North CS, Nixon SJ, Shariat S, Mallonee S, McMillen JC, Spitznagel EL, Smith EM. Psychiatric Disorders Among Survivors of the Oklahoma City Bombing. JAMA. 1999;282(8):755-762. doi:10.1001/jama.282.8.755
Author Affiliations: School of Medicine, Department of Psychiatry (Drs North and Smith), George Warren Brown School of Social Work (Dr McMillen), and Department of Mathematics and Statistics (Dr Spitznagel), Washington University, St Louis, Mo; and Department of Psychiatry and Behavioral Sciences, University of Oklahoma Health Sciences (Dr Nixon), and the Oklahoma State Department of Health (Mss Shariat and Mallonee), Oklahoma City. †Dr Smith died March 7, 1997.
Context Disasters expose unselected populations to traumatic events and can
be used to study the mental health effects. The Oklahoma City, Okla, bombing
is particularly significant for the study of mental health sequelae of trauma
because its extreme magnitude and scope have been predicted to render profound
psychiatric effects on survivors.
Objective To measure the psychiatric impact of the bombing of the Alfred P. Murrah
Federal Building in Oklahoma City on survivors of the direct blast, specifically
examining rates of posttraumatic stress disorder (PTSD), diagnostic comorbidity,
functional impairment, and predictors of postdisaster psychopathology.
Design, Setting, and Participants Of 255 eligible adult survivors selected from a confidential registry,
182 (71%) were assessed systematically by interviews approximately 6 months
after the disaster, between August and December 1995.
Main Outcome Measures Diagnosis of 8 psychiatric disorders, demographic data, level of functioning,
treatment, exposure to the event, involvement of family and friends, and physical
injuries, as ascertained by the Diagnostic Interview Schedule/Disaster Supplement.
Results Forty-five percent of the subjects had a postdisaster psychiatric disorder
and 34.3% had PTSD. Predictors included disaster exposure, female sex (for
any postdisaster diagnosis, 55% vs 34% for men; χ21=8.27; P=.004), and predisaster psychiatric disorder (for PTSD,
45% vs 26% for those without predisaster disorder; χ21=6.86; P=.009). Onset of PTSD was swift, with
76% reporting same-day onset. The relatively uncommon avoidance and numbing
symptoms virtually dictated the diagnosis of PTSD (94% meeting avoidance and
numbing criteria had full PTSD diagnosis) and were further associated with
psychiatric comorbidity, functional impairment, and treatment received. Intrusive
reexperience and hyperarousal symptoms were nearly universal, but by themselves
were generally unassociated with other psychopathology or impairment in functioning.
Conclusions Our data suggest that a focus on avoidance and numbing symptoms could
have provided an effective screening procedure for PTSD and could have identified
most psychiatric cases early in the acute postdisaster period. Psychiatric
comorbidity further identified those with functional disability and treatment
need. The nearly universal yet distressing intrusive reexperience and hyperarousal
symptoms in the majority of nonpsychiatrically ill persons may be addressed
by nonmedical interventions of reassurance and support.
Disasters offer unique opportunities to study mental health effects
of traumatic events in unselected populations. Because most disasters strike
randomly, studies of disasters circumvent the limitations of research on trauma
to individuals in the community, where risk for traumatic events is confounded
with vulnerability to psychopathology.1 The
extreme magnitude and intensity of the Oklahoma City, Okla, bombing made it
a particularly significant subject for the study of mental health effects
of trauma because of the profound effects anticipated among its survivors,
including persons with no predisaster psychiatric history.2- 6
The bombing of the Alfred P. Murrah Federal Building in Oklahoma City
on April 19, 1995, was the most severe incident of terrorism ever experienced
on American soil.7,8 The death
count totaled 167, including 19 children; the number of persons injured totaled
684. The fatality rate inside the Murrah Building was 46%, and 93% of survivors
who were in the building were injured.9 The
explosion demolished or damaged more than 800 building structures, with an
estimated property damage of $625 million.
We studied direct survivors of the blast. Our research objectives included
documenting rates of postdisaster psychopathology, examining functional impact,
and identifying predictors of these difficulties to help guide mental health
intervention workers in future disasters. We anticipated that the scope and
severity of this event would elicit higher rates of psychopathology than previous
disasters studied using similar research methods.
