ArticlePDF AvailableLiterature Review

Clinician-Administered PTSD Scale: A review of the first ten years of research

January 2001
Depression and Anxiety 13(3):132 - 156

DOI:10.1002/da.1029

Authors:

Frank Weathers

Auburn University

Terence M Keane

VA National Center for PTSD & Boston University

Jonathan Davidson

Duke University Medical Center

The Clinician-Administered PTSD Scale (CAPS) is a structured interview for assessing posttraumatic stress disorder (PTSD) diagnostic status and symptom severity. In the 10 years since it was developed, the CAPS has become a standard criterion measure in the field of traumatic stress and has now been used in more than 200 studies. In this paper, we first trace the history of the CAPS and provide an update on recent developments. Then we review the empirical literature, summarizing and evaluating the findings regarding the psychometric properties of the CAPS. The research evidence indicates that the CAPS has excellent reliability, yielding consistent scores across items, raters, and testing occasions. There is also strong evidence of validity: The CAPS has excellent convergent and discriminant validity, diagnostic utility, and sensitivity to clinical change. Finally, we address several concerns about the CAPS and offer recommendations for optimizing the CAPS for various clinical research applications. Depression and Anxiety 13:132–156, 2001 © 2001 Wiley-Liss, Inc.

Content uploaded by Jonathan Davidson

Content may be subject to copyright.

132 Weathers et al.

DEPRESSION AND ANXIETY 13:132–156 (2001)

Review Article

CLINICIAN-ADMINISTERED PTSD SCALE: A REVIEW OF

THE FIRST TEN YEARS OF RESEARCH

Frank W. Weathers, Ph.D.,1* Terence M. Keane, Ph.D.,2 and Jonathan R.T. Davidson, M.D.3

The Clinician-Administered PTSD Scale (CAPS) is a structured interview for

assessing posttraumatic stress disorder (PTSD) diagnostic status and symptom

severity. In the 10 years since it was developed, the CAPS has become a stan-

dard criterion measure in the field of traumatic stress and has now been used

in more than 200 studies. In this paper, we first trace the history of the

CAPS and provide an update on recent developments. Then we review the

empirical literature, summarizing and evaluating the findings regarding

the psychometric properties of the CAPS. The research evidence indicates

that the CAPS has excellent reliability, yielding consistent scores across

items, raters, and testing occasions. There is also strong evidence of valid-

ity: The CAPS has excellent convergent and discriminant validity, diag-

nostic utility, and sensitivity to clinical change. Finally, we address several

concerns about the CAPS and offer recommendations for optimizing the

CAPS for various clinical research applications. Depression and Anxiety

Key words: posttraumatic stress disorder; structured interview; diagnosis;

assessment; reliability; validity

1Auburn University, Auburn, Alabama

2Boston Veterans Affairs Medical Center and Boston Univer-

sity School of Medicine, Boston, Massachusetts

3Duke University Medical Center, Durham, North Carolina

*Correspondence to: Dr. Frank W. Weathers, Department of Psy-

chology, 226 Thach Hall, Auburn University, Auburn, AL 36849-

5214. E-mail: weathfw@auburn.edu

Received for publication 1 May 2000; Accepted 23 October 2000

INTRODUCTION

Since its development in 1990 at the National Center

for Posttraumatic Stress Disorder (PTSD), The Clini-

cian-Administered PTSD Scale [CAPS; Blake et al.,

1990] has become one of the most widely used struc-

tured interviews for diagnosing and measuring the se-

verity of PTSD. Initially validated on combat veterans,

the CAPS has now been used successfully in a wide va-

riety of trauma populations, including victims of rape,

crime, motor vehicle accidents, incest, the Holocaust,

torture, and cancer. It has served as the primary diag-

nostic or outcome measure in more than 200 empirical

studies on PTSD and has been translated into at least

ten languages. In addition, a child and adolescent ver-

sion of the CAPS has been developed and is now un-

dergoing field testing and psychometric evaluation.

Originally based on the PTSD criteria in the DSM-III-

R, the CAPS has been revised several times in response

to user feedback and changes in the PTSD diagnostic

criteria, with the most significant revision occurring af-

ter the publication of the DSM-IV in 1994.

The present paper is an update on the CAPS and a

critical review of the first 10 years of CAPS-related

research. It was prompted by the increasing popularity

of the CAPS, the rapid accumulation of empirical evi-

dence that supports its use, and the need to inform cur-

rent and potential CAPS users about the latest revisions

and recommendations for administration and scoring.

This paper consists of three sections. First, we provide

a brief overview of the CAPS, describing the rationale

for its development, its key features, and its evolution

through an extensive revision for DSM-IV, as well as a

description of other minor modifications. Second, we

review the published literature on the CAPS, focusing

in particular on psychometric studies of the CAPS and

on pharmacological and psychosocial treatment studies

that employed the CAPS as an outcome measure.

Third, we discuss the implications of the findings and

offer recommendations for using the CAPS in a range

of research and clinical applications.

Review Article: Clinician-Administered PTSD Scale 133

This paper was not intended as an in-depth critique

of the methodology or conceptual implications of the

studies we reviewed, nor did we seek to reach any gen-

eral conclusions about the current status of PTSD re-

search. Rather, our main purpose was simply to identify

studies that have used the CAPS and summarize the

empirical findings that bear directly on its psychomet-

ric properties and utility for assessing PTSD. Finally,

since the child and adolescent version is still undergo-

ing validation, we focus here only on research examin-

ing the adult CAPS.

OVERVIEW OF THE CAPS

In developing the CAPS, the primary goal was to

create a comprehensive, psychometrically sound inter-

view-based rating scale that would be widely accepted

as a standard criterion measure of PTSD. In this sense

it was intended to serve a role in the field of traumatic

stress analogous to that of the ubiquitous Hamilton

Depression Rating Scale [HAM-D; Hamilton, 1960]

in the field of depression. The CAPS was designed

with a number of features intended to improve exist-

ing PTSD interviews and enhance the reliability and

validity of PTSD assessment [see Blake et al., 1995,

for a full discussion and a comparison of the CAPS

with other PTSD interviews]. First, the CAPS can be

used either as a dichotomous (present/absent) diag-

nostic measure or as a continuous measure of PTSD

symptom severity. Second, the CAPS assesses both the

frequency and intensity of individual PTSD symptoms

on separate five-point (0–4) rating scales, and these rat-

ings can be summed to create a nine-point (0–8) sever-

ity score for each symptom. This permits considerable

flexibility in scoring: CAPS users can focus on the fre-

quency, intensity, or severity ratings for individual

PTSD symptoms, for the three PTSD symptom clus-

ters (re-experiencing, avoidance and numbing, and

hyperarousal), and for the PTSD syndrome as a whole.

Third, the CAPS promotes uniform administration

and scoring through carefully phrased prompt ques-

tions and explicit rating scale anchors with clear be-

havioral referents. Initial prompt questions explicitly

target each symptom, and follow-up prompts help in-

terviewers clarify the inquiry as needed, anticipating

typical points of ambiguity or confusion regarding the

PTSD criteria. These features enhance standardiza-

tion across interviewers and ensure comparability of

scores across diverse settings, raters, and trauma popu-

lations. Fourth, the CAPS provides complete coverage

of the PTSD syndrome. The original version of the

CAPS included 17 items assessing the DSM-III-R

symptoms of PTSD, 8 items assessing associated fea-

tures (e.g., guilt, hopelessness, memory impairment),

and 5 items assessing response validity, global severity,

global improvement, and social and occupational im-

pairment. As described below, the current version of the

CAPS assesses all DSM-IV diagnostic criteria for PTSD,

including Criterion A (exposure to a traumatic event),

Criteria B–D (core symptom clusters of re-experiencing,

numbing and avoidance, and hyperarousal), Criterion E

(chronology), and Criterion F (functional impairment),

as well as the associated symptoms of guilt and dissocia-

tion. Finally, the CAPS assesses current and lifetime

PTSD symptom status. The prompts for lifetime di-

agnosis help the interviewer establish explicitly that

any endorsed symptoms occurred as a syndrome

within the same one-month period.

Initially it was decided that two parallel versions of

the CAPS were needed in order to address two dis-

tinct assessment needs. The CAPS-1, or current and

lifetime diagnostic version, was designed to assess

PTSD symptom severity and diagnostic status over

the past month, or for the worst month since the

trauma. The CAPS-2, or 1-week symptom status ver-

sion, was designed to measure PTSD symptom sever-

ity over the past week and was intended primarily for

repeated assessment over relatively brief time intervals

in pharmacological research. Apart from the different

time frames assessed, the main difference between the

CAPS-1 and CAPS-2 is that for the ten CAPS items

where symptom frequency is rated in terms of a count

(i.e., how often) as opposed to a percentage (i.e., how

much of the time), the rating scale anchors on the

CAPS-2 were based on a 1-week time frame, whereas

for the CAPS-1, they were based on a 1-month time

frame. The distinction between these two original ver-

sions of the CAPS led to some confusion in the field,

such that the CAPS-2 was thought by some to be a

revised version of the CAPS. In response to this con-

fusion, as part of the DSM-IV revision, the CAPS-1

was renamed the CAPS-DX (i.e., CAPS-Diagnostic

version), and the CAPS-2 was renamed the CAPS-SX

(i.e., CAPS-Symptom Status version). As discussed be-

low, these two versions were recently combined into a

single instrument now simply known as the CAPS.

Following the publication of the DSM-IV in 1994,

the CAPS was revised, both to bring it up to date with

changes in the PTSD criteria and to incorporate user

feedback accumulated since its release in 1990. The

overarching goal for the revision was to ensure back-

ward compatibility with the original CAPS. This was

accomplished by retaining the basic structure, most of

the prompt questions, and the values and stems for the

rating scale anchors. The revision included four major

modifications and a number of relatively minor ones.

Major modifications included the following.

1. Adding a brief protocol for assessing Criterion A

(exposure to a traumatic event). This consists of a

17-item self-report checklist of potentially trau-

matic events and follow-up questions to help the

interviewer determine if a stressful event satisfies

both parts of the DSM-IV definition of a trau-

matic event (i.e., the event involves life threat, se-

rious injury, or threat to physical integrity; and

the person responds with intense fear, helpless-

ness, or horror).

134 Weathers et al.

2. Rewording some of the descriptors for the intensity

rating scale anchors. This was done to achieve a

consistent focus across items on the three key di-

mensions of intensity (duration, subjective distress,

and functional impairment), to achieve roughly

equal gradations of intensity between each of the

rating scale values, and to provide examples appli-

cable to a range of trauma populations.

3. Adding a three-point rating scale (“definite,”

“probable,” and “unlikely”) that requires inter-

viewers to determine if a reported symptom is at-

tributable to a specific traumatic event. This scale

only applies to the last 9 of the 17 symptoms of

PTSD (emotional numbing and hyperarousal)

because the first 8 symptoms (re-experiencing,

effortful avoidance, and amnesia) are all inher-

ently trauma-linked.

4. Replacing six of the eight original associated fea-

tures. The two items assessing guilt were re-

tained, but the other items were felt to be either

too population-specific (e.g., homicidality and

disillusionment with authority) or too broad or

complex to be assessed with a single item (e.g.,

sadness and depression). Also, feedback indicated

that they were not routinely administered in most

settings. They were replaced with three items

that assess the dissociative symptoms of acute

stress disorder: reduction in awareness, derealiza-

tion, and depersonalization. The addition of

these items meant that the CAPS could be used

to assess acute stress disorder, either currently, if

administered within 1 month of the trauma, or

retrospectively.

The minor modifications included a) reordering the

items to correspond to the order of the DSM-IV diag-

nostic criteria; b) adding items to fully assess Criterion E

(duration requirement) and Criterion F (subjective dis-

tress and functional impairment requirement); c) renam-

ing the CAPS-1 and CAPS-2, as described earlier; d)

improving the formatting and typeface conventions; e)

eliminating the “at its/their worst” convention for the in-

tensity prompts; f) eliminating the phrase “without being

exposed to something that reminded you of the event”

from the frequency prompt for the first item assessing

intrusive recollections; and g) adding an instruction to

the interviewer to specify the basis of any QV (question-

able validity) ratings.

Completing this discussion on the development of

the CAPS are two significant, quite recent develop-

ments. One development is the decision to eliminate

the “two CAPS” system (i.e., the distinction between

the CAPS-1 or CAPS-DX and the CAPS-2 or CAPS-

SX) and create a single CAPS scale that can be used to

assess PTSD symptoms over the past week, past

month, or worst month since the trauma. As noted

earlier, the CAPS-2 or CAPS-SX was designed to

monitor changes in symptom status over a 1-week

time frame, and it appears to work well for this pur-

pose, demonstrating excellent psychometric properties

[Nagy et al., 1999]. The problem, however, is that for

the ten CAPS items where symptom frequency is

measured as “how often” versus “how much of the

time” (i.e., as the number of occurrences rather than

as a percentage of time) the CAPS-SX and CAPS-DX

had different values because of the different time

frame (i.e., for the past week time frame on the CAPS-

SX 0=never, 1=once, 2=two or three times, 3=four or

five times, and 4=daily or almost every day, but for the

past month time frame on the CAPS-DX 0=never,

1=once or twice, 2=once or twice a week, 3=several

times a week, and 4=daily or almost every day).

This means that scores on the two versions were not

directly comparable, with CAPS-SX scores tending to

yield lower scores when the reported frequency is in

the 3–5 times a week range. As a result, investigators

who wanted to use the CAPS to establish a PTSD di-

agnosis as an inclusion criterion, but were interested

in weekly assessment intervals over the course of the

study, needed to administer a CAPS-DX in the initial

evaluation, then administer a CAPS-SX at baseline,

mid-treatment, and post-treatment, and then a CAPS-

DX at long-term follow-up if they wished to assess

end-point diagnostic status. In general this is a work-

able scheme but proved to be needlessly cumbersome.

Therefore, on the recommendation of the CAPS Ad-

visory Group for the National Center for PTSD, the

CAPS-DX and CAPS-SX were combined into a single

version, which is now simply known as the CAPS.

This was accomplished by two minor modifications to

the CAPS-DX. First, the word “week” was provided as

an alternative to “month” in the prompt questions for

frequency [e.g., “How often have you had these

memories in the past month (week)?”]. Second, for

each item a space was provided to record frequency

and intensity ratings for “past week,” in addition to

“past month” and “lifetime.” When the new combined

version of the CAPS is used to assess 1-week symptom

status, frequency ratings for the ten items for which

frequency is rated as a count are scored as 0=never,

2=once or twice a week, 3=several times a week, and

4=daily or almost every day, skipping the value 1=once

or twice (a month). Thus, the combined CAPS is ap-

propriate for assessing 1-month or 1-week intervals

and yields comparable scores from either application.

The second development involved new options for

interpreting CAPS scores. First, nine scoring rules for

deriving a PTSD diagnosis have been developed and

compared on their psychometric properties and utility

for different assessment tasks [Weathers et al., 1999].

It should be emphasized that although several of these

rules appear to be quite useful, more research is

needed before firm recommendations can be made. A

number of other rules are possible and may prove to

have greater utility for some applications. Second, five

rationally derived severity score ranges for interpret-

ing CAPS total severity scores have been proposed

and are currently being evaluated. These categories

Review Article: Clinician-Administered PTSD Scale 135

are 0–19=asymptomatic/few symptoms, 20–39=mild

PTSD/subthreshold, 40–59=moderate PTSD/thresh-

old, 60–79=severe PTSD symptomatology, and >80=

extreme PTSD symptomatology. Finally, a rationally

derived 15-point change in CAPS total severity score

has been proposed as a marker of clinically significant

change. Again, it should be emphasized that these se-

verity score ranges and the 15-point marker are pre-

liminary, and unlike the scoring rules have not been

empirically evaluated, but they offer some guidance to

clinicians and investigators who use the CAPS to mea-

sure change.

In summary, the format and the procedures for ad-

ministering and scoring the CAPS have evolved in the

10 years since it was first developed. However, the

changes can be characterized as refinements rather

than major revisions, and the goal of backward com-

patibility of the latest CAPS with the original version

appears to have been accomplished [Weathers et al.,

1999]. The CAPS now provides a range of options re-

garding administration and scoring. Interviewers can

administer only the 17 core symptoms, all DSM-IV

criteria (A-F), or add the associated symptoms. Cur-

rent symptom status can be assessed for the past week

or past month, and lifetime status can be assessed for

the worst month since the trauma. By administering

the 17 core symptoms plus the 3 dissociative items the

CAPS can also be used to assess acute stress disorder.

In terms of scoring options, the CAPS can be used to

derive a PTSD diagnosis by using one or more of the

available scoring rules, or a continuous severity score

for each item, for the three symptom clusters or for

the entire syndrome. Total severity scores summed

over the 17 core symptoms can be interpreted with re-

spect to the five proposed severity score ranges, from

asymptomatic to extreme, and a 15-point change in

CAPS scores can be used to indicate clinically signifi-

cant change.

REVIEW OF THE CAPS-RELATED

LITERATURE

LITERATURE SEARCH AND SELECTION

OF STUDIES

We developed an initial list of studies to be included

by searching the phrase “Clinician-Administered

PTSD Scale” in the “Instruments” index of the PI-

LOTS database. PILOTS is the most comprehensive

database for the field of traumatic stress, containing

virtually every relevant citation in journals and book

chapters. This search, conducted in October 1999,

yielded 241 citations. We excluded book chapters, re-

view papers, dissertations, letters to the editor, an ar-

ticle on the child and adolescent version of the CAPS,

and several studies in which CAPS-related data were

included, but not in a form suitable for our purpose.

