The September 11, 2001 (9/11), attacks remain unprecedented among terrorist events in the United States in terms of human casualties, property destruction, and societal effects. The consequences of the attacks reverberated across the nation, leading to modifications in infrastructures and systems related to emergency response, civil defense, disaster management, transportation, and commerce. Graphic and unrelenting media coverage of the attacks depicted dramatic scenes of human harm and property damage. The attacks and media coverage of them generated concern about the effects of media contact, as well as the physical and emotional outcomes for those most highly affected and for the general public. A sizeable research literature has examined the relationship between contact with media coverage of mass trauma events and various psychological outcomes, especially posttraumatic stress outcomes – posttraumatic stress disorder (PTSD) and posttraumatic stress symptoms or reactions.Reference Houston1-Reference Pfefferbaum, Nitiéma and Newman3 In particular, media coverage of the 9/11 attacks in adults has been widely studied.Reference Houston1,Reference Pfefferbaum, Newman and Nelson2,Reference Ahern, Galea and Resnick4-Reference Rubin, Haridakis and Hullman20 Media research has also explored more recent terrorist events.Reference Dougall, Hayward and Baum21-Reference Monfort and Afzali25 For example, a series of reports on a survey of the Boston Marathon bombingReference Garfin, Holman and Silver22-Reference Holman, Garfin and Silver24 with a follow-up assessmentReference Holman, Garfin, Lubens and Silver23 examined media effects on acute stressReference Garfin, Holman and Silver22-Reference Holman, Garfin and Silver24 and posttraumatic stress.Reference Holman, Garfin, Lubens and Silver23
Most studies of the 9/11 attacks have used samples of predominantly trauma-unexposed individuals drawn from the general community in the New York City (NYC) areaReference Ahern, Galea and Resnick4-Reference Bernstein, Ahern and Tracy7,Reference Schlenger, Caddell and Ebert16 and from distant communitiesReference Cho, Boyle and Keum10 or nationwide,Reference Cardeña, Dennis, Winkel and Skitka8,Reference Holman, Silver and Poulin14-Reference Silver, Holman and McIntosh19 using self-report symptom surveys rather than studying samples of directly exposed survivors using diagnostic interviews. Studies of more recent terrorist incidents have also assessed samples drawn from general populations in the disaster or distant communities and/or from national samples using self-report symptom surveys.Reference Dougall, Hayward and Baum21-Reference Monfort and Afzali25 Noteworthy exceptions to the use of general community samples are 9/11 studies of US Pentagon employees, some of whom were onsite at the time of the Pentagon attack,Reference Grieger, Waldrep, Lovasz and Ursano13 and of undergraduate college students who, on the day of the attacks, evacuated the university they attended in Ohio when United Air 93 circled overhead prior to crashing in Pennsylvania.Reference Cardenas, Williams and Wilson9 An exception to the use of self-report symptom measures in 9/11 disaster studies, Gershoff and colleaguesReference Gershoff, Aber, Ware and Kotler12 administered diagnostic interviews to assess the mothers of children in an NYC school-based sample. Although the authors reported “direct exposure” in many of the mothers,Reference Gershoff, Aber, Ware and Kotler12 it is unclear how many actually met the PTSD exposure criterion.
THE CURRENT STUDY
Differences in sampling and assessment are of more than academic interest and may have important clinical implications. Relatively few studies have assessed directly exposed samples of survivors. Although early 9/11 studies suggested that media contact may lead to PTSD, the Diagnostic and Statistical Manual (DSM), fifth edition (DSM-5),26 disallowed media contact as a qualifying exposure for the diagnosis. Unfortunately, few studies have distinguished psychopathological and normative outcomes in exposed and unexposed samples. In light of these methodological concerns, the goal of the current study was to extend the extant literature by exploring both psychopathology and normative distress associated with media contact in a sample recruited to include large numbers of 9/11 trauma-exposed individuals and to clarify distinctions between these outcomes in exposed and unexposed groups. Specifically, this study of adult employees of NYC area businesses after the 9/11 World Trade Center (WTC) attacks examined the associations between media contact before and after the disaster and both DSM, fourth edition (DSM-IV),27 PTSD and the 3 PTSD symptom groups (re-experiencing, avoidance and numbing, and hyperarousal) in trauma-exposed survivors, as well as the association between media contact before and after the disaster and the number of posttraumatic stress symptoms in both the exposed and unexposed participants. DSM-IV27 defined PTSD as a disorder that occurs after exposure to a traumatic event with sufficient numbers of new symptoms in 3 symptom groups – groups B (re-experiencing), C (avoidance and numbing), and D (hyperarousal) – lasting for more than 1 month and producing clinically significant distress or functional impairment. The current study also explored how disturbed people were by graphic 9/11 media images. The results are discussed in the context of the extant terrorism media research.
