Abstract
This review proposes a model of Long-COVID where the constellation of symptoms are in fact genuinely experienced persistent physical symptoms that are usually functional in nature and therefore potentially reversible, that is, Long-COVID is a somatic symptom disorder. First, we describe what is currently known about Long-COVID in children and adults. Second, we examine reported “Long-Pandemic” effects that create a risk for similar somatic symptoms to develop in non-COVID-19 patients. Third, we describe what was known about somatization and somatic symptom disorder before the COVID-19 pandemic, and suggest that by analogy, Long-COVID may best be conceptualized as one of these disorders, with similar symptoms and predisposing, precipitating, and perpetuating factors. Fourth, we review the phenomenon of mass sociogenic (functional) illness, and the concept of nocebo effects, and suggest that by analogy, Long-COVID is compatible with these descriptions. Fifth, we describe the current theoretical model of the mechanism underlying functional disorders, the Bayesian predictive coding model for perception. This model accounts for moderators that can make symptom inferences functionally inaccurate and therefore can explain how to understand common predisposing, precipitating, and perpetuating factors. Finally, we discuss the implications of this framework for improved public health messaging during a pandemic, with recommendations for the management of Long-COVID symptoms in healthcare systems. We argue that the current public health approach has induced fear of Long-COVID in the population, including from constant messaging about disabling symptoms of Long-COVID and theorizing irreversible tissue damage as the cause of Long-COVID. This has created a self-fulfilling prophecy by inducing the very predisposing, precipitating, and perpetuating factors for the syndrome. Finally, we introduce the term “Pandemic-Response Syndrome” to describe what previously was labeled Long-COVID. This alternative perspective aims to stimulate research and serve as a lesson learned to avoid a repeat performance in the future.
Keywords
Introduction
According to the World Health Organization, Post COVID-19 condition [i.e., Long-COVID] occurs in individuals with a history of probable or confirmed SARS-CoV-2 infection, usually 3 months from the onset of COVID-19 with symptoms and that last for at least 2 months and cannot be explained by an alternative diagnosis. Common symptoms include fatigue, shortness of breath, cognitive dysfunction but also others and generally have an impact on everyday functioning. Symptoms may be new onset following initial recovery from an acute COVID-19 episode or persist from the initial illness. Symptoms may also fluctuate or relapse over time.
1
Expert groups have emphasized that Long-COVID is a feared and common complication in the population. For example, a Canadian policy report wrote in their executive summary that “an important element of this pandemic response should now be focused on preventing and treating post-COVID-19 syndrome in patients . . . .” 2 A multinational Delphi consensus recommended that “REC2.9 Public health policy should take better account of the potential long-term impact of the unchecked spread of COVID-19, given ongoing uncertainties about the prevalence, severity, and duration of post-COVID-19 morbidity (long COVID),” and to “REC5.6 Prioritize research funding for long COVID . . .,” also writing that “continued uncertainty about the widespread consequences of long COVID and its implications for public health policy is an ongoing concern.” 3 Many studies and reviews have hypothesized that symptoms of Long-COVID are due to irreversible tissue damage, possibly from the viral infection itself, the immune response to the virus with systemic hyperinflammation and neuroinflammation, hypercoagulation, and/or tissue scarring.4–10
This review addresses the problem that most cases of Long-COVID remain “medically unexplained,” without reproducible structural organ pathology, which has led patients to accuse physicians of “medical gaslighting”—patient concerns have been dismissed and patients have been told there is “nothing we can do.” 11 To address this problem, we aim to develop the idea that the constellation of symptoms ascribed to Long-COVID are most often genuinely experienced persistent physical symptoms that are functional in nature and hence manageable and potentially reversible, that is, that Long-COVID is a somatic symptom disorder (SSD). First, we describe what is known about Long-COVID in children and adults, including a close look at limitations of existing studies, the incidence of and risk factors for the syndrome, and evidence that the symptoms are “medically unexplained.” Second, we examine so-called Long-Pandemic effects that create a risk for similar symptoms to develop in non-COVID-19 patients. Third, we describe functional syndromes (identified before the COVID-19 pandemic), including SSD, related disorders (syndromes), and functional neurological disorders (FNDs), and suggest that by analogy, most Long-COVID phenotypes are best conceptualized as one of these functional syndromes, with similar symptoms and predisposing, precipitating, and perpetuating factors. Fourth, we describe mass sociogenic illnesses in the past and discuss the concept of nocebo effects. Again, we suggest that, by analogy, Long-COVID is compatible with these descriptions. Fifth, we describe the current theory of the mechanism for functional disorders, the Bayesian predictive coding model (BPCM) for perception, including moderators that can make perceptual inferences functionally inaccurate. We describe how functional symptoms, and predisposing, precipitating, and perpetuating factors can be understood within this model. Finally, we discuss some implications of this framework for improved public health messaging and management of Long-COVID. We believe the current approach inducing fear of Long-COVID in the population, with constant messaging about disabling symptoms of Long-COVID and theorizing irreversible tissue damage as the cause of Long-COVID, has created a self-fulfilling prophecy by introducing the very predisposing, precipitating, and perpetuating factors for the syndrome.