Declaration of all bombing-associated injuries and illnesses as reportable
cases by the Oklahoma State Department of Health commissioner led to the development
of a confidential registry of survivors, from which the study sample was drawn.
The registry contained 1092 names of survivors directly exposed to the blast
based on their proximity to the Murrah Building. Persons exposed only indirectly
through search and rescue or clean-up efforts or by bereavement alone were
not listed. Further detail on the development of the Oklahoma State Department
of Health registry is provided in an earlier publication.8
Participation in the study was limited to subjects at least 18 years
old. Those too severely injured to participate were excluded as ineligible.
To commence with interviewing with minimal delay, the first 20 registry members
to complete and return a preliminary health department survey of their demographics,
exposure to the blast, injuries, and medical treatment were selected for this
study. Subsequently, using an SAS computer program (SAS 6.12; SAS Institute
Inc, Cary, NC), names of 242 additional eligible registry entries were randomly
selected from the registry, which included 1 of the 20 nonrandomized subjects.
Of the 261 subjects thus selected, 3 had left the country, 1 did not speak
English, and 2 had died in the interim, precluding their participation. Of
the 255 remaining subjects, 32 (13%) could not be contacted, 35 (14%) refused
to be interviewed, 6 (2%) were not interviewed with no reason available, and
182 (71%) were interviewed.
Thirty-one percent of the study subjects were within 46 m of the bomb
(distance selected for 99% of all deaths occurring within this radius) at
the instant of detonation. All of these subjects were located in the Murrah
building, except for 1 who was in the Athenian building, which stood across
the street and was in the direct path of the blast. Another 51% were 46 to
184 m from the point of detonation in heavily damaged (YMCA, Water Resources,
and Journal Record buildings) or in less damaged buildings, or outdoors. The
remaining 18% were more than 184 m from the detonation point.
The 35 individuals (16% of those located and eligible) who did not participate
did not differ from study participants in age, sex, injury rates, or medical
treatment received. Significantly more participants than nonparticipants were
in the most heavily damaged buildings (79% vs 55%; χ21=9.69, P=.002), indicating that persons with
less intense exposure may have been less likely to participate in the study.
The sample was representative of the health department's registry population
with respect to sex and age. Compared with the registry population, significantly
more study subjects had been in the most heavily damaged buildings (79% vs
66%; χ21=11.25, P<.001)
and specifically in the Murrah building (31% vs 18%; χ21=15.88, P<.001).
The nonrandomly selected subjects did not differ from the 162 selected
by randomization in demographics, predisaster psychiatric disorder, or any
diagnosis made after the disaster. A higher proportion of them, however, were
in the Murrah building (50% vs 27% of the others; χ21=4.45, P=.04). Removal of the 20 nonrandomly
selected subjects from the sample effectively reduces the proportion of Murrah
building occupancy from 31% to 27%, but the difference from the registry remains
significant (χ21=9.68, P=.002).
Approval for the study was obtained from the Washington University School
of Medicine Institutional Review Board, St Louis, Mo. All subjects provided
written informed consent prior to participating.
An average of 6 months (range, between August and December 1995, 4-8
months for most) after the event, subjects were interviewed using the Diagnostic
Interview Schedule (DIS)/Disaster Supplement, which is based on the Diagnostic and Statistical Manual of Mental Disorders, Revised
Third Edition10 (DSM-III-R), the operating criteria available during the design of longitudinal
disaster studies.11,12 Diagnostic
information was obtained for 8 psychiatric disorders: posttraumatic stress
disorder (PTSD), major depression, panic disorder, generalized anxiety disorder,
somatization disorder, alcohol use disorder, drug use disorder, and antisocial
personality disorder. The interview also documented demographic data, level
of functioning, and treatment received. The Disaster Supplement elicited subjects'
disaster-related experience including exposure to the event, involvement of
family and friends, and physical injuries. All interviews were administered
by members of the Washington University disaster research team who received
formal training to administer the DIS. Sixty-three percent of the interviews
were conducted in person, but due to logistics in the field, 25% were conducted
by telephone, and another 12% initiated in person were completed by telephone.
No associations of any relevant variables with telephone interviews were identified
in the data.
