Treatable traits for long COVID

Long COVID, or post‐acute COVID‐19 sequelae, is experienced by an estimated one in eight adults following acute COVID‐19. Long COVID is a new and complex chronic health condition that typically includes multiple symptoms that cross organ systems and fluctuate over time; a one‐size‐fits‐all approach is, therefore, not likely to be appropriate nor relevant for long COVID treatment. ‘Treatable Traits’ is a personalized medicine approach, purpose‐built to address the complexity and heterogeneity of complex chronic conditions. This comprehensive review aimed to understand how a treatable traits approach could be applied to long COVID, by first identifying the most prevalent long COVID treatable traits and then the available evidence for strategies to target these traits. An umbrella review of 22 systematic reviews identified 34 symptoms and complications common with long COVID, grouped into eight long COVID treatable trait clusters: neurological, chest, psychological, pain, fatigue, sleep impairment, functional impairment and other. A systematic review of randomized control trials identified 18 studies that explored different intervention approaches for long COVID prevention (k = 4) or management (k = 14). While a single study reported metformin as effective for long COVID prevention, the findings need to be replicated and consensus is required around how to define long COVID as a clinical trial endpoint. For long COVID management, current evidence supports exercise training or respiratory muscle training for long COVID treatable traits in the chest and functional limitation clusters. While there are studies exploring interventions targeting other long COVID treatable traits, further high‐quality RCTs are needed, particularly targeting treatable traits in the clusters of fatigue, psychological, pain and sleep impairment.


INTRODUCTION
Long COVID is the (now) widely recognized term used by the general public and researchers to describe the collection of chronic symptoms experienced by some people following acute infection with the SARS-CoV-2 virus. 1,2The virus, which causes acute COVID-19 disease, spread across the globe as a pandemic and resulted in high rates of hospitalization and death shortly after its discovery in 2019. 3Many people experienced incomplete recovery and persistent symptoms in the months that followed acute infection. 46][7] Concerningly, post-acute sequelae were also reported among people who did not experience severe initial disease and were managed in the community. 5,8This was unexpected as previous experience with coronaviruses (e.g., Middle Eastern Respiratory Syndrome [MERS] and SARS-CoV-1) causing severe acute respiratory distress syndrome (ARDS) had not identified chronic symptoms among non-hospitalized patients. 9However, the total number of people infected with previous coronaviruses was relatively small, thus chronic symptoms and complications among non-hospitalized cases were not studied.The situation with SARS-CoV-2 could not be more different.There are now more than 700 million global cases of COVID-19 and almost 7 million deaths. 10everal terms and definitions have been used to describe the clinical syndrome of persistent symptoms following SARS-CoV-2 infection. 11,12The World Health Organization (WHO) led a Delphi process in October 2021 (published April 2022) to create a consensus clinical case definition for research and healthcare: 12 'post COVID-19 condition occurs in individuals with a history of probable or confirmed SARS-CoV-2 infection, usually three months from the onset of COVID-19 and with symptoms that last for at least two months and cannot be explained by an alternative diagnosis'. 12The definition of 'post-COVID-19 syndrome' from the United Kingdom National Institute for Health and Care Excellence (NICE) is similar to that by the WHO but adds 'signs' to symptoms and states their duration should be at least 12 weeks. 13The Australian National Clinical Evidence Taskforce has adapted the WHO and NICE definitions, including a minimum duration of 12 weeks but noting that symptoms may change or fluctuate over time and the diagnosis of post-COVID-19 condition or syndrome can be considered prior to 12 weeks whilst alternate diagnoses are being investigated. 14In North America, the National Institute of Health (NIH) utilizes an umbrella term of 'post COVID conditions' that includes the variety of health conditions that occur beyond the first 28 days (the acute illness), their definition, therefore, states that symptoms should be present for at least 4 weeks. 15Another widely used term in research is post-acute sequelae of COVID-19 or SARS-CoV-2 infection (PASC), 11,16 which is typically defined by sequelae that persist for ≥28-30 days following initial onset; however, the definition is inconsistent across research studies.For this review, as it was developed by people with lived experience, 1,2 and is widely recognized across diverse audiences including the general public, we have used the term long COVID to collectively refer to definitions by the WHO, NICE and Australian Clinical Evidence Taskforce with the common time point of at least 12 weeks from acute COVID-19 onset.
The number of people living with long COVID is still debated.3][24] This has been done by comparing the presence of symptoms in individuals prior to and following COVID-19 or in COVID-19 positive and negative cohorts matched for demographic and clinical factors. 16,22Following this latter approach, current estimates suggest around 7%-13% of people who have had COVID-19 will experience long COVID; 5,23,25 incidence of long COVID may be higher, at around 20%-30%, in those who were hospitalized during the acute phase. 7,16,26][29][30] The clinical syndrome of long COVID is complex and typically includes multiple symptoms that cross organ systems and fluctuate over time. 7,23,26,27,30Emerging evidence also suggests there may be distinct long COVID symptom clusters or phenotypes, with the presence and severity of long COVID symptoms manifesting differently in different people and contexts. 7,16,25,27,30hilst over 200 symptoms have been described, there are some symptoms that occur more frequently than others and appear to cause the most morbidity and disability. 7,11,23,25,26,30Consistently reported common symptoms include fatigue, breathlessness, cognitive dysfunction (e.g., 'brain fog'), musculoskeletal pain, impaired sleep, chest pain and palpitations. 7,23,26,27,30Other symptoms that have been reported include, but are not limited to, anosmia, dysgeusia, memory loss, sexual dysfunction and 'COVID toes' (chilblains or chilblain-like lesions) and other skin discolorations or rashes. 7,23,26,27,30he major challenge facing healthcare providers and patients is how to treat long COVID-an emergent 'public health disaster'. 31Long COVID is complex and a 'one size fits all' approach is not likely to be appropriate nor effective.This review aimed to highlight how a precision medicine 'Treatable Traits' (TTs) approach could be applied to long COVID.Accordingly, this review first aimed to identify the most prevalent long COVID symptoms and complications ≥12 weeks following acute COVID-19 onset (i.e., TTs) and summarize the current evidence around long COVID treatment approaches.

