Psychiatric and neuropsychiatric sequelae of COVID-19 – A systematic review

It has become evident that coronavirus disease 2019 (COVID-19) has a multi-organ pathology that includes the brain and nervous system. Several studies have also reported acute psychiatric symptoms in COVID-19 patients. An increasing number of studies are suggesting that psychiatric deficits may persist after recovery from the primary infection. In the current systematic review, we provide an overview of the available evidence and supply information on potential risk factors and underlying biological mechanisms behind such psychiatric sequelae. We performed a systematic search for psychiatric sequelae in COVID-19 patients using the databases PubMed and Embase. Included primary studies all contained information on the follow-up period and provided quantitative measures of mental health. The search was performed on June 4th 2021. 1725 unique studies were identified. Of these, 66 met the inclusion criteria and were included. Time to follow-up ranged from immediately after hospital discharge up to 7 months after discharge, and the number of participants spanned 3 to 266,586 participants. Forty studies reported anxiety and/or depression, 20 studies reported symptoms- or diagnoses of post-traumatic stress disorder (PTSD), 27 studies reported cognitive deficits, 32 articles found fatigue at follow-up, and sleep disturbances were found in 23 studies. Highlighted risk factors were disease severity, duration of symptoms, and female sex. One study showed brain abnormalities correlating with cognitive deficits, and several studies reported inflammatory markers to correlate with symptoms. Overall, the results from this review suggest that survivors of COVID-19 are at risk of psychiatric sequelae but that symptoms generally improve over time.


Introduction
The typical presentation of coronavirus disease 2019  caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) includes fever and respiratory difficulties. However, studies have shown that COVID-19 has a multi-organ pathology (Gavriatopoulou et al., 2020). Recent studies have reported that more than one-third of the infected patients develop neurological symptoms in the acute phase of the disease, and that 34% show brain abnormalities such as white matter hyperintensities and hypodensities as well as microhemorrhages, hemorrhages, and infarcts (Egbert et al., 2020;Helms et al., 2020a).
Intriguingly, several studies have reported a high incidence of acute psychiatric symptoms in COVID-19 patients. It has been suggested that at least 35% of the patients display symptoms of anxiety and depression Kong et al., 2020). Studies on psychiatric sequelae have been performed for the related corona-viruses SARS-CoV and Middle East respiratory syndrome coronavirus (MERS-CoV). A systematic review and meta-analysis on SARS-CoV and MERS-CoV patients concluded that both during the illness and post-illness, an increased incidence of cognitive disabilities, as well as depressed mood and anxiety, was found (Rogers et al., 2020). Data on long term effects of SARS-CoV-2 infection is currently being published at a high rate and include a multitude of symptoms including breathlessness, cough, muscle pains, chest pains, headache, and fatigue. Further, more and more studies demonstrate that psychiatric symptoms might also persist after recovery from the initial infection. These long-term symptoms are collectively referred to as 'long COVID' (Long COVID, 2020;Yelin et al., 2020) and it is evident that such sequelae will have severe personal and socioeconomic consequences. It is therefore of utmost importance to characterize and disseminate evidence of long-COVID.
One possible way SARS-CoV-2 could affect the brain is indirect, through the host's immune response to the infection. It has been suggested that COVID-19 patients experience a cytokine storm syndrome and that this is one of the main factors in the pathogenesis of the disease (Kempuraj et al., 2020;Mehta et al., 2020). Furthermore, evidence is beginning to emerge that neuroinflammation is present in some COVID-19 sufferers (Divani et al., 2020;Muccioli et al., 2020;Pilotto et al., 2020;Tang et al., 2021). Increased levels of cytokines, peripherally as well as centrally, can lead not only to lung inflammation and dysfunction (Kempuraj et al., 2020;Wu and Yang, 2020) but also to the development of psychiatric disease (see Dantzer et al. (2008); Drzyzga et al. (2006); Young et al. (2014) for reviews). This link has already been suggested in patients with acute COVID-19, where a tendency for higher levels of the cytokine interleukin (IL)-1β was found in subjects with depressive and/or anxiety symptoms compared to COVID-19 patients who did not display such symptoms Kong et al., 2020).
As mentioned above, a systematic review with meta-analysis examining psychiatric sequelae after SARS-CoV and MERS-CoV has already been published, providing valuable information on these related coronaviruses (Rogers et al., 2020). Also, a systematic review focussing primarily on the indirect effects of the COVID-19 pandemic on mental health has been performed (Vindegaard and Benros, 2020). In the current systematic review, the aim was to provide an overview of the current evidence of psychiatric complications in long-COVID after primary symptoms of acute COVID-19 have ceased. Furthermore, we aimed to identify risk factors and molecular mechanisms which could give rise to psychiatric symptoms.