This narrowed the list to a total of 210 studies deemed

eligible for potential inclusion in our review.

For the purposes of this review, we divided the eli-

gible studies into three categories: a) psychometric

studies, which provided direct evidence of the reliabil-

ity and validity of the CAPS; b) pharmacotherapy and

psychotherapy studies, which provided evidence of the

sensitivity of the CAPS to clinical change; and c) case-

control studies, which provided additional validity evi-

dence based on conceptually meaningful differences

between individuals diagnosed with and without PTSD

using the CAPS. In the following sections, we summa-

rize all of the available studies in the first two categories

since there was a manageable number of them and they

provided the richest information regarding the utility of

the CAPS. However, due to space constraints, we limit

our discussion of studies in the third category to several

representative examples, since these were more nu-

merous and provided more limited validity evidence.

As noted earlier, the purpose of this review was to

examine all available research addressing the psychomet-

ric characteristics of the CAPS and its usefulness as a

standard criterion measure of PTSD. Accordingly, we

placed few restrictions in selecting the studies to be in-

cluded, realizing that the final set of studies would vary

widely in their quality of design and interpretability of

results. We felt that a consistent pattern of positive re-

sults across a large number of studies would provide un-

ambiguous support for the CAPS, and that if the studies

varied in quality, it would make an even stronger case

with regard to the generalizability of the findings. In the

process of evaluating a psychological assessment instru-

ment, each study, regardless of how well-designed and

executed it is, only contributes one piece of evidence and

can never be considered definitive. Conclusive answers

can be reached only by considering the accumulation of

several different types of evidence across different trauma

populations, settings, and research designs. In the next

section, we briefly review some fundamental psycho-

metric concepts in order to provide a conceptual

framework for organizing and evaluating the evidence

regarding the effectiveness of the CAPS.

PSYCHOMETRIC CONSIDERATIONS

Psychological assessment instruments are evaluated

with respect to two important characteristics: reliabil-

ity and validity. Reliability refers to the consistency of

test scores over repeated observations. Three com-

monly reported types of reliability include internal

consistency, test-retest reliability, and interrater reli-

ability, each of which addresses a different potential

source of error in test scores. Internal consistency re-

fers to consistency over different items on a test. Re-

quiring only a single administration of a test, it is

usually indexed by coefficient alpha (Cronbach’s al-

pha), which ranges from 0.00 to 1.00, with higher val-

ues reflecting a greater degree of intercorrelation

among the items. Item-scale total correlations, which

reflect how well each item correlates with the remain-

ing items, are another useful source of information

about internal consistency. Test-retest reliability refers

136 Weathers et al.

to consistency of test scores over repeated administra-

tions. It is estimated by administering a test twice and

calculating the correlation between the two scores.

Interrater reliability refers to consistency of test scores

over different raters. It is estimated by having two or

more raters evaluate and score responses and then cal-

culating either a correlation (only two raters) or

intraclass correlation (more than two raters) on the

scores. When an instrument is used to obtain a

dichomotomous score, as in the case of a present/ab-

sent diagnostic decision, interrater or test-retest reli-

ability is estimated by calculating a kappa coefficient, a

chance-corrected measure of agreement.

Two different research designs are typically em-

ployed to evaluate the reliability of a structured inter-

view such as the CAPS. In a simple interrater design,

two or more raters independently rate the same inter-

view. One rater administers and scores the interviews as

usual, while additional raters either observe the inter-

view live or, if more convenient, observe an audiotape

or videotape of the interview. Since the information

available to the raters is identical, the only potential

source of error is inconsistency in scoring among raters.

In a test-retest design, two independent raters adminis-

ter and score the interview on separate occasions. This

is a more stringent test of reliability because it involves

inconsistency in scoring plus two additional potential

sources of error: inconsistency in how raters ask the

questions and inconsistency in respondents’ answers.

Although we follow common practice in referring to it

as test-retest reliability, the reliability estimate this de-

sign yields is more precisely known as a coefficient of

stability and interrater equivalence because it involves

both occasions and raters as potential sources of error.

An important consideration for the test-retest design

is the interval between interviews. If the interval is too

brief, respondents’ answers in the second interview

may be influenced by their memory of their answers in

the first interview. If the interval is too long, genuine

change in clinical status may occur, meaning that in-

consistencies in responses are legitimate and not a

source of error. In the assessment of PTSD, an inter-

val of a few days to a week is probably reasonable for

most applications.

Although reliability clearly is a desirable character-

istic of an assessment instrument, a more important

concern is validity, which refers to the extent to which

evidence exists to support the various inferences, in-

terpretations, conclusions, or decisions that will be

made on the basis of a test. Traditionally, three types

of validity have been identified. The first type is con-

tent validity, which refers to evidence that items on a

test adequately reflect the construct being assessed.

The second type is criterion-related validity, which re-

fers to evidence that the test can predict some variable

or criterion of interest. The criterion may be mea-

sured either at the same time the test is administered

(concurrent validity) or at some point after the test

(predictive validity). The third type is construct valid-

ity, which refers to evidence that the test measures the

construct of interest and not other constructs. This

can be demonstrated, for example, by showing that the

test correlates strongly with other measures of the

same construct (convergent validity) but not with

measures of other constructs (discriminant validity).

However, this traditional approach to validity has

recently been superseded by the latest revision of the

Standards for Educational and Psychological Testing

[APA, 1999], which maintains the following.

“[Different] sources of evidence may illuminate dif-

ferent aspects of validity, but they do not represent

distinct types of validity. Validity is a unitary concept.

It is the degree to which all the accumulated evidence

supports the intended interpretation of test scores for

the proposed purpose. Like the 1985 Standards, this

edition refers to types of validity evidence, rather than

distinct types of validity.” (p. 11)

Thus, the new Standards argues for an integrative

approach to validity, emphasizing a confluence of va-

lidity evidence from different sources, and its updated

scheme for categorizing validity evidence represents a

marked departure from previous editions. Categories

include a) evidence based on test content; b) evidence

based on response processes, which focuses on respon-

dents’ behavior during the test process; c) evidence

based on internal structure, which focuses on relation-

ships among test items and components; d) evidence

based on relations to other variables, which includes

convergent and discriminant evidence, criterion-re-

lated evidence, and the generalization of validity to

new testing situations; and e) evidence based on con-

sequences of testing, which focuses on both the in-

tended and unintended outcomes of test use.

The new Standards also emphasizes that the process

of validation applies not to tests themselves but rather

to any specific interpretations that will be made on the

basis of test scores. Therefore, stating that a test is

valid begs the question: Valid for what purpose? To

address this question specifically with regard to the

CAPS, two main uses of the CAPS have been pro-

posed. One is to establish a dichotomous PTSD diag-

nosis and the other is to provide a continuous measure

of PTSD symptom severity. Thus, the two main inter-

pretations of CAPS scores that should be the focus of

validation are the following.

1. CAPS scores reflect severity of PTSD symptoms,

for individual symptoms, symptom clusters, or

the syndrome as a whole.

2. CAPS diagnoses reflect the presence or absence

of PTSD.

One source of validity evidence that applies to these

inferences is content-based evidence. This refers to

the extent to which the content of a test corresponds

to the construct being assessed. In this regard, the

CAPS was written and revised by a team of experts in

traumatic stress at the various branches of the Na-

Review Article: Clinician-Administered PTSD Scale 137

tional Center for PTSD. It was based directly on the

diagnostic criteria for PTSD in the DSM-III-R, and

now DSM-IV, and represents these criteria faithfully.

As noted earlier, the major revision of the CAPS that

followed the publication of the DSM-IV not only re-

flected changes in the PTSD criteria but also took

into account formal and informal feedback from a

broad cross-section of CAPS users in other clinical re-

search settings. Although difficult to quantify, there is

clearly a consensus among those familiar with the

CAPS that the content of the CAPS corresponds

veridically to the construct of PTSD.

A second source of validity evidence has to do with

the internal structure of the CAPS. As currently con-

ceptualized in the DSM-IV criteria, PTSD is a multi-

faceted syndrome that consists of three closely related

but distinct symptom clusters: re-experiencing, avoid-

ance and numbing, and hyperarousal. If PTSD is a syn-

drome, then there should a reasonably high degree of

correlation among all of the symptoms. If there are dis-

tinct but overlapping symptom clusters, then the items

within the clusters should correlate more strongly with

each other than they do with symptoms in other clus-

ters. These relationships would be reflected in alpha

coefficients and item-total correlations. Factor analysis,

especially confirmatory factor analysis, in which com-

peting hypotheses about the nature of PTSD can be

directly compared, is another means of evaluating the

internal structure of the CAPS.

A third, and particularly important, source of valid-

ity evidence involves the relationship between the

CAPS and other variables. As conceptualized in the

latest Standards, this source of evidence includes what

used to be referred to as construct and criterion-re-

lated validity, and encompasses a broad range of evi-

dence that the CAPS corresponds in theoretically

meaningful ways with measures of other constructs.

Relevant findings might include a) convergent evi-

dence, showing relatively strong correlations between

the CAPS and other measures of PTSD; b) discrimi-

nant evidence, showing relatively weak correlations

between the CAPS and measures of different con-

structs; c) evidence of test-criterion relationships,

showing the correspondence between the CAPS and a

criterion such as a PTSD diagnosis or an indicator of

clinically significant improvement in PTSD symptom

severity; d) evidence that groups formed on the basis

of the CAPS differ as hypothesized on some character-

istic or behavior; and e) evidence that PTSD preva-

lence, severity, or symptom profile based on the CAPS

vary as hypothesized in different groups.

PSYCHOMETRIC STUDIES

In this section, we describe the results of studies

that emphasized the psychometric properties of the

CAPS, including studies in which the CAPS was ei-

ther the primary instrument being investigated or was

included as a validational measure for another PTSD

instrument. First, we summarize studies that examined

reliability and convergent and discriminant validity.

Then we summarize studies that address two other

psychometric issues: the factor structure of the CAPS

and the utility of various scoring rules for converting

CAPS frequency and intensity scores into a dichoto-

mous PTSD diagnosis. In reviewing these studies, we

found that investigators often neglected to specify the

version of the CAPS they administered and the scor-

ing rule they used to determine a PTSD diagnosis.

The version could usually be readily inferred, and

with few exceptions was the CAPS-1 or CAPS-DX,

the current and lifetime diagnostic version. In our dis-

cussion of the studies in this section, then, “CAPS” re-

fers to the CAPS-1 or CAPS-DX, and “CAPS-2” is

used explicitly to refer to the weekly symptom-rating

version. Unless explicitly stated, however, the scoring

rule could not be determined. For the purposes of this

review we assumed, unless stated otherwise, that in-

vestigators used the original scoring rule, whereby a

frequency of “1” or higher and an intensity of “2” or

higher for a given CAPS item indicated symptom en-

dorsement.

Reliability, convergent and discriminant validity,

and diagnostic utility. The CAPS has been the pri-

mary focus of several psychometric investigations.

Blake et al. [1990] reported the first psychometric data

on the CAPS. In a pilot study they administered the

CAPS, the Combat Exposure Scale [CES; Keane et

al., 1989], the Mississippi Scale for Combat-Related

PTSD [Mississippi Scale; Keane et al., 1988], and the

Keane PTSD Scale of the MMPI [PK scale; Keane et

al., 1984] to 25 male combat veterans. To determine

interrater reliability for the CAPS, a second rater ob-

served and independently rated seven interviews. Ex-

cellent agreement was found between the two raters,

with reliability coefficients for frequency and intensity

scores across the three symptom clusters (re-experi-

encing, numbing and avoidance, and hyperarousal)

ranging from .92 to .99. The raters also demonstrated

perfect diagnostic agreement for the seven partici-

pants, five of whom had a positive diagnosis. Internal

consistency for the three PTSD symptom clusters was

high, with alpha coefficients ranging from .73 to .85

for the three symptom clusters. Regarding convergent

validity, the CAPS correlated strongly with the Missis-

sippi Scale (.70) and the PK scale (.84). It also corre-

lated .42 with the CES, a moderate correlation that is

typical for correlations between measures of trauma

exposure and measures of PTSD.

Hovens et al. [1994] examined the psychometric

properties of the CAPS in a Dutch sample, employing

translations of the CAPS and other PTSD measures.

Participants were 76 Dutch trauma survivors (51

males, 25 females), including combat veterans, resis-

tance veterans, and concentration camp survivors. Par-

ticipants were first diagnosed with or without PTSD,

using DSM-III-R criteria, on the basis of an unstruc-

tured clinical interview. They were then administered

the CAPS, the Mississippi Scale, the PK scale, and the

138 Weathers et al.

IES. Interrater reliability on the CAPS was evaluated

through simultaneous ratings of nine interviews by

two independent clinicians. Diagnostic agreement was

perfect for these nine participants. Furthermore, reli-

ability coefficients for frequency and intensity scores

for individual items were strong, ranging from .59 to

1.00 for frequency, with a mean of .92, and .52 to 1.00

for intensity, with a mean of .86. At the symptom clus-

ter level, reliability coefficients ranged from .92 to

1.00 for frequency and .92 to .98 for intensity. Re-

garding internal consistency, Hovens et al. [1994]

found alphas of .63 for re-experiencing, .78 for avoid-

ance and numbing, .79 for hyperarousal, and .89 for

all 17core PTSD symptoms. No rationale was given

for the decision to report internal consistency for in-

tensity scores but not for frequency or severity (fre-

quency + intensity) scores.

By using the clinical interview as the criterion,

Hovens et al. [1994] found that a CAPS-based PTSD

diagnosis had 74% sensitivity, 84% specificity, and

79% efficiency, and a kappa of .58. Because these fig-

ures were lower than expected, they examined discrep-

ancies between the clinical interview and the CAPS.

They concluded that in the clinical interview clini-

cians primarily emphasized re-experiencing symptoms

in making a PTSD diagnosis, failing to give sufficient

attention to the other two symptom clusters, particu-

larly avoidance and numbing. They further found that

many of the participants with discrepant diagnoses

were only mildly symptomatic and thus more diagnos-

tically ambiguous. As evidence of convergent validity,

the total CAPS score correlated .73 with the Missis-

sippi Scale, .74 with the PK scale, and .62 with the

IES total score. Finally, with the exception of amnesia,

the prevalence of each of the 17 core PTSD symptoms

on the CAPS was significantly greater in participants

with PTSD than in those without PTSD, indicating

robust discrimination between the two groups.

As part of an effort to develop and evaluate a com-

puter-administered version of the CAPS, Neal et al.

[1994] administered both the computerized and the

original interview versions of the CAPS to 40 military

personnel (36 males and 4 females) with mixed trauma

exposure, including combat, non-combat-related as-

saults, accidents, and disasters, and childhood physical

and sexual abuse. To evaluate the reliability of the

CAPS interview, ten participants were interviewed

twice by independent clinicians, resulting in perfect

diagnostic agreement. Treating the CAPS interview as

the criterion, the computerized version had 95% sen-

sitivity and 95% specificity, with a kappa of .90. Al-

though the interval between the two versions was not

specified, they appear to have been administered in a

single session, which could have inflated this high

level of agreement. An initial finding of a high corre-

lation (.96) between total frequency and total intensity

scores on both the interview and computerized ver-

sions of the CAPS led Neal et al. [1994] to use inten-

sity scores alone as a continuous measure of severity in

all further analyses. Internal consistency of intensity

scores was high for both versions, with an alpha of .90

and a median item-total correlation of .77 for the in-

terview version, and an alpha of .92 and a median

item-total correlation of .70 for the computerized ver-

sion. In addition, intensity scores on the two versions

were strongly correlated, ranging from .55 to .92 for

individual items and from .87 to .92 for the three

symptom clusters. The correlation for total intensity

score between the two versions was .95.

Hyer et al. [1996] investigated the utility of the

CAPS for assessing older combat veterans. Participants

were 125 male World War II and Korean combat veter-

ans. They were administered a computer-assisted version

of the SCID (SCID-DTREE), including the PTSD

module, as well as the CAPS, by two clinicians. They

also completed the Mississippi Scale, the IES, and the

CES. To assure the comparability of the SCID-DTREE

and the SCID, 25 participants were administered the

SCID in a separate testing session by an independent cli-

nician. In this subsample there was perfect agreement as

to PTSD diagnostic status, not only between the SCID-

DTREE and the SCID, but between the CAPS and

the SCID. In the full sample, against a PTSD diagno-

sis based on the SCID-DTREE, the CAPS had 90%

sensitivity, 95% specificity, and 93% efficiency, and a

kappa of .75. The CAPS also demonstrated high inter-

nal consistency, with alphas of .88 for re-experiencing,

.87 for avoidance and numbing, .88 for hyperarousal,

and .95 for all 17 core items. CAPS diagnosis was cor-

related .81 with the IES, .61 with the Mississippi

Scale, and .26 with the CES. The relatively low corre-

lation with the CES is likely attributable in part to a

restricted range on the CES, since most participants

had moderate to heavy combat exposure.