What is the TTs approach?
Chronic health conditions, like long COVID, can be difficult to manage because they are complex, meaning they have many different components or pathogenic pathways that can contribute to the illness experience.Furthermore, not all pathways operate in all people at the same time, meaning there is heterogeneity in the manifestations of chronic health conditions.Personalized and precision medicine are new treatment approaches that can address the complexity and heterogeneity of chronic disease and could be useful for long COVID.
These issues were identified in asthma, especially severe asthma and asthma in the elderly, and a personalized management approach was designed to address these issues.The strategy is based on the identification and treatment of so-called TTs, 32 and many papers related to TT have now been published, including two systematic reviews 33,34 and two proofof-concept clinical trials: one in severe asthma 35 and another in chronic obstructive pulmonary disease (COPD). 36Collectively, this body of evidence strongly supports the TT strategy for the management of people living with chronic airway diseases.

What is a TT?
A TT is a recognizable phenotypic or endotypic characteristic that can be assessed and successfully targeted by therapy to improve a clinical outcome in a person with a chronic disease. 32Importantly, multiple TTs can co-exist in the same individual, so they are not mutually exclusive.TTs must share three characteristics: 37 (1) clinical relevance, which means that they are linked to a relevant clinical outcome such as symptoms, quality of life or objective lung function impairment, such as lung function decline; (2) detectable either 'phenotypically' (by clinical assessment) and/or based on deep understanding of critical causal pathways ('endotypes') through validated 'trait identification markers' (TIM) (e.g., circulating eosinophils, see below); and, (3) treatable, meaning that effective treatment is available and accessible, and when implemented, leads to a clinically important outcome.Admittedly, there are currently untreatable traits, but they may also be important to consider in the clinical management of the patient and they clearly constitute targets for research.A well-recognized example of a TT in asthma is eosinophilic airway inflammation. 38This is an endotype of airway disease mediated by specific cytokines such as interleukin (IL)-5. 38Increased levels of eosinophilic inflammation are associated with increased exacerbation risk in both asthma and COPD, indicating it is clinically relevant. 38his trait can be detected via blood eosinophil count (a TIM) and can be treated via inhaled corticosteroids, oral corticosteroids or T2-directed monoclonal antibody (mAbs) therapy.The endotype of eosinophilic airway inflammation meets each of the three criteria named above for designation as a TT because it is relevant, detectable and treatable.

How is the TTs approach applied in clinical practice?
The TTs strategy proposes to investigate, in each individual person with chronic disease, the number and type of TTs present, and to treat each of them according to guideline recommendations. 32This strategy is agnostic (i.e., independent) to the traditional diagnostic labels (asthma, COPD and bronchiectasis), so it can be applied to any person with airway disease. 32In clinical practice, these traditional diagnostic labels may constitute a good starting point of the diagnostic process, not its end.That is, after having established a first diagnostic approximation (label), it is imperative to deconstruct the clinical and biologic complexity of each patient using a TTs strategy.

