Predictors
of Posttraumatic Samuel
Perry,
M.D.,
Allen
J. Frances,
Stress JoAnn
M.D.,
Disorder
Difede,
M.A.,
and Lawrence
After
Gloria
Jacobsberg,
Burn
Musngi,
R.N.,
M.D.,
Ph.D.
Injury
Objective: The authors’ goal was to examine subjective and objective predictors of posttraumatic stress disorder (PTSD). Method: Hospitalized burn patients were assessed 1 week after injury with both objective predictors (percent of burned area and facial disfigurement) and
subjective
predictors
(emotional
distress
and
perceived
social
support).
The
patients
were
then assessed 2, 6, and 12 months later for development ofPTSD. Results: Among 51 patients, 1 8 (35.3%) met PTSD criteria at 2 months. High rates ofPTSD were also found at 6 months (N=16, 40.0% ofthe 40 available patients) and 12 months (N=14, 45.2% ofthe 31 available patients). PTSD was predicted by subjective variables assessed at baseline, but patients with more severe burns were not more likely to develop PTSD. Conclusions: The DSM-HI-R diagnosis ofPTSD relies on an objective evaluation ofthe stressor’s severity. The prospective data in this study support those who argue that evaluations of the severity of the stressor might also take into account subjective factors. (Am J Psychiatry 1992; 149:931-935)
A
lthough the emotional effects of severe trauma were recognized by Chancot, Freud, Janet, and others, the diagnosis of these effects has been inconsistently defined (1-3). In DSM-I, “gross stress reaction” excluded patients with other psychopathology, DSM-II deleted this category altogether, and DSM-III introduced explicit criteria for posttraumatic stress disorder (PTSD), but then DSM-III-R narrowed the stresson critenon and added that the symptoms must have a duralion
of at least
1 month.
One reason for this inconsistency has been the paucity of prospective data. Most studies have not assessed subjects immediately after the trauma and instead have evaluated symptomatic groups months or years after the stressful event. Such methods not only raise questions about sample bias and retrospective reporting but also preclude determining if the development of PTSD is related to the individual’s emotional state at the time of the trauma. Further, studies have not quantified the stressor to determine if its severity predicts PTSD. The DSM-IV Task Force is considering whether the current criteria for PTSD should once again be modifled, especially regarding the definition of the stresson criterion. The DSM-III criterion (“a recognizable stressor that would evoke significant symptoms of distress in almost everyone”) may have been too broad Received
1991;
June
accepted
Cornell
University
13, 1990;
received May 15 and Oct. 23, From the Department of Psychiatry, College and Columbia University College New York City. Address reprint requests revisions
Medical
in part by grant
of General Medical Sciences. Copyright © 1992 American
Am
J
Psychiatry
wrongfully
excludes
some
the concept, whereas the event outside the range of may be so narrow that it patients
with
posttraumatic
symptoms. Also at issue is whether the stresson cnitenon should include an assessment of the individual’s subjective response to the trauma, such as emotional distress and perceived helplessness. Finally, there is debate as to whether avoidant mechanisms reliably and validly characterize the phenomenology of PTSD. Perhaps social withdrawal and numbing of responsiveness (“emotional anesthesia”) are not characteristic and only intrusive criteria (e.g., “flashbacks”) should be used diagnostically. In response to the call for prospective data to address these issues (2-5), we evaluated patients shortly after hospitalization for a burn injury and followed a representative sample for 1 year to determine the development of PTSD. We chose to study burn patients because the severity of the stressor can b#{231} objectively quantified and has a wide variance, because the distress associated with the trauma can be assessed shortly after the event, and because burn injury requiring hospitalization contains those elements presumed to be associated with PTSD (threat of disability or death; exposure to overwhelming stimuli; suddenness, unexpectedness, and unpreparedness; loss of loved ones and/or property; and direct involvement with the stresson) (6).
Dec. 2, 1991.
of Physicians and Surgeons, to Dr. Perry, 525 East 68th St., New York, NY 10021. Supported
and may have trivialized DSM-III-R criterion (“an usual human experience”)
149:7,
GM-26145
from the National
METHOD
Institute
Every Psychiatric
July
1992
Association.
center
adult patient was approached
admitted within
to our hospital’s burn the first week of hospi-
931
PTSD
AFThR
BURN INJURY
talization. If the patient was able and willing to give informed consent, four subjective variables were measured: 1) emotional distress as measured by the Profile of Mood States (POMS) (7), which includes an assessment of state anxiety and helplessness, 2) perceived social
support
as measured
by the
Interpersonal
Support
Evaluation List (8), 3) intrusive thoughts during the past week about the burn injury as measured by the Impact of Event Scale (9), and 4) avoidant thoughts about the injury as measured by the Impact of Event Scale (9). The selection of these four subjective predictons of PTSD was based on a review of existing litenatune (2, 4-6, 10). Many patients are unable to write during
the
first
few
weeks
following
their
injuries.
