Implementation of Evidence-Based Assessment, Treatment, and Research Programs Following the World Trade Center Disaster on September 11, 2001

Clinician-administered measures

Domain

Clinician-administered PTSD scale (CAPS; Blake et al., 1995)

PTSD

Standard clinical interview for DSM-IV (SCID; First et al., 1997)

Depression

Panic disorder

Generalized anxiety disorder

Acute stress disorder (peri-traumatic dissociation)

WTC exposure questionnaire (developed by Weill Cornell trauma experts)

Individual’s exposure to the WTC attacks

Trauma history questionnaire (Green, 1993)

Lifetime trauma history

Alcohol use disorders identification test (AUDIT; Babor, Higgins-Biddle, Saunders, & Monteiro, 2001)

Alcohol use

Self-report measures

Domain

PTSD checklist (PCL; Weathers et al., 1993)

PTSD

Beck depression inventory-II (BDI-II; Beck et al., 1996)

Depression

State-trait anger expression inventory-2 (STAXI-2; Spielberger, 1999)

Anger

Brief symptom inventory (BSI; Derogatis & Spencer, 1982)

Psychological and physical symptoms

Sheehan disability scale (SDS; Sheehan, 1983)

Social, occupational, and family functioning

Challenges in Implementing a Large-Scale Screening Program

One major challenge of a workplace interview format was the possibility that individuals might be motivated to minimize or deny symptoms. This factor weighed into our implementation strategy in multiple ways. For example, although we strongly considered including a formal substance use assessment in our initial interviews or self-report materials, we decided at the outset of our screening process to ask these questions indirectly as we were concerned that including these questions could cause participants to become guarded due to fear of negative work-related consequences and potentially threaten the validity of the remainder of the assessment because of social desirability bias as substance abuse was cause for termination (e.g., Klassen, Hornstra, & Anderson, 1975; Paulhus, 1991; Phillips & Clancy, 1970, 1972) or response set bias. In the third year of the screening program, once rapport had been strongly established and our presence at the workplace clinic had become routine for the employees, we added a structured assessment of alcohol intake to our clinical interviews, using the Alcohol Use Disorders Identification Test (Babor et al., 2001).

The issue of confidentiality in a workplace interview also had to be carefully considered to allow participants adequate protection to discuss their symptoms with our clinicians. To ensure this, we arranged to provide only aggregate data to the utility company in the form of percentage of PTSD and depression in participants overall; no individual clinical information was released and all interview and self-report data was stored in our office. In the rare case when an individual reported intent to harm self or others, the incident was managed as a clinical issue between patient and therapist, following customary practices for emergencies.

The work conditions required a high degree of flexibility. Because the company was unable to give an estimate of daily expected participants (varying from 2 to 12) ahead of time, our staffing evolved from one clinician on site to two, and later to a primary clinician with a second clinician available if needed.

Our project staffing included psychologists as well as administrative staff to schedule participants, create and maintain the database, input and check the data, and photocopy the screening materials. Additional postdoctoral fellows and junior faculty members were hired and intensively trained in the CAPS, SCID, and other screening instruments. After first observing senior faculty conducting several assessments, these junior colleagues were then observed interviewing a participant and received feedback and supervision from senior faculty. Interrater reliability was continuously monitored with senior faculty observing junior faculty interviews on a random basis.

In the third year of the program, we scaled back the assessments to screen the entire group via self-report, and only interviewed a selected high-risk group and random comparison group (for a detailed description of this methodology, see Cukor, Wyka, Mello et al., 2011). This decision was made as a strategy to continue to assess those experiencing symptoms while reducing the burden on non-symptomatic participants of an additional hour-long interview for the third consecutive year. We defined high-risk individuals as those who were present at the WTC during the attacks, had a previous psychiatric history, or met criteria for PTSD or subthreshold PTSD at either of the previous assessments. Those who were not designated as high risk completed the self-report measures and then met briefly with a psychologist. If these individuals endorsed distress on self-report measures, the full assessment interview was conducted; otherwise the assessment was stopped. See Table 19.2 describing the evolution of our assessment process.

