Interventions for mental health, cognition, and psychological wellbeing in long COVID: a systematic review of registered trials

Background Among patients diagnosed with COVID-19, a substantial proportion are experiencing ongoing symptoms for months after infection, known as ‘long COVID’. Long COVID is associated with a wide range of physical and neuropsychological symptoms, including impacts on mental health, cognition, and psychological wellbeing. However, intervention research is only beginning to emerge. This systematic review synthesizes currently registered trials examining interventions for mental health, cognition, and psychological wellbeing in patients with long COVID. Methods Standard systematic review guidelines were followed. Trials registered in two large trial registries in 2020 to May 2022 were reviewed. Included studies were narratively synthesized by type of intervention and a risk-of-bias assessment was conducted. Results Forty-two registered trials were included, with a total target sample size of 5814 participants. These include 11 psychological interventions, five pharmacological and other medical interventions, and five evaluating herbal, nutritional, or natural supplement interventions. An additional nine trials are examining cognitive and neurorehabilitation interventions and 12 are examining physiotherapy or physical rehabilitation. Most trials are randomized, but many are feasibility trials; trials are evaluating a wide spectrum of outcomes. Conclusions While there is a newly emerging body of research testing interventions for mental health, cognition, and psychological wellbeing in long COVID, the breadth and scope of the research remains limited. It is urgently incumbent on researchers to expand upon the intervention research currently under way, in order to generate high-quality evidence on a wide range of candidate interventions for diverse long COVID patient populations.

, as social interaction, pro-social activities, physical activity, and everyday life have been radically transformed. The research on mental health in long COVID remains scant (Vannorsdall & Oh, 2021). However, it appears that long COVID can be accompanied by anxiety, depression, and posttraumatic stress disorder, as well as neurocognitive issues (Raveendran et al., 2021); these, in turn, can be complicated by the physiological and neurological symptoms that prevent people from returning to their previous level of functioning (Aiyegbusi et al., 2021). People with long COVID are at risk of combining the mental health impacts of long COVID with those associated with population-level pandemic response strategies (Brüssow & Timmis, 2021). Unfortunately, as mental healthcare needs have increased, the mental healthcare system has also been disrupted (Mann, Chen, Chunara, Testa, & Nov, 2020). A shift to virtual care has exacerbated the digital divide and changed willingness to seek services (Hoyer et al., 2021), compromising access to timely mental health support.
The National Institute for Health and Care Excellence (NICE) has issued clinical practice guidelines for the treatment of long COVID, in which they recommend integrated and interdisciplinary models of care to meet the wide range of long-term needs with which these individuals may present (International Foundation of Integrated Care, 2020). As part of integrated treatment models, it is critical that we combine physical healthcare with social services, mental health supports, cognitive rehabilitation, and psychiatric treatments when indicated (Aiyegbusi et al., 2021). Given the complexity of the physical, cognitive, psychological, and social impacts of long COVID in the context of the ongoing pandemic, there is a need for multi-facetted, complex interventions that are adapted to the individual and the local context. This intervention complexity requires appropriate evaluation, ideally following the Medical Research Council's framework for evaluating complex interventions (Skivington et al., 2021).
Interventions that support mental health and psychological wellbeing have been shown to help people with physical health challenges (Ferrier et al., 2021;Gilbert et al., 2012;Jenkins et al., 2021b). By building resiliency in vulnerable populations, it is possible to build positive mental health that supports disease management and improves quality of life. Pharmacological interventions are also sometimes indicated for mental health problems secondary to physical health diagnoses (National Collaborating Centre for Mental Health, 2010). It is therefore important to consider such interventions within multi-component, integrated models of care for long COVID. However, since long COVID is an emerging clinical entity, effective interventions for this complex condition constitute a critical gap in the literature. To advance research agendas in this area, it is important to understand the current state of the research, even at this early stage, in the absence of a large body of published evidence.
This systematic review aims to support the rigorous planning of research agendas by synthesizing the currently registered trials examining interventions for mental health, cognition, and psychological wellbeing in long COVID.

Methods and analysis
This systematic review of registered trials follows the Preferred Reporting Items for Systematic Reviews and Meta Analyses (PRISMA) .

