key: cord-0786620-emxgwaxy
authors: Tasleem, Amina; Wang, Yutong; Li, Kexin; Jiang, Xiaowen; Krishnan, Ankur; He, Chen; Sun, Ying; Wu, Yin; Fan, Suiqiong; Boruff, Jill T.; Markham, Sarah; Rice, Danielle B.; Bonardi, Olivia; Santo, Tiffany Dal; Li, Letong; Thombs-Vite, Ian; Agic, Branka; Fahim, Christine; Martin, Michael S.; Sockalingam, Sanjeev; Benedetti, Andrea; Thombs, Brett D.
title: Effects of mental health interventions among people hospitalized with COVID-19 infection: A systematic review of randomized controlled trials
date: 2022-04-09
journal: Gen Hosp Psychiatry
DOI: 10.1016/j.genhosppsych.2022.04.002
sha: f74847fedb077e0af0aa487430ce96d615a4f5c8
doc_id: 786620
cord_uid: emxgwaxy

OBJECTIVE: We evaluated the effects of mental health interventions among people hospitalized with COVID-19. METHODS: We conducted a systematic review and searched 9 databases (2 Chinese-language) from December 31, 2019 to June 28, 2021. Eligible randomized controlled trials assessed interventions among hospitalized COVID-19 patients that targeted mental health symptoms. Due to the poor quality of trials, we sought to verify accuracy of trial reports including results. RESULTS: We identified 47 randomized controlled trials from China (N = 42), Iran (N = 4) and Turkey (N = 1) of which 21 tested the efficacy of psychological interventions, 5 physical and breathing exercises, and 21 a combination of interventions. Trial information could only be verified for 3 trials of psychological interventions (cognitive behavioral, guided imagery, multicomponent online), and these were the only trials with low risk of bias on at least 4 of 7 domains. Results could not be pooled or interpreted with confidence due to the degree of poor reporting and trial quality, the frequency of what were deemed implausibly large effects, and heterogeneity. CONCLUSION: Trials of interventions to address mental health in hospitalized COVID-19 patients, collectively, are not of sufficient quality to inform practice. Health care providers should refer to existing expert recommendations and standard hospital-based practices. Registration: PROSPERO (CRD42020179703); registered on April 17, 2020.

People infected with COVID-19, including those who have been hospitalized, are at risk for negative mental health outcomes [1] [2] [3] [4] [5] . In addition to stressors faced by anybody with a serious medical condition, including the risk of disability and death, people hospitalized due to COVID-19 may be at particular risk of poor mental health outcomes due to (1) limited social contact between healthcare workers and patients, (2) smaller health care worker to patient ratios, (3) families being excluded from intensive care units, (4) limited resources such as beds and mechanical ventilators, (5) long-term health sequelae of and (6) stigma and discrimination associated with being infected [2] [3] [4] [5] . Strategies for mental health management for people hospitalized due to COVID-19 are needed to reduce short-and long-term negative mental health outcomes.

Two previous systematic reviews have evaluated the effects of non-pharmacological interventions to reduce psychological distress among intensive care unit (ICU) patients, but both concluded that methodological limitations and risk of bias, small sample sizes, and heterogeneity in populations and approaches limited confidence in results [6, 7] . Two systematic reviews [8, 9] have sought to evaluate the effects of interventions on mental health among COVID-19 patients.

One [8] searched through April 2020 and identified 16 studies but no randomized controlled trials [RCTs] . The other [9] searched for RCTs through July 2020, identified 5 (768 participants), and synthesized results quantitatively; however, low risk of bias ratings were applied to trials with major shortcomings, and results were pooled across highly heterogeneous interventions (muscle relaxation, respiratory rehabilitation, "life intervention", traditional Chinese nursing, internet-based intervention), which reduced confidence in synthesized results.

We are conducting a series of living systematic reviews [10] of changes in mental health symptoms during COVID-19 and the effects of interventions designed to improve mental health during the pandemic [11, 12] . Living systematic reviews are systematic reviews that are updated frequently and provide ongoing access to results as they become available [10] . They are logistically challenging but provide value beyond conventional systematic reviews in situations where (1) important decisions need to be made that merit the resources involved; (2) the certainty in existing evidence is low or very low, posing a barrier to decision-making; and (3) there is likely to be new research evidence emerging that would inform decisions [10] .

The objective of the present sub-study is to synthesize evidence from RCTs on the effects of mental health interventions for people hospitalized with COVID-19 infection. Results from trials of interventions for people not quarantined or undergoing treatment due to infection have been reported elsewhere [13] .

Our systematic reviews on COVID-19 mental health [11, 12] were registered (CRD 42020179703), and a protocol was posted (https://osf.io/96csg/) prior to initiation. Results are posted online as data are extracted (https://www.depressd.ca/research-question-3-intervention).

This manuscript adheres to the Preferred Reporting Items for Systematic Reviews and Metaanalysis statement [14] .

