While concepts of ‘stress’, ‘fear’ and ‘burnout’ have dominated the discourse on responses to ‘traumatic events’ like COVID-19 (Annaloro et al., 2021; Kira et al., 2021; Lis et al., 2020; Marvaldi et al., 2021), a less common narrative is that of ‘tragedy’. Baran et al.’s (2023) qualitative study explored the working experiences of intensive care unit (ICU) nurses during the pandemic and highlighted different themes relating to relationships, compassion and fear. However, less was revealed in terms of their own emotional responses to the realities of the deaths and the process of dying; that is, their own empathic grief to these ‘life tragedies of the losses’. This is important because of the known link between compassionate engagement with patients and vulnerability to empathic distress and coping with grief in certain clinical areas (Behan and Kelly, 2025). Although depression has been identified as one of the experiences of HCPs (Kira et al., 2021), it is important to distinguish sadness and grieving from depression because they are very different (Harris and Ho, 2023; Horwitz and Wakefield, 2007).

The narrative and definitions of tragedy have a different history, different meanings, and invite exploration of stress-traumatic experiences in new ways to those currently focused on ‘fear-based trauma’. Some years ago, one of the authors (PG) noted that when he empathically reflected on some clients’ experiences with the term tragedy rather than trauma, the client typically shifted into focusing on grief, a sense of loss and common humanity. Moreover, those who seemed blocked in their ability to process grief, which could interfere with the therapeutic journey, seemed to be more open to it with discussions about tragedy. These were anecdotal observations, linked to concepts of compassion but with no clear data (Gilbert, 2009; Gilbert, 2022a, pp. 337–340).

Currently, the distinctions and overlaps between fear-focused vs tragedy-focused trauma are rarely articulated. For example, the last 30 years has seen considerable research into the causes and experiences of, and therapies for, people who have been through significant threatening life events and traumas (Gold, 2017). Trauma can be defined in terms of an event, the psychophysiological reactions to the event(s) or context, and their interaction. The American Psychological Association (2024) offer the following definition:

Problematic in these areas is that terms such as ‘stress,’ ‘trauma’ and ‘disorder’ are not without controversies and ‘fuzzy’ boundaries (Richter-Levin and Sandi, 2021; Richter-Levin et al., 2019). Distinctions regarding ‘a disorder’ have been made on the basis of whether stressors promote resilience, both psychologically and biologically, or dysfunctional change, with the emergence of unique symptoms (Richter-Levin and Sandi, 2021). Using resilience as a distinction can be tricky however, because people’s resilience can emerge over time and become a source of growth (Tedeschi and Calhoun, 1996). Hence, a number of researchers have highlighted that both the definitions of a trauma as an event, as distinct from a stressor, and as a disorder distinct from a ‘normal’ reaction, are complex (Richter-Levin and Sandi, 2021; North et al., 2016). North et al. (2016) plotted some of the changes and problems in the history of the American Diagnostic and Statistical Manual of Mental Disorders (DSM-5; APA, 2013) and note that “the PTSD criterion requires not just the occurrence of a traumatic event, but also a qualifying exposure to it” (p. 206). Key to trauma is what the actual threat is. Harris and Machado (2017) note that stressful events can become traumatic partly because of the way they disrupt our experience of the world and throw us into states of uncertainty and threat vigilance:

Although there is some agreement around three major symptom clusters underpinning PTSD: hyperarousal and vigilance, avoidance and numbing, and flashbacks-re-experiencing, which are core to a diagnosis (Grey, 2009; North et al., 2016), it is also recognised that the impact and experience of trauma are highly heterogeneous and require different therapeutic interventions. Traumatic events can shatter people’s beliefs about the world and their ability to deal with it, particularly in terms of it becoming more unpredictable and frightening. Trauma implies injury to one or more of: the physical-body, self-identity, sense of safety and social trust. Hence, traditional approaches to PTSD tend to focus on different types of injury and fear of injury. In contrast, tragedy tends to be focused on sadness and grief. PTSD is also associated with complex unique patterns of emotions such as (for example) anger, anxiety, sadness/grief and disgust. Some individuals can feel intensely rageful while for others, the dominant emotions may be ones of fear or grief/sadness (Behan and Kelly, 2025). Some can experience mood shifts such as anhedonia and numbing (Eskelund et al., 2018), or hopelessness and defeat, which are different to grief states (Siddaway et al., 2015). Traumas can also be linked to self-blame, shame, shame memories and guilt (Lee et al., 2001; Pinto-Gouveia and Matos, 2011) that can impact a long-term sense of one’s identity (Lee, 2022; Matos et al., 2020). These variations can depend on the nature and context of the trauma, personal history and supports for coping.

However, there remain major research gaps exploring the impact of ‘traumatic events’. In a review of the neuroscience of trauma, Fonzo (2018) notes that, “there remains a stark imbalance in the degree to which the neuroscience of each affective domain has been probed and characterised in PTSD, with our knowledge of post-trauma diminished positive affect remaining comparatively underdeveloped” (p. 214). Seidemann et al. (2021) also highlights that “to date, the majority of PTSD research has focused on the inhibition system—fear learning, maintenance, and extinction—due to the conspicuous nature of the disorder’s fear-related symptoms. Though less researched, PTSD also involves depressive symptoms, such as anhedonia and emotional numbing (which relate to the reward system)” (p. 1). This is partly because research on trauma emerged from studies of veteran trauma that became a template for trauma studies, but this might not be a good template for other types of trauma, particularly those linked to the compassion motivation (Gilbert, 2022a, pp. 337–340).