TTs in long COVID
To explore how the TT approach could be applied to long COVID, requires first identifying potential TTs, and subsequently, evidence-based treatment approaches for those specific TT.To do this, we conducted two separate systematic reviews.Review 1 was an umbrella review, 39 that aimed to identify the most prevalent long COVID symptoms or complications (i.e., TT).We chose to perform an umbrella review (or, review of reviews) rather than a traditional systematic review due to the anticipated high number of studies reporting on prevalence of COVID-19 sequelae in the post-acute phase.Review 2 aimed to summarize the available evidence on pharmacological and non-pharmacological therapies for long COVID prevention or management.The outcomes of these two reviews were then combined to present a TTs strategy for long COVID.

METHODS
The systematic reviews were reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). 40

Eligibility
For both reviews, the time period considered for long COVID was ≥12 weeks from acute COVID-19 onset, consistent with definitions by the WHO, NICE and Australian Clinical Evidence Taskforce.Studies were eligible for inclusion in Review 1 if they were a systematic review with metaanalysis of the prevalence of ongoing sequelae at least 12 weeks following COVID-19 onset and published in English.All other study designs were excluded, including systematic reviews with qualitative synthesis only, narrative reviews, primary studies (observational or experimental), editorials or opinion pieces, conference abstracts, protocols, and so on.
Studies were eligible for inclusion in Review 2 if they were a randomized control trial (RCT) exploring the effects of pharmacological and/or non-pharmacological intervention for the prevention or management of long COVID.Regarding long COVID prevention, studies were eligible if: (1) participants had confirmed acute COVID-19; (2) they reported on the effects of intervention administered during the acute COVID-19 phase compared with any other comparator type; and (3) outcomes pertained to incidence of long COVID at ≥12 weeks since acute COVID-19 onset, which could include presence of persistent symptoms or complications and/or self-reported or medical diagnosis of long COVID.For long COVID management, studies were eligible for inclusion if they: (1) included people diagnosed with, or suspected of having, long COVID-19 due to ongoing sequelae ≥12 weeks following acute COVID-19 onset; (2) reported on the effects of intervention on COVID-19 sequelae compared with any other comparator type; and (3) had outcomes related to COVID-19 sequelae that could include, but were not limited to, objective measures of physiology or function and subjective measures of symptoms or health-related quality of life (HRQoL).Studies were excluded from Review 2 if criteria relating to long COVID were not defined or long COVID was defined as <12 weeks; were of observational or non-randomized experimental design, not a primary study (review, clinical practice guideline or consensus statement, editorial or opinion piece, conference abstract, protocol, etc.) or not published in English.

Information sources and search strategy
Electronic databases Ovid MEDLINE, Ovid Embase, CINAHL, Scopus and Cochrane (Review 1 only) were searched for eligible studies by a single reviewer between the 13th and 16th of January (Review 1) and on the 17th of April 2023 (Review 2).The search strategy was developed in consultation with an academic librarian and included a search string of key words pertaining to 'long COVID' combined with a MeSH term for systematic review for Review 1, or a search string of MeSH and keywords pertaining to 'disease management' for Review 2. Limits were applied to the search for Review 2 for study type (RCT only) and publication date (since 2019).The search strategies initially developed for use in Ovid MEDLINE are presented in Appendix S1 of the Supporting Information; MeSH terms were adapted for use in other databases.

Selection process
References identified by the search process were imported into Endnote, where duplicates were removed.Remaining references were imported into Covidence for screening against eligibility criteria by two independent researchers.Titles and abstracts were first screened, followed by remaining full texts; discrepancies on eligibility were resolved by discussion and consensus.A third reviewer was consulted when required to resolve discrepancies.

Data collection process and data items
For both reviews, a data extraction template was developed a priori, and pilot-tested by two independent researchers on two randomly selected studies.Following refinement of the template, data were extracted from all included studies.Data items extracted are provided in Appendix S2 of the Supporting Information.

Data synthesis methods
For Review 1, the overlap of primary studies included in the eligible systematic reviews was explored using a citation matrix and quantified using the corrected covered area (CCA) measure proposed by Pieper et al. 41 The CCA informed whether all or only non-overlapping systematic reviews would be included in the qualitative synthesis. 42haracteristics of included systematic reviews were tabulated and summarized descriptively.For each systematic review, the top 10 most prevalent symptoms or complications (name verbatim, pooled prevalence, 95% CI, k = number of studies included in meta-analyses) were extracted and tabulated from most to least prevalent.Symptoms/complications were extracted from the systematic reviews and included in the synthesis if prevalence data were available from meta-analyses of at least two primary studies.The frequency that each unique symptom/complication appeared in the top 10 most prevalent across all included systematic reviews were calculated and reported.Symptoms and complications were categorized into common long COVID TT clusters based on expert opinion and consensus of the authorship team.
For Review 2, study characteristics were tabulated and summarized descriptively, according to the primary outcome and by intervention type.Details were reported around the primary outcome and relevant secondary outcomes, outcome measures, post-intervention between group differences, and if and what adverse events were reported.