For trials to be included in the present review, all participants had to be enrolled after December 31, 2019, when China first reported on COVID-19 to the World Health Organization [15] . The population was restricted to patients receiving hospital-based care due to infection. Eligible interventions include any intervention described as designed primarily to J o u r n a l P r e -p r o o f address mental health symptoms from COVID-19. Trials that tested non-mental health interventions and primarily targeted non-mental health outcomes (e.g., exercise with primary outcome physical activity) were excluded. Eligible comparators included: (1) inactive control conditions (e.g., no treatment, waitlist) and (2) other eligible interventions. Eligible outcomes were defined broadly and included general mental health, mental health quality of life, anxiety symptoms, depression symptoms, stress, loneliness, anger, grief, burnout, and other emotional states. Eligible studies had to be RCTs that included at least 10 participants. There were no restrictions on language or publication format.

We did not include non-randomized studies because such studies are highly prone to bias when intervention and control groups are self-selected or there is no control group. Results from pre-post analyses of non-randomized studies without a control group are not possible to interpret unless there is a precise knowledge of the natural trajectory of symptoms or if one can safely assume that symptoms will not change over time without intervention. Even in normal times, however, this is not the case for mental health trials. Participants often seek mental health services and enrol in trials when they are experiencing high levels of symptoms, and regression to the mean is common [16] [17] [18] [19] [20] . Approximately 40% of participants assigned to placebo groups in drug trials or no-treatment groups in psychological intervention trials for major depression, for instance, achieve remission [21] . The Cochrane Collaboration discourages inclusion of evidence from non-randomized studies when conducting trials is feasible and when evidence from nonrandomized trials is subject to these kinds of biases [22] .

We searched MEDLINE (Ovid), PsycINFO (Ovid), CINAHL (EBSCO), EMBASE (Ovid), Web of Science Core Collection: Citation Indexes, China National Knowledge Infrastructure, Wanfang, medRxiv (preprints) , and Open Science Framework Preprints (preprint server aggregator), using a search strategy designed and built by an experienced health sciences librarian. The China National Knowledge Infrastructure and Wanfang databases were searched using Chinese-language terms based on the English-language strategy. (See Appendix A).

An initial search was conducted from December 31, 2019 to April 13, 2020, then automated searches were set for daily updates. As of December 28, 2020, the automatic daily search updates were converted to weekly updates for feasibility. The present report includes trials from searches conducted up to June 28, 2021. We conducted a pre-submission study review up to searches done on February 14, 2022 to determine if any verified studies had been published since the main search date.

Search results were downloaded into the systematic review software DistillerSR (Evidence Partners, Ottawa, Canada) where duplicate references were identified and removed.

Two reviewers independently screened titles and abstracts in random order to identify potentially eligible trials. If either reviewer deemed a study potentially eligible then a full-text review was conducted by two reviewers independently. Any disagreements at the full-text level were resolved through discussion and consensus, involving a third reviewer if necessary. An inclusion/exclusion coding guide was developed and pre-tested, and all team members were trained over several sessions. See Appendix B.

J o u r n a l P r e -p r o o f For each included RCT, one reviewer extracted data using a pre-specified standardized form in DistillerSR. A second reviewer validated the extracted data. Reviewers extracted (1) publication characteristics (e.g., first author, publication year, journal, funding source); (2) population characteristics and demographics (e.g., country, study eligibility criteria, number of participants, age, sex or gender, recruitment setting; (3) COVID-19 characteristics (e.g., severity); (4) intervention components, and (5) mental health outcomes. If sufficient information to calculate effect sizes was not provided, we attempted to obtain from authors.

We encountered many trials of unclear origin and funding, of very poor quality, and with effect sizes that exceeded plausibility. Thus, we emailed all included study authors twice (if no response to initial email) and requested that they verified the (1) authenticity of methods and results and (2) accuracy of our extracted outcomes. Authors of Chinese-language studies were contacted in both English and Chinese.

We used the 2011 Cochrane Risk of Bias tool [23] , which includes 7 domains that are rated as high risk, low risk, or unclear risk of bias. Two independent reviewers assessed each study independently; disagreements were resolved through discussion and consensus with a third reviewer consulted as necessary.

We used the Template for Intervention Description and Replication (TIDieR) checklist

[24] to evaluate intervention reporting adequacy. The checklist is comprised of 12 items that assess reporting of intervention name; rationale for use of the intervention; materials used; intervention provider and background; delivery mode; location and necessary infrastructure; intervention frequency and duration; any tailoring; any modifications; if adherence or fidelity J o u r n a l P r e -p r o o f was assessed; and, if assessed, degree to which intervention was delivered as planned. TIDieR was done independently by two reviewers with consensus-based resolution of conflicts, including a third reviewer if necessary.

We calculated between-groups standardized mean difference (SMD) effect size for each outcome using Hedges" g with 95% confidence interval (CI) when possible. We did not pool results quantitatively in a meta-analysis due to substantial heterogeneity of populations, interventions, and outcomes and because of serious concerns about risk of bias.

Protocol amendments are shown in Appendix C.