Because individuals could have had PTSD resulting from other traumatic
events besides the bombing, diagnoses and symptoms of bombing-associated PTSD
were tabulated separately from those associated with other traumas. It is
well established that traumatic events experienced by individuals in the community
disproportionately strike persons with proclivities to psychopathology, suggesting
that PTSD following sporadic traumas to individuals in the community may represent
a somewhat different phenomenon from the PTSD arising from a community-wide
disaster such as the bombing.1 Therefore, 4
cases of PTSD unrelated to the bombing were excluded from calculation of postdisaster
Associations between categorical variables were tested using χ2 analyses, substituting Fisher exact tests when expected numbers in
cells were less than 5. Linear regression analyses were performed to compare
numeric variables. For comparisons of means on repeated measures, McNemar
tests were performed. Statistical significance was set at α<.05.
Table 1 displays the demographics
of the sample, which had roughly equal sex representation and was largely
white. Eighty-seven percent of study participants reported injuries sustained
in the blast, and 77% overall had required medical intervention, including
hospitalization (20%) and surgery (15%). The most prevalent injuries were
lacerations (76%), followed by contusions (50%), skin-embedded glass or metal
shards (46%), hearing loss (34%), and smoke or dust inhalation (23%).
Eighty-two percent of survivors reported witnessing others being injured
or killed at the bombing scene, and 46% recalled thinking they were going
to die during the event. Forty-three percent reported loss of a family member
or friend in the bombing, and 92% personally knew someone injured or killed.
Table 2 displays rates of
predisaster and postdisaster disorders. We determined incident and recurrent
or persistent disorders by assessing whether the individual had met criteria
for the same disorder at any time before the bombing. Overall, nearly half
the sample met criteria for 1 or more psychiatric diagnoses after the disaster,
with more than one third qualifying for a diagnosis of PTSD specific to the
bombing. Women had at least twice the rate of PTSD as men (45% vs 23%, respectively; χ21=9.44; P=.002), major depression
(32% vs 13%, respectively; χ21=9.82; P=.002), and generalized anxiety disorder (9% vs 0%, respectively, P=.007). Women were more likely to qualify for any postdisaster
diagnosis (55% vs 34% for men; χ21=8.27; P=.004). No subjects met criteria for somatization disorder or antisocial
Table 2 also shows that
15% of the sample had experienced PTSD at some time before the bombing, and
43% had any predisaster lifetime diagnosis. Seventy-four percent of the subjects
who experienced PTSD had not experienced it before the bombing, and 56% of
subjects who experienced major depression had no history of it before the
bombing. Preexisting major depression was especially likely to persist or
recur after the bombing (78% of predisaster cases). Incident postdisaster
substance use disorders were not observed. The majority of predisaster alcohol
and drug use disorders were reported as inactive after the disaster. For all
diagnoses except generalized anxiety disorder, postdisaster occurrence of
the disorder was statistically associated with predisaster history of the
Fifty-seven percent of subjects with bombing-related PTSD had a predisaster
lifetime history of psychiatric illness. Subjects with a predisaster disorder
were more likely than others to experience bombing-related PTSD (45% vs 26%
for those without a predisaster disorder; χ21=6.86; P=.009). All postdisaster disorders were significantly
associated with history of predisaster psychopathology. Sixty-three percent
of subjects with any active postdisaster psychiatric disorder had a predisaster
diagnosis; ie, more than one third of those with a postdisaster disorder had
never had a psychiatric disorder before the bombing. Conversely, 66% of subjects
who had at least 1 predisaster psychiatric disorder had an active disorder
afterward, compared with 29% of those who had no psychiatric history (χ21=24.32; P<.001).
In 63% of the cases, PTSD was accompanied by postdisaster comorbidity,
occurring most often in 55% of the subjects with PTSD who also were diagnosed
as having major depression. Only 9% of the sample subjects had a non-PTSD
postdisaster diagnosis in the absence of PTSD. Only 4% of subjects without
any predisaster disorder and no PTSD after the bombing had any non-PTSD diagnosis
afterward compared with 48% of those with no predisaster disorder but who
had bombing-related PTSD (Fisher exact P<.001).
Conversely, 74% of the subjects who had preexisting psychopathology and who
were diagnosed as having postdisaster PTSD also had postdisaster comorbidity
vs 30% of those with a predisaster disorder who did not experience PTSD (χ21=14.98; P<.001). Sixty percent
of the bombing survivors had experienced a psychiatric disorder at some time
in their lives either before or after the bombing.