Review 1-Long COVID symptoms and complications (TTs)
Of the 792 records identified from the search process (Figure S1 in the Supporting Information), 22 systematic reviews with meta-analyses (1 of 22 = preprint) met the inclusion criteria and were included in this umbrella review, published in 2021 (k = 6), 2022 (k = 14) and 2023 (k = 2).The primary reasons for exclusion were long COVID not ≥12 weeks or no data were reported on symptom/ complication prevalence (Figure S1 in the Supporting Information).There were 583 primary studies or preprints (23 of 583 = preprints) included in the 22 systematic reviews.The CCA was 5.7%, indicating slight to moderate overlap of primary studies across systematic reviews.Symptom/complication prevalence data from all included systematic reviews was, therefore, included in the qualitative synthesis.

Review 2-Long COVID prevention or management
Electronic database searching identified 364 unique records, of which, 308 were excluded based on title and abstract screening (Figure S2 in the Supporting Information).Of the 56 full texts screened for eligibility, 38 were excluded, with the primary reason being that no clear definition was provided for what constituted long COVID, or long COVID was defined as <12 weeks (k = 22).Therefore, 18 RCTs were included in this systematic review (2 = preprints), published in 2021 (k = 1), 2022 (k = 14) or 2023 (k = 3).There were two pairs of RCTs reporting duplicate data: Jimeno-Almazan et al. 64 and Jimeno-Almazan et al. 65 both reported data from the RECOVE trial, with the 2023 study having enrolled a further 44 participants; and Catalogna et al. 66 and Zilberman-Itskovich et al. 67 both reported outcomes from hyperbaric oxygen therapy (HBOT) in 79 long COVID participants.

Interventions for the prevention of long COVID
Four of the 18 (22%) RCTs [68][69][70][71] explored the effect of different intervention approaches administered during the acute COVID-19 phase (while hospitalized) for the prevention of long COVID (Table S3 in the Supporting Information).Körper et al. 70 explored the effects of convalescent plasma therapy compared with standard care; Barros et al. 68 compared intravenous methylprednisolone sodium succinate with placebo; Bramante et al. 69 compared metformin, ivermectin or fluvoxamine with placebo; and Nevalainen et al. 71 compared remdesivir with standard of care.
For these studies, the outcome of long COVID was assessed by symptom or HRQoL questionnaires, 68,70,71 objective assessments of function, 68 self-reported symptoms (presence and severity), 70 self-reported rating of perceived recovery from COVID-19 (numeric rating scale), 71 and/or self-reported diagnosis of long COVID from a medical provider. 69ramante et al. 69 (preprint) was the single study to report significant benefits for long COVID prevention.Participants (n = 564) who received metformin for 14 days (day

Interventions targeting neurological TTs
Four RCTs included neurological symptoms as their primary outcome, including: cognition 66,67 and olfactory dysfunction. 74,81Catalogna et al. 66 and Zilberman-Itskovich et al. 67 reported results from the same intervention; hyperbaric oxygen therapy (HBOT) compared with sham treatment for cognition assessed via the NeuroTrax MindStreams Computerized Battery. 82Two months of HBOT (40 sessions, 5/week within a 2-month period) significantly improved global cognitive score, attention and executive function compared with a sham protocol. 66,67A B L E 3 Frequency that long COVID symptoms and complications were in the top 10 most prevalent across systematic reviews, grouped into eight long COVID treatable trait clusters.
Long COVID treatable trait n = (%) Targeting olfactory dysfunction, Yan et al. 81 explored the effects of three intranasal platelet-rich plasma (PRP) injections compared with sterile saline injections, while De Luca et al. 74 explored the effects of 90-days of olfactory training with palmitoylethanolamide and luteolin (PEA-LUT) supplementation versus PEA-LUT alone.The PRP injections, each administered 2 weeks apart, significantly improved olfaction (Sniffin' Sticks TDI score 83 ) at 3-months (but not 1-month) post-treatment. 81Olfactory training (3 times/day for 90 days) with PEA-LUT supplementation did not result in a change in olfaction (Sniffin' Sticks TDI score 83 ) compared with PEA-LUT alone. 74our RCTs included neurological symptoms as secondary outcomes, however, no significant benefits were reported for cognitive outcomes 73,74 or smell/taste disturbance. 66,67

