Mental health in COVID-19 survivors from a general hospital: association with sociodemographic, clinical, and inammatory variables

Background: The current COVID-19 (coronavirus disease 2019) pandemic constitutes a signicant problem for the world's public health and generates mental health problems. Objective: To describe the characteristics of mental health in survivors of COVID-19 and the main sociodemographic, clinical, and immune factors related. Method: A Cross-sectional and correlational study was conducted on 318 patients (196 women, mean age 54.4 ± 15.1 years) surviving COVID-19 from one hospital in Peru in which sociodemographic, clinical, and immune characteristics were explored. Through telephone interviews, an evaluation of the presence of depressive, anxious, somatic, and distress symptoms was carried out using standardized scales. Adjusted prevalence ratios (PRa) were estimated. Results: A signicant proportion of the patients have depressive (30.3%), anxious (29.9%), somatic (33.7%), and distress (28.7%) symptoms. In the regression analysis, the variables associated with a higher frequency of clinically relevant mental symptoms were female sex (depression: aPR = 2.29; anxiety: PRa = 2.71; somatic symptoms: PRa = 2.04; distress: PRa = 2.11), proceeding outside the capital (depression: PRa = 1.61; anxiety: PRa = 1.53), the self-perception of a greater severity of the infection (depression: PRa = 5.53; anxiety: PRa = 2.29; distress: PRa = 14.78), the presence of persistent COVID-19 symptoms (depression: PRa = 8.55; anxiety: PRa = 11.38; somatic symptoms: PRa = 5.46; distress: PRa = 20.55), a history of psychiatric treatment (depression: PRa = 2.29; somatic symptoms: PRa = 2.90 ; distress: PRa = 3.80), the history of a family member infected by COVID-19 (anxiety: PRa = 4.71; somatic symptoms: PRa = 1.99), and a neutrophil-lymphocyte index greater than 6.5 (depression: PRa = 1.67; anxiety: PRa = 1.82). Conclusion: COVID-19 survivors show a high prevalence of negative mental symptoms. Some useful variables have been found when identifying vulnerable patients requiring psychiatric care.


Introduction
The current pandemic of the novel coronavirus disease , caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), represents a signi cant problem for world mental health (1). Until January 12, 2021, worldwide, more than 89 million cases and 1,940,352 deaths were reported (2). In Peru, during the same period, more than a million con rmed cases and 38,335 deaths were reported, with lethality of 3.70% (3). The great infectivity, the harmfulness of the virus, the daily increase in the number of con rmed cases and deaths, Due to the scarce information regarding the impact of this novel pandemic on mental health in patients surviving COVID-19, we decided to carry out this study, which aims to describe the characteristics of mental health and the main sociodemographic, clinical, and immune factors related to the disease.

Study design
This study is observational, analytical, and cross-sectional. It evaluates the in uence of sociodemographic, clinical, and immune characteristics on the levels of depression, anxiety, somatic symptoms, and distress in outpatients who survived COVID-19.

Clinical context
This study was carried out at Hospital Nacional Guillermo Almenara Irigoyen (HNGAI), which is the second-largest hospital of the "Seguridad Social de Salud del Perú" (EsSalud), with a total of 815 hospital beds. Furthermore, it is a reference center of the third level of attention and has all the medical specialties. By 2019, the Almenara network met the health needs of 1,634,990 insured (15).
During the COVID-19 pandemic, HNGAI was a national referral center for the care of COVID-19 patients. Due to the high demand for care, hospital beds of different specialties had to be redistributed to care for these COVID-19 patients. Moreover, new hospital environments had to be created. The diagnoses of COVID-19 were made with serological tests, which were later con rmed with molecular tests.
According to EsSalud's records, from the beginning of the pandemic until September 2020, in HNGAI, there were a total of 3,238 hospital discharges of patients diagnosed with COVID-19. For this study, minors (n=369), deceased (n=843), escaped (n=1), referred to another center (n=88), voluntarily withdrawn (n=13), as well as those patients who had two or more hospitalizations during March to September 2020 (n=14), were excluded. Hence, we only considered patients who were discharged from the hospital (n=1,910).

Participants
The sample was obtained from a population of 1,910 patients with COVID-19 who were discharged from HNGAI hospitalization services between March and September 2020. We used the Paz et al. study (16) to calculate the sample size. In this research, an expected frequency of 55% of mental health problems in patients with COVID-19 was estimated. With this data, and considering a margin of error of 5%, with a design effect of 1 and a single group, we obtained a total of 318 individuals with a 95% con dence interval.
The sample was selected through a simple random sampling using the Epidat v4.2 program (Dirección Xeral de Saúde Pública da Consellería de Sanidade, Galicia, España). Each of the 1,910 patients had a coding.
The variables were collected on a virtual le through telephone calls to the patients between October 22 and November 28, 2020. The cell phone number that was registered in the electronic medical record of each patient was used. The calls were made by the co-investigators, who are psychiatrists with clinical and research experience. If after two calls the selected participant did not answer, they were excluded from the study, and we then made a new random selection of another participant. In addition, a group of selected participants did not wish to continue with the interview (n=8). Therefore, they were removed from the database, and new participants were randomly selected.