Search strategy and selection criteria
We searched the databases PubMed and Embase for studies on psychiatric sequelae following SARS-CoV-2 infection. The search was performed on June 4th, 2021 and included all relevant articles published since January 1st 2020. Only primary articles published in peerreviewed journals in English with a quantitative outcome related to mental health were included, and time since acute COVID-19 had to be defined. Studies were included as sequelae when they were conducted after cessation of acute symptoms. As a minimum two negative qPCR tests had to be provided for the study to be included. Opinion articles, commentaries, reviews, and other articles without original data were excluded. Pre-prints without peer-review and case studies were not included.

Study selection and data extraction
Study selection and data extraction were performed using Covidence systematic review software (Veritas Health Innovation, Australia). Duplicate references were removed electronically and manually. Title/ abstract screening and full-text screenings were performed independently by two reviewers (CBR and TMS/GW). In cases of disagreements on the inclusion of an article at the title/abstract screening level, it was retained for the next screening stage. Upon disagreement on the inclusion or reason to exclude, the study was forwarded to a third reviewer (SJ), who made the final decision. We included studies containing primary data on psychiatric symptoms in adult patients with prior SARS-CoV-2 infection. We excluded studies without specified psychiatric presentations but included neuropsychiatric manifestations such as cognitive impairments and dyssomnia. Following descriptive variables were extracted for each study and presented in Table 1: Reference ID and country in which the research was conducted; Primary aim of the study; Study design; Study instruments related to the psychiatric assessment; Numbers of participants; Number of males; Mean age; Time since acute COVID-19 or since negative SARS-CoV-2 test; Inclusion criteria or population description; Exclusion criteria, when reported; Main findings related to psychiatric sequelae and other relevant results.

Results
The search resulted in 1725 unique references. Of these, 1540 were excluded as irrelevant during the title-abstract screening. Mostly these references were related to indirect effects of the COVID-19 pandemic on mental health. 172 references were full-text screened. Of these, 116 were excluded, primarily because they 1) did not contain primary data or 2) examined acute effects of SARS-CoV-2 infection rather than sequelae. After this thorough selection process, 66 articles were included in this review (Table 1). The study selection process is illustrated in Fig. 1. Included studies were from countries in Asia, Europe and North America. The patient material in the included articles was from subjects who were outpatients or had previously been hospitalized with COVID-19 either at the intensive care unit (ICU), in a ward, or the Emergency Department. Study designs include cohort studies, case-control studies, and case series. Time to follow-up ranged from day 1 after recovery until 7 months post-COVID-19 (see Fig. 2 for an overview of follow-up periods).

