On 22 July 2011, the Norwegian Anders Breivik killed 77 adults and children in a bombing in Oslo and a subsequent shooting on the nearby island Utøya, where the Norwegian Labor Party’s youth organization was having their summer camp.1
The Breivik attacks represent the most devastating act of terrorism occurring in Scandinavia in modern times. In addition to the casualties and the severe physical sequelae of the wounded, the Breivik attacks had a profound psychological impact. Among the survivors of the Utøya massacre, the posttraumatic stress levels 4–5 months after the attack were more than six times higher than those in the general population.2 The attacks also had an immediate psychological impact on Norwegians who were not directly exposed,3 which confirms findings from other terrorist attacks, most extensively documented for the 11 September 2011 (9/11) attacks in the United States.4,5
We have recently demonstrated that terrorist attacks can be associated with substantial deterioration of mental wellbeing far beyond the borders of the country in which the attacks take place. Indeed, in Denmark, the 9/11 attacks were followed by a substantial increase in the incidence rate of trauma- and stressor-related disorders.6 Based on these findings, we hypothesized that the Breivik attacks would also be followed by an increase in the incidence rate of trauma- and stressor-related disorders in Denmark, and that this increase would likely be larger than for 9/11 due to the geographic and cultural proximity of the locus of the attacks.
The methods employed here are analogous to those used in our study of the 9/11 attacks6 and are described in detail in eAppendix 1 (http://links.lww.com/EDE/B226). We drew our data from the Danish Psychiatric Central Research Register (DPCRR),7,8 which contains diagnoses on all individuals assessed/treated for mental disorders at psychiatric services in Denmark (including Greenland and the Faroe Islands). For each patient contact, a primary diagnosis is assigned by the treating psychiatrist in accordance with the criteria from the 10th edition of the International Classification of Disease (ICD-10).9 The study was approved by the Danish Data Protection Agency, the State Serum Institute and Statistics Denmark. Ethical review board approval is not required for register-based studies in Denmark.
Using the day of the Breivik attacks (Friday, 22 July 2011) as the intervention date, we applied time series intervention analysis10–12 to investigate whether the average daily number of diagnoses of trauma- and stressor-related disorders (ICD-10 code: F43 = reaction to severe stress, and adjustment disorders) in a given week increased after the attacks. Below we refer to this average daily number of diagnoses as the “incidence rate.” The incidence rates for the study period are available in eAppendix 2 (http://links.lww.com/EDE/B227) and an R script replicating the findings presented here is available in eAppendix 3 (http://links.lww.com/EDE/B228).
In accordance with our study of the effects of the 9/11 attacks,6 we also carried out a series of “placebo” tests (eAppendix 1; http://links.lww.com/EDE/B226). These tests involved estimating the exact same intervention models as in the main analysis, but with the intervention moved to a different date than that of the Breivik attacks to assess whether similar changes could be detected where no intervention had occurred. Further, to check whether the attacks were followed by an increase in the rate of trauma- and stressor-related disorders specifically, rather than an increase in mental disorders more generally, we carried out a similar intervention analysis on a combined time series of incidence rates of all other mental disorders in Denmark registered in the DPCRR. Finally, we ran a series of intervention models with different limits on the length of the preintervention period to make sure that the results were not an artifact of the long preintervention time series used.
Between 1995 and 2012, 159,618 acute contacts with Danish psychiatric hospital services were followed by a main diagnosis of a trauma- and stressor-related disorder. Figure 1 illustrates the development in these disorders over the time period.
Strikingly, the increase in the incidence rate of trauma- and stressor-related disorders after the Breivik attacks is actually discernible in the raw time series. To test this formally, we followed the procedure described in eAppendix 1 (http://links.lww.com/EDE/B226) and specified the following intervention model:
is the time series of trauma- and stressor-related disorders,
is a constant,
is a white noise error term,
denotes the first difference such that
denotes the backshift operator such that
is a function of time elapsed after the Breivik attacks. In the main model below,
denotes the intervention date. With the time series in first differences,
estimates the drift of the series (see eAppendix 1; http://links.lww.com/EDE/B226 for more details).
Figure 2 plots the observed incidence of trauma- and stressor-related disorders around the week of the attacks along with the predictions from this model.
Figure 2 clearly shows that right after the Breivik attacks, there was a marked elevation in the incidence of trauma- and stressor-related disorders (grey line with open circles) compared with the incidence rate predicted by the preintervention trend (dashed black line). The parabolic increase shown in Figure 2 (solid black line) is precisely estimated (linear term: 95% confidence interval (CI) = 0.318, 0.601; quadratic term: 95% CI = −0.008, −0.004; eTable 1; http://links.lww.com/EDE/B226). Specifically, the results demonstrate that the Breivik attacks were followed by an increase of approximately 16% in the incidence rate of trauma- and stressor-related disorders over the first 76 postintervention weeks (i.e., until the end of the time series), adding up to a total of 2736 estimated extra cases. The increase in cases was computed by using the coefficients of the linear and the quadratic terms to calculate the total estimated parabolic increase over the 76 postintervention weeks and then dividing this by the total number of diagnoses in the 76 preintervention weeks to arrive at the 16%.
