Prevalence of post-traumatic stress disorder, acute stress disorder and depression following violence related injury treated at the emergency department: a systematic review

Relevant studies were identified through systematic literature searches in the EMBASE, MEDLINE, Cochrane Central, PubMed, and PsycINFO databases. The search strategies were developed in consultation with a medical librarian. A detailed description of the search strategy can be found in the Additional file 1. Reference lists and citation indices of the included papers were inspected to identify additional relevant citations. We restricted searches to English-language papers, published in peer-reviewed journals before November 2017.

Studies reporting the prevalence of PTSD, ASD, or depression after ED or hospital treated injury following interpersonal violence were included in this review. We defined the following inclusion and exclusion criteria:

Titles and abstracts of all identified studies were screened for relevance by one reviewer (RO, BO, or JH). After initial selection, the remaining records were independently read in full-text by two reviewers (RO and BO) for the eligibility assessment. Discrepancies were discussed and resolved by consulting a third reviewer (JH). Two reviewers (RO and BO) extracted data on the study populations, study setting, injury details, prevalence rates, diagnostic instruments, and follow-up. If possible, we provided prevalence rates at different points in time. We used approximations when specific time points were not reported. For example, when ‘within two weeks after ED admission’ was reported as time indication, the midpoint (one week) was used. We reported gender-specific prevalence rates and measures of injury severity if provided.

A quality assessment in terms of risk of bias was performed with the Quality in Prognosis Studies (QUIPS) tool [12], which was developed for assessing the risk of bias of prognostic studies. Although the current systematic review does not focus on prognostic studies, we used the QUIPS tool because it covers general quality criteria on risk of bias. We considered these general criteria as appropriate because of the variety of study designs included in our study. The following domains of the QUIPS were selected in order to assess the risk of bias: study participation, study attrition, outcome measurement, and statistical analysis. Two reviewers (RO and BO) independently used the QUIPS tool to assess the risk of bias. Each domain was scored as ‘low risk’, ‘moderate risk’ or ‘high risk’. Any discrepancies in the domain scores were resolved via discussion until consensus was reached.

In total, the literature search yielded 3556 articles. After excluding 1537 duplicates, the titles and abstracts of 2019 articles were screened for relevance. The screening of titles and abstracts resulted in the exclusion of 1979 articles. Forty studies were left for full-text eligibility assessment, of which 24 were excluded for several main reasons: no prevalence reported, no violence related injury, no ED or hospital admission, literature review. Finally, sixteen studies were included in the systematic review. A flow chart of the study identification process is presented in Fig. 1.

The majority of the studies were conducted in the United States (n = 10) [13‐22] (Table 1). The remaining studies were conducted in the United Kingdom (n = 3) [23‐25], Denmark (n = 1) [26], and Norway (n = 2) [27, 28]. Seven studies included patients aged eighteen years and older [14, 20‐22, 26‐28] and two studies included patients aged sixteen years and older [23, 25]. In two studies [13, 24], the age of the participants was not specified. The remaining five studies applied different age criteria (Table 1) [15, 19].

All studies included patients who presented to the ED, trauma center, or hospital with injury following intentional violence. Alarcon et al. [13] included patients with the ICD-9-CM injury codes 800–995, covering injury such as open wounds and fractures. Injury related to sexual assault was excluded in three studies [14, 15, 19] and injury caused by domestic violence was excluded in four studies [19, 24, 27, 28]. In four studies on children and adolescents, injury caused by child abuse was excluded [15, 16, 18, 19].

A full structured clinical interview was used as diagnostic instrument in four out of sixteen studies [15, 20, 22, 24]. All DSM IV or V diagnostic criteria for PTSD (n = 4) and depression (n = 1) were met in these studies (Table 2). The Child and Adolescents Trauma Survey for assessing PTSD symptoms was used as diagnostic instrument in two studies [16, 18]. In both studies, patients were considered having PTSD when they scored 27 or higher. The Immediate Stress Response Checklist for ASD was used in the same studies [16, 18], although one of these studies did not report a cut-off score [18]. The diagnostic instruments used in the other studies were all different from each other. Twelve out of sixteen studies (75%) used brief questionnaires based on self-report or screening measures to obtain probable diagnoses. Therefore, these studies may have included individuals who would not have met the diagnostic criteria for ASD, PTSD, or depression if a full diagnostic interview would have been conducted. Brief questionnaires are mainly focused on symptoms whereas in a full diagnostic interview impairment is assessed as well.