Depression and anxiety
di-sc-          harge from the hospital (Huang et al., 2021a). Disease severity was in several studies suggested to be a risk factor. Al-Aly et al. (2021) Taquet et al. (2021a) and (Halpin et al., 2021) but not (de Graaf et al., 2021) reported that anxiety and/or depression was highest in patients recovering from severe COVID-19, Gennaro et al. (2021) showed that duration of hospitalization correlated with depressive symptoms, Alemanno et al. (2021) reported a correlation between depressive symptoms and severity of the initial disease and de Graaf et al. (2021) showed that worse post-COVID functional status was associated with depression. Alemanno et al. (2021) also reported that the depressive symptomatology did not improve compared to depressive status at admission. In contrast Gennaro et al. (2021) reported anxiety to decrease from 1 to 3 months follow-up. Chevinsky et al. (2021) reported that anxiety was significantly higher in COVID-19 survivors than controls up to 60 days after recovery, but symptoms improved 90 days post-recovery. Depression is significantly higher than controls until 30 days after recovery whereafter the symptoms improve and no significant differences between cases and controls were found. Matalon et al. (2021) reported that depression and anxiety had normalised at the one month follow-up, but were predictors of longer lasting PTSD. Mazza et al. (2021) found persistent depressive symptomatology at 3 month follow-up, whereas anxiety had improved at this timepoint. Taquet et al. (2021a) found HR for anxiety and mood disorders to be elevated at 6 months follow-up but lower than at the 3 months follow-up. Iqbal et al. (2021) and Lorenzo et al. (2021) found that depression and anxiety did not correlate with time since recovery. Tomasoni et al. (2021) reported that neither anxiety nor depression was predicted by clinical parameters or disease severity but that the patients reported a higher degree of persistence of physical symptoms. Three studies examined baseline inflammatory markers concerning depression and anxiety at follow-up. Mazza et al. (2020) showed that women displayed lower baseline inflammatory markers but suffered more from both anxiety and depression. Patients with a previous psychiatric diagnosis showed high scores on most psychopathological measures, with similar baseline inflammation. Baseline systemic immuneinflammation index (SII) was positively associated with scores of depression and anxiety at follow-up Mazza et al., 2021). Gennaro et al. (2021) showed that systemic inflammation at admission predicted severity of depressive psychopathology at the 3 months follow-up. Baseline comorbidities have also been suggested to be essential for the development of depression or anxiety. Wong et al. (2020) reported that 22% percent of the patients with comorbidities at baseline suffered from anxiety or depression, whereas this was only 9% without baseline comorbidities. Gennaro et al. (2021), , Romero-Duarte et al. (2021) and (Sykes et al., 2021) reported that previous psychiatric history and female sex were predicters of depression and anxiety. Yuan et al. (2020) on the contrary reported depression not to correlate with sex, age, comorbidity, severity of initial infection, or initial illness duration. Instead, they found that these patients exhibited an elevated immune response as measured by increased white blood cell and neutrophil counts.

2020
; Wang et al., 2020b). Gennaro et al. (2021) and Mazza et al. (2021) reported that PTSD symptoms improved from 1 to 3 months follow-up. On the contrary, Lorenzo et al. (2021) found no improvement from 1 to 3 months follow-up. Of potential risk factors, several elements were highlighted. One study found no difference between ICU or no ICU admission (de Graaf et al., 2021), whereas both Halpin et al. (2021) and Horn et al. (2020) showed that PTSD was most pronounced after ICU admission. (Al-Aly et al., 2021) Another study found that being hospitalized had a preventive effect on PTSD . Sex has also been suggested as a risk factor. Bellan et al. (2021) indicated that being male increased the risk of developing PTSD, whereas De Lorenzo et al. (2020) and Poyraz et al. (2021) showed that female sex was a predictor of the disorder. De ; Horn et al. (2020); Poyraz et al. (2021) all showed that a previous psychiatric history or past traumatic events increased the risk of PTSD and Matalon et al. (2021) found that depressive and anxiety symptoms during acute COVID-19 were predictors of PTSD. Gennaro et al. (2021) and Mazza et al. (2020) examined whether baseline SII correlated with PTSD at follow-up, but did not find any such relationship.

