The risk of hospitalization for psychotic disorders following hospitalization for COVID-19: a French nationwide longitudinal study

COVID-19, like other infectious diseases, may be a risk factor for psychotic disorders. We aimed to compare the proportions of hospitalizations for psychotic disorders in the 12 months following discharge from hospital for either COVID-19 or for another reason in the adult general population in France during the first wave of the pandemic. We conducted a retrospective longitudinal nationwide study using the national French administrative healthcare database. Psychotic disorders were first studied as a whole, and then chronic and acute disorders separately. The role of several adjustment factors, including sociodemographics, a history of psychotic disorder, the duration of the initial hospitalization, and the level of care received during that hospitalization, were also analyzed. Between 1 January 2020 and 30 June 2020, a total of 14,622 patients were hospitalized for psychotic disorders in the 12 months following discharge from hospital for either COVID-19 or another reason. Initial hospitalization for COVID-19 (vs. another reason) was associated with a lower rate of subsequent hospitalization for psychotic disorders (0.31% vs. 0.51%, odds ratio (OR) = 0.60, 95% confidence interval (CI) [0.53-0.67]). This was true for both chronic and acute disorders, even after adjusting for the various study variables. Importantly, a history of psychotic disorder was a major determinant of hospitalization for psychotic disorders (adjusted OR = 126.56, 95% CI [121.85-131.46]). Our results suggest that, in comparison to individuals initially hospitalized for another reason, individuals initially hospitalized for COVID-19 present a lower risk of hospitalization for first episodes of psychotic symptoms/disorders or for psychotic relapse in the 12 months following discharge. This finding contradicts the hypothesis that there is a higher risk of psychotic disorders after a severe COVID-19.


INTRODUCTION
From the beginning of the COVID-19 pandemic and related restrictions, concerns have been raised about its impact on mental health in the general population, specifically concerning subjects without a history of psychiatric disorders-who may experience a first episode of psychiatric disorders-and also those with such a history, especially psychotic disorders [1,2].Severe infectious diseases, both bacterial and viral, were already known to be a risk factor for psychotic disorders, and several longitudinal studies have found consistent significant associations [3][4][5].Regarding these associations, many studies hypothesized that the development of psychotic disorders (or relapse in those with a history of psychotic disorders) is not related to the infectious diseases themselves but to abnormal immune response and neuroinflammation [6][7][8].It is possible that such phenomenon concerns SARS-CoV-2 infection, which could be a risk factor of psychotic disorders [9].Moreover, COVID-19-especially severe forms-represents a stressor and a negative life event, and thus could cause or trigger a psychotic disorder [10][11][12].
Subjects with a history of schizophrenia or other psychotic disorder have been shown to be at risk of developing COVID-19, especially severe forms [13][14][15].Severe infection is associated with anxiety symptoms, sleep disorders, and delusional ideations; it therefore could be a risk factor for psychotic relapse [16].
Moreover, isolation measures imposed on patients with COVID-19 may have increased the risk of psychotic relapse through reduced access to mental health care and/or interrupted medication, especially during lockdown periods [17][18][19].
Following these studies, we analyzed the whole French Metropolitan general population [33] and found a slightly-but significantly-higher rate of hospitalization for psychiatric disorders in the 12 months following hospitalization for COVID-19 in comparison to hospitalization for another reason (unadjusted odds ratio (OR) = 1.20, 95% confidence interval (95% CI) [1.18-1.23]).The rate was also higher when looking specifically at psychotic disorders (OR = 1.11, 95% CI [1.05-1.19]).Importantly, the significance of the associations varied according to the adjustment factors.In the fully adjusted model, (i.e., taking into account socio-demographic characteristics, history of psychiatric disorders, and the levels of care intensity), this association was no more significant.
In the present study, we focused on psychotic disorders.More specifically, we described the different kinds of psychotic disorders, and separately analyzed acute vs. chronic psychotic disorders.Indeed, showing different patterns of association for different psychotic disorders-especially acute psychotic disorders that could reflect FEP-would add significant knowledge concerning a putative role of COVID-19 as risk factor for psychotic disorders.Moreover, we could analyze if the psychotic disorder was the main reason of the hospitalization ("principal" diagnosis) or if it was considered a comorbid condition of the main reason of the hospitalization ("associated" diagnosis).In this last case, we could investigate how COVID-19 could increase global health vulnerability among subjects with psychotic disorders, and the different (sociodemographic and clinical) risk factors of hospitalizations for psychotic disorders (by analyzing different types of hospitalization, in psychiatric and non-psychiatric wards).
To act on these research possibilities, using the same dataset employed for our study on psychiatric disorders in the whole French Metropolitan general population [33], the present longitudinal study aimed to determine the risk of hospitalization for different non-affective psychotic disorders in the 12 months following discharge from hospital for COVID-19 during the first wave of pandemic, in comparison to discharge from hospital for another reason.Different types of psychotic disorders were described, as well as acute and chronic disorders.The role of adjustment variablesincluding the level of social deprivation, a history of psychotic disorders, and the characteristics of hospitalization for COVID-19 (duration, level of care received)-were also separately analyzed.