Norcross and Wampold (2019) have also raised concerns about how we define and use terms like trauma and, in particular, how we can turn what can be ‘normal’ reactions to abnormal events into biomedical-type pathologies and disorders, e.g., PTSD (see also Harris and Machado, 2017; Horwitz and Wakefield, 2007; Lee, 2018). They note that an over-focus on such diagnostic approaches risks limiting people’s experiences to a set of ‘diagnostic core symptoms’ and ignoring the multiplicity of experiences and their therapeutic engagement. Indeed, whether mental health difficulties, such as certain depressions and PTSD, should be seen as disorders, reactions or postures has a long history (Hill, 1968) and has invited political critique (Lee, 2018; Harris and Machado, 2017; Horwitz and Wakefield, 2007). It is also a theme central to evolutionary approaches to mental health difficulties (Gilbert, 1989/2016, 2006; Nesse, 2019). Crucial, too, is the degree to which people experience trauma alone and in isolation such as various forms of violence or abuse, in contrast to collective trauma such as from slavery, war, famine, natural disasters and pandemics. All these concerns relate to how contexts influence how people conceptualise, narrate and give voice to life events and their impacts on them. One person may experience a trauma as a test by God (Jung and Pauli, 1952; Jung, 1969) while another as an example of the harshness, meaninglessness and callousness of life (Gilbert, 1989/2016, 2009; Felski, 2008). These conceptualisations impact scientific exploration and clinical engagement.

The variety of experiences of trauma can have many sources and causes. These can include: non-social versus social, intentional versus unintentional, direct versus vicarious. While most people will deliberately avoid harmful stressors and traumas, there are some circumstances and occupations where individuals deliberately put themselves at risk to help others, such as in the military, police, rescue services, and care and medical services (Fraess-Phillips et al., 2017). Given that compassion motivation is defined as ‘a sensitivity to suffering in self and others with a commitment to try to alleviate and prevent it’ (Gilbert, 2014), compassion motivation is at the heart of those who deliberately and intentionally move out of their own comfort zones, with personal risk, to help others in states of suffering (Gilbert, 2009, 2017, 2020; Ricard, 2015; Seppälä et al., 2017). In a study of COVID-19 public health care, Bhuiyan (2022) found mixed reasons for helping that included: ‘altruistic behaviour, service to profession, adherence to bureaucratic accountability, and a desire to help mankind’ (p. 269).

Deliberate engagement with suffering exposes people to potential trauma experiences, from risks of physical injury faced by the police, military and rescue services, for example, and infections and contamination faced by healthcare workers. In addition, the emotional experiences of witnessing the suffering of others, and efforts to help them, is also a source of mental distress. Hence, compassion is very different to kindness, and unlike kindness, requires courage, wisdom and commitment, and with some awareness of potential risk to self (Gilbert, 2020; Gilbert et al., 2019; Kirby et al., 2022). The mental distress arising from engagement with the distress-suffering of others has sometimes been labelled as secondary traumatic stress (STS; Kalaitzaki et al., 2022) and also, depending on context and circumstances, moral trauma (Litam and Balkin, 2021).

In regard to STS, frontline COVID-19 HCPs suffered many adverse reactions indicating high threat arousal, such as anxiety, depression, insomnia, somatisation and symptoms of PTSD (Conti et al., 2020; D’Ettorre et al., 2021; Wild et al., 2022) and burnout distress (Annaloro et al., 2021; Barello et al., 2020). Greene et al. (2021) surveyed 1,194 health and social care workers in May–July 2020, of which 47% met diagnostic criteria for anxiety and depression and 22% for PTSD. Marvaldi et al.’s (2021) meta-analysis estimated 20.2% of frontline HCPs suffered with PTSD symptoms and 31.1% experienced depression. Greenberg et al. (2021) found that 13% of frontline professionals reported suicidal ideation and self-harm. Newman et al. (2022) used social media to survey 395 NHS staff in April through to May, 2020. Their open-ended interviews generated three core themes: (1) Despair and uncertainty (feeling overwhelmed trying to protect everyone); (2) Behavioural and psychological impacts (affecting wellbeing and functioning) and (3) Coping and employer support (getting the right help). They conclude by saying [NHS staff] “felt enormous burden to adequately complete their professional, personal and civil responsibility to keep everyone safe leading to negative psychological and behavioural consequences and desire to receive better support from the NHS employers.” (p. 789). Another source of distress was empathic distress arising from the ‘tragedies’ of the deaths, the impact on their families and the way that HCPs often had to stand in for families and priests (Lake et al., 2022). Clearly then, the level of stress and distress for those engaging in this kind of compassionate action is high.

Linked to STS is moral trauma and moral injury (Fischer et al., 2022) and moral distress (Lake et al., 2022). Here, again, there can be elements of tragedy. Litam and Balkin (2021) give the following definition:

Hence, moral distress can arise when individuals feel they acted against, or have failed to act in line with, their moral principles and values that are key to their sense of self-identity (Fischer et al., 2022; Kip et al., 2022; Mottaghi et al., 2020; Browne et al., 2015). It can also relate to empathic distress from being moved by the tragedies of the dying patients, their families and having to act as stand-ins (Lake et al., 2022). Aiken et al. (2023) conducted a large-scale survey involving 21,050 physicians and nurses from US hospitals. Over 40% of respondents said they lacked confidence in management. They also selected management interventions (to improve the delivery of care to patients) as more important to their own mental health and well-being than interventions directly targeting clinicians’ mental health, implicating the importance of adequate patient care for their own wellbeing.

Among the crucial traumas and stresses that arise from compassion motivation is what happens if our efforts fail – if the people we are trying to help, be they ‘the rescued’, patients, relatives, or friends, continue to suffer and in some cases, worsen or die. Indeed, Williamson et al. (2020) indicated moral trauma to be especially noted when the people one is trying to help, die. This was a common experience in some COVID-19 care contexts. Although there is scant research on the impact of the death of those one is trying to help or save on those trying to help, the data so far suggests it can be a traumatising experience with many symptoms associated with PTSD (Masia et al., 2010).