As of June 28, 2021, our search identified 56,854 unique titles and abstracts. Of these, 56,503 were excluded after title and abstract review and 196 at the full-text level, leaving 155 trials, of which 108 were excluded from the present sub-study because they were not RCTs (N = 21) or were not conducted with people infected with included RCTs. See Figure 1 . Additional searches from the search date to February 14, 2022, did not include any eligible RCTs that were registered prior to enrolling participants, that were verified, or that were low risk of bias on at least 4 of 7 domains.

Of the 47 included RCTs, 42 were from China [25, [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [39] [40] [41] [42] [43] [44] [45] [46] [47] [49] [50] [51] [52] [53] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] 4 from Iran [26, 37, 38, 54] , and 1 from Turkey [48] . All were conducted in 2020 or did not report when conducted. One trial was registered prior to initiation but reported outcomes that differed from tested psychological nursing with relaxation and empathic listening components [28, 35, 36, [40] [41] [42] ].

Out of the 21 trials, only 3 verified their results [31, 37, 38] . Few trials reported interventions in sufficient detail to fully understand what had occurred. Of the 8 main domains in the TIDieR checklist (4 additional domains were not applicable for most studies), all trials had 1-5 partially or not reported (median = 4). The number of trials that adequately reported was 21 (100%) for intervention name, 20 (95%) for rationale, 1 (5%) for materials used, 12 (57%) for procedures, 7 (33%) for provider, 19 (90%) for how administered, 6 (29%) for where administered, and 9 (43%) for when and how much provided. See Appendix D. Risk of bias was unclear or high in 20 of the 21 trials for blinding of participants and personnel and blinding of outcome assessment [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [39] [40] [41] [42] [43] [44] [45] , in 5 trials for random sequence generation [25, 26, 29, 32, 35] , in 17 trials [25-30, 32, 33, 35, 36, 39-45] for allocation concealment, in one trial for incomplete outcome data [43] , and in all 21 trials for selective reporting as none were adequately registered pre-trial . Excluding the "other bias" domain, the median number of the other 6 domains rated as unclear or high risk was 3. There were 3 trials [31, 37, 38] , the same trials that were verified, that were rated as low risk on at least 4 domains and not high risk on any domains other than blinding. See Table 2 .

Of the 21 trials, 20 reported anxiety symptoms [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [39] [40] [41] [42] [43] [44] [45] , 15 depression symptoms [25-31, 33-36, 39, 41-43, 45 ], 4 general mental health function [26, 33, 35, 38] , and 3 stress outcomes [26, 31, 38] . Hedges" g ranged from 0.24 to 17.87 for anxiety symptoms, 0.04 to19.61 J o u r n a l P r e -p r o o f for depression symptoms, 0.16 to 1.47 for mental health function, and 0.10 to 1.25 for stress. See Table 3 .

Among the 3 verified RCTs [31, 37, 38] rated low risk of bias on 4 of 7 domains, one (N = 93) [31] compared daily cognitive behavioral therapy (4 weeks) to routine care in hospitalized patients with mild COVID-19 in China; it did not find significant reductions in anxiety symptoms (Hedges" g = 0.24, 95% CI -0.17 to 0.65), depressive symptoms (Hedges" g = 0.04, 95% CI -0.37 to 0.45), or stress (Hedges" g = 0.10, 95% CI -0.51 to 0.31). Another trial (N = 110) [37] tested twice daily guided imagery (5 days) against routine care among hospitalized patients in Iran with oxygen saturation > 90% and reported statistically significant reductions in state (Hedges" g = 0.72, 95% CI 0.34 to 1.11) and trait anxiety (Hedges" g = 0.56, 95% CI 0.18 to 0.94). A third trial (N = 48) [38] compared a multi-faceted online psychological intervention to an offer to receive telephone counselling among mild to severely infected patients in a general hospital in Iran. It reported a non-statistically significant reduction in mental health function (Hedges" g = 0.49, 95% CI -0.09 to 1.08) and a significant reduction in stress (Hedges" g = 0.76, 95% CI 0.17 to 1.35)

The number of participants in the 5 physical and breathing exercise trials [46-50] ranged from 14 to 40 (mean = 51.0; SD = 20.5; median = 51). The percentage of female participants ranged from 39% to 55%. Mean age ranged from 35 to 69 years. Two trials tested breathing exercises [47, 50] , and three tested physical exercises [46, 48, 49] .

None of the 5 trials were verified. Out of 8 core TIDieR domains, each trial had 3 to 5 partially or not reported (median = 3). The number of trials that adequately reported was 5 J o u r n a l P r e -p r o o f (100%) for intervention name, 3 (60%) for rationale, 0 (0%) for materials used, 4 (80%) for procedures, 2 (40%) for provider, 4 (80%) for how administered, 0 (0%) for where administered, and 4 (80%) for when and how much provided. See Appendix D. Risk of bias was unclear or high in 2 trials for random sequence generation [47, 50] , one for incomplete outcome reporting [39] , and all 5 for allocation concealment, blinding of participants and personnel, blinding of outcome assessment, and selective outcome reporting [46] [47] [48] [49] [50] . Across trials, excluding "other bias", the median number of the other 6 domains rated as unclear or high risk was 4. No trials were at low risk of bias for ≥ 4 of 7 domains. See Table 2 .

