Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Comparing Screening Instruments to Predict Posttraumatic Stress Disorder

  • Joanne Mouthaan,

    Affiliation Department of Psychiatry, Centre for Anxiety Disorders, Research Group Psychotrauma, Academic Medical Centre, Amsterdam, The Netherlands

  • Marit Sijbrandij,

    Affiliations Clinical Psychology, VU University, Amsterdam, The Netherlands, EMGO Institute for Health and Care Research, Amsterdam, The Netherlands

  • Johannes B. Reitsma,

    Affiliations Department of Clinical Epidemiology, Biostatistics and Bioinformatics, Academic Medical Centre, Amsterdam, The Netherlands, Julius Centre for Health Sciences and Primary Care, University Medical Centre, Utrecht, The Netherlands

  • Berthold P. R. Gersons,

    Affiliations Department of Psychiatry, Centre for Anxiety Disorders, Research Group Psychotrauma, Academic Medical Centre, Amsterdam, The Netherlands, Arq Psychotrauma Expert Group, Diemen, The Netherlands

  • Miranda Olff

    m.olff@amc.uva.nl

    Affiliations Department of Psychiatry, Centre for Anxiety Disorders, Research Group Psychotrauma, Academic Medical Centre, Amsterdam, The Netherlands, Arq Psychotrauma Expert Group, Diemen, The Netherlands

Comparing Screening Instruments to Predict Posttraumatic Stress Disorder

  • Joanne Mouthaan, 
  • Marit Sijbrandij, 
  • Johannes B. Reitsma, 
  • Berthold P. R. Gersons, 
  • Miranda Olff
PLOS
x

Abstract

Background

Following traumatic exposure, a proportion of trauma victims develops posttraumatic stress disorder (PTSD). Early PTSD risk screening requires sensitive instruments to identify everyone at risk for developing PTSD in need of diagnostic follow-up.

Aims

This study compares the accuracy of the 4-item SPAN, 10-item Trauma Screening Questionnaire (TSQ) and 22-item Impact of Event Scale-Revised (IES-R) in predicting chronic PTSD at a minimum sensitivity of 80%.

Method

Injury patients admitted to a level-I trauma centre (N = 311) completed the instruments at a median of 23 days and were clinically assessed for PTSD at 6 months. Areas under the curve and specificities at 80% sensitivity were compared between instruments.

Results

Areas under the curve in all instruments were adequate (SPAN: 0.83; TSQ: 0.82; IES-R: 0.83) with no significant differences. At 80% sensitivity, specificities were 64% for SPAN, 59% for TSQ and 72% for IES-R.

Conclusion

The SPAN, TSQ and IES-R show similar accuracy in early detection of individuals at risk for PTSD, despite differences in number of items. The modest specificities and low positive predictive values found for all instruments could lead to relatively many false positive cases, when applied in clinical practice.

Introduction

A proportion of individuals exposed to a traumatic event develops psychopathology, such as posttraumatic stress disorder (PTSD) [1]. PTSD consists of symptoms of intrusion, avoidance, and hyperarousal. It is a disabling disorder which may lead to chronic psychiatric morbidity and loss of normal daily functioning [2], [3]. Traumatic injury is one of the most common traumatic events worldwide, accounting for 9% of global mortality [4]. Reported rates of PTSD within the first 6 months of injury range between 10–20% [1], [2], [5]. Early interventions for everyone involved in the traumatic event are unsuccessful in preventing PTSD (e.g., psychological debriefing) [6], whereas there is evidence that early treatment of acute PTSD with trauma-focused cognitive behavioural therapy can ward off a chronic course of PTSD [7], [8]. To identify trauma survivors at risk for PTSD in need of early treatment, the use of self-report instruments has been proposed [9], such as the SPAN [10], the Trauma Screening Questionnaire (TSQ) [11] and the Impact of Event Scale-Revised (IES-R) [12]. These instruments may be used as selection tools as part of a “triage” [13] or “screen and treat” [14] strategy. Only those individuals scoring above the cut-off are referred for further diagnostics. If the instrument is sufficiently accurate, such a strategy may save resources and costs [13].

In evaluating the accuracy of early screening instruments, its sensitivity (i.e., the probability that someone with PTSD has a positive test result) and specificity (i.e., the probability that someone without PTSD has a negative result) are important. The higher the sensitivity, the lower the specificity, and vice versa. In studies of diagnostic accuracy, an optimal cut-off point is usually chosen based on the trade-off between sensitivity and specificity [15]. However, the post-hoc nature in choosing an optimal cut-off has resulted in many different cut-offs reported across studies and populations. This limits comparability between screening instruments, and may complicate the professionals' choice for an early PTSD risk screening instrument. Related to sensitivity and specificity, but more important for clinical practice, are the positive predictive value (PPV) and the negative predictive value (NPV) of a screening instrument. The PPV refers to the proportion of people scoring positive on the screening instrument who indeed have or will develop PTSD, whereas the NPV refers to the proportion of people scoring negative on the screener who do not have or will not develop PTSD. PPV and NPV are both dependent on the population prevalence of a disorder. Typically, when the prevalence of a disorder in the population is low, as is the case in PTSD after injury (10-20%, [1], [2], [5]), PPV will be low and NPV will be high.