Interventions targeting chest TTs
For two RCTs, chest symptoms/complications were the primary outcome: pulmonary function 72 and dyspnoea. 78Dal Negro et al. 72 conducted a double-blind cross-over RCT in eight participants, exploring the effects of a beta-blocker (nebivolol, 2.5 mg per day for 1 week) compared with placebo.The results reporting was limited to pre-postnebivolol or placebo; no between condition comparisons were reported.
Romanet et al. 78 (preprint) explored the effects of endurance exercise training following international pulmonary rehabilitation guidelines 84 (intervention group) compared with standard physiotherapy care for dyspnoea.Both groups attended two sessions per week for 10 weeks; no specific details were provided around what constituted standard physiotherapy care.At 3-months, there were significant differences between groups for dyspnoea (Multidimensional Dypnoea Profile [MDP] 85 ) favouring the intervention group (MDP A1 score: mean difference [MD] = À1.74 [95% CI À2.81, À0.67], p = 0.0006).
Del Coral et al. 73 and McNarry et al. 76 both explored effects of respiratory muscle training on respiratory muscle strength and dyspnoea as secondary outcomes, reporting significant benefits compared with control conditions. 73,76

Interventions targeting psychological TTs
No study included psychological symptoms as a primary outcome.]75,76,79 Significant intervention benefits for depression, but not anxiety, symptoms were shown following HBOT 66,67 and exercise training, 64 compared with a sham condition and exercise advice, respectively.A study exploring high-definition transcranial direct current stimulation (HD-tDCS) reported significant benefits for anxiety symptoms compared with a sham. 79A B L E 4

Interventions targeting pain TTs
No study included pain as a primary outcome.Three RCTs 66,67,79 included pain as a secondary outcome, of which, one study reported significant benefits for pain assessed via the Brief Pain Inventory following HBOT. 64

Interventions targeting fatigue TTs
Two RCTs explored fatigue as a primary intervention outcome, Hawkins et al. 75 and Santana et al. 79 Hawkins et al. 75 tested the effects of aromatherapy, involving inhalation of essential oils for 15-min twice per day for 2 weeks, compared with inhalation of inert, odourless oils; all participants (n = 40) were women.Positive treatment effects were achieved in the intervention group compared with the control group for fatigue assessed with the Multidimensional Fatigue Symptom Inventory-Short Form. 86Santana et al. 79 explored the application of HD-tDCS or a sham (10 sessions over 5 weeks) during low-intensity exercise based on PASCrelated fatigue guidelines; 87 both groups also received education.Compared with the sham treatment, participants in the intervention group achieved significantly greater reduction in fatigue assessed with the Modified Fatigue Impact Scale; 86 significant benefits were achieved for the cognitive and psychosocial fatigue subscales but not the physical subscale. 79hree RCTs reported on the effects of fatigue as a secondary outcome, of which, two reported significant interventions benefits following L-arginine and liposomal vitamin C supplementation (proportion of people reporting fatigue most/all the time) 80 and exercise training (Chalder Fatigue Scale, Fatigue Severity Scale). 64

Interventions targeting sleep TTs
No RCT explored sleep as a primary outcome.A single study included sleep quality (Pittsburgh Sleep Quality Index) as a secondary outcome, reporting significant benefits following HBOT compared with a sham condition. 67

Interventions targeting functional impairment TTs
Six RCTs tested intervention effects for measures of functional impairment as the primary outcome, including exercise capacity 64,65,77,80 and HRQoL. 73,76Of the four studies that included exercise capacity as a primary outcome, three were exercise-based interventions 64,65,77 and one explored benefits of L-arginine and vitamin C supplementation. 80alau et al. 77 and Tosato et al. 80 both reported significant intervention effects for exercise capacity.For Palau et al., 77 participants who were randomized to a 12-week homebased inspiratory muscle training (IMT) programme had significantly better maximum rate of peak oxygen consumption (V 0 O 2peak ) post-intervention compared with the usual care group.Tosato et al. 80 reported significant benefits for six-minute walk test (6MWT) post-intervention for participants who received twice-daily oral supplementation with L-arginine (1.66 g) and liposomal vitamin C (500 mg) for 28 days, compared with placebo.While Jimeno-Almazan et al. 64 (n = 39) reported significantly greater improvements in V 0 O 2peak for participants who received an 8 weeks exercise training programme (3 days/week) compared with exercise advice, Jimeno-Almazan et al. 65 (n = 80) found no significant between group differences in V 0 O 2peak when comparing the exercise training program combined with IMT versus exercise or IMT alone versus exercise advice.
McNarry et al. 76 and Del Corral et al. 73 both explored different forms of respiratory muscle training for HRQoL.McNarry et al. 76 compared IMT (three times per week for 8 weeks) to usual care, reporting significant between-group differences in HRQoL (per protocol but not intention to treat analyses) favouring the intervention group.Del Corral et al. 73 compared 8 weeks of IMT or respiratory (inspiratory and expiratory) muscle training (40 min/day, 6 days/week) with a sham intervention; only the respiratory muscle training (not IMT) significantly improved HRQoL compared with the sham intervention.
Eight RCTs included measures of functional impairment as secondary outcomes.Three assessed muscular strength: compared with placebo, L-arginine and liposomal vitamin C supplementation was associated with significantly greater benefits for hand grip strength; 80 respiratory muscle training significantly improved 1-min sit-to-stand compared with sham; 73 and exercise training achieved significantly greater benefits for five times sit-to-stand, and 50% 1-repetition maximum bench press and half squat compared with exercise advice. 64Five studies included HRQoL as a secondary outcome, with all studies reporting significant intervention effects in global 79 or sub-domain 64,67,77,78 scores.