Of the 5 trials, 5 reported anxiety symptom outcomes [46] [47] [48] [49] [50] , and 3 reported depression symptom outcomes [46, 49, 50] . Hedges" g ranged from -0.40 to 4.06 for anxiety symptoms and 0.78 to 2.99 for depression symptoms (Table 3) .

In the 21 mixed interventions, the number of participants ranged from 15 to 130 (mean=92.6; SD =49.3; median =88). The proportion of female participants ranged between 29% to 59%, and mean age ranged from 36 to 64. There were 12 trials that tested some combination of "humanized" nursing involving psychological care or a dietary intervention [52, 54-59, 61, 64-67] , 2 trials that assessed interventions with components of both music therapy and physical and breathing exercises [60, 63] , and 6 interventions of general medicine and health education or traditional Chinese medicine including acupoint massage [51, 53, [68] [69] [70] [71] . Intervention duration, frequency, and session length were inconsistently reported.

J o u r n a l P r e -p r o o f Of the 21 trials, none were verified. Interventions were generally poorly described, and each trial had 3 to 5 of the 8 core TIDieR domains either partially or not at all reported (median = 4). The number of trials that adequately reported was 21 (100%) for intervention name, 18 (86%) for rationale, 1(5%) for materials used, 14 (60%) for procedures, 2 (9.5%) for provider, 19 (91%) for how administered, 4 (19%) for where administered, and 2 (10%) for when and how much provided. See Appendix D. For risk of bias, random sequence generation was unclear in 9 trials [52, 54, 55, 61, [64] [65] [66] [67] 71] , and incomplete outcome data was unclear for 5 [51, 54, 59, 65, 66] .

All 21 trials were rated as unclear or high risk for allocation concealment procedures, blinding of participants and personnel, blinding of outcome assessment, and selective reporting .

Across trials, excluding "other bias", the median number of the other 6 domains rated as unclear or high risk was 4. No trials were at low risk of bias for ≥ 4 of 7 domains (Table 2) .

Among the 21 trials, 20 reported anxiety symptom outcomes [51-64, 66-71], 16 depression symptoms [51-53, 56, 58-65, 67, 68, 70, 71] , and 3 mental health function [57, 61, 64] . Hedges" g ranged from 0.45 to 26.98 for anxiety symptoms, 0.85 to 15.20 for depression symptoms and 1.76 to 2.28 for mental health function (Table 3 ).

We investigated the effects of mental health interventions among patients with COVID-19 infection. We identified and extracted data from 47 eligible RCTs from China (N=42), Iran (N=4) and Turkey (N=1), of which 21 tested psychotherapy interventions, 5 physical and breathing exercises, and 21 a combination of other types of interventions (e.g., "humanistic" nursing care). All trials were conducted in hospital settings and evaluated short-term, in-hospital mental health.

Overall, poor reporting and risk of bias limited our ability to draw conclusions about the effects of interventions. Of the 47 included trials, only 21 included contact information for authors, and institutional origin was unclear for many trials. Because of the very poor quality and uncertain origin of many trials, we attempted to contact authors to verify that trials had been conducted as reported and the accuracy of our extracted results; however, authors of only 3 trials verified results. We used the TIDieR tool to describe adequacy of intervention reporting. In each of the intervention categories (psychological, physical and breathing exercise, mixed), a median of 3 to 4 of 8 core TIDieR domains were partially or not at all reported. For risk of bias, across the 3 intervention categories, the median number of unclear or high risk of bias domains per trial was 3 for psychological interventions, 4 for physical and breathing exercises, and 4 for mixed interventions. Of the 47 trials, 25 (53%) reported SMD effect sizes greater than 2.0, which is several times the effect typically seen in mental health interventions, including in ICU settings [7] . Only 3 trials [31, 37, 38] , all of psychological interventions, were rated as low risk of bias on at least 4 risk of bias domains. These same 3 trials were the only 3 trials for which authors verified results. A cognitive behavioral intervention (N = 93) reported Hedges" g effects between 0.04 and 0.24 (all not statistically significant) for anxiety symptoms, depressive symptoms, and stress; a guided imagery intervention (N = 110) reported significant reductions in state (Hedges" g = 0.72, 95% CI 0.34 to 1.11) and trait anxiety (Hedges" g = 0.56, 95% CI 0.18 to 0.94); and an online multicomponent intervention reported a non-statistically significant reduction in mental health function (Hedges" g = 0.49, 95% CI -0.09 to 1.08) and a significant reduction in stress (Hedges" g = 0.76, 95% CI 0.17 to 1.35). It is difficult, however, to draw generalizable J o u r n a l P r e -p r o o f conclusions from these studies. They tested different types of interventions and, although better conducted than other included trials, still had important methodological limitations.

