Design and study population

This is a retrospective cohort study that included virtually all individuals with laboratory-confirmed SARS-CoV-2 infection in the region of Aragon, Spain, with a reference population of 1.3 million inhabitants. Approximately 95% of the population are provided by the Spanish health system free-of-charge healthcare with universal coverage [22]. The data used in this study was obtained through the PRECOVID (Prediction in COVID–19) Study Cohort. That cohort was created through the linkage of real-world data from clinical-administrative databases and patients’ electronic health records (EHRs). Its objective was to serve as the basis for observational studies on the epidemiology and trajectory of the COVID–19 disease [4]. Users of the public health system of all ages with a laboratory-confirmed infection by SARS-CoV-2 were consecutively included in this cohort along the pandemic, and their socio-demographic and clinical information linked at the patient level and then pseudonymized. Patients did not have the option of opting out of providing their data for use in this study. The authors who accessed the pseudonymized data belong to the EpiChron Research Group of Aragon Health Sciences Institute (IACS), and received permission from IACS to utilize the data for this specific study, thus implying its exclusive use by the researchers appearing in the project protocol approved by the Clinical Research Ethics Committee of Aragon (CEICA).

For this study, we accessed data from the PRECOVID Study Cohort on March 24, 2022 and we included all the individuals from the PRECOVID Study Cohort with confirmed SARS-CoV-2 infection between 4 March 2020 and 22 July 2021 (enrollment period). We followed patients for a minimum of 30 days from the date of inclusion in the cohort to study whether death occurred within 30 days of SARS-CoV-2 infection confirmation. We also assessed hospital admissions during the 15 days before and after infection. Patients were followed-up at most until 22 August 2021 (date of last data available in the cohort). We have analyzed hospitalization and mortality risk in patients with mental health disorders and COVID-19, because these two outcomes define infection severity following the definition of the COVID-19 Treatment Guidelines provided by National Institutes of Health [23].

In Aragon, Spain, different measures of geographic confinement and social restrictions were taken throughout this study period depending on the incidence of cases. In the first three months of the pandemic (March—May 2020) testing capacity in our region was limited at the time and the population was in lockdown. One this period finished, diagnostic tests were performed on all patients who were admitted to hospital and on symptomatic patients, and isolation of infected patients was carried out for 14 days. Contacts of positive patients were also actively studied. The use of masks was mandatory in the whole studied period.

The ethics committee endorsed the PRECOVID Study research protocol (PI20/226), and due to the use of anonymized data and the epidemiological nature of the project, it was not necessary to obtain the informed consent of the patients.

Variables of study and data sources

For each subject, sociodemographic variables were analysed: age, sex, ethnicity (native vs. migrant), nationality, residence area (urban—those that concentrate in one of its municipalities at least 80% of the population of the area, and rural—the rest), and deprivation index of the area. This index was specifically developed for Aragon in a previous work [24], calculated at an aggregated level by basic healthcare area according to 26 socio-economic indicators including information on education, housing and neighbourhood conditions, unemployment rates, types of employment, ageing of the population, and immigration, and divided into four quartiles from least (Q1) to most (Q4) deprived. As clinical variables, we analysed all chronic disease diagnoses present at the date of infection and registered in primary care EHRs, the presence of multimorbidity (defined as the coexistence of two or more chronic conditions [15]), all drugs dispensed in community and hospital pharmacies during the month prior to infection, and the presence of polypharmacy defined as the simultaneous prescription of five or more drugs.

The registered diagnoses of chronic diseases were coded using the International Classification of Primary Care, First Edition (ICPC–1). A mapping system developed to codify temporary disability was used to mapped the codes to the International Classification of Diseases, Ninth Edition, Clinical Modification (ICD–9–CM) [25]. Then, theses diagnoses were subsequently sorted into 226 clinically relevant categories thought the utilization of the Clinical Classifications Software (CCS) for ICD–9–CM [26]. In some cases, diagnoses were renamed to facilitate their clinical interpretation (e.g., the term “mood disorders” was renamed as “depression and mood disorders”). The different diseases were classified as chronic or not chronic using the Chronic Condition Indicator software tool [27]. 153 conditions were categorized as chronic if they were coded during the last 12 months at least, and meeting one or both of the following criteria: (i)requires of ongoing interventions using medical products, services, and special equipment; (ii) entails limitations on self-care, social interactions, and independent living. All drugs were analysed at the third level (i.e., pharmacological subgroup) of their Anatomical–Therapeutic–Chemical (ATC) classification system code [28].