Trial retrieval
An electronic search was conducted of two large, international registries for clinical trials: clinicaltrials.gov, a trial registry by the United States National Library of Medicine, and the International Clinical Trials Registry Platform, an aggregator of international trial registries by the World Health Organization. The search was conducted on 4 January 2022 and updated on 31 May 2022, covering the first 5 months of 2022.
Recognizing that this is a new, emerging literature, multiple search strategies were piloted to identify keywords. Based on the relatively small number of trials on the topic, it was decided not to use intersecting search terms, to balance risk of missing trials v. over-screening. Therefore, search terms were long COVID, OR post COVID, OR post-acute COVID, OR long haul*, OR COVID sequelae, OR sequelae of COVID, OR COVID survivor, OR post-SARS-COV-2. There were two filters: (1) the date-of-registration of the trial had to be between 1 January 2020 and 31 December 2021 or between 1 January 2022 and 31 May 2022 for the update; (2) only interventional trials were included (filter available on clinicaltrials.gov only). All trials found with these search parameters were uploaded into Covidence systematic review software (Veritas Health Innovation, 2021), where duplicates were automatically deleted and visually confirmed. For records in the final record set, published articles with outcome data were manually sought in MEDLINE and PsycINFO using the study ID and lead researcher names.

Eligibility criteria
From a PICO (population, intervention, control, outcomes) perspective (Richardson, Wilson, Nishikawa, & Hayward, 1995), studies were eligible if they addressed individuals with long COVID, used any intervention, with or without a comparison group, and had either a primary intervention aim or primary outcome specific to mental health, cognition, or psychological wellbeing. Studies with quality of life as a primary outcome were included only if the quality of life measure contained a mental health or psychological wellness subscale, ensuring that a mental/psychological component is to be included in the outcomes rather than limiting the outcome to physical health-related quality of life. Each study's own definition of long COVID was accepted, provided that the record referred to the concept of long COVID and recruited individuals at a minimum of 1 month after acute COVID-19. Records could originate from any country and could report on participants of any age group and with any sociodemographic characteristics. Excluded were trials registered prior to 2020 (i.e. before the pandemic) or after 31 May 2022, noninterventional trials, and trials that did not focus on mental health, cognition, or psychological wellbeing.

Study selection
The initial search yielded a total of 912 records, among which 150 duplicates were automatically removed by Covidence (Veritas Health Innovation, 2021), for an initial search of 762 records. The update produced an additional 158 records, including 18 duplicates, for a total of 902 records. These records were reviewed first at the title and project summary level based on inclusion and exclusion criteria. After a training and calibration review of 25 records by a research lead and a research staff, the research staff and lead both independently screened 152 records (152/762 = 20%), achieving 92.8% agreement, with an inter-rater agreement of κ = 0.74 [substantial agreement (Sim & Wright, 2005)]. Any conflicts were resolved by consensus. The research staff independently screened the remaining records, with open discussion of any uncertainties. An additional 30 duplicates were removed manually during screening. Records that were retained were then screened at the full record level by both the research lead and research staff for 21 initial records (21/100 = 21%), with 90.5% agreement [κ = 0.74, substantial agreement (Sim & Wright, 2005)]. The research staff screened the remaining records at the initial search and all records at the updated search independently, bringing any uncertainties to the project lead for discussion and resolution by consensus. The final record set was reviewed and confirmed by the research lead.

Data extraction and synthesis
Data were extracted into an Excel spreadsheet by the study staff, with ongoing discussion. Data extraction included descriptive information about the (1) trial as a whole (i.e. the study ID, funder/sponsor, date of registration, country of the principal investigator, countries of recruitment, scientific title), (2) intervention (i.e. name, type, description, delivery mode, dose, frequency, length), (3) study design (i.e. allocation, model, masking, arms, recruitment status, start, and expected end date), (4) sample (i.e. long COVID definition if available, age, sample size, inclusion, exclusion criteria), and (5) outcome measures (i.e. primary outcome(s), secondary outcome(s), measures, timing of measurement). Data were synthesized and summarized in narrative and table format based on the type of intervention.