METHODS
The current analysis used data collected from a volunteer sample of 379 employees recruited from 8 businesses affected by the 9/11 attacks on the WTC in NYC.Reference North, Pollio and Smith28 Information about the study was distributed by participating businesses to their employees who contacted the research team to participate in the research. The original sample included 176 participants from 3 agencies located in the WTC towers and 203 participants from 5 other agencies in the geographical area. Interviews were conducted a mean of 35 months (range, 27–52 months) after the attacks. The institutional review boards of the collaborating academic institutions for this study approved the project. All participants provided written informed consent.
Assessment Interviews and Questionnaires
Structured interviews used in this study included the Diagnostic Interview Schedule for DSM-IV,Reference Robins, Cottler and Compton29 which has demonstrated reliability for psychiatric disorders, and the Disaster Supplement.Reference North, Pfefferbaum, Robins and Smith30 These interviews were administered by mental health professionals who received formal standardized training on these instruments by authors of the instruments. Both of these instruments have been used successfully in disaster research for decades. Pre-disaster and post-disaster psychiatric disorders were assessed by obtaining onset and recency information for psychiatric disorders relative to the date of the 9/11 attacks. A DSM-IV27 diagnosis of disaster-related PTSD at any time after the disaster (ie, post-disaster prevalence) required meeting criteria for exposure to 9/11 trauma, as well as meeting symptom criteria and duration and distress or impairment requirements. Detailed information was collected in the administered interviews about participants’ disaster experiences, including trauma exposures (direct; witnessed in person; and/or indirect through close family members, friends, or coworkers) according to DSM-IV27 criteria for PTSD.
Respondents also completed the self-report Disaster Supplement Questionnaire (DSQ),Reference North and Pfefferbaum31 an adaptation of the Disaster Supplement providing more detail of respondents’ disaster experience and its effects on them. This study included new questions that were informed by a large focus group study.Reference North, Pfefferbaum and Hong32,Reference North, Pfefferbaum and Hong33 The dataset analyzed for this report comprised 254 participants (67% of the sample) who completed the DSQ. The only relevant difference between those who did and did not complete the DSQ was that DSQ completion was associated with a higher prevalence of disaster-related PTSD (22% vs 11%; χ2 = 7.55, df = 1, P = 0.006). The DSQ obtained information about contact with news coverage before and after the 9/11 attacks and reactions to the coverage. Specifically, respondents were asked the number of hours per day they spent “obtaining news from all sources” in the month before the 9/11 attacks and in the week after the attacks. They categorized how “bothered” they were by graphic media images as “not at all,” “somewhat,” “moderately,” or “extremely,” and this variable was dichotomized for analysis as “moderately/extremely” versus “somewhat/not at all.”
Data Analysis
The public sector R statistical package (version 3.5.2) was used for data analysis (R Foundation for Statistical Computing, Vienna, 2018). Descriptive findings are presented with raw numbers and proportions for categorical variables and with means and standard deviations (SD) and median values for numerical variables. Categorical variables were compared using 2-sided χ2 analyses, substituting Fisher’s exact texts when expected cell sizes were < 5. Continuous variables were predicted from categorical variables using t-tests, substituting Satterthwaite tests for pooled analysis in cases of unequal variance as determined by Levene’s test. Paired t-tests were used to compare 2-paired continuous variables. Associations between 2 continuous variables were assessed with Pearson’s correlations. Because a series of 6 multivariate models was tested (Table 1), a Bonferroni correction was applied to this series of multiple comparisons, for which the α level of statistical significance was reduced to < 0.008. Otherwise, the statistical significance was set at α ≤ 0.050. Because unexposed individuals cannot, by definition, develop PTSD in relation to a traumatic event, the diagnosis of PTSD and the PTSD symptom groups were examined only in exposed participants, although symptom counts were examined in both exposure groups.