Onset of PTSD was acute. Of 62 subjects with bombing-related PTSD, 47
(76%) reported immediate (same day) onset, another 11 (94% cumulative) in
the first week, 3 more by the end of the first month (98% cumulative), and
only 1 more between 1 and 6 months. Due to the timing of the index interviews,
subjects had little or no time to develop delayed PTSD, defined in the DSM-III-R10 as onset more
than 6 months after the traumatic event. Comparing onset information with
44 PTSD-producing traumas at some other time in their lives (for which there
had been ample opportunity to observe timing of onset well beyond 6 months),
32 (73%) of the cases reportedly began the same day, and another 10 (95% cumulative)
the same week. Two delayed-onset cases (5% of the total PTSD) were reported:
1 beginning between 6 and 12 months, and 1 more than 3 years afterward. Even
though PTSD onset was very acute after the bombing, its course was chronic.
Eighty-nine percent of subjects with bombing-related PTSD reported that they
were still symptomatic during the month before the interview (which was at
least 3 months after the bombing), defining their PTSD as chronic.10 Sixty-three percent of the entire sample, including
those not meeting full PTSD criteria, acknowledged having experienced some
disaster-related PTSD symptoms within the past month.
The effects of PTSD on occupational and social functioning reported
by the subjects suggest the clinical importance of this disorder (Table 3). More than half of subjects with
PTSD alone and the vast majority of those with comorbid PTSD reported that
their PTSD symptoms interfered with their activities; similar numbers in each
group were dissatisfied with their work performance after the disaster. Negative
changes in personal relationships as a result of the bombing were acknowledged
by 75% of subjects with PTSD compared with 27% of those without this diagnosis
Diagnostic comorbidity with PTSD was specifically associated with effects
on relationships with spouses and other household members. Nearly 40% of all
the survivors used medication to cope, including about 25% of those who did
not experience any postdisaster psychiatric disorder, and 73% of those with
comorbid PTSD (Table 3). Only
those persons with PTSD that was complicated by comorbidity were using medication
or alcohol as a coping mechanism. Regardless of diagnostic status, turning
to others for support was a nearly universal response.
Mental health treatment was abundant. Sixty-nine percent of the survivors
received some kind of mental health intervention after the disaster; 40% had
participated in debriefings and 41% had sought professional mental health
treatment, but only 16% had been treated by a psychiatrist. Few individuals
had received mental health intervention from their primary care physician
(5%) or pastor (3%). The highest rates of professional mental health services
utilization were among those with comorbid PTSD (72%). Subjects who had PTSD
but who did not have a postdisaster comorbid disorder did not use mental health
services more often than those who had a comorbid disorder (39% vs 29%, respectively, P=.32; χ21=0.978).
Figure 1 shows the rates of
each of the PTSD symptoms that are arranged by DSM-III-R symptom groups: group B (intrusive reexperience), group C (avoidance
and numbing), and group D (hyperarousal). To be diagnosed as having PTSD,
according to DSM-III-R criteria, a subject must first
be exposed to a traumatic stressor, which the DSM-III-R defines as criterion A, and were exposed to an "event that is outside
the range of usual human experience and that would be markedly distressing
to almost anyone."10 The subject must then
present with at least 1 of the symptoms from the intrusive and reexperience
category (group B), have at least 3 symptoms from the avoidance and numbing
category (group C), and at least 2 of the symptoms from the hyperarousal category
(group D). These symptoms must last for at least 1 month and must be severe
enough to cause subjective distress or functional impairment. PTSD symptoms
were nearly universal: only 7 subjects (4%) reported no bombing-related PTSD
symptoms. The 2 most commonly experienced symptoms were in the hyperarousal
category: difficulty concentrating (78%) and exaggerated startle response
(77%). The 3 least experienced symptoms were in the avoidance and numbing
category: sense of foreshortened future (19%), restricted range of affect
(13%), and psychogenic amnesia (12%).
The vast majority of survivors fulfilled criteria for intrusive reexperience
and hyperarousal categories. Only one third of the total subjects fulfilled
the avoidance and numbing criteria. The avoidance and numbing criteria were
highly specific for the diagnosis of PTSD: 94% of the subjects who had fulfilled
avoidance and numbing criteria met full PTSD criteria related to the bombing.