Data sources
The present study used data from France's national administrative healthcare database Système National des Données de Santé (SNDS), which covers almost the country's entire population of 67 million inhabitants (i.e., 66.3 million in 2020).It includes admission and discharge data for all public and private hospital stays.The different data sources which comprise the SNDS are described elsewhere [33].For all hospitalizations, medical diagnoses are coded according to the International Classification of Diseases 10 th edition (ICD-10), while the main medical and surgical procedures performed are coded according to France's Classification Commune des Actes Médicaux (CCAM, or Common Classification of Medical Procedures).Medical diagnoses are coded as principal (i.e., the main reason for hospitalization), or associated (i.e., a comorbidity that is not the main reason for hospitalization).
For the present study, we also used the Cartographie des Pathologies et des Dépenses (CPD, or Diseases and Expense Mapping) database in the SNDS to identify the presence of a history of psychotic disorder over the five years preceding the study period for each individual included.The CPD database's medical algorithms are based on SNDS data on hospitalizations, filled prescriptions, and free complete healthcare coverage for long-term diseases; the algorithms are publicly available in French [34,35].

Study design and participants
This retrospective longitudinal study aimed to compare the risk of hospitalization for a psychotic disorder during the 12 months following hospital discharge from a medical, surgical, or obstetrics ward in adult patients (i.e., 18 years or over) who had either been hospitalized for COVID-19 or for another reason between 1 January 2020 and 30 June 2020 (i.e., first wave of the pandemic) in Metropolitan France.Subjects brought to an emergency department (ED) visit but not subsequently admitted to hospital were not considered.For each individual, a reference hospital stay (i.e., either COVID-19 or another reason) was selected.For patients hospitalized more than once during the study period, if at least one of these was COVID-19 related (N = 85,514), then it was considered the reference hospital stay.For patients hospitalized more than once for COVID-19 (N = 7521), the hospitalization where the most intensive level of medical care was provided was considered the reference stay.This care intensity criterion was also adopted to define the reference stay for patients with more than one hospitalization for pathologies other than COVID-19 during the study period.More information concerning this data is available elsewhere [33].

Outcomes
The main outcome was hospitalization where a psychotic disorder of any type was diagnosed in the 12 months following discharge from the reference hospital stay (i.e., COVID-19 or another reason).We searched for inpatient admissions to medical, surgical, obstetrics, and psychiatric wards for the following non-affective psychotic disorders (ICD-10 codes): F20: Schizophrenia; F21: Schizotypal disorder; F22: Delusional disorder; F23: Acute and transient psychotic disorders F24: Shared psychotic disorder; F25: Schizoaffective disorders; F28: Other nonorganic psychotic disorders; F29: Unspecified nonorganic psychotic disorders.
More specifically, the two types of outcomes considered were: i) hospitalizations for a psychotic disorder (i.e., principal diagnosis only), and ii) hospitalizations with a diagnosis (i.e., principal or associated diagnosis) of a psychotic disorder.Psychotic disorders were studied first as a whole B. Pignon et al.

