorbid Conditions 189 (61.0%)

MILITARY MEDICINE, Vol. 179, August 2014 881

Psychometric Properties of the BDI-II for Veterans in a Polytrauma Sample

Downloaded from publications.amsus.org: AMSUS - Association of Military Surgeons of the U.S. IP: 203.056.241.002 on Aug 10, 2014.

BDI-II is a self-report screening measure that asks the patient

to evaluate symptoms based on a series of sentences (for each

item) presented in ascending order from no symptoms (score

of 0) to severe symptoms (score of 3). The patient is asked

to rate symptoms in the context of experiences within the

last 2 weeks. The instrument has been demonstrated to have

good reliability and validity. 10

Statistical Analysis

Descriptive statistics were calculated with IBM/SPSS Statis-

tics (IBM/SPSS version 21.0, IBM/SPSS, Chicago, Illinois,

2012). Data were examined for skewness and kurtosis for

each item of the BDI-II. Cutoff scores of greater than + 1 were used as criteria for determining skewness and/or kurtosis of

the distributions. Items 6 (Punishment Feelings), 9 (Suicidal

Thoughts or Wishes), 10 (Crying), and 14 (Worthlessness)

exceeded values + 1 for skewness and/or kurtosis. Therefore, log transformations were conducted for these items before

further analyses. Internal reliability of the sample was calcu-

lated with a coefficient.17 There was very little missing data (i.e., <1%). Therefore, full data imputation was not necessary, and missing data were replaced by the median of nearby points.

Confirmatory factor analysis for this study was performed

with Amos (Amos Version 21.0, IBM/SPSS, Chicago, Illinois,

2012). A covariance matrix of BDI-II items was developed

with the maximum-likelihood estimation. The marker variable

strategy was used for analysis. The marker variable strategy

assigns one of the factor loadings with a value of 1.0 for each

factor. According to recommendations for model goodness of

fit proposed by other researchers, 18,19

a combination approach

to evaluating model fit was used.

The goodness of fit for each model was examined using

(1) the c2 statistic, (2) the incremental fit index (IFI), (3) the comparative fit index (CFI), (4) the root mean square error of

approximation (RMSEA), and (4) the expected cross-validation

index (ECVI). Nonsignificant c2 values and values higher than 0.90 on the IFI and CFI are considered adequate model

fit. In addition, IFI and CFI values at 0.95 or greater are

considered good model fit. RMSEA values of up to 0.08

indicate reasonable errors of approximation for the model.

For ECVI, the smallest value of all models is indication of

having the best fit.

RESULTS Internal reliability of the BDI-II with the sample was excellent

(a = 0.93). The mean BDI-II total score was 21.20 (SD = 11.76; range = 0–51) for the sample. A score of 21 is in the range for moderate symptoms of depression on the BDI-II.

Confirmatory factor analyses were first conducted to com-

pare the three-factor TBI, two- factor psychiatric, three-factor

substance abuse, and two-factor neurorehabilitation samples

described previously. Results of the analyses are presented

in Table II. To test our hypothesis that the three-factor TBI

model provided the best fit for the data, a CFA was conducted

by comparing models in the following order: the three-factor

TBI model, three-factor substance abuse model, two-factor

psychiatric model, and two-factor neurorehabilitation model.

Unexpectedly, the three-factor TBI model provided the poorest

fit for the data.

Chi-square tests for all models were significant (p < 0.001). For CFA, a nonsignificant result is desired for the c2 statistic, as this would suggest the model fits the data. However, a

significant c2 finding for CFA is not unusual, especially as the sample size gets larger.

20,21 The CFI and IFI were within

acceptable levels for all models except the three-factor TBI

model; however, the CFI and IFI were highest for the three-

factor substance abuse model (CFI = 0.930 and IFI = 0.931). Next, RMSEA was examined for each of the models, and

again the three-factor substance abuse model provided the

best fit as evidenced by the lowest score and lowest error of

approximation for the model (RMSEA = 0.061). For ECVI, the smallest value of all models is an indication

of having the best fit. The three-factor substance abuse model

provided one of the lowest score of all models (ECVI = 1.593), with the exception of the two-factor neurorehabilitation sam-

ple. Given the finding regarding the EVCI, the c2 mean dif- ference test was conducted between these two models, a

significant difference between c2 values would mean that the model with the smallest c2 value would provide a better model fit for this variable. The finding was not significant

(Dc2 (35) = 24.257, p > 0.05), indicating neither of the two models provided a better fit with regard to the c2 statistic.