Quality assessment
The Cochrane Risk of Bias 2.0 (Higgins, Savović, Page, Elbers, & Sterne, 2021) guidelines were used to examine study quality. Since the review was conducted on registered trials rather than publications presenting outcomes, only minimal variables were available. Of the five domains in the Cochrane guidelines, domain 1 (randomization process) was partially reviewed for randomization and allocation concealment; domain 2 (deviations from the intended intervention) was partially reviewed for participant and care provider masking only; domain 3 (missing outcome data) was not reviewed, since no outcome data are available; domain 4 (outcome measurement process) was fully reviewed. Domain 5 (reported results) was not reviewed, since no results have been reported. A partial risk-of-bias determination was made, based on the retained domains. The risk-of-bias assessment was conducted collaboratively, with the study lead and a study staff member completing 11/28 (39.3%) of the initial studies together, and the staff member completing the remaining alone, bringing any questions forward for discussion. Results of the partial risk-of-bias assessment are narratively reported.

Results
Of a total of 902 identified records, 42 were eligible for inclusion; see the PRISMA diagram in Fig. 1  . Associated with the selected trials, one published protocol was found (Gao et al., 2021); publications presenting outcome data were not available for any of the trials. General trial characteristics are described in Table 1 and detailed trial information is provided in Table 2. The 42 records report on trials of psychological interventions, cognitive or neurorehabilitation interventions, pharmacological and other medical interventions; herbal, nutritional, or natural supplements; and physiotherapy or physical rehabilitation interventions.
Intervention duration ranges from 2 weeks to 5-6 months in the four of five studies reporting a duration. Four of five trials are randomized, three with placebo control groups and one with a waitlist control group. The medicinal cannabis study is a single-group pilot trial (Iveson et al., 2021). Primary and secondary outcomes include depression and anxiety, cognitive function, quality of life, fatigue, long COVID symptoms in general, and a variety of physical health metrics. None of the studies require that individuals have impairments to mental health or wellbeing to participate, but one study requires cognitive deficits (Guzman-Velez et al., 2021). Four trials list severe, chronic, or pre-existing mental illness as an exclusion criterion (Blane et   Jacobs, 2021), and three exclude individuals with substance use disorders (Guzman-Velez et al., 2021;Iveson et al., 2021;Rice & Jacobs, 2021). None list any cognitive impairment factors as exclusion criteria.

Limited risk-of-bias assessment
Overall risk-of-bias assessment is reported in Table 2.

Randomization
Most of the registered trials report randomization (35, 83.3%), a low risk-of-bias indicator. However, only four records confirm that they will implement allocation concealment (9.5%); information about the randomization process is missing for all other records.

Masking
Fifteen studies (35.7%) report that participants and/or treatment providers are masked, lowering the risk of bias.

Outcome measurement
Nineteen trials (45.2%) achieved a low risk-of-bias rating, with appropriate masked measurement processes that would not be expected to be influenced by bias. Four trials (9.5%) are associated with some concern, and 19 trials (45.2%) have a high risk-of-bias rating, generally due to open label designs and possible interviewer or self-report biases in outcome assessments.

Discussion
Given the rapid emergence and global spread of COVID-19, it has taken time to move from identifying long COVID to testing treatments for it. A small international body of research is assessing interventions for mental health, cognition, and psychological wellbeing in long COVID. Several psychological interventions are being tested, but few full-scale psychotherapeutic interventions are being trialed to date. Only a few interventions with pharmacological and other medical treatments were found, complemented by a similar number of herbal, nutritional, or natural supplement interventions. Several physical and cognitive rehabilitation interventions are also being examined. Randomized-controlled trials and randomized-controlled feasibility trials dominate the trial landscape.
We laud researchers who have quickly registered trials and begun testing interventions for this new clinical entity. At the same time, we highlight that the number, size, and quality of trials and the breadth of interventions are limited. Given the potential long-term disability associated with long COVID (Alwan, 2021;Brown & O'Brien, 2021), we call on funders to support research in this area at a level commensurate with symptomatic burden. We also call on interventionists to rapidly pursue large-scale, rigorous, high-quality clinical trials on interventions that address the full range of long COVID symptoms, including mental health, cognition, and psychological wellbeing (Crook et al., 2021;Malik et al., 2022;Maury et al., 2020;Sudre et al., 2021). Adaptive trials may be the most promising design approach to address the mental health symptoms of long COVID, in the context of an evolving pandemic and emerging knowledge base ( These are feasibility or pilot trials not intended to produce complete unbiased outcomes.