Adverse events
Ten (71%) of fourteen studies reported on adverse events (Table 4).Four studies reported no adverse events. 64,65,77,80There were 13 adverse events recorded during HBOT (and 14 during control condition), including barotrauma, palpitations, pain, fever, hypertension and hospitalization; no participant receiving HBOT was required to discontinue treatment due to adverse events. 66,67Intranasal PRP injections were not associated with long-term adverse effects; short-term effects were reported relating to the injection site. 81One participant receiving aromatherapy 75 reported experiencing headaches and withdrew from the study on day 13 (of 14).For Santana et al. 79 more participants in the HD-tDCS intervention group reported skin redness compared with the sham protocol; no serious adverse events were reported.One participant in the sham IMT group of del Corral et al. 73 withdrew due to symptom exacerbation.

DISCUSSION
This review aimed to explore how a TTs approach could be applied to long COVID.We conducted an umbrella review to identify long COVID TTs, and a systematic review of RCTs to summarize the available evidence on intervention approaches targeting potential TTs.The main findings were: (1) long COVID is complex, with symptoms or complications in eight different TT clusters common; (2) no metaanalyses compared prevalence of persistent long COVID symptoms in relevant sub-populations, including people with versus without previous acute COVID-19 or vaccinated versus unvaccinated for SARS-CoV-2; (3) limited research is available for strategies to prevent long COVID; a single study reported benefits for metformin, however, the findings need to be replicated and the consensus is required around how to define long COVID as a clinical trial end-point; (4) there is evidence to support exercise training or respiratory muscle training for traits within chest and functional limitation long COVID TT clusters and (5) while there are studies exploring interventions targeting other long COVID TT clusters, further high-quality RCTs are needed, particularly targeting fatigue, psychological, pain and sleep TTs.
Consistent with previous research, this review highlights the complexity of long COVID, associated with at least eight long COVID TT clusters.We propose a TT approach for the management of long COVID (Figure 2) as it is purposebuilt to address such complexity.Rather than treatment being informed by the underlying diagnosis, a criteria for which is still being defined for long COVID, 14,16 a comprehensive multidimensional assessment is conducted to identify TT, and then an individualized treatment plan is developed and implemented based on (1) this assessment and (2) the patient's perspective of the most important TTs.The eight most prevalent long COVID TT clusters identified in our umbrella review (neurological, chest, psychological, pain, fatigue, sleep impairment, functional impairment and other) are consistent with a recently developed core outcome set (COS) for long COVID, which includes 11 items identified by healthcare practitioners, researchers, and people living with long COVID and their carers. 88This COS represents a minimum set of outcomes that should be considered in long COVID clinical and research practice, as they are what matters most to key stakeholders.Importantly, however, the COS and our identified list of long COVID TTs do not represent an exhaustive list of potential therapeutic targets for every individual.Rather, these highly prevalent TTs represent a starting point for applying the TTs approach to long COVID.To facilitate this, we have proposed (Figure 2) traitidentification markers (TIMs) for some of the long COVID TTs.Further research is needed to validate these TIMs in people living with long COVID and to establish populationspecific minimal clinically important differences.
In our umbrella review, no meta-analyses compared the prevalence of long COVID symptoms or complications in people with versus without a history of SARS-CoV-2 exposure.The prevalence of different symptoms, therefore, cannot be fully attributed to acute COVID-19 and may be related to the presence of other underlying health conditions.Where reported, the mean age of participants included in systematic reviews ranged from 47 ± 18 years 62 to 56 ± 6 years, 59 an age associated with increased odds of living with two or more chronic conditions. 89,90A population-based cohort study from the Netherlands, which compared prevalence of symptoms of at least moderate severity in people with versus without previous acute COVID-19 (≥90 days prior), identified smell/taste disturbances, difficulty breathing, chest pain, pain with breathing, lump in throat, heavy arms/legs, general tiredness, painful muscles, tingling extremities and feeling hot and cold alternatively, as being significantly more prevalent in the COVID-19 cohort compared with controls.Participants in this study were exposed to SARS-CoV-2 prior to April 2021; 9.8% of the cohort were considered fully vaccinated, and alpha was the dominant SARS-CoV-2 variant.The National Institute of Health's RECOVER Initiative (USA) similarly compared symptoms of at least moderate severity in people previously exposed (on or after December of 2021) versus unexposed to SARS-CoV-2. 16Twelve symptoms were attributed to long COVID: post-exertional malaise, fatigue, brain fog, dizziness, GI symptoms, palpitations, changes in sexual desire or capacity, loss of or change in smell/taste, thirst, chronic cough, chest pain and abnormal movements. 16Over half (58%) of the cohort were fully vaccinated. 16While there are differences between the findings of our umbrella review and these (and other) cohort studies for the most prevalent long COVID symptoms or complications, likely at least in part due to use of a control cohort and differences in vaccination status and/or SARS-CoV-2 variant, 21 the TTs approach is designed to account for such disease heterogeneity.Further research is, however, needed to identify the TTs of greatest patient and clinical relevance (i.e., associated with deleterious health outcomes), which will inform treatment priorities.