Midwives can be confronted by many tragedies, e.g., baby health problems, deformities, stillbirth or subsequent death, threats to the mother and post-birth complications. Linetsky et al. (2024) found that in midwives the rate and intensity of PTSD symptoms were associated with self-criticism, lower self-compassion and pain catastrophising involving ruminating, magnifying and a sense of helplessness. Grief was not explored, but when their patients suffered despite the carer’s best efforts, there could be complex mixtures of self-blame, doubt, guilt, anxiety, anger and grief, depending on the circumstances (Behan and Kelly, 2025). Moral trauma can be linked to elevated levels of self-blame and self-criticism. These problems have been found in COVID-19 HCPs (Zerach and Levi-Belz, 2022). Elevated levels of self-criticism are associated with mental health difficulties (Werner et al., 2019). Importantly, it is the harshness and what is called self-attacking forms of self-criticism (Gilbert et al., 2004) that are particularly problematic (Gilbert, 2022b; Whelton and Greenberg, 2005).

Blaming external factors is also prominent in moral trauma. During COVID-19, HCPs faced limited resources that constantly confronted them with what could or should be possible with what was possible (Greenberg et al., 2020; Litam and Balkin, 2021). Being sensitive to suffering without being able to act as one would like to can lead to more distress (Gilbert et al., 2017; Di Bello et al., 2021). Hence, these stresses were linked to a sense of guilt and shame that one had not been able to care in the way that one would have liked to care. In addition, these were significant stressors in having to constantly step in as a proxy relative at the point of dying, again with worries of how one had been able to act as such. Although some of the ‘symptoms’ linked to STS and moral trauma have been associated with burnout or compassion fatigue, Litam and Balkin (2021) did not find evidence for compassion fatigue. Indeed, it is important to distinguish a genuine loss of motivation (for example for people you do not like) in contrast to wanting to be compassionate but feeling overwhelmed and helpless. However, again this research did not explore empathic grief, although it is common in certain clinical areas (Behan and Kelly, 2025).

As with other pandemics before it, the advent of COVID-19 created multiple arenas for fears, traumas and tragedies in the general population and with the HCPs trying to help. However these efforts were contextualised in multiple sources of stress that included: the nature of the disease, the number of people requiring help, the intense struggles to try to keep people alive with aided respiratory systems and other means, the displacement of other patients from their wards or places of care, the stresses of wearing appropriate protective clothing, not being able to facilitate relative contact with dying patients, and anger with the system in which HCPs were working. What was different in this pandemic to those that had come before was the extraordinary advance in medical technologies that enabled HCPs to save lives, but which also put major stresses on those delivering such services. As a number of studies have noted, unfamiliarity with the technologies and concerns about lacking the skills necessary was a major stress, and in some ways the more ‘miraculous’ medicine becomes (e.g., complex lifesaving surgery) the more HCPs can worry if they are keeping up in regards to their own competencies.

A dimension of trauma research that we think is relatively absent is the focus on the experience of tragedy rather than (just) fear, secondary traumatic stress and moral trauma. Further, we propose that understanding the psychology of the nature of tragedy as rooted in how we process certain types of loss, including the inevitabilities of disease and death, and including those associated with failed efforts to alleviate suffering, can advance understanding and help us to work with certain forms of trauma.

All biological life forms decay and die, and there is evidence that many species will try to help conspecifics who are injured. Some species of ant will even carry injured ants back to the colony (Kessler, 2020). Indeed, one of the evolutionary challenges that may well have driven human caring behaviour was concern and care for the sick and injured (Spikins, 2015). Caring for the sick and injured has always carried risks, not only of diverting resources away from self-care but also because bacterial and viral infections are dangerous and have driven species to extinction. The European bubonic plague (1346–53) killed more than 50 million, over 50% of the population at the time. More people died as a result of the Spanish flu of 1918 than the First World War (National Archives and Records Administration, n.d.).

The point is that part of human existence is being confronted with the normal processes of disease, decay and death but also the extraordinary mass killings from natural disasters and pandemics. Moreover, with heightened human capacities for compassion comes a capacity for empathic grief and sadness for the losses of others. Rahmani et al. (2023) reported that during and after COVID-19, HCPs were at a high risk of suffering from complicated grief (grief that significantly affects daily life for over 1 year; Marie Curie, 2024). Tragedy is a useful concept therefore because most dictionary definitions of tragedy define it as an event of great loss associated with sadness, sorrow and grief. While trauma is associated with fear, threat and numbing (Seidemann et al., 2021), some traumas are associated with intense emotions such as sadness and states of grieving (Gilbert, 2022a, 2023; Harris and Ho, 2023). However, just as the role of positive affect in trauma is understudied, so is the role of sadness and grief. Indeed, Horwitz and Wakefield (2007) brought a number of authors together to highlight how sadness has been gradually edged out of ‘normal human experience’ and can be become seen as forms of depressive disorders, which it is not. While there has been a lot of grief work in close relationships (Gruppi et al., 2022; Harris and Ho, 2023) there is also a long history of philosophical traditions that focus on sadness and grief in relation to life’s suffering, conceptualised as tragedy. Tragedy offers a different focus for conceptualising and narrating traumatic events. For example, we tend to focus on the experience of tragedy when thinking about suffering that has befallen another as opposed to ‘self-tragedy.’ This aspect of tragedy invites compassion via the experience of empathic distress; we feel distressed by the distress of the one experiencing the life difficulty, loss or misfortune. Indeed, it is empathic sharing that was essential to the Greek dramas. The importance of narrating experiences of (compassion-associated) traumatic events as also ones of tragedy are not currently conceptualised in trauma research or therapies. The efforts to heal moral trauma can involve working with forgiveness and compassion (Litz et al., 2009) but have not been considered in terms of tragedies of evolved human brains that can behave in the harmful ways that they do (Gilbert, 1989/2016, 2009).

The word tragedy has a different origin to trauma, and its meaning and textured experience is more elusive (Conversi and Sewall, 2025; Felski, 2008; Poole, 2005; Steiner, 2004). The Greek origin of tragedy is from tragos, meaning ‘goat’ and ‘ōidē’ meaning song (Poole, 2005). The link to its contemporary meaning is unknown, but it might have been associated with the singing of sad songs and laments, with ‘goat’ relating to sacrifice. It became a term to describe a certain type of theatre and drama and it was seen as a reflection on and response to the ordeals of life, as in the Greek tragedies (Felski, 2008; Nietzsche, 1872/2008; Poole, 2005). In theatre, the fates of Romeo and Juliet or Achilles are presented as tragedy rather than trauma. Grief and tragedy vary with meaning. For example, the death of a 90-year-old grandmother from dementia will be seen very differently from the death of a teenager in a car accident or the death of Romeo and Juliet – tragedy conveys a sense of loss of what could have been.

Felski (2008) and Poole (2005) discuss how the concept of tragedy was central to many western philosophers such as Plato, Aristotle, Kant, Hegel and Schopenhauer, and the more recent existential movement. Many of these philosophers contextualised the inevitable hardships and suffering of life and its ending (death, including the death of loved ones) as tragedies that all must face. Much of our lives can be spent trying to avoid processing these realities or coming to terms with them. This has been linked to terror management (Juhl and Routledge, 2016). Goldenberg et al. (1999) considered an explanation for why we are attracted to storylines of tragedy that are sometimes regarded as ‘tearjerkers’ like the Greek tragedies, because they can touch a part of us that is aware that ‘all that we love and all that we are, will pass away and be lost’, i.e., ‘all is impermanent’, as is stressed in the contemplative traditions (Anālayo, 2015). They found that priming people with this kind of morality awareness made them more sensitive and emotionally moved by tragedy scenarios. Key to tragedy is grief and sense of unfairness (Felski, 2008). Although sadness is discussed in the diagnosis of PTSD, it is linked to depression (Seidemann et al., 2021), that can disengage it from normal, intense grieving (Horwitz and Wakefield, 2007). Grief and sense of unfairness are also rarely discussed, but for pandemics and disasters, the arbitrariness of suffering can itself be traumatic and spiritually challenging. In the context of caring, HCPs can grieve for themselves (e.g., the losses of their own life disruptions) but also they can suffer immense empathic grief for the death and suffering of their patients and relatives.

How we narrate people’s distressing life experiences in terms of trauma and/or tragedy will impact our engagement and training. For example, Cook et al. (2019) explored many important dimensions for training HCPs in ‘sensitivity to trauma’ yet the concept of tragedy and the importance of guiding people through grief is less explored. Most therapies for trauma focus on exposure and desensitisation of some kind, although some trauma therapists also highlight that for some clients, grief can be profound and requires careful attention, analysis and time (Ogden et al., 2006) especially with its many varieties (Gilbert, 2022a; Harris and Ho, 2023). One dimension is grief for the life one has lost or has been robbed of by the trauma and its effects (Gilbert, 2023).

Grief work is central to compassion focused therapy and plays a major role in working with shame and the sense of disconnection that shame creates, which compounds trauma experiences (Gilbert, 2022b; Harris and Ho, 2023). Over a number of years, one of the authors (PG) has observed that narrating certain distressing life events in terms of tragedy-based trauma can lead more naturally into explorations of grief, whereas narrating them as trauma tends to create a more fear-based and damaged focus. An area where these differences may be especially important is when intense losses emerge within caring relationships, such as how clinical staff encountered COVID-19. As of June 2025, the World Health Organisation (WHO, 2025) reports 7.1 million deaths from COVID-19, and it is estimated to have led to excess deaths as high as 27 million (Our World in Data, 2025).

There is now considerable evidence that the quality of our social relationships plays a huge role in the biopsychosocial processing of threat and trauma events, impacting the autonomic nervous system, numerous neurocircuits, hormones and immune system functions (Slavich et al., 2023). Hence, the social contexts in which threats appear is crucial to their impact and subsequent effects. There is a large literature on the importance of the emotional and practical helpfulness and support of others in dealing with life crises (Taylor, 2011). Having a compassionate orientation to self and others has been shown to be a major coping strategy for stress with a variety of physiological effects (Singer, 2025). In a large multinational study of 4,057 adults across 21 countries using the Compassionate Engagement and Action scales (Gilbert et al., 2017), Matos et al. (2022) found that:

Behan and Kelly (2025) have highlighted how compassion engagement automatically gives rise to empathic distress and grief in the face of suffering and dying and the need to develop appropriate supports for working through grief. Miotto et al. (2020) highlighted the importance of providing emotional support for HCPs. It is important that care and support can come from the individual’s own support system in terms of friends and family, and from their organisation. Feelings of safety and safeness have profound effects on a range of physiological systems (Gilbert, 2024; Slavich et al., 2023). These will texture people’s abilities to cope with the tragedies that are unfolding around them. One of the complications is that fear of infecting family, friends and relatives might have reduced intimacy between people. Hence, understanding how people can recruit their personal networks to provide a secure base and safe haven and how these texture the experiences confronting them is an important research area.

As highlighted above, there is general agreement that experiences of trauma can vary vastly in terms of the patterns of emotional responses and focus (Norcross and Wampold, 2019). Within this heterogeneity, we suggest that the concept of tragedy, which is loss and grief-focused, can be distinguished from fear and injury-focused experiences, and that most traumas will have some combination to different degrees. For example, working through the trauma from abusive parenting may also require grieving for the ‘archetypal’ parent that children evolved to need and yearn for (Gilbert, 2023). We also suggest that in some kinds of contexts, such as compassionate efforts to help others (as for COVID-19 HCPs), empathic distress at the suffering of others can be experienced and narrated in terms of tragedy. We further suggest that fear-based aspects of trauma and tragedy-based aspects of trauma require different therapeutic interventions. Moreover, compassion focused interventions are particularly suited to address the tragedy aspects of trauma (Lee, 2022). Figure 1 offers an overview of how trauma experiences can be textured in different ways according to whether the focus is on a tragedy or a fear-based process, with recognition that there are many overlapping features. These variations can also indicate different patterns of intervention.