Which levels of sensitivity and specificity are acceptable, depends on the aim of administering the screening instrument and the costs and benefits of the decision based on the screening instrument [15]. Thus, there are no standard sensitivity and specificity levels that are considered acceptable. However, with respect to early PTSD risk screening after injury, a few recommendations can be made. First, in the early phase following traumatic injury, sensitivity may be important, to identify as many cases as possible that may be referred for more extensive diagnostic examination. In addition, it has been argued that in populations with a low condition prevalence, more sensitive screening instruments may be especially useful in selecting individuals for further assessment because PPV in that population will be low whereas NPV will be high [16]. Since previous studies in injury victims have shown that sensitivities of early PTSD risk screening instruments vary between 80% to 90% (for a review, see O'Donnell et al. [17]), it may be argued that for being acceptable as an early PTSD risk screener, an instrument should at least have a sensitivity of 80%. To illustrate, when 10% of a trauma population of 1,000 individuals will develop PTSD, 80% sensitivity would imply that 80 PTSD cases will be correctly detected and 20 PTSD cases remain unnoticed by the test. An important question, then, is which early PTSD risk screening instrument performs best at this sensitivity level?

Until now, most of the early PTSD risk screening instruments were developed to predict acute PTSD, but were not evaluated on their accuracy in predicting a chronic PTSD diagnosis. Yet, from a clinical and cost-effectiveness perspective we want to pick out the individuals with the worst long term prognosis and focus our resources on them. In this study, we will compare the diagnostic performance of three well-established PTSD self-report instruments when predicting PTSD at 6 months after injury at 80% sensitivity, aiming to evaluate whether the two shorter instruments, the SPAN [10] and the TSQ [11], perform as well as the longer instrument, the IES-R [12].

Method

Ethics statement

Medical ethical approval was obtained from the local institutional review boards of the Academic Medical Centre and the VU medical centre, Amsterdam, The Netherlands.

Participants

Participants were recruited from September 2005 to March 2009 for participation in a prospective study on the psychiatric morbidity following traumatic injury (for more details on the setting, see also Mouthaan et al. [18]). All adult patients presenting at a level-1 trauma centre in Amsterdam, The Netherlands, with injuries sustained in a traumatic event according to the DSM-IV A1-criterion for PTSD [19] and with sufficient understanding of the Dutch language were eligible for participation. Patients were excluded if they had injuries due to deliberate self-harm, an organic brain condition, current psychotic symptoms or disorder, bipolar disorder or depression with psychotic features, moderate to severe traumatic brain injury (according to the Glasgow Coma Scale score of less than 13) [20], or if they permanently resided outside the Netherlands.

Procedure

Research assistants selected eligible patients for participation from the hospitals' registrations and contacted them within 72 hours of the injury in-hospital or by telephone. Participants provided oral and written informed consent prior to data collection and completed an extensive clinical and self-report examination of psychological symptoms, including the index tests described in this study, within 8 weeks of their injury. All patients received clinical assessments of PTSD at 6 months. These were performed at the Centre for Anxiety Disorders of the Academic Medical Centre, at bedside or at the private home of the patient by master's- and doctoral level assessors who were blind to the outcomes of the baseline tests. Formal training in the Clinician Administered PTSD Scale (CAPS) [21] was provided to all interviewers by the original translators of the Dutch version of the CAPS 5.0 of the Research Centre for Military Mental Health Care, Utrecht, The Netherlands, and by experienced clinicians from the Centre for Anxiety Disorders of the Academic Medical Centre, Amsterdam, The Netherlands.

Measures

PTSD self-report instruments (index tests).

The 4-item SPAN (Startle, Physiological arousal, Anger, and Numbness) is derived from the 17-item Davidson Trauma Scale (DTS), which assesses frequency and severity of PTSD symptoms over the past week [22]. The items are rated on a 5-point scale from 0 (not at all distressing) to 4 (extremely distressing) with a total score range of 0–16. The Dutch translation of the DTS, including the SPAN items, has been previously reported to show high agreement with the Structured Interview for PTSD (SI-PTSD) [23], [24]. In our sample, the SPAN correlated moderately (Pearson r = .65, p<.001) with a concurrent CAPS assessment that was collected as a part of the larger prospective study [18]. Missing items were replaced with the mean item response (1 case with 1 missing item). Previous studies found sensitivity varying between 0.79–0.84 and specificity varying between 0.80–0.91 in relation to a concurrent PTSD diagnosis [10], [25], [26]. The only prospective study of the SPAN found a cut-off of 10 at 1 week post-trauma to be predictive of 2 month PTSD with a sensitivity and specificity of 0.86 [24].