Post hoc analyses were conducted to determine if a higher

order factor structure might improve the hypothesized models.

Table II provides results of the analyses. However, the

TABLE II. Fit Indices for Several Models

Model c2 df CFI IFI RMSEA (90% CI) ECVI

Three-Factor TBI 13

577.525* 186 0.874 0.875 0.083 (0.075–0.090) 2.160

Three-Factor Substance Abuse 15

402.097* 186 0.930 0.931 0.061 (0.053–0.070) 1.593

Two-Factor Psychiatric 10

424.199* 188 0.924 0.924 0.064 (0.056–0.072) 1.651

Two-Factor Neuro 14

377.840* 151 0.918 0.919 0.070 (0.061–0.079) 1.475

Three-Factor Higher Order TBI 577.525* 186 0.874 0.875 0.083 (0.075–0.090) 2.160

Three-Factor Higher Order Substance Abuse 440.054* 188 0.919 0.919 0.066 (0.058–0.074) 1.702

Two-Factor Higher Order Psychiatric 424.199* 188 0.924 0.924 0.064 (0.056–0.072) 1.651

Two-Factor Higher Order Neuro 377.840* 151 0.918 0.919 0.070 (0.061–0.079) 1.475

CI, confidence interval. *p < 0.001.

MILITARY MEDICINE, Vol. 179, August 2014882

Psychometric Properties of the BDI-II for Veterans in a Polytrauma Sample

Downloaded from publications.amsus.org: AMSUS - Association of Military Surgeons of the U.S. IP: 203.056.241.002 on Aug 10, 2014.

addition of higher-order factors to each of the models added

no better fit when compared with the other models.

Finally, the factor structure for the three-factor substance

abuse model was explored (see Fig. 1). Results of analyses

for the factors and information regarding BDI-II items are

presented in Table III. Cognitive (M = 6.90, SD = 5.28), Affective (M = 4.45, SD = 2.89), and Somatic (M = 9.85, SD = 4.82) factors were calculated for the sample. Internal reliabilities for the three factors were moderate to good

(a’s = 0.79 to 0.88).

DISCUSSION The three-factor model for individuals with substance abuse

described in previous research provided the best fit of all

models that were compared with our polytrauma study sam-

ple. This finding was somewhat surprising, as factor struc-

tures comparable to previous research for individuals with

TBI were expected to provide the best model fit. Given the

number of individuals with comorbid conditions, it was also

surprising that the psychiatric factor structure did not pro-

vide the best fit (although the psychiatric model did provide

an adequate fit). This finding might best be explained by

several factors.

First, the sample consisted of OEF/OIF veterans who were

referred to a level 3 polytrauma clinic. The majority of these

individuals met criteria for mTBI. The Rowland et al three-

factor TBI model 13 had a sample that consisted of a sample of

subjects that consisted of greater than 50% of the subjects

with severe TBI. In contrast, our sample was significantly

larger and consisted of a majority of subjects that were con-

sidered to have only mTBI (77.1%). Further, their findings

were based on a very small sample size, and CFA of the

solution was not conducted.

Second, our polytrauma sample was heterogeneous com-

pared to some of the other groups from which other models

were derived. For example, when compared to the Beck et al 10

psychiatric model, the current sample had a number of patients

with co-occurring conditions. Over 62% of this sample had

comorbid conditions, which included subjects with psychiatric

and chemical abuse/dependency diagnoses. The sample is

likely representative of what might be found in veteran clin-

ical populations, particularly with those who have served in

an OEF/OIF context. Despite the fact that our sample was

somewhat heterogeneous, the sample is comparable to veterans

served in typical polytrauma settings. Further, the findings of

this study suggest that results may be most appropriate for

other settings that serve veterans given the fact that the model

identified in Buckley et al 15

provided the best fit for our

sample as well. An increasing number of OEF/OIF veterans

are presenting with overlapping conditions (e.g., mTBI,

PTSD, depression) that provide unique challenges for achiev-

ing optimal treatment outcomes. 22

Comorbidity of mTBI with

other mental health conditions (i.e., PTSD and depression) in

the OEF/OIF veteran population is common. For example, at

least one study has reported that 42% of OEF/OIF veterans

with mTBI also had symptoms of PTSD. 23

A number of individuals (40%) met criteria for alcohol

abuse or dependence issues. Other substances used by the

sample included cannabis, methamphetamine, and cocaine.