Obsessive-compulsive disorder and psychotic disorders
Symptoms of obsessive-compulsive disorder (OCD) was examined in two studies Mazza et al., 2020). Mazza et al. (2020) screened patients for OCD and found that 20% suffered from OCD symptoms at follow-up. Gennaro et al. (2021) reported that any signs of OCD improved from 1 to 3 months follow-up. (Taquet et al., 2021a) found an increased incidence of psychotic diagnoses following COVID-19 compared to control cohorts.
One study showed that memory loss was present in 13% of the patients in the acute phase, whereas at the follow-up, 3 months later, 28% of the patients were affected by memory loss . In contrast, Alemanno et al. (2021) showed that cognitive deficits correlated with disease severity and had improved at follow-up 1 month after discharge compared to at admission. Al-Aly et al. (2021) also reported follow-up symptoms to be most pronounced after severe acute disease. In contrast, de Graaf et al. (2021) did not find differences in cognitive deficits between ICU and non-ICU patients. McLoughlin et al. (2020) did not find delirium to be a predictor of cognitive impairments. Sykes et al. (2021) described that at 100 day follow-up, memory impairments had improved, but 31% were still affected. Gennaro et al. (2021) reported that cognitive dysfunction was not predicted by sex, previous psychiatric diagnoses, or hospitalization duration, but by severity of depressive symptoms. Lu et al. (2020) performed MRI scans on patients at the 3-month follow-up and showed that patients displayed higher bilateral grey matter volumes (GMV) in the hippocampus. This correlated negatively with lactate dehydrogenase (LDH). Global mean diffusion (MD) of white matter (WM) correlated with memory loss. Furthermore, Zhou et al. (2020) but not Gautam et al. (2021) reported that cognitive dysfunction was associated with increased C-reactive protein (CRP) levels. Mazza et al. (2021) found baseline SSI levels to predict cognitive impairments at follow-up and Gennaro et al. (2021) found that systemic inflammation at hospital admission predicted neurocognitive performance in a multivariate analysis of variance whereas oxygen saturation or duration of hospitalization did not. Miskowiak et al. (2021) reported that cognitive impairments were associated with d-dimer levels during acute illness and residual pulmonary dysfunction (Ortelli et al., 2021).
Two studies reported fatigue to be independent of severity of acute COVID-19 (Sami et al., 2020;Townsend et al., 2020) whereas Halpin et al. (2021) and Sudre et al. (2021) found fatigue to be most pronounced in patients recovering from severe acute disease. The severity of fatigue improved from the acute phase to follow-up. In support of this, Chevinsky et al. (2021) and Iqbal et al. (2021) found that fatigue was still elevated in COVID-19 survivors at follow-up, but improved with time since recovery.
Acute levels of serum troponin-I correlated with fatigue at follow-up (Liang et al., 2020). In contrast, no association was found between markers of inflammation and cell turnover (leukocyte, neutrophil or lymphocyte counts, neutrophil-to-lymphocyte ratio, lactate dehydrogenase, CRP) or pro-inflammatory molecules (IL-6 or sCD25) and fatigue post COVID-19 (Townsend et al., 2020).

Discussion
We identified 5 major areas of deficits, namely depression/anxiety, PTSD, cognition, fatigue, and sleep disturbances. Additionally, OCD was reported in two studies and one article described an increased incidence of psychotic disorders following COVID-19. The results suggest that survivors of COVID-19 are at risk of psychiatric sequelae but that symptoms generally improve over time.
In summary, we identified and included 66 studies that provided information on psychiatric and neuropsychiatric sequelae of COVID-19. The studies were performed in Asia (16), Europe (37), North America (12), and Oceania (1) and follow-up periods ranged from immediately after recovery to 7 months post-recovery (Fig. 2). Thirteen studies compared results to one or more comparison group. The majority of the included studies were based on patients who had been hospitalized. This should be considered when evaluating the results and before extrapolating to, e.g., patients with mild symptomatology.