The TSQ is a 10-item self-report scale with a yes/no-response format and assesses the presence of 5 intrusion items (e.g., “Upsetting dreams about the event”) and 5 hyperarousal items (e.g., “Difficulty falling or staying asleep”) over the past week [11]. It was adapted from the PTSD Symptom Scale-Self-Report Version (PSS-SR) [27]. The TSQ, translated into Dutch by the Research Group Psychotrauma of the Centre for Anxiety Disorders, shows adequate agreement with a concurrent CAPS assessment in the present study (Pearson r = .72, p<.001). Missing data were replaced with the scale mean (6 cases with 1 missing item). At a cut-off of 6 symptoms in any combination the TSQ showed a sensitivity/specificity of 0.86/0.93 in predicting PTSD in rail crash survivors at 6–12 months post-trauma and a sensitivity/specificity of 0.76/0.97 in crime victims within 1 month post-trauma [11]. Walters et al. [28] replicated this cut-off in a sample of emergency unit patients with a sensitivity/specificity of 0.85/0.89 for future PTSD at 1 month and of 0.88/0.78 for 6 month PTSD. At a cut-off of 7, the Dutch version of the TSQ showed a sensitivity and specificity of 0.87 and 0.69 respectively in a sample of civilian trauma survivors for future PTSD at 1 month [29].

The 22-item IES-R measures intrusion (8 items, e.g., “Any reminder brought back feelings about it”), avoidance (8 items, e.g., “I tried not to talk about it”), and hyperarousal (6 items, e.g., “I felt watchful and on guard”) [12]. The items are scored on a 5-point scale, from 0 (not at all) to 4 (extremely), with respect to how distressing each item has been in the past week. Total scores range from 0–88 with higher scores representing greater severity. In the present study, both the subscales and the total score of the Dutch IES-R, translated by the Research Group Psychotrauma of the Centre for Anxiety Disorders, show moderate to adequate similarities with the subscales and total score of the CAPS (Intrusion: r = .79, Avoidance: r = .57, Hyperarousal: r  = .65, total score: r = .75, all p<.001). Missing responses were replaced with the individual subscale mean (18 cases with 1 missing item, 6 cases with 2 missing items and 1 case with 3 missing items). Proposed cut-offs to indicate probable concurrent presence of PTSD were 19-20 in adolescent floods and mudslides survivors with a sensitivity/specificity of 0.86/0.84 [30], 22 in treatment-seeking substance abusers with a sensitivity/specificity of 0.92/0.57 [31] and in psychiatric patients at a sensitivity/specificity of 0.95/0.80[32], 24/25 in survivors of terrorist and natural disasters with a sensitivity of 0.83 and 1.00 and specificity of 0.67 and 0.84 [33] 33 in treatment-seeking and community-sample military veterans with a sensitivity/specificity of 0.91/0.82 [34], and an item mean score of 1.6, which corresponds to a cut-off score of 35, in acute lung injury survivors with a sensitivity/specificity of 1.00/0.85 [35].

Clinical interview.

The 30-item semi-structured interview of the CAPS was used to obtain the final diagnosis of PTSD [21]. The CAPS is generally considered to be the best reference standard for diagnosing PTSD [36]. Frequency and intensity of the 17 DSM-IV symptoms of PTSD are scored from 0-4, with higher scores indicative of more severe symptoms. The internal consistency of the scales in the Dutch translation of the CAPS is good to excellent (Intrusion: α = 0.63, Avoidance: α = 0.78, Hyperarousal: α = 0.79, CAPS total: α = 0.89) [37]. In the current study, the CAPS subscales and total score showed similar high internal consistency (Intrusion: α = .91, Avoidance: α = .83, Hyperarousal: α = .86, CAPS total: α = .95). Missing items were replaced with the mean item response of the corresponding subscale (5 cases with 1 missing item, 1 case with 2 missing items and 1 case with 3 missing items). Inter-rater reliability was high (ICC = .98, 95% CI = .95-.99). We used the rule of Weathers et al. [38] to establish a PTSD diagnosis, in which symptoms need at least a frequency of 1 and intensity of 2 with a total score of at least 45 points.

Analyses

Receiver operating characteristic (ROC) curves were computed for the total scores of the SPAN, the TSQ and the IES-R in relation to the 6 month PTSD diagnosis as the reference standard. As our main analysis, we performed paired t-tests to test for significant differences in areas under the curve (AUCs) using the method of Hanley and McNeil [39], which accounts for the paired data (subjects received all instruments under evaluation) within our study. Across all analyses, p-values of P = 0.05 (two-tailed) were considered to indicate statistical significance. Second, to illustrate the clinical implications of the three self-report instruments in predicting PTSD, we compared the values for specificity at 80% sensitivity. We calculated the exact specificity for 80% sensitivity by means of linear interpolation between the two data points directly before and after the 80% sensitivity. PPV, NPV and cut-off values are reported that correspond most closely to 80% sensitivity.

