Main findings and significance of the results
There is extensive discussion on the role of biological and psychological variables in the occurrence of depressive symptoms in patients with COVID-19. Previous studies found higher values in inflammatory markers (i.e., NLR) in patients with depression, compared with the non-depression groups [25, 57]. Our research presents evidence that a model about subjective perception of symptoms (psychosomatic and COVID-19 symptoms), as well as anxiety and post-traumatic stress explains the presence of depressive symptoms in hospitalized COVID-19 with and without severe inflammatory response in NLR.
Using the anxiety-PTSD-depression triad as a base model, we added the variables related to the subjective perception of symptoms, and finally our model explained 85% of depressive symptoms in the overall sample. This demonstrates that several mood disorders occur at the same time prior to the development of depression .
The model also showed that psychosomatic symptoms and anxiety are the variables with the highest influence. These both variables can be explained from biological perspectives due to an increase of anxiety has been found to be associated with presence of somatic symptoms such as headaches and shoulder and limb pain . Besides, the relationship between psychosomatic symptoms and post-traumatic stress represents a relevant effect, which can be explained due to somatic symptoms being more prevalent during periods of stress .
Contrasting findings with existing literature
Prevalence and indicators
In our study, the most prevalent mental health problems were anxiety (11.2%), depression (7.1%) and PTSD (6.1%). These results are below those reported in previous studies. A systematic review indicated a prevalence of depression at 45% (95% CI: 37–54%) and 47% for anxiety (95% CI: 37–57%) in COVID-19 patients . In the case of PTSD, a systematic review, which evaluated 13 studies in 1093 participants with severe cases of COVID-19, reported a prevalence of PTSD at 16% (95%CI: 9% to 23%) . Similarly, another study conducted with 190 patients indicated that the prevalence of PTSD was 22.6% . These differences among researchers could be explained due to the different socio-demographic compositions, different study designs, and measurement instruments used, which may influence the degree of prevalence. Even so, results evidenced that the patients with COVID-19 present several mental health problems at the same time. This could mean that, during the treatment, patients may develop multiple related psychiatric diseases which form a mutually influential symptom network; in turn, influences their recovery .
Sleep problems were one of the most frequent indicators of mental health problems. Although a systematic review reported that sleep problems are common during the COVID-19 pandemic, being associated with higher levels of mental health problems . Another study showed that this indicator had been prevalent in healthcare workers even before COVID-19 . The prevalence of sleep problems could be explained by fear of COVID-19. Due to worries about the disease, patients cannot take rest and consequently develop insomnia . Moreover, if the individual cannot manage the fear for a specific time, it leads to mental health problems (e.g. depression, anxiety) . We also found that low energy was the second indicator with the highest prevalence, which is a comorbid symptom of many psychiatric problems in patients with COVID-19 . In addition, low energy could be caused by lack of sleep or one of the other mental health problems such as depression.
The main indicators related to anxiety were nervousness, irritability and worry. This finding was also reported in other studies in people with COVID-19, being the most common symptoms of anxiety, insomnia, irritability, restlessness, and excessive worrying . As previously mentioned, being hospitalized, patients may experience fear of COVID-19 and worries related to their health, family, financial issues, and environmental conditions (e.g., isolation, uncertainty about the evolution of the disease). On the other hand, for somatic complaints, backache, arms and legs pain, and feeling exhausted were the most prevalent in our sample. Reasons might be related to similarities with the physiological symptoms of COVID-19.
Structural equation model and relationship between variables
Previous studies that identified pose predictive models between mental health problems in COVID-19 (i.e., depression, anxiety, stress, fear of COVID-19, among others) indicated that concern of COVID-19 has a negative impact on some mental health. Two studies proposed models in which fear of COVID is significantly and positively related to depression, anxiety and insomnia [74, 75]. Moreover, studies that reported the influence of biological responses (i.e., C-reactive protein, neutrophil/lymphocyte ratio, monocyte/lymphocyte ratio, among others) during COVID-19 on mental health problems reported COVID-19 virus could lead to system immune changes, which in turn could reflect in mental health problems. These psychiatric outcomes can be influenced by other factors as well (i.e., biological, factor social isolation, adverse effects of treatments, among others) [12, 76]. However, a small number of studies has proposed predictive models about the relationship of biological responses with these mental clinical problems. For example, a study, using SEM, proposed a model that evening salivary cortisol (as an indicator of Hypothalamic pituitary adrenal) predicts depression, which predicts circulating pro-inflammatory cytokines (IL-2, IL-6, TNF-α) in patients diagnosed with Chronic Fatigue Syndrome (CFS) . Another study explored the relationship between biological factors (i.e., sex, disease duration, self-perceived illness severity, and inflammatory markers) and mental health status in inpatients with COVID-19. In the SEM, inflammatory markers (i.e., NLR, IL-1β as observed variables) and mental health (i.e., insomnia, depression and anxiety as observed variables) were set as latent variables. Results indicated the inflammatory markers had a significant and direct effect on mental health. Moreover, the disease duration and inflammatory markers indirectly influenced mental health, through self-perceived illness severity as a mediator . Although these models are not the same as ours, these findings suggest that inflammatory responses could be related to psychological disorders.