Further variables
In addition, information regarding the following variables was collected: a) Sociodemographic data, including sex, age, educational degree, job status, place of origin before hospitalization, current place of residence, with whom they live, history of a relative infected and/or death by COVID-19; b) Clinical and hospitalization data, including history of diagnosis and/or treatment for psychiatric diagnosis, self-perception of the severity of COVID-19, hospitalization area, days of hospitalization, days from discharge to the interview, presence of COVID-19 symptoms at the time of the interview; and c) Immune data, including neutrophil-to-lymphocyte ratio (NLR) and monocyte-to-lymphocyte ratio (MLR) at the beginning of hospitalization. Immune variables were obtained by reviewing electronic medical records.

Statistical analyses
The relative and absolute frequency of the categorical variables were described. For quantitative variables with normal distribution, the mean and standard deviation (SD) are described, and for those without normal distribution, the median and interquartile range (IQR) are described.
Then, the outcome variables of mental health, depression, anxiety, somatic symptoms, and distress were dichotomized into clinically irrelevant (normal-mild) and clinically relevant (moderate-severe). At the same time, hospitalization time was categorized into 1-7 days, 8-14 days, and more than 14 days.
In the dichotomous analysis, the association between each covariate with each dichotomized mental health outcome variable was evaluated.
Chi-2 and Fisher's exact tests were used for categorical covariates, depending on the ful llment of assumptions. For numerical covariates, Student's T-test was used when they had a normal distribution, and the Mann Whitney's U test was used when they had a non-normal distribution.
Prior to the regression analysis, the age variable was categorized into quartiles. Additionally, the NLR was categorized as <6.5 and ≥ 6.5 (26), and the MLR was categorized as <0.364 and ≥ 0.364 (27). The cut-off points were chosen considering their ability to predict in-hospital mortality in patients with COVID-19. The COVID-19 symptoms variable during the interview was categorized into asymptomatic (without any symptoms at the time of follow-up) and with symptoms (at least one symptom at the time of follow-up).
In order to calculate the Prevalence Ratio (PR) with their 95% con dence interval (95% CI), Poisson regression analyses were performed with their adjusted variances, using the service variable as a cluster. All regression models were adjusted (PRa) for follow-up time. Furthermore, the regression models of the covariates "Severity of COVID-19" and "COVID-19 symptoms during the interview" were also adjusted for the presence of COVID-19 symptoms at the moment of admission to hospitalization. A p-value <0.05 was considered as a statistically signi cant result.
Stata MP v.16.0 statistical software was used for all analyses.

Ethical aspects
Informed consent was requested from each participant verbally. This included an explanation of the objectives of the investigation as well as an explanation of the rights of the participants (anonymity and the right to refrain from participating in case they considered appropriate).
Furthermore, psychiatric help was offered when the interviewer considered it necessary at some point, either during or after the phone interview.
This investigation was carried out with the authorization of the EsSalud's Research Ethics Committee Speci c to COVID-19.

Results
In a representative sample of outpatient survivors of COVID-19, the majority were women (61.6%), professed a religion (92.8%), were from Lima (94.3%), with an average age of 54.4 years, lived with a partner, and/or children (80.5%), had at least one family member infected (83.0%), and at least one relative died (30.5%) from COVID-19. 31.1% of survey respondents perceived that their COVID-19 disease had been severe, and 33.7% perceived that it had been moderate. The median NLR was 6 (IQR: 3.8-11.1), and the MLR was 0.3 (IQR: 0.2-0.5). Regarding COVID-19 symptoms, 7.9% of survey respondents were asymptomatic at the time of admission to the hospital, while 39% were asymptomatic at the time of the interview. The median time of hospitalization was 8 days (IQR: 4-15), and the mean follow-up time was 100.8 days ± SD: 37.7.
In the bivariate analysis, women had a higher frequency of clinically relevant symptoms (moderate-severe) of the depressive type (p=0.009), anxious type (p=0.011), somatic type (p=0.011), and distress type (p=0.059). Patients with history or treatment of psychiatric illness presented a higher frequency of clinically relevant mental symptoms (p <0.05). Those patients who self-reported having had severe or critical COVID-19 had a higher frequency of clinically relevant depressive symptoms (p=0.004) and distress (p=0.009). NLR, in its numerical nature, was associated with clinically relevant depressive symptoms (p=0.037). Further associations can be found in Table 2.
In the regression analysis, it was found that women have a higher frequency of clinically relevant symptoms of depression (PRa=2.29; CI 95%: 1.06-4.92), anxiety (PRa=2.71; CI 95%: 1.62-4.53), somatic (PRa=2.04; CI 95%: 1.06-3.89), and distress (PRa=2.11; CI 95%: 1.05-4.24). On the other hand, those patients with a higher level of education had a lower frequency of having mental symptoms than those with no education or only a primary level. Likewise, those patients with a job or who are retired also have less frequency of having an adverse mental health outcome, compared to those who do not work. Those patients who had had a family member infected or who died by COVID-19, or those who have a history of psychiatric diagnosis, have a higher frequency of developing adverse mental health outcomes. Patients who self-reported having severe or critical COVID-19 are more likely to have clinically relevant depression, anxiety, somatic symptoms, and distress, compared to those who self-reported a mild infection. Having a high NLR (≥ 6.5) was associated with a higher frequency of depressive symptoms