Results

Participant characteristics

Of the total study sample of 852 consecutively included injury patients, 311 patients (36.5%) were included in the final sample. Patients who did not complete any of the index tests (38.7%, n = 330) or did not attend the 6 month assessment of PTSD (24.8%, n = 211) were excluded from the final sample. Figure 1 shows a flow chart of the study sample. Patients completed the SPAN, TSQ and/or the IES-R at a median of 23 days post-injury (Inter Quartile Range = 10–16) and were clinically assessed for a final diagnosis of PTSD at 6 months. Compared to the excluded patients (n = 541), participants were older (mean = 44.9 (s.d. = 15.9) versus mean = 42.6 (s.d. = 15.8); t = −2.1, d.f. = 850, P = .037), more often female (39.9% versus 32.9%; χ2 = 4.2, d.f. = 1, P = .041), more often had a Dutch cultural background (89.3% versus 76.0%; χ2 = 21.9, d.f. = 1, P<.001), and were less often diagnosed with PTSD at 6 months (5.8% versus 13.7%; χ2 = 8.4, d.f. = 1, P = .004).

Eighteen patients (5.8%) of the final sample were diagnosed with 6 month PTSD. In addition, 22 patients (7.1%) were diagnosed with major depressive episode (MDE) and 24 (7.8%) with an anxiety disorder (AD) other than PTSD. Of the patients with PTSD, nine were diagnosed with comorbid MDE, two with comorbid AD and one with both. Thirty-seven patients reported having had contact with a professional mental health caregiver (i.e., psychologist, psychiatrist, social worker, victims aid worker) in the past 6 months, four of which were diagnosed with 6 month PTSD. Sixty-five patients in the final sample (20.9%) were offered a brief web-based early intervention within the first month of injury, aimed at preventing PTSD symptoms, as part of a randomized controlled trial [40]. Except for more prior traumatic experiences (t = −2.05, d.f. = 308, P = 0.011), there were no baseline differences between patients with and without PTSD. Table 1 presents baseline characteristics of participants and differences in baseline variables for patients with and without 6 month PTSD.

thumbnail
Table 1. Baseline characteristics of the total sample and participants with and without 6 month PTSD.

https://doi.org/10.1371/journal.pone.0097183.t001

Accuracy of the SPAN, TSQ and IES-R as screening instruments for PTSD

AUCs of the SPAN (0.83, 95% CI = 0.66–1.00), TSQ (0.82, 95% CI = 0.66–0.98) and IES-R (0.83, 95% CI = 0.72–0.94) were adequate (see Figure 2). There were no statistical significant differences between the AUCs of the SPAN, TSQ and IES-R (P SPAN-TSQ = 0.84, P SPAN-IES-R = 0.97, P TSQ-IES-R = 0.85). Figure 2 shows the ROC curves of the original data points of sensitivity and specificity values of the SPAN, TSQ and IES-R for 6 month PTSD using linear interpolation between the data points.

thumbnail
Figure 2. ROC curves of the SPAN, TSQ and IES-R for 6 month PTSD.

Note: ROC curves represent original sensitivity and specificity values using linear interpolation between the observed data points. ROC, Receiver Operating Characteristic; SPAN, Startle, Physiological Arousal, Anger & Numbness; TSQ, Trauma Screening Questionnaire; IES-R, Impact of Event Scale-Revised; PTSD, posttraumatic stress disorder.

https://doi.org/10.1371/journal.pone.0097183.g002

The specificity of the SPAN at 80% sensitivity was 64% (Table 2). This specificity indicates that in a population of 1,000 patients and assuming a prevalence of PTSD of 6%, the number of subjects receiving a false positive SPAN result will be 338. The cut-off most adjacent to 80% sensitivity was 4. PPV at this cut-off was 8% and NPV 99%, indicating an 8% chance of 6 month PTSD at a positive score and a 99% chance of no PTSD at a negative score. At 80% sensitivity, the TSQ's specificity was 59%, indicating that out of a 1,000 patients, 385 patients would unjustly score positive. The cut-off for the TSQ that most closely corresponded to 80% sensitivity was 5 (PPV = 0.19, NPV = 0.98). The specificity of the IES-R at 80% sensitivity was 72%, which means 263 false positives. A cut-off of 23 corresponded best with this sensitivity (PPV = 0.15, NPV = 0.99). There were no significant differences in specificities between tests.

thumbnail
Table 2. Specificity values of the SPAN, the TSQ and the IES-R at 80% sensitivity, and corresponding cut-off values, PPVs and NPVs.

https://doi.org/10.1371/journal.pone.0097183.t002

Discussion

Summary of the findings and comparison to previous studies

In this study, we compared the accuracy of three widely used early PTSD risk screening instruments in predicting a PTSD diagnosis at 6 months in injury victims. The results showed that with AUCs varying between 0.82–0.83, all instruments adequately distinguished between individuals with and without PTSD at 6 months. As we aimed for high sensitivity (80%) in order for these instruments to not miss potential clinical cases of PTSD, the specificities were modest for all instruments. This implies that while the instruments are adequate in identifying PTSD cases, they are poor in identifying non-cases. Thus, the instruments could well be used as a first selection step of possible cases, but a second, more comprehensive, diagnostic examination is needed to identify individuals in need of treatment. Importantly, the specificities did not significantly differ between tests, suggesting that the briefer, SPAN and TSQ, are as accurate as the longer, IES-R.