While issues of poor study design, methodology and reporting of results were a pervasive issue in research prior to COVID-19, concerns have been raised about the exceptionally high volume and unprecedented poor quality of research being generated in COVID-19. This is especially the case in COVID-19 research among trials with time pressures and insufficient research infrastructure, particularly in the earlier phases of the pandemic, which contributed to the conduct and publication of many poorly designed studies with sample sizes too small to answer the research questions being posed [72] . The set of trials included in the present review exceed what we have seen prior to COVID-19 or in other areas of COVID-19 mental health (e.g., [13] ) in terms of failing to provide minimally transparent reporting, verification of origin, overall study quality, and the unusually large effects reported. For comparison, over 50% of trials reported SMD effect sizes of 2.0 or greater versus only 1 of 11 (9%) trials in the most recent pre-COVID-19 systematic review of ICU-based interventions [7] .

Previous systematic reviews have, however, also reported that there are few wellconducted and reported RCTs of interventions for hospitalized patients infected with acute illnesses [5, 6] . The most recent systematic review [6] , which included meta-analyses of effects of different types of interventions (exercise, psychosocial, information, diaries, other) reported that diary interventions significantly improved depression and anxiety but there were only two small trials included with 88 total participants. Meta-analysis of 4 RCTs (N = 231) found that exercise interventions did not improve anxiety or depression symptoms; but meta-analysis of 7 different RCTs (N = 664) found that exercise improved general mental health function.

The limited evidence available on interventions to support mental health in acutely hospitalized patients prior to COVID-19, coupled with the very poor quality of trials done during COVID-19, reduces the ability of health care providers to use trial evidence to guide their clinical care of patients hospitalized with COVID-19. Furthermore, the resources to provide inhospital psychological care during COVID-19 may be limited due to demands on health care systems across the globe during the pandemic [73] as well as measures taken to reduce transmission. In the absence of actionable evidence, health care providers will need to rely on expert recommendations for ICU and other hospital care. These, generally, involve initial close attention to symptom stabilization, including strategies to manage pain, maintaining contact with loved ones through virtual calls, reduce fear and anxiety, and support sleep; the normalization of experiences; and reassurance that symptoms are expected to decrease in frequency and intensity with time. Types of interventions that may be helpful include stress reduction (e.g., breathing, meditative interventions), education, diaries, supportive therapy, and movement, as possible. As patients achieve increased stability, assessment of and strategies to address anxiety, depression, and posttraumatic stress can be initiated [74] .

Well-designed and conducted, adequately powered trials for different interventions to support stress reduction and improve short-and long-term mental health among hospitalized acute care patients are desperately needed in the context of COVID-19 and otherwise.

Additionally, interventions are needed to address mental health for people who experience longterm effects of COVID-19, as they appear to be at risk of negative cognitive and mental health outcomes [75, 76] . We did not include studies done with people with long COVID-19 outside of hospital, but these trials would have been included in our main living systematic review, and as of February 11, 2022 , no such trials had been identified.

There are several strengths of the present living systematic review. Firstly, it differs from prior systematic reviews as it investigated the effects of psychological interventions specifically among hospitalized COVID-19-infected patients based on randomized trials. Second, we included many more trials than have been included in any other review. Related to this, we included two Chinese databases in our search and included Chinese-language trials; these trials comprise the largest proportion of all trials that have been conducted. Third, we employed a rigorous methodology in conducting the present review based on best practices per the Cochrane

The findings of the current review must be interpreted in the context of limitations.

Chiefly, the quality of reporting and conduct of included trials was very poor, generally.

Documenting this rigorously is important for supporting health care providers to determine how usable the evidence might be and to underline the need for higher quality research in this area.

We were not, however, able to use the evidence to make recommendations about which interventions may be effective in practice. Any other limitations are secondary to this main concern. For instance, we did not attempt to examine the possibility of publication bias. It is possible that well-conducted trials with disappointing outcomes may have been conducted, but we do not believe this is likely given the overall quality issues we uncovered. It is likely that producing well-designed and conducted trials of mental health interventions in the initial phases of the pandemic was simply not feasible; with less intense pressure on health care systems in more recent phases, it is possible that trials will be conducted that will be of better quality and greater utility. Additionally, all trials were from 3 countries. It is possible that more trials in later stages of the pandemic will be done in countries that normally conduct the largest numbers of mental health and other health care service trials [78] .