Table 19.2

WTC screening program interview format

Program year	Interview structure	PTSD prevalence and comorbid psychopathology
Year 1–2	• In-person full structured interview^a with all participants (45–60 min, conducted by doctoral level psychologists) • Self-report measures^b administered to all participants	Time frame: July 2002–April 2004 PTSD prevalence (CAPS-based): – Full PTSD—8 % – Subthreshold PTSD −9.3 % PTSD prevalence (PCL-based^c): – Full PTSD—9.5 % – Subthreshold PTSD −9.3 % Comorbid psychopathology – MDD—6 % – GAD—3.5 % – Panic disorder −2.5 %
Year 3–4	• In-person full structured interview^a with high-risk group and random comparison group (45–60 min, conducted by doctoral level psychologists), including questions on alcohol use^d • Self-report measures^b administered to all participants and brief meeting with a psychologist – Follow-up full interview if distress reported on self-report measures	Time frame: May 2004–Dec 2005 PTSD prevalence (CAPS-based, high-risk, and control comparison subsample): – Full PTSD—8.4 % – Subthreshold PTSD −8.9 % PTSD Prevalence (PCL-based): – Full PTSD—4.8 % – Subthreshold PTSD −3.6 % Comorbid Psychopathology – MDD—3.2 % – GAD—1.8 % – Panic disorder −2.9 %
Year 5–6	Same as above	Time frame: Jan 2007–Dec 2008 PTSD prevalence (CAPS-based, high-risk, and control comparison subsample): – Full PTSD—5.8 % – Subthreshold PTSD −7.7 % PTSD Prevalence (PCL-based): – Full PTSD—2.4 % – Subthreshold PTSD −1.8 %

^aClinician-administered measures: CAPS, SCID, Trauma History Questionnaire, WTC Exposure Questionnaire

^bSelf-report measures: PCL, BDI, BSI, STAXI-2, Sheehan

^cPCL cluster scoring method

^dAUDIT

This method provided the opportunity to meet our clinical and public health goals. At the utility company’s request, we conducted screenings in this manner for 3 additional years, for a total of 7 years of assessment. Our access to this population of non-rescue disaster workers across nearly a decade has provided unique insights into the longitudinal mental health sequelae of the most significant terror attacks in the United States.

Research Findings on WTC Disaster Workers

The program we developed to assess WTC-related PTSD and other psychiatric symptomatology provided a wealth of data demonstrating the short- and long-term mental health consequences of terrorism.

Our screening program found that a substantial number of participants met criteria for comorbid psychopathology between one and 2 years following the trauma (Cukor, Wyka, Jayasinghe, et al. 2011). Within a sample of 2,960 individuals, 8 % met criteria for PTSD, 9.3 % for subthreshold PTSD,¹ 6 % reported symptoms consistent with MDD, 3.5 % met criteria for generalized anxiety disorder (GAD), and 2.5 % reported symptoms consistent with panic disorder. Thirty-nine percent of those with full PTSD reported comorbid current MDD, 14 % comorbid current GAD, and 10.5 % comorbid current panic disorder.

Several variables significantly predicted probable PTSD, including prior life trauma, prior psychopathology, and level of trauma exposure. Notably, those who perceived their lives to be in danger on the day of the attacks were twice as likely to endorse symptoms of full PTSD as the overall sample, making this the single best predictor of probable PTSD. Extent of exposure predicted 69 % of PTSD cases in the most vulnerable group (i.e., those with a psychiatric and trauma history) as well as 89 % of PTSD cases in the low-risk group (i.e., those with no such history).

Four years following the WTC attacks, probable PTSD was 4.8 % and subthreshold PTSD was 3.6 % using a PCL-based estimate for those individuals with complete data (n = 2,626) (Cukor, Wyka, Mello et al., 2011). Probable PTSD was 2.4 % and subthreshold PTSD was 1.8 % 6 years post-9/11 (n = 1,983). Several variables emerged as significant predictors of PTSD diagnosis at the final assessment. The presence of MDD at the initial assessment (OR = 2.80, 95 % CI [1.17, 6.71]), a prior history of trauma (OR = 2.27, 95 % CI [1.06, 4.85]), and extent of occupational exposure (OR = 1.31, 95 % CI [1.13, 1.51]) were the strongest risk factors.

This large sample allowed us to determine whether individuals with subthreshold PTSD represented a distinct group that should be the subject of research and treatment (Cukor, Wyka, Jayasinghe, & Difede, 2010). The subthreshold PTSD group displayed a moderate level of impairment between the non-PTSD group and the full PTSD group. While the largest percentage of individuals (58.8 %) had subthreshold PTSD at their first assessment and then no longer met criteria for subthreshold or full PTSD at subsequent yearly assessments, 19 % of participants had symptoms that progressed to a diagnosis of full PTSD over the next 2 years. This long-standing and considerable impairment makes a strong case for the clinical significance of subthreshold PTSD. Furthermore, 26 % of those with subthreshold PTSD had comorbid MDD, 11.4 % had GAD, and 7 % had comorbid panic disorder.