Our results concerning the accuracy of the SPAN, TSQ and IES-R are in line with those found in other studies that focused on screening for future PTSD [17]. Specificity results are often lower in replication studies as compared to the original validation studies, in which the items of the instruments are selected based on the performance in that particular population [29], [41][43]. Our proposed cut-offs of 4 for SPAN, 5 for TSQ and 23 for IES-R, best corresponding to 80% sensitivity, slightly differ from the cut-offs previously published [10], [11], [24][26], [28][35]. Our cut-off of 4 for the SPAN is much lower than found in a previous study administering the SPAN at 10 days after the trauma [24]. Perhaps the later timing of administration in our study (i.e., 3 weeks following trauma) explains this difference. At 3 weeks most initial PTSD symptoms, such as re-experiencing and hyperarousal symptoms that are usually found in the first weeks after trauma [17], are likely to have decreased naturally. Previous cut-offs reported for the TSQ were 6 [11], [28] and 7 [29]. Our cut-off of 23 for the IES-R is the first prognostic cut-off to be published, but matches the range of diagnostic cut-offs of 22 to 25 found previously [30][32], with the exception of studies of veterans [34] and acute lung injury victims [35] who reported cut-offs of 33 and 35 respectively.

It is important to consider that a PTSD incidence at 6% is low, which leads to low PPVs. In our sample, the highest chance of someone with a positive result developing PTSD at 6 months was 19%. It is an issue found in several previous screening studies, especially within injury populations [24], [28], [32], [44]. As our PPVs imply, at most only one in five persons scoring positive on the instruments will actually develop PTSD. This hampers the usefulness of screening initiatives in these types of population. Among patients who did not complete the early PTSD risk screening instruments (n = 330), PTSD prevalence was double that of the final sample (13.7%). This could indicate a reluctance to screening or research participation among patients with a higher risk of later PTSD. This issue has also been raised previously [41] and limits the generalizability of the findings to samples with a higher PTSD prevalence.

Previous studies have shown good results in identifying individuals at risk for depressive and anxiety disorders using screeners of 5, 3, 2 or even a single item [45][47]. Our finding that the short instruments perform equal to the longer one are in line with these results. However, comparable to our results, these screeners generally suffered from moderate to low specificity (44% to 77%) [45][47]. Thus, it may be worthwhile to evaluate the diagnostic accuracy of even shorter early PTSD risk screening instruments than in the current study, but only when followed by a comprehensive clinical evaluation, to rule out any false positives as a result of the lower specificity.

Strengths and limitations

A strength of our study is that we tested the accuracy of three instruments that are already in use globally in various populations and practices. Our study employed a comparative accuracy design, meaning that all patients received all tests under evaluation as well as the best available reference standard. Such a design has both higher validity and greater efficiency than comparing instruments between samples. Higher validity comes from the fact that all external factors can be kept constant when comparing the instruments; efficiency is enhanced because of the within patient comparisons that can be made similar to a paired t-test. This knowledge helps us make informed decisions on which instrument to best choose in a certain situation or population. Because these instruments do not rely on information specific to a particular trauma population (e.g., injury characteristics or the context in which the trauma occurred), but instead on symptoms, they may be useful for multiple groups or situations. A strong point of our sample is that, compared to other studies of screening in acutely traumatised populations [24], [28], [44], ours included a heterogeneous, consecutively included sample of injury victims, reflecting the broad range of injury victims of a level-1 trauma centre. We included various accident and assault victims, admitted to the hospital for a long period or immediately released. Finally, this study was the first to provide a prognostic cut-off for the IES-R and the first to examine the Dutch versions of the TSQ and the SPAN in a prospective design and using a consecutively included sample.

A limitation is that the screening instruments were administered quite late, at around 3 weeks following injury. This limits the generalizability of our findings to screening efforts in the immediate aftermath of traumatic events. Additionally, we were unable to collect a clinical PTSD diagnosis of 211 patients at 6 months, whose results on the index tests could not be included. However, post-hoc analyses revealed no significant differences in demographics, trauma history or scores on the screening instruments between participants and these dropouts. The determination of specificity values was hampered due to the relatively small sample size in the study. The total number of patients with a final diagnosis of PTSD was relatively low in this study. This hampered in particular the exact determination of cut-off values to obtain a specific value of sensitivity. Due to the low PTSD incidence in our sample, the ROC curves were surrounded with considerable uncertainty, which may have led to small, but potentially significant, differences between instruments remaining undetected.