The PTSD, ASD, and depression prevalence rates at different points in time are reported in Table 2. Fifteen studies reported the prevalence of PTSD following violence related injury [13, 15‐28], five studies reported the prevalence of ASD [14, 16, 18, 24, 26], and five studies reported the prevalence of depression [14, 15, 17, 18, 21]. The PTSD prevalence at 1, 3, 6, and 12 months post-injury ranged between 11.0–60.9%, 5.8–30.4%, 1.9–23.9%, and 16.3–27.1% respectively. The following range of ASD prevalence rates were reported < 1 week post-injury and 1–2 weeks post-injury: 24.0–24.6% and 11.7–40.6%. Four studies reported depression prevalence rates < 1 month post-injury ranging between 3.0 and 35.3%. Beyond one month post-injury, a prevalence rate of 16.8% was reported. Heterogeneity resulting from the use of different diagnostic instruments strongly limited the comparability of the reported prevalence rates of PTSD, ASD, and depression. In total, one study reported injury severity of the target population [26] and one study reported gender-specific prevalence rates [27] (Table 2).

Of all 64 possible scoring options (four quality domains times sixteen studies), the reviewers disagreed on five scoring options resulting in a disagreement rate of 7.8%. Two of the disagreements belonged to the study participation domain and three to the outcome measurement domain. Disagreements were resolved after discussion. Table 3 describes the risk of bias per domain (study participation, study attrition, outcome measurement, and statistical analysis) for all studies included. The study by Pailler et al. [18] was the only study with a low risk of bias on all four domains. The study attrition domain was mainly scored as high risk (83%) because of low participation rates and/or poor descriptions of the patients lost to follow-up. One study scored ‘low risk’ in this domain [18]. The statistical analyses and the presentation of the results were adequate in all studies. Therefore, all studies scored ‘low risk’ on the statistical analyses domain. The outcome measurement domain was mainly scored as low risk (67%). The majority had a low risk score for the study participation domain (67%), but one study had a high risk of bias because the recruitment process, inclusion criteria, and baseline characteristics were not reported adequately [26].

One of the strengths of our study is that standard methods for conducting and reporting systematic reviews were followed [11]. Furthermore, psychological, medical, and other relevant literature databases were searched exhaustively. Another strength is that we assessed the quality of the included studies. A limitation of our review is that the search was restricted to studies published in scientific peer-reviewed journals in English language. We did not consider dissertations, unpublished material or studies in non-English language, which could have biased our findings.

For future research, we recommend uniformity in diagnostic procedures. Structured diagnostic interviews by a clinician are preferred, but this is often not feasible. These interviews are time consuming and costly as they require involvement of trained professionals. Nevertheless, validated questionnaires can be used as an approximation. Our findings show that a large variety of questionnaires are available, however. Estimates of PTSD prevalence tend to vary according to the diagnostic criteria used, which underpins the need for uniformity in diagnostic procedures. These differences in diagnostic procedures could be reduced by establishing international guidelines on assessing mental health problems among trauma patients. Although international uniformity in diagnostic procedures would increase the comparability of PTSD, ASD and depression estimates, one should pay attention to ethnocultural differences. The validity of responses to measures may vary between populations, cultures, and countries [33]. Values and norms associated with culture guide perception and individual responses, including psychiatric symptoms [34]. Marshall et al. [35] investigated posttraumatic stress among a sample of Hispanic, non-Hispanic Caucasian, and African American survivors of physical injury. They found that the Hispanic group reported different symptoms and higher levels of overall posttraumatic distress. Such results raise questions regarding whether certain cultures truly experience higher levels of distress after experiencing a traumatic event, or whether cultural factors have an impact on the symptom manifestation only.

One of the sixteen included studies reported gender-specific prevalence rates. We recommend to report gender specific prevalence rates, since it is known that women are more likely to develop PTSD after trauma than men [29, 36]. Trajectories of PTSD, ASD, and depression can be better understood when distinguishing gender specific prevalence rates.

Prevalence rates should also be reported separately for injury types, such as sexual versus physical assault injuries and injuries caused by strangers versus family. The studies in the current review included patients with different injury types but prevalence rates were not reported separately. Identifying injury types that are associated with higher rates of PTSD, ASD, or depression may lead to earlier identification of high risk patients. Furthermore, ethnocultural differences in prevalence estimates should be considered in future studies. Cultural factors shape the subjective meaning of traumatic events, which in turn influences symptom expression [37].

Only few studies had follow-up measurements beyond one year after the violent incident. Previous studies suggest that the course of PTSD may vary over time. Prospective assessments are required to study the course of mental disorders following violence related injury treated at the ED or hospital. Since there are indications that the prevalence of PTSD among victims of intentional violence increases over time [5] it is relevant to know what the trajectories of PTSD and other mental disorders are for individuals who sustained injury following violence. For future research, extending the follow-up could contribute to better understanding of mental disorder trajectories following violence related injury.