Main ndings and meaning of the results
This study sought to describe the characteristics of mental health in Peruvian patients surviving COVID-19, as well as the main sociodemographic, clinical, and immune factors related to this. We found that mental symptoms were present after an average of 100 days after discharge of the patients. In a previous respiratory infection epidemic in South Korea, it was reported that mental symptoms might be present for up to a year after the epidemic outbreak (28).
The reported prevalence of depressive (30.3%) and anxious (29.9%) symptoms in COVID-19 survivors is similar to what was reported in other studies that include patients with mild COVID-19 (29,30). Notwithstanding, a recent meta-analysis has documented a prevalence of depression of 52% and anxiety of 47% (13), which is higher than that reported in our study. This difference may be because most of the included studies were conducted in hospitalized patients. The mentioned meta-analysis also reported that the prevalence of depression and anxiety was lower in outpatients (35% and 33%, respectively) compared to hospitalized patients (48% and 42%, respectively) (13). The prevalence of clinically relevant depressive symptoms (moderate-severe) (10.8%) reported in this study is higher than what was documented in the Peruvian population prior to the pandemic, in which a prevalence of 6.4% was reported using the PHQ-9 (31). This would suggest that the COVID-19 infection has some impact on mental health in the Peruvian population by increasing the prevalence of clinically relevant depressive symptoms.
We found a high prevalence of distress symptoms (28.7%) and somatic symptoms (33.7%). In a study of 34 patients with COVID-19 conducted in Italy, distress was found in 82% of the participants. However, after 4 months, there was a decrease in these symptoms up to 46.6% (32).
Regarding the somatic symptoms, we did not nd any research in which this variable was evaluated in survivors of COVID-19. Nonetheless, in a study carried out in the general population in China, during the peak of the pandemic, researchers found a prevalence of 45.9% of somatic symptoms (33), higher than what was reported by our study.

Variables associated with mental symptoms
Some sociodemographic variables had a relation to a higher prevalence of clinically relevant mental symptoms. Being a woman is one of the main variables, which is similar to what was reported in other studies (16,29,(34)(35)(36). These could probably be because women tend to have greater symptoms of hyperactivity, recurrent distressing memories, and negative cognitive and mood disturbances (37). Moreover, proceeding out of the capital before hospitalization was associated with a higher frequency of clinically relevant depressive and anxiety symptoms. This result was expected since leaving home, family, and community could have an impact on mental health (38). Furthermore, it is likely that those patients who were required to travel to Lima are also those with a more serious illness.
Patients who had family members who were diagnosed and died of COVID-19 had a higher prevalence of somatic and anxious symptoms.
This nding was also reported in other studies in China (30,39). Speci cally, in a study carried out in Wuhan in patients with COVID-19, having family members diagnosed and/or deceased by the same disease were independent predictors of both the depression severity index, as well as presenting higher anxiety scores (39). Likewise, a history of psychiatric diagnosis and treatment has been associated with a higher frequency of clinically relevant mental symptoms in COVID-19 survivors. This could be because the current pandemic causes reactive symptoms such as stress, depression, and anxiety, which in combination with hospitalization, can aggravate the mental health of people with a previous psychiatric diagnosis (40).
Self-perception regarding a greater severity of COVID-19 was a variable associated with a higher prevalence of mental symptoms in survivors of the disease. A study carried out in patients hospitalized for COVID-19 in Wuhan reported a similar result (35). This could be because the patients' concerns about their disease would be added to their psychological burden, which would be associated with anguish and a poor result in their mental health. Similarly, another study from Turkey reported that perception of the severity of COVID-19 infection was associated with the risk of having PTSD symptoms, although this association became non-signi cant after controlling the effects of other variables (41).
We found a high frequency of persistent COVID-19 symptoms (61%). This nding is similar to what was reported in a study conducted in Chinese patients who survived COVID-19, where it was found that up to 76% of patients reported at least one COVID-19 symptom after 6 months of follow-up (42). The persistence of these symptoms was associated with a higher prevalence of mental symptoms.