Adjustment variables
Sociodemographic characteristics.Age, sex, and region of residence (among the 13 regions in metropolitan France) were considered.Age, calculated from the individual's year of birth, was categorized into four age groups (18-39 years, 40-59 years, 60-74 years, 75+ years).Socioeconomic status of the city of residence was also considered.This was measured using the French Deprivation Index (Fdep), an ecological-level indicator based on median household income, the percentage of higher education graduates in the population over 15 years old, the percentage of manual workers in the labor force, and the unemployment rate of the individual's city of residence [36].
History of psychotic disorders.The CPD database (8th version) was used to assess whether a patient had a history of psychotic disorders in the five years preceding the study period (summarized using a dichotomous variable).An individual was considered to have had a psychotic history for a specific year if one of the following elements was found in the SNDS for that year: i) declaration by a healthcare professional that the patient had a psychotic disorder officially recognized as a long-term disease; (ii) hospitalization(s) with a diagnosed psychotic disorder in a psychiatric or non-psychiatric hospital ward or in a healthcare center during at least one of the previous two years; (iii) hospitalization(s) with a diagnosed psychotic disorder in a psychiatric ward or non-psychiatric health establishment during the previous 5 years (i.e., n to n-4) and receiving prescribed antipsychotics drugs on at least three different occasions during year n [35].
Characteristics of the reference hospital stay.Median duration (in days) and level of clinical care received were used to characterize the reference hospital stay.We defined three levels of care intensity for these stays (i.e., whether for COVID-19 or for another reason).These three levels were defined according to the care provided in general for different degrees of COVID-19 severity (for the complete list of procedures used to define the levels of clinical care provided in this context, see Decio et al. [33]).The first level corresponded to patients with the mildest level of respiratory difficulty and admitted to a general hospital ward; these patients required no or low-flow oxygen (up to 15 L/minute).The second level corresponded to patients admitted to an ICU, irrespective of the level (i.e., type and flowrate) of oxygen supply therapy, and patients who received high-flow nasal oxygen or non-invasive ventilation.The third and most intense level of care corresponded to patients who were hospitalized in an ICU and required at least invasive ventilatory support.

Statistical analysis
We used estimated and compared the risks of hospitalization according to the reference hospital stay type (i.e., COVID-19 vs. for another reason) where there was a diagnosis of i) a psychotic disorder (all kinds), (ii) an acute psychotic disorder, and (iii) a chronic psychotic disorder, first using chi-square test.For each of these three types of diagnosis, we first considered only principal diagnoses of a psychotic disorder (i.e., hospitalization for a psychotic disorder).We then considered hospitalization with (i.e., principal or associated diagnosis) a psychotic disorder.For these six outcomes, when the association was significant, four logistic regression nested models were successively performed as follows: Model 1 described the crude associations between the outcome and the reason for the reference hospital stay (i.e., COVID-19 vs. another reason); Model 2: model 1 plus socio-demographic variables; Model 3: model 2 plus the history of psychotic disorders variable; Model 4: model 3 plus the two characteristics of the reference hospital stay variables (i.e., median duration and level of clinical care).
In order to analyze whether the role of hospitalization for COVID-19 was different in patients with vs. without history of psychotic disorders, we searched for multiplicative interaction between history of psychotic disorder and the type of reference hospitalization, concerning all psychotic disorders as principal diagnosis.If the interaction was statistically significant, we subsequently conducted stratified analyses by psychotic history, adjusted for socio-demographic variables and the characteristics of the reference hospital stay (Model 5).