As part of a large prospective study on the effects of

trauma, Shalev et al. [1997] employed signal detection

methodology to determine whether the CAPS or any

of several questionnaire measures of PTSD, dissocia-

tion, and anxiety administered at 1 week or 1 month

post-trauma could predict PTSD diagnostic status at 4

months post-trauma. Participants included 207 (98

male and 109 female) victims of civilian trauma re-

cruited from the emergency room of a hospital. In

most cases, the traumatic event involved a motor ve-

hicle accident. Within a week of their trauma, partici-

pants completed the IES, the State form of the State

Trait Anxiety Inventory [STAI; Spielberger et al.,

1970], and the Peritraumatic Dissociative Experiences

Questionnaire [PDEQ; Marmar et al., 1997]. Assess-

ments at one month and 4 months post-trauma added

the CAPS and the civilian version of the Mississippi

Scale to this battery. They found that all of the

questionnaires administered at either 1 week or 1

month post-trauma were predictive of PTSD diag-

nostic status at 4 months, but that none of the ques-

tionnaires differed significantly in terms of accuracy

of prediction. In contrast, the CAPS at 1 month

post-trauma, used as a continuous measure, was a

Review Article: Clinician-Administered PTSD Scale 139

significantly better than all of the questionnaires in

predicting a 4-month diagnostic status that was also

based on the CAPS. Although Shalev et al. [1997]

did not identify an optimal cutoff score for CAPS

total severity, they did provide diagnostic utility

data for a range of selected cutoff scores. These data

indicate that a CAPS score of 40 yielded 93% sensi-

tivity and 80% specificity.

To determine the prevalence of PTSD in veterans

with spinal cord injuries, Radnitz et al. [1995] admin-

istered the CAPS and the SCID PTSD module to 126

male veterans receiving medical care for spinal cord

injuries in inpatient and outpatient settings. Current

and lifetime diagnostic status was assessed on both the

CAPS and the SCID. To determine diagnostic status

on the CAPS, Radnitz et al. [1995] used a variant of

the original scoring rule (i.e., frequency ≥ 1, intensity

≥ 2), whereby either the frequency or intensity of an

item had to be “2” or higher and the other dimension

had to be a “1” or higher. As described below, this

scoring rule was referred by Blanchard et al. [1995a,b]

as the “Rule of 3.” Although Radnitz et al. [1995] did

not provide kappas or other diagnostic utility statistics

except efficiency, we were able to calculate these from

data provided in the tables. Treating the SCID as the

criterion, for current diagnosis the CAPS had 83%

sensitivity, 94% specificity, 93% efficiency, and a

kappa of .73. For lifetime diagnosis the CAPS had

84% sensitivity, 90% specificity, 88% efficiency, and a

kappa of .74. Although not explicitly stated, it appears

that both interviews were administered by the same

research assistant in the same session. Both of these

factors, i.e., the lack of a time interval between inter-

views and the lack of an independent rater, could have

inflated the correlation between the CAPS and the

SCID. Finally, CAPS total severity scores appeared to

strongly differentiate between participants with and

without a PTSD diagnosis, although these mean dif-

ferences were not evaluated by statistical test.

Although all of these studies provide valuable infor-

mation, the most comprehensive investigations of the

psychometric properties of the CAPS, based on data

collected at the National Center for PTSD, are de-

scribed in two articles currently submitted for publica-

tion. Weathers et al. [1999a] examined the reliability and

validity of the CAPS-1/CAPS-DX in five samples of

male Vietnam veterans, including 267 veterans from four

different research projects and 571 veterans seen for

clinical services. To evaluate the test-retest reliability

(i.e., stability and rater equivalence) of the CAPS-1, 60

veterans were administered the CAPS twice, at a 2–3 day

interval, by independent clinicians. For the three symp-

tom clusters intraclass correlations ranged from .86 to

.87 for frequency, .86 to .92 for intensity, and .88 to .91

for severity. Across all 17 symptoms intraclass correla-

tions were .93 for total frequency, .95 for total intensity,

and .95 for total severity. Following the revision of the

CAPS for DSM-IV, the same design was implemented

for the CAPS-DX in a smaller sample of 24 veterans.

This study also yielded robust estimates of reliability,

with intraclass correlations of .91 for total frequency, .91

for total intensity, and .92 for total severity. Using the

optimal scoring rule, kappa, indicating test-retest reli-

ability for a CAPS-based PTSD diagnosis, was .89 in the

first sample and 1.00 in the second sample.

Examining internal consistency, Weathers et al.

[1999a], in a combined research sample of 243 veter-

ans, found alphas for the three symptom clusters rang-

ing from .78 to .87 for frequency, .82 to .88 for

intensity, and .82 to .88 for severity. Alphas for all 17

items were .93 for frequency, .94 for intensity, and .94

for severity. In the clinical sample, alphas for the three

symptom clusters ranged from .64 to .73 for fre-

quency, .66 to .76 for intensity, and .69 to .78 for se-

verity. Alphas for all 17 items were .85 for frequency,

.86 for intensity, and .87 for severity. The lower al-

phas in the clinical sample were likely due in part to a

restricted range in CAPS scores, since most veterans

referred for clinical services at the National Center re-

port moderate to severe PTSD symptoms; they may

also be due to a much larger and more diverse pool of

clinicians, relative to the small number of well-cali-

brated clinicians who administered the CAPS to the

research samples. Nonetheless, these scores provide

excellent evidence supporting the CAPS as used in a

clinical setting.

Weathers et al. [1999a] also reported validity evi-

dence for the CAPS, focusing primarily on convergent

and discriminant validity evidence and the diagnostic

utility of the CAPS against a PTSD diagnosis based

on the SCID. In the first research sample of 123 vet-

erans, the CAPS total severity score correlated .53

with the CES, .91 with the Mississippi Scale, .77 with

the PK scale, .89 with the number of PTSD symptoms

endorsed on the SCID, and .94 with the PTSD

Checklist [PCL; Weathers et al., 1993], a 17-item self-

report measure of PTSD. CAPS total severity corre-

lated somewhat less strongly, but still robustly, with

measures of depression (.61 to .75) and anxiety (.66 to

.76), findings that were expected given the substantial

overlap between PTSD, depression, and anxiety.

Much weaker correlations were observed between

CAPS total severity and measures of antisocial person-

ality (.14 to .33), a disorder conceptually distinct from

PTSD. In an effort to bring these convergent and dis-

criminant correlations into sharper relief, Weathers et

al. [1999a] then calculated partial correlations, con-

trolling first for nonspecific distress and symptom ex-

aggeration by using the F scale of the MMPI-2, then

for nonspecific distress again using the Global Sever-

ity Index (GSI) of the Symptom Checklist-90-Revised

(SCL-90-R; Derogatis, 1983]. After controlling for

the F scale, the CAPS demonstrated strong partial

correlations with measures of PTSD, including the

Mississippi Scale (.83), the PCL (.89), and the number

of PTSD symptoms on the SCID (.82). As predicted,

however, partial correlations between the CAPS and

measures of depression (.37 to .53) and anxiety (.37 to

140 Weathers et al.

.55) were markedly lower, and those between the

CAPS and measures of antisocial personality were es-

sentially zero (–.05 to .02). A similar, but even more

striking pattern was found after controlling for the

GSI. Further, these results involving the F scale were

generally replicated in a second research sample.

Finally, again focusing on the sample of 123 partici-

pants, Weathers et al. [1999a] reported the diagnostic

utility of three CAPS scoring rules for predicting a

SCID-based PTSD diagnosis. The original, rationally

derived scoring rule (frequency ≥ 1, intensity ≥ 2, or

F1/I2) had 91% sensitivity, 71% specificity, and 82%

efficiency, with a kappa of .63. These figures reveal the

F1/I2 rule to be relatively lenient, with excellent sensi-

tivity but only moderate specificity, suggesting that it

tends to somewhat overdiagnose PTSD relative to the

SCID. The two other rules were empirically derived on

this sample. The second rule, which assigns a positive

diagnosis if the CAPS total severity score is 65 or

greater (TSEV65), had 82% sensitivity, 91% specificity,

and 86% efficiency, with a kappa of .72. Although the

higher kappa indicates a better correspondence with the

SCID than the F1/I2 rule has, the TSEV65 rule ap-

pears to be relatively stringent, tending to somewhat

underdiagnose PTSD relative to the SCID. The third

rule, derived by empirically calibrating each CAPS

symptom with the analogous SCID symptom (SX-

CAL), had the closest correspondence to the SCID,

with 91% sensitivity, 84% specificity, and 88% effi-

ciency, with a kappa of .75. Although these results re-

quire cross-validation, the SXCAL rule appears to be

the optimally efficient rule, and therefore the best

choice for differential diagnosis.

In the second article based on National Center data,

Nagy et al. [1999] described the only comprehensive

investigation of the CAPS that focused specifically on

the CAPS-2. To evaluate interrater reliability, Nagy et

al. administered the CAPS-2 to 30 (29 male, 1 female)

inpatients and outpatients in treatment for PTSD, all

but two of whom were combat veterans. Interviews

were videotaped and scored by three additional raters,

resulting in four ratings for each participant. Intraclass

correlations ranged from .76 to .99 for the 17 core

PTSD symptoms, and from .92 to .97 for the three

symptom clusters, with values of .98 for total frequency,

.96 for total intensity, and .98 for total severity. Internal

consistency and convergent and discriminant evidence

were examined in two additional samples of male com-

bat veterans: 20 veterans enrolled in a pharmacologic

trial and 37 veterans in inpatient PTSD treatment pro-

gram. All participants were administered the CAPS and

the IES. In addition, the 20 participants in the drug

trial were administered the Hamilton scales for depres-

sion and anxiety (HAM-D and HAM-A), and the 37 in-

patients completed the BDI and the Beck Anxiety

Inventory [BAI; Beck et al., 1988].

In the combined sample alphas were .25 for re-ex-

periencing, .69 for avoidance and numbing,.70 for

hyperarousal, and .79 for all 17 items. In the com-

bined sample, the CAPS correlated .37 with the IES.

For the participants in the drug trial the CAPS corre-

lated .34 with the HAM-D and .36 with the HAM-A.

In the inpatient sample the CAPS correlated .67 with

the BDI and .51 with the BAI. Taken together, these

results are generally in line with results from studies

involving the CAPS-1. However, the alpha for the re-

experiencing cluster and the correlation of the CAPS

with the IES were lower than those found previously.

It is unclear whether these findings are sample-specific

and reflect some idiosyncrasies of the particular par-

ticipants or settings in the study or whether they are

attributable to some aspect of the CAPS-2.

Although not designed primarily as psychometric

investigations of the CAPS per se, other investigations

have nonetheless provided additional evidence of its

reliability and validity. Hovens et al. [1994] used the

CAPS as a criterion measure in the evaluation of a

new self-report measure of PTSD, the Self-Rating In-

ventory for Posttraumatic Stress Disorder (SIP). The

SIP consists of 51 items, 22 assessing DSM-III-R

PTSD symptoms and 29 measuring other trauma-re-

lated sequelae, particularly those associated with the

proposed diagnostic category of disorders of extreme

stress not otherwise specified (DESNOS). This study

included two samples: the same 76 participants used in

their previous study on the CAPS plus 59 (22 male

and 37 female) psychiatric outpatients. Although the

psychiatric outpatients were not selected on the basis

of a known trauma history, 18 of them reported expo-

sure to various types of civilian trauma, including

sexual and physical assault, traumatic loss of a loved

one, and motor vehicle accidents. Combining all par-

ticipants with a trauma history across the two samples,

Hovens et al. [1994] found that the CAPS correlated

.73 with total SIP score, .75 with the DSM-III-R

items on the SIP, .70 with the civilian version of the

Mississippi Scale, .72 with the PK scale, and .61 with

the IES. Correlations for the DSM-III-R symptom

clusters between the CAPS and the SIP were .54 for

re-experiencing, .69 for avoidance and numbing, and

.71 for hyperarousal.

Two studies by Neal and colleagues also provide

convergent validity evidence. First, Neal et al. [1994]

assessed 70 (59 male and 11 female) military personnel

with mixed military and civilian trauma exposure, pre-

sumably similar to, or overlapping with, the sample

they evaluated for their study on the computerized

CAPS described earlier. They examined the correla-

tions of two CAPS variables, total intensity and num-

ber of symptoms endorsed, with the PK scale, the IES,

and the GSI of the SCL-90. Although no rationale

was offered for why they used intensity rather than se-

verity scores, presumably this was because of the high

degree of correlation between frequency and intensity

scores they found in their previous study on the com-

puterized CAPS. Similar patterns of correlations were

found for both CAPS variables. Total CAPS intensity

correlated .85 with the PK scale, .78 with the IES, and

Review Article: Clinician-Administered PTSD Scale 141

.77 with the SCL, whereas the number of CAPS

symptoms correlated .84 with the PK scale, .81 with

the IES, and .74 with the SCL-90. The strong corre-

lations with the PK scale and the IES offer convergent

evidence, but the nearly as strong correlations with

the SCL-90 failed to provide strong discriminant evi-

dence. However, given the high rates of comorbidity

found in PTSD and given that the SCL-90 primarily

reflects nonspecific distress, the SCL-90 is not an op-

timal measure for discriminant evidence. Second, Neal

et al. [1995] administered the CAPS, the IES, the PK

scale, and the Mississippi scale to 30 (29 male and 1

female) World War II prisoners of war. In this study

the CAPS correlated .63 with the IES, .71 with the

PK scale, and .81 with the Mississippi Scale, again

providing convergent evidence for the CAPS as a

measure of PTSD.

Two studies by Blanchard and colleagues provide evi-

dence regarding interrater reliability and convergent

validity. Blanchard et al. [1995b] employed the CAPS as

the primary diagnostic measure of PTSD in a study of

male and female motor vehicle accident victims. All

CAPS interviews were audiotaped and an independent

rater re-scored 15 randomly selected interviews. Inter-

rater reliability for individual items ranged from .82 to

.99, with a mean of .98, and kappa for a PTSD diagno-

sis was .81. Blanchard et al. [1996a,b] also used the

CAPS as the criterion measure in a psychometric evalu-

ation of the PCL. Participants were 27 (3 male and 24

female) motor vehicle accident victims and 13 female

sexual assault victims. Interrater reliability for the

CAPS, based on 19 audiotaped and independently re-

scored interviews, was again quite strong. Coefficients

for individual items ranged from .84 to .99 for indi-

vidual items, with a mean of .94, and kappa for a

PTSD diagnosis was .84. Correlations between the

PCL and the CAPS supplied convergent evidence.

Correlations between PCL items and corresponding

CAPS items ranged from .39 to .79, with all but three

correlations above .60 and seven correlations above

.70. In addition, the correlation between the total

scores on the PCL and the CAPS was .93.

Finally, two studies utilized the CAPS in the valida-

tion of the Davidson Trauma Scale [DTS; Davidson et

al., 1997], a 17-item self-report measure of DSM-IV

PTSD symptoms. Like the CAPS, the DTS assesses

PTSD symptoms on two dimensions: frequency, which

corresponds to the frequency dimension on the CAPS,

and severity, which corresponds to the intensity dimen-

sion of the CAPS. The DTS assesses symptoms over

the previous week. To obtain convergent evidence for

the DTS, Zlotnick et al. [1996] administered the DTS

and the CAPS to 50 female sexual abuse survivors.

They found correlations of .72 between DTS total fre-

quency and CAPS total frequency and .57 between

DTS total severity and CAPS total intensity. For total

DTS and total CAPS scores for each of the three symp-

tom clusters, they found correlations of .70 for re-ex-

periencing, .53 for avoidance, and .73 for hyperarousal.

As part of a comprehensive psychometric investigation

of the DTS, Davidson et al. [1997] administered the

DTS and the CAPS to a mixed sample of 102 female

sexual assault victims and male combat veterans, find-

ing a correlation of .78 between total scores on the

DTS and CAPS.

CAPS factor structure. The final two issues we

will discuss address the factor structure of the CAPS

and the development and evaluation of various scoring

rules for deriving a CAPS-based PTSD diagnosis.

Two studies have examined the factor structure of the

CAPS using confirmatory factor analysis. Buckley et

al. [1998] tested a single hypothesized factor structure

consisting of two factors: a) Intrusion and Avoidance

and b) Hyperarousal and Numbing. Although these

factors cut across the three DSM-III-R and DSM-IV

symptom clusters of PTSD, there is theoretical and

empirical justification for this two-factor structure. In

fact, Buckley et al. [1998] sought to replicate a previ-

ous study by Taylor et al. [1998], in which this struc-

ture was derived in an exploratory factor analysis.

Analyzing CAPS scores from a combined sample of

217 male and female motor vehicle accident victims,

Buckley et al. [1998] found support for the hypoth-

esized two-factor structure across several indices of

model fit.

In a more comprehensive analysis, King et al. [1998]

conducted a confirmatory factor analysis of CAPS

scores in 524 male combat veterans seen for clinical ser-

vices at the National Center for PTSD in Boston. In

this study King et al. [1998] tested four competing

models, three of which involved dividing Criterion C

(numbing and avoidance) into two distinct factors of

effortful avoidance (criteria C1 and C2) and emotional

numbing (criteria C3–C7). The first model was a four-

factor, first-order solution consisting of four correlated

primary factors: re-experiencing, effortful avoidance,

emotional numbing, and hyperarousal. The second

model, which was similar to the one Buckley et al.

[1998] evaluated, was a two-factor, higher-order solu-

tion, with one factor comprising re-experiencing and

effortful avoidance and the other comprising emotional

numbing and hyperarousal. The third model was a

single factor, higher-order solution that hypothesized a

single PTSD factor comprising the four symptom clus-

ters. The fourth model was a single-factor, first-order

solution that hypothesized that all 17 symptoms load on

a single PTSD factor. King et al. [1998] found that the

first model provided the best fit to the data, suggesting

that PTSD, as assessed by the CAPS, consists of four

correlated but distinct symptom clusters. This finding

supports the CAPS as a measure of PTSD in that the

internal structure of the CAPS corresponds to the

DSM PTSD symptom clusters, albeit with the addi-

tional, conceptually meaningful distinction between

effortful avoidance and emotional numbing.

CAPS scoring rules. Finally, one of the recent de-

velopments in the CAPS has been the explication and

evaluation of various rules for converting continuous

142 Weathers et al.

CAPS scores into a dichotomous PTSD diagnosis.

