Participants and procedure
There have been two school shooting incidents in Finland during the last decade; at Jokela High School, 2007 and at Kauhajoki vocational school and college, 2008. We have studied the recovery process of the adolescents and young adults who were students of these institutions at the time of the incidents. The overall protocol for the study has been described elsewhere [
20,
21].
Here we present the results from an interview arm of the study. Clinical psychiatric assessments were performed about 16 months after the index incident. Those who had consented to take part in the follow-up study were contacted by phone to schedule interviews. Subjects could refuse an interview but otherwise continue in the study. Ethical permissions were given and study protocols were accepted by the Ethics Committee of Helsinki University Central Hospital and the Ethics Committee of the Hospital District of South Ostrobothnia.
Of the 474 Jokela middle and high school students, 124 were interviewed, 26.2 % of all the students and 53.7 % of the students taking part in the baseline questionnaire (N = 231) four months after the incident. Of the 389 Kauhajoki educational institution students, 104 were interviewed, 26.7 % of all the students and 44.1 % of the students taking part in the baseline questionnaire (N = 236). There were 228 interviewed students in total, of which 32 (14.0 %) were interviewed by phone due to inconveniently long distances, the remainder being face-to-face interviews. There were 184 (80.7 %) females and 44 males (19.3 %) and the mean age was 17.6 years (SD = 3.7, median age 17, range 12–30 years). Two thirds of the interviewed sample had no previous trauma exposure (65.0 %), while one quarter (24.3 %) disclosed one previous traumatic event, and about one in ten (10.7 %) had experienced two or more previous traumatic experiences. New trauma had been experienced by 19 students (8.5 %) after the index trauma.
Those who took part in the interview arm of the study did not differ from the whole study sample on age, socioeconomic status, previous need for psychosocial support, previous trauma exposure, exposure level of the index trauma, or baseline levels of symptoms measured by self-report scales. Boys did not take part in the interview as often as girls in Jokela, p = 0.001. Those who had experienced a new trauma after the index event took part in the interview more often in Jokela, p < 0.001.
Measures
The interview included basic background information. Exposure to the school shooting was used as the index incident. The students were asked to tell about their experience of the event in their own words, while structured questions were asked about fearing for their own or others lives or physical injury, about feelings of not being able to stop the events happening, requiring help, or acting in panic or being overwhelmed. The PTSD A2 criterion (exposure to a traumatic event is accompanied by intense fear, helplessness, or horror) required in DSM-IV was categorized based on these answers as either present (=1) or not present (= 0). An assessment of the severity of the exposure was based on the level of threat-to-life and losses suffered, as reported in the baseline questionnaire. The answers were categorized into mild-to-moderate, significant, and severe-to-extreme exposure [
20].
The semi-structured Schedule for Affective Disorders and Schizophrenia for School-Age Children—Present and Lifetime version (K-SADS-PL) [
22] was used with those under 18 years to assess major psychiatric diagnoses according to DSM-IV. The K-SADS-PL has from good to excellent test–retest reliability, high concurrent validity and inter-rater agreement for the original and translated versions [
22‐
25]; the Finnish translation has previously been used in different kinds of study settings. Adult age students were interviewed using the Structured Clinical Interview for DSM-IV (SCID-I) to assess major psychiatric diagnoses [
26]. However, the K-SADS-PL was used with all age groups for PTSD symptoms to ensure item by item consistency within the data. Psychosocial functioning was approximated using the Global Assessment Scale (GAS), using the children’s version when appropriate [
27].
The interviewers were either adolescent psychiatrists or experienced psychiatric nurses trained to use the instrument. Scorings by the nurse interviewers were reviewed with psychiatrists. Ambiguities were settled by consensus between two psychiatrists.
Although the used K-SADS-PL interview is based on DSM-IV, it includes items that are the same or close approximates of the proposed six ICD-11 symptoms. The re-experiencing items were especially considered, since the ICD-11 criteria of nightmares and flashbacks require that the event is experienced as occurring again and typically with overwhelming emotions. This excludes, for example, the use of the K-SADS-PL item Recurrent thoughts or images of event for the ICD-11 diagnosis, since the question allows for a voluntary contemplation and not necessarily a re-experience of the trauma with strong emotions.