TABLE 1 Associations of Posttraumatic Stress Symptoms With Media Variablesa
a Statistical significance threshold is P < 0.008 because α level is corrected for a series of 6 multiple comparisons.
b Moderately/extremely.
RESULTS
At the time of the assessment, the sample (N = 254) for the current analysis was almost one-half (45%, n = 115) male with a median age of 44 years, more than two-thirds (69%, n = 174) white, two-thirds (67%, n = 168) college-educated, and approximately one-half (51%, n = 129) currently married. Approximately 40% (41%, n = 105) of the sample had a PTSD-qualifying 9/11 disaster trauma exposure according to DSM-IV27 criteria. Specifically, more than one-fifth (22%, n = 57) were directly exposed to 9/11 trauma, including 36 individuals who were in the WTC towers and another 9 who were immediately outside of the towers during the attacks. Among those directly exposed to 9/11 trauma in the sample analyzed for this report, disaster-related PTSD developed in 30% (n = 17). The mean (SD) number of hours per day of news media contact was higher in the first week after the attacks compared with the month before the attacks (mean = 5.7, SD = 4.8 vs mean = 2.1, SD = 2.1; t = 12.67, df = 225, P < 0.001). More than one-half (56%) of the sample reported that they were either “extremely” (33%) or “moderately” (23%) bothered by graphic media images of the 9/11 attacks; only 18% reported that they were “not at all” bothered by graphic 9/11 media images.
No demographic variables were associated with the amount of contact with news coverage in the month before or the first week after the disaster or with feeling bothered by graphic 9/11 media images. Neither exposure to 9/11 trauma nor working in the WTC towers was associated with the mean number of hours per day of news media contact in the month before or the first week after the attacks or with feeling bothered by graphic 9/11 media images. Injury during the attacks was associated with a higher number of media contact hours per day only in the first week after the attacks (t = 2.49, df = 238, P = 0.014) and was not associated with feeling bothered by graphic 9/11 media images. The number of hours per day of news media contact in the month before the attacks or in the week after the attacks was not associated with feeling bothered by graphic media images in the full sample in this analysis or in either exposure group. Because of this general lack of associations of media variables with demographic and trauma exposure variables, subsequent comparisons of media variables were conducted using a simple bivariate analysis without adjusting for these other variables in multivariate models.
Table 1 presents associations of media variables with posttraumatic stress symptoms in the 9/11 trauma-exposed and trauma-unexposed groups. Separate bivariate comparisons were made for each of the posttraumatic symptom variables, rather than including them together in multivariate models, because variance inflation testing identified excessive multicollinearity among these variables within multivariate models. In the bivariate comparisons, hours of media contact before and after 9/11 were not associated with the total number of posttraumatic stress symptoms in either exposure group. In both the 9/11 trauma-exposed and trauma-unexposed groups, a higher number of hours per day of media contact in the first week after the attacks was associated with a higher number of re-experiencing but not avoidance and numbing or hyperarousal symptoms. Pre-disaster media contact was not associated with any posttraumatic stress symptom variables. Feeling “moderately/extremely” bothered by graphic 9/11 media images was associated with a higher number of re-experiencing symptoms in both exposure groups and also with total posttraumatic stress symptoms and hyperarousal symptoms in the exposed group only.
DISCUSSION
Media news consumption in this sample of employees of NYC area businesses increased after the 9/11 attacks. The number of hours per day of post-disaster media contact was associated with re-experiencing symptoms in both exposure groups but was not associated with 9/11 trauma exposures or with feeling bothered by graphic media images. Feeling bothered by graphic 9/11 media images was associated with the number of re-experiencing symptoms in both exposure groups and with the number of total posttraumatic stress symptoms and number of hyperarousal symptoms in the trauma-exposed group.