During the follow-up period, we analysed as outcome variables all-cause mortality within the first 30 days after SARS-CoV-2 diagnosis and the need for hospitalization (including admission to the Intensive Care Unit) during the 15 days before and after infection confirmation.

In our study, mental health disease was defined as the presence of at least one of the following conditions: schizophrenia and other psychotic disorders, anxiety, cognitive disorders, depression and mood disorders, substance abuse, and personality and eating disorders. Each of these conditions corresponded to a CCS category that was created based on the ICD–9–CM codes registered in patients’ EHRs.

Statistical analysis

We conducted a descriptive analysis of the population with COVID–19 infection according to the presence or not of mental illness; differences were assessed using t-tests and chi-square tests. Multivariable logistic regression models were used to examine the likelihood of 30-day all-cause mortality and COVID-19-related hospitalization (dependent variables) according to baseline demographic and clinical variables, including the diagnosis of specific mental conditions. A total of four models were performed, one for each outcome and gender combination. Results were presented as odds ratios (ORs) accompanied by their 95% confidence intervals (CI) and adjusted by all the variables included in the model. We used logistic regression instead of Cox regression considering we had the date of inclusion in the cohort (date of the first confirmatory test result), but not the exact date of infection; moreover, follow-up was truncated at 30 days after the index date for all individuals.

Stata software (Version 12.0, StataCorp LLC, College Station, TX, USA) was used to conduct the analyses, and we considered statistical significance at p<0.05.

Principals results and comparison with other studies

This population-based study in a region of Spain described the main socio-demographic and clinical characteristics of individuals with SARS-CoV-2 infection and the most prevalent mental health conditions, observing differences in the risk of death and hospitalization in men and women.

In our cohort of confirmed COVID–19 cases, mental illness was more frequent in women than in men, as it has been shown in previous similar studies [20, 29]. Anxiety was the most frequent mental condition, followed by depression and mood disorders, and cognitive disorders, although some gender-differences in the distribution of mental illnesses were shown. Wang et al. observed that patients with a recent diagnosis of a mental health condition had significantly higher odds of COVID–19 infection than those free of mental illness, with depression and schizophrenia as the strongest predictors after adjusting for age, gender and ethnicity [20]. Another study showed, however, that patients with a previous diagnosis of a mental health condition had the same risk for testing positive for SARS-CoV-2 infection than people without mental illness [8].

Most of the patients in our cohort were natives; however, the higher proportion of migrants observed in the population without mental illness may be due to the “healthy migrant effect”, which suggests that migrants are initially healthier than natives, and that their health worsens as the duration of their stay increases [30–34]. In addition, the lower diagnosis of mental health disorders among the migrant population is also related to immigrants´ health disadvantage explained by their experience of discrimination, access barriers to care (language culture, legal, etc.) and directly to the inherent stress of being an immigrant [30].

We found no association between deprivation index and residence area and COVID–19 infection. In contrast, in a previous study conducted in Aragon, COVID–19 incidence was slightly higher in patients belonging to the more deprived areas [29], which was associated with health areas’ inequalities due to precarious conditions involving overcrowded accommodation and limited access to outdoor spaces, with more risk of exposure to the virus and different susceptibility to infection [29, 35].

In general, patients with mental illness showed a higher number of chronic diseases and multimorbidity and polypharmacy rates, which has already been related to lower incomes and educational levels [12], and a higher burden of cardiometabolic and respiratory comorbidities such as diabetes and chronic obstructive pulmonary disease (COPD) that reduce life expectancy [16].

One of the major findings of this cohort study was that mental health conditions were associated to an increase in the risk of mortality and hospitalization in COVID–19 patients, and that their effect depended on gender. Gender is one of the determinants of health that produces considerable consequences for health outcomes [20]. Mental health differences between gender have been attributed to sex and gender differences: genetic predisposition; different sex hormones; dysregulations in the hypothalamic-pituitary-adrenal axis, especially for mood disorders; low self-esteem; gender-based violence; belonging to a gender minority; and other factors like education, employment, housekeeping and socioeconomic status [36]. These differences in the prevalence of mental diseases depending on gender has been related to gender disparities in COVID‐19 infection rates [20]. Through it is not an objective of our work, other studies have determined that hospitalizations and deaths are more frequent in men [20].