In the placebo tests (eFigures 11–14; http://links.lww.com/EDE/B226), we did not find consistent evidence that increases similar to that observed for the Breivik attacks could be identified on any other dates. Furthermore, we found no increase in other mental disorders similar to that seen in trauma- and stressor-related disorders after the Breivik attacks (eFigures 15–16; http://links.lww.com/EDE/B226). Finally, the results did not change when we restricted the preintervention period (eFigure 17; http://links.lww.com/EDE/B226).
This study is the first to investigate extra-national deteriorations of mental health in the wake of the Breivik attacks. By means of time series intervention analyses of population-based data, we have shown that the Breivik attacks were followed by a 16% increase in the incidence rate of trauma- and stressor-related disorders in Denmark over the 1½ years after the attacks. A similar increase was not observed at other dates or for other mental disorders. This finding is consistent with the results of our recent study of the 9/11 attacks6 and further supports the notion that the psychological aftermath of terrorist attacks are not limited to the country under attack, but extends beyond borders.
The magnitude of the increase in the incidence rate of trauma- and stressor-related disorders after the Breivik attacks (16% increase over 76 weeks) was substantially larger than that observed for the 9/11 attacks (4% increase over 76 weeks). Although both attacks were exceptionally violent, this marked difference in the magnitude of the effects is somewhat surprising as the 9/11 attacks were substantially more severe in terms of death toll than the Breivik attacks (almost 3,000 vs. fewer than 80 individuals). Several factors—perhaps operating in conjunction—may explain the more pronounced effect of the Breivik attacks. First, the Breivik attacks in Norway took place much closer to Denmark than the 9/11 attacks—the distance between Oslo and Copenhagen (300 miles) is less than one tenth of that between Copenhagen and New York (3840 miles)—and there is evidence suggesting that geographic proximity to terrorist attacks is associated with increased psychological burden.3,13–15 Second, the geographic proximity is also reflected in a high degree of cultural similarity. The two Scandinavian countries have been united in various constellations for centuries up until 1814 and share many cultural features including mutually intelligible languages. This cultural similarity likely strengthens Danes’ emotional identification with Norwegians, which could also have contributed to the large spike in the incidence rate in Denmark after the Breivik attacks. Finally, the empathy of Danes may have been heightened by the fact that there was a Danish citizen among Breivik’s victims at Utøya.
While the overall increase in the incidence rate of trauma- and stressor-related disorders in Denmark after the Breivik attacks appeared roughly parabolic, a closer look at Figure 2 reveals a biphasic-like trend with a large surge immediately after the actual attacks and a secondary surge occurring a little over a year later. Given that previous research shows a link between exposure to media coverage of terrorist attacks and negative psychological reactions,16–18 a plausible explanation for this biphasic trend could be that the first surge in the incidence rate of trauma- and stressor-related disorders was driven by the media coverage of the actual attacks, while the second surge was driven by coverage of subsequent developments in the case—most notably the trial and verdict of Breivik—which did indeed occur just over a year after the attacks. To subject this (post hoc) hypothesis to empirical testing, we reanalyzed our data in two different ways both shown in Figure 3. First, we simply allowed for a more flexible estimation of the shape of the postintervention increase (see eAppendix 1; http://links.lww.com/EDE/B226 for details), which we then could visually compare with the development of the media coverage of Breivik after the attacks (measured by the average daily news items mentioning Breivik in a given week, cf. eFigure 3; http://links.lww.com/EDE/B226). Second, we developed a formal test of the association through an intervention model that used the timing of major spikes in news coverage as intervention dates instead of the date of the actual attacks. The model took the same parabolic form as the model using the true intervention week, but replaced this with the weeks showing spikes in coverage—defined as weeks when media coverage exceeded the 95th percentile of the postintervention distribution. Three weeks during the postattack period emerged as interventions based on this criterion: the week of the attacks, the beginning of the trial, and the passing of the sentence. The weeks were specified as regular interventions and assigned a quadratic and a linear term each (see eFigures 5–10; http://links.lww.com/EDE/B226 for more details on the analysis and for robustness checks using different criteria for classifying media attention).
The sextic polynomial used by the best-fitting flexible model estimated a clear biphasic trajectory in the incidence rate of trauma- and stressor-related disorders (Figure 3, top panel). A visual comparison of this with the development of the media coverage (bottom panel) suggested an initial surge after the actual attacks (extended by the beginning of the trial), followed by a second, but shorter-lived, surge upon the passing of Anders Breivik’s sentence on 24 August 2012 (imprisonment for 21 years with the potential of extension to life imprisonment).