Study cohort and description of patients hospitalized for psychotic disorders
Between 1 January 2020 and 30 June 2020, there were a total of 96,313 hospitalizations (i.e., reference hospital stays) for COVID-19 and 2,979,775 hospitalizations for another reason in metropolitan France.The characteristics of the whole hospitalized population are available in supplementary materials (Supplementary Table 1).Among these patients, in the 12 months following discharge from their reference hospital stay, 14,622 were hospitalized for a psychotic disorder (principal diagnosis only, all kinds of psychotic disorders).The characteristics of these patients are presented in Table 1.Those initially hospitalized for COVID-19 were older than those initially hospitalized for another reason (mean age: 68.5 years for COVID-19 vs. 59.9 for another reason).They were also more likely to live in the least socially deprived areas (FDep: 30.00% vs. 17.71%), and to have a history of psychotic disorders (89.83% vs. 78.35%).Furthermore, concerning the reference hospital stay (i.e., COVID-19 or another reason), they received a higher level (i.e., more intense) of clinical care during their reference hospital stay (ICU: 12.88% vs. 6.19%,ICU with invasive procedure: 8.14% vs. 2.44%), and to have had a longer stay (median [interquartile interval] = 8 [3][4][5][6][7][8][9][10][11][12][13][14] vs. 1 [0-3] days).The characteristics of both principal and associated diagnoses of the psychotic disorders were similar (see Supplementary Table 2).Among patients hospitalized with acute psychotic disorders, those without a history of psychotic disorders (44.14% for principal diagnosis only and 54.73% for principal or associated diagnosis) were more frequent than among patients with chronic psychotic disorders (25.21% and 20.15%).
Hospitalization rates for psychotic disorders Principal diagnoses.Between 1 January 2020 and 30 June 2020, among patients hospitalized in metropolitan France in the 12 months following discharge from their reference hospital stay, 27,614 were hospitalized with a diagnosis of a psychotic disorder (i.e., principal or associated diagnosis).When only principal (i.e., hospitalized for a psychotic disorder) diagnoses of psychotic disorders were considered, the rate of hospitalization for a psychotic disorder (all kinds) after a COVID-19-related reference hospital stay (0.31%) was lower than the rate after a stay for another reason (0.51%, p value of chi-square test < 0.001).Schizophrenia rates were consistent with this overall pattern (0.17% vs. 0.25%, p value < 0.001), as were the rates of most of the different psychotic disorders, and for acute (0.02% vs. 0.08%, p value < 0.001) and chronic (0.27% vs. 0.41%, p value < 0.001) psychotic disorders.Details on hospitalization rates for the different psychotic disorders (principal diagnosis) are shown in Fig. 1 and in the supplementary materials (Supplementary Table 3).
Principal and associated diagnoses.When principal and associated (i.e., hospitalized with a psychotic disorder) diagnoses of psychotic disorders were considered, the rate of hospitalization for a psychotic disorder (all kinds) after a COVID-19-related reference hospital stay (1.06%) was higher than the rate after a stay for another reason (0.95%, p value < 0.001).This significant difference was found for all the different psychotic disorders studied, except schizophrenia (non-significant variation: 0.45% after initial COVID-19 hospitalization vs. 0.43% after hospitalization for another reason, p value = 0.215).Similarly, hospitalizations for subjects with a chronic psychotic disorder were not significantly associated (0.76% vs. 0.73%, p value = 0.345, subjects hospitalized with a chronic psychotic disorder were thus not studied in the multivariable models).Unlike chronic disorders, acute psychotic disorders were significantly more frequent after a COVID-19-related reference stay (0.05% vs. 0.11%, p value < 0.001).Details on hospitalization with a psychotic disorder can be found in the supplementary materials (Supplementary Table 3).