The interviewed items corresponding to the ICD-11 Beta Draft [
3] PTSD diagnostic criteria for re-experiencing symptoms are
nightmares (Probes:
Has there ever been a time when you had a lot of nightmares? … How did you feel when you woke up from one of your nightmares?) and
dissociative flashbacks (Probes:
Has there ever been a time when you felt like it was happening again? … Was the feeling so strong that it was hard to tell whether or not it was happening again? Have you ever seen or heard things that you knew weren’t really there that reminded you of what happened?). Items corresponding to the avoidance symptoms were
efforts to avoid thoughts or feelings associated with the trauma (Probes:
What kind of things do you do or have you done to keep from thinking about what happened? To get rid of bad thoughts, some kids, read, do things to keep busy, or go to sleep. Did you ever do any of these things or other things to get rid of those bad thoughts and/or feelings?), and
efforts to avoid activities or situations that brought up recollections of the trauma (Probes:
You said before that sometimes __ reminds you of what happened. Did you try to avoid __?). Items corresponding to the hyperarousal symptoms were
hypervigilance (Probes:
Since __ happened, are you more careful? Do you feel like you always have to watch what’s going on around you? Do you double check the doors or windows to make sure they are locked?”), and
exaggerated startle response (Probes:
Since __happened, are you more jumpy? Do little noises really scare you?). Scorings of the PTSD symptom items in the K-SADS-PL are: 0 =
no information, 1 =
the symptom is not present, 2 =
the symptom is present, i.e. the symptom criterion is fulfilled. Rare missing items were replaced by 0 =
no information.
Scorings of the separate symptom and impairment items were used to compose PTSD diagnoses according to DSM-IV, ICD-10, and the proposed ICD-11 two- and three-factor solutions criteria. DSM-IV diagnosis required a fulfilling of stressor criterion A1 and A2 as well as1/5 of re-experiencing or intrusive symptoms, 3/7 avoidance symptoms, 2/5 hyperarousal symptoms, and impairment. The ICD-10 diagnosis was made when stressor criterion A1 was present as well as 1/4 re-experiencing symptoms, 1/2 avoidance symptoms, and specific amnesia or 2/5 hyperarousal symptoms. ICD-11 three-factor diagnosis required stressor criterion A1 as well as 1/2 of re-experiencing symptoms, 1/2 avoidance symptoms, 1/2 hyperarousal symptoms, and impairment. In comparison, the ICD-11 two-factor model combines re-experiencing symptoms and avoidance symptoms when 2/4 of these symptoms are required for the diagnosis (Table
1).
Table 1
Proportions of the studied subjects meeting PTSD symptom criteria and diagnoses
Stressor criterion A | | | | |
A1. Traumatic event | 228 (100.0) | 228 (100.0) | 228 (100.0) | 228 (100.0) |
A2. Emotional response | 186 (81.6) | | | |
Re-experiencing criterion B | | | | |
B1. Distressing recollections | 133 (58.3) | 133 (58.3) | | |
B2. Distressing dreams | 117 (51.3) | 117 (51.3) | 117 (51.3) | 117 (51.3) |
B3. Sense or reliving, illusions, hallucinations, or dissociative flashbacks | 52 (22.8) | | | |
(B3.) Dissociative flashbacks only | | 44 (19.3) | 44 (19.3) | 44 (19.3) |
B4. Psychological reactivity | 89 (39.0) | 89 (39.0) | | |
B5. Physiological reactivity | 62 (27.2) | | | |
Avoidance criterion C | | | | |
C1. Avoiding internal reminders | 90 (39.5) | 90 (39.5) | 90 (39.5) | 90 (39.5) |
C2. Avoiding external reminders | 55 (24.1) | 55 (24.1) | 55 (24.1) | 55 (24.1) |
C3. Specific amnesia | 39 (17.1) | 39 (17.1) | | |
C4. Diminished interest | 48 (21.1) | | | |
C5. Detachment | 26 (11.4) | | | |
C6. Restricted affect | 45 (19.7) | | | |
C7. Foreshortened future | 17 (7.5) | | | |
Hyperarousal criterion D | | | | |
D1. Difficulty sleeping | 107 (46.9) | 107 (46.9) | | |
D2. Irritability | 64 (28.1) | 64 (28.1) | | |
D3. Difficulty concentrating | 96 (42.1) | 96 (42.1) | | |
D4. Hypervigilance | 88 (38.6) | 88 (38.6) | 88 (38.6) | 88 (38.6) |
D5. Exaggerated startle response | 106 (46.5) | 106 (46.5) | 106 (46.5) | 106 (46.5) |
Criterion fulfilled | | | | |
Exposure and symptom criteria positive | 52 (22.8) |
85 (37.3)
| 66 (28.9) | 74 (32.5) |
Exposure and symptom criteria positive with positive impairment criteria |
43 (18.9)
| 62 (27.2) |
51 (22.4)
|
56 (24.6)
|
PTSD diagnoses occurring after the index incident until the time of the interview were included in the analyses. Other psychiatric disorders present after the index incident were considered when studying comorbidity. Depression included a major depressive disorder single or recurrent episode, dysthymic disorder and depressive disorders NOS. Anxiety disorders included a general anxiety disorder, panic disorder with or without agoraphobia, agoraphobia, specific phobia, social phobia and anxiety disorder NOS. Rates of alcohol use disorders were low in this partly adolescent sample. We categorized alcohol use as a no use or non-problem use and problem use. Problem use was coded when at least one alcohol use disorder diagnostic criterion was met or when an adolescent engaged in heavy binge drinking.