The news spike model tested this more formally by replacing the actual intervention with the three spike weeks. Strikingly, this model produced predictions that displayed the same clear, biphasic trend as those of the best-fitting polynomial, and which were just as accurate. This meant that one could accurately model the development of the incidence rate of trauma- and stressor-related disorders after the Breivik attacks by completely ignoring the actual attacks and only using the news coverage surrounding them. This strongly suggests that media coverage of terrorist attacks—even in the extended aftermath of the actual attacks—can re-evoke trauma and bring about substantial negative psychological reactions.
In conclusion, this study has bolstered the finding that deterioration of mental health in the wake of terrorist attacks is not limited to the country under attack, but extend beyond borders. More specifically, the Breivik attacks in Norway were followed by a substantial increase in the incidence rate of trauma- and stressor-related disorders in neighboring Denmark. The increase was much more pronounced than the one observed after the 9/11 attacks, highlighting the moderating role of geographic and cultural proximity for psychological responses. Furthermore, we find evidence of a renewed surge in incidence during Breivik’s trial, when the news media revisited the attacks, thus suggesting that news media play a key role in triggering psychological reactions to terrorism.
The authors are grateful to Professor Kim M. Sønderskov (Department of Political Science, Aarhus University, Aarhus Denmark) for significant assistance in relation to the acquisition of data and discussion of the results.
1. Melle I. The Breivik case and what psychiatrists can learn from it. World Psychiatry. 2013;12:16–21.
2. Dyb G, Jensen TK, Nygaard E, et al. Post-traumatic stress reactions in survivors of the 2011 massacre on Utøya Island, Norway. Br J Psychiatry. 2014;204:361–367.
3. Thoresen S, Aakvaag HF, Wentzel-Larsen T, Dyb G, Hjemdal OK. The day Norway cried: proximity and distress in Norwegian citizens following the 22nd July 2011 terrorist attacks in Oslo and on Utoya Island. Eur J Psychotraumatol. 2012;3. [Epub ahead of print].
4. Schuster MA, Stein BD, Jaycox L, et al. A national survey of stress reactions after the September 11, 2001, terrorist attacks. N Engl J Med. 2001;345:1507–1512.
5. Stein BD, Elliott MN, Jaycox LH, et al. A national longitudinal study of the psychological consequences of the September 11, 2001 terrorist attacks: reactions, impairment, and help-seeking. Psychiatry. 2004;67:105–117.
6. Hansen BT, Østergaard SD, Sønderskov KM, Dinesen PT. Increased incidence rate of trauma- and stressor-related disorders in Denmark after the September 11, 2001, terrorist attacks in the United States. Am J Epidemiol. 2016;184:494–500.
7. Munk-Jørgensen P, Østergaard SD. Register-based studies of mental disorders. Scand J Public Health. 2011;39(7 suppl):170–174.
8. Mors O, Perto GP, Mortensen PB. The Danish Psychiatric Central Research Register. Scand J Public Health. 2011;39(7 suppl):54–57.
9. World Health Organization. The ICD-10 Classification of Mental and Behavioural Disorders. Diagnostic Criteria for Research. 1993.Geneva, Switzerland: WHO.
10. Box GEP, Tiao GC. Intervention analysis with applications to economic and environmental problems. J Am Stat Assoc. 1975;70:70–79.
11. Cryer J, Chan K. Time Series Analysis: With Applications in R. 2010.New York, NY: Springer.
12. Hansen BT, Sønderskov KM, Hageman I, Dinesen PT, Østergaard SD. Daylight savings time transitions and the incidence rate of unipolar depressive episodes. Epidemiology. 2017;28:346–353.
13. Nordanger DØ, Hysing M, Posserud MB, et al. Posttraumatic responses to the July 22, 2011 Oslo Terror among Norwegian high school students. J Trauma Stress. 2013;26:679–685.
14. Schlenger WE, Caddell JM, Ebert L, et al. Psychological reactions to terrorist attacks: findings from the National Study of Americans’ Reactions to September 11. JAMA. 2002;288:581–588.
15. Miguel-Tobal JJ, Cano-Vindel A, Gonzalez-Ordi H, et al. PTSD and depression after the Madrid March 11 train bombings. J Trauma Stress. 2006;19:69–80.
16. Holman EA, Garfin DR, Silver RC. Media’s role in broadcasting acute stress following the Boston Marathon bombings. Proc Natl Acad Sci U S A. 2014;111:93–98.
17. Neria Y, Sullivan GM. Understanding the mental health effects of indirect exposure to mass trauma through the media. JAMA. 2011;306:1374–1375.
18. Suvak M, Maguen S, Litz BT, Silver RC, Holman EA. Indirect exposure to the September 11 terrorist attacks: does symptom structure resemble PTSD? J Trauma Stress. 2008;21:30–39.