In conclusion, the objective of the present review was to evaluate the effects of interventions to support mental health among hospitalized COVID-19 infected patients. Our inability to verify what occurred in the studies, very poor reporting and study quality, high risk of bias, and implausible reported effects, in many trials, did not allow us to draw conclusions about the likely effects of any included interventions. Our inability to draw conclusions from this evidence base is compounded by limitations in the quality of ICU-based mental health intervention trials available prior to COVID-19 [6, 7] . Accordingly, our review highlights the importance of and need for carefully conducted and reported RCTs. Without trustworthy evidence, health care providers should use strategies that are considered best practice to help patients hospitalized with COVID-19 achieve symptom stability as quickly as possible and to prevent and address long term mental health ramifications. Trials are needed, both in general ICU contexts and, to the degree as possible as the COVID-19 pandemic continues, that test and compare procedures that are commonly used with acutely hospitalized patients, including different approaches for symptom stabilization and management, stress reduction, and psychological tools designed to support recovery [74] . The limitations in existing trials may highlight limitations in infrastructure. Multi-center collaborations with an ongoing commitment to adaptive trial designs, such as platform trials [79] , might be considered to address the current lack of well-designed and conducted trials of interventions to support mental health in ICU settings, both prior to and during COVID-19. (MH "Stay at Home Orders") † S12 (MH "Quarantine") S11 S7 OR S8 OR S9 OR S10 S10 ( (MH "Pneumonia+") or TI (pneumonia) OR AB (pneumonia) ) AND ( TI (wuhan) OR AB (wuhan) OR AF (wuhan) ) S9 TI ( (severe acute respiratory syndrome coronavirus 2 or "SARS CoV-2" or "SARSCoV 2" or SARSCoV2 or cov2 or "sars 2" or COVID or "coronavirus 2" or covid19 or nCov or ((new or Novel) N3 coronavirus*) ) OR AB ( (severe acute respiratory syndrome coronavirus 2 or "SARS CoV-2" or "SARSCoV 2" or SARSCoV2 or cov2 or "sars 2" or COVID or "coronavirus 2" or covid19 or nCov or ((new or Novel) N3 coronavirus*) ) or (MH "Covid 19") † S8 TI ( (betacoronavirus* or beta coronavirus* or coronavirus* or corona virus*) ) OR AB ( (betacoronavirus* or beta coronavirus* or coronavirus* or corona virus*)) S7 S5 AND S6 S6 S1 OR S2 S5 S3 OR S4 S4 TI ( (china or chinese or hubei or wuhan) ) OR AB ( (china or chinese or hubei or wuhan) ) OR AF ( (china or chinese or hubei or wuhan) ) OR SO ( (china or chinese or hubei or wuhan) )

World Health Organization Coronavirus Disease (COVID-19) Dashboard

mental-health-and-psychosocial-support-emergency-settings/interim-briefing-noteaddressing-mental-health-and-psychosocial-aspects-covid-19-outbreak

Intensive care management of coronavirus disease 2019 (COVID-19): challenges and recommendations

Long covid -mechanisms, risk factors, and management

Attributes and predictors of long COVID

Nonpharmacological interventions to reduce ICU-related psychological distress: a systematic review

Nonpharmacologic interventions to prevent or mitigate adverse long-term outcomes J o u r n a l P r e -p r o o f Journal Pre-proof among ICU survivors: a systematic review and meta-analysis

Comparative effectiveness of multiple psychological interventions for psychological crisis in people affected by coronavirus disease 2019: a Bayesian network meta-analysis

Effects of non-drug interventions on depression, anxiety and sleep in COVID-19 patients: a systematic review and meta-analysis

Living systematic review: 1. Introduction-the why, what, when, and how

Curating evidence on mental health during COVID-19: A living systematic review

Living systematic review of mental health in COVID-19

Effects of COVID-19 mental health interventions among children, adolescents, and adults not quarantined or undergoing treatment due to COVID-19 infection: a systematic review of randomized controlled trials

The PRISMA 2020 statement: an updated guideline for reporting systematic reviews

World Health Organization. Rolling updates on coronavirus disease (COVID-19)

GRADE guidelines: 3. Rating the quality of evidence

Grading the strength of a body of evidence when assessing health care interventions: an EPC update

Systematic reviews in health care: Assessing the quality of controlled clinical trials

Explanation and Elaboration: updated guidelines for reporting parallel group randomised trials

Causal Inference: What If

The Challenges of Improving Treatments for Depression

Chapter 24: Including non-randomized studies on intervention effects

Quarantine* or Self-isolation or isolation or social distanc* or shelter*-in-place or psych* or mental health or mental illness* or mental disorder* or stigma or fear* or anxiety or anxious or depression or depressive or loneliness or stress* or trauma* or post-traumatic or posttraumatic or anger or mood* or irritability or irritable or emotional disturbance* or grief or burned out or burnout).tw,kf

* or beta coronavirus* or coronavirus* or corona virus*).mp.) and (exp china/ or (china or chinese or hubei or wuhan).af.)) or (coronavirus* or corona virus* or betacoronavirus* or beta coronavirus*).mp. 18. (severe acute respiratory syndrome coronavirus 2 or "SARS CoV-2" or "SARSCoV 2" or SARSCoV2 or cov2 or "sars 2" or COVID or "coronavirus 2" or covid19 or nCov or ((new or Novel) adj3 coronavirus*) or ncp).mp. or

20191231" or 2020* or 2021*).dt,ez,da

(severe acute respiratory syndrome coronavirus 2 or "SARS CoV-2" or "SARSCoV 2" or SARSCoV2 or cov2 or "sars 2" or COVID or "coronavirus 2" or covid19 or nCov or ((new or Novel) adj3 coronavirus*) or ncp).mp. 11

mental disorder* or Quarantine* or Self-isolation or isolation or social distanc* or shelter*-in-place or psych* or mental health or mental illness* or stigma or fear* or anxiety or anxious or depression or depressive or loneliness or stress* or trauma* or post-traumatic or posttraumatic or anger or mood* or irritability or irritable or emotional disturbance* or grief or burned out or burnout

