Linguistic Markers of Chronic PTSD in World Trade Center Rescue and Recovery Workers: A Computer-Based Natural Language Processing Study

Details

• Personal Author:
  Braide B ; Corcoran C ; Feder A ; Monti E ; Perez-Rodriguez MM ; Pietrzak RH ; Schafler T ; Schreiber Z ; Zonshayn D
• Description:
  Background: An emerging literature suggests the value of automated natural speech analysis in characterizing posttraumatic stress disorder (PTSD). Automated analyses of speech transcripts aim to identify linguistic patterns differentiating individuals with PTSD from trauma-exposed individuals who did not develop the disorder, with the ultimate goal of improving diagnostic characterization and treatment interventions for this chronic and disabling disorder. Studies to date have generally included heterogeneous samples of individuals with PTSD stemming from a range of trauma exposures and with varying degrees of chronicity. In the present study, we applied automated language processing methods to participant responses during open-ended interviews with World Trade Center (WTC) rescue and recovery workers, who were asked to describe their experience during the 9/11 terrorist attacks and their aftermath. Novel aspects of this study include a unique sample of WTC responders all exposed to a single, shared, and well-documented trauma, and the inclusion of a comparison group of highly resilient WTC responders. Methods: WTC responders recruited from the WTC Health Program Responder Cohort (group-matched by age, race, marital status, education, and word count) completed in-person diagnostic interviews, including the Structured Clinical Interview for DSM-5 (SCID-5) and the Clinician-Administered PTSD Scale for DSM-5 (CAPS-5), administered by trained clinicians. Participants were also recorded during open-ended interviews while responding to the following prompt: "Please describe your experience of the 9/11 attacks and your experience during your involvement in the Work Trade Center recovery work". Interviews were designed to be non-directive; interviewers could further prompt participants by stating, "Is there anything more you'd like to say about your WTC recovery experiences?". Participants (n = 10 per group) were classified into "highly resilient" [high WTC-related exposure severity, no lifetime psychiatric disorders, mean (SD) past-month CAPS-5 = 1.4 (1.2) and "chronic PTSD" [met past-month DSM-5 criteria for WTC-related PTSD; mean (SD) past-month CAPS-5 = 30.7 (10.7)]. Analyses of interview transcripts were conducted using Language Inquiry and Word Count (LIWC) software. Bivariate analyses were conducted to identify linguistic categories that differed between groups at the p < 0.20 level. These categories were then entered into a multivariate analysis of variance to identify between-group differences in linguistic categories used in interview responses. We then conducted LASSO regression analyses to examine associations between individual PTSD symptoms and linguistic categories. Analyses were adjusted for full-scale IQ and responder type (police vs. non-traditional [e.g., construction worker]). Results: Bivariate analyses revealed 19 linguistic categories that differed between groups. In a multivariate model, 3 categories reflecting biological processes (i.e., body words [e.g., cheek, hands, spit], Cohen d = 2.0); perceptual processes (i.e., feel words [e.g., feels, touch], d = 1.7); and drives (i.e., affiliation words [e.g., ally, friend, social], d = 1.3) showed large magnitude between-group differences, with WTC responders with PTSD using a greater proportion of feel and body words, and a lower proportion of affiliation words than highly resilient responders. Sleep disturbance was associated with a greater proportion of body (β = 0.67) words; intrusive thoughts about the WTC disaster with a greater proportion of feel (β = 0.61) words; and feeling distant or cut off from others with a lower proportion of affiliation words (β = -0.53). Conclusions: This study identified potential linguistic markers of chronic PTSD in a sample of individuals exposed to a common traumatic event. Findings point to specific differences in speech patterns in responders with WTC-related chronic PTSD, compared to highly resilient WTC responders. Further, preliminary evidence indicates a close alignment between certain linguistic markers and specific PTSD symptom dimensions, suggesting that linguistic markers might map to specific underlying biological abnormalities. Further study in larger samples, in combination with biomarker studies (e.g., neural circuitry function), has the potential to deepen our understanding of this heterogeneous disorder and enhance personalized treatment interventions for individuals with chronic PTSD. [Description provided by NIOSH]
• Subjects:
  Cohort Studies Mental Health Nervous System Occupational Health Safety September 11 Terrorist Attacks
• Keywords:
  Author Keywords: PTSD Automated Natural Speech Analysis Cohort Studies Disaster Response Linguistic Markers Mental Health Post-traumatic Stress Disorder PTSD Resilience World Trade Center World Trade Center Responders WTC

  More +
• ISSN:
  0893-133X
• Document Type:
  Text
• Funding:
  Cooperative Agreement
• Genre:
  Abstract or Summary
• Place as Subject:
  New York ; OSHA Region 2
• CIO:
  National Institute for Occupational Safety and Health (NIOSH)
• Topic:
  Workplace Safety & Health
• Location:
  North America
• Pages in Document:
  50
• Volume:
  46
• NIOSHTIC Number:
  nn:20069557
• Citation:
  Neuropsychopharmacology 2021 Dec; 46(Suppl 1):50
• Federal Fiscal Year:
  2022
• Performing Organization:
  Icahn School of Medicine at Mount Sinai, New York
• Peer Reviewed:
  False
• Start Date:
  20170701
• Source Full Name:
  Neuropsychopharmacology. Selected Abstracts from the American College of Neuropsychopharmacology (ACNP) 60th Annual Meeting, December 5-8, 2021, San Juan, Puerto Rico
• Supplement:
  1
• End Date:
  20210630
• Collection(s):
  National Institute for Occupational Safety and Health
• Main Document Checksum:
  urn:sha-512:63439c7105b317d191aa51d85964b0687a10479d404070ae7677b20bc3e72f7b85ea0fba85963ac453defa3c79860fb3293c5117a3ae4e72f0d9d46699176612
• Download URL:
  https://stacks.cdc.gov/view/cdc/207971/cdc_207971_DS1.pdf
• File Type:
  [PDF - 89.12 KB ]