Psychological Medicine
Ioannidis, 2021). Given the urgent need to build a new evidence base, juxtaposed with typically high rates of non-publication of clinical trials (Lee, Bacchetti, & Sim, 2008), researchers are encouraged to publish their findingspositive or negativeat the earliest possible date (Mlinaric, Horvat, & Supak Smolcic, 2017). Likewise, publishers and peer reviewers are encouraged to welcome both positive and negative findings to accelerate the construction of a balanced and comprehensive evidence base in this new domain. NICE guidelines recommend integrated, interdisciplinary treatments for long COVID (International Foundation of Integrated Care, 2020), but the current trials demonstrate limited service integration. Integrated, interdisciplinary models of care that directly address a broad range of symptoms are needed, and they should be rigorously evaluated using methodologies appropriate for complex interventions (Skivington et al., 2021). Many of the registered trials are broadly scanning for outcomes in an integrated manner, across biological and psychological spheres, which is also important to continue. The ongoing use of virtual service features will provide important advancements for the evidence base on virtual healthcare interventions (Torous, Jän Myrick, Rauseo-Ricupero, & Firth, 2020).
While many of trials currently registered are addressing mental health in some way, comparatively few trials focus explicitly on mental health, v. cognition and psychological wellbeing. Importantly, interventions targeting individuals with severe mental illness or pre-existing mental illness are absent, and a number of trials explicitly identify mental illness as an exclusion criterion. Not only can long COVID be associated with the emergence of new mental health challenges (Aiyegbusi et al., 2021), but some long COVID patients will have pre-existing mental illness, which is a risk factor for long COVID (Gebhard et al., 2021) and may affect the experience of long COVID. Similarly, substance misuse is a very common comorbidity among people with mental illness (Lai, Cleary, Sitharthan, & Hunt, 2015), yet none of the registered trials mentioned substance misuse, except as an exclusion criterion. We therefore call on interventionists to develop and evaluate interventions that integrate evidencebased treatments for mental illness and substance misuse with treatments for the physiological symptoms of long COVID, while also addressing the potential interaction between mental and physical health.
Given the novelty of this clinical entity, it is unsurprising that trials are recruiting from the general population of patients with long COVID. A next, critical step is to test interventions adapted to vulnerable subpopulations. With a focus on equity, diversity, and inclusion, interventions should attend to individuals with different sociodemographic characteristics, including youth and seniors, and subgroups of people who are facing challenges with various social determinants of health, physical and mental health comorbidities, limited access to digital technologies, and other treatment access barriers. While doing so, attending to generalizability within interventions and trial designs may provide gains for other disorders with overlapping symptomatology (Wong & Weitzer, 2021). Researchers are encouraged to reflect on additional knowledge gaps and opportunities, from their unique disciplinary perspectives, and to move forward with addressing them in a timely manner.
We further call on the research community to engage patients in the research and service design process to address long COVID, from a pragmatic, patient-oriented research perspective (Allemang, Sitter, & Dimitropoulos, 2022; Canadian Institutes of Health Research, 2019). Only two of the registered trials refer to patient-engaged research processes (Busse & Potter, 2022;Martin & Lynall, 2022). However, patients first identified long COVID as a clinical entity (Callard & Perego, 2021), demonstrating their important insights into their lived experience and their ability to advocate for themselves to drive change. Through co-creation, patients can make meaningful contributions to research and service design (Canadian Institutes of Health Research, 2019; Hamilton et al., 2018).
This review has a number of limitations. Notably, the pace of COVID-19 research is extremely rapid (Liu et al., 2021b). This review is limited to trials registered by 31 May 2022; any trials registered after this date, or not registered, are not included. Given the limited amount of information available in trial registries, only a partial quality appraisal was possible. Due to the lack of trial results to date, a meta-analytical report was not possible. Researchers are encouraged to register their trials, consult the trial registries for studies aligning with their area of work, and report their results rapidly to members of the scientific and clinical care communities, many of whom are eagerly awaiting their findings.

Conclusions
An emerging body of research has begun to test interventions for mental health, cognition, and psychological wellbeing in long COVID. However, this review highlights that the scope of the associated intervention research currently in progress is not yet commensurate with the scope of this important new clinical entity. Despite a great deal of uncertainty around the evolution of long COVID, it is incumbent on researchers to build upon the trials currently under way and to rapidly generate rigorous evidence in this entirely new domain. We therefore call on researchers around the world to develop high-quality clinical trials testing a wide range of candidate interventions addressing mental health, cognition, and psychological wellbeing in diverse patient populations experiencing the symptoms of long COVID.
Financial support. This work is supported by Canadian Institutes of Health Research (Funding reference number WI1-179893).