Long COVID is vaguely defined, but well recognised

Long COVID (LC) may occur in adults after an infection with SARS-CoV-2, the virus that causes COVID-19 (referred to here as COVID) [1]. Estimated prevalence varies from 10–20% [2] to 50% in adults [1]. The prevalence of LC and the often profound effect on people’s quality of life highlight the need to identify effective treatments [1], but even defining LC has been difficult [2, 3].

According to the definition set by the WHO, LC usually occurs 3 months after a probable or confirmed SARS-CoV-2 infection. LC is typically accompanied by the presence of new or persistent symptoms that affect everyday functioning and cannot be explained by an alternative diagnosis, which last ≥ 2 months [2]. The US definition simply includes signs and symptoms occurring ≥ 4 weeks after a COVID infection; the UK splits LC into symptomatic COVID lasting for 4–12 weeks and post-COVID-19 (PC) syndrome beyond this timeframe [2, 3].

The signs and symptoms, as well as the patterns of relapse and remission with LC are heterogeneous, and most organ systems may be potentially affected [2, 3]. However, fatigue, shortness of breath, and cognitive dysfunction are among the most commonly reported symptoms [2], with fatigue being an important factor associated with depression [1].

Depressive symptoms are clinically significant in 30–40% of patients 1–12 months after a COVID infection [1]. While also heterogenous and variably defined, LC depression (LCD) shares features of major depressive disorder (MDD), including abnormal brain imaging, symptoms such as fatigue [1], distress, apathy, insomnia and cognitive impairment, and negative thinking styles [1, 4]. Depression is also part of the neurological manifestations of COVID (neuro-COVID or nCOVID) and may incorporate conditions as diverse as post-traumatic stress disorder and new or worsening Alzheimer’s disease [4].

This article summarises early data on the epidemiology, pathophysiology and potential treatments of LCD, as reviewed by Mazza et al [1]. Other large studies including LCD among broader LC symptoms were also referenced [5, 6]. The data presented in this article are current as of October 2022.

A large study has indicated potential risk factors for long COVID

Risk factors for LC in general were identified in a US Veterans Affairs database study including ≈ 200,000 people who were infected with COVID [6]. The risk for LC correlated with higher Charlson Comorbidity Index scores (a weighted measure of comorbidity, with higher values indicating greater severity); underlying asthma, diabetes, chronic kidney disease, chronic obstructive pulmonary disease, congestive heart failure and other cardiovascular disease were significant risk factors [adjusted odds ratios (ORs) of 1.07–1.42, with the ranges of the 95% CIs all excluding 1] [6].

More severe COVID infections may be associated with greater risk for LC [6]. The presence of symptomatic illness [OR 1.46 (95% CI 1.42–1.52) in people with 1 or 2 symptoms; higher risk in people with ≥ 3 symptoms], hospitalisation [OR 2.6 (95% CI 2.51–2.69)] and mechanical ventilation [OR 2.5 (95% CI 2.26–2.69)] were risk factors for LC [6].

Regarding depression-related risk factors, an increased risk of LC was observed in people who made 7–19 mental health visits [OR 1.05 (95% CI 1.01–1.09)] or ≥ 20 visits [OR 1.16 (95% CI 1.11–1.21)] in 2 years; but not in people who made ≤ 6 visits [OR 1.02 (95% CI 0.98–1.05)] [6]. However, the effects of comorbid depression were not directly assessed in this study. The presence of fatigue was also a risk factor for LC [OR 1.51 (95% CI 1.44–1.59)], but antidepressant use did not appear to increase the risk of LC [OR 1.02 (0.99–1.05)] [6].

A lower risk for LC was observed in people who received two doses of vaccines versus those receiving no vaccination [OR 0.78 (95% CI 0.68–0.90)], in contrast to people who received one dose of vaccine [OR 1.03 (95% CI 0.95–1.12)] [6].

Long COVID depression may be common...

The prevalence of clinically-relevant depressive symptoms in people who were infected with COVID varies from approximately 20–50% [1, 3]. The use of different screening instruments may have influenced estimates in meta-analyses [1]; the most commonly used tool was the Patient-Health-Questionnaire-9, which identified depressive symptoms in about half of all patients (47–52%), versus ≈ 20–22% if the Hospital Anxiety and Depression scale or Symptom Checklist-90 were used [7]. In two analyses, mild, moderate and severe LCD occurred in ≈ 30%, ≈ 15% and ≈ 8% of all people who were infected with COVID [1].

It may be difficult to distinguish LCD from depression caused by broader COVID-related stress and social isolation [8] and a full discussion of those issues is outside the scope of this article. However, based on a cohort in one study published in early 2021 [9], LCD represented an increase in the incidence of depression [1]. Approximately 42% of COVID-infected patients, versus 32% of the general population and 31% of health workers developed depressive symptoms [9].