Depression and anxiety
Of the 46 studies screening for anxiety and depression, 10 found >30% of the patients affected. These results correspond with SARS-CoV and MERS-CoV sequelae, where an increased incidence of anxiety and depression has also been reported (Rogers et al., 2020). Intriguingly, a recent meta-analysis concluded that the prevalence of anxiety and depression in the background population (with unknown COVID-19 status) during the pandemic was >30% (Salari et al., 2020) suggesting that the increased incidence of depression/anxiety is caused by indirect effects of the pandemic. Speaking against this, is several of the included studies in the current review where large cohorts of COVID-19 survivors are compared to matched comparison groups (e.g. patients having survived other respiratory diseases during the pandemic). In these studies, COVID-19 survivors were in significantly increased risk of developing depression/anxiety at follow up (Al-Aly et al., 2021;Chevinsky et al., 2021;Daugherty et al., 2021;Mattioli et al., 2021;Taquet et al., 2021a;Taquet et al., 2021b). Only Noviello et al. (2021) who included 164 COVID-19 cases failed to find an increased risk compared to a control cohort 5 months after acute disease.
The variation in results between studies is likely to be affected by the use of very different study instruments, and that time to follow-up examinations differs substantially between studies. Disease severity and duration of symptoms also vary between reports and are even highlighted as risks factor in several of the included studies. Generally, anxiety and depressive symptomatology was reported to improve with time from acute disease. (Alemanno et al., 2021;de Graaf et al., 2021;Halpin et al., 2021;Huang et al., 2021a;Tomasoni et al., 2021). Further, only 7 studies contrast their results to comparison groups making it difficult to differentiate between direct and indirect effects of the COVID-19 pandemic.
Inflammatory markers were examined in some articles; one study showing elevated immune response at the time of the follow-up in patients with self-reported depression (Yuan et al., 2020) and two studies showing a correlation between baseline SII and anxiety/depression Mazza et al., 2020;Mazza et al., 2021). Also, Gennaro et al. (2021) reported that changes in SII predicted changes of depression during follow-up. Wong et al. (2020) reported that patients with baseline comorbidities were more likely to suffer from depression and/or anxiety. It therefore appears that an elevated inflammatory response is likely involved in the development of these symptoms. This is supported by a vast array of work on the relationship between neuroinflammation and depression (Dantzer et al., 2008;Raison et al., 2006). Interestingly, in the two studies which did not find increased incidence of anxiety or depression (Daher et al., 2020;Zhou et al., 2020) blood levels of IL-6 were not elevated. This is particularly interesting as IL-6 has been linked to depressive symptomatology in several studies (Achtyes et al., 2020;Dahl et al., 2014;Lindqvist et al., 2009). It should also be noted that the two studies which failed to report depression or anxiety were of a smaller sample size (n < 34).

Post-traumatic stress disorder
Symptoms-or diagnoses of PTSD was screened for and reported in 20 studies. Two of these studies reported that as many as 43% of the patients suffered from post-traumatic stress symptoms (Bellan et al., 2021). It is important to note that PTSD has been reported continuously throughout the pandemic as an indirect consequence of living under stress, uncertainty, and altered daily life rather than due to the disease itself Wang et al., 2020a). Also, surviving a critical illness has been shown to induce PTS symptoms (Sparks, 2018). Nonetheless, the reported levels of PTS are higher than what has been outlined in the background population, where affected individuals are reported to be 7-10% Tan et al., 2020).
The severity of COVID-19 has been highlighted as a risk factor for PTSD. Two studies reported that patients discharged from ICU were more likely to develop PTSD than non-ICU patients (Halpin et al., 2021;Horn et al., 2020), whereas one study found no difference between ICU and non-ICU patients (de Graaf et al., 2021). This difference could be caused by the fact that non-ICU patients can be a heterogeneous group, including severely ill patients who do not qualify for ICU treatment.
Male sex was in one study shown to be a risk factor for developing moderate-severe PTS (Bellan et al., 2021) whereas three studies described female sex as the predictor for PTSD symptoms in general Gennaro et al., 2021;Poyraz et al., 2021). Supporting the notion of a role of the female sex is a previous study examining PTSD in SARS survivors 30 months after recovery where female sex was also found to be an independent predictor of the disorder (Mak et al., 2010). A diagnosis of depression/anxiety is frequently reported as risk factors for the development of PTSD (Brady et al., 2000). In the current study, depression and anxiety during acute disease and early follow-up time-points were also reported to be predictors of subsequent development of PTSD Matalon et al., 2021). Mazza et al. (2021) but not Lorenzo et al. (2021) found PTSD symptoms to improve over time. Interestingly, Mak et al. (2009) reported PTSD as the most prevalent long-term psychiatric morbidity in SARS survivors.
Inflammation has previously been reported to be a pathophysiological mechanism in the development of PTSD (Lindqvist et al., 2014;Passos et al., 2015). Of the included studies Gennaro et al. (2021) and Mazza et al. (2020) examined whether a relationship between baseline SII and PTSD existed, but did not find such relationship. de Graaf et al. (2021) and van den Borst et al. (2020) showed that CRP and leukocyte count was elevated at admission, but normalised at follow-up. None of the included articles specifically examined cytokine levels in relation to PTSD, which will be highly relevant to explore in future studies.

