key: cord-0944186-4idnjykz
authors: Liu, Dong; Baumeister, Roy F.; Zhou, Yong
title: Mental Health Outcomes of Coronavirus Infection Survivors: A Rapid Meta-Analysis
date: 2020-10-21
journal: J Psychiatr Res
DOI: 10.1016/j.jpsychires.2020.10.015
sha: ecbbaf7351199d015c6389af6435ecd2fc21afcc
doc_id: 944186
cord_uid: 4idnjykz

BACKGROUND: The current COVID pandemic is happening while the long-term effects of coronavirus infection remain poorly understood. The present article meta-analyzed mental health outcomes (anxiety, depression, etc.) from a previous coronavirus outbreak in China (2002). METHOD: CNKI, Wanfang, PubMed/Medline, Scopus, Web of Science, Baidu Scholar, and Google Scholar were searched up to early June 2020 for articles in English or Chinese reporting mental illness symptoms of SARS patients. Main outcome measures include SCL-90, SAS, SDS, and IES-R scales. 29 papers met the inclusion criteria. The longest follow-up time included in the analysis was 46 months. FINDINGS: The systematic meta-analysis indicated that mental health problems were most serious before or at hospital discharge and declined significantly during the first 12 months after hospital discharge. Nevertheless, average symptom levels remained above healthy norms even at 12 months and continued to improve, albeit slowly, thereafter. INTERPRETATION: The adverse mental health impact of being hospitalized with coronavirus infection long outlasts the physical illness. Mental health issues were the most serious for coronavirus infected patients before (including) hospital discharge and improved continuously during the first 12 months after hospital discharge. If COVID-19 infected patients follow a similar course of mental health development, most patients should recover to normal after 12 months of hospital discharge.

We searched the published Chinese and English literature examining SARS to identify the long-term psychiatric status for the SARS survivors. We meta-analyzed the sustained psychiatric symptoms at different follow-up time points to examine how the mental status of SARS survivors changed after infection and after hospital release. We recognize one previous meta-analysis for SARS and MERS survivors, but it included only 7 studies, and in particular the long-term mental health outcomes of SARS patients remained unclear (Rogers et al., 2020) . The present meta-analysis included four times as many studies, some of which followed up almost four years after hospital discharge, enabling a much better picture of the long-term impact.

We searched CNKI, WANFANG, Baidu Scholar, Google Scholar, PsycINFO, and Medline databases for studies or abstracts published until June 10, 2020. We used a combined set of keywords to identify SARS related studies. The search terms combination was: (SARS OR severe acute respiratory syndrome OR coronavirus) AND (mental health OR anxiety OR depression OR SCL-90 OR SAS OR SDS OR Post traumatic stress disorder OR PTSD OR Impact of Event Scale -Revised OR IES-R OR life quality). Inclusion criteria were original articles in English or Chinese that reported statistics of SCL-90, SAS or SDS scores of SARS patients.

Articles were excluded for the following reasons: lack of original data (reanalyses of previously analyzed datasets); failing to report essential mental health scores; use of nonstandard mental health measures (indeed we relied only on studies using SCL-90, SAS , SDS, or IES-R);

reporting only SCL-90 total scores (i.e., failing to provide subscale data); reporting only J o u r n a l P r e -p r o o f percentages of positive cases; focusing on other infectious diseases such as MERS; and failing to report data by specific follow-up times.

Data were extracted by the first author and one graduate student. Descriptive variables extracted were average score and standard deviation of SCL-90, SAS or SDS scores, percentage of positive symptoms, number of cases, age, female proportion, and follow-up time. All nine scores of SCL-90 symptom dimensions were coded separately. The 3 subscale scores of IES-R (intrusion, avoidance, and hyperarousal) were coded separately. If a study simultaneously reported the score at several follow-up time points, all effects were coded.