Despite these concerning findings, the majority of individuals demonstrated great resilience. Approximately 70 % of the overall sample never met criteria for full or subthreshold PTSD according to the PCL. We found a substantial reduction in PTSD symptomatology, so that improvement over time was the common course in our sample. Presumed rates of PTSD based on self-reports (using the PCL) at 6 years post-9/11 were nearly a quarter of the rates at the initial assessment, and rates had decreased by more than 50 % by 6 years for the subsample when examined using the CAPS. However, 10.6 % and 13.9 % of the subsample continued to meet criteria for full PTSD or subthreshold PTSD, respectively, on the CAPS at the final assessment 6 years post-9/11.

Treatment Program

It was evident at the inception of our assessment program that some individuals would report significant symptomatology and need treatment. Our overarching goal was to provide treatments with the strongest research demonstrating their efficacy. At the time of the September 11th terrorist attacks, there was limited agreement on the best forms of treatment for PTSD and little emphasis in the field of psychiatry on the need for evidence-based treatment in general (U.S. Department of Health and Human Services, 1999). The first expert consensus guidelines for PTSD treatment had only recently been published (Foa, Davidson, & Frances, 1999), recommending CBT with exposure therapy as the first line of treatment. Additionally, sertraline was the only medication with FDA approval to treat PTSD. As research scientists and clinicians, therefore, when asked to develop a free, formal treatment program, we sought to develop a model offering treatment with the most evidence of its effectiveness.

To increase confidentiality and the likelihood of treatment acceptance, utility company employees were offered weekly sessions in our private offices at New York-Presbyterian Hospital and were allowed to come to the appointment on work time. In an additional effort to increase confidentiality, our staff communicated all names and appointment times to a single company liaison who emailed the supervisor of each employee stating only that they had an unspecified “mandatory medical appointment” at our hospital at the designated time, a standard practice in the company for all medical appointments. Supervisors at the utility company were advised by their leadership to respect these medical appointments as well as the confidentiality of employees.

While these steps likely minimized dropout, it was initially noted that approximately half of participants did not attend their first appointment even after agreeing to come in for treatment. Speculating that a psychotherapy appointment with an unknown mental health provider might be anxiety-provoking, we altered our approach so that the psychologist performing the assessment on location could become that individual’s treatment provider. In this way, the rapport that developed during the screening could be carried into future treatment, thereby increasing chances of first session attendance.

Treatment consisted of evidence-based practices, using exposure therapy as the first-line psychological treatment and sertraline as the first-line pharmacologic agent. All psychologists were trained to employ evidence-based practices to ensure patients would benefit from techniques that had been shown to work in research trials.

Given the large volume of traumatized patients seen, we developed and implemented a secondary stress prevention program to address the potential for vicarious traumatization among our program staff and the larger hospital staff treating WTC survivors. This secondary stress prevention program contained three components: (1) individual supervision, (2) group supervision, and (3) an additional wellness component in the form of yoga classes for our staff and the wider hospital community. Individual supervision was conducted strategically, with our entire group organized into teams for weekly supervision, with a senior faculty member supervising a junior faculty member who in turn supervised a postdoctoral fellow. All senior faculty were directly supervised weekly by the program director. In addition, we offered all clinicians weekly group process supervision, facilitated by the program director. Outside trauma specialists presented at a biweekly seminar series on their respective research areas to provide further discussion and support to our providers. A weekly yoga program was also developed and made available free of charge for faculty and staff across departments at Weill Cornell Medical College and New York-Presbyterian Hospital working with patients affected by the WTC attacks. The program was partially funded by a grant from the Greater New York Hospital Foundation and directed by JoAnn Difede, Ph.D., and yoga expert Beryl Bender Birch.

Pioneering a Clinical Research Program with Virtual Reality Treatment

Knowing that substantial numbers of those expected to be diagnosed with PTSD might not improve using the emerging evidence-based treatments, we sought to develop an effective alternative. While empirical evidence at the time was limited, preliminary evidence suggested that even after imaginal exposure therapy, approximately one-third of patients failed to achieve remission from PTSD (Foa et al., 1999). Effective imaginal exposure requires patients to be willing and able to describe their traumatic experience to their therapist repeatedly, in the present tense, and to actively engage their traumatic memories. Avoidance is a hallmark symptom of PTSD, leaving some patients refusing to discuss their trauma entirely and others exhibiting psychic numbing, unable to engage emotionally with their memories, which predicts poor treatment response (Jaycox, Foa, & Morral, 1998).