Clinical implications

An important clinical implication of our study is that shorter instruments perform as well as longer ones in early PTSD risk screening. The use of shorter instruments may enhance the response rate to these screening instruments [48]. This means more individuals could be monitored or even referred for follow-up. Our results also imply that early screening cannot be used in clinical practice without adequate diagnostic follow-up. As a triage strategy, early screening can assist in filtering out anyone in need of PTSD treatment, without having to conduct time-consuming clinical interviews in low-risk individuals. This may not be problematic, since mental health treatment is usually preceded by a more thorough clinical interview.

Research implications

Future studies should investigate the potential disadvantages of early screening, such as the hazard of stigmatizing individuals when wrongly identifying them as potential PTSD cases. Lastly, the design and method of our study of comparing the diagnostic accuracy within the same patients and at a fixed sensitivity level should also be replicated in trauma populations with a higher PTSD prevalence, such as victims of rape or violence [49], to examine whether the instruments perform equally accurate in samples with a higher proportion of PTSD cases.

Acknowledgments

We thank the patients of the Trauma TIPS study for their participation in the study. We also thank the interviewers and students who helped collect and enter the data for this study. Special thanks are extended to research assistants Susanne van Buschbach, Nina A. van der Togt and Hendrieke Bolding for coordinating the data collection.

Author Contributions

Conceived and designed the experiments: JM MS MO. Performed the experiments: JM MS. Analyzed the data: JM MS JBR. Contributed reagents/materials/analysis tools: JBR. Wrote the paper: JM MS JBR BPG MO.