By DSM-III-R requirement, all subjects meeting criteria
for PTSD fulfilled the avoidance and numbing criteria (100% sensitivity).
Figure 2 shows that the avoidance
and numbing criterion group was significantly associated with predisaster
psychopathology and with postdisaster comorbidity, associations generally
not observed in conjunction with intrusion and hyperarousal symptom groups
alone (when avoidance and numbing criteria were not met). The avoidance and
numbing criterion group was also associated with receiving treatment, whereas
intrusion and hyperarousal in its absence were not. As seen in Figure 2, the avoidance and numbing group (and to a much smaller
extent, hyperarousal) was associated with reports of functional interference.
The avoidance and numbing criterion group was also associated with dissatisfaction
with work performance.
Subjects with postdisaster PTSD reported a mean (SD) of 5.7 (4.2) injuries,
compared with 3.1 (2.4) injuries among others (Wilcoxon z=3.14, P=.002). Those reporting injury or
death of a family member or friend in the bombing had higher rates of PTSD
than others (43% vs 25%; χ21=5.02, P=.03). Variables associated with a non-PTSD disorder after the bombing
included female sex (39% women vs 21% men; χ21=7.71, P=.006) and number of disaster-related injuries (4.6 [4.2]
vs 3.4 [2.7]; t=2.35; P=.02).
Postdisaster major depression was not more prevalent among those who had lost
a friend or relative in the disaster, nor was the number of depressive symptoms
higher in this group. Controlling for the confounding effects of sex on education
and marital status (women having less education and being more often divorced
or separated compared with men), these 2 variables were not associated with
PTSD or other postdisaster psychopathology.
The Oklahoma City bombing provided a rare opportunity to study mental
health effects resulting from a severely traumatic event in an essentially
unselected population. This study documented extensive psychopathology in
a highly exposed sample of direct victims of the blast.
Nearly half the bombing survivors studied had an active postdisaster
psychiatric disorder, and full criteria for PTSD were met by one third of
the survivors. PTSD symptoms were nearly universal, especially symptoms of
intrusive reexperience and hyperarousal.
An explosion in a Norwegian paint factory studied by Weisæth13,14 represented a similar type of disaster,
although it was considerably smaller in scope and magnitude, with far less
mortality and morbidity (6 fatalities, 2 incapacitating injuries, and another
21 minor injuries among 125 survivors). The high-exposure group had a 43%
rate of PTSD at 10 weeks after the bombing, diminishing to 36% by 7 months.
Only a few other disaster studies of nonmilitary populations have reported
higher PTSD rates: 44% after the Buffalo Creek Dam break and floods,15 53% after Australian bushfires,16
54% after an airplane crash landing,17 and
50% to 100% after a plane crash into a shopping mall.18
Differences in research methods, such as use of unstructured interviews and
self-report scales known to be associated with higher estimates of psychopathology,19 unfortunately preclude meaningful comparisons.
Comparison across disasters is possible within the Washington University
research database on several different disaster events studied using uniform
methods. The 34% rate of PTSD after the bombing was the highest of all the
disasters studied to date by this team.4,20- 24
Rates of PTSD in these other studies were 2% following a tornado,22 28% after a mass shooting episode,23
and 29% after a plane crash into a hotel.20
The degrees of both occupational and social impairment associated with
PTSD after the bombing demonstrate the clinical importance of this disorder.
These functional effects of PTSD appeared to be mediated in large part by
its psychiatric comorbidity. Major depression was the most commonly associated
disorder, and most preexisting depression recurred or persisted in the period
after the bombing. No new cases of substance abuse were observed, consistent
with previous findings20- 23,25
pertaining to new postdisaster alcohol use disorders after other events studied
by this team.
This study found several predictors of bombing-related PTSD: degree
of disaster exposure (represented by number of injuries), female sex, preexisting
psychopathology, and secondary exposure through loved ones (injury and death).