Most studies did not include control groups, which made it impossible to compare across different scales even when they measured the same symptoms. To deal with that, we limited our review to studies that used the most frequently reported scales, for which published norms are available. We calculated all analyses using the Comprehensive Meta-Analysis (CMA 3.0). We used I 2 and τ to estimate heterogeneity variance. Values of I 2 greater than 35% were deemed indicative of study heterogeneity. We used random-effects models for the analysis because of the high heterogeneity. Because of the paucity of studies reporting percentage of positive mental health symptoms, we only meta-analyzed the score of mental illness scales.

Results of meta-analyses were grouped by follow-up time (admission, in hospital, hospital discharge, 1 month after discharge etc.). Furthermore, we performed a subgroup analyses to compare the effects at different follow-up time points. Funnel plots, Begg and Egger tests were conducted to check for the publication bias.

The initial search yielded 1124 results. Initially screening of abstracts left 50 articles. We further excluded 14 studies based upon our exclusion criteria. For the remaining 36 studies, 16 studies used the SCL-90, 13 studies used the SAS, 14 studies used the SDS, and 5 studies used the IES-R. Ultimately, 29 studies and 385 effects were included in our meta-analyses. This rapid meta-analysis followed PRISMA guidelines. Details of the selection of studies can be found 

We used the criteria established by Rogers (2020), which was adapted from the Newcastle Ottawa Scale, to assess the quality of the study, see appendix. The coders rated the quality of the included studies. 21 of the 29 studies were rated poor or medium quality and only 8 were of high quality.

For symptom severity scores, the weighted mean symptom score for the SCL-90 anxiety subscale at 12 months after hospital discharge, was 1.49 on a scale from 0 to 4, with higher scores meaning more symptoms (95% CI 1.41-1.58, N= 257). The anxiety subscale score was the highest at hospital admission (M=2.12; 95% CI 1.24-3, N=942). The weighted mean symptom score for the SCL-90 depression subscale at 12 months after hospital discharge, was 1.57 (95% CI 1.48-1.67, N= 257). The depression subscale score was the highest in hospital 2.26 (95% CI 1.74-2.77, N=320).

Somatization is the tendency to experience physical symptoms of a psychiatric condition such as depression. The weighted mean symptom score for the SCL-90 somatization subscale J o u r n a l P r e -p r o o f score was the highest at hospital admission 1.96 (95% CI 1.6-2.33, N=1034), declined during hospital time to 1.81 (95% CI 1.61-2.01, N=320), and rebounded at hospital discharge (M= 1.93; 95% CI 1.79-2.08, N=799), and then dropped during the first two months and maintained at a high level even after 12 months.

The weighted mean symptom score for the SCL-90 interpersonal sensitivity subscale fluctuated following a similar pattern of depression score. It reached the highest point at hospital discharge 1.79 (95% CI 1.34-2.24, N=942), and dropped to its lowest level at 12 months after hospital discharge 1.63 (95% CI 1.54-1.73, N=257).

The weighted mean symptom score for the SCL-90 hostility subscale score increased to the highest at hospital discharge 1.78 (95% CI 1.56-2, N=591), and then keep falling and dropped to 1.47 (95% CI 1.39-1.56, N=257) at 12 months. The weighted mean symptom score for the SCL-90 phobic anxiety subscale score was the highest at hospital admission 1.88 (95% CI 0.72-3.04, N=942).

Patients at discharge had the most severe symptoms on obsessive-compulsive disorder: the weighted mean score was 2.01 (95% CI 1.83-2.19, N=591). The paranoid ideation and psychoticism symptoms level were relatively low across all time periods compared with other SCL-90 symptom dimensions.

All sub-scores of SCL-90 dropped significantly after release from hospital. The effect changes (d) from the highest point to the end of the first 12 months were respectively 5. 58, 7.93, 6.29, 3.29, 5.74, 3.88, 4.71, 3, 3 .61 for anxiety, depression, somatization, interpersonal sensitivity, hostility, phobic anxiety, obsessive-compulsive disorder, paranoid ideation and psychoticism. Depression symptoms improved the most and paranoid ideation symptoms J o u r n a l P r e -p r o o f improved the least (though the latter were low throughout, yielding therefore relatively little room for improvement).