Posttraumatic stress symptom severity was estimated with Impact of Event Scale version that has 22 items (IES-22) [
21,
28]. This self-report form includes symptom statements (e.g.
I stayed away from reminders of it) that are rated on the basis of how frequently they occurred during the past seven days; 0 =
not at all, 1 =
rarely, 3 =
sometimes, 5 =
often. The General Health Questionnaire 12-item version (GHQ-12) was used to evaluate general psychological symptoms [
29]. The symptoms enquired (e.g.
Over the past few weeks, have you been feeling unhappy or depressed? 0 =
not at all, 1 =
no more than usual, 2 =
rather more than usual, 3 =
much more than usual) are scored in a bimodal fashion (0–0–1–1). Sum scores for the scales were calculated and used as continuous variables. Missing items were replaced by the the respondent’s mean score of the other items on a given scale. The entire scale was omitted from the analyses when more than 15 % of the items were missing. Internal consistencies (Cronbach’s α) were 0.941 for IES-22 and 0.897 for GHQ-12 with this sample.
Statistical analyses
The distributions of variables were presented as percentages for categorical variables and means (M) and standard deviations (SD) for continuous variables.
Confirmatory factor analysis (CFA) was used to compare three- and two-factor models for the six dichotomous K-SADS-PL symptom variables corresponding to the proposed ICD-11 PTSD criteria. The factors (and their indicators) in the three-factor model were 1) re-experiencing (distressing dreams, dissociative flashbacks), 2) avoidance (avoiding internal reminders, avoiding external reminders), and 3) hyperarousal (hypervigilance, exaggerated startle response). In the two-factor model, the four items from the re-experiencing and avoidance factors were combined to form one factor, while the hyperarousal factor remained intact. Models were analysed using the Weighted Least Squares Mean and Variance adjusted (WLSMV) estimator. Model fit was assessed using the
χ
2 statistic, the Root Mean Squared Error of Approximation (RMSEA), the Comparative Fit Index (CFI), and the Tucker–Lewis Index (TLI). A RMSEA below 0.06 and a CFI/TLI above 0.95 was considered to indicate a good fit [
30]. Additionally, Bayesian Information Criteria (BIC) were obtained from models estimated with a maximum likelihood (ML) estimator, where lower BIC values suggest a better fit.
Comparisons between proportions of the PTSD cases (prevalence rates) diagnosed by the different criteria were made with a Z-score test, while differences in background information and clinical characteristics among youth having a particular PTSD diagnosis, compared to those not having the diagnosis, were tested using an analysis of variance (ANOVA) and chi-square test. Cohen’s kappa was calculated to measure the diagnostic agreement between the proposed three-factor ICD-11 PTSD caseness and the other diagnostic systems.
Those individuals who met the three-factor ICD-11 PTSD criteria also met the two-factor ICD-11 criteria, and the ICD-10 criteria with and without impairment, the first one having the strictest criteria. Differences between clinical characteristics were tested between those meeting both diagnostic criteria and those meeting the less strict criteria only. Comparison between the three-factor ICD-11 and DSM-IV PTSD criteria was made between groups of those meeting both diagnoses, and those meeting the three-factor ICD-11 or DSM-IV diagnoses only. ANOVAs, with post hoc multiple comparisons (Bonferroni) when appropriate, and chi-square tests were used.
A significance level of
p < 0.05, two-tailed, was chosen. Analyses were performed using IBM SPSS Statistics version 22 and Mplus 7.1 software [
31].