Associations Between Media Contact and Posttraumatic Stress Variables
The use of diagnostic assessment in a sample with many individuals who were directly exposed to the 9/11 attacks offered the opportunity to extend the extant media effects research to examine psychopathological (PTSD) and normative (subdiagnostic distress) responses in both the exposed and unexposed groups.
PTSD Diagnosis Variables
The current study found no association of media contact in the first week after the attacks with PTSD or with any of the PTSD symptom groups in exposed participants. The use of dichotomous PTSD and symptom group variables in this analysis and restricting the analysis to the exposed group may have limited statistical power for these analyses.
Posttraumatic Stress Symptoms
The analysis examined the association between news media contact and posttraumatic stress symptoms in the unexposed group, as well as the exposed group. In individuals without trauma exposures or PTSD, these symptoms do not represent components of the disorder (PTSD) but rather indicators of subdiagnostic distress. Statistical power was amplified in this separate examination of stronger numeric posttraumatic stress symptom variables in place of dichotomous diagnosis and symptom group variables. The analysis of symptom count variables also permitted the analysis of the full sample, including not only exposed participants, but also unexposed participants for whom an examination of PTSD was not justifiable. Media contact was not associated with the number of total posttraumatic stress symptoms in either exposure group. This finding is in contrast to a descriptive review of disaster media effects, which found a significant association between disaster-related television viewing and posttraumatic stress in most studies,Reference Pfefferbaum, Newman and Nelson2 and with meta-analyses of terrorismReference Houston1 and mass traumaReference Pfefferbaum, Nitiéma and Newman3 media studies, which found small but significant associations between media contact and posttraumatic stress symptomsReference Houston1,Reference Pfefferbaum, Nitiéma and Newman3 in samples with proximal and distal proximities to the disaster.Reference Pfefferbaum, Nitiéma and Newman3 The association of greater media contact with the number of re-experiencing symptoms in both the exposed and unexposed groups in the current study is consistent with the possibility that media consumption may contribute to intrusive re-experiencing of the disaster, regardless of personal trauma exposure.
It appears that graphic 9/11 media images, like media contact, in general, may contribute to intrusive re-experiencing of trauma regardless of personal exposure to the disaster. In contrast, associations between feeling bothered by graphic images and both the total number of posttraumatic stress and hyperarousal symptoms depended on exposure in the current study. Perhaps their real-life horrific exposures sensitized directly exposed survivors to hyperarousal and posttraumatic stress symptoms in response to these graphic media images. The relative importance of the role of the 3 groups of PTSD symptoms in the nosology of the disorder, in general, has been explored in prior studies. A number of investigators have found that symptoms in the re-experiencing and hyperarousal groups in the absence of prominent avoidance and numbing consist predominantly of responses reflecting non-pathological distress and that avoidance and numbing symptoms represent the core psychopathology of PTSD.Reference Breslau, Lucia and Davis34-Reference Whitman, North, Downs and Spitznagel39 Outside of the context of the diagnosis of PTSD, then, re-experiencing and hyperarousal symptoms may represent normative distress arising from trauma exposure or distress associated more broadly with other trauma-related stressors. The current study adds to the limited research examining media effects on posttraumatic symptoms in the 3 symptom groups in the disaster context.Reference Cardenas, Williams and Wilson9,Reference Collimore, McCabe, Carleton and Asmundson11,Reference Dougall, Hayward and Baum21,Reference Monfort and Afzali25 A recent meta-analysis found an association between posttraumatic stress symptoms and trauma-related television contact both in studies that examined the content of coverage (eg, graphic images) and in those that did not,Reference Pfefferbaum, Nitiéma and Newman3 but the meta-analysis did not explore associations with posttraumatic stress symptoms in the 3 symptom groups. Additional research is needed to clarify the influence of the content of coverage and/or graphic images on these outcomes in both exposed and unexposed samples.