...especially in lonely people after severe COVID

A history of severe COVID infection consistently increased the risk for LCD in published studies, with sociodemographic risk factors also evident [1, 6, 10]. In meta-analyses, women and patients with severe COVID were more prone to LCD [1]. Other studies highlighted COVID-related social isolation [4], low or high education and wide exposure to social media as additional risk factors [10].

Biomarkers may indicate the likelihood of developing LCD [10, 11]. Consistent with a link to severe COVID (as is the case for all CNS complications [4]), high levels of cortisol, C-reactive protein (CRP) and interleukin (IL)-1β increase depression risks [10]; while other biomarkers are under investigation [1, 4, 12]. A study preprint reports that current HIV infection or transient reactivation of Epstein Barr virus (EBV) during acute COVID increased LC symptoms (including fatigue and cognitive dysfunction) [11], while prior cytomegalovirus infection was protective against LC symptoms [11].

Inflammation and depression may be related

A known, but complex two-way association between systemic inflammation and nitro-oxidative stress, and mood disorders, poor cognitive functioning and fatigue is supported by early COVID data [1, 13]. Long-term depression is not unique to COVID, as infection with SARS, influenza and EBV are all known risks for persistent mental health issues [8, 14].

Aside from numerous social stressors, the persistence of both the virus and inflammation, cerebrovascular damage including strokes, disrupted brain glucose metabolism, hypoxia and hypoxia-induced mitochondrial damage are all possible underlying mechanisms for LCD [1, 8]. In terms of biomarkers, investigations have linked CRP, IL-6, the neutrophil/lymphocyte ratio, the neutrophil count and the systemic inflammation index with LCD [1]. Sphingomyelinase activity (Tables 1, 2) correlates with inflammation and COVID severity [15].

Table 1 Initial data on selective serotonin reuptake inhibitors based on a review by Mazza et al. [1]
Full size table
Table 2 Initial data on pharmacological therapies for acute COVID that reduce the incidence of long COVID depression based on a review by Mazza et al. [1]
Full size table

Based on imaging, structural changes in the brain associated with LCD are similar to those sometimes seen in patients with MDD [1]. A small study revealed decreased grey and white matter in patients with LCD, 3 months following a COVID infection [16].

SARS-Cov-2 infects and inflames the brain...

SARS-CoV-2 affects the brain via systemic inflammation, appearing to preferentially target neurones [1, 4], and may be a neurotropic virus, acting via direct CNS infection [4]. While even a mild infection may prompt a damaging systemic response [1], severe inflammation and consequent hypercoagulability are most associated with COVID-related cerebrovascular disease [4].

SARS-CoV-2 infection leading to severe inflammation harms the brain in many ways [1, 4, 8]. For example, it increases the permeability of the blood-brain barrier (BBB), allowing cytokines to enter the brain [1]. Although cytokines are involved in learning and memory, abnormally high levels can reduce the plasticity of brain synapses, thereby decreasing learning and memory [17]; these changes likely underlie many aspects of depression [1].

Inflammation may also cause microglia and astrocytes to be abnormally and sometimes persistently activated [1, 4]. These cells normally support brain homeostasis and regulate immune responses [1, 4]. Activated microglia produce cytokines such as ILs and tumour necrosis factor (TNF)-⍺ [1], as well as reactive oxygen and nitroxygen species (ROS and NOS) [1]. Additionally, severe inflammation also causes hyperferritinaemia (elevated blood iron levels), which leads to the production of ROS [4]. Brain cells are especially prone to oxidative stress, and stress biomarkers have been associated with MDD [1].

... with specific depression-related effects

Depression, similar to other psychological disorders, is associated with a dysregulated immune response, which is amplified by reduced serotonin and increased cortisol levels [12]. The production, release and reabsorption of neurotransmitters, including serotonin, dopamine and glutamate, are sensitive to inflammation [1]. The availability of serotonin in particular can decrease due to the abnormal diversion of tryptophan (a precursor) [1], especially in women [13]. The beneficial effects of serotonin against viral inflammation (in addition to their effects on depression) includes the promotion of natural killer cells, reduced production of proinflammatory cytokines and inhibiting the cell entry of other viruses (HIV and dengue virus) [12].

In the short-term, the hypopituitary-adrenal axis tends to dampen the immune response via the production and release of glucocorticoids (e.g. cortisol) as a negative feedback mechanism [1]. However, over the long term, high glucocorticoid levels may result in resistance to this mechanism, thereby promoting inflammation. This may increase cortisol levels, which can precipitate depressive symptoms in vulnerable patients [1].

Treat depression with established options

A published trial in 60 patients with established LCD reported a potential benefit with selective serotonin reuptake inhibitors (SSRIs) [5], which have anti-inflammatory, anti-thrombotic and anti-viral properties [18]. SSRIs significantly improved LCD in 89 and 95% of patients with and without previous psychiatric history, respectively [5]. Furthermore, improvements in depression-related symptoms were reported with the mean Hamilton Depression Rating Scale score decreasing from 23.3 to 6.7 over 4 weeks.

