Background
Healthcare is associated with relevant risks not only for patients, but also for healthcare professionals [
1]. Besides well-known risks to physical integrity like needle stick injuries [
2] or psychological stress [
1,
3,
4], unanticipated clinical events or outcomes (often caused by mistakes in healthcare) do not only harm patients. They also traumatise healthcare professionals, who may thus become so-called second victims [
5,
6]. Being a second victim may lead to dysfunctional coping strategies [
7], resulting in a change in work behaviour and leading to further negative employee-related outcomes. These outcomes are psychological and psychosomatic symptoms [
8], including isolation, reduced quality of life up to post-traumatic stress disorder (PTSD) [
7,
9,
10], or even suicide [
11]. Furthermore, the care of future patients (i.e., practising defensive medicine [
12,
13]) can be negatively affected, leading to overall reduced quality of care [
14]. Previous surveys in English-speaking countries indicate prevalences up to 42% of second victims among healthcare professionals [
15,
16]. Based on the research of the natural history of second victim traumatisation [
5], several interventional programs for healthcare professionals were launched, mainly in English-speaking countries [
17,
18]. They showed beneficial evidence regarding employee-related outcomes [
17,
19], and cost-effectiveness [
20].
In Germany, the association of statutory accident insurances defined standards for employees’ care after traumatising events [
21]. The first recommendation regarding handling of traumatisations after severe complications in patient care was published in 2013 by the German Society and the German Association of Anaesthesiologists [
22]. Contrary to sectors like rail services [
23] or air traffic [
24], where psychological support for employees after traumatising events has been addressed already, no systematic assessment of this phenomenon in the German-speaking healthcare sector has been published so far.
For this reason, we initiated the SeViD (Second Victims im Deutschsprachigen Raum/Second Victims in German-speaking Countries) project. As a first step, we developed and validated a German-language questionnaire for the assessment of second victim incidents [
25].
This study describes the new questionnaire’s first application in a sample of young German physicians, who are in training for general internal medicine or an internal medicine subspecialty. This research was planned and conducted before the SARS-CoV-2 pandemic. Nevertheless, the pandemic is an example, how an unanticipated adverse event can put many healthcare professionals under extreme pressure. Specialists experience high workloads and deal with uncertainty and death while at risk of contracting the illness themselves.
This study aims at adding evidence to the second victim phenomenon by evaluating the following hypotheses:
-
The prevalence of second victims among young physicians being trained in internal medicine in Germany is different from reported prevalences from other countries, specialties and age groups.
-
Distinct factors can predict the risk of becoming a second victim, the magnitude of symptoms and the time to self-perceived recovery after traumatic events.
-
Second victims favour certain support strategies.
Discussion
The survey aimed to investigate the prevalence, influencing factors on occurrence and course as well as support strategies for second victim traumatisations in a cross-sectional fashion among young German physicians working in internal medicine in inpatient care. International studies, especially from the US, suggest [
16] that second victim traumatisations are frequent among healthcare professionals and potentially carry a high impact on affected and future patients, the professionals themselves, their colleagues and thus the whole healthcare system. Data from Germany is scarce, which implies a considerable need for more research and campaigns in this country.
Nine out of ten participants of this survey had no knowledge of the term “second victim”. That does not automatically imply that these physicians were unaware of potentially occurring traumatisations at work, but the tendency seems obvious. In contrast, the study of Edrees et al. reported in 2011 that 46% (
n = 139 participants, manly nurses from Johns Hopkins University/ US) were aware of the term second victim and its definition [
27]. One reason could be that healthcare professionals’ traumatisations are until today, to our knowledge, neither mentioned in German medical school or specialty training curricula, nor many support programs exist at German hospitals and medical universities. In 2016, the European Board of Internal Medicine published European standards of postgraduate medical specialist training [
28]. Even this modern and comprehensive curriculum addresses traumatisations of healthcare professionals only superficially within the so-called milestones belonging to the CanMEDS framework (e.g. one milestone of the role healthcare advocate “identify, reflect on, and learn from critical incidents such as near misses and preventable medical errors” or one milestone of the role professional “recognise and address personal, psychological, and physical limitations that may affect performance”). Another limitation might be that until today parallel definitions of the term second victims exist (for three frequent definitions, see [
16]). In this context, it should be mentioned that the term second victim is criticised by some experts, who argue that it diminishes the importance and seriousness of the injury or complication of the patient and affected relatives [
29].
The prevalence of single or multiple second victim traumatisations found in our study among young Germany physicians in internal medicine was high (59% all-over), with 35% of the physicians affected in the last 12 months. A review [
16] reported prevalence rates from three studies varying from 10 to 43.3%: A study from Lander et al. from 2006 among otolaryngologists reported a 6-month prevalence of 10% [
30], whereas the study of Scott et al. from 2010 among various healthcare professionals including students found a 12-month prevalence of 30% [
18] and, finally, the study by Wolf and colleagues from 2000 described the prevalence of 43.1% again among various healthcare professionals [
31]. These studies mostly included older populations from different specialties in the United States. Compared to the results of this study, there is no clear signal that prevalences vary substantially with these factors.
Most traumatising incidents from this study where related to situations with direct harm to a patient or even their death. A minor number of cases were near misses or aggressive patients or their relatives. Remarkably, Waterman et al. in a study from 2007 stated that a third of the physicians who “only” have been involved in near misses were suffering from typical second victim traumatisations as well [
7].
