Clinical Hub,UW Health Clinical Tool Search,UW Health Clinical Tool Search,Questionnaires,Related

Psychometric Properties of the PTSD Checklist (PCL)

Psychometric Properties of the PTSD Checklist (PCL) - Clinical Hub, UW Health Clinical Tool Search, UW Health Clinical Tool Search, Questionnaires, Related

Behav. Res. Ther. Vol. 34, No. 8, pp. 6694i73, 1996
Copyright © 1996 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
0005-7967/96 $15.00 + 0.00
Center for Stress and Anxiety Disorders, University at Albany-SUNY, 1535 Western Avenue,
New York, NY 12203, U.S.A.
(Received 2 January 1996; Revised 12 April 1996)
Summary--The psychometric properties of the PTSD Checklist (PCL), a new, brief, self-report
instrument, were determined on a population of 40 motor vehicle accident victims and sexual assault
victims using diagnoses and scores from the CAPS (Clinician Administered PTSD Scale) as the criteria.
For the PCL as a whole, the correlation with the CAPS was 0.929 and diagnostic efficiency was
0.900 versus CAPS. Examination of the individual items showed wide ranging values of individual item
correlations ranging from 0.386 to 0.788, and with diagnostic efficiencies of 0.700 or better for symptoms.
We support the value of the PCL as a brief screening instrument for PTSD. Copyright © 1996 Elsevier
Science Ltd
A research team from the National Center for PTSD (Weathers, Litz, Herman, Huska & Keane,
1993) has recently introduced a brief self-report inventory for assessing for the 17 symptoms of
Post-Traumatic Stress Disorder (PTSD), the PTSD Checklist (PCL). They have provided initial
psychometric data on it including test-retest reliability (0.96; retest interval was not specified) and
validity as indicated by a kappa of 0.64 for diagnosis of PTSD from the SCID \[Structured Clinical
Interview for DSM-III-R: Spitzer, Williams, Gibbon & First, 1990)\], one of the 'gold standards'
for diagnosing PTSD. The latter correlation results were obtained on a sample of 123 male Vietnam
theater veterans.
The PCL thus joins the growing list of relatively new self-report inventories for assessing PTSD
and post-traumatic stress symptoms (PTSS) in a traumatized population. It joins measures such
as that of Foa (Foa, Riggs, Dancu & Rothbaum, 1993) and Davidson's Self-Rating PTSD Scale
(Davidson, Book & Colket, 1995) and the older Impact of Events Scale (Horowitz, Wiimer &
Alvarez, 1979). We believe it is important that new instruments be subjected to independent
cross-validation. It would also seem appropriate to conduct such a cross-validation of the PCL
on populations who had been the victims of different types of trauma. Such was the primary
purpose of the present study.
A secondary purpose was to examine the psychometric properties of the 17 individual
items of the PCL. Weathers et al. reported data on the 3 symptom clusters which make up
PTSD, including internal consistency (alpha coefficients) values which ranged from 0.89 to 0.92,
but no data on the individual items. The properties of the individual items seem important if,
following Weathers et al.'s advice, one wanted to make diagnostic judgments about the possible
presence of PTSD based on the PCL. This is especially important because of the unusual
diagnostic criteria requiring at least 1 reexperiencing symptom \[out of 5 according to DSM-IV
(American Psychiatric Association, 1994)\], 3 avoidance and psychic numbing symptoms (out of
7), and 2 hyperarousal symptoms (out of 5). If different PCL items have different levels of validity
(or even require different cut-off scores to maximize 'hits' and minimize 'false positives' and
'false negatives' or 'misses'), then such item information would be of value for possible diagnostic
*Author for correspondence.

670 Edward B. Blanchard et aL
Forty adults who either had been involved in a severe automobile accident (n = 27) or had been
the victim of a sexual assault (n = 13) participated. Demographic and diagnostic information on
the sample are contained in Table 1. Participants were characterized as PTSD if they met the full
DSM-IV (American Psychiatric Association, 1994) criteria, or as sub-syndromal PTSD if they met
Criterion B and either Criterion C or Criterion D (but not both). Participants were characterized
as non-PTSD if they met either one criterion or no criteria, based upon the CAPS interview.
Of the 18 participants with PTSD, 9 were suffering from co-morbid Major Depression while
9 were suffering from Generalized Anxiety Disorder (GAD). Of the 9 participants with sub-
syndromal PTSD, 2 were suffering from co-morbid Major Depression while 3 were suffering from
As is obvious from Table 1, our sample was predominantly female (92%) of average age 40.0 yr.
The motor vehicle accident (MVA) victims were all recent trauma victims whereas the sexual assault
survivors (SASs) had experienced their trauma considerably longer ago.
All participants were paid volunteers in other research projects. They all gave written informed
consent before the study. The MVA victims were paid $30 whereas the SASs were paid $20. The
reason for the discrepancy was because funding for the two populations came from different
Potential participants were recruited by means of advertisements or stories in local media.
Some of the MVA victims were referred by local medical practitioners whereas some of the SASs
were referred by a local rape crisis center.
Two common sources of data were used with all participants. The PCL was mailed to the
participant with instructions to fill it out (along with other questionnaires which are not part of
this report) and bring it to the interview. Interviewers did not have knowledge of the participant's
PCL responses prior to the interview.
Participants also took part in a lengthy (2-4 hr) individual interview which included adminis-
tration of the CAPS (Clinician-Administered PTSD Scale: Blake, Weathers, Nagy, Kaloupek,
Klauminzer, Charney & Keane, 1990) by a trained interviewer, who was either a doctoral level
psychologist with much experience with the CAPS, or an advanced doctoral student who had been
trained in the administration of the CAPS by the senior interviewer. The other portions of the
interviews were more idiosyncratic to the individual's source of trauma.
All CAPS interviews were tape recorded. Nineteen were rescored from the tape by a doctoral
student in clinical psychology, blind to the original diagnosis. The kappa for categorical agreement
on diagnosis was 0.836, P < 0.001. Correlations on individual item scores ranged from 0.843 to
0.991 with an average of 0.945, P = < 0.001.
For the PCL the 17 PTSD symptoms are rated by the participant for the previous month on
a scale indicating the degree to which the respondent had been bothered by a particular symptom
from 1 (not at all) to 5 (extremely). For the CAPS the assessor makes two ratings for each symptom,
its relative frequency (or percent of time or events to which it applies) for the past month on a
0-4 scale and its severity at its worse for the past month, also on a 0-4 scale. The two values are
Table 1. Demographic and diagnostic information on the sample
Type of trauma
Current CAPS Gender Age Years since trauma PCL scores CAPS scores
diagnosis M/F X (SD) X (SD) X (SD) X (SD)
Motor vehicle accidents PTSD
Sub-syndromal PTSD
Sexual assault PTSD
Sub-syndromal PTSD
3/12 44.00 (I 1.58) 1.17 (0.79) 60.0 (9.4) 73.8 (18.5)
0/5 39.90 (8.30) 0.80 (0,27) 42.8 (5.4) 38.2 (5.5)
0/7 48.50 05.59) 1.43 (0.93) 26.6 (4.6) 10.9 (9.5)
0/3 21.33 (2.31) 7.67 (3.51) 55.0 (16.7) 61.7 (27.3)
0/4 28.75 (7,54) 11.50 (6.86) 43.8 (12.0) 35.0 (12.5)
0/6 36.67 (14.57) 23.50 (16.83) 22.8 (11.8) 31,7 (10.2)

Psychometric properties of the PCL
Table 2. Prediction parameters for total PCL score vs CAPS diagnosis
Total PCL Positive Negative
score cut off Sensitivity Specificity predictive power predictive power Diagnostic efficiency
50 0.778 0,864 0.824 0.826 0.825
44 0.944 0.864 0.850 0.950 0.900
Notes: Sensitivity = true positives/\[true positives + misses (i.e. false negatives)\].Specificity = true negatives/\[true negatives + false positives\].
Positive predictive power = true positives/\[true positives + false positives\].Negative predictive power = true negatives/\[true negatives +
misses\]Diagnostic efficiency = (true positives + true negatives)/total sample.
summed to arrive at an individual symptom score (range 0-8). Previous research (Blanchard,
Hickling, Taylor, Forneris, Loos & Jaccard, 1995) has suggested using as a minimum sum either
a score of 3 or 4 on a symptom for it to count as positive towards the diagnosis.
PCL total score
PCL scores ranged from 17 to 74 with a mean of 45.8 (SD = 16.1), while the CAPS scores ranged
from 0 to 119 with a mean of 45.9 (SD = 29.1). The overall correlation of PCL total score with
CAPS total score was r(38) = 0.929, P < 0.0001. Using the recommended PCL cutoff score of 50,
we could correctly classify 14 of 18 participants with PTSD (and thus missed 4) while including
3 participants with a sub-syndromal form of PTSD; this yields a sensitivity of 0.778, a specificity
of 0.864 and an overall diagnostic efficiency of 0.825. If we lowered the cutoff score to 44, we
improve overall diagnostic efficiency to 0.900, yielding a sensitivity of 0.944 and specificity of 0.864
and correctly identifying 17 of 18 participants with PTSD. The internal consistency coefficient
(Cronbach's alpha) for the total scale is 0.939. Internal consistency for the items in Criterion B
was 0.935, Criterion C 0.820, and Criterion D 0.839. Prediction parameters for the total score from
the PCL with a criterion of the CAPS diagnosis are presented in Table 2.
Table 3. Parameters for individual PCL items vs individual CAPS items
PCL Positive Negative Pearson r
item predictive predictive Diagnostic with CAPS
Item number score Sensitivity Specificity power power efficiency item
1. Intrusive Recollect 3 0.727 0.778 0.727 0.700 0.750 0.7t5"***
4 0.591 1.000 1.000 0.667 0.775
2. Flashbacks 3 0.800 0.767 0.533 0.920 0.775 0.617"***
4 0.400 0.933 0.667 0.824 0.800
3. Upset by Reminders 3 0.826 0,824 0.864 0.778 0.825 0.668****
4 0.652 0,938 0.940 0.667 0.775
4. Distressing Dreams 3 0.643 0,923 0.818 0.828 0.825 0.788****
4 0.286 1,000 1.000 0.722 0.750
5. Physical Reactions to Reminders 3 0.913 0,882 0.913 0.882 0.900 0.765****
4 0.435 0,941 0.909 0.552 0.650
6. Avoid Thoughts 3 0.571 0,895 0.857 0.654 0.725 0.570****
4 0.429 0.947 0.900 0.600 0.675
7. Avoid Reminders 3 0.810 0.789 0.8t0 0.789 0.800 0.710"***
4 0.571 0.895 0.857 0.654 0.725
8. Psychogenic Amenesia 3 0.786 0.846 0.733 0.880 0.825 0.479**
4 0.500 0.885 0.700 0.767 0.750
9. Anhedonia 3 0.944 0.591 0.654 0.929 0.750 0.660****
4 0.722 0.864 0.813 0.792 0.800
10. Estrangement from Others 3 0.933 0.600 0.583 0.938 0.725 0.688****
4 0.667 0.800 0.667 0.800 0.750
I1, Psychic Numbing 3 0.636 0.944 0.933 0.680 0.775 0.742****
4 0.364 1.000 1.000 0.563 0,400
12. Foreshortened Future 3 0.846 0.630 0.524 0.895 0.700 0.643****
4 0.769 0.741 0.588 0.870 0.750
13. Sleep Difficulty 3 0.885 0.714 0.852 0.769 0.825 0.743****
4 0.577 1.000 1.000 0.560 0.725
14. Irritability/Anger 3 0.708 0,688 0.773 0,611 0.700 0,670****
4 0.417 1.000 1.000 0.533 0.650
15. Concentration Impaired 3 0.913 0.824 0.875 0.875 0.625 0.744****
4 0.652 0.882 0.882 0.652 0.750
16. Hypervigilant 3 0.833 0.500 0.714 0.667 0.770 0.386*
4 0.542 0.813 0.813 0.813 0.650
17. Exaggerated Startle 3 0.909 0.611 0.74t 0.846 0.775 0.630****
4 0.636 0.889 0.875 0.667 0.750
*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