Transcranial direct current stimulation was also helpful in one case study [1] and post-infection vaccination reduced depressive symptoms and other symptoms associated with LC in a large (n > 28,000) observational cohort study in the UK [5]:

  • The first dose of vaccine led to average 13% decrease in the odds of any LC symptoms and a 12% decrease for activity-limiting LC symptoms; however, improvements may or may not have been sustained

  • The second dose (a median 72 days later) led to a further immediate 9% decrease in the odds of any LC symptoms with a 0.8% decrease for each follow-up week

  • Additionally, the second dose led to an immediate 9% decrease in the odds of activity-limiting LC symptoms with a 0.5% decrease for each follow-up week

  • Following the second dose, the symptoms with the largest decrease (9–10%) were fatigue, headache and trouble sleeping

Trials for other options for the management of LCD are urgently needed [1]. Many have been discussed in the literature (e.g. alternative medicine, oxytocin and non-pharmacological support), but no trial data are available [1]. Trials in LC have been initiated, including a trial investigating temelimab (an antibody that targets human endogenous retroviral envelope protein HERV-W-Env) in patients with post-COVID neuropsychiatric symptoms [19].

While referral to psychological or psychiatric services is suggested in some guidelines [3], this may be impractical due to the high demand for these services [1].

Reducing the severity of COVID infections may reduce the risk of long COVID depression

Consistent with findings linking severe COVID infections to LCD, several studies confirm that pre-empting severe illness both before and during infection decreases the incidence of LCD [1, 13]. Agents with anti-inflammatory, anti-thrombotic and/or anti-viral properties are all helpful, with many readily available to outpatients via primary care providers (Tables 1, 2) [1].

The antidepressants in Tables 1 and 2 with COVID-moderating effects, can also reduce stress and anxiety [12] and some have other known or emerging beneficial actions [1, 12]. SSRIs may minimise the severity of COVID infections and/or depression mediated by increased cortisol, while clomipramine has anti-inflammatory properties and can easily cross the BBB [1, 12]. Melatonin in depression can re-establish normal circadian rhythm, which can be dysregulated with depression [1].

Agents with functional inhibition of acid sphingomyelinase (FIASMA) activity have recently been highlighted, as their ability to block the activity of acid sphingomyelinase can reduce viral cell entry and inflammation [15, 20]. Many antidepressants which are used for depression have been identified as posessing FIASMA activity, which may contribute to their efficacy in LCD (Tables 1, 2).

Drugs with other mechanisms of action have been reported to have potential benefits during COVID infections. The serotonin antagonist ondansetron improved outcomes in hospitalised patients, as it counteracts elevated serotonin levels that occur during COVID infections [15]. Casirivimab is a monoclonal antibody against SARS-CoV-2 that blocks viral attachment to ACE-2 receptors to prevent cell entry [4]. Metformin, aside from its benefits for the management of diabetes, has anti-inflammatory, anti-thrombotic and cardioprotective effects [8]; it is readily available and may be useful in both acute and long COVID [8].

Vaccination, as well as the administration of antidepressants or other neurological agents prior to infection may all prevent COVID hospitalisations and reduce LC (Tables 1, 2) [5, 8, 15, 20]. COVID vaccinations lessen disease severity and, in one phone application study, reduced the incidence of LC by 50% [5].

Two recent retrospective analyses of large health databases confirm that many antidepressants were protective against more severe illness (Tables 1, 2) [15, 20]. Anti-depressants with FIASMA activity appear to be particularly effective (Tables 1, 2) [15, 20], in addition to bupropion in one study [15].

Clinical trials a long way from treatments

Current clinical trials are focusing on risk factors and pathophysiology for LCD, fatigue and/or cognitive symptoms, rather than treatment [12]. Vortioxetine, a multimodal antidepressant, is being evaluated for cognitive deficits during or after COVID infection, but no other studies for the treatment of LCD were identified [12]. Potential research topics in this field include the investigation of the link between severe COVID infections and depression [1], the efficacy of phytochemicals, the relevance of FIASMA activity and examining pathways that are implicated in various neurodegenerative conditions, and potentially, depression [4, 8].

Take home messages

  • LCD is a vaguely defined, but common, condition that is associated with depressive symptoms lasting 1–12 months after a COVID infection

  • Although the risk factors for LCD are not directly known, the presence of comorbidities, greater disease severity and more frequent mental health visits were associated with LC

  • Reducing the severity of COVID infections, via pharmacotherapy or prior vaccination may also decrease the risk of LCD

  • Current data suggest pro-inflammatory cytokines and conditions may play a role in LCD, thus anti-inflammatory agents may be useful

  • SSRIs typically used for the treatment of depression may also be effective in LCD, owing to their anti-inflammatory, anti-thrombotic and anti-viral properties; though more trials are needed in this therapeutic area