Comparison of the Current Findings With Other Research
This study’s finding that media contact was not associated with disaster-related PTSD is supported by 2 studies that also used structured diagnostic interviews, including a study of directly exposed Oklahoma City bombing survivorsReference Pfefferbaum, North and Pfefferbaum40 and a 9/11 study of mothers of NYC school children.Reference Gershoff, Aber, Ware and Kotler12 Other studies have yielded positive associations between media contact and posttraumatic stress measures, perhaps due, in part at least, to methodological differences. Unlike most of the extant research that used survey methodology with samples drawn from disaster or distant communities or national samples, the current study used structured diagnostic interviews to assess full DSM-IV27 criteria for PTSD, specifically adhering closely to the trauma exposure criterion in a sample with a large number of directly exposed survivors. For example, a 9/11 study that assessed posttraumatic stress symptoms in a survey of Manhattan residents found that viewing media images of people falling or jumping from the towers was associated with posttraumatic stress symptoms only among those who were “directly involved in or affected by” the disaster.Reference Ahern, Galea and Resnick4 Unfortunately, some of the 9/11 experiences considered to reflect direct involvement in the study did not constitute qualifying trauma exposures required for the diagnosis of PTSD according to DSM-IV27 criteria. Further, the study’s measure of posttraumatic stress did not consider duration of the disturbance or distress or impairment. A Boston Marathon bombing study did measure functional impairment, and investigators clarified that they reported posttraumatic stress symptoms rather than PTSD because many participants did not meet the exposure criterionReference Holman, Garfin, Lubens and Silver23 (therefore only representing distress that is not part of the diagnostic construct). Trauma exposure and psychopathology may be overestimated in studies that do not assess all diagnostic criteria or fully conform to the criteria.Reference North, Suris, Davis and Smith38 Thus, the current study augments the literature by its assessment of all diagnostic criteria for PTSD in evaluating psychopathological outcomes.
Strengths and Limitations
A number of strengths and weaknesses of the current study warrant attention. A major limitation was the use of a non-representative sample of voluntary participants. The recruitment of the sample from agencies affected by the 9/11 attacks, rather than from the general community, permitted assessment of a large number of directly exposed individuals from a range of businesses of varying proximity to the WTC disaster site for comparison with trauma-unexposed individuals. The sample was small, relative to samples in studies that use general population survey methodology but would be considered a large sample for studies using diagnostic interview assessments, which are more burdensome because they require a greater commitment of time and effort for both the researchers and respondents. Furthermore, the response rate in this volunteer sample was similar to that in a study of the WTC Health Registry, which used interviewer-administered symptom surveys.Reference Murphy, Brackbill and Thalji41
Much of the extant research has not sufficiently considered the PTSD exposure criterion, resulting in the inclusion of unexposed as well as exposed participants in analyzing posttraumatic stress outcomes. Because individuals who do not meet the exposure criterion do not qualify for a diagnosis of PTSD, strengths of the current study were in limiting the analysis of PTSD to exposed individuals and in assessing all diagnostic criteria using methodologically rigorous structured interviews, differentiating PTSD in exposed individuals from distress in unexposed individuals who cannot have PTSD in response to a trauma to which they were not exposed.
The use of an inclusive global measure of media forms, referencing news from all sources rather than querying specific media forms (eg, newspaper, television, radio, Internet), may have influenced the results. A number of studies have used a single item to measure amount of television contactReference Bernstein, Ahern and Tracy7,Reference Cardenas, Williams and Wilson9,Reference Collimore, McCabe, Carleton and Asmundson11,Reference Grieger, Waldrep, Lovasz and Ursano13,Reference Schuster, Stein and Jaycox17,Reference Silver, Holman and McIntosh19 or a global item to measure amount of contact with multiple forms of media.Reference Dougall, Hayward and Baum21 A recent meta-analysis of mass trauma media effects on posttraumatic stress outcomes revealed significant associations regardless of measurement approaches, including studies that examined television or combined media (television and other media forms) coverage of specific incidents, studies that used numeric or subjective quantification of media contact, and studies that did and did not measure the content of media coverage.Reference Pfefferbaum, Nitiéma and Newman3 The assessment of feeling bothered by graphic media images in the current study was limited, as graphic images may have generated a psychological effect even if not experienced as bothersome. Other unexamined factors related to media variables may have also influenced outcomes, including the content or specific characteristics of coverage, participants’ motivations for media contact, and avoidance or discontinuation of media contact.