From the outset it was recognized that the original,

rationally derived F1/I2 rule described earlier was

only an initial working rule that might be replaced by

others once sufficient empirical evidence had accumu-

lated. Over time, a number of new rules have been

proposed and have recently appeared in the literature.

Blanchard et al. [1995a] were the first to compare the

impact of adopting different scoring rules. In an inves-

tigation of 100 (35 male and 65 female) motor vehicle

accident victims they proposed and evaluated three

different scoring rules, all of which involved convert-

ing CAPS frequency and intensity scores into a di-

chotomous score for each symptom, then following

the DSM requirements (one re-experiencing symp-

tom, three numbing and avoidance symptoms, and

two hyperarousal symptoms) to derive a PTSD diag-

nosis. According to the Rule of 2, a symptom is consid-

ered present if the severity score for an item (frequency

+ intensity) is ≥ 2 (i.e., frequency and intensity are both

≥ 1). Similarly, the Rule of 3 requires an item severity

score > 3 (either frequency or intensity is ≥ 2 and the

other is ≥ 1). This is similar to but more inclusive than

the original F1/I2 rule. Last, the Rule of 4 requires an

item severity ≥ 4. Blanchard et al. found that the three

rules yielded markedly different PTSD prevalence esti-

mates, with 44% for the Rule of 2, 39% for the Rule of

3, and 27% for the Rule of 4. Furthermore, they found

that participants who met the Rule of 4 had higher

scores on measures of depression and anxiety, and

greater functional impairment, relative to those who

only met the Rule of 3.

More recently, Weathers et al. [1999b] described

and compared nine scoring rules, drawing on data

from the same five samples in the Weathers et al.

[1999a] psychometric article described earlier. Four of

the nine rules were rationally derived, including the

original F1/I2 rule, the Item Severity ≥ 4 (ISEV4)

rule, which is identical to Blanchard’s Rule of 4, and

two rules based on clinicians’ judgments regarding

which frequency/intensity combinations constitute a

symptom. The other five rules were empirically de-

rived, including four rules calibrated in various ways

against the SCID PTSD module, and one rule identi-

fied by Orr [1997], based on a study of physiological

reactivity in female incest survivors. Kappa coeffi-

cients indicating test-retest reliability for the rules

ranged from .72 to .90 in an initial sample of 60 veter-

ans, and from .68 to 1.00 in a follow-up sample of 24

veterans. Kappa coefficients for predicting a PTSD di-

agnosis based on the SCID ranged from .63 to .75. As

in the Blanchard et al. [1995a] study, the nine rules

yielded widely varying prevalence estimates, ranging

from 26% to 49% in a combined research sample of

243 veterans and 47% to 82% in a clinical sample of

571 veterans. The F1/I2 rule was the most lenient in

the clinical sample and second most lenient in the re-

search sample. The two rules based on clinicians’ rat-

ings were the most stringent in both samples. Also,

compared to participants who met criteria only by the

F1/I2 rule, those who met criteria for the most strin-

gent rule had significantly higher scores on measures

self-report measures of PTSD, depression, anxiety,

and nonspecific distress.

A third study, by Fleming and Difede [1999], exam-

ined the impact of adopting different scoring rules on

the CAPS-2 in a sample of hospitalized burn patients.

Although they recognized that the CAPS-2 was not

suitable for a diagnosis of PTSD because of the one-

week time frame, they deliberately chose it for their

study because they were interested in acute PTSD

symptoms within the first 2 weeks after the trauma.

Administering the CAPS-2 to 69 (48 male and 21 fe-

male) participants, they compared the effects of adopt-

ing essentially the same scoring rules described by

Blanchard et al. [1995a]. The one exception was that

Fleming and Difede appear to have used the F1/I2

rule rather than the more inclusive Rule of 3 of

Blanchard et al. [1995a]. Compared to the previous

two studies, they found less variability among the dif-

ferent rules in terms of estimated prevalence of

PTSD. The Rule of 3 and the Rule of 4 both yielded a

prevalence of 25%, while the Rule of 2 yielded a

prevalence of 32%. Furthermore, they found no sig-

nificant differences on the IES or self-report measures

of acute stress and nonspecific distress between par-

ticipants who met criteria only by the Rule of 2 and

those who met criteria by the Rule of 3 or the Rule of

4. However, differences were found on all self-report

measures between all participants who met criteria for

PTSD by at least the Rule of 2 and those who did not

meet criteria for PTSD by any of the rules.

Taken together these three studies of scoring rules

for the CAPS indicate that there are important conse-

quences to adopting a particular rule. Prevalence esti-

mates can vary considerably and participants who

meet criteria by lenient rules may be less symptomatic

and less impaired relative to those who meet criteria

by more stringent rules. Weathers et al. [1999b] dis-

cuss three implications of these findings. First, investi-

gators should always explicitly describe and defend

their choice of a CAPS scoring rule. Second, for many

applications, an efficient and informative strategy

would be to use several scoring rules, ranging from le-

nient to stringent, and compare the different results

obtained. Third, when using different scoring rules is

not feasible, investigators should select scoring rules

that are best suited for the purpose of the study. Le-

nient scoring rules are most appropriate for screening,

when a lower threshold for diagnosis is needed to

avoid false negatives. Stringent rules are most appro-

priate for confirming a diagnosis or creating an unam-

biguous PTSD group for case-control research, when

a higher threshold is need to avoid false positives.

Moderate rules are most appropriate for differential

diagnosis, when false negatives and false positives are

weighted equally and the goal is to minimize the over-

all number of diagnostic errors.

Review Article: Clinician-Administered PTSD Scale 143

Finally, we note that it is possible that some of the

diagnostic utility data cited for the CAPS in this sec-

tion, even though it is consistently high, might actu-

ally have been stronger had different scoring rules

been applied. In discussing the articles in this section,

we assumed that unless stated otherwise investigators

used the original F1/I2 rule to derive a PTSD diagno-

sis from CAPS scores. However, the Weathers et al.

[1999b] article, in particular, demonstrated that the

F1/I2 rule is a relatively liberal rule and may not be

optimal for differential diagnosis.

Discussion. Considering all the accumulated evi-

dence, the CAPS appears to have excellent psycho-

metric properties across a wide variety of clinical

research settings and trauma populations. Interrater

reliability for continuous CAPS scores was consis-

tently at the .90 level and above, with diagnostic

agreement at times reaching 100%. Test-retest reli-

ability, a more stringent measure of agreement, was

nearly as strong, although it was only evaluated in one

study and needs replication. These findings suggest

that trained and calibrated raters can achieve a high

degree of consistency in using the CAPS to diagnose

PTSD and rate PTSD symptom severity. In addition,

internal consistency was generally high, with alphas

typically in the .80 to .90 range for the three PTSD

symptom clusters and for the entire syndrome.

Although somewhat more variable and therefore

more difficult to easily summarize, evidence of validity

was also strong. Regarding convergent evidence, the

CAPS generally demonstrated correlations at the .70

level and above with self-report measures of PTSD

such as the Mississippi Scale, the PK scale, the IES,

the PCL, and the DTS, often reaching the .80 to .90

range. Diagnostic utility of the CAPS was evaluated in

five studies, and with one exception in which the crite-

rion was a clinical diagnosis based on an unstructured

interview, was quite robust, with sensitivities and

specificities above .80, and often above .90, and kappas

above .70. To date, however psychometric studies of

the CAPS offer little in terms of discriminant evi-

dence. More data on this are needed. Because indi-

viduals with PTSD, especially chronic PTSD, often

have comorbid disorders and experience high levels of

distress, it may prove to be difficult to obtain un-

equivocal discriminant evidence, particularly with

measures of depression and anxiety, since these two

constructs overlap conceptually with PTSD. Weathers

et al. [1999a] tried to address this problem by includ-

ing measures of a construct conceptually unrelated to

PTSD (antisocial personality) and by partialing out

the effects of nonspecific distress. These two ap-

proaches appeared to be successful in providing dis-

criminant evidence, but more creative research on this

issue is needed.

We close this section with a brief discussion of some

fundamental questions regarding the psychometric in-

vestigation of the CAPS. First, regarding convergent

and discriminant evidence there are no absolute stan-

dards for what constitutes “good” evidence. How large

should convergent validity coefficients be? How small

should discriminant validity coefficients be? How

large a difference should there be between convergent

and discriminant coefficients? Reasonable answers to

these questions must be informed by a well-articulated

theoretical model and ultimately based on expert

judgment.

Second, is it appropriate to evaluate a putative “gold

standard” such as the CAPS against self-report mea-

sures? When a correlation between the CAPS and an-

other measure is lower than expected it is unclear if the

“problem” lies with the CAPS or with the alternative

measure, or a combination of both. This question is

particularly important with respect to self-report mea-

sures of PTSD, which are subject to misinterpretation

and to response biases such as social desirability, exag-

geration, minimization, and even random responding.

In addition, they vary significantly in format, including

their correspondence with DSM criteria for PTSD, the

dimension of symptom severity they emphasize (e.g.,

subjective distress, functional impairment, and fre-

quency), and the time frame they assess (past week and

past month). Finally, they vary in the quality of their

psychometric properties. Any of these characteristics,

alone or in combination with characteristics of different

samples, could affect their correlation with the CAPS.

In general, in PTSD research, as in other areas of psy-

chopathology, the diagnostic standard is a clinical inter-

view because interviewers can clarify as needed, ask for

examples, observe clinically relevant behaviors, and

evaluate potential response bias. Most importantly, with

an interview it is ultimately the clinician who makes the

final rating, not the participant.

This, then, raises a third question. What measure

should serve as the criterion for evaluating the diagnos-

tic utility of the CAPS? Part of the problem is that

there is no other single measure that has been widely

accepted as a criterion measure of PTSD. The SCID

PTSD module comes the closest, but there is evidence

that suggests that it may not be as reliable as the CAPS,

which sets an upper limit on how well the CAPS can

perform in predicting it. In fact, as Weathers et al.

[1999b] have argued, the CAPS appears to be more

strongly associated with the SCID PTSD module than

the SCID PTSD module is with itself. Another possi-

bility might be to use a multiple converging measures

approach, such as was used in the National Vietnam

Veterans Readjustment Study [NVVRS; Kulka et al.,

1990] or the so-called LEAD standard approach pro-

posed by Spitzer and colleagues. Both approaches could

readily be applied to the CAPS and would provide valu-

able new information.

TREATMENT OUTCOME STUDIES

Design and analysis issues. In this section, we de-

scribe pharmacological and psychosocial treatment

outcome studies that employed the CAPS as a primary

outcome measure. Our main focus in this section is on

144 Weathers et al.

the ability of the CAPS to detect genuine changes in

PTSD symptom severity in the context of a clinical

intervention. A key question addressed in this section

is “What empirical results would constitute evidence

supporting the claim that the CAPS is in fact sensitive

to change?” We hypothesize four results that we

would expect to occur in a treatment outcome study if

this claim is true. First, we would expect to find a re-

duction in CAPS scores from pre-treatment to post-

treatment. This should be true for virtually any

intervention, for any of the following reasons:

1. Possible placebo effects.

2. Possible statistical regression (i.e., participants

selected on basis of extreme scores tend to show

less extreme scores on subsequent testing).

3. The fact that repeated assessment, particularly in-

terview-based assessment, may be considered an in-

tervention in and of itself since it includes many

putative active ingredients of psychotherapy, in-

cluding a) a safe, professional interpersonal context;

b) therapeutic exposure and emotional and cogni-

tive processing through disclosure of painful aspects

of the trauma and trauma-related symptoms; and c)

education about PTSD symptoms and self-moni-

toring.

Second, if a study includes one or more comparison

groups, there should be greater improvement in the

group or groups that receive a more potent treatment

or a treatment with more putative active ingredients

of therapy. Third, changes on the CAPS should paral-

lel changes in other measures of PTSD. Finally, if the

active therapy ingredient targets PTSD specifically,

then the CAPS should show greater reduction relative

to measures of other constructs such as depression,

anxiety, and global distress and impairment.

In reviewing these studies, we focused only on data

related specifically to the CAPS. It was not our intent

to address the effectiveness of pharmacological or psy-

chosocial treatments for PTSD per se or to rigorously

critique the research methodology of the various stud-

ies. Nonetheless, within this limited scope of our re-

view, we identified several issues with regard to the

reporting of CAPS data that required several decisions

about how to extract and summarize CAPS-related re-

sults and present them in a standard format. First,

studies varied considerably in terms of the outcome

measures they included and how the data were re-

ported and analyzed, differing on a) which CAPS

scores were included (e.g., frequency, intensity, or se-

verity scores for individual items, for the three symp-

tom clusters, or for the syndrome as a whole); b)

which additional measures were included; c) how

scores were presented (e.g., means and totals); d) how

change was quantified (e.g., change scores, percent

change, statistical significance, effect size, and graphic

presentation only); and e) how complete the data

analyses were. In general, in response to this variabil-

ity, we tried to extract the results most relevant to the

CAPS and present them as uniformly as possible. For

the purposes of this review, we used percent change as

the primary metric for comparing results across stud-

ies and across instruments within the same study. This

is a commonly reported metric, particularly in the

pharmacology literature. It is easily calculated when

not provided, readily comprehensible, and applicable

for any type of study, from case studies to large ran-

domized trials. Where possible, we identified or calcu-

lated percent change for the primary outcome

variables in each of the studies. In addition, we in-

cluded the results of statistical significance tests of key

comparisons when they were provided.

Second, studies varied in terms of how many mea-

surement points they included. All studies included as-

sessments at pre-treatment and post-treatment, but

others included assessments at screening, extended

baseline, pre-treatment, post-treatment, additional in-

tervals during treatment, and one or more long-term

follow-ups. To simplify our presentation, whenever

possible, we examined only pre-post changes for all

studies. These data were available for almost all stud-

ies and were sufficient as evidence of the sensitivity of

the CAPS to clinical change. Also, in the studies that

presented additional follow-up data, pre-post changes

were generally sustained and sometimes continued to

improve, so little would have been gained by examin-

ing additional assessment periods.

Third, there was some ambiguity with regard to the

terms investigators used to describe their study de-

signs. Terms such as open trial, uncontrolled trial, and

open label do not adequately characterize the essential

aspects of the research designs they were used to de-

scribe nor were they used consistently across studies.

The questions we used as a guide in depicting the

various research designs were the following.

1. Is the treatment condition known to the partici-

pant?

2. Is the treatment condition known to the assessor?

3. Is there at least one comparison condition?

4. Is the comparison condition within-subjects, as in

a crossover design, or between-subjects, as in a

randomized controlled trial?

Answers to these questions were not always stated

explicitly, although the investigators may have in-

tended to imply them by the labels they used to de-

scribe their studies. In particular, unless otherwise

specified, we assumed that assessments were not

blinded. Fourth, studies often did not explicitly iden-

tify which version of the CAPS was used. This could

sometimes be inferred, but in general, unless there

was some specific indication that the CAPS-2/CAPS-

SX was used, we assumed that the CAPS-1/CAPS-DX

was used. Finally, the final sample size often differed

from the initial one due to attrition and inclusion/ex-

clusion criteria. We report the sample size on which

the final data analyses were based.

Review Article: Clinician-Administered PTSD Scale 145

Pharmacological and psychosocial treatment

studies. In this section, we review 10 pharmacological

and 19 psychosocial treatment studies that used the

CAPS as a primary outcome measure. These studies

and their key findings relevant to the CAPS are pre-

sented in Tables 1 and 2. We consider the results with

respect to the four issues outlined above with regard to

evidence of sensitivity to clinical change, including

within-groups effects (pre-post change), between-

groups effects (differential change due to nature of in-

tervention, e.g., drug versus placebo), change on the

CAPS relative to change on other measures of PTSD,

and change on the CAPS relative to measures of other

constructs (e.g, anxiety, depression, and global distress

and functional impairment). Whenever possible we

present the percent change values for each measure de-

scribed in Tables 1 and 2. However, some studies only

reported the results of significance tests and did not in-

clude actual values for one or more key measures. The

studies in Tables 1 and 2 are arranged chronologically

and numbered within each table. For ease of presenta-

tion in the following sections, we refer to studies by

number rather than by author(s) and year.

Within-groups effects. Among the pharmacological

studies, there was a significant reduction in CAPS to-

tal score in eight of the nine studies that reported in-

ferential statistics (Table 1, all but Study 5 reported

significance levels; all of those but Study 8 were sig-

nificant). Considering only participants who received

a drug, for the nine studies that reported actual CAPS

score values (all but Study 2), the reduction in CAPS

total score ranged from 10–63%, with a median of

33%. The psychosocial studies yielded similar find-

ings, with evidence of even greater improvement.

There was a significant reduction in CAPS total score

in 10 of the 13 studies that reported inferential statis-

tics (Table 2, Studies 1–4, 6, 8, 10–13, 15, 16, and 19

reported significance levels; all of those but Studies 1,

6, and 8 were significant). Considering the partici-

pants who received an active intervention and showed

the most improvement, for the studies that reported

actual CAPS score values (all but Studies 1, 8, and 16),

the reduction in CAPS total score ranged from 19–

100%, with a median of 50%.