Cognition
Delirium is a recognised complication of, e.g., respiratory illnesses in older adults. For acute COVID-19, the number of patients experiencing this complication is very high, with reports of up to 84% in severe illness cases at all ages (Helms et al., 2020b). The acute symptoms of delirium include disturbances of attention, awareness, and cognition. Not only do acute reports of cognitive deficits exist, but in the current review we have also identified 27 studies that all report cognitive decline at follow-up. The studies use different study instruments and measure various aspects of cognition. The deficits range from concentration problems (Halpin et al., 2021) to memory deficits (Halpin et al., 2021;Lu et al., 2020;Negrini et al., 2021) and praxis disabilities (Negrini et al., 2021) as well as formal diagnoses of dementia (Taquet et al., 2021a;Taquet et al., 2021b). It should be noted that McLoughlin et al. (2020) examined whether delirium was a predictor of cognitive impairments and did not find an association. It is, therefore, unclear whether initial cognitive deficits are related to the long-term effects. Interestingly, Lu et al. (2020) reported a higher degree of memory deficits at the 3 months follow-up compared to during the acute phase whereas Sykes et al. (2021) found memory impairments to improve with time. It should also be noted that Taquet et al. (2021a) found 0.67% of the COVID-19 survivors to suffer from dementia 6 months after the diagnosis and that patients with more severe acute disease were more likely to receive a dementia diagnosis than patients with milder COVID-19. Lu et al. (2020) showed that COVID-19 patients displayed brain abnormalities at a 3-month follow-up and that this correlated with memory loss and lactate dehydrogenase (LDH). LDH levels have previously been highlighted as markers for the severity of COVID-19 . The maintenance of normal cognition in older adults is highly correlated with low levels of LDH (Waters et al., 2020). No correlation was found between oxygen saturation and cognitive dysfunction, suggesting that the deficits were not caused by brain hypoxia. This is in contrast to Miskowiak et al. (2021) who suggested that restricted cerebral oxygen delivery could be involved in the development of cognitive sequelae. Zhou et al. (2020) reported that cognitive dysfunction correlated positively with CRP levels at follow-up and Mazza et al. (2021) showed that SSI predicted cognitive impairments at follow-up. This could indicate that neuroimmune alterations are involved.

Fatigue
COVID-19 induced fatigue can be defined as 'a decrease in physical and/or mental performance that results from changes in central, psychological, and or/peripheral factors due to the COVID-19 disease' (Rudroff et al., 2020). Fatigue was reported during acute COVID-19 but is also found to persist following recovery with debilitating effects for the individuals affected. In the present review we identified 32 studies that all found fatigue at follow-up. Fatigue was reported in both more severe COVID-19 cases (requiring hospitalization) as well as in milder cases. For SARS and MERS it was reported that fatigue was one of the most persistent long-term symptoms with accounts up to 39 months after the initial infection (Rogers et al., 2020). For COVID-19 several studies found fatigue to improve from acute disease to follow-up (Chevinsky et al., 2021;Iqbal et al., 2021;Petersen et al., 2020;Tenforde et al., 2020). Sun et al. (2021) reported that the median duration of fatigue in patients with mild COVID-19 was 14 days and 32 days in patients with severe disease. This could suggest that fatigue is not as severe as in the cases of SARS and MERS. It should though be noted that Albu et al. (2021) found that fatigue was the most debilitating long-COVID symptom, and the main reason patients contacted a COVID rehabilitation program.
Risk factors were female sex and pre-existing psychiatric diagnoses. This is consistent with a review identifying risk factors for persistent fatigue following acute infections (Hulme et al., 2017).
Of biological measures, acute levels of serum troponin-I was in one study reported to correlate with fatigue at follow-up suggesting that fatigue is possibly related to myocardial injury (Liang et al., 2020). Several investigators have noted similarities between post-COVID fatigue and myalgic encephalomyelitis/chronic fatigue syndrome (ME/ CSF). A recent systematic review found a strong overlap between the post-COVID symptomatology and the clinical presentation of ME/CFS (Wong and Weitzer, 2021). The pathogenesis of ME/CFS is not yet fully understood and likely multifactorial, but interestingly ME/CSF has previously been linked to infection with Epstein-Barr virus (EBV) and the EBV-induced gene-2 leading to neurological and immune-related symptoms (Kerr, 2019). Also enterovirus, cytomegavirus, human herpesvirus-6, human parvovirus B19 and Chlamydophila pneumoniae have been associated with ME/CSF (Chia and Chia, 2008) and it is therefore plausible that a SARS-CoV-2 infection may also be involved in the pathogenesis of ME/CSF. Future studies will be able to shed light on whether the pathophysiology of post-COVID fatigue and ME/CSF is comparable. Townsend et al. (2020) explored whether markers of inflammation and cell turnover could be related to fatigue but did not find any significant associations. Sample size for the study was n = 20/group.