The immune system's role in mental symptoms
The presence of an NLR greater than 6.5 at hospital admission was a variable associated with a higher prevalence of clinically relevant depressive and anxiety symptoms. In a study conducted in patients who survived COVID-19, in their convalescent phase, researchers found that those who reported depressive symptoms showed a greater immune response evidenced by a higher mean of NLR (2.4 vs 1.8; p<0.001) (43). In another study from China, conducted with hospitalized patients, it has been documented that those with mental symptoms have higher levels of interleukins (IL) 1β, NLR, and lower levels of IL-10 and lymphocyte count (35).
When SARS-CoV-2 infects the respiratory tract, it can cause an acute respiratory syndrome with the consequent release of pro-in ammatory cytokines, such as IL-1β and IL-6, producing a "cytokine storm" (44). These cytokines may be increased in psychiatric disorders such as depression, schizophrenia, and PTSD (9). However, in some circumstances, it is not possible to make a study of these cytokines. Therefore, we can indirectly value their increase through an elevation of various in ammatory parameters with the NLR. This provides a quick and easy way to value the state of systemic in ammation, which can be calculated from a complete blood count. NLR elevations have been associated with an increase in cytokines, such as IL-6 and IL-8 (45).
Given the relationship between elevated levels of cytokines in COVID-19 as well as in psychiatric disorders, the immune/in ammatory pathways can be considered as one of the mechanisms involved in the mental health problems of this infection (9). Considering the impact of COVID-19 infection on mental health and the involvement of the immune system, it is necessary to evaluate the psychopathology of survivors of COVID-19. Consequently, the investigation of in ammatory biomarkers should be deepened in order to adequately diagnose, treat, and monitor emerging psychiatric conditions (36). Nonetheless, it is important to consider that related biological factors (such as advanced age, female sex, and excess fat), along with other factors inherent to COVID-19 (such as social isolation, nancial stress, and adverse effects of treatments), can in uence psychiatric outcomes. As a consequence, it is probable that the psychiatric symptoms in COVID-19 patients that were observed are due to a combination of the processes involved in the virus-host relationship and the psychosocial and therapeutic problems associated with the pandemic and the disease (9).

Implications for public health and decision making
This study evidences a high impact on the mental health of survivors of COVID-19. Hence, Peruvian Public Health must focus on the early diagnosis and treatment of the mental health problems of these patients. It is indispensable to evaluate all factors that in uence these problems, including sociodemographic factors, proper clinical factors, and immune factors. This allows health care professionals to establish individual management to improve the psychological well-being of the survivors.
On the other hand, it is suggested to implement health policies that aim to implement diverse mental health services. These policies include screenings with standardized online evaluations, educational intervention in mental health, provision of psychological support after the detection of vulnerable patients, assessment of the state of systemic in ammation using a complete blood count, adequate psychiatric care for mental health management, and timely management of emergency situation in public health. All of these measures will empower Peru in the containment and future eradication of the COVID-19 pandemic (46).

Strengths and limitations of the study
This study has strengths that include the size of the sample and the fact that a random sample was made, which ensures the results' representativeness of the prevalence of mental disorders in the target population at a hospital level. Furthermore, to our knowledge, the study provides rst-time exploratory data on different factors associated with mental health problems in patients that survived COVID-19 in Peru, during a follow-up.
Notwithstanding, this study must be understood in the context of its methodological limitations. Since the study is cross-sectional, we were neither able to evaluate a causal relationship nor how the different mental health outcomes evolve from hospital discharge to the moment of the evaluation. Future studies should evaluate how anxiety, depression, and distress levels are modi ed as the pandemic evolves in Peru. Since we only evaluate patients from one single hospital, we cannot generalize the results to all patients from other hospitals. Nonetheless, our ndings are in line with what was reported in other countries.

Conclusions
The results of this study indicate a high prevalence of mental symptoms in COVID-19 survivors from a third-level hospital in Peru. Female sex, being from outside the capital, self-perception of greater severity of the infection, the presence of persistent COVID-19 symptoms, prior psychiatric diagnosis or treatment, and peripheral in ammatory markers have been associated with a higher prevalence of clinically relevant mental symptoms. All these variables could be useful when identifying vulnerable patients who require timely psychiatric care.