References

  1. 1. de Vries GJ, Olff M (2009) The lifetime prevalence of traumatic events and posttraumatic stress disorder in the Netherlands. J Trauma Stress 22: 259–267 10.1002/jts.20429 [doi].
  2. 2. O'Donnell ML, Creamer M, Pattison P, Atkin C (2004) Psychiatric morbidity following injury. Am J Psychiatry 161: 507–514.
  3. 3. Neria Y, Nandi A, Galea S (2008) Post-traumatic stress disorder following disasters: a systematic review. Psychol Med 38: 467–480 S0033291707001353 [pii];10.1017/S0033291707001353 [doi].
  4. 4. Peden M, McGee K, and Sharma G (2002) The injury chart book: a graphical overview of the global burden of injuries. Geneva: World Health Organization.
  5. 5. Zatzick DF, Jurkovich GJ, Gentilello L, Wisner D, Rivara FP (2002) Posttraumatic stress, problem drinking, and functional outcomes after injury. Arch Surg 137: 200–205. soa1164 [pii].
  6. 6. Sijbrandij M, Olff M, Reitsma JB, Carlier IV, Gersons BP (2006) Emotional or educational debriefing after psychological trauma. Randomised controlled trial. Br J Psychiatry 189: 150–155 189/2/150 [pii];10.1192/bjp.bp.105.021121 [doi].
  7. 7. Roberts NP, Kitchiner NJ, Kenardy J, Bisson J (2009) Multiple session early psychological interventions for the prevention of post-traumatic stress disorder. Cochrane Database Syst Rev CD006869. 10.1002/14651858.CD006869.pub2 [doi].
  8. 8. Sijbrandij M, Olff M, Reitsma JB, Carlier IV, de Vries MH, et al. (2007) Treatment of acute posttraumatic stress disorder with brief cognitive behavioral therapy: a randomized controlled trial. Am J Psychiatry 164: 82–90 164/1/82 [pii];10.1176/appi.ajp.164.1.82 [doi].
  9. 9. Sijbrandij M, Reitsma JB, Roberts NP, Engelhard IM, Olff M, et al. (2013) Self-report screening instruments for post-traumatic stress disorder (PTSD) in survivors of traumatic experiences. Cochrane Database of Systematic Reviews.
  10. 10. Meltzer-Brody S, Churchill E, Davidson JR (1999) Derivation of the SPAN, a brief diagnostic screening test for post-traumatic stress disorder. Psychiatry Res 88: : 63–70. S0165-1781(99)00070-0 [pii].
  11. 11. Brewin CR, Rose S, Andrews B, Green J, Tata P, et al. (2002) Brief screening instrument for post-traumatic stress disorder. Br J Psychiatry 181: 158–162.
  12. 12. Weiss DS, Marmar CR (1997) The Impact of Event Scale-Revised. In: Wilson JP, Keane TM, editors. Assessing psychological trauma and PTSD.New York: Guilford Press.pp. 399–411.
  13. 13. Bossuyt PM, Irwig L, Craig J, Glasziou P (2006) Comparative accuracy: assessing new tests against existing diagnostic pathways. BMJ 332: 1089–1092 332/7549/1089 [pii];10.1136/bmj.332.7549.1089 [doi].
  14. 14. Brewin CR, Scragg P, Robertson M, Thompson M, d'Ardenne P, et al. (2008) Promoting mental health following the London bombings: a screen and treat approach. J Trauma Stress 21: 3–8 10.1002/jts.20310 [doi].
  15. 15. Smits N, Smit F, Cuijpers P, De GR (2007) Using decision theory to derive optimal cut-off scores of screening instruments: an illustration explicating costs and benefits of mental health screening. Int J Methods Psychiatr Res 16: 219–229 10.1002/mpr.230 [doi].
  16. 16. Baldessarini RJ, Finklestein S, Arana GW (1983) The predictive power of diagnostic tests and the effect of prevalence of illness. Arch Gen Psychiatry 40: 569–573.
  17. 17. O'Donnell ML, Bryant RA, Creamer M, Carty J (2008) Mental health following traumatic injury: toward a health system model of early psychological intervention. Clin Psychol Rev 28: 387–406 S0272-7358(07)00129-8 [pii];10.1016/j.cpr.2007.07.008 [doi].
  18. 18. Mouthaan J, Sijbrandij M, Reitsma JB, Gersons BP, Olff M (2011) Internet-based prevention of posttraumatic stress symptoms in injured trauma patients: design of a randomized controlled trial. Eur J Psychotraumatol 2 . 10.3402/ejpt.v2i0.8294 [doi];EJPT-2-8294 [pii].
  19. 19. American Psychiatric Association (2000) Diagnostic and statistical manual of mental disorders (Revised 4th ed.). Washington, D.C.: American Psychiatric Association.
  20. 20. Teasdale G, Jennett B (1974) Assessment of coma and impaired consciousness. A practical scale. Lancet 2: 81–84. S0140-6736(74)91639-0 [pii].
  21. 21. Blake DD, Weathers FW, Nagy LM, Kaloupek DG, Gusman FD, et al. (1995) The development of a Clinician-Administered PTSD Scale. J Trauma Stress 8: 75–90.
  22. 22. Zlotnick C, Davidson J, Shea MT, Pearlstein T (1996) Validation of the Davidson Trauma Scale in a sample of survivors of childhood sexual abuse. J Nerv Ment Dis 184: 255–257.
  23. 23. Davidson JR, Malik MA, Travers J (1997) Structured interview for PTSD (SIP): psychometric validation for DSM-IV criteria. Depress Anxiety 5: 127–129 10.1002/(SICI)1520-6394(1997)5:3<127::AID-DA3>3.0.CO;2-B [pii].
  24. 24. Sijbrandij M, Olff M, Opmeer BC, Carlier IV, Gersons BP (2008) Early prognostic screening for posttraumatic stress disorder with the Davidson Trauma Scale and the SPAN. Depress Anxiety 25: 1038–1045 10.1002/da.20441 [doi].
  25. 25. Chen CH, Shen WW, Tan HK, Chou JY, Lu ML (2003) The validation study and application of stratum-specific likelihood ratios in the Chinese version of SPAN. Compr Psychiatry 44: 78–81 10.1053/comp.2003.50013 [doi];S0010440X03000336 [pii].
  26. 26. Seo HJ, Chung S, Lim HK, Chee IS, Lee KU, et al. (2011) A validation study of the Korean version of SPAN. Yonsei Med J 52: 673–679 201107673 [pii];10.3349/ymj.2011.52.4.673 [doi].