For the prevention of long COVID, our systematic review of RCTs (Review 2) identified four studies exploring different intervention approaches.One study (published as a preprint), exploring 14-days of metformin administered during the acute COVID-19 phase compared with placebo, reported significant intervention effects. 69Of the 564 participants randomized to receive metformin during hospitalization for acute COVID-19, 6.3% self-reported receiving a medical diagnoses of long COVID at 10-month follow-up compared with 10.6% of the 561 participants randomized to placebo. 69Metformin may help to prevent long COVID via anti-inflammatory and antithrombotic actions. 91,92Other proposed mechanisms include preventing SARS-CoV-2 infection (or progression to severe COVID-19 disease) by changing ACE2 receptor formation, which the virus binds to enter human cells. 91However, despite metformin showing early promise for long COVID prevention, currently, it is not recommended for use outside of clinical trials by the National Clinical Evidence Taskforce. 14 to >8 months after acute COVID-19 onset.No study explicitly reported matching intervention and control groups for time since acute COVID-19, and there were differences between intervention and control groups for mean time since acute COVID-19 by up to >60 days. 65Given the dynamic nature of long COVID, with the presence and severity of symptoms fluctuating over time, 23 this may lead to spurious interpretations of intervention effects.The predominant focus of RCTs for the management of long COVID, was on long COVID TTs in the chest and functional limitation clusters.Eleven of fourteen (79%) RCTs reviewed included primary or secondary outcomes relevant to these TT clusters.Of these, six RCTs explored the benefits of exercise training or respiratory muscle training, demonstrating significant benefits for respiratory 73,76 and peripheral muscle strength, 64,73 dyspnoea, 73,76,78 exercise capacity, 64,77 and HRQoL 64,73,76-78 compared with control conditions.These findings are in line with evidence from non-COVID-19 chronic respiratory disease populations.There is substantial evidence to support the benefits of supervised exercise training for dyspnoea, exercise capacity and HRQoL in people with COPD, 93 and there is emerging evidence in interstitial lung disease, 94 and asthma. 95Benefits have also been observed with IMT compared with usual care for dyspnoea, exercise capacity and HRQoL; 96 however, no significant differences in these outcomes have been shown when IMT was compared with exercise training. 96Exercise training combined with IMT is also not superior to exercise training alone. 96IMT may, therefore, be an effective strategy for people who are unable to participate in sufficient volumes and/or intensities of exercise (e.g., due to mobility or other limitations, barriers to adherence) to achieve maximal health benefits.The single remaining study reporting significant benefits for functional impairment (exercise capacity), tested L-arginine and liposomal vitamin C vitamin supplementation, which is not recommended for use outside clinical trials. 14hile exercise training and IMT may be effective for targeting TTs within chest and functional limitation clusters, caution is required for people experiencing persistent fatigue, with graded exercise training contraindicated in this population. 878][99] The goals of rehabilitation are to return to pre-illness levels of daily activities and quality of life; exercise is to only be introduced once these goals are achieved and when possible without exacerbating fatigue. 97his may explain why alternative intervention approaches have been explored for fatigue, including inhalation of essential oils 75 and HD-tDCS applied during low-intensity exercise. 79HD-tDCS is a form of non-invasive brain stimulation, which modulates neuronal activity via electrodes placed on the scalp that emit an electronical current targeting predefine brain regions. 100,101This intervention approach has previously been combined with rehabilitative exercise in the acute COVID-19 phase, 102 and used to target fatigue in multiple sclerosis, 103 stroke 104 and Parkinson's disease. 105To better understand the benefits of HD-tDCS for people experiencing persistent fatigue associated with long COVID, further RCTs are required to replicate findings of Santana et al. 79 In fact, high-quality RCTs are urgently needed, which target all traits of high priority to people living with long COVID.These interventions should be co-designed with people with long COVID, and explicitly report adverse outcomes to inform the safety of interventions in this new disease population.