Although the clinic served patients on an outpatient basis, a

percentage of individuals had completed a residential treat-

ment program for PTSD and/or substance abuse before or

during evaluation (18.1%). Because of the retrospective

nature of the study, a portion of the sample also received

residential services at some point after the TBI evaluations

were conducted. These findings might partially account for

findings of the three-factor substance abuse model (residen-

tial setting) described by Buckley et al 15

being the best fit

for our sample.

Finally, it should also be noted that the previous study also

used a sample of veterans for their study. What is unclear is

whether these findings regarding the factor structure with the

BDI-II might be consistent with other veteran samples. Fur-

ther research is needed with other samples of veterans (e.g.,

veterans with moderate to severe TBI, patients with medical

diagnoses) to see if a similar factor structure exists.

One potential limitation of the study is the smaller sample

size compared to some previous studies using CFA. For some

studies that have performed CFA with larger samples, cutoff

scores for the CFI and IFI have been increased to 0.95 as

evidence of a good model fit. Given this study’s smaller

sample size, we chose the lower cutoff score as evidence of

TABLE III. Descriptive Statistics for the Three Factor Substance Abuse Model

Factor Item M SD Cronbach’s a

Cognitive 6.90 5.28 0.88

(1) Sadness 0.71 0.75

(2) Pessimism 0.94 0.76

(3) Past Failure 0.94 0.88

(5) Guilty Feelings 0.87 0.86

(6) Punishment Feelings 0.57 0.90

(7) Self-Dislike 0.98 0.91

(8) Self-Criticalness 1.02 0.89

(9) Suicidal Ideation 0.23 0.47

(14) Worthlessness 0.67 0.84

Affective 4.45 2.89 0.79

(4) Loss of Pleasure 1.32 0.87

(10) Crying 0.73 0.98

(12) Loss of Interest 1.28 0.91

(13) Indecisiveness 1.13 0.94

Somatic 9.85 4.82 0.85

(11) Agitation 1.36 0.93

(15) Loss of Energy 1.09 0.71

(16) Sleep Disturbance 1.58 0.86

(17) Irritability 1.43 0.90

(18) Appetite Disturbance 1.07 0.93

(19) Concentration Difficulty 1.38 0.77

(20) Fatigue 1.12 0.84

(21) Loss of Interest in Sex 0.83 0.95

M, mean; SD, standard deviation.

MILITARY MEDICINE, Vol. 179, August 2014 883

Psychometric Properties of the BDI-II for Veterans in a Polytrauma Sample

Downloaded from publications.amsus.org: AMSUS - Association of Military Surgeons of the U.S. IP: 203.056.241.002 on Aug 10, 2014.

an acceptable model fit. Several studies have reported that

0.90 is an acceptable cutoff score. 24–26

Future research will be helpful to further evaluate the

BDI-II with veterans, particularly OEF/OIF veterans with

comorbid conditions. This study provides some replication

of the findings regarding a three-factor model consisting of

Cognitive, Affective, and Somatic concerns. Confirmatory

factor analysis with the BDI-II with other clinically relevant

and common groups of veterans (e.g., veterans with physical

injuries, medical illnesses) will also be helpful to determine

if the factor structure only pertains to select groups of

veterans (e.g., substance abuse, polytrauma samples), or if

overall it might serve as a factor structure to be used with

all veterans.

ACKNOWLEDGMENT

This manuscript is the result of work supported with resources and the use of

facilities at the St. Cloud VA Health Care System.

REFERENCES

1. United States Department of Veterans Affairs. Understanding Traumatic

Brain Injury: Definition and Background, 2011. Available at http://

www.polytrauma.va.gov/understanding-tbi/definition-and-background.asp;

accessed November 21, 2011.

2. Halbauer JD, Ashford JW, Zeitzer JM, Adamson MM, Lew HL,

Yesavage JA: Neuropsychiatric diagnosis and management of chronic

sequelae of war-related mild to moderate traumatic brain injury. J Rehabil

Res Dev 2009; 46: 757–96.

3. Kay T, Harrington DE, Adams R, et al: Definition of mild traumatic

brain injury. J Head Trauma Rehabil 1993; 8: 86–7.

4. Glenn MB, O’Neil-Pirozzi T, Goldstein R, Burke D, Jacob L: Depres-

sion amongst outpatients with traumatic brain injury. Brain Inj 2001; 15:

811–8.