Importantly, these variations, as well as the associations, were consistently significant when considering only admissions to a hospital psychiatric department (see Supplementary Table 4).Interestingly, younger age group, male sex, and to a lesser extent the level of deprivation of the city of residence (negative association), were associated with a higher risk of hospitalization for psychotic disorder, irrespective of the initial hospital type.After adjustment for a history of psychotic disorders (Model 3), this negative association was stronger (aOR = 0.57), whereas the association with a history of psychotic disorders was strong (aOR = 126.56).Finally, in Model 4, which took into account the duration and the level of care of the initial hospital stay, the association remains significantly negative (aOR = 0.67).Moreover, the duration of the reference hospital stay (aOR = 0.96) and ICU care (aOR = 0.76) were negatively associated with the risk of hospitalization for a psychotic disorder.The details of the different models for principal diagnoses are available in Table 2.
With regard to acute psychotic disorders, the association with the main study outcome was stronger than for psychotic disorders (all kinds): OR = 0.26, 95% CI [0.16-0.40],p value < 0.001 (Model 1).This significant association was attenuated after the various adjustments: Model 2: aOR = 0.38, Model 3: aOR = 0.36, Model 4: aOR = 0.50 (all p values < 0.001).The details of the different models concerning hospitalization for an acute psychotic disorder are available in Table 3.
With regard to chronic psychotic disorders, the different associations were also consistent with psychotic disorders (all kinds) analyses (Model 1: OR = 0.63, Model 2: OR = 0.72, Model 3: OR = 0.87, Model 4: OR = 0,69, all p values < 0.001).The details of the different models concerning hospitalization for a chronic psychotic disorder are available in Table 4.
Finally, concerning all psychotic disorders as principal diagnosis, we found a significant multiplicative interaction between the reference hospital stay type and history of psychotic.After stratification by history of psychotic, analyses showed a significant negative association between initial hospitalization for COVID-19 and the risk of subsequent hospitalization for a psychotic disorder among patients without history of psychotic disorder (Model 4: aOR = 0.41, 95% CI [0.28-0.59]),whereas the association was not significant among patients with history of psychotic disorder (aOR = 0.94, 95% CI [0.82-1.06]).
Principal or associated diagnosis of a psychotic disorder.Model 1 shows that, after COVID-19 hospitalization, in comparison to hospitalization for another reason, the risk of hospitalization with a diagnosis of a psychotic disorder (all kinds, principal or associated diagnosis) was higher (OR = 1.11, 95% CI [1.05-1.19],p value < 0.001).Model 2, which was adjusted for socio-demographic characteristics, was consistent with Model 1 (aOR = 1.16).Younger age groups, male sex, and to a lesser extent the level of deprivation of the city of residence (negative association), were associated with a higher risk of hospitalization with a diagnosis of psychotic disorder, irrespective of the initial hospital type.After adjustment for a history of psychotic disorders (Model 3), the association between the risk of hospitalization with a diagnosis of psychotic disorder and the initial reference hospitalization (i.e., COVID-19 vs. another reason hospitalization) was no longer significant (p value = 0.058), whereas the association with a history of psychotic disorders was strong (aOR = 116.28).Finally, in Model 4, the association between hospitalization for COVID-19 and the risk of subsequent hospitalization with a psychotic disorder was reversed: aOR = 0.90, 95% CI [0.84-0.97](p value = 0.006).Moreover, the duration of the reference hospital stay (positive association: aOR = 1.01),ICU care (negative association: aOR = 0.89) and care in an ICU with an invasive procedure (positive association: aOR = 1.09), were all associated with the risk of hospitalization with a diagnosis of psychotic disorder.The details of the different models for hospitalization with a diagnosis (i.e., principal or associated) of a psychotic disorder (all kinds) are available in the supplementary materials (Supplementary Table 5).
Unlike what was found for psychotic disorders in general (i.e., all kinds), Model 1 revealed a negative association with hospitalization with an acute psychotic disorder (aOR = 0.49, 95% CI [0.38-0.66],p value < 0.001).The three other models were consistent with Model 1.The details of all the different models concerning hospitalization with an acute psychotic disorder are available in the supplementary materials (Supplementary Table 6).