To estimate the degree of recovery, we consulted a sample of normal and healthy people in China (N=1890, year=2003) (Tong, 2006) . The SCL-90 scores of SARs survivors across the studies reviewed here at 12 months after hospital discharge were still slightly higher than the scores of the general population sample at most dimensions. Thus, the mental health problems diminished over the first year after having SARS but did not entirely disappear even after a year.

We also meta-analyzed the SARS patients' anxiety and depression score with SAS and SDS score. The results showed that the SDS score was the highest in hospital 48.87 (95% CI 42.53, 55.21) (no data at hospital admission was reported) and dropped to the lowest level 33.44 (95% CI 18.75, 48.14) at 12 months after discharge. SDS scores can range from 20 to 80, with most depressed people scoring 50-69, and above 70 indicating severe depression. The SAS score fluctuated in a similar pattern to SDS score and declined from 50.21 (95% CI 42.99-57.42, N=325) in hospital to 29.72 (95% CI 15.52,43.92) at 12 months after discharge. SAS total scores can range from 20 to 80, and 36 is the cutoff score for clinical screening. The effect of change for SAS and SDS score were d =7.80 and 4.92. These indicate quite large drops in mental health symptoms during the first year after release from hospital.

All 3 subscales of IES-R scores reduced slowly during the first 12 months after hospital discharge. The changes of subscale scores were not obvious. The effect of change for IES-R avoidance, intrusion, hyperarousal, and total scores were d =1.11, 1.06, 1.13, 3.24. Unfortunately, there were not enough data on PTSD to permit reliable meta-analysis.

We used funnel plot and Egger's test for publication bias at each time point. Only time points with more than 5 effect sizes were analyzed. The funnel plots were symmetrical, and the Egger's tests were not significant. Thus, no evidence of bias was found.

To our knowledge, this is the first systematic meta-analysis of the long-term mental health status of coronavirus infection on hospitalized patients. Across the 29 studies included in the meta-analyses, mental symptoms were widespread at clinically significant levels upon release from hospital. They declined significantly during the ensuing year, and on average dropped out of the clinically significant range -but the symptoms remained higher than norms for healthy individuals, and some individuals continued to have clinical levels of symptoms beyond a year.

We identified 16 independent studies that reported specific statistics of SARS patients' mental health effect using SCL-90. The SCL-90 scale, consisting of 9 dimensions, is the most widely used psychological inventory in China to measure patient mental health status in 2000s.

The 9 dimensions it includes are somatization, obsessive-compulsive disorder, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychoticism.

Among them, somatization, depression, anxiety, phobic anxiety, and obsessive-compulsive disorder are the most severe symptoms that SARS patients reported at the initial stage of infection. SARS patients also experienced a variety of physical symptoms such as shortness of breath and pain.

During the first 12 months after hospital discharge, all 9 dimensions of symptoms declined significantly. The aggregate scores of SDS and SAS likewise dropped sharply during the first 3 months after hospital discharge. There was some evidence of continued improvement (symptom reduction) beyond 12 months. The SAS scores dropped slowly from 31.94 (95% CI J o u r n a l P r e -p r o o f 27.68-36.2, N=18) at 12 months after hospital discharge to 29.72 [15.52-43.92 , N=114] at 46 months after hospital discharge. The SDS scores dropped significantly from 39.98 (95% CI 37.9-42.05, N=116) at 12 months after hospital discharge to 33.44 [18.75-48.14, N=114] at 46 months after hospital discharge, suggesting the depression symptoms continued to diminish after the 1 st year of hospital discharge.

Data from IES-R scores suggest similar patterns, but there were relatively few studies using this scale, so our findings with it may be less stable than with the other measures. All 3 subscales of IES-R scores reduced slowly during the first 12 months after hospital discharge.

Unfortunately, we did not find any reports on SARS' patients' IES-R scores beyond 12 months.