Physical injuries and involvement of loved ones may represent specific mechanisms
for generation of psychiatric sequelae of disasters. The predominance in postdisaster
psychopathology in women has been reported in previous disaster studies4,13,14,23,26- 29
and was not unexpected because the disorders classically observed after disasters—depression
and anxiety disorders—are more prevalent among women in the general
population. Preexisting psychopathology has also been identified as a robust
predictor of PTSD by previous studies of this team4,20- 24
and others.13,14,30- 32
The 43% rate of lifetime predisaster psychiatric illness in the Oklahoma sample
does not exceed the expected general population lifetime rates of 48% in a
large population assessed with structured interviews33
and is not significantly higher than rates of preexisting illness in other
disaster sites studied by this disaster research team.20,23,24,34
This study provided important observations on the character and early
course of PTSD following a particularly severe disaster. Symptom onset was
rather immediate—usually the same day—and few other cases developed
after the first month. This rapidity of onset is consistent with other traumatic
events subjects had experienced and with findings of 2 other studies.24,34 In the Weisæth13,14
paint factory explosion study, 114 of 117 symptomatic subjects reported symptom
onset within 5 hours, and the remaining 3 within 32 days.
The relatively uncommon postdisaster avoidance and numbing symptoms
were virtually tantamount to the diagnosis. Avoidance and numbing symptoms
were associated with preexisting and comorbid postdisaster psychopathology,
functional impairment, use of medication and alcohol to cope, and treatment
received—unlike the more prevalent intrusive reexperience and hyperarousal
symptoms only, which did not show these associations. These observations confirm
this team's previously published findings in studies of an earthquake in Northridge,
Calif,24 and a mass murder episode at a cafeteria
in Killeen, Tex.34
Because virtually all the cases of PTSD started acutely after the bombing,
the most efficient plan would be to expedite large-scale efforts to identify
survivors with psychiatric illness as soon as possible. Because most individuals
with any psychiatric disorder had PTSD, focusing on PTSD could identify most
cases for triage to psychiatric care. Shortages of resources encountered in
acute disaster settings make it important to focus attention on those at greatest
risk for PTSD. This study found highest risk among women, individuals with
more direct and indirect disaster exposure (defined by the number of personal
injuries and secondary exposure through loved ones), and subjects with a predisaster
psychiatric history. The data indicate that PTSD may be readily and efficiently
identifiable with truncated assessment for avoidance and numbing criteria
Once PTSD is identified, as suggested by the comorbidity data collected
from the Oklahoma bombing, clinicians would be well advised to continue searching
for other psychopathology, a finding verified elsewhere in the literature.20,24,34 These data suggest
that subjects with comorbidity will be significantly more impaired by their
psychopathology. The chronicity of PTSD identified in this study (with 9 of
10 cases still symptomatic at interview an average of 6 months after the disaster,
criteria of at least 3 months' duration for chronicity) indicates that availability
of ongoing treatment of PTSD is essential.
In the absence of avoidance and numbing, the nearly ubiquitous intrusive
reexperience and hyperarousal symptoms were associated with little to no functional
impairment or psychiatric comorbidity. This suggests different management
strategies for these normative yet distressing symptoms from the professional
assessment and intervention generally advised for avoidance and numbing responses.
The nonpathological nature of intrusion and hyperarousal symptoms uncomplicated
by avoidance and numbing suggests that after major psychiatric illness is
ruled out, these symptoms may be managed by nonphysician mental health professionals
with nonmedical interventions such as public and workplace debriefings. The
therapeutic tools for these uncomplicted intrusion and hyperarousal syndromes
will be education, general support, and reassurance that the symptoms are
normal and not evidence of impending psychiatric illness.
Because the study sample was slightly skewed toward proximity to the
blast, the findings may reflect a mildly elevated estimate of psychiatric
impact of the disaster population as a whole. Two major strengths of the study
are its random sampling that maximized the general representativeness of the
registry population and the structured research interview that generated psychiatric
diagnoses. The cross-sectional nature of assessment at a single time, however,
limits findings to description of only the first 6 months after the event.
Further study is needed to chart the course of postdisaster psychiatric disorders
over a longer period , to observe for the development of delayed cases of
PTSD, and to identify predictors of chronicity vs recovery, including potential
effects of treatment. Naturalistic observation studies such as this one suffer
from confounding of outcomes with seeking treatment whose benefits cannot
be assessed under the available design. Additional study with uniformly applied
methods across various disaster events will allow merging the data to generate
statistical power for untangling disaster-specific confounders, modeling complex
hypotheses, and generalizing across events.36