5. Jorge RE, Starkstein SE: Pathophysiologic aspects of major depression

following traumatic brain injury. J Head Trauma Rehabil 2005; 20:

475–87.

6. Kreutzer JS, Seel RT, Gourley E: The prevalence and symptom rates of

depression after traumatic brain injury: A comprehensive examination.

Brain Inj 2001; 15: 563–76.

7. Rapoport MJ: Depression following traumatic brain injury. CNS Drugs

2012; 26: 111–21.

8. Seel RT, Kreutzer JS: Depression assessment after traumatic brain injury:

an empirically based classification method. Arch Phys Med Rehabil 2003;

84: 1621–8.

9. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J: An inventory for

measuring depression. Arch Gen Psychiat 1961; 4: 561–71.

10. Beck AT, Steer RA, Brown GK: Beck Depression Inventory Manual,

Ed 2. San Antonio, Psychological Corporation, 1996.

11. Christensen BK, Ross TP, Kotasek M, Rosenthal M, Henry RR: Factor

structure of the Beck Depression Inventory in a sample of persons with

traumatic brain injury [Abstract]. J Int Neuropsych Soc 1995; 1: 122.

12. Green A, Felmingham K, Baguley IJ, Slewa-Younan S, Simpson S:

The clinical utility of the Beck Depression Inventory after traumatic

brain injury. Brain Inj 2001; 15: 1021–8.

13. Rowland SM, Lam CS, Leahy B: Use of the Beck Depression Inventory-II

(BDI-II) with persons with traumatic brain injury: analysis of the factorial

structure. Brain Inj 2005; 19: 77–83.

14. Siegert RJ, Walkey FH, Turner-Stokes L: An examination of the factor

structure of the Beck Depression Inventory-II in a neurorehabilitation

sample. J Int Neuropsychol Soc 2009; 15: 142–7.

15. Buckley TC, Parker JD, Heggie J: A psychometric evaluation of the

BDI-II in treatment-seeking substance abusers. J Subst Abuse Treat

2001; 20: 197–204.

16. United States Department of Veterans Affairs. Putting polytrauma care

‘on the map.’ VA Research Currents, 2008; October 5–6. Available at

http://www.research.va.gov/resources/pubs/docs/va_research_currents_

oct_08.pdf; accessed June 12, 2010.

17. Cronbach LJ: Essentials of Psychological Testing, Ed 4. New York,

Harper Row, 1984.

18. Byrne BM: Structural Equation Modeling with AMOS: Basic Concepts,

Applications, and Programming, Ed 2. New York, Routledge, 2010.

19. Hoyle RH ( editor): Structural Equation Modeling: Concepts, Issues,

and Applications. Thousand Oaks, California, Sage, 1995.

20. Brown TA: Confirmatory Factor Analysis for Applied Research.

New York, Guilford Press, 2006.

21. Meterko M, Baker E, Stolzmann KL, Hendricks AM, Cicerone KD,

Lew HL: Psychometric assessment of the Neurobehavioral Symptom

Inventory −22: the structure of persistent postconcussive symptoms fol-

lowing deployment-related mild traumatic brain injury among veterans.

J Head Trauma Rehab 2012; 27: 55–62.

22. Lew HL, Vanderploeg RD, Moore DF, et al: Overlap of mild TBI

and mental health conditions in returning OIF/OEF service members

and veterans. J Rehabil Res Dev 2008; 45: xi–xvi.

23. Lew HL, Poole JH, Vanderploeg RD, et al: Program development and

defining characteristics of returning military in a VA polytrauma network

site. J Rehabil Res Dev 2007; 44: 1027–34.

24. Linley PA, Andrews L, Joseph S: Confirmatory factor analysis of the

Posttraumatic Growth Inventory. J Loss Trauma 2007; 12: 321–32.

25. Palmer GA, Graca JJ, Occhietti KE: Confirmatory factor analysis of the

Posttraumatic Growth Inventory in a veteran sample with posttraumatic

stress disorder. J Loss Trauma 2012; 17: 545–56.

26. Schinka JA, Brown LM, Borenstein AR, Mortimer JA: Confirmatory

factor analysis of the PTSD Checklist in the elderly. J Trauma Stress

2007; 20: 281–9.

MILITARY MEDICINE, Vol. 179, August 2014884

Psychometric Properties of the BDI-II for Veterans in a Polytrauma Sample

Downloaded from publications.amsus.org: AMSUS - Association of Military Surgeons of the U.S. IP: 203.056.241.002 on Aug 10, 2014.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.