Several recent studies available for COVID-19 patients' mental status currently provide preliminary information about how COVID-19-related psychiatric symptoms develop and change. During their hospital stay, a significantly high proportion of patients reported depression (60.2 %), anxiety (55.3%) (Guo et al., 2020) and PTSD (96.2%) (Bo et al., 2020) . Liu et al. (2020) found that the prevalence rate of clinically significant depression, anxiety, and PTSD symptoms for hospital discharged COVID-19 patients are respectively 19%, 10.4%, and 12.4%, which is a significant drop compared with Bo's finding. But no longer-term follow-up data after hospital discharge are available for COVID, because the pandemic is still less than a year old. Differences may emerge between SARS and the more recent COVID-19, but for now, the data on SARS provide a basis for speculatively predicting what will happen to people suffering from COVID-19 in the coming months.

This study provides a comprehensive data set of mental health outcomes and changes of coronavirus infected patients. Nevertheless, several limitations must be acknowledged. Most J o u r n a l P r e -p r o o f studies included in our analyses were of low to moderate quality. All studies used Chinese adult samples, which limits the generalizability of our findings. In particular, no adolescent or child samples were available. Most studies were cross-sectional and lacked baseline psychiatric assessments before coronavirus infection. Most studies collected data on patients' mental health status within the 1 st year after hospital discharge, so longer-term data beyond 12 months were scarce.

The data mainly concern people who were hospitalized and thus presumably had severe forms of the illness. With COVID, many people have no or minimal physical symptoms (while others become intensely sick), and it seems reasonable to assume that the people with the worst physical symptoms will also be at risk for the most severe mental health symptoms. In other words, our findings should not be generalized to everyone who is infected with the coronavirus but rather only to the more severe cases.

We relied on the most commonly used measures, but inevitably these omit mental health issues that depend on other measures. In our view, the most serious gap in the literature we reviewed was PTSD. Our sample did not have enough PTSD data to analyze. Future work should attend particularly to PTSD, given that these symptoms sometimes last far longer than others.

The coronavirus causes physical illness, but it also has lasting mental health consequences (at least for people whose illness is severe enough to warrant hospitalization). The present data cannot address the important question of what causes these mental health problems.

They may be due to direct action by the virus on the brain and central nervous system.

Alternatively, they may arise from the stresses caused by hospitalization with poorly understood illness amid widespread societal concern, and/or experiences such as exposure to deaths of other hospital patients and family members.

Our review of studies done on people afflicted with the 2002 SARS coronavirus found that people who were hospitalized with that virus retained significantly elevated levels of mental illness symptoms even 12 months after hospital discharge -although, fortunately, all symptoms declined by substantial amounts during that first year, and the majority of people were no longer in the clinically significant range after one year. The problems were not confined to one particular symptom but rather were diverse, indeed covering all nine subscales of the SCL-90 measure (though paranoid ideation and psychoticism scores were generally lower than the others). Nearly all symptoms were worst at or before hospital discharge, so there is a general trend toward improved mental health over the months after discharge. Nevertheless, it seems fair to conclude that the mental symptoms stemming from coronavirus infection endure much longer than the physical symptoms of the disease. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Table 9 Studies included in the meta-analysis of SCL-90 psychoticism subscale score. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample. Full-text articles assessed for eligibility (n =80)

Full-text articles excluded with reasons (n =135)

Studies assessed with excluding criteria (n = 51)

Studies included in quantitative synthesis (meta-analysis) (n = 29)

Full-text articles excluded, with excluding criteria (n = 22)

Posttraumatic stress symptoms and attitude toward crisis mental health services among clinically stable patients with COVID-19 in China

Immediate psychological distress in quarantined patients with COVID-19 and its association with peripheral inflammation: a mixed-method study

A follow-up study of post-traumatic stress disorder of SARS patients after discharge

Mental morbidities and chronic fatigue in severe acute respiratory syndrome survivors: long-term follow-up

Risk factors associated with mental illness in hospital discharged patients infected with COVID-19 in Wuhan

All authors declare that they have no conflicts of interest. Note: -2 refers to hospital admission; -1 refers to in hospital; 0 refers to hospital discharge. Time = months after hospital discharge; N = sample size; Female = female proportion in the sample.