Sleep disturbances
Sleep disturbances of various degrees were reported in 24 studies. Such results are similar to reports of SARS and MERS' long term effects >6 months after acute infection (Rogers et al., 2020). Sleep disturbances were by some authors found to be independent of severity of acute COVID-19, but to decrease with time since recovery Iqbal et al., 2021;Sami et al., 2020). Others report severity of acute disease to be a predictor of sleep disturbances at follow up (Huang et al., 2021a;Sudre et al., 2021;Taquet et al., 2021a) It should be noted that indirect effects of the pandemic have also led to an increase in reports of sleep disturbances. One study found that up to 20% of the included subjects suffered from insomnia . Such numbers are still lower than what is presented in the included articles and can therefore not account for the full extent of dyssomnia reported herein. Further, Taquet et al. (2021a) reported that significantly more COVID-19 survivors than comparison subjects suffered from insomnia 6 months after COVID diagnosis and that patients with a more severe acute disease course were at higher risk of insomnia than patients with milder COVID-19 symptoms.

Limitations
The main limitation of this review is the lack of studies comparing results to appropriate comparison groups. Without such comparisons it is not possible to completely differentiate between direct and indirect effects of COVID-19. In this review 13 out of 66 references compared results to comparison groups.
In the present review the goal was to include all available data at this specific time of the pandemic. By doing so, we have created a full overview of primary studies on psychiatric and neuropsychiatric sequelae, but it should also be stressed that the quality of the studies was very varied. Sample sizes ranged from 3 subjects to 266,586 and study instruments were not always standardized or the most appropriate for a specific outcome. Additionally, to be able to fully evaluate the indirect effects of the pandemic a more thorough evaluation of the temporal aspects will be needed. Further, the very fast time to publication of COVID-19 related studies is a possible concern that is highlighted in the number of retracted articles on the topic (Soltani and Patini, 2020). When more studies of high data quality have been published, meta analyses of the data should also be performed.

Conclusion
In summary, the results from the included studies suggest that survivors of COVID-19 are at risk of psychiatric sequelae and that psychiatric and neuropsychiatric sequelae are indeed an essential part of the long-COVID syndrome. The severity and time-frame of such sequelae are, at present, difficult to evaluate. Overall, a tendency towards symptom improvement over time exist. Likewise, severity of the acute infection also appears to be of importance for subsequent sequelae.
Only a few of the included studies contained comparison groups. It is therefore difficult to fully differentiate between the effects of the disease itself and indirect factors, such as living in a pandemic or having survived a severe disease. As highlighted previously, most of the included studies examined patients discharged from hospital, such that patients with mild or asymptomatic COVID-19 are not well represented. Essential for these aspects are the included reports of large cohorts of both hospitalized and non-hospitalized COVID-19 survivors compared to several appropriate comparison groups (Al-Aly et al., 2021;Chevinsky et al., 2021;Daugherty et al., 2021;Taquet et al., 2021a;Taquet et al., 2021b).
Highlighted risk factors were disease severity, duration of symptoms as well as female sex. Increased inflammatory markers were also reported, suggesting that neuroimmune alterations could be at least some of the disease's underlying course. Supporting this notion is the fact that the receptor for SARS-CoV-2 (ACE2) is expressed on neurons and glial cells (Gowrisankar and Clark, 2016;Nemoto et al., 2020), that SARS-CoV-2 can be detected in the brain, and that astrocytes and microglia cells are activated during COVID-19 (Matschke et al., 2020).