Another potential limitation was the timing of this study conducted nearly 3 years after the 9/11 attacks. Studies of retrospective recall of symptoms of acute stress disorderReference Harvey and Bryant42 and of psychological distressReference Brennan, Stewart and Jamhour43 in individuals directly exposed to trauma have found retrospective recall to be accurate with slight discrepancies. While they did not measure the accuracy of recalled assessment of media contact in their study of 9/11 media consumption in Canadian undergraduate students 2 years after the attacks, Collimore and colleaguesReference Collimore, McCabe, Carleton and Asmundson11 inferred support for people’s ability to recall distress and television consumption after salient events. A study of the reactions of Oklahoma City bombing survivors to 9/11 media coverage revealed that the amount of time that participants reported spending in contact with 9/11 media coverage in the first week after the attacks was independent of the time that elapsed between the attacks and the research assessment.Reference Pfefferbaum, Nitiéma and Pfefferbaum44 Nonetheless, recall bias should not be discounted as a potential source of bias in the interpretation of the results of the current analysis.
CONCLUSIONS
Differences in media effects in exposed and unexposed individuals in the current sample underscore the importance of considering exposure groups in studying media effects and of distinguishing psychopathological and normative reactions to media coverage. In both exposure groups, media contact in the week after the 9/11 attacks and feeling bothered by graphic 9/11 media images were associated with re-experiencing symptoms. Media contact and bothersome graphic media images may have heightened posttraumatic distress or, alternatively, greater posttraumatic distress may have driven participants to consume media coverage. Feeling bothered by graphic media images was associated with hyperarousal symptoms in the exposed group only, perhaps because these survivors were sensitized by their disaster exposures. Relatively few disaster studies have examined media effects on the 3 PTSD symptom groups. Thus, additional research is needed to validate and explicate the current findings.
In contrast to much of the terrorism media research, which has surveyed community and national samples, the current sample was recruited from 9/11-affected agencies and included large numbers of directly exposed individuals. The current study extends the extant research by using rigorous diagnostic interviews to explore pathological and normative outcomes in association with media contact separately in exposed and unexposed participant groups. The failure to find significant associations between media contact and PTSD in exposed participants suggests that media contact did not further contribute to psychopathological outcomes, although it was associated with normative re-experiencing symptoms regardless of 9/11 trauma exposure.
Increased access to traditional and new media forms that may carry graphic content and images of terrorist attacks has intensified interest in terrorism media effects. The current results underscore the importance of attention to media contact in both the exposed and unexposed individuals following incidents like the 9/11 attacks. Because of the potential for adverse effects associated with media contact, clinicians and public health professionals are encouraged to discuss concerns about mass trauma media contact with their patients and the public at large. Both exposed and unexposed individuals should be cautioned that disaster media contact and graphic images may lead to or exacerbate distress. Although research on disaster media effects has grown substantially over the last 2 decades, many issues remain unaddressed or unresolved. In particular, more studies using rigorous methodology and improved media measures along with greater attention to exposure groups and a host of moderators are needed to fully understand terrorism media effects.
Financial Support
This research was supported in part by the Oklahoma City National Memorial Institute for the Prevention of Terrorism (MIPT; Oklahoma City, OK) and the Office of Justice Programs, National Institute of Justice (Washington, DC), US Department of Justice award MIPT106-113-2000-020 (BP), and by the National Institute of Mental Health (NIMH; Bethesda, MD) grant MH68853 (CSN).
Conflict of Interest Statement
The authors have no conflicts of interest to declare.
Disclaimer
Points of view in this document are those of the authors and do not necessarily represent the official position of MIPT, NIMH, US Department of Justice, The University of Oklahoma Health Sciences Center, Metrocare Services, or The University of Texas Southwestern Medical Center.