DISCUSSION
In this nationwide French study, the 12-month risk of hospitalization for a psychotic disorder (i.e., principal diagnosis only) after discharge from hospital for COVID-19 during the first wave of the pandemic was lower than the risk after discharge after hospitalization for another reason.This was consistent for acute and chronic psychotic disorders.The stratified analyses revealed that this risk was lower in subjects without history of psychotic disorder.

Methods considerations
Two points are important to highlight to explain these results.First, we need to make an important distinction between principal and associated diagnoses: the former represents the reason for hospitalization (i.e., patient is hospitalized for a psychotic disorder), while the latter represents a comorbidity at the time of hospitalization (i.e., patient is hospitalized with a psychotic disorder).Accordingly, associated diagnoses may reflect to a large extent patients who are not hospitalized in a psychiatric ward (i.e., hospitalization with a diagnosis of psychotic disorders).In the present study, we were interested in the risk of FEP or psychotic relapse following a hospitalization for COVID-19.This is why the main finding of this study is the lower rate of hospitalization for a psychotic disorder among patients previously hospitalized for COVID-19 than in patients previously hospitalized for another reason.Second, we cannot conclude that hospitalization for COVID-19 is a protective factor against hospitalization for psychotic disorders, given that the control group (i.e., hospitalized for another reason) was a particularly vulnerable sample and not the general population.Indeed, the 12-month rate of hospitalization for and with psychotic disorders in those previously hospitalized for a reason other than COVID-19 was high (0.95% considering principal or associated diagnosis and 0.51% considering principal diagnosis only).In comparison, these rates were of the same order of magnitude as the prevalence of psychotic disorders in the general population (0.46%) according to a French urban-area based study [38]), and the majority of those with psychotic disorders in that study was not hospitalized.

Etiological considerations
The rate of hospitalization for acute psychotic disorders can be considered a proxy for the rate of incidence of psychotic disorders in the population of patients hospitalized given the majority of subjects with FEP are hospitalized in France [39].This consideration is approximate, as 44.86% of patients with acute disorders had a history of psychotic disorders (whereas this rate was 74.39% in patients with chronic disorders).According to this hypothesis, the rates of FEP were also high (COVID-19 patients: 0.02% considering hospitalization for acute psychotic disorders, 0.05% considering hospitalization with acute psychotic disorders; non-COVID-19 patients: 0.08% and 0.11%).The only recent incidence study of psychotic disorder in France took place in two areas, one urban and one rural, and found lower rates (0.036% per year and 0.017%, respectively) [39].This confirms that hospitalization, irrespective of the reason (i.e., COVID-19 or another reason), is a risk factor for FEP, but not specifically COVID-19 hospitalization.This contrasts with what was initially feared about the role of this diseases as a risk factor for psychotic disorder [3][4][5].
In an Australian study in early psychosis intervention services, O'Donoghue et al. found a higher rate of hospitalization of FEP during the pandemic than the pre-pandemic period (this difference was not significant: incidence rate ratio = 1.14, 95% CI [0.92-1.42])[40].Longitudinal studies with longer follow-up durations, especially for children with a history of COVID-19, are necessary to investigate the role of COVID-19 as a putative risk factor for FEP.Indeed, the associated risk may increase over time, as is the case for other infections.Interestingly, in the study by Taquet et al., among the 14 neuro-psychiatric outcomes studied, psychotic disorders were the only psychiatric disorders whose risk was still higher after 2 years of follow-up [31].In contrast, in an English cohort study of more than 8 million subjects focusing on the incident psychotic disorders after initial hospitalization for COVID-19 or for severe acute respiratory infection hospitalizations, Clift et al. found non-significant differences between rates for the two types of hospitalization.[32].