2. (severe acute respiratory syndrome coronavirus 2 or "SARS CoV-2" or "SARSCoV 2" or SARSCoV2 or cov2 or "sars 2" or COVID or "coronavirus 2" or covid19 or nCov or ((new or Novel) adj3 coronavirus*) or ncp).mp. or ((exp pneumonia/ or pneumonia

TI ( (betacoronavirus* or beta coronavirus* or coronavirus* or corona virus*) ) OR

AB ( (betacoronavirus* or beta coronavirus* or coronavirus* or corona virus*) )

Coronavirus+") OR (MH "Coronavirus Infections+

Self-isolation" or isolation or "social distanc*" or "shelter*-inplace" or psych* or "mental health" or "mental illness*" or "mental disorder*" or stigma or fear* or anxiety or anxious or depression or depressive or loneliness or stress* or trauma* or "post-traumatic" or posttraumatic or anger or mood* or irritability or irritable or "emotional disturbance*" or grief or "burned out" or burnout) AND TOPIC: ((coronavirus* or "corona virus*" or betacoronavirus* or "beta coronavirus*" or "severe acute respiratory syndrome coronavirus 2" or "SARS CoV-2" or

IC Timespan=Year to date China National Knowledge Infrastructure Restricted to disciplines

+摘要:("隔离"+封城+"社交距离"+方舱+心理+"心理健康"+"精神卫生"+"精神

2020 are all for this purpose. To make this process more efficient, the disciplines of the China National Knowledge Infrastructure database were restricted to Medical and Public Health AND Social science subgroup 2 and those of Wanfang database were restricted to Medicine and Health AND Culture

理疾病"+污名+耻辱+羞辱+恐惧+焦虑)*("新冠"+"新型冠状")+摘要:("隔离"+封城+"社交

题名:("隔离"+封城+"社交距离"+方舱+抑郁+孤独+压力+应激+创伤+"创伤后"+愤怒+情

心理障碍"+哀伤+悲伤+悲痛+悲哀+忧郁+倦怠)*("新冠"+" 新型冠状")+摘要:("隔离"+封城+"社交距离"+方舱+抑郁+孤独+压力+应激+创伤+"创伤

理疾病"+污名+耻辱+羞辱+恐惧+焦虑)*("新冠"+"新型冠状")+摘要:("隔离"+封城+"社交

题名:("隔离"+封城+"社交距离"+方舱+抑郁+孤独+压力+应激+创伤+"创伤后")*("新冠 "+"新型冠状")+摘要:("隔离"+封城+"社交距离"+方舱+抑郁+孤独+压力+应激+创伤+"创

题名:("隔离"+封城+"社交距离"+方舱+愤怒+情绪+心情+易怒+"情绪障碍"+"心理障碍"+

新冠"+"新型冠状")+摘要:("隔离"+封城+"社交距

理疾病")*("新冠"+"新型冠状")+摘要:("隔离"+封城+"社交距离"+方舱+心理+"心理健康 "+"精神卫生"+"精神疾病"+"心理疾病")*

题名:("隔离"+封城+"社交距离"+方舱+污名+耻辱+羞辱+恐惧+焦虑+抑郁+孤独+压

新冠"+"新型冠状")+摘要:("隔离"+封城+"社交距离"+方舱+污名+耻辱+羞辱+恐惧

理疾病")*("新冠"+"新型冠状")+摘要:("隔离"+封城+"社交距离"+方舱+心理+"心理健康 "+"精神卫生"+"精神疾病"+"心理疾病")*

题名:("隔离"+封城+"社交距离"+方舱+污名+耻辱+羞辱+恐惧+焦虑+抑郁)*("新冠

+摘要:("隔离"+封城+"社交距离"+方舱+污名+耻辱+羞辱+恐惧+焦虑+抑郁)*(" 新冠"+"新型冠状")

题名:("隔离"+封城+"社交距离"+方舱+孤独+压力)*("新冠"+"新型冠状")+摘要:("隔离"+

+方舱+孤独+压力)*("新冠"+"新型冠状")

题名:("隔离"+封城+"社交距离"+方舱+应激+创伤+"创伤后"+愤怒+情绪+心情+易怒+

China National Knowledge Infrastructure Restricted to disciplines: Medical and Public Health & Social science subgroup 2 disciplines: Medicine and Health & Culture, Science, Education and PE 题名:("隔离" or 封城 or "社交距离" or 方舱 or 心理 or "心理健康" or "精神卫生" or

心理疾病" or 污名 or 耻辱 or 羞辱 or 恐惧 or 焦虑 or 抑郁 or 孤独 or 压力 or 应激 or 创伤 or "创伤后" or 愤怒 or 情绪 or 心情 or 易怒 or "情绪障碍" or "心理障碍" or 哀伤 or 悲伤 or 悲痛 or 悲哀 or 忧郁 or 倦怠) and ("新冠" or "新型冠状") or 摘要:("隔离" or 封城 or "社交距离" or 方舱 or 心理 or

理疾病" or 污名 or 耻辱 or 羞辱 or 恐惧 or 焦虑 or 抑郁 or 孤独 or 压力 or 应激 or 创伤 or "创伤后" or 愤怒 or 情绪 or 心情 or 易怒 or "情绪障碍" or "心理障碍" or 哀伤 or 悲伤 or 悲痛 or 悲哀 or 忧郁 or 倦怠) and

Open Science Framework (pre-prints)

OR psychology OR psychological OR psychosocial OR "mental health" OR "mental illness" OR "mental disorder" OR anxiety OR depression OR stress or trauma) AND (coronavirus OR COVID OR covid19)

Search 2: (psychology OR psychological OR psychosocial OR anxiety OR depression OR stress or trauma) AND (COVID OR covid19)

Ovid MEDLINE All †New subject heading added to original search on January 27, 2021 4. ("20191231" or 2020* or 2021*) .up.