Although, recent studies confirmed that pre-existing mental health conditions were associated with COVID–19 mortality and an increased risk of hospitalisation [7, 14], this is one of the few studies that analyse the risk of mortality and hospitalizations considering the most frequent patterns of mental illness in the population at the same time, and differentiating by sex. The risk of COVID-19 mortality and hospitalisation has been related to patients´ high morbidity burden [20], and that it was also more frequent in patients with psychotic disorders like schizophrenia and schizotypal disorders, compared with patients with mood disorders [13, 14, 37]. The fact that cognitive disorders were the mental illness with the highest prevalence of mortality and hospitalization has been observed in other studies [2, 38]. It could be explained by the fact that patients with dementia may have less capacity to communicate their medical concerns [38], the atypical symptoms that may impede the early recognition of the disease, and their association with age and comorbid conditions [2]. Schizophrenia spectrum disorders have been associated with an increased risk of death [3, 6, 9, 10, 12, 21] and hospitalization, and this risk was not only limited to psychiatric diagnoses in hospitals [12, 16]. In another study, schizophrenia also increased mortality; however, it did not associate an increase in admissions to the Intensive Care Unit compared with non- schizophrenia patients [10, 21]. Another one showed that schizophrenia spectrum disorders were related to mortality, while mood and anxiety disorders were not [3].

In addition, the fact that mental illness increased the risk of negative outcomes of COVID–19 infection has also been linked to the relationship between severe mental illness (SMI), unhealthy lifestyles (e.g., physical inactivity, poor diets, sleep disturbances [12, 37], reduction in self-care, social isolation, lack of a caregiver or family support [8] and high alcohol and tobacco use) and a higher burden of somatic disease [12, 37]. Furthermore, SMI has been associated with dysregulation of biological processes such as immune-inflammatory alterations, which may predispose these patients to more severe infections or to secondary bacterial infections [12, 37]. Even before COVID–19, the incidence of pneumonia was higher in SMI, and was associated with antipsychotic medications, tobacco use, and other factors. Furthermore, clozapine, which is the antipsychotic reserved for treatment-resistant schizophrenia patients, can suppress immune functions, and increase susceptibility to infections like pneumonia [38]. Toubasi and Lee et al. observed that patients who were diagnosed with mental health conditions needed more mechanical ventilation, were more frequently admitted to the Intensive Care Unit, and had higher mortality [8, 14]. Fond et al. described an increased risk of mortality and poor COVID–19 outcomes in patients with mental health conditions irrespective of the main clinical risk factors for severe COVID–19 [7]. This suggests that other factors can lead to these health inequities, like a lack of communication skills, cognitive impairment [8, 21], limited comprehension of medical advice, poor self-awareness [21], lack of caregivers or family support [8] and the stigmatization of mental health conditions, which might aggravate social isolation of patients with mental health conditions and complicate the diagnosis of COVID–19 [12]. Thus, it is possible that patients with severe SMI, may have a harder time complying with protective hygiene measures, stay-at-home advice, and other health guidance during this pandemic [10, 39], or they might reside in congregate facilities, such as psychiatric inpatient units, homeless shelters, community housings, and prisons, where risk of COVID–19 transmission is increased because of the inability to effectively socially distance and/or quarantine [13]. Long-term use of alcohol, tobacco and other drugs is associated with pulmonary (COPD, pulmonary hypertension), cardiovascular (myocardial infarction, arrhythmias and cardiac insufficiency), and metabolic (hypertension and diabetes) diseases, all of which are risk factors for COVID-19 infection and its negative outcomes [7]. In addition, tobacco, whose use is much more common in SMI [10, 20, 39], is associated with a worsening of the prognosis in COVID-19, because of its effect on the respiratory system and immune responsiveness [10, 13], and the higher angiotensin converting enzyme 2 (ACE–2) levels in the airways [39]. All these factors could result in a delay in the access to hospital care and a worsening of the respiratory condition, which are important indicators of serious illness [21].

The influence of migration as a whole variable in the risk of mortality and hospitalization was not significant, but we observed some differences analysing the different nationalities. In the case of mortality, no significant differences were observed in either gender, but in hospitalization, Latin America men had 52% more risk than Spanish, and all women from the foreign nationalities, with the exception of EU and North America, had more risk of hospitalization than Spanish women. This increase in risk of hospitalization in the migrant population has been an unexpected outcome. In general, other studies have shown that this population suffers fewer hospitalizations than the native population [40, 41]. This finding would merit specific studies aimed at finding a more detailed explanation.

About the influence of deprivation index in the risk of mortality and hospitalization, we observed differences in the models analysed. The risk of hospitalization was higher in patients living in less deprived areas. This could be explained because migrants may have more difficulties to access the health system, they do not receive adequate care in the initial stages and are treated in more serious situations requiring hospitalization [30]. On the contrary, in this study the risk of mortality decreased in most deprived areas in men. This result must be interpreted with caution since deprivation index is an ecological index calculated in an aggregated form data, so it would be necessary to examine more deeply the influence of this variable in mortality risk in specifics studies.

As we saw, in some cases, multimorbidity [9, 12, 37] and having an additional chronic disease were associated with COVID–19 complications, especially in diabetes, obesity, cardiovascular disease, COPD, immunodeficiency, cancer and hypertension [39, 42, 43], and these diseases are more common in patients with SMI [10]. It is believed that certain comorbidities increase the inflammatory response, a biological factor common to severe COVID–19 pathophysiology (cytokine storm), and the chronic low-grade inflammation caused by mental illnesses. This response could induce the development of acute respiratory distress syndrome and death in patients with COVID–19 [6, 9, 12, 20]. The reasons why underlying medical conditions cause more severe COVID–19 cases are not yet fully understood, but ACE–2, the receptor to which SARS-CoV-2 binds to cause infection, are highly expressed in the heart and lungs. Whatever the mechanism, the high rate of smoking and comorbid medical conditions in SMI, in combination with the medications routinely used to treat SMI, may create the perfect storm for COVID–19 complications [39]. In addition, polypharmacy has been found to be associated with a higher risk of developing severe COVID–19 [6]. Other studies have reported that exposure to anxiolytic and antipsychotic drug treatments was associated with higher risk of pneumonia [13] and severe COVID–19 outcomes [6, 37]. Although the cause is unclear [6], it is thought that antipsychotics precipitate cardiovascular and thromboembolic risk, interfering with an adequate immune response, and might cause pharmacokinetic and pharmacodynamic interactions with drugs used to treat COVID–19 [37].

Although further research is required to determine the underlying mechanisms of these associations, our findings highlight the need for targeted approaches to manage and prevent COVID–19 in the at-risk groups that were identified in this study [37], as well as the need for the development of new clinical decision-making guidelines, including improvements in follow-up and targeted interventions [3]. Other recommendations are to maintain active surveillance, prompt and timely treatments when needed [8, 12] and the prioritization of these risk groups during vaccination campaigns [12, 14]. Hence, an important in the prevention of unfavourable COVID–19 outcomes in these patients would be an optimal management of their pulmonary, metabolic and cardiovascular diseases, and improve their treatment adherence [12].

Strengths and limitations

One of the main strengths of this study is the fact that we analysed virtually all the individuals of the reference population with a confirmed infection by COVID–19, and we exhaustively analysed all the chronic diagnoses registered by physicians in their EHRs. Many of the published studies focus on the study of the risk of hospitalizations or mortality, or include one or two mental illnesses. Few studies study all factors at the same time, considering the multimorbidity and sociodemographic variables. In addition, it is interesting to analysed this population-based data in a region of Spain to add comparability with the situation generated by this pandemic in other regions of the world. Furthermore, a free software with an open algorithm was used to manage chronic conditions in the analyses, facilitating reproducibility and comparability of results. One of the main limitations lies on the observational, cross-sectional nature of the study, which prevents the establishment of causal relationships between the explanatory variables and infection severity; nonetheless, our results point to certain factors that should be further investigated to confirm their influence. Another limitation of our study was related to the lack of relevant variables such as tobacco use, body mass index, reduction in self-care, isolation from society, difficulty of communication, and social determinants (e.g lack of a caregiver or family support) that could influence the severity of COVID–19 and that were not available in our cohort. Given that smoking, alcohol use and being overweight are especially common among patients with psychiatric disorders, we cannot rule out the possibility that our results could be influenced by some of these confounding factors. Moreover, we collected data on the use of oral psychotropic drugs but the information on intravenously administered psychotropic drugs was not recorded. Finally, the COVID–19 pandemic occurred in a series of waves that, as other studies have pointed out, has led to differences in infection diagnosis due to the availability of diagnostic tests and the updates in diagnostic protocols, resulting in a potential misclassification of cases.