Most second victims recover soon after traumatising events. Nevertheless, a small but relevant proportion - in our study 12% who need more than 1 year or have not recovered so far - recover late or never. Gazoni et al. report that 19% of traumatised anaesthesiologists have never fully recovered [
32]. Especially these colleagues need early and effective help to reduce the risk for severe outcomes like dysfunctional coping strategies, which potentially could harm other patients [
7], lead to physical and psychological morbidity [
16], or could lead to leaving the profession [
19].
Logistic regression models in of our study suggest that women are at greater risk of becoming a second victim (OR 2.5) and having higher symptom loads (OR 2) than men. Besides methodical limitations (e.g. women were overrepresented in the study sample) published studies reported several gender-related differences regarding the second victim phenomenon. Tolin and Foa conclude in a quantitative review from 2006 that females are generally more likely to meet PTSD criteria than men. However, they are less likely to experience potentially traumatic events [
33]. Studies by Kaldjian et al. [
34], Muller and Ornstein [
35] and Wu et al. [
36] report more distress among women after traumatising events on one side (e.g. feeling more guilt, being more afraid of losing confidence or reputation), but more constructive patterns of handling the situation compared to men on the other side (e.g. more motivated to discuss errors or to support changes in practice).
In our study, being in advanced training stages (6 years and more) was associated with a higher risk of becoming a second victim (OR 2). In a study by West et al. [
15], the prevalence of second victims increased with time from a 3-months prevalence of 14.3% to a 3-years prevalence of 34%. Some authors argue that almost every healthcare professional will experience at least one traumatic event throughout their career [
9].
In this study, shorter duration until self-perceived full recovery after a traumatic event was associated with predominantly working in acute care (OR 0.5). An explaining hypothesis could be that adverse events happen more frequently in these fields, so physicians might be better prepared through more routine and expectancy in dealing with such situations.
All regression models show a deficiency in predicting the outcome of the dependent variable. The possible explanation is that the relevant factors have not been included in these models and/or that multiple factors and their complex interactions influence the outcome. Van Gerven et al. lists personal, situational and organisational aspects that impact the outcome [
37]. Therefore, further research could concentrate on individual factors like personality characteristics, details of the traumatising events, or environmental conditions to explain differences.
Second victims of this study report that support in overcoming the traumatising event originated mainly from colleagues and friends or relatives, namely from the closest surrounding persons at work and home. Furthermore, second victims ask in particular for support strategies which include a prompt debriefing with discussion of the event and related emotional/ ethical aspects.
Today, nationwide support programs do not exist in the US [
5,
7,
38] nor in Europe [
39‐
41]. Single programs have been developed (e.g. in the US: “Medically Induced Trauma Support Service (MITTS)” in Boston [
42], the “forYOU” program at the University of Missouri Health Care [
18] or the „Resilience in Stressful Events (RISE)” program at Johns Hopkins Hospital [
17] and in Europe: “PSUakut”, which is a support program for healthcare professionals working in acute care in Germany [
43], the “Mitigating Impact in Second Victims (MISE)” online support program in Spain [
40] or “Collegial Help (Kollegiale Hilfe/ KoHi)”, a support program for second victims which is currently established at the Hietzing hospital in Vienna/ Austria [
44]). All programs include graduated levels of support. Scott et al. for example describe the following three levels (three tiers): Tier 1 with local unit/department support by direct colleagues, tier 2 with support through trained peer supporters and tier 3 with support through an established referral network (including professional support up to psychologists). The authors estimate that on these levels 60, 30 and 10%, respectively, of all second victims will receive sufficient support. Our study shows that most traumatised physicians who received support get it from their colleagues or friends. According to the just mentioned estimations by Scott et al., up to 40% might not receive the right support they need if professional support programs are not in place.
All support programs for second victims and their prevention aim for strengthening the resilience of healthcare professionals. The term “resilience” has been significantly shaped by the work of Aaron Antonovsky. He defined the sense of coherence as a prerequisite for resilience that is based on three components: viewing the world as comprehensible, meaningful and manageable. Regarding the SARS-CoV-2 pandemic and drawing on current recommendations by Wu et al. we recently published recommendations for healthcare leadership which take these three above-mentioned components into account [
45].
Our findings may be limited in several important ways. The cross-sectional design can describe associations but will never link causation. Our sample of young physicians in internal medicine was a convenience sample that is liable to selection bias and thus might lack representativity. More physicians with traumatic incidents in their past could have taken advantage of the survey. More women than expected were among our study participants. Furthermore, investigating only members of one medical society could harbour bias because members could have specific characteristics that might distinguish them from others. Another limitation is the low response rate and the number of dropouts which increased with the duration of the survey (at the last questions around 11%). The fear of potential participants to admit that something went wrong could have negatively influenced the response rate. There is often still a culture of blame in the workplace and fear of recrimination. Furthermore, due to the anonymous conduction of the survey we cannot exclude multiple participations of certain participants. Nevertheless, study characteristics like the response or dropout rates of our electronic survey were among expected limits for such designs. The item “time to full recovery” is difficult to define. Participants might feel that they have fully recovered, but the traumatic incident could still influence their behavior. Additionally, recovery might be a process with ups and downs. Finally, we did not correct for multiple-hypothesis testing. Our analysis is mainly explorative and is supposed to generate hypotheses and a basis for further research in Germany and Europe. Thus, we leave space in drawing the line between statistical significance and clinical relevance to the reader.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.