  27. 27. Foa EB, Riggs DS, Dancu CV, Rothbaum BO (1993) Reliability and Validity of A Brief Instrument for Assessing Posttraumatic-Stress-Disorder. Journal of Traumatic Stress 6: 459–473.
  28. 28. Walters JT, Bisson JI, Shepherd JP (2007) Predicting post-traumatic stress disorder: validation of the Trauma Screening Questionnaire in victims of assault. Psychol Med 37: 143–150 S0033291706008658 [pii];10.1017/S0033291706008658 [doi].
  29. 29. Dekkers AM, Olff M, Naring GW (2010) Identifying persons at risk for PTSD after trauma with TSQ in the Netherlands. Community Ment Health J 46: 20–25 10.1007/s10597-009-9195-6 [doi].
  30. 30. Chen CS, Cheng CP, Yen CF, Tang TC, Yang P, et al. (2011) Validation of the Impact of Event Scale-Revised for adolescents experiencing the floods and mudslides. Kaohsiung J Med Sci 27: 560–565 S1607-551X(11)00154-9 [pii];10.1016/j.kjms.2011.06.033 [doi].
  31. 31. Rash CJ, Coffey SF, Baschnagel JS, Drobes DJ, Saladin ME (2008) Psychometric properties of the IES-R in traumatized substance dependent individuals with and without PTSD. Addict Behav 33: 1039–1047 S0306-4603(08)00097-X [pii];10.1016/j.addbeh.2008.04.006 [doi].
  32. 32. Adkins JW, Weathers FW, McDevitt-Murphy M, Daniels JB (2008) Psychometric properties of seven self-report measures of posttraumatic stress disorder in college students with mixed civilian trauma exposure. J Anxiety Disord 22: 1393–1402 S0887-6185(08)00053-4 [pii];10.1016/j.janxdis.2008.02.002 [doi].
  33. 33. Asukai N, Kato H, Kawamura N, Kim Y, Yamamoto K, et al. (2002) Reliability and validity of the Japanese-language version of the impact of event scale-revised (IES-R-J): four studies of different traumatic events. J Nerv Ment Dis 190: 175–182.
  34. 34. Creamer M, Bell R, Failla S (2003) Psychometric properties of the Impact of Event Scale - Revised. Behav Res Ther 41: 1489–1496.
  35. 35. Bienvenu OJ, Williams JB, Yang A, Hopkins RO, Needham DM (2013) Posttraumatic stress disorder in survivors of acute lung injury: evaluating the impact of event scale-revised. Chest 144: 24–31 1393098 [pii];10.1378/chest.12-0908 [doi].
  36. 36. Weathers FW, Keane TM, Davidson JR (2001) Clinician-administered PTSD scale: a review of the first ten years of research. Depress Anxiety 13: 132–156 10.1002/da.1029 [pii].
  37. 37. Hovens JE, van der Ploeg HM, Klaarenbeek MT, Bramsen I, Schreuder JN, et al. (1994) The assessment of posttraumatic stress disorder: with the Clinician Administered PTSD Scale: Dutch results. J Clin Psychol 50: 325–340.
  38. 38. Weathers FW, Ruscio AM, Keane TM (1999) Psychometric properties of nine scoring rules for the Clinician-Administered Posttraumatic Stress Disorder Scale. Psychological Assessment 11: 124–133.
  39. 39. Hanley JA, McNeil BJ (1983) A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 148: 839–843.
  40. 40. Mouthaan J, Sijbrandij M, de Vries GJ, Reitsma JB, van de Schoot R, et al. (2013) Internet-based early intervention to prevent posttraumatic stress disorder in injury patients: randomized controlled trial. J Med Internet Res 15: e165 v15i8e165 [pii];10.2196/jmir.2460 [doi].
  41. 41. Brewin CR, Fuchkan N, Huntley Z, Scragg P (2010) Diagnostic accuracy of the Trauma Screening Questionnaire after the 2005 London bombings. J Trauma Stress 23: 393–398 10.1002/jts.20529 [doi].
  42. 42. Shalev AY, Freedman S, Peri T, Brandes D, Sahar T (1997) Predicting PTSD in trauma survivors: prospective evaluation of self-report and clinician-administered instruments. Br J Psychiatry 170: 558–564.
  43. 43. O'Donnell ML, Lau W, Tipping S, Holmes AC, Ellen S, et al. (2012) Stepped early psychological intervention for posttraumatic stress disorder, other anxiety disorders, and depression following serious injury. J Trauma Stress 25: 125–133 10.1002/jts.21677 [doi].
  44. 44. O'Donnell ML, Creamer MC, Parslow R, Elliott P, Holmes AC, et al. (2008) A predictive screening index for posttraumatic stress disorder and depression following traumatic injury. J Consult Clin Psychol 76: 923–932 2008-16943-012 [pii];10.1037/a0012918 [doi].
  45. 45. Cuijpers P, Smits N, Donker T, ten HM, de GR (2009) Screening for mood and anxiety disorders with the five-item, the three-item, and the two-item Mental Health Inventory. Psychiatry Res 168: 250–255 S0165-1781(08)00163-7 [pii];10.1016/j.psychres.2008.05.012 [doi].
  46. 46. van Ballegooijen W, Riper H, Donker T, Martin AK, Marks I, et al. (2012) Single-item screening for agoraphobic symptoms: validation of a web-based audiovisual screening instrument. PLoS One 7: e38480 10.1371/journal.pone.0038480 [doi];PONE-D-12-04312 [pii].
  47. 47. Donker T, van SA, Marks I, Cuijpers P (2010) Brief self-rated screening for depression on the Internet. J Affect Disord 122: 253–259 S0165-0327(09)00338-3 [pii];10.1016/j.jad.2009.07.013 [doi].
  48. 48. Edwards PJ, Roberts I, Clarke MJ, Diguiseppi C, Wentz R, et al. (2009) Methods to increase response to postal and electronic questionnaires. Cochrane Database Syst Rev MR000008. 10.1002/14651858.MR000008.pub4 [doi].
  49. 49. Bisson JI, Weltch R, Maddern S, Shepherd JP (2010) Implementing a screening programme for post-traumatic stress disorder following violent crime. Eur J Psychotraumatol 1 . 10.3402/ejpt.v1i0.5541 [doi];EJPT-1-5541 [pii].