Following this hypothesis, studies have found a heterogeneous influence of NLR in psychological mental problems. First, one study, using regression analysis, demonstrated the influence of NLR markers in both the prevalence of depression and anxiety in Chinese patients with gastric cancer . In contradiction to this finding, a multi-linear regression study showed a weak association between inflammatory biomarkers and depression in a three-month cohort of stroke patients . Reasons could be related to there are immune responses in both COVID-19 infection and mood disorders, suggesting a similar biological response between them. Both states induce the production of abnormal levels of cytokines, chemokines, and other inflammatory mediators , showing a hyperinflammatory state . While patients with mental health problems show high levels of biomarkers , a meta-analysis, with 16 studies, evidenced higher counts of biomarkers (i.e. IL-6, CRP, PCT, among others) in severe cases of COVID-19 .
Another interesting result was the high influence of anxiety on depression in all three models. This finding is in accordance with other studies which evidenced that anxious symptoms had a direct and significant relationship with depression. One study that proposed a model of the triad fear-anxiety-stress in the development of depression symptoms in pandemic disease symptoms in health-workers, indicated that the fear to COVID-19, anxiety and posttraumatic symptoms explains depression symptoms. The SEM demonstrated that anxiety was the most influential variable in the depression symptoms in comparison with post-traumatic stress . Even before COVID-19 exploded, researchers have shown that anxiety may contribute directly or as mediating variables in depression. These results show how the different variables (i.e. stress, self-esteem, stressful negative events) influences depression, where increases in anxiety may lead to increases in depression [83, 84]. To sum up, these findings suggest that the role of anxiety in the occurrence of depressive symptoms is significant and is even maintained in the COVID-19 pandemic. Anxiety is a common adaptive response against threatening situations, which could be increased due to factors such as stress or fear, and could trigger prolonged anxiety. Thus, pathological anxiety can affect functioning in the daily routine of patients, which in turn may cause or be comorbid with other mental disorders such as depression .
Besides, another result was the influence of the PTSD variable on depression. Our results demonstrated that PTSD does not present a significant influence on depression in hospitalized patients both with and without severe inflammatory markers. This finding might be related to the PTSD symptoms changes over time. Other studies have found the different prevalences of PTSD symptoms in each stage of COVID-19 disease (i.e. recovering from COVID-19 infection, being quarantined) [86, 87]. Likewise, another reason could be the similarity between our variables. There are studies that report a high association between PTSD and somatic symptoms, whose findings support that somatic symptom may be related to the patient’s psychophysiological dysregulation and lead to psychological symptoms (e.g. PTSD) [88, 89].
Implications in public health and making decisions
These findings provide a theoretical model, which permits establishing policies to prevent depression in inpatients. Specifically, the model revealed that somatic and anxious symptoms are the most relevant predictors to develop depression. Health workers could employ screening measures for anxiety and somatic symptoms in order to prioritize the care of patients with high levels in these conditions, and thus avoid possible cases of depressive symptoms. It is a necessity because Peru is one of the countries that reported worse mental health levels than overall average during the pandemic  and the prevalence of depression in 2020 was five times higher than previous years .
Interventions to reduce symptoms of anxiety, fear and worry in hospitalized patients could be an effective strategy to prevent subsequent cases of mental illness . Telephone-based intervention also has been useful to reduce symptoms of anxiety and depression, which provides psychological support, information about the process of the disease and promotes a sense of emotional stability [93, 94]. Thus, the implementation of telephones during hospitalizations could be a strategy to prevent psychological problems in hospital isolated patients. This also could be used as a facility for patients to have access to make calls or send messages to their relatives.
Strengths and limitations
This study has limitations that should be mentioned. First, some patients did not have inflammatory markers recorded, so they were eliminated. This elimination could lead to an information bias. Second, NLR was evaluated as the only inflammatory measure, however this is not a gold standard so it could cause errors in grouping people with and without severe inflammatory response. Third, this study has a cross-sectional design and we cannot infer causality in the interpretation of the findings. Fourth, we employed self-reported measures, which could be influenced by social desirability or memory bias. Fifth, the data includes a single hospital in a Peruvian city, so the results should not be extrapolated to other cities or contexts. Sixth, we used a validated scale such as the IES-R to measure PTSD, however the IES-R apparently does not include the entire concept of PTSD, since the Diagnostic and Statistical Manual of Mental Disorders (DSM–5) considers four dimensions and the IES-R only assesses three of these dimensions. This could imply a partial evaluation of the symptoms of PTSD. Finally, other confounding variables were not considered, such as fear of COVID-19 ( and coping , so it is possible that the model is partial or influenced by other variables.
On the other hand, our study has three main strengths. This investigation presents a larger sample compared to previous studies evaluating hospitalized patients [25, 96]. Besides, we employed structural equation modeling, as a solid technique that allows us to assess several variables at the same time. Moreover, to our knowledge, this is the first study that provides a framework of biological and psychological variables that explain depressive symptoms as an outcome in the context of the COVID-19.