Risk factors of hospitalization for psychotic disorders
This study confirmed that hospitalization for non-psychiatric reasons (i.e., for COVID-19 or another reason) was a major risk factor for psychotic relapse for patients with a history of psychotic disorders, as hospitalization represents a major stressful event [41].Moreover, during hospitalization in non-psychiatric wards, these patients' psychotropic (including antipsychotics) treatments are often stopped or modified because of their side effects [42][43][44]; this too constitutes a major risk factor for psychotic relapse [45].
Crude hospitalization rates in non-psychiatric wards were also very high in our study (hospitalization rate for or with a diagnosis of a psychotic disorder: 0.95%, including 0.49% for hospitalization in psychiatric wards and 0.46% in non-psychiatric wards).This confirms that psychotic disorders are major risk factors (and therefore very comorbid) for non-psychiatric diseases (with high associated mortality rates) [46][47][48][49].These high rates of hospitalization must be interpreted in the context of the beginning of the COVID-19 pandemic when the use of hospitalization for non-COVID-19 pathologies was greatly reduced in France-both in psychiatric and non-psychiatric wards -because of bed availability issues [50][51][52][53][54].
In terms of adjustment variables, as expected, having a history of psychotic disorders was a major determinant of the risk of hospitalization for and with a psychotic disorder.Moreover, male sex was associated with a higher risk of subsequent hospitalization for psychotic disorders, which is consistent with the fact that males present more severe forms of psychotic disorders than females [55].Likewise, the young age was associated with a higher risk of hospitalization for and with psychotic disorders, which also reflects the literature [56,57].

Comparisons with previous findings
A comparison of our results with the literature is difficult, as studies on the role of severe or hospitalized COVID-19 are very rare.Taquet et al. studied larger and international samples in several studies [29][30][31]; one of these can be compared to our present work [29].Contrary to our results, the risk of psychotic disorders post-COVID-19 (with or without hospitalization) in that study was higher in comparison to influenza (HR = 2.03, 95% CI [1.78-2.31]),and other respiratory infections (HR = 1.66, 95% CI [1.53-1.81]).Interestingly, the authors compared COVID-19 with and without hospitalization, and found an even greater risk for patients requiring hospitalization (HR = 2.22, 95% CI [1.92-2.57]).This finding further increases the discrepancy with the present study.Moreover, their analyses of ICU care were consistent with this last association (HR = 1.48, 95% CI [1.14-1.92]).However, once again, their choice of specific health control events (i.e., other respiratory infections), and their sample-which was very different (over 81 million patients in different countries, primarily in the USA) from the present study-make the comparison difficult.Elsewhere, Clift et al. found less inconsistent results [32].In that study, which examined psychotic relapse in a register database of more than 8 million subjects, the rates of relapse were not different between persons previously hospitalized for COVID-19 and those previously hospitalized for a severe acute respiratory infection, although they were both higher than in the general population (HR = 3.05 95% CI [1.58-5.90]for COVID-19; 3.63 95% CI [1.88-7.00]for severe acute respiratory infection).

Explanations of the main findings
There are several putative explanations for the lower risk of hospitalization for psychotic disorders after COVID-19 hospitalization in comparison to hospitalization for another reason.The first regards the control population.COVID-19 may have represented a less severe condition than the other reasons requiring hospitalizations.This hypothesis is particularly likely in the context of the first wave of the pandemic, during which reduced access to care (e.g., lower rate of hospitalization for all the other diseases during this period [52,53]) could have led to a selection of more severe cases leading to hospitalizations for other reasons.However, contrary to this hypothesis, subjects hospitalized for COVID-19 had higher levels of care and longer hospital stays.
A second explanation is that hospitalization for a reason other than COVID-19 may have concerned patients with diseases that were more chronic than COVID-19, and who therefore had a higher and/or longer lasting vulnerability to relapse.Another explanation is that vulnerability to psychotic disorders may have been associated with reduced access to hospitalization for COVID-19-especially admission to an ICU-during the first wave of the pandemic [13].This hypothesis is consistent with the negative association that we found between hospitalization for psychotic disorders and previous admission to an ICU ward compared to non-ICU wards.Finally, patient mortality may have been higher in subjects hospitalized with COVID-19 than in those hospitalized for another reason; this could partially explain their lower rate of subsequent hospitalization for or with a psychotic disorder.Nevertheless, contrary to-but without excluding-this hypothesis, a recent French nationwide study highlighted that non-COVID-19 mortality increased more among patients with psychotic disorders than among controls during 2020 in comparison to 2019 (aOR = 1.18, 95% CI [1.11-1.25])[58].