S26 S11 AND S25 S25 S12 OR S13 OR S14 OR S15 OR S16 OR S17 OR S18 OR S19 OR S20 OR S21 OR S22 OR S23 OR S24 S24 TI ( (mental disorder* or Quarantine* or Self-isolation or isolation or social distanc* or shelter*-in-place or psych* or mental health or mental illness* or stigma or fear* or anxiety or anxious or depression or depressive or loneliness or stress* or trauma* or post-traumatic or posttraumatic or anger or mood* or irritability or irritable or emotional disturbance* or grief or burned out or burnout) ) OR AB ( (mental disorder* or Quarantine* or Self-isolation or isolation or social distanc* or shelter*-in-place or psych* or mental health or mental illness* or stigma or fear* or anxiety or anxious or depression or depressive or loneliness or stress* or trauma* or post-traumatic or posttraumatic or anger or mood* or irritability or irritable or emotional disturbance* or grief or burned out or burnout) ) S23 (MH "Burnout, Professional") S22 (MH "Grief+") S21 (MH "Anger") S20 (MH "Stress, Physiological") OR (MH "Stress, Psychological") S19 (MH "Depression") S18 (MH "Anxiety") S17 (MH "Fear") S16 (MH "Stigma") S15 (MH "Mental Health") or (MH "Mental Disorders") S14 (MH "Psychology") S13 (MH "Social Isolation") OR (MH "Loneliness") or (MH "Social Distancing") or J o u r n a l P r e -p r o o f

Title and Abstract Review:Exclude: not original human data or a case study or case series. If it is clear from the title and abstract that the article is not an original report of primary data, but, for example, a letter, editorial, systematic review or meta-analysis, or it is a single case study or case series, then it is excluded. Studies reporting only on animal, cellular, or genetic data are also excluded. Conference abstracts are included.Exclude: not a study of any population affected by the COVID-19 outbreak. Eligible studies must be initiated after China's first announcement to the WHO on December 31, 2019. If it is clear from the title or abstract that the study is not about any population affected by the COVID-19 outbreak, it is excluded. Studies that include fewer than 10 subjects, are excluded.

If it is clear from the title or abstract that the study is not about an intervention or is an intervention, but the intervention does not primarily target mental health, then it will be excluded. Mental health must be the primary trial outcome if a primary outcome or outcomes are stated.

If it is clear from the title or abstract that the study is not a RCT or nonrandomized controlled trial that compares an intervention designed to improve any aspect of mental health during the COVID-19 pandemic to (1) any inactive control condition (e.g., no treatment, waitlist control) or to (2) another eligible intervention designed to mental health, then it will be excluded.

Full-text Review:Exclude: not original human data or a case study or case series. If the article is not an original report of primary data, but, for example, a letter, editorial, systematic review or meta-analysis, or it is a single case study or case series, then it is excluded. Studies reporting only on animal, cellular, or genetic data are also excluded. Conference abstracts are included.

Eligible studies must be initiated after China's first announcement to the WHO on December 31, 2019. If the study is not about any population affected by the COVID-19 outbreak, it is excluded. Studies that include fewer than 10 subjects, are excluded.

The series of living systematic reviews was designed and initiated quickly early in COVID-19, and several amendments were made that are relevant to the present sub-study.First, we changed the automatic daily search updates to weekly updates as of December 28, 2020 to allow for more efficient processing of studies. Second, on January 27, 2021 we made a change to the MEDLINE search strategy to include a new subject heading on physical distancing created for COVID-19. Third, we made several amendments to the Chineselanguage search strategies to allow more efficient processing of studies (see Appendix A).Fourth, we added the TIDieR checklist to assess intervention reporting adequacy. Fifth, we clarified that we only included trials which began participant enrolment after December 31, 2019. Sixth, we did not include non-randomized trials. Such studies are highly prone to bias when intervention and control groups are self-selected or there is no control group. Results from pre-post analyses of non-randomized studies without a control group are not possible to interpret unless there is a precise knowledge of the natural trajectory of symptoms or if one can safely assume that symptoms will not change over time without intervention. Even in normal times, however, this is not the case for mental health trials, and the Cochrane Collaboration discourages inclusion of evidence from non-randomized studies in systematic reviews when conducting trials is feasible [1] . Seventh, because we encountered many trialreports of poor quality with seemingly implausible results, some with unclear origin, we added the author verification process.[1] Reeves BC, Deeks JJ, Higgins JPT, Shea B, Tugwell P, Wells GA. Chapter 24: