Depression, post-traumatic stress, anxiety, and fear of COVID-19 in the general population and health-care workers: prevalence, relationship, and explicative model in Peru

This triad involves short-term or medium-term emotional responses such as the fear and anxiety response to a highly stressful situation (i.e., a pandemic). Fear is defined as the fear or aversion response to a concrete, known or defined situation (i.e., the possibility of catching the virus) [10]. In contrast, the anxiety response is a non-specific, diffuse sensation that is not necessarily associated with concrete or well-defined element (i.e., the context of the pandemic) [10]. Fear, anxiety, and a highly stressful event are elements that correspond to the fear circuit that could trigger acute stress and subsequent post-traumatic stress [10]. Acute stress usually begins immediately after the highly stressful event and lasts from 3 days to 1 month; however, post-traumatic stress can manifest as early as the first month after the event. Therefore, both persons directly exposed to the virus (i.e., healthcare workers) and persons exposed but not exposed but anxious about the uncertainty of the future, the possible infection of family members, and the avalanche of news about COVID-19 (i.e., the general population and healthcare workers) [11], could trigger a state of post-traumatic stress disorder resulting from a highly stressful event such as facing the first few months of the pandemic.

Although there is ample evidence about the triad of fear, anxiety, and post-traumatic stress, it is not understood how this triad influences the presence of depressive symptoms. Fear of COVID-19 is weakly related to depressive symptoms but more strongly related to anxiety and post-traumatic stress [5]. However, there are several possibilities as to how depression can be explained by the triad of fear, anxiety, and post-traumatic stress (see Fig. 1b-e). Therefore, this study raises four possible hypotheses (hypotheses B, C, D, and E) that could explain the role of depression within this relationship.

In many cases, post-traumatic stress is not the only condition resulting from traumatic experiences (i.e., living in a pandemic), but also the onset of other comorbid conditions, including depression, somatization, or physical problems [12]. Longitudinal studies have identified that post-traumatic stress predicts depression [13, 14], so the hypothesis that post-traumatic stress influences depression is plausible, considering that this stress is preceded by the fear of COVID-19 and by anxious symptoms [5].

This model assumes that the anxious and depressive symptoms are not directly related, but together with post-traumatic stress, they are mediators. However, there is abundant evidence that anxiety and depression are closely related [15, 16]. It is feasible that the force of the relation between anxiety and depression diminishes if post-traumatic stress acts like a mediator [17]. This could support the hypothesis, but it is necessary to prove it with data.

In hypothesis B, the relationship between post-traumatic stress and depression was justified, but in this model, as evidence indicates, post-traumatic stress is considered to be a mediating factor in the relationship between anxiety and depression [18]. Some studies have also identified that post-traumatic stress can mediate depression with other mental health problems. Studies in refugees exposed to different forms of trauma indicate the mediating effect of post-traumatic stress on some mental disorders, such as depression, substance abuse, and personality disorders [19]. In addition, post-traumatic stress can also act as a partial mediator of the relationships between trauma and the severity of depression and between trauma and general mental functioning [20]. Likewise, there is abundant evidence that anxious and depressive symptoms are strongly correlated with each other [15], even in the context of COVID-19 [16]. Therefore, it is suggested that this may be a tentative model of the relationship between depression with fear, anxiety, and post-traumatic stress. For our study, we considered the mediating role of post-traumatic stress since the evidence suggests that this variable behaves as a mediator and not as a moderator.

This model assumes the same assumptions as to the previous hypothesis (model D) but considers that fear of COVID-19 and depression are directly related. It should be noted that although fear of COVID-19 and depression in bivariate analyses have found a positive but low correlation [5], this correlation may increase if mediated by factors such as anxiety or post-traumatic stress. Additionally, by considering all possible relationships, it is possible to see more clearly which dimensions are more or less related.

A better understanding of the relationship of the triad of fear, anxiety, and post-traumatic stress with the appearance of depressive symptoms will allow the identification of how these variables would trigger emotional problems during the context of the COVID-19 pandemic. An important element to consider is that the model could be affected if the groups evaluated have a higher exposure to or knowledge about the virus. Therefore, it is necessary to identify whether the models evaluated are equivalent among health-care workers and the general population. Because the former has greater knowledge of the treatment and evolution of the virus and greater average exposure to the virus due to the nature of their work, the relationships between the variables of the triad with depressive symptoms could be affected.

Therefore, there are two objectives related to the previously presented hypotheses. The first is to determine if “hypothesis A” about the relationship between the fear of COVID-19, anxiety, and post-traumatic stress fits adequately with the data collected from the general population and health-care workers (see Fig. 1a). The second is to determine which of the hypotheses presented above best explains the relationship between depression and the triad of fear, anxiety, and post-traumatic stress in both groups (see Fig. 1b-e). Our study considers a differentiated assessment of the general population and healthcare workers as they are exposed to different concrete and unspecific conditions of fear, anxiety, and stress.

A cross-sectional study was conducted that included the use of self-reported questionnaires. An online survey was used to avoid physical contact and the spread of SARS-CoV-2 among participants. Demographic and social data of the participants were obtained. Anxiety, fear of COVID-19, depression, and post-traumatic stress were measured using validated questionnaires and scoring systems. Data collection took place over 1 week, April 17–23, 2020, 1 month after the state of emergency was declared and mandatory self-quarantine was ordered in Peru. The sampling was not probabilistic and we used networks of contacts through social networks and other digital media to circulate the online survey.

Information was collected from adults living in Lima, the capital of Peru and the most populous city in the country. Inclusion criteria included: 18 to 80 years of age and an agreement to participate in the online survey. The participants were divided into two groups: the general population and health-care workers. Data collection sought to provide a sufficient number of cases from each group to perform the analyses (at least 150 participants per group) [21].

Data collection focused on collecting a sufficient number of participants for the general population and health professionals, so the questionnaire was socialized through social networking groups specific to health workers and through profile pages where there is a greater reach to the general population.

All analyses were done in R Studio, with the packages “lavaan,” “semTools,” and “semPlot.”

The study protocol and the instruments used for the evaluation were approved by the ethics committee of the Universidad San Martin de Porres (Oficio No. 227–2020-CIEI-FMH-USMP).

The characteristics of the participants are found in Table 1. The average age of health workers was 38.8 (SD = 11.2) and of the general population was 38.3 (SD = 13.2). The majority of participants were women. The overall prevalence for health-care workers and the general population was 16% for clinically relevant depressive symptoms, 11.7% for clinically relevant anxious symptoms, and 14.9% for post-traumatic stress. It was found that 23.1% of the participants had one of these mental health problems. A higher prevalence of depressive, anxious, or stress symptoms was identified in the general population (28.6%) compared to health-care workers (17.9%).

The relationship between depressive and anxious symptoms was high and very similar in the general population and health-care workers (r_s > 0.70). The general population presented a strong relationship between anxious symptoms and post-traumatic stress (overall score and three dimensions), while health-care workers reached a moderate relationship (r_s > 0.50) [39]. However, these values did not change much (see Table 2). All the instruments evaluated presented adequate levels of internal consistency.

A moderate relationship (r_s > 0.50) was found between post-traumatic stress (overall score and three dimensions) with fear of COVID-19 and depressive symptoms for both the general population and health workers. It should be noted that a small relationship was found between fear of COVID-19 and depressive and anxious symptoms in both groups (r_s > 0.30).

The “model A” or baseline model of the relationship of fear of COVID-19, anxiety, and post-traumatic stress showed adequate goodness-of-fit indices (see Table 3), indicating that this model is stable and could be used to evaluate more complex models to explain depressive symptoms in health-care workers and the general population. Therefore, models to explain depressive symptoms were evaluated using “model A” as a basis (see Fig. 1b-e).

It was identified that the four models evaluated (Model B to E) identified adequate goodness-of-fit indices in all cases. However, the models that most explained the depressive symptoms were model D (explaining 73.9% of the variance) and model E (explaining 74.8% of the variance). Although model E contributed 0.7% more variance, the relationship between fear of COVID-19 and depressive symptoms was − 0.104. This finding would be contrary to what is theoretically expected since it is estimated that the greater the fear of the COVID-19, the greater the levels of depressive symptoms are expected, as was found in the bivariate analysis, where these variables presented a direct, significant, and positive correlation (see Table 2). Therefore, it was decided that model D would best explain the depressive symptoms because it presents sufficient goodness-of-fit indices, explains almost the same amount of variance as model E, and fits best with the theoretical assumptions (see Fig. 2).

Analysis of model D for the total number of participants, the general population, and healthcare workers (see Fig. 2 and Fig. 3), identified that fear of COVID-19 is directly associated with anxiety (p < 0.001), and with the post-traumatic stress response (p < 0.001). Also, anxiety is directly associated with post-traumatic stress and depressive symptoms (p < 0.001). When the relationships between the variables in model D were evaluated by type population, it was found that the general population and health professionals had very similar values between the variables (see Fig. 3). However, some variables are no longer significant between the two groups. On the one hand, age in health professionals does not associate with anxiety levels (β = − 0.065; p = 0.479), while in the general population it does (β = − 0.337; p < 0.001). On the other hand, post-traumatic stress is not associated with the presence of depressive symptoms in the general population (β = 0.122; p = 0.025), but not in healthcare professionals where it presents a very small and non-significant coefficient (β = − 0.015; p = 0.876).

Mediation analysis identified that in the overall participants the indirect effect of post-traumatic stress on the relationship of anxiety and depression was significant (β = 0.10; 95%CI = 0.05 to 0.16). However, the results varied depending on the population assessed. In the general population, post-traumatic stress mediated the relationship between anxiety and depression (β = 0.12; 95%CI = 0.06 to 0.18), but the mediation was not significant in health care workers (β = 0.03; 95%CI = − 0.11 to 0.19).

A sub-analysis was conducted to assess the role of other sociodemographic variables such as sex and age with COVID-19 fear, anxiety, and post-traumatic stress variables. It was identified that only significant relationships were between fear of COVID-19 and age; post-traumatic stress and sex; and age and anxious symptoms.

Our study proposes and tests different models based on structural equations models that allowed testing hypothetical models based on substantive theories that use the triad of fear, anxiety, and post-traumatic stress to explain depressive symptoms in the general population and health-care workers during the COVID-19 pandemic.

We found that this triad, along with three additional variables (number of COVID-19 symptoms, age, and sex), explained more than 71% of depressive symptoms in the general population and health care workers. The proposed model (model D) allows us to understand the role of fear of COVID-19, age, and the number of perceived COVID-19 symptoms in the presence of anxiety symptoms. In turn, it is found that anxiety symptoms and sex affect the stress symptoms, and all of these are associated with depressive symptoms. Anxiety symptoms are the most influential in the occurrence of depressive symptoms, in compassion to stress. It was identified that the model presents a different behavior for the general population and health-care workers since in the latter group the variables of age and post-traumatic stress are not associated with depressive symptoms.

This explanatory model can be used in health contexts and populations for how emotional factors (fear, anxiety, and stress) can affect depressive symptoms. Also, the model can be used to understand a differential response of health-care workers and the general population in the genesis of depressive symptoms.

Analysis of the triad of fear-anxiety-stress to predict depressive symptoms identified that only anxiety is strongly associated with depressive symptoms, as PTSD has a small effect, and there is no direct relationship with fear with COVID-19. However, our study was supported by the fact that this triad mechanism represents a joint response to emotionally charged events such as a pandemic. Therefore, we consider that its assessment as a whole (fear, anxiety, and stress) may better explain the presence of depressive symptoms.

Also, there is a higher prevalence of depressive, anxious, and stress symptoms in the general population than in health workers. Therefore, it is necessary to focus on mental health interventions and prevention activities for both groups.

Previous studies show the key role that organizations and public health bodies play in promoting adaptive coping and reducing health worries and the emotional and psychological distress caused by the pandemic. Evidence highlights particular groups at risk of developing mental health problems (contact with infected patients, having children), and time points where risk may increase (initial response phase, when quarantined) [58]. Our study raises three main implications for public and global health in Peru and other Low and Middle-Income Countries (LMIC) with similar characteristics. First, the high prevalence of mental health problems recorded during the pandemic [3] makes it necessary to establish national policies and strategies for screening and epidemiological surveillance of the components of the triad of fear of COVID-19, anxiety, and post-traumatic stress, since these three elements predict the presence of depressive symptoms and other emotional problems. Health-care workers from Peru and other LMICs are overburdened by the pandemic and are in a fragmented health system [59]. It is recommended that technological tools such as apps or short (2 or 4 items) virtual self-reporting systems be used to collect information on emotional problems (i.e., anxiety, stress, fear, or depression) from the general population and health workers. These strategies have proven useful for addressing mental health needs and referring users with severe emotional problems in other countries during a pandemic emergency. Second, it is recommended to develop and implement preventive activities focused on the three elements of the triad (fear of COVID-19, anxiety, and post-traumatic stress). It is not only necessary to provide care to people who have moderate or severe mental health problems (depression, anxiety, or post-traumatic stress) but also to develop strategies aimed at people with mild and non-specific mental health indicators such as stress or fear of COVID-19 [60]. These strategies could prevent the latter individuals from evolving to a more advanced stage involving significant health problems and generating years of life lost due to disability [60]. Third, in Peru and other LMICs, the treatment gap for mental health problems such as depression is high, and it is estimated that only 1 in 10 people who require treatment receive it [44]. This gap may have increased during the pandemic, so actions to increase access to appropriate care need to be targeted. This model can be used to identify cases with a high fear of COVID-19 and facilitate their access to the health care system so that cases with greater severity of depression do not develop in the future.

At the level of mental health decision-makers, increased resources in mental health care are strongly recommended. In Peru, only 2% of all GDP is allocated to health, and only 0.2% of GDP is allocated to mental health [61], thus requiring increased human, financial, and political capital resources to improve mental health during and after the COVID-19 pandemic.

Our study presents and assesses different models for understanding depressive symptoms using the triad of fear, anxiety, and stress. Our model applies a solid framework on the influence of this triad on mental health and applies it in the presence of depressive symptoms [6‐9]. Another strong point of the study is the use of statistical methods that consider all these variables within their analysis. However, our study is not free of limitations. First, this cross-sectional study was conducted during the beginning of the first wave of COVID-19 in Peru (April 2020), so this model could vary during the later stages of the pandemic. Second, the data are from an urban city with the highest population density in the country (Lima, the capital of Peru), so the results could vary in a rural or less densely populated context (risk of infection). Third, the prevalence of depressive, anxious, and stress symptoms was evaluated with validated psychometric instruments, but this evaluation is not a substitute for clinical assessment, so misclassification can be a problem. Fourth, the scale for measuring PTSD (IES-R) is based on the criteria of the DMS-IV, which proposes three dimensions (intrusion, avoidance, and hyperactivity). However, the criteria for defining PTSD is modified in the DSM-5 where four dimensions are proposed (Reexperiencing, Avoidance, Negative Alterations in Cognitions and Mood, Hyperarousal), therefore, it is possible that the PTSD construct is being partially measured. This is not the case in the instruments used to measure depression and anxiety since the DMS-IV and DSM-5 criteria are equivalent. Fifth, other variables that could be useful to explain the model such as intolerance of uncertainty [53], pregnancy in the female participants [50], anxiety caused by COVID-19 [50], or income level [16] could not be included. Sixth, our study is a cross-sectional study so that causality statements cannot be made only about the association. Seventh, no differentiation was made between front-line staff working directly with COVID-19 patients and other types of health professionals who may be doing remote work. Thus, it is possible that not all health care workers had the same level of exposure to the virus.