Between-groups effects. Overall, there were relatively

few controlled trials. Of the ten pharmacological stud-

ies, only three were randomized, placebo-controlled

trials (Table 1, Studies 2–4). Two of these (Studies 2 and

3) found significantly greater reduction in CAPS scores

for the drug group relative to the placebo group. The

third study (Study 4) found slightly greater improve-

ment for drug versus placebo, although the effect was

not significant. Similarly, of the 19 psychosocial studies,

only 6 were randomized, controlled trials (Table 2,

Studies 1, 4, 12, 13, 16, and 19), although an additional

3 studies (2 crossover designs and 1 program evalua-

tion) included a comparison condition (Studies 8, 10,

and 17). Only two of the six randomized, controlled tri-

als (Studies 12 and 16) found significant between-

groups effect, with significantly greater reduction in

CAPS scores for a more active, trauma-focused inter-

vention than for a control condition. Two of the other

four studies (Studies 4 and 13) found greater improve-

ment for active interventions relative to control condi-

tions, but the effects were not significant. Of the

remaining two studies, Study 19 included two active

interventions, which showed substantial, equivalent

improvement but no minimal intervention control con-

dition; Study 1 employed a very brief intervention and

found no within-groups or between-groups changes on

any measures. Finally, in Study 10, a quasi-experimental

program evaluation, between-groups differences were

found among three types of PTSD inpatient programs.

CAPS versus other PTSD measures. In general,

CAPS results matched the results for self-report PTSD

measures, particularly the IES. Among the pharmaco-

logical studies, the CAPS had comparable results to the

IES in three studies (Table 1, Studies 1, 4, and 6) and

to the DTS in two other studies (Studies 5 and 9), with

differences ranging from 0–7 percentage points. For

the psychosocial studies differences between the CAPS

and other PTSD measures were more variable and

somewhat larger. Eight studies (Table 2, Studies 2, 4, 5,

11, 12, 15, 16, and 19) found a greater reduction on the

CAPS relative to the IES, with differences ranging

from 1–24 percentage points. On the other hand, four

studies (Studies 3, 9, 14, and 18) found a greater reduc-

tion on the IES, with differences ranging from 7–26

percentage points. In addition, the CAPS showed a

comparable or greater reduction relative to the PK

scale (Study 1), the Mississippi Scale (Study 5), the Ci-

vilian Mississippi Scale (Study 15), the PSS (Study 9),

the MPSS-SR (Study 7), the PCL (Study 11), and the

Penn Inventory (Study 19).

CAPS versus measures of depression. All but two of

the pharmacological studies included a measure of de-

pression, primarily the HAM-D and MADRS. Four

studies (Table 1, Studies 1, and 8–10) found greater re-

duction on the HAM-D relative to the CAPS, with

differences ranging from 2–13 percentage points. A

fifth study (Study 2) found significant within-groups

and between-groups effects for both the CAPS and

the HAM-D but did not report actual rating scale val-

ues. However, two studies (Studies 6 and 7) found

greater reduction on the CAPS relative to the MADRS,

with differences of 9 and 10 percentage points, respec-

tively. One study (Study 5) found slightly greater reduc-

tion on the CAPS relative to the BDI. The BDI was

also included in 12 of 19 psychosocial studies, with ten

(Table 2, Studies 2, 6, 7, 11–13, 15, 16, 18, and 19) find-

ing greater reduction on the CAPS and two (Studies 5

and 9) finding equivalent reduction on the two scales.

Except for a case study (Study 7), which found a 48%

reduction on the CAPS and an 18% increase on the

BDI, the greater reduction on the CAPS ranged from

2–18 percentage points.

CAPS versus measures of anxiety. Four pharmaco-

logical studies included the HAM-A. Three (Table 1,

TABLE 1. Summary of CAPS findings from pharmacological treatment studies of posttraumatic stress disorder*

Authors (year) Participants Design Drug Duration Key CAPS-related findings

1. Nagy et al. [1993] Male combat 1. Non-blinded, Fluoxetine 10 weeks 1. Significant reduction in CAPS-2 total score (34%)

veterans (N=19) uncontrolled 2. Comparable reduction on IES (39%)

2. CAPS-2 3. Somewhat larger reduction on HAM-D (47%) and HAM-A (41%)

4. With response defined as 50% reduction in CAPS total score, a 2-point

improvement on CAPS global severity rating, and consensus of two clinicians, 7

participants (37%) had good responses, 5 (26%) had partial response, and 7 (37%)

did not respond

2. van der Kolk et al. 1. Civilian with mixed Double-blind, Fluoxetine 5 weeks 1. Significantly greater reduction in total CAPS score for drug relative to

[1994] trauma (N=23, 12 randomized, placebo, after adjusting for initial CAPS score and site

male/23 female) placebo-controlled 2. Greater reduction in CAPS total score for civilian sample relative to

2. Combat veterans veteran sample

and civilians with 3. Significant reduction in numbing and hyperarousal symptoms but not

mixed trauma reexperiencing or avoidance

(N=24, 23 male/1 female) 4. Significantly greater reduction in HAM-D score for drug relative to placebo

3. Katz et al. [1994/1995] Combat veterans and Double-blind Brofaromine 14 weeks 1. Significant reduction in CAPS total score for both groups (drug=48%,

civilians with mixed randomized, placebo=29%), with significant, between-groups difference

trauma (N=45, 34 male/ placebo-controlled 2. 55% of drug group and 26% of placebo group no longer met diagnostic

11 female) multi-center criteria for PTSD

3. On CGI, drug group had significantly greater mean improvement and more

participants rated as very much improved

4. Baker et al. [1995] Combat veterans and Double-blind Brofaromine 10 weeks 1. Significant reduction in CAPS total score for both groups (drug=33%,

civilians with mixed randomized, placebo=31%), but no between-groups difference

trauma (N=114, 92 placebo-controlled 2. Comparable results for IES, with somewhat smaller reduction in IES

male/22 female) Multi-center total score in both groups (26%) and no between-groups differences

3. No between-groups difference on DTS or Physician’s Global Evaluation (within-

groups analyses not presented)

5. Hertzberg et al. [1996] Male combat veterans Multiple baseline, Trazodone 4 months 1. Reduction in CAPS total score (15%)

(N=6) open label but 2. Comparable reduction on DTS (15%)

assessment blind 3. Somewhat smaller reduction on BDI (10%), little change on STAI-S (+1%)

4. Four of 6 participants rated as much improved on CGI, 2 rated as minimally

improved

6. Neal et al. [1997] Military personnel 1. Non-blinded, Moclobemide 12 weeks 1. Significant reduction in computerized CAPS total score (50%)

and civilians with uncontrolled 2. Comparable reduction on IES (49%)

mixed trauma (N=20, 2. Computerized 3. Somewhat smaller reduction on MADRS (41%), HAM-A (44%), and

18 male/2 female) CAPS, intensity CIS (39%)

scores only 4. Computerized CAPS change score correlated .76 with IES change score, but only

.31 with MADRS and .32 with HAM-A change scores

7. Bouwer and Stein Male torture victims Routine clinical care, Sertraline (n=9) 8 weeks 1. Significant reduction in CAPS total score (63%)

[1998] (N=14) non-blinded, un- Imipramine 2. Somewhat smaller reduction on MADRS (53%)

controlled (n=2) 3. 12 of 14 participants rated as very much or much improved on CGI

Fluoxetine

(n=2)

Clomipramine

(n=1)

(continued)

TABLE 1. (Continued).

Authors (year) Participants Design Drug Duration Key CAPS-related findings

8. Cañive et al. [1998] Male combat veterans Routine clinical care, Bupropion 6 weeks 1. Trend for reduction in CAPS total score (10%), significant

(N=14) non-blinded, un- reduction (16%) in CAPS hyperarousal score, but not in

controlled reexperiencing (+1%) or avoidance/numbing (9%) scores

2. Ten of 14 participants rated as very much or much improved on CGI

3. Significant reduction on HAM-D (26%) but not HAM-A (12%)

9. Hertzberg et al. [1998] Male combat veterans Non-blinded, Nefazadone 12 weeks 1. Significant reduction in CAPS total score (32%)

(N=10) uncontrolled 2. Significant, somewhat smaller reduction on DTS (28%)

3. Significant reduction on HAM-D (34%) but not BDI (7%)

4. Ten of 10 participants rated as much improved or very much improved CGI

10. Clark et al. [1999] Male combat veterans Open label but Divalproex 8 weeks 1. Significant reduction in CAPS total (18%), reexperiencing 21%), and

(N=13) assessment (except hyperarousal (29%) scores, nonsignificant reduction in avoidance/

CGI) blind, numbing score (7%)

uncontrolled 2. Significant, somewhat larger reduction on HAM-D (31%) and HAM-A (27%)

3. Eleven of 13 participants rated as much improved or very much improved on CGI

*BDI, Beck Depression Inventory; CGI, Clinical Global Impressions; CIS, Clinician Impression of Severity; DTS, Davidson Trauma Scale; HAM-A, Hamilton Rating Scale for Anxiety, HAM-D, Hamilton Rating Scale

for Depression; IES, Impact of Event Scale, MADRS, Montgomery-Asberg Depression Rating Scale; STAI, State-Trait Anxiety Inventory.

TABLE 2. Summary of CAPS findings from psychosocial treatment studies of posttraumatic stress disorder*

Number

of sessions/

Authors (year) Participants Design Intervention duration Key CAPS-related findings

1. Boudewyns et al. Male combat veterans Ramdomized, 1. EMD Two 90-minute 1. No significant reduction in any CAPS symptom or symptom cluster scores

[1993] (N=20) controlled trial, 2. exposure EMD or 2. No significant reduction on Mississippi Scale or IES

assessments not control exposure 3. No significant reduction in psychophysiological responding

blinded 3. routine sessions in

clinical care 2 weeks

(group therapy

without exposure)

2. Busuttil et al. Military personnel, Uncontrolled, Inpatient group 12 days 1. Significant reduction in CAPS total intensity (54%), global improvement

[1995] veterans and civilians assessments not therapy (59%), and global severity (55%) scores

with mixed trauma blinded 2. Significant, somewhat smaller reduction in IES (42%) and PK (48%)

(N=34, 28 male/6 3. Significant, somewhat smaller reduction on SCL-90 (43%), and BDI (39%)

female) 4. 26 or 34 (76%) participants no longer met PTSD diagnostic criteria

3. Thompson et al. Civilians with mixed Uncontrolled, Multicomponent 8 weekly sessions 1. Significant reduction in CAPS total score (35%)

[1995] trauma (N=23, 17 assessments not cognitive-behavioral 2. Significant, somewhat larger reduction on IES (42%)

male/6 female) blinded protocol (imaginal 3. Comparable reduction on SCL-90 (38%), larger reduction on GHQ (61%)

and in vivo exposure,

cognitive

restructuring)

4. Boudewyns and Male combat veterans Randomized, 1. EMDR 5-7 EMDR or 1. Significant reduction in CAPS total score for all three groups (EMDR=33%,

Hyer [1996] (N=61) controlled trial, 2. exposure exposure sessions exposure=21%, routine care=17%), but no significant between-groups

assessments control in 6 weeks differences

blinded 3. routine 2. No significant reduction on IES

clinical care (group 3. Significant between-groups differences on POMS anxiety scale and heart rate

therapy without reactivity, with EMDR and exposure group showing reduction in scores and

exposure) no-exposure control group showing slight increase

5. Carlson et al. Male combat veterans Single-subject EMDR 12 sessions, 2 1. At 3-month followup, reduction in CAPS total score across 4 participants

[1996] (N=4) replication sessions per week ranged from 34–100%, with 3 of 4 showing > 80% improvement

series 2. Comparable reduction on IES (34–88%), but smaller reduction on Mississippi

Scale (6–46%)

3. More variable outcome on BDI and STAI-S and STAI-T (1 participant

showing slight increase on these scales, other 3 showing reduction of 50–

100% reduction on BDI and 8–41% on STAI)

6. Frueh et al. [1996] Male combat veterans Uncontrolled, Multicomponent 29 sessions in 1. Trend for reduction in CAPS total score (21%)

(N=11) assessments not cognitive- 17 weeks 2. Significant reduction on HAM-A (31%), CGI (34%) and heart rate reactivity

blinded behavioral protocol (14%)

(education, imaginal 3. No significant reduction on BDI, SPAI, or STAXI

and in vivo exposure,

social skills training,

anger management)

7. Hall and Female sexual 1. Case study Cognitive processing 17 weekly sessions 1. Reduction in CAPS-2 total scores (48%)

Henderson [1996] abuse victim 2. CAPS-2 therapy 2. Smaller reduction on MPSS-SR (31%)

(N=1) 3. Somewhat smaller reduction on SCL-90 (26%), and slight increase on BDI

(+18%)

(continued)

TABLE 2. (Continued).

Number

of sessions/

Authors (year) Participants Design Intervention duration Key CAPS-related findings

8. Pitman et al. Male combat veterans Crossover, EMDR, with and 12 weekly sessions 1. Little change in CAPS total score, with slight increase after eye movement

[1996] (N=17) assesssments without eye move- (6 in each condition and slight decrease after no eye movement condition

blinded ment condition) 2. Comparable result for Mississippi Scale, with slight increase after both

conditions, and mixed results for IES, with significant reductions for intrusion

or avoidance subscale depending on condition and trauma memory evaluated

3. Significant reduction on SCL-90 in eye movement condition

4. Therapy integrity ratings significantly correlated with CAPS change score in

both conditions (.55, .62), but with SCL-90 in eye movement condition only

(.69)

9. Thrasher et al. Male physical assault Single-subject Cognitive 10 sessions 1. Substantial reduction in CAPS total score for both participants (67–90%)

[1996] victims (N=2) replication series restructuring 2. Comparable reduction in IES (76–91%) and PSS (79–80%)

3. Comparable reduction on BDI (65–92%)

10. Fontana and Male combat veterans Quasi-experimental 1. Long-stay Variable 1. Significant reduction in CAPS total score for all three programs (long-stay=

Rosenheck [1997] (N=785) program evaluation PTSD program (approximately 13%, short-stay=19%, psychiatric=16%)

2. Short-stay 1–3 months) 2. Significant between-groups effect, with veterans in short-stay PTSD

PTSD program programs and general psychiatric inpatient units showing greater

3. General improvement

psychiatric unit 3. No significant reduction on Mississippi Scale (long-stay=0%, short-stay=3%,

psychiatric=3%)

4. Significant, larger reduction on ASI psychiatric score (long-stay=24%, short-

stay=26%, psychiatric=26%) and significant, smaller reduction on BSI (long-

stay=2%, short-stay=11%, psychiatric=12%), both with significant between-

groups effects similar to those for the CAPS

11. Hicling and Motor vehicle Uncontrolled Multi- 10 weekly sessions 1. Significant reduction in CAPS total score (68%)

Blanchard [1997] accident victims trial, nonblinded component 2. Comparable reduction on IES (66%), significant but smaller reduction on

(N=10, 1 male/ assessments cognitive- PCL (39%)

9 female) behavioral 3. Significant, somewhat smaller reduction on BDI (50%) and significant,

protocol smaller reduction on STAI-S (19%) and STAI-T (20%)

(education, 4. Five of 8 participants with full PTSD and 1 of 2 with subsyndromal

relaxation, PTSD at pre-test no longer met diagnosis at post-test; 3 of 8 with full

exposure, PTSD at pre-test were subsyndromal at post-test

cognitive

restructuring)

12. Carlson et al. Male combat veterans Randomized, 1. EMDR 12 sessions in 1. Significant Group x Time interaction at 3-month followup, with EMDR

[1998] (N=35) controlled trial, 2. Biofeedback- 6 weeks group showing significantly greater reduction on CAPS total score (69%)

non-blinded assisted relaxa- compared to relaxation group (20%0

assessments tion 2. Similar pattern with smaller reduction on IES (EMDR=45%, relaxation=14%)

except at 9- 3. Routine 3. Similar pattern with smaller reduction on BDI (EMDR=57%, relaxation=

month follow- clinical care 22%), and substantially smaller reduction in both groups on STAI-S

up (EMDR=14%), relaxation=18%) and STAI-T (EMDR=22%, relaxation=11%)

4. Of participants completing first follow-up 7 or 9 (78%) in EMDR group

versus 2 of 9 (22%) in relaxation group no longer met PTSD diagnostic

criteria

(continued)

TABLE 2. (Continued).

Number

of sessions/

Authors (year) Participants Design Intervention duration Key CAPS-related findings

13. Conlon et al. Motor vehicle Randomized, 1. Debriefing Single 30- 1. Significant reduction in CAPS total score for total sample (53%; debriefing=

[1998] accident victims, 1 controlled trial, 2. Monitoring minute 70%, monitoring=36%), but no significant between-groups difference at

week post-accident non-blinded (assessment- debriefing follow-up

(N=40, 19 male/21 assessments only control) session 2. Interpretation of CAPS change scored is somewhat ambiguous because

female) CAPS-2 used at baseline and CAPS-1 used at followup

3. Comparable reduction on IES (total sample=50%; debriefing=55%,

monitoring=44%)

14. Lazrove et al. Civilians with mixed Uncontrolled EMDR 3 weekly sessions 1. Substantial reduction in CAPS total score (70%)

[1998] trauma (N=8, 2 male/ trial, assessments 2. Larger reduction on IES-R (87–.96% for intrusion, avoidance, hyperarousal

6 female) conducted by non- subscales)

treating research 3. Comparable reduction on BDI (68%), smaller reduction on SCL-90 (42%)

assistant 4. All of the participants who completed treatment no longer met diagnostic

criteria for PTSD

15. Lubin et al. Female victims of Uncontrolled Trauma-focused, 16 weekly sessions 1. Significant reduction in CAPS total score (39%)

[1998] mixed civilian trauma trial, assessments cognitive behavioral 2. Significant, smaller reduction on Civilian Mississippi Scale (9%) and IES

conducted by non- group therapy (16%)

treating research 3. Significant, somewhat smaller reduction on BDI (33%) and smaller reduction

assistants on DES (21%) and SCL-90 (23%)

16. Marks et al. Civilians with mixed 1. Randomized, 1. Imaginal and 10 sessions in 1. Significant reduction in CAPS-2 total score, with effect sizes ranging from

[1998] trauma (N=87, 56 male/ controlled trial, in vivo exposure an average of 1.30 to 2.00 for three active intervention groups and .60 for relaxation group

31 female) blinded assessments 2. Cognitive 16 weeks 2. Significant between-groups effect for CAPS-2 total score, with greater

2. CAPS-2 restructuring reduction for three active intervention groups pooled versus relaxation group

3. Exposure plus 3. Similar within-groups and between-groups results for IES (within-groups

cognitive restruc- effect sizes from 1.30 to 1.50 for active intervention groups, .08 for relaxation

turing group)

4. Relaxation 4. Similar within-groups and between-groups results for BDI (within-groups

(placebo control) effects sizes from 1.20 to 1.70 for active intervention groups, .07 for

relaxation group)

5. With improvement defined as > 2 SDs, 47–53% of participants in active

intervention groups showed improvement on CAPS-2 total score, versus 15%

in relaxation groups. Somewhat higher rates found in IES (50–60% for active

intervention groups, 20% for relaxation group)

6. 63–75% of participants in active intervention groups versus 55% in relaxation

group no longer met diagnostic criteria for PTSD

17. Pantalon and Male combat veterans 1. Crossover, 1. Implosive therapy 12 weekly sessions 1. Reduction in CAPS-2 score (reexperiencing and avoidance only; hyperarousal

Motta [1998] (N=6) single-subject (imaginal exposure) scores not reported) ranged from 46–88% (M=71%) across the six participants

replication series, 2. Anxiety management 2. Lower but substantial reduction on PCL (8–100%, M=50% across the six

non-blinded training participants; reexperiencing and avoidance only; hyperarousal scores not

assessments reported)

2. CAPS-2

(continued)

TABLE 2. (Continued).

Number

of sessions/

Authors (year) Participants Design Intervention duration Key CAPS-related findings

18. Rothbaum et al. Male combat veterans Case study Virtual reality 14 sessions, 2 1. Reduction in CAPS total score (34%)

[1999] (N=1) exposure session per week 2. Larger reduction on IES (45%)

3. Smaller on BDI (19%) and STAXI-T (21%), substantially larger reduction on

STAXI-S (63%)

19. Tarrier et al. Civilians with mixed Randomized, 1. Imaginal Average of 10-12 1. Significant within-groups reduction in CAPS total score for both groups

[1999] trauma (N=62, 36 male/ controlled trial, exposure sessions over 6 (32% for exposure group, 35% for cognitive therapy group), but no between

26 female) assessments 2. Cognitive months groups difference

blinded therapy 2. Comparable within-groups reduction on IES (31–33% for intrusion, 25–34% for

avoidance), somewhat smaller reduction on Penn Inventory (22–27%); no

between-groups difference on either

3. Somewhat smaller within-groups reduction on BDI (27–31%) and BAI (23–

25%); no between-groups difference

4. Comparable to somewhat larger within-groups reduction on GHQ (30–46%),

but no between-groups difference

5. 59% of exposure group versus 42% of cognitive therapy group no longer met

diagnostic criteria for PTSD

*ASI, Addiction Severity Index; BDI, Beck Depression Inventory; BSI, Brief Symptom Inventory; DES, Dissociative Experiences Scale; EMDR, Eye Movement Desensitization and Reprocessing; CHQ, General Health

Questionnaire; HAM-A, Hamilton Rating Scale for Anxiety; IES, Impact of Event Scale; MPSS-SR, Modified PTSD Symptom Scale — Self-Report; PCL, PTSD Checklist; PK, Keane MMPI PTSD scale; POMS,

Profile of Mood States, PSS, PTSD Symptom Scale; SCL-90, Symptom Checklist-90; SPAI, Social Phobia and Anxiety Inventory; STAI, State-Trait Anxiety Inventory; STAXI, State-Trait Anger Expression Inventory.

152 Weathers et al.

Studies 1, 8, and 10) found greater reduction on the

HAM-A relative to the CAPS, with differences rang-

ing from 2–9 percentage points. The fourth study

found a 5 percentage point greater reduction on the

CAPS. One pharmacological study (Study 5) found a

greater reduction on the CAPS relative to the STAI-S.

Three psychosocial studies (Table 2, Studies 5, 11, and

12) included the STAI–S, two included the STAI-T

(Studies 11 and 12), and one (Study 19) included the

BAI. In each case the CAPS showed greater reduction,

ranging from 2–59 percentage points.

CAPS versus global measures of distress and impair-

ment. In the pharmacological studies, reduction in CAPS

scores were accompanied by global measures of func-

tioning, including the CGI in the six studies that em-

ployed it (Table 1, Studies 3, 5, 7, and 8–10), the CIS

(Study 6), and a relatively stringent consensus definition

of treatment response (Study 1). Five psychosocial stud-

ies (Table 2, Studies 2, 3, 7, 13, and 15) included the

SCL-90 and one (Study 10) included the BSI. In each

case the CAPS showed greater reduction, ranging from

3–28 percentage points. In contrast, two studies (Studies

3 and 19) found greater reduction on the GHQ relative

to the CAPS, and one study (Study 10) found greater

reduction on the ASI psychiatric score.

Discussion. The 29 treatment outcome studies re-

viewed in this section provide ample evidence of the

sensitivity of the CAPS to clinical change. We summa-

rize the results by returning to the four hypothesized

results discussed at the outset of this section. First,

there was clear and consistent evidence of within-

groups effects in both the pharmacological and the psy-

chosocial treatment studies. Stronger within-groups

effects were found in the psychosocial studies. This

could be due to the fact that with one exception the

drugs used in the studies reviewed were all antidepres-

sants, and their efficacy for treating PTSD has not been

clearly established. The symptom relief they bring

about may be due more to their antidepressant effects

rather than to specific effects on PTSD symptoms such

as re-experiencing and effortful avoidance. In contrast,

all of the psychosocial interventions involved some type

of trauma-specific component, and most included some

form of direct therapeutic exposure or cognitive pro-

cessing, which have been shown to have specific effects

on PTSD symptoms. This finding could also be due to

the fact that in general the psychosocial interventions

involved considerably more patient-therapist contact

than did the pharmacological trials.

Second, there was some evidence of between-groups

effects, although relatively few studies included a com-

parison condition. Two of the three pharmacological

trials with a placebo control found greater reduction on

the CAPS in participants who received the drug. Re-

sults were more inconsistent for the psychosocial trials.

Only two of the six randomized trials, plus one quasi-

experimental program evaluation, found a significant

between-groups effect. However, two of the non-sig-

nificant trials employed quite limited interventions, and

a third trial compared two active interventions, expo-

sure and cognitive restructuring. Clearly, more ran-

domized, placebo-controlled trials are needed before

this issue can be resolved.

Third, reduction in CAPS scores was mirrored by

reduction in self-report measures of PTSD, particu-

larly the IES. The CAPS showed a slightly greater

reduction than the IES in 2 of 3 pharmacological

studies and 7of 11 psychosocial studies, although

the margins, especially in the pharmacological tri-

als, were generally small. Fourth, there was some

evidence of greater reduction on the CAPS than on

measures of depression, anxiety, and global distress,

particularly on self-report measures such as the

BDI, STAI, and SCL-90.

Finally, in commenting about the populations

studied, although the CAPS was developed in a male

combat veteran population, and many of the early

studies focused exclusively on this population, the

CAPS has now been extended to increasingly diverse

samples that include females and victims of various

types of civilian trauma. Of the studies reviewed in

this section, 11 of 29 included at least some females

and 15 of 29 included at least some participants with

civilian trauma.

VALIDITY EVIDENCE FROM

CASE-CONTROL DESIGNS

In this section, we consider validity evidence from

studies in which participants were designated as

PTSD-positive (“cases”) or PTSD-negative (“con-

trols”) based on the CAPS and then compare this evi-

dence on some biological or psychological measure or

experimental task. Such case-control studies were too

numerous and diverse to summarize briefly. Instead,

we describe several representative examples from dif-

ferent research domains to illustrate that groups

formed on the basis of a CAPS diagnosis differ in con-

ceptually meaningful ways on a variety of characteris-

tics or behaviors.

The first example involves the psychophysiology

of PTSD. Physiological reactivity to reminders of

the trauma is a core symptom of PTSD, and a

growing number of studies have found that indi-

viduals with PTSD show greater reactivity than

those without PTSD in laboratory-based physi-

ological assessments. Much of the early work was

conducted with male combat veterans, but more re-

cent studies have examined male and female victims

of civilian trauma. Blanchard et al. [1996a] used the

CAPS to classify 105 male and female motor vehicle

accident victims as PTSD, subsyndromal PTSD,

and non-PTSD. They also included a control group

of 54 participants who had not experienced an acci-

dent. They found that compared to participants

without PTSD, those with PTSD showed a signifi-

cantly greater increase in heart rate in response to

brief audiotapes depicting each participant’s unique

traumatic experience. They also found that an in-

Review Article: Clinician-Administered PTSD Scale 153

crease of two beats per minute had reasonable diag-

nostic utility, yielding 69% sensitivity and 78%

specificity among accident victims.

The second example comes from a more recent line

of research on auditory event-related potentials (ERP)

in PTSD. Several different investigators have docu-

mented abnormal ERPs in individuals with PTSD and

have suggested that such characteristic responses may

be associated with the attention and concentration dif-

ficulties often seen in PTSD. Metzger et al. [1997]

used the CAPS to classify male Vietnam combat veter-

ans as PTSD or non-PTSD groups and then further

divided the PTSD participants into medicated and

unmedicated groups. Administering a three-tone audi-

tory “oddball” task, they found significantly smaller

P3 amplitudes in the unmedicated PTSD group, rela-

tive to the medicated PTSD group and the non-

PTSD controls.

The third example involves research on the associa-

tion of chronic PTSD and physical health problems.

Beckham et al. [1998] used the CAPS to classify 276

male Vietnam combat veterans as PTSD or non-

PTSD, then assessed participants’ current health status

and reviewed their medical records. Health measures

included health complaints, current and lifetime physi-

cal conditions, number of physician-rated medical cat-

egories, and total number of physician-rated illnesses.

After controlling for a variety of potentially confound-

ing third variables, including age, socioeconomic status,

ethnicity, combat exposure, alcohol problems, and

smoking history, they found that veterans with PTSD

had significantly more health problems across all indi-

cators compared to veterans without PTSD.

The last example represents an effort to identify

potential risk factors for PTSD. Yehuda et al. [1995]

used the CAPS to classify a community sample of

72 Nazi concentration camp survivors as PTSD or

non-PTSD. They also included a comparison group

of 19 demographically matched participants who

had not experienced the Holocaust. The purpose of

the study was to examine the relationships among

lifetime trauma history, recent stressful life events,

and severity of current PTSD symptoms. As ex-

pected, Yehuda et al. [1995] found that Holocaust

survivors with PTSD had greater lifetime trauma

exposure and more recent stressful life events than

did survivors without PTSD or comparison partici-

pants. By using the CAPS as a continuous measure

of PTSD symptom severity, they found that lifetime

trauma was significantly associated with avoidance

and hyperarousal, but not with re-experiencing,

within a combined sample of all Holocaust survi-

vors. In a similar analysis, they found that recent

stressful life events were significantly associated

with all three CAPS symptom clusters.

These examples, and the other case-control studies

that we did not discuss, provide additional evidence that

the CAPS is a valid measure of PTSD diagnostic status

and symptom severity. They demonstrate that when the

CAPS is used to classify trauma-exposed individuals as

PTSD or non-PTSD, the resulting groups differ sig-

nificantly in a theoretically consistent way on key de-

pendent variables.

GENERAL DISCUSSION AND

RECOMMENDATIONS

In the 10 years since it was developed, the CAPS

has proven to be a psychometrically sound, practical,

and flexible structured interview that is well-suited for

a wide range of clinical and research applications in

the field of traumatic stress. Moreover, it has been

successfully used with many different traumatized

populations. It has excellent reliability, yielding con-

sistent scores across items, raters, and testing occa-

sions. There is also considerable validity evidence that

supports the use of the CAPS as a measure of PTSD

diagnostic status and symptom severity. Evidence of

content validity derives first from its direct correspon-

dence with the DSM-IV diagnostic criteria for PTSD

and second from the fact that it was developed by ex-

perts in the field of traumatic stress and revised based

on feedback from many clinicians and investigators

who used it in real-world settings. Evidence from a

growing number of psychometric investigations indi-

cates it has strong convergent and discriminant valid-

ity, strong diagnostic utility, and is sensitive to clinical

change. In addition, factor analyses, especially confir-

matory factory analyses, have shown that the factor

structure of the CAPS corresponds well to current

conceptualizations of PTSD. Finally, when the CAPS

is used in case-control designs, individuals designated

as PTSD differ from those without PTSD in predict-

able, theoretically meaningful ways. Clearly more re-

search on the CAPS is needed, but at this point the

CAPS is the most extensively investigated structured

interview for PTSD.

Criticism of the CAPS tends to focus on three con-

cerns. The first concern is that the CAPS is cumber-

some and lengthy. In response, the CAPS clearly is

longer on paper than other PTSD interviews, but it

does not necessarily take longer to administer. Most of

the CAPS questions are optional probes, only some of

which would likely be administered during a given in-

terview. A standard administration of the CAPS in-

volves asking the initial probe under frequency for

each item. With an articulate, motivated respondent

this single questions may elicit all the information

necessary to rate both the frequency and intensity of a

given symptom. All other probes are to be used only

if: a) a response is incomplete, vague, confusing, or in

some way insufficient to make a rating, and therefore

needs to be clarified, or b) the respondent does not

understand what is being asked.

In our experience, even with ideal respondents,

some degree of clarification is inevitable. To enhance

uniformity of administration, we have included a

154 Weathers et al.

number of follow-up probes that address the most

common points of clarification. This reduces variabil-

ity due to idiosyncratic questioning across different in-

terviewers and provides a helpful structure for less

experienced interviewers. Furthermore, the CAPS was

designed as a comprehensive yet flexible instrument

that would meet the demand of almost any PTSD as-

sessment task, including diagnosis, evaluating symp-

tom severity, and conducting a functional analysis of

symptoms for case conceptualization and treatment

planning. Therefore, in some assessment contexts it

may opted not to assess Criterion A, elicit descriptive

examples of symptoms, administer the global ratings

or the guilt and dissociation items, or rate lifetime

PTSD.

The second concern, closely related to the first, is

that the CAPS is too complicated and difficult to

learn. In response, our own experience, based on doz-

ens of training sessions, is that after a 2-hour orienta-

tion trainees naïve to the CAPS can make highly

reliable ratings of a role-played interview. With some

self-study and a few practice interviews, they can

achieve a uniform, clinically sensitive administration.

CAPS trainees, including those with little or no expe-

rience with structured interviews or assessing PTSD,

typically find that the CAPS is very straightforward to

learn. In fact, less experienced interviewers tend to

have the most favorable responses because they appre-

ciate the structure the CAPS provides.

The third concern centers on the question of whether

frequency and intensity ratings overlap to such an extent

as to be essentially redundant. Clearly, they appear to be

strongly correlated at the syndrome level and even at the

symptom cluster level. At the item level, however, the

correlations between frequency and intensity are moder-

ate, suggesting that they measure correlated but distinct

dimensions. We have several responses to this concern.

First, the separate assessment of frequency and intensity

explicitly defines what is meant by symptom severity,

thereby reducing variability in clinical judgment, espe-

cially among less experienced interviewers. Second, this

is a meaningful, theoretical distinction, employed suc-

cessfully for example in the substance abuse literature,

where typologies of drinkers are based on how often a

person drinks, as well as how much they consume at any

given setting. Third, adding frequency and intensity to-

gether yields a nine-point scale (0–8) that allows finer

gradations of severity. This increases variance attribut-

able to individual differences, thereby avoiding a restric-

tion of range that could lower estimates of reliability and

validity. Fourth, it allows the assessment of the differen-

tial impact of treatment on the frequency versus the in-

tensity of symptoms.

Last, we close with some recommendations for the

use of the CAPS in clinical research and the presenta-

tion of CAPS data in empirical reports. First, for

newly initiated research, investigators should use what

is now the sole version of the CAPS, the combined

DSM-IV version, and explicitly identify it as such. For

research already underway or completed, investigators

should explicitly identify the version used, either the

CAPS-1 or CAPS-2 (DSM-III-R versions) or the

CAPS-DX or CAPS-SX (DSM-IV versions). Also, if

the CAPS is used as a diagnostic measure, investiga-

tors should specify the scoring rule used to obtain a

diagnosis. Second, investigators should briefly specify

the experience and training of CAPS interviewers,

both in terms of their general background in psycho-

pathology and structured interviewing, and in terms of

their specific experience with the CAPS. Also, when-

ever possible they should attempt to collect and report

reliability data on the interviewers and participants in-

volved. Even something as modest as inter-rater reli-

ability on a small number of audiotaped interviews is

helpful for documenting the quality of the CAPS data.

Third, investigators should take greater advantage

of the flexibility of the CAPS in analyzing their data.

Some examples include a) using multiple CAPS scor-

ing rules and comparing the results for lenient, mod-

erate, and stringent rules; b) using the CAPS as both a

dichotomous and a continuous measure, reporting not

only diagnostic status but symptom severity scores,

which would be valuable for comparing findings

across studies; c) breaking out CAPS symptom sever-

ity scores into the three DSM-IV symptom clusters

and examining the results by cluster; d) examining the

symptom clusters further by separating Cluster C into

effortful avoidance (C1 and C2) and emotional numb-

ing; and e) dividing scores even further into frequency,

intensity, and severity scores for each of the symptom

clusters. Finally, although considerable progress has

been made in the development and evaluation of

PTSD assessment measures, including the CAPS, reli-

ance on a single instrument should be avoided. We ad-

vocate multimodal assessment of PTSD, an approach

that relies on converging evidence from multiple

sources, and we encourage investigators to include

multiple measures of PTSD and comorbid disorders

whenever possible.

Acknowledgments. The authors wish to thank

Alethea Smith for her assistance with the literature

search and manuscript preparation.

REFERENCES

American Psychological Association. 1999. Standards for Educa-

tional and Psychological Testing. Washington DC: Author.

Baker DG, Diamond BI, Gillette GM, Hamner MB, Katzelnick D,

Keller TW, Mellman TA, Pontius EB, Rosenthal M, Tucker P,

Van der Kolk BA, Katz RJ. 1995. A double-blind, randomized,

placebo-controlled, multi-center study of brofaromine in the

treatment of post-traumatic stress disorder. Psychopharmacol-

ogy 122:386–389.

Beck AT, Epstein N, Brown G, Sterr RA. 1988. An inventory for

measuring clinical anxiety: psychometric properties. J Consult

Clin Psychol 56:893–897.

Beckham JC, Moore SD, Feldman ME, Hertzberg MA, Kirby AC,

Fairbank JA. 1998. Health status, somatization, and severity of

posttraumatic stress disorder in Vietnam combat veterans with

posttraumatic stress disorder. Am J Psychiatry 155:1565–1569.

Review Article: Clinician-Administered PTSD Scale 155

Blake DD, Weathers FW, Nagy LM, Kaloupek DG, Klauminzer

G, Charney DS, Keane TM. 1990. A clinician rating scale for

assessing current and lifetime PTSD: the CAPS-1. Behav Ther

13:187–188.

Blake DD, Weathers FW, Nagy LM, Kaloupek DG, Gusman FD,

Charney DS, Keane TM. 1995. The development of a Clinician-

Administered PTSD Scale. J Trauma Stress 8:75–90.

Blanchard EB, Hickling EJ, Taylor AE, Forneris CA, Loos WR,

Jaccard J. 1995a. Effects of varying scoring rules of the Clini-

cian-Administered PTSD Scale (CAPS) for the diagnosis of

post-traumatic stress disorder in motor vehicle accident victims.

Behav Res Ther 33:471–475.

Blanchard EB, Hickling EJ, Taylor AE, Loos WR. 1995b. Psychiat-

ric morbidity associated with motor vehicle accidents. J Nerv

Ment Dis 183:495–504.

Blanchard EB, Hickling EJ, Buckley TC, Taylor AE, Vollmer A,

Loos WR. 1996a. Psychophysiology of posttraumatic stress dis-

order related to motor vehicle accidents: Replication and exten-

sion. J Consult Clin Psychol 64:742–751.

Blanchard EB, Jones-Alexander J, Buckley TC, Forneris CA.

1996b. Psychometric properties of the PTSD Checklist (PCL).

Behav Res Ther 34:669–673.

Boudewyns PA, Hyer, Leon A (Lee). 1996. Eye movement desensi-

tization and reprocessing (EMDR) as treatment for post-trau-

matic stress disorder (PTSD). Clin Psychol Psychother 3:

185–195.

Boudewyns PA, Stwertka SA, Hyer, Leon A (Lee), Albrecht JW,

Sperr EV. 1993. Eye movement desensitization for PTSD of

combat: a treatment outcome pilot study. Behav Ther 16:29–33.

Bouwer C, Stein DJ. 1998. Survivors of torture presenting at an

anxiety disorders clinic: symptomatology and pharmacotherapy. J

Nerv Ment Dis 186:316–318.

Buckley TC, Blanchard EB, Hickling EJ. 1998. A confirmatory fac-

tor analysis of posttraumatic stress symptoms. Behav Res Ther

36:1091–1099.

Busuttil W, Turnbull GJ, Neal LA, Rollins JW, West AG, Blanch

N, Herepath R. 1995. Incorporating psychological debriefing

techniques within a brief group psychotherapy programme for

the treatment of post-traumatic stress disorder. Br J Psychiatry

167:495–502.

Cañive JM, Clark RD, Calais LA, Qualls CR, Tuason VB. 1998.

Bupropion treatment in veterans with posttraumatic stress disor-

der: an open study. J Clin Psychopharmacol 18:379–383.

Carlson JG, Chemtob CM, Rusnak K, Hedlund NL. 1996. Eye

movement desensitization and reprocessing treatment for com-

bat PTSD. Psychotherapy 33:104–113.

Carlson JG, Chemtob CM, Rusnak K, Hedlund NL, Muraoka MY.

1998. Eye movement desensitization and reprocessing (EMDR)

treatment for combat-related posttraumatic stress disorder. J

Trauma Stress 11:3–24.

Clark RD, Cañive JM, Calais LA, Qualls CR, Tuason VB. 1999.

Divalproex in posttraumatic stress disorder: an open-label clini-

cal trial. J Trauma Stress 12:395–401.

Conlon L, Fahy TJ, Conroy RM. 1999. PTSD in ambulant RTA

victims: a randomized controlled trial of debriefing. J Psychosom

Res 46:37–44.

Davidson JRT, Book SW, Colket JT, Tupler LA, Roth SH, David

D, Hertzberg MA, Mellman TA, Beckham JC, Smith RD,

Davison RM, Katz RJ, Feldman ME. 1997. Assessment of a new

self-rating scale for posttraumatic stress disorder. Psychol Med

27:153–160.

Derogatis LR. 1983. SCL-90-R administration, scoring, and proce-

dures manual-II for the revised version. Towson, MD: Clinical

Psychometric Research.

Fleming MP, Difede J. 1999. Effects of varying scoring rules of the

Clinician Administered PTSD Scale (CAPS) for the diagnosis of

PTSD after acute burn injury. J Trauma Stress 12:535–542.

Fontana A, Rosenheck RA. 1997. Effectiveness and cost of the in-

patient treatment of posttraumatic stress disorder: comparison of

three models of treatment. Am J Psychiatry 154:758–765.

Frueh BC, Turner SM, Beidel DC, Mirabella RF, Jones WJ. 1996.

Trauma Management Therapy: a preliminary evaluation of a

multicomponent behavioral treatment for chronic combat-re-

lated PTSD. Behav Res Ther 34:533–543.

Hall CA, Henderson CM. 1996. Cognitive processing therapy for

chronic PTSD from childhood sexual abuse: a case study. Coun-

sel Psychol Quart 9:359–371.

Hertzberg MA, Feldman ME, Beckham JC, Davidson JRT. 1996.

Trial of trazodone for posttraumatic stress disorder using a mul-

tiple baseline group design. J Clin Psychopharmacol 16:294–298.

Hamilton M. 1960. A rating scale for depression. J Neurol Neuro-

surg Psychiatry 23:56–62.

Hamilton M. 1969. Diagnosis and ratings of anxiety. Br J Psychia-

try 3:76–79.

Hertzberg MA, Feldman ME, Beckham JC, Moore SD, Davidson

JRT. 1998. Open trial of nefazodone for combat-related post-

traumatic stress disorder. J Clin Psychiatry 59:460–464.

Hickling EJ, Blanchard EB. 1997. The private practice psychologist

and manual-based treatments: post-traumatic stress disorder sec-

ondary to motor vehicle accidents. Behav Res Ther 35:191–203.

Hovens JEJM, Van der Ploeg HM, Bramsen I, Klaarenbeek MTA,

Schreuder BJN, Rivero VV. 1994. The development of the Self-

Rating Inventory for Posttraumatic Stress Disorder. Acta Psychiatr

Scand 90:172–183.

Hyer, Leon A (Lee), Summers MN, Boyd S, Litaker M, Boudewyns

PA. 1996. Assessment of older combat veterans with the Clini-

cian-Administered PTSD Scale. J Trauma Stress 9:587–593.

Katz RJ, Lott MH, Arbus P, Crocq L, Herlobsen P, Lingjaerde O,

Lopez G, Loughrey, Gerry C (Gerard), MacFarlane DJ, McIvor

R, Mehlum L, Nugent D, Turner SW, Weisath L, Yule W.

1994–1995. Pharmacotherapy of post-traumatic stress disorder

with a novel psychotropic. Anxiety 1:169–174.

Keane TM, Caddell JM, Taylor KL. 1988. Mississippi Scale for

Combat-Related Posttraumatic Stress Disorder: three studies in

reliability and validity. J Consult Clin Psychol 56:85–90.

Keane TM, Fairbank JA, Caddell JM, Zimering RT, Taylor KL,

Mora CA. 1989. Clinical evaluation of a measure to assess com-

bat exposure. Psychol Assess 1:53–55.

Keane TM, Malloy PF, Fairbank JA. 1984. Empirical development

of an MMPI subscale for the assessment of combat-related post-

traumatic stress disorder. J Consult Clin Psychol 52:888–891.

King DW, Leskin GA, King LA, Weathers FW. 1998. Confirma-

tory factor analysis of the Clinician-Administered PTSD Scale:

evidence for the dimensionality of posttraumatic stress disorder.

Psychol Assess 10:90–96.

Kulka RA, Schlenger WE, Fairbank JA, Hough RL, Jordan BK,

Marmar CR, Weiss DS. 1990. The National Vietnam Veterans

Readustment Study: tables of findings and technical appendices.

New York: Brunner/Mazel.

Lazrove S, Triffleman EG, Kite L, McGlashan TH, Rounsaville B.

1998. An open trial of EMDR as treatment for chronic PTSD.

Am J Orthopsychiatry 68:601–608.

Lubin H, Loris M, Burt J, Johnson DR. 1998. Efficacy of psycho-

educational group therapy in reducing symptoms of posttrau-

matic stress disorder among multiply traumatized women. Am J

Psychiatry 155:1172–1177.

Marks IM, Lovell K, Noshirvani H, Livanou M, Thrasher S. 1998.

Treatment of posttraumatic stress disorder by exposure and/or

cognitive restructuring: a controlled study. Arch Gen Psychiatry

55:317–325.

156 Weathers et al.

Marmar, CR, Weiss DS, Metzler TJ. 1997. The Peritraumatic Dis-

sociative Experiences Questionnaire. In: Wilson JP, Keane TM,

editors. Assessing psychological trauma and PTSD. New York:

Guilford Press. p 412–428.

Metzger LJ, Orr SP, Lasko NB, Pitman RK. 1997. Auditory event-

related potential to tone stimuli in combat-related posttraumatic

stress disorder. Biol Psychiatry 42:1006–1015.

Nagy LM, Morgan CA, Southwick SM, Charney DS. 1993. Open

prospective trial of fluoxetine for posttraumatic stress disorder. J

Clin Psychopharmacol 13:107–113.

Nagy LM, Blake DD, Schnurr P, Southwick SM, Charney D,

Weathers F, Horner B. 1999. The Clinician-Administered

PTSD Scale – Weekly Version (CAPS-2): Reliability and valid-

ity. Manuscript submitted.

Neal LA, Busuttil W, Herepath R, Strike PW. 1994. Development

and validation of the computerized Clinician Administered Post-

Traumatic Stress Disorder Scale-1-Revised. Psychol Med 24:

701–706.

Neal LA, Hill N, Hughes JC, Middleton A, Busuttil W. 1995. Con-

vergent validity of measures of PTSD in an elderly population of

former prisoners of war. Int J Geriatric Psychiatry 10:617–622.

Neal LA, Shapland W, Fox C. 1997. An open trial of moclobemide

in the treatment of post-traumatic stress disorder. Int Clin

Psychopharmacol 12:231–237.

Orr SP. 1997. Psychophysiologic reactivity to trauma-related imag-

ery in PTSD: diagnostic and theoretical implications of recent

findings. Ann NY Acad Sci 821:114–124.

Pantalon MV, Motta RW. 1998. Effectiveness of anxiety manage-

ment training in the treatment of posttraumatic stress disorder: a

preliminary report. J Behav Ther Exp Psychiatry 29:21–29.

Pitman RK, Orr SP, Altman B, Longpre RE, Poiré RE, Macklin

ML. 1996. Emotional processing during eye movement desensiti-

zation and reprocessing therapy of Vietnam veterans with chronic

posttraumatic stress disorder. Compr Psychiatry 37:419–429.

Radnitz CL, Schlein IS, Walczak S, Broderick CP, Binks TM,

Tirch DD, Willard J, Perez-Strumolo L, Festa J, Lillian LB,

Bockian N, Cytryn A, Green L. 1995. The prevalence of post-

traumatic stress disorder in veterans with spinal cord injury. SCI

Psychosocial Process 8:145–149.

Rothbaum BO, Hodges L, Alarcón RD, Ready DJ, Shahar F,

Graap K, Pair J, Hebert P, Gotz D, Wills B, Baltzell D. 1999.

Virtual reality exposure therapy for PTSD Vietnam veterans: a

case study. J Trauma Stress 12:263–271.

Shalev AY, Freedman SA, Peri T, Brandes D, Sahar T. 1997. Pre-

dicting PTSD in trauma survivors: prospective evaluation of self-

report and clinician-administered instruments. Br J Psychiatry

170:558–564.

Spielberger CD, Gorsuch RL, Lushene RE. 1970. STAI manual for

the State-Trait Anxiety Inventory. Palo Alto, CA: Consulting

Psychologists Press.

Standards for educational and psychological testing. 1999. Wash-

ington DC: American Educational Research Association. 194 p.

Tarrier N, Pilgrim H, Sommerfield C, Faragher B, Reynolds M,

Graham E, Barrowclough C. 1999. A randomized trial of cogni-

tive therapy and imaginal exposure in the treatment of chronic

posttraumatic stress disorder. J Consult Clin Psychol 67:13–18.

Taylor S, Kuch K, Koch WJ, Crockett DJ, Passey G. 1998. The

structure of posttraumatic stress symptoms. J Abnorm Psychol

107:154–160.

Thompson JA, Charlton PFC, Kerry R, Lee D, Turner SW. 1995.

An open trial of exposure therapy based on deconditioning for

post-traumatic stress disorder. Br J Clin Psychol 34:407–416.

Thrasher SM, Lovell K, Noshirvani M, Livanou M. 1996. Cogni-

tive restructuring in the treatment of post-traumatic stress disor-

der: two single cases. Clin Psychol Psychotherapy 3:137–148.

Van der Kolk BA, Dreyfuss D, Michaels MJ, Shera D, Berkowitz

R, Fisler RE, Saxe GN. 1994. Fluoxetine in posttraumatic stress

disorder. J Clin Psychiatry 55:517–522.

Weathers FW, Litz BT, Herman DS, Huska JA, Keane TM. 1993.

The PTSD Checklist (PCL): Reliability, validity, and diagnostic

utility. Paper presented at the 9th Annual Meeting of ISTSS.

Weathers FW, Blake DD, Krinsley KE, Haddad W, Ruscio AM,

Keane TM, Huska JA. 1999a. Reliability and validity of the clini-

cian-administered PTSD scale. Manuscript submitted for publi-

cation.

Weathers FW, Ruscio AM, Keane TM. 1999b. Psychometric prop-

erties of nine scoring rules for the Clinician-Administered Post-

traumatic Stress Disorder Scale. Psychol Assess 11:124–133.

Yehuda R, Kahan B, Schmeidler J, Southwick SM, Wilson S, Giller

EL. 1995. Impact of cumulative lifetime trauma and recent stress

on current posttraumatic stress disorder symptoms in Holocaust

survivors. Am J Psychiatry 152:1815–1818.

Zlotnick C, Davidson JRT, Shea MT, Pearlstein T. 1996. Valida-

tion of the Davidson Trauma Scale in a sample of survivors of

childhood sexual abuse. J Nerv Ment Dis 184:255–257.

A quasi-experimental trial of narrative reconstruction for prolonged grief disorder: Symptomatic improvement and enhanced memory integration

Article

Aug 2023
J CLIN PSYCHOL

Background: Prolonged grief disorder (PGD) was recently approved as a formal diagnosis in the DSM-5-TR. The implementation of bereavement interventions is frequently requested, but their effectiveness has been controversial. Narrative reconstruction (NR) is a time-limited integrative therapy, originally developed for the treatment of post-traumatic stress disorder (PTSD) and adapted for the treatment of PGD. NR consists of exposure to the loss memory, a detailed written reconstruction of the loss memory narrative, and an elaboration of the personal significance of that memory for the bereaved. Objectives: In this study we evaluated the efficacy of NR for PGD. Method: In this study, 33 participants with PGD were quasi-randomized-that is, assigned to an immediate (n = 20) or delayed (n = 13) 16-session NR intervention. PGD, intrusion, avoidance and depression symptoms, as well as levels of the loss memory integration, were assessed at pretreatment, post-treatment, and at a 3-month follow-up. Results: Mixed linear models showed significant intervention effects for PGD and intrusive symptomatology. Results also showed an increase in integration of the loss memory, and improvements remained stable for all outcomes at follow-up. Conclusion: In this study we established NR as an effective intervention for PGD and call for further validation in future studies. Integrating this intervention into the routine care of people with PGD seems important and beneficial.

Emotional freedom technique versus written exposure therapy versus waiting list for post-traumatic stress disorder: protocol for a randomised clinical MRI study

Article

Full-text available

Jun 2023

Introduction: The emotional freedom technique (EFT) is an acupuncture-based psychotherapy that combines tapping on acupoints with cognitive reframing. EFT has been previously shown to have potential for treating post-traumatic stress disorder (PTSD). However, further clinical evidence and underlying mechanisms of EFT are yet to be fully explored. This proposed clinical trial aims to examine the effect of EFT on patients with PTSD compared with the waitlist (WL) and active controls. Methods and analysis: This study was designed as a randomised, assessor-blinded, three-arm clinical MRI study. A total of 120 eligible patients with PTSD will be recruited and randomised into EFT, written exposure therapy (WET) or WL groups. EFT and WET will be applied once a week for 5 weeks. For patients in the WL group, EFT will be performed after 12 weeks. PTSD symptoms, depression, anxiety, somatic symptoms and quality of life will be evaluated. Assessments will be conducted at baseline (week 0), post-treatment (week 6) and follow-up (week 12). Structural and functional brain images and recording videos of facial expressions to emotional stimuli will be obtained before and after treatment. Sixty participants without lifetime traumatic experiences will be enrolled as healthy controls. The primary objective of the study is to compare the change from baseline in the Clinician-Administered PTSD Scale after treatment (week 6) between EFT and WL groups and between EFT and WET groups. Ethics and dissemination: Ethical approval was obtained from the Institutional Review Board of the Kyung Hee University Korean Medicine Hospital. The research findings will be shared at national and international conferences and will be published in peer-reviewed journals. Trial registration number: Clinical Research Information Service KCT0007360 https://cris.nih.go.kr/cris/search/detailSearch.do/21974.

Inhibitory control and alcohol use history predict changes in posttraumatic stress disorder symptoms

Article

Jun 2023

Objective: Posttraumatic stress disorder (PTSD) is associated with significant disability and can become chronic. Predictors of PTSD symptom changes over time, especially in those with a PTSD diagnosis, remain incompletely characterized. Method: In the present study, we examined 187 post-9/11 veterans (Mage = 32.8 years, 87% male) diagnosed with PTSD who performed two extensive clinical and cognitive evaluations approximately 2 years apart. Results: We found that greater PTSD symptom reductions over time were related to lower lifetime drinking history and better baseline inhibitory control ability (Color-Word Inhibition and Inhibition/Switching), though not performance on other executive function tasks. Further, groups with reliably Improved, Worsened, or Chronic PTSD symptoms demonstrated significant differences in baseline inhibitory control and lifetime drinking history, with marked drinking differences starting in the early-to-mid 20s. We also found that PTSD symptom changes showed little-to-no associations with changes in inhibitory control or alcohol consumption. Conclusions: Together, these findings suggest that, in those diagnosed with PTSD, inhibitory control and alcohol use history reflect relatively stable risk/resiliency factors predictive of PTSD chronicity. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

High executive functioning is associated with reduced posttraumatic stress after trauma exposure among male U.S. military personnel

Article

Full-text available

Sep 2023

Introduction Evidence suggests that executive function (EF) may play a key role in development of PTSD, possibly influenced by factors such as trauma type and timing. Since EF can be improved through intervention, it may be an important target for promoting resilience to trauma exposure. However, more research is needed to understand the relation between trauma exposure, EF, and PTSD. The goal of this study was to improve understanding of EF as a potential antecedent or protective factor for the development of PTSD among military personnel. Method In a cohort of U.S. Marines and Navy personnel ( N = 1,373), the current study tested the association between exposure to traumatic events (pre-deployment and during deployment) and PTSD severity, and whether EF moderated these associations. Three types of pre-deployment trauma exposure were examined: cumulative exposure , which included total number of events participants endorsed as having happened to them, witnessed, or learned about; direct exposure , which included total number of events participants endorsed as having happened to them; and interpersonal exposure , which included total number of interpersonally traumatic events participants’ endorsed. EF was measured using the Penn Computerized Neurocognitive Battery. Results EF was associated with less PTSD symptom severity at pre-deployment, even when adjusting for trauma exposure, alcohol use, traumatic brain injury, and number of years in the military. EF also moderated the relation between cumulative trauma exposure and interpersonal trauma exposure and PTSD, with higher EF linked to a 20 and 33% reduction in expected point increase in PTSD symptoms with cumulative and interpersonal trauma exposure, respectively. Finally, higher pre-deployment EF was associated with reduced PTSD symptom severity at post-deployment, independent of deployment-related trauma exposure and adjusting for pre-deployment PTSD. Conclusion Our results suggest that EF plays a significant, if small role in the development of PTSD symptoms after trauma exposure among military personnel. These findings provide important considerations for future research and intervention and prevention, specifically, incorporating a focus on improving EF in PTSD treatment.

Evaluation of a Telehealth-Enabled Pilot Program to Address Intensive Care Unit Health Care Worker Mental Health Distress

Article

Full-text available

Aug 2023

Introduction: Health care workers (HCWs) are at heightened risk of adverse mental health events (AMHEs) and burnout with resultant impact on health care staffing, outcomes, and costs. We piloted a telehealth-enabled mental health screening and support platform among HCWs in the intensive care unit (ICU) setting at a tertiary care center. Methods: A survey consisting of validated screening tools was electronically disseminated to a potential cohort of 178 ICU HCWs. Participants were given real-time feedback on their results and those at risk were provided invitations to meet with resiliency clinicians. Participants were further invited to engage in a 3-month longitudinal assessment of their well-being through repeat surveys and a weekly text-based check-in coupled with self-help tips. Programmatic engagement was evaluated and associations between at-risk scores and engagement were assessed. Qualitative input regarding programmatic uptake and acceptance was gathered through key informant interviews. Results: Fifty (28%) HCWs participated in the program. Half of the participants identified as female, and most participants were white (74%) and under the age of 50 years (93%). Nurses (38%), physicians-in-training (24%), and faculty-level physicians (20%) engaged most frequently. There were 19 (38%) requests for an appointment with a resiliency clinician. The incidence of clinically significant symptoms of AMHEs and burnout was high but not clearly associated with engagement. Additional programmatic tailoring was encouraged by key informants while time was identified as a barrier to program engagement. Discussion: A telehealth-enabled platform is a feasible approach to screening at-risk HCWs for AMHEs and can facilitate engagement with support services.

Treatment length and symptom improvement in prolonged exposure and present-centered therapy for posttraumatic stress disorder: Comparing dose-response and good-enough level models in two manualized interventions

Article

Jul 2023
J CONSULT CLIN PSYCH

Objective: The dose-response model of change in psychotherapy posits that each session of therapy is incrementally beneficial across patients. The contrasting good-enough level model suggests that patients improve at different rates in therapy and discontinue treatment when they are satisfied with their improvement. Support for each theory has been mixed, and many prior studies have relied on samples of patients receiving unstructured treatment approaches. We conducted this study to compare these two theories across two manualized treatments for posttraumatic stress disorder (PTSD). Method: Two hundred eighty-four female veterans and military service members with PTSD (Mage = 44.79; 54.6% White non-Hispanic, 6.7% Black non-Hispanic, 37% other) were randomized to receive 10 sessions of prolonged exposure (PE), a trauma-focused therapy, or present-centered therapy (PCT), a non-trauma-focused therapy. Participants completed the PTSD Checklist (PCL) at even-numbered treatment sessions, and the timing of dropout/treatment completion was monitored. Results: The point of highest risk for dropout differed between the treatments, with risk in PE corresponding to the beginning of imaginal exposures. In the PE condition, but not in PCT, a higher number of sessions completed increased the likelihood of achieving reliable clinically significant improvement. Across treatments, the rate of change in PTSD symptoms did not differ according to the number of sessions completed (b = 0.06, p = .687). Conclusions: Findings support the dose-response model of change in psychotherapy. There were notable differences in dropout across the treatment conditions, including rates, timing, and implications for outcomes. These differences likely reflect differences in content between the protocols. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Çocukluk Döneminde Maruz Kalınan İstismar Nedeniyle Travma Sonrası Stres Bozukluğu Yaşayan Bireylere Özgü Etkili Psikolojik TedavilerEFFECTIVE PSYCHOLOGICAL TREATMENTS SPECIFIC TO INDIVIDUALS SUFFERING FROM POST-TRAUMATIC STRESS DISORDER DUE TO CHILDHOOD ABUSE

Article

Jul 2023

Çocukluk döneminde yaşanan duygusal, fiziksel, cinsel ve ekonomik istismar kişiyi ruhsal yönde olumsuz olarak etkilemektedir. Bu istismarları yaşayan kişiler birden fazla psikopatolojiyi aynı anda geliştirebilirken bu kişilerde en sık rastlanan psikopatoloji Travma Sonrası Stres Bozukluğu (TSSB)’dur. Özellikle fiziksel ve cinsel istismar yaşayan çocukların bu bozukluğu geliştirme riski daha yüksektir. Bu çalışmanın amacı çocukluk döneminde maruz kalınan istismardan dolayı TSSB geliştiren bireyler için etkili, kanıtlanmış, güncel psikolojik tedavileri araştırmaktır. Cinsel istismardan dolayı TSSB geliştiren bireyler için kanıtlanmış müdahale teknikleri: Travma Odaklı Bilişsel Davranışçı Terapi (BDT), Diyalektik Davranış Terapi (DDT), Duygu ve Kişilerarası İlişkiyi Düzenleme Becerisi Eğitimi (STAIR), Travma için Duygu Odaklı Terapi (DOT), Göz Hareketleriyle Duyarsızlaştırma ve Yeniden İşlemleme (EMDR), Yeniden Senaryolaştırma (Imagery Rescripting, IR)’dır.

Associations of Warzone Veteran and Intimate Partner PTSD Symptoms with Child Depression, Anxiety, Hyperactivity, and Conduct Problems

Article

Aug 2023

Altered functional connectivity between cortical networks associated with inhibitory control in trauma-exposed females

Article

Jun 2023
PSYCHIAT RES-NEUROIM

Post-traumatic stress disorder (PTSD) is associated with impaired inhibitory control and alterations in large-scale brain network connectivity. However, few studies to date have examined the construct of inhibitory control as it relates to resting-state functional connectivity (rsFC) in a population with PTSD or trauma-exposure. The present study investigated the relationship between impaired inhibitory control and rsFC within the default mode network (DMN), central executive network (CEN), and salience network (SN) in a sample of females exposed to interpersonal trauma with and without PTSD (n = 67). Participants completed a classic Color-Word Stroop task as a measure of inhibitory control and two resting-state fMRI scans. We conducted voxelwise rsFC analyses with seed regions in the DMN, CEN, and SN and voxelwise linear regression analyses to examine the relationship between inhibitory control and rsFC of these networks across the sample. Better Stroop performance was negatively associated with total self-reported PTSD symptoms. An analysis of PTSD symptom clusters indicated that better Stroop performance was also associated with re-experiencing and hyperarousal symptoms, but not avoidance PTSD symptoms. Decreased coupling between the CEN and the DMN was associated with better inhibitory control in this sample of trauma-exposed females. These findings lend support to the hypothesis that efficient switching between these networks may contribute to better performance on cognitive and attentional tasks in trauma-exposed individuals.

The Cognitive Profiles of Partial PTSD in Vietnam War Veterans

Article

Full-text available

May 2023

Objectives This study identifies the distinct cognitive function profiles of partial posttraumatic stress disorder (PTSD). Methods Medical records of Vietnam War Korean Veterans, who visited a psychiatric clinic and underwent a Clinician-Administered PTSD scale (CAPS) and neurocognitive test on the same day, were retrospectively reviewed. Based on the CAPS interview, participants were divided into partial PTSD (n=10) and non-PTSD (n=36) groups. The neurocognitive test score profiles of both groups were compared. Analysis of covariance was performed to adjust the contribution of possible confounders to cognitive function. Results Cognitive profile analysis displayed significantly poorer performance of executive function and mental flexibility (measured by the Trail Making Test B) and concentration (measured by mini mental status exam) in the partial PTSD group. When adjusted with age, depressive symptom measure, and education level, no significant differences were obtained in the cognitive profiles between both groups. Conclusion Results of the current study revealed a deficit in executive function and concentration in partial PTSD subjects compared to non-PTSD. However, when adjusted with possible confounders such as depressive symptoms, the cognitive profiles of partial PTSD displayed no significant difference with the cognitive profiles of non-PTSD subjects.

Psychometric properties of nine scoring rules for the Clinician-Administered Posttraumatic Stress Disorder Scale

Article

Full-text available

Jun 1999

The use of structured interviews that yield continuous measures of symptom severity has become increasingly widespread in the assessment of posttraumatic stress disorder (PTSD). To date, however, few scoring rules have been developed for converting continuous severity scores into dichotomous PTSD diagnoses. In this article, we describe and evaluate 9 such rules for the Clinician-Administered PTSD Scale (CAPS). Overall, these rules demonstrated good to excellent reliability and good correspondence with a PTSD diagnosis based on the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders (3rd ed., rev.; DSM-III—R ; American Psychiatric Association, 1987). However, the rules yielded widely varying prevalence estimates in 2 samples of male Vietnam veterans. Also, the use of DSM-III—R versus DSM-IV criteria had negligible impact on PTSD diagnostic status. The selection of CAPS scoring rules for different assessment tasks is discussed. (PsycINFO Database Record (c) 2012 APA, all rights reserved)

A clinician rating scale for assessing current and lifetime PTSD: The CAPS-1

Article

Full-text available

Jan 1990
BEHAV THER

Empirical development of an MMPI subscale for the assessment of combat-related posttraumatic stress disorder.

Article

Full-text available

Jan 1984

• 200 male Vietnam combat veterans who were assigned to a posttraumatic stress disorder (PTSD) group (n = 100) or a non-PTSD group (n = 100) were administered the MMPI to develop empirically based criteria for use of the MMPI to aid in assessment and diagnosis of PTSD. Standard clinical profiles demonstrated that PTSD Ss had overall higher mean elevations, and discriminant function analysis correctly classified 74% of Ss in each group. A special PTSD subscale was developed and cross-validated that improved diagnostic hit rates to 82% of Ss. (9 ref) (PsycINFO Database Record (c) 2012 APA, all rights reserved)

Pharmacotherapy of post-traumatic stress disorder with a novel psychotropic

Article

Jan 1994

This was a multicenter, randomized, double-blind, parallel trial conducted in outpatients in three countries. Following screening and placebo washout, patients received brofaromine (a combined MAO-A inhibitor/5-HT transport inhibitor) or placebo in a !exible dosing design. Based upon the CAPS, a standardized post-traumatic stress disorder (PTSD) interview, "ndings from a cohort involving both subchronic and chronic traumatic stress marginally favored brofaromine over placebo; however, not to a statistically signi"cant degree. With a more conservation de"nition of the syndrome, employing a primary cohort of patients with PTSD of one year or greater duration, brofaromine signi"cantly reduced PTSD symptoms in comparison with placebo. In all analyses a substantial proportion of patients in both drug and placebo groups remained symptomatic throughout. Findings were supported by an analysis of secondary measures. Brofaromine may be of bene"t in the therapy of PTSD. Anxiety 1:169–174 (1994/1995).

An inventory for measuring clinical anxiety: Psychometric properties

Article

Jan 1988
J CONSULT CLIN PSYCH

The development of a clinician administered post-traumatic stress disorder (PTSD) scale

Article

Jan 1990
J TRAUMA STRESS

SCL-90-R administration, scoring and procedures manual-II towson

Article

Jan 1994

L.R. Derogatis

State-trait anxiety inventory (STAI)

Article

Eye movement desensitization for PTSD of combat: A treatment outcome pilot study

Article

Jan 1993

Three- to Four-Year Follow-Up to an Open Trial of Nefazodone for Combat-Related Posttraumatic Stress Disorder

Article

Jan 2002

Multiyear (37–51 months) follow-up data was obtained on patients who had participated in an open label trial of nefazodone that originally showed nefazodone may be useful for symptom management in posttraumatic stress disorder (PTSD) patients. Ten patients with combat-related DSM-IV posttraumatic stress disorder (PTSD) entered an open-label 12-week trial of nefazodone, beginning with 100 mg/day and increasing as necessary to achieve a maximal response or until reaching a maximum dosage of 600 mg/day. All 10 patients were followed for over 3–4 years and used nefazodone with dosages of 400–600 mg a day. The entire dosage was shifted to bedtime to facilitate sleep in 7 patients. Data on PTSD symptoms, depression, sleep, and anger were examined. Nefazodone was well tolerated and no significant changes in sexual function were reported. All participants reported compliance with the prescribed nefazodone over 3–4 years. Nine patients reported that it remained effective, and expressed a desire to remain on the medication. On the basis of clinician global impression ratings (compared to baseline), 10 patients were rated as much improved at 12 weeks. Seven of the 10 patients continued to be much improved, 2 were minimally improved, and 1 was rated as worse (compared to baseline assessment) on 3–4-year follow-up. At 3–4-year follow-up, improvements in PTSD symptoms, sleep, and anger were maintained. These improvements were statistically significant with moderate-to-large effect sizes. These data suggest that clinical improvement in PTSD patients administered nefazodone may be maintained with continued treatment. The medication was tolerated well in long-term treatment, compliance was high, and improvement was maintained over several years. Length of treatment, appropriate dose, long-term efficacy, and compliance are all clinically significant issues with little guiding data available. Controlled studies are needed to (a) further investigate the long-term efficacy of nefazodone in the treatment of PTSD; (b) provide information for length of treatment guidelines; and (c) document if discontinuation is possible and efficacious.

Clinician-Administered PTSD Scale: A review of the first ten years of research

Abstract

Recommendations

Clinician-Administered PTSD Scale: A review of the first ten years of research

Psychometric properties of nine scoring rules for the Clinician-Administered Posttraumatic Stress Di...

Psychometric properties of nine scoring rules for the Clinian-Administered PTSD Scale (CAPS)

The Criterion A problem revisited: Controversies and challenges in defining and measuring psychologi...