Many studies have explored the high levels of stress and trauma experienced by HCPs in relation to COVID-19 (e.g., Greene et al., 2021; Lamb et al., 2021; Duarte et al., 2022; Jordan et al., 2023; Newman et al., 2022). However, at the time of developing our study in 2020, some of these studies were unavailable. Hence, we explored the same themes from our discussions with colleagues and staff working in these areas. We wanted to explore a multi-dimensional and multi-textured set of experiences in terms of severity and perceived change. In addition, we wanted to contextualise the research in terms of the flows of compassion; that is the experiences people had in regard to their own mental states, the experiences that arose from compassion to others (patients) and the effects of compassionate support and help from others.

As Norcross and Wampold (2019) indicate, current trauma measures are focused on diagnosis rather than capturing the multi-textured experiences associated with the traumas and tragedies of life. This study, therefore, sought to develop a multicomponent survey that tapped into experiences of HCPs treating COVID-19 patients, such as those working in Intensive Care Units (ICUs). Rather than developing specific questionnaires, we wanted to measure patterns of experience. We were interested in whether such a multi-textured approach would enable us to detect patterns linking to how people would experience tragedy-based and fear-based trauma.

This was a cross-sectional, survey-based design. We began planning and developing the study in the October of 2021 and received ethical permission in the UK in March 2022 from the Health, Psychology and Social Care Research Ethics Committee at the University of Derby. Data were gathered for the UK sample between April 2022 and January 2023. The survey and measures were completed via Qualtrics (www.qualtrics.com, Qualtrics, Provo, UT). Participants were provided with an information sheet and then gave their consent before filling in the surveys. They were provided a debriefing sheet at the end.

Initially, the UK team had hoped to collect data from ICU units within an NHS Trust. However, due to administrative concerns and staff changes, we instead collaborated with ICU charities to collect data. These charities provide community hubs and support to ICU HCPs, hence ethical approval took longer than anticipated and data was gathered over a longer period. Thus, after March 2022, when the UK research team started collecting data, almost all COVID-19 restrictions had been lifted and rates of hospital admission had fallen substantially (Office for National Statistics, 2022), allowing useful insight into the change in HCPs’ experiences. Despite the lifting of most restrictions within that time, there was large strain on the NHS,¹ the removal of social distancing measures increased stress levels in HCPs and there were worries about COVID-19 returning in the winter of 2023.

In Portugal, the research team translated the scales for use in the Portuguese sample, and the back translations were examined by a bilingual researcher for accuracy and fidelity to the original scales. After ethical approval was granted by the Ethics Committee of the Faculty of Psychology and Educational Sciences at the University of Coimbra in November 2021, the research team contacted clinical boards of national general hospitals and health centers, inviting units directly involved in the treatment of COVID-19 patients to collaborate in the study (e.g., ICUs, Internal Medicine, Pneumonology, Emergency). The healthcare institutions that agreed to collaborate disseminated the study (via email) to the clinical boards of all the health units directly involved in the treatment of COVID-19 patients and respective HCPs. Data were gathered from the Portuguese sample between November 2021 and April 2022 in an online format, through the University’s Institutional LimeSurvey platform. Before completing the self-report questionnaires, participants provided their informed consent. Contact details for the researchers were made available if participants needed any clarification, and psychological support lines were also provided in case participants experienced distress during or after the study.

HCPs working closely with COVID-19 patients were recruited via email and completed the questionnaires using the online survey tools detailed above.

As this is a preliminary exploration of a complex array of varied experiences in the context of caring for COVID patients, we chose to use the survey method. Ponto (2015) offers a good summary, noting that:

Participants were also asked four (voluntary) open-ended questions such as “What was the most difficult aspect of dealing with people outside of your working environment?” and “What did you find helpful?.” These were included to allow participants to express any additional thoughts they may have had or which came up for them during the survey.

All quantitative data was analysed using SPSS version 28. We experienced delays with accessing the data due to issues with the data gathering program being updated. The data was screened for normality using parameters of skewness and kurtosis. Although all values were within the cut-off points of 2 for skewness and 7 for kurtosis (West et al., 1995), some variables had a non-normal distribution as shown by the histograms. Given that statistical methods to normalise data distribution (e.g., log transformations, square root transformations) can lead to biased estimates and difficulties of interpretation (McCuen et al., 1990; Lee, 2020), and can substantially impact Pearson correlations, we chose to use nonparametric Spearman’s rank correlations to explore the relationships between variables.

As the multicomponent survey of experiences of tragedy-based and fear-based trauma used single item questions, frequency analysis was conducted. Ratings on this survey were grouped into ‘low scores’ (1, 2, 3) and ‘medium to high scores’ (4, 5, 6, 7). Mean imputation is recommended for small amounts of ‘missing completely at random’ data (Tabachnick and Fidell, 2007) and performs as well as other imputation techniques (Peyre et al., 2011; Shrive et al., 2006). For this study, nine participants in the UK sample had a small amount of data (one or two data points) missing, which were replaced with the mean score. Only one participant in the UK sample had over 25% missing data in a section of the survey, so their data for that section was removed. A sensitivity analysis for missing data was conducted but it had no impact on results (Leurent et al., 2018). There was no missing (quantitative) data for the Portuguese sample. There were no outliers on more than one scale, therefore none were removed. Independent samples t-tests and Mann–Whitney tests showed significant differences between the UK and Portuguese data for many items, thus data from the two samples were analysed separately and compared.

Descriptives (means and standard deviations), reliability analyses and Spearman’s rank correlations were conducted on suitable study data. Data gathered from four voluntary open-ended questions was reviewed although it was not suitable for in-depth qualitative analysis due to the questions’ concise nature, and because not all participants provided answers. Instead, we briefly discuss the main findings in prose in the following section.

There are unique characteristics of fear-based and tragedy-based trauma when people choose to enter into states of other people’s suffering. These are forms of empathic, other-focused distress. This makes them a very different type of distress-trauma to, say, the trauma of a car accident or violence against the self and one’s own injuries. These are explored in Table 6.

There is considerable evidence that social relationships, particularly those that provide support and caring, have major impacts on a range of biopsychosocial systems that can impact not only mental states but also immune systems which can play a role in virus susceptibility (Slavich et al., 2023). Hence, over and above practical supports, relational supports are crucial. HCPs, like all of us, can be socially embedded and supported in relations with partners, family, friends, professional colleagues and networks. These can be significant, compassionate resources for coping with stress, recovery and posttraumatic growth. Our questions in Table 7 explored how HCPs experienced this network of relationships to their clinical work. On the other side however, people who are avoidant or fearful of seeking and responding to compassionate support and help can be at a disadvantage (Matos et al., 2021a, 2021b).

In our informal discussions with HCPs, they noted several organisational stresses which we tried to capture in the questions shown in Table 9.

The context in which people work, over and above the nature of the work, can also be a source of support or stress and trauma (Dutton et al., 2014). Anecdotal evidence from our informal discussions suggested this was something HCPs felt strongly about. It is important to keep in mind that the HCPs who participated in this study were from diverse groups and some will have been more frontline and intensively involved than others.

There is increasing interest in how people come through these life experiences and are changed by them for better or for worse. Crucial can be the feeling of making a meaningful contribution that increases our sense of self-efficacy and value to others. Also, coming through trauma helps us recognise that we can cope with things in a way that maybe we thought we would not be able to. One of the distinctions that is sometimes made between PTSD and severe stress is that of resilience. These themes are explored in Table 10.

Tables 11–14 provide data on the four standard measures used in the study.

Most research investigating the experiences of HCPs involved in COVID-19 contexts focuses on trauma. Table 11 provides the means and standard deviations for trauma (IES-R) in the Portuguese and UK samples.

Scores of 33 and above suggest a PTSD clinical diagnosis (Creamer et al., 2003) and 88 is the highest possible score. The UK participants’ mean total scores were high (39.52) compared to the Portuguese participants’ (22.36). Despite variation, some studies show a similar mean score to the UK data. For example, Nie et al. (2020) showed a mean score of 39.8 for frontline nurses regarding the impact of the pandemic, and Couper et al.’s (2022) study reported a mean of 32.2 during the first wave of COVID-19 with a sample of UK nurses and midwives. The large difference between the samples’ scores might reflect the demographic data on the healthcare workplace, as 84.7% of the UK sample worked primarily in an ICU compared to only 11.3% of the Portuguese sample. A higher number of UK participants also reported they had contracted COVID-19 and lost colleagues to the virus compared with the sample in Portugal. Moreover, the mean scores for the Portuguese sample seem to be similar to other studies where data was collected after the height of the pandemic (Figueiredo et al., 2024) and with non-COVID-19 samples (King et al., 2009; Lee et al., 2018).

Table 12 provides the data for the Social Safeness and Pleasure scale, which measures the degree to which people feel generally secure in their relationships, with a sense of belonging, connectedness, being cared for, wanted and valued (Gilbert et al., 2009). There is now good evidence that this is significantly related to mental health measures and well-being (Armstrong et al., 2020). The scores are consistent with other studies, indicating that HCPs felt as socially connected and safe as the general population (Basran et al., 2019; Kelly et al., 2012). The two samples are equivalent in terms of their sense of social safeness.

The Posttraumatic Growth Inventory data is shown in Table 13. As mentioned, this scale was only administered to UK participants. The highest possible score is 105. Mazor et al. (2016) suggests scores of 46 and above represent medium to high posttraumatic growth (PTG) levels. This indicates that many of the UK respondents have experienced high levels of posttraumatic growth, which aligns with similar PTGI mean scores reported by HCPs in other pandemic-related studies (e.g., Prekazi et al., 2021; Sarıalioğlu et al., 2022; Kalaitzaki et al., 2022; Kapur et al., 2022).

Burnout is explored in Table 14. A number of studies have indicated that burnout is a risk when healthcare professionals are subjected to this degree of distress for long periods and when under pressure because of staff shortages and other factors.

This data reflects the burnout levels of the HCPs within the month leading up to study completion. The maximum individual mean score is 7. Lundgren-Nilsson et al. (2012), Gerber et al. (2018) and Schilling et al. (2019) reported clinically significant burnout as a score of ≥ 4.4. This therefore shows that the UK (M = 3.65), and Portuguese (M = 3.88) samples do not have clinical burnout. The Portuguese sample show clinically significant physical fatigue (M = 4.61) whereas the UK sample do not (M = 4.16).

Tables 15–19 provide the Spearman’s rank correlations between variables linked to fear-based trauma and variables linked to tragedy-based trauma in both populations, scored when self-report measures were completed (i.e., the ‘do’ data). The tragedy-based trauma data was based on the question: ‘[do you] feel overwhelmed with sadness for the tragedy of pandemic deaths you experienced?’ This item is key to the experience of tragedy. We deliberately used the term overwhelmed to try to measure people’s intensity of sadness. To see how this links to symptoms and experiences of fear-based trauma as measured in standard trauma measures, we use the IES-R (Weiss and Marmar, 1997). The relationship between tragedy-based and fear-based trauma is explored in Table 15.

The British sample scored highly on the trauma scale at (M = 39.52), in other words there was significant endorsement of PTSD symptoms. Despite this high rate of symptoms, in the UK sample, there is no correlation with this and the endorsement of feeling overwhelmed (which we labelled as tragedy). This would indicate that fear-based and tragedy-based trauma, or at least the concept and use of these words, are associated with different experiences of distress. The small but significant positive correlation between fear-based and tragedy-based trauma shown for the Portuguese sample perhaps reflects the vastly low proportion of those HCPs working in an ICU compared to the UK sample. ICU staff, typically, have more pre-pandemic experience of witnessing distressing cases, i.e., patients dying, compared to HCPs in general and community wards, thus possibly explaining the increased sadness and sense of feeling overwhelmed in the Portuguese participants. To investigate this further we looked at the pattern of emotions associated with fear-based and tragedy-based trauma.

Table 16 provides correlations for the emotions that HCPs felt about the pandemic in general (given in Table 1) with fear-based and tragedy-based trauma.

This analysis also revealed different patterns of emotions associated with the experience of fear-based vs tragedy-based trauma. For the UK data, feeling overwhelmed by the tragedy is linked to anxiety, sadness and depression, but the same emotions were not associated with fear-based trauma (apart from a small correlation with anger). Again, this indicates these are different types of experience.

Table 17 provides correlations for the emotions associated with flashbacks (given in Table 2) with tragedy-based and fear-based trauma.

Given that flashbacks are very common when going through traumatic events, we were interested in the emotions that textured flashbacks in relation to fear-based and tragedy-based trauma. Surprisingly perhaps, in the UK sample, the standard trauma measure (tapping into fear-based trauma) was not linked to the emotions texturing flashbacks whereas tragedy-based trauma was. Especially notable is that sadness and grief were strongly correlated with conceptualising their experiences of tragedy-based trauma, as well as anxiety and to a lesser degree, anger.

Table 18 shows correlations between tragedy-based and fear-based trauma and compassionate support from others such as feeling understood, family and colleagues’ support and being able to turn to others.

Table 18 indicates that the symptoms of fear-based trauma were not affected by the degree to which people experienced being understood, family support, colleagues’ support, openness and being able to turn to others. We were surprised by this, but again it may mean fear-based trauma is not the appropriate experience in these stressful contexts. When individuals were asked to consider their experiences relating to the concept of tragedy-based trauma however, they did find these sources of support facilitated their ability be aware of, and tolerate feelings of, overwhelming distress—supporting the importance of distinguishing experiences of tragedy-based trauma from experiences of fear-based trauma. However, this was only for the UK sample (mostly frontline workers) and not for the Portuguese sample (mostly community workers). These variables (being understood, family support, colleagues’ support, openness and being able to turn to others) were important for social safeness in both the Portuguese and UK samples, interestingly slightly more so for the Portuguese sample.

Regarding the measure of posttraumatic growth, our data suggests the support people gain from family and colleagues, feeling understood and being open to receiving help are significantly related to this type of change and personal development. This is seen in existing research (Scrignaro et al., 2011). These experiences contributed to a sense of general social safeness and connectedness. Interesting, too, is that these social support factors were not related to burnout, indicating they may operate separately from the challenges of the work itself. Wei et al. (2024) found that secure attachment histories enabled nurses to engage and tolerate painful feelings linked to death. Regarding specific emotions, small variations link to these variables, for example sadness and the ability to tolerate sadness are linked to various forms of support. It is an area requiring further research.

Table 19 explores tragedy-based trauma and fear-based trauma in relation to posttraumatic growth, social safeness and burnout.

Table 19 shows that fear-based trauma and social safeness are significantly negatively correlated in the Portuguese sample whereas in the UK sample, they are significantly positively correlated. While the latter result is less typical, it may reflect the difference between samples regarding the participants’ primary workplaces. As referenced earlier when discussing Table 11, 84.7% of UK HCPs and only 11.3% Portuguese HCPs worked in an ICU. ICUs are known for nurturing greater connectedness and camaraderie between staff due to the high-pressure nature of the workplace (Defilippis et al., 2020; Mealer et al., 2012; Baran et al., 2023; Walsh, 2020; Montgomery et al., 2021) in comparison with community and general wards, possibly explaining the difference between each country’s results.

We invited people to write about and reflect on their experiences in their own way by posing four open-ended questions. The number of participants who chose to answer these questions was very inconsistent, with as low as 34% of all participants answering one of them. Hence, we only comment briefly on some of the themes that emerged, most of which are complementary to the quantitative data above.

To try to explore the sense of change in HCPs’ experiences, we designed the survey such that people could reflect on what they remember doing or feeling and what they are doing and feeling now. Of course, one cannot assume that memory is always accurate but it offers impressionistic data of change over time. As can be seen from the data, some experiences and symptoms have changed over time, occasionally by over 50% but for other experiences less so. Regarding Table 1 and primary emotions, there has been a substantial drop in anxiety, anger, depression and sadness but even so the rates are still quite high. For example, retrospectively 83.5% of UK participants scored high to medium for anxiety which dropped to 47.2%. Although that is under 50%, it still means that anxiety is quite highly endorsed in this group and we can see similar patterns for the other emotions too.

In Table 2, relating to the nature and content of flashbacks, we see a similar story that there has been some drop in the emotional intensity of the flashbacks but still relatively high numbers of people are experiencing them at the time of the study.

Table 3 explores the spectrum of experiences and here we can see that at the time of the study the distress from the unpredictability of the pandemic had dropped, but in the UK sample was still at over 70%. Notable too is the fact that distress linked to feeling emotionally numb hardly dropped at all.

Table 4 suggests also relatively little change in self-evaluations, indicating that these may be problematic ways of self-relating that are relatively stable and may need specific attention.

From Table 5, regarding the emotions that people want help with, it seems to be ‘finding joy’ that is particularly salient, with nearly half the group seeing this is an issue. Indeed as Fonzo (2018) highlighted, not enough attention is paid to issues of positive affect and the recovery of positive affect after these kinds of traumas. And this is interesting given the data on feeling numb noted above.

Regarding data from Table 6, although many of the items for empathic distress have dropped over time, they are still quite high. For example, even at the time of data collection, empathic distress is 74.1% for the UK sample.

Exploration of the other items in Table 6 show similar processes where change has occurred but nevertheless some areas remain quite high. What we do not know is the degree to which this is standard for people working in these high stress areas or whether scores are elevated as a result of COVID. If it is the former then we require more detailed studies on these different aspects of experiences of working in these clinical areas and how to support and help people because some of these experiences are not only distressing but are affecting quite high numbers.

We have argued that trauma experiences which are associated with a sense of tragedy are different to those that are fear-based and need to be included in conceptualising, narrating and working with certain forms of trauma. Our data seems to support this. For example, in Table 15 we show that there is no relationship between tragedy-based trauma and fear-based trauma on the IES-R in the UK sample, and only small correlations in the Portuguese sample. Furthermore, in Table 16 we found that tragedy-based trauma was significantly correlated with anxiety, sadness and depression across both samples, whereas in the UK sample, fear-based trauma showed a smaller correlation with anger. There was a similar finding when we looked at the emotions that textured flashbacks. Here, again, we find that sadness consistently textures tragedy-based trauma but shows no correlation with fear-based trauma in the UK sample.

In Table 18, we see that the degree of support individuals were receiving do not seem to be associated with symptoms of fear-based trauma, but with the ability to experience sadness (particularly empathic distress). Importantly, posttraumatic growth was also linked to the support people received as they were going through this experience, which corroborates findings from the general population during COVID-19 (Xie and Kim, 2022) and HCPs (Mo et al., 2022; or a review, see Henson et al., 2021). Feeling socially safe and connected was also importantly linked to a range of support from others. In terms of emotions, the support of others had small effects on anxiety, anger and distress. Forms of support increase the ability to acknowledge and experience some of these emotions. It may be that close relationships facilitate safeness for open discussion and greater awareness of one’s emotions. This fits with the finding that attachment history enables empathic engagement with client distress (Wei et al., 2024).

Finally, we looked at the relationships between tragedy-based trauma, fear-based trauma, posttraumatic growth, social safeness and burnout (Table 19). From our data, it seems to be consistent that tuning into sadness and thinking about it as tragedy-based trauma was strongly linked to posttraumatic growth and feeling socially safe, whereas symptoms of fear-based trauma were not.

The different types of traumatic events are experienced in different ways. Our study was stimulated by considering that while the issue of fear, hypervigilance and flashbacks along with other dimensions of anxiety and safety are central to trauma, conceptualising some traumas as tragedies opens new dimensions of narrating and conceptualising experiences of suffering. Indeed, the experiences of HCPs who choose to engage with suffering are more nuanced and focused on the issues of empathic distress, sadness and grief (Behan and Kelly, 2025).

While exposure therapies and enabling people to ‘feel safe in the body’ (Van der Kolk, 2014) can help with the tolerance of the fear aspects such as the emotional avoidance of trauma, and safety, processing grief can also be an involved process and requires a ‘safe and sharing relationship’ to enable it (Behan and Kelly, 2025; Gilbert H., 2022; Gilbert, 2022a; Harris and Ho, 2023; Harris and Machado, 2017; Horwitz and Wakefield, 2007). The focus of grief can also vary in ways that we have not explored here. For example, grief can involve a recognition that one cannot go back to being the person one was before, particularly if these kinds of traumas have disrupted one’s conceptualisation and model of the world and oneself in it; called the assumptive world (Harris and Ho, 2023). Traumas can significantly disrupt people’s beliefs about the world (Janoff-Bulman, 2010). They can lead some people to lose (for example) their spiritual or religious beliefs, which can also give a sense of now being separate and different from a group they previously identified with (Leo et al., 2021). People can also lose a sense of connectedness from those who have not been through the same experience, increasing the sense of disorientation and separateness. These are also personal tragedies of loss. Hence the general outcome of this study supports the value of further research in the conceptualisation, nature and treatment of experiencing trauma as tragedy.

Due to various issues raised by ethical committees, our original research agenda was delayed. In addition, we were not able to access COVID-19 HCPs directly, despite efforts particularly by one of the authors. For this reason, HCPs were recruited via charity support groups for COVID-19. As time had progressed, we were past the peak of the crisis and therefore chose to explore people’s reflections on what they felt during the peak and what they were feeling at the time of gathering the data. Also, recruiting primarily through these charities might have led to a selection bias that favoured ICU HCPs. Clearly this is not ideal, but nevertheless we think this study provides a useful platform for subsequent research.

More accurate data on the people involved would have been useful but that was difficult to obtain, again indicating improvements for subsequent studies. This study has not looked at gender differences or differences in age and clinical experience which may be important mediators of personal experiences. Some respondents felt the scales were too long and we were too ambitious. We agree with this, although we faced a dilemma because we knew the time window for getting this data was closing. Retrospective data on trauma was not collected using the IES-R (Weiss and Marmar, 1997) and data only represented participants’ experiences at the time they completed the study. This is not consistent with the past (‘Did’) and present (‘Do’) data gathered in the Multicomponent Survey of Experiences of Tragedy-Based and Fear-Based Trauma, and therefore we were unable to assess change scores for these factors.

The present study utilised the survey method which enabled us to explore a diverse range of subtopics. This is a versatile method that allows the gathering of a large amount of data from diverse groups (Aiken et al., 2023; Greene et al., 2021; Newman et al., 2022). Although our survey items were directed by feedback from clinicians and may inspire future studies, the survey remains open and relatively loose. As raised by one reviewer, it is important to note these limitations and to consider the next stages of research. This would be to develop new insights and therefore measures of tragedy-based trauma. Further qualitative research could be used to explore changes in experience over time. For example, as fear becomes less of a focus, grief and the sense of tragedy can become more prominent in experience. Greater clarity about the textures of tragedy-based trauma could help guide clinicians to variations in intervention. For example, exposure to fear and avoided mental events may not be so helpful for tragedy-based work, whereas grief and anger work might. This is one of balance, not ‘either or’ and a research endeavour. Hence, there are many areas for improvement and development for this kind of research. However, we hope it is robust enough to establish some key principles in terms of trying to distinguish between the complexity of people’s personal experiences beyond diagnostic categories. It also highlights that the older concepts of tragedy, with particular focus on issues of loss, sadness and grieving, remain salient areas of research along with those of the fear-based focus of trauma. This links us to how we all have to deal with the harsh realities of disease, decay, and death of ourselves and others.