Strengths and limitations
This review was strengthened by the comprehensive approach to summarize the available evidence on the most prevalent long COVID TTs, and potential intervention approaches to target these traits.We conducted two rigorous systematic reviews including over 600 primary studies.This review was limited by not conducting a meta-analyses to quantify the pooled prevalence of long COVID symptoms and complications across included systematic reviews.However, systematic reviews were heterogenous in their methodological design and results reporting.We also did not explore differences in long COVID symptom prevalence according to acute COVID-19 disease and management characteristics.Based on available evidence, 21 it is expected that vaccination against SARS-CoV-2 reduces the incidence of long COVID.The SARS-CoV-2 strain likely also impacts long COVID incidence and presentation, however, it is difficult to discern the impact of the strain versus vaccination status; COVID-19 was predominantly caused by wild type virus, alpha, beta and delta variants prior to widespread vaccination, while the omicron wave occurred after vaccination implementation.The systematic reviews included in Review 1 were published from 2021 to 2023 with search dates from January 2021 to September 2022.It is likely that new evidence has emerged around common long COVID symptoms/ complications not captured by these reviews.For example, there is now evidence that respiratory muscle weakness is a common long COVID complication, 106 which has also been identified by this review as an intervention target.The aim of this review, however, was not to provide a definitive list of the most prevalent long COVID TTs in all contexts and populations, but to consolidate a large body of evidence to identify common long COVID symptoms and complications to provide a starting point for clinicians and researchers to explore a TTs strategy for long COVID management.

CONCLUSION
A precision medicine strategy based on the identification and treatment of TTs was proposed 6 years ago to improve the clinical management of patients with chronic airway diseases. 328][109][110] Thus, the question is no longer 'what' but 'how'.To address this and facilitate the use of a TT strategy for long COVID, we have provided a list of common long COVID TTs in eight clusters (neurological, chest, psychological, pain, fatigue, sleep impairment, functional impairment and other) identified via a comprehensive umbrella review, and summarized the available evidence for strategies to target these traits.Our findings enable a TT approach for the benefit of your patients living with long COVID.

1 =
500 mg; days 2-5 = 500 mg/d; days 6-14 = 500 mg mornings and 1000 mg evenings) in the acute COVID-19 phase, had a reduced incidence of developing long COVID at 10-months (self-report medical diagnosis) compared with participants who received placebo (n = 561) (hazard ratio [HR] 0.58 [95% CI 0.38, 0. Furthermore, for future studies exploring long COVID preventative strategies, consensus is required around how to define long COVID as a clinical trial end-point.Use of self-reported medical long COVID diagnoses as an outcome is limited by lack of knowledge or criteria (in the health and general community) around what constitutes long COVID.It is likely that the incidence of long COVID is substantially underestimated, particularly in disadvantaged populations with limited access to high-quality healthcare.For the management of long COVID, a range of different interventions were administered from 158 days ($5-6 months) F I G U R E 2 Treatable traits for long COVID: what are treatable traits, and recommended approaches for assessment and management.Reproduced with permission from the Centre of Excellence in Treatable Traits, originally developed as part of the Centre of Excellence in Treatable Traits (https:// treatabletraits.org.au).
Characteristics of systematic reviews reporting on overall long COVID symptom and complication prevalence.Characteristics of systematic reviews reporting on prevalence of specific long COVID symptoms or complications.
18,43-45,47-50,53-55 Breathlessness or T A B L E 1 Abbreviations: F, female; ICU, intensive care unit; MA, meta-analysis; NR, not reported.a Sub meta-analyses for prevalence of long COVID symptoms ≥12 weeks after follow-up for factors related to acute COVID-19 disease.T A B L E 2 Characteristics and findings of RCTs exploring interventions for the management of long COVID treatable traits.
Abbreviations: GI, gastrointestinal; PFT/CT, pulmonary function test/computerized tomography; PTSD, post-traumatic stress disorder; QoL, quality of life; TT, treatable trait.F I G U R E 1 Long COVID treatable trait cluster profile-long COVID symptom and complication types in the top 10 most prevalent across systematic reviews.T A B L E 4