672 Edward B. Blanchard et al.
PCL individual items
As noted above, the symptom criteria for PTSD are fairly complex, with differing numbers of
symptoms required from the different symptom clusters. If the PCL is to be used to make diagnostic
judgments, then the psychometric properties of individual items, rather than for the scale as a
whole, become important. For this reason we believed it would be of value to examine the
psychometric properties of the 17 individual items which make up the PCL and compare them to
our 'gold standard', the CAPS score for the same individual item. In Table 3 are the PCL
psychometric values for each of the 17 individual items.
Examining Table 3, we find all 17 PCL items correlate significantly with the corresponding
individual CAPS items. For two of the items, the correlations are relatively low; 0.479 for
psychogenic amnesia and 0.386 for hypervigilance. Other psychometric properties of the items are
in the table.
Turning to an examination of the other parameters of the individual CAPS items, we call
attention primarily to diagnostic efficiency, that is, the fraction of true positives and true negatives
in the total sample. This parameter, in our opinion, is a good gauge of how well an item can identify
the presence or absence of a PTSD symptom. In the absence of apparent agreement on an
acceptable value for diagnostic efficiency, we have assumed a value of 0.70, that is, 70% of Ss
correctly identified as positive or negative for that symptom. On this basis, all 17 PCL items have
a value of 0.70 or higher for diagnostic efficiency. Interestingly, for 11 items one obtains better
diagnostic efficiency with a PCL value of 3 whereas for the other 6 items one obtains a better
diagnostic efficiency with a PCL score of 4.
Overall, our results with a predominantly female population of trauma victims (MVA victims
and sexual assault survivors) support the value of the PCL, taken as a whole, as a screening device
for possible presence of PTSD. Our data thus supplement the data from Weathers et al., 1993
whose strongest validational data were derived from their first sample, 123 male Vietnam veterans
who were diagnosed by structured interview. Our sample had been, for the most part, traumatized
more recently and was predominantly female. We did find a slightly lower cut-off score (44 as
opposed to 50) led to greater diagnostic efficiency with our sample. This could mean there are
possible gender differences for the appropriate cut-off score or that more recently traumatized
individuals respond somewhat differently than those who are 20 yr or more post-trauma. This latter
point cannot be easily tested in Weathers et al.'s second sample, 1006 Desert Storm veterans, since
no diagnostic interview was used with this latter sample. Further research is needed on this point.
Given that our two instruments are trying to assess the same symptoms and the same
disorder in two different ways (self-report rating on the PCL vs clinician rating on the CAPS),
the overall correlation of 0.929, indicating 86% common variance, seems somewhat remarkable.
Unfortunately, one does not find the same level of correlation for the individual items. This may
reflect a statistical artifact related to the attenuated range found in individual symptoms in
comparison to the whole scale, analogous to one's typically finding greater reliability when a scale
has relatively more items.
While the results for the PCL scale as a whole are quite positive, when one turns to the individual
items (which is necessary for the diagnosis), a less optimistic picture is obtained. For 7 of the 17
symptoms, the correlation coefficients indicate over 50% of the variance in the CAPS item score
is accounted for by the PCL item score. In two instances, psychogenic amnesia and hypervigilance,
the correlation coefficients, although significant, are below 0.5.
Weathers et al., 1993 endorse the PCL to make diagnostic judgements about the presence of
PTSD based on scores of 3 or greater on an appropriate mix of symptoms. We would urge caution
on this point, in part because of the relatively low correlations on the two symptoms mentioned
above and in part because the diagnostic efficiency for some items is lower if one uses a PCL score
of 3 vs PCL score of 4: intrusive recollection, flashbacks, anhedonia, estrangement from others,
foreshortened future, and impaired concentration. If one plans to use the PCL in this fashion (to
make diagnostic judgements), it would seem wise to use different cut off levels for different items.

Psychometric properties of the PCL 673
In any event, we endorse the scale as a screening instrument and as a self-report measure of
degree of post-traumatic stress symptoms, but caution the user to be sensitive to possible gender
differences or trauma type differences in setting a cut-off score.
American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders (4th edn). Washington,
DC: American Psychiatric Association.
Blake, D., Weathers, F., Nagy, L., Kaloupek, D., Klauminzer, G., Charney, D. & Keane, T. (1990). Clinician-Administered
PTSD Scale (CAPS). National Center for Post-Traumatic Stress Disorder, Behavioral Science Division Boston-VA,
Boston, MA.
Blanchard, E. B., Hickling, E. J., Taylor, A. E., Forneris, C. A., Loos, W. R. & Jaccard, J. (1995). Effects of varying scoring
rules of the Clinician-Administered PTSD Scale (CAPS) for the diagnosis of post-traumatic stress disorder in motor
vehicle accident victims. Behaviour Research and Therapy, 33, 471475.
Davidson, J. R. T., Book, S. W. & Colket, J. T. (1995). Davidson Self-Rating PTSD Scale. Available from Multi-Health
Systems, Inc., 908 Niagara Falls Boulevard, North Tonawanda, NY 14120.
Foa, E. B., Riggs, D. S., Dancu, C. V. & Rothbaum, B. O. (1993). Reliability and validity of a brief instrument for assessing
post-traumatic stress disorder. Journal of Traumatic Stress, 6, 45%473.
Horowitz, N. J., Wilmer, N. & Alvarez, N. (1979). Impact of Events Scale: a measure of subjective stress. Psychosomatic
Medicine, 41, 209-218.
Spitzer, R. L., Williams, J. B. W., Gibbon, M. & First, M. B. (1990). Structured Clinical Interview .for DSM-Ill-R,
non-patient edition (SCID-NP) (Version 1.0). Washington, DC: American Psychiatric Press.
Weathers, F. W., Litz, B. T., Herman, D. S., Huska, J. A. & Keane, T. M. (1993). The PTSD checklist: reliability, validity,
& diagnostic utility. Paper presented at the Annual Meeting of the International Society for Traumatic Stress Studies,
San Antonio, TX, October.