Limitations and strengths
Several study limitations must be underlined.The control population (discussed in detail elsewhere [33]) was not specific (different ward types, a wide variety of indications).Second, contrary to Taquet et al. we were not able to study different durations after the first hospitalization [31].Third, in terms of incidence of psychotic disorders (especially FEP), given the etiology of psychotic disorders-for which risk factors are involved years (or even decades) before the onset of the disease-our follow-up duration (12 months) was relatively short and, as stated before, the hospitalization for acute psychotic disorders was a questionable proxy.Moreover, the present study considered only hospitalization for COVID-19, which could be considered as a proxy of severe COVID-19, and the results could be explained by unmeasured variables (e.g., tobacco use, treatment, etc.).Any causal conclusions must therefore be interpreted with caution.Fourth, we had no information on patients who left the study (i.e., patients who died or moved outside of metropolitan France).Finally, as it is based on health data register database, this study and its conclusions concerns only treated psychotic disorders.The treatment gap phenomenon (i.e., the gap between experiencing a disease and using treatment for this disorder) is well-known concerning psychotic disorders, and concerns both psychiatric and non-psychiatric disorders (thus, potentially COVID-19) [59].Thus, our results are only generalizable to a population of people with disorders requiring hospital care (for both COVID-19 and psychotic disorders), and having access to such care.
Despite these limitations, the study has several strengths.The first is the exhaustiveness of the data which covered all hospitalizations in metropolitan France, thereby limiting selection and memory biases, and ensuring analyses with good statistical power.Second, the study of both principal and associated diagnoses of psychotic disorders enabled us to distinguish the reasons for hospitalization.Moreover, by studying acute vs. chronic psychotic disorders separately, we were able to study to what degree severe COVID-19 is a risk factor for FEP.

Conclusion
Overall, during the first wave of the COVID-19 pandemic in France, the risk of hospitalization for psychotic disorders in the 12 months following discharge from hospital for COVID-19 was lower than the risk following discharge after hospitalization for another reason.This was true for psychotic disorders as a whole, for acute and chronic psychotic disorders taken separately, and for schizophrenia.These findings contradict the hypothesis that the risk of psychotic disorders after hospitalization for COVID-19 is higher than that after hospitalization for another reason.Other studies on this topic with longer follow-up duration, and which integrate COVID-19 vaccination data are necessary.Indeed, vaccination may modify the consequences of COVID-19, including the rate of hospitalization or the risk of further psychotic disorders.
Finally, this study confirms that hospitalization for nonpsychiatric reasons (i.e., COVID-19 or another reason) is a major risk factor for psychotic relapse; this finding should encourage clinicians to provide greater care to patients with psychiatric disorders during hospital stays and after discharge.

DATA AVAILABILITY
Risk of hospitalization for a psychotic disorder according to the type of reference hospital stay Principal diagnosis of psychotic disorders.The unadjusted model (Model 1) shows that, after COVID-19 hospitalization, in comparison to hospitalization for another reason, the risk of hospitalization for a psychotic disorder (all kinds, but principal diagnosis only) was lower (OR = 0.60, 95% CI [0.53-0.67],p value < 0.001).Model 2, which was adjusted for socio-demographic characteristics, was consistent with Model 1 (adjusted OR (aOR) = 0.69).

Fig. 1
Fig. 1 Rate of hospitalization for psychotic disorders after hospitalization during the first wave of the COVID-19 pandemic.Rate of hospitalization for psychotic disorders (principal diagnoses) after hospitalization during the first wave of the COVID-19 pandemic in metropolitan France (%, denominator: patients hospitalized during the 6 first months of 2020).

Table 1 .
Comparison of patients previously hospitalized for COVID-19 and those previously hospitalized for another reason among patients hospitalized for a psychotic disorder (principal diagnoses only) in the 12 months following discharge from the previous hospitalization.
a chi-square tests or Student tests.B. Pignon et al.

Table 4 .
Association of COVID-19 hospitalization with further hospitalization for a chronic psychotic disorder (i.e., principal diagnosis only) among patients hospitalized between 1 January 2020 and B.Pignon et al.