For
consistency in data collection, investigators read the questionnaires to all patients and marked their responses on the standardized instruments. Objective severity of the stressor was assessed by using two measures: percent of total body surface area of the burn and percent of patients with facial disfigurement. Although not the primary focus of the study, we also estimated psychopathology before the burn by reviewing charts and interviewing available informants about any history of alcoholism, drug abuse, psychosis, and psychiatric hospitalization. Patients were subsequently assessed at 2, 6, and 12 months after their burn injury. A clinical rating of PTSD was obtained by using the Structured Clinical Interview for DSM-III-R (SCID) (1 1). To examine the concurrent association between PTSD and psychological factors and to determine if these psychological factors predicted future PTSD, patients then completed the POMS and the Impact of Event Scale. The interviewers were blind to responses on these self-report measures. Patients geographically distant on otherwise unable on unwilling
to attend
follow-up
cally rated by a 30-minute and returned questionnaires Logistic regression analyses extentto
predicted determine
of Event
which
PTSD
baseline
assessments
telephone by mail. were used
subjective
were
SCID
clini-
interview
to determine
and objective
the
measures
at 2-month-12-month follow-up and to scores on the POMS and the Impact at 2-month and 6-month follow-up pre-
whether
Scale
dicted PTSD at later assessments. examinedconcurrentnelationships psychological measures (POMS
Analyses
of variance
between
PTSD and the and the Impact of Event thoughts). Since others (12)
Scale intrusive and avoidant have reported a correspondence between PTSD and Impact of Event Scale cutoff scores (greater than 19 on both intrusive and avoidant thoughts about the injury), we cxamined the specificity and sensitivity of these cutoffs in our sample. Finally, we explored whether PTSD was predicted by sociodemographic variables and estimates of premorbid psychopathology.
RESULTS
Two-month and 17 women;
932
follow-up 22 (43%)
data were obtained ofthese Si patients
for 34 men were white,
were black, 11 (22%) were Hispanic, and were “other”; 23 (45%) were married, 16 (31%) were single, and 12 (24%) were divorced or separated; 24 (47%) were skilled laborers, nine (18%) were homemakers, eight (16%) were unemployed, two (4%) were professionals, and eight (16%) did not provide data on their occupation. A history of alcoholism was documented in 12 (24%), other drug abuse in nine (18%), and psychosis or psychiatric hospitalization in eight (16%). Burns were due to flame injury (N=38, 75%), scalding (N=11, 22%), or chemical exposure (N=2, 4%). 16
(31%)
two
(4%)
Six-month tients, and
follow-up data were available for 40 pa12-month data were available for 31 patients.
An additional 78 patients were assessed within the first week of hospitalization but were lost to follow-up by 2 months after the injury. Two of these patients had died; 68 (89%) ofthe remaining 76 could not be located after discharge. (A change of address is a common consequence of burn injury and its associated physical, familial, and residential disruption.) Eight ( 1 1 %) of the dropouts were located but refused to participate at the 2-month follow-up. At no follow-up assessment was attnition significantly related to age, sex, race, marital status, PTSD at a previous predictor variables (e.g.,
assessment, on hypothesized severity of distress or burn).
To assess the relative contributions of the independent variables (social support, body surface area burned, general distress, intrusive and avoidant thoughts, facial disfigurement) assessed at baseline in predicting PTSD, a forward stepwise logistic regression analyses was performed for each assessment time. As shown in table 1, patients with smaller burns were more likely to meet criteria for PTSD at the 2-month assessment.
PTSD
at this
assessment
was
also
predicted
by less perceived emotional support (Interpersonal Support Evaluation List) and greaten emotional distress (POMS). Regarding subjective variables, less perceived social support at baseline (mean=7.2 days, SD=1.1, aften burn injury), the first variable selected for inclusion in the regression model, predicted acute PTSD at 2month follow-up (mean=10.S weeks, SD=1.9, later). According to model chi-square (table 1), this variable made a significant improvement over the constant in predicting
PTSD
in this
group
of patients.
Severity
of
emotional distress at baseline contributed to the prediction of acute PTSD (table 1). The severity of intrusive or avoidant thoughts during the first week of hospitalization did not predict acute PTSD. Regarding objective variables, the presence of facial disfigurement did not predict
PTSD.
Eleven
(22.4%)
of the
18 patients
with
PTSD at 2 months
and 16 (30.6%) of the 33 patients without PTSD at 2 months had facial disfigurement at baseline. More extensive burn injury was also not associated with the development of PTSD. However, patients with smaller burns were significantly more likely to meet criteria for acute PTSD (table 1). In the final model, the goodness-of-fit chi-square of 40.8 (df=43, p=0.6) had a high probability of occurrence; i.e., the null hypothesis that the model’s predictions differ from the observed data was not confirmed.
Am
J
Psychiatry
149:7, July 1992
DIFEDE,
PERRY,
TABLE
1. Physical
Variable
and Psychological
at Baseline
Interpersonal Evaluation POMS
Support List
Impact of Event Scale Intrusive thoughts Avoidant thoughts
TABLE
2. Psychological
Patients
With
Patients
PTSD Months
at 2
-
(N=18)
Mean
Burned area (%) Scores on psychological measures at baseline
Thus,
thoughts
a regression
sis of social
support,
model body
J
Psychiatry
149:7,