It is widely theorized that emotional engagement or fear activation plays a critical role in exposure therapy. Foa and Kozak (1986) proposed that in order for a reduction in fear to occur, fear relevant information associated with the patient’s memory for the traumatic event (i.e., the fear structure) must be accessed and activated through emotional engagement. After the fear structure is aroused through emotional engagement, new or corrective information is incorporated into the patient’s memory. These authors suggest that repeated engagement with the feared stimulus in a safe environment is necessary for the fear structures to change, thereby allowing long-term habituation to take place.

The few studies that had addressed the question of treatment failures had concluded that failure to engage emotionally predicts a poor treatment outcome. One of the few studies to examine treatment variables that mediate outcome investigated the impact of the variables emotional engagement and habituation on successful outcome of exposure therapy for chronic PTSD in female assault victims (Jaycox et al., 1998). Results showed that although all participants made treatment gains, those with high emotional engagement in the treatment and habituation to emotion-eliciting stimuli were eight times more likely to meet stringent criteria for good end-state functioning (e.g., a 50 % reduction in PTSD symptom scores and normal scores on measures of depression and anxiety).

We reasoned that virtual reality technology might provide a tool to facilitate high emotional engagement. At the time of the WTC attack, multiple studies had documented that virtual reality exposure therapy was an effective treatment for anxiety disorders other than PTSD (Carlin, Hoffman, & Weghorst, 1997; Emmelkamp, Bruynzeel, Drost, & van der Mast, 2001; Rothbaum et al., 1995, 1999; Rothbaum & Hodges, 1999; Rothbaum, Hodges, Ready, Graap, & Alarcon, 2001; Rothbaum, Hodges, Watson, Kessler, & Opdyke, 1996). Similarly to PTSD, patients with specific phobias avoid the feared stimulus but must confront it to get well. Virtual reality environments afford opportunities not only to capitalize on the patient’s imaginative and memory capacities but also to augment them with visual, auditory, and even haptic computer-generated experiences. Because patients’ fear-relevant cues for a specific trauma include not just thoughts and feelings but also sensory stimuli, such as the sights, sounds, and tactile sensations associated with the traumatic memory, we hypothesized that the virtual environment’s ability to evoke these sensory stimuli would facilitate emotional engagement.

Development of the Virtual World Trade Center Simulation

The attack on the World Trade Center was particularly well suited to the development of a virtual scenario because those who witnessed it experienced numerous shared sensory elements (for example, survivors typically saw either or both planes hit the Twin Towers, were enveloped in the dust cloud after the collapses, and heard screams and cries from others). Virtual reality offered the opportunity to approach feared memories in a multisensory environment, rather than solely verbally, and to create a systematic, graded exposure to increasingly distressing aspects of the traumatic experience. Thus, those reluctant or unable to engage in their traumatic memories could be presented with an external, evocative environment that would allow them to process their trauma therapeutically at their own pace.

To develop the WTC virtual reality scenario, we collaborated with colleagues at the University of Washington Human Interface Technology Laboratory. As part of the software development process, 45 disaster workers who had experienced significant trauma exposure during the WTC attacks were interviewed to incorporate their diverse experiences into the creation of the virtual environment.

The initial WTC virtual world shows the Twin Towers from a distance with a clear, sunny sky as it appeared on September 11, 2001 prior to the attacks. Graded exposure to the WTC virtual scenario involves systematically progressing through multiple stages: (1) the patient sees a jet fly past the Twin Towers without crashing and hears normal New York City street noise; (2) a jet crashes into the first tower, but there is no explosion or sound effects; (3) visuals of the explosion are added; and (4) the explosion sound effects are added. Next, (5) the patient sees the first tower with fire and smoke escaping from the location where the jet hit the building but does not hear screaming sound effects; (6) screaming is added; and (7) people begin jumping from the first tower. In the final stages, (8) the second jet crashes into the second tower with the explosion and screaming sound effects; (9) the second tower collapses and creates a dust cloud; and (10) the first tower collapses and produces a dust cloud. The concluding step is when the patient views the full sequence of events in succession (11) (Fig. 19.1 ).