Table 1 gives a summary of the main points made in each section, to help readers follow the narrative review leading to our conclusions. We offer this alternative perspective to argue that the pandemic response created the perfect storm of factors causing the fallout known as Long-COVID. We hope that the lessons learned will prevent a repeat performance in the future.
Summary of our main findings.
BPCM, Bayesian predictive coding model; SSD, somatic symptoms disorder. For more details and many references, please see the text.
What is known about Long-COVID
In children
A striking finding is that severe limitations in the available studies make definitive statements difficult. Systematic reviews list study limitations that include the following: lack of a control group while including a large range and number of nonspecific symptoms that are highly prevalent in the general population, and that could be new or intermittent or fluctuating, without pathognomonic features (i.e., almost anything could be Long-COVID); missing data on symptom severity; selection bias due to often self-selected unblinded patients with high nonresponse bias or more severe cases being more closely tested and monitored; outcome measurement bias due to unblinded subjective self-report of symptoms; missing data bias due to attrition; unmeasured confounding bias; recall bias in the face of substantial public awareness; and inaccurate denominators due to including cases rather than all infections.11–17 Hirt et al. concluded that “none [of the studies] provided evidence with reasonable certainty . . . inadequate, over-confident interpretation of the findings by media and decision-makers may cause potentially unnecessary fears and worries among parents and children with SARS-CoV-2 infection.” 12 Lopez-Leon concluded that “all studies had a high probability of bias.” 15 This may explain why there was marked heterogeneity between studies.13–17 In addition, most studies determined the cross-sectional prevalence of symptoms, while longitudinal study showed that most individual cases resolved over time, and many new cases developed adverse symptoms months after the infection at rates similar to controls. 18
Several studies with controls have found that Long-COVID may not occur at a higher rate among COVID-19 cases compared to control groups,19–24 or that Long-COVID was rare, with some symptoms worse in the control group (including worse objective scoring of somatic symptoms distress).22,25–30 Studies with controls have also found a higher sense of well-being 24 or quality of life (with reduced psychological symptoms),26,27 or no difference in well-being or quality of life in Long-COVID cases. 31 Some studies without controls suggested a high incidence of Long-COVID, but correcting those study denominators for MIS-C incidence (which occurs in <1/3000 infections), 32 the rates were <0.5%.33,34 Systematic reviews have reached similar conclusions. Zimmerman et al. 14 found a difference in persistent symptoms of <4% compared to controls and suggested that “infection-associated symptoms are not necessarily more common or severe than pandemic-associated symptoms,” 13 and that “nearly all symptoms are also reported in similar frequencies in those without evidence of infection [i.e., may be due to pandemic-related symptoms].” The Global Burden of Disease Collaborators found a Long-COVID incidence of 2.7% in non-hospitalized COVID-19 cases aged <20 years. 7 Hirt et al. 12 found that the “two largest studies had symptoms without infection in 34% and 53% [much higher than any estimate for children with infection in the uncontrolled studies].” Stephenson et al. 16 and Behnood et al. 17 found well-being to be similar in cases and controls. For individual symptoms, systematic reviews had conflicting findings, with Lopez-Leon et al. 15 finding no differences between cases and controls in mood symptoms, fatigue, headache, rhinitis, concentration, myalgia/arthralgia, cough, sore throat, or nausea/vomiting, but more dyspnea, fever, and anosmia/ageusia, while Behnood et al. 17 found no difference in abdominal pain, cough, fatigue, myalgia, insomnia, diarrhea, fever, dizziness, or dyspnea, but more cognitive difficulties, headache, loss of smell, sore throat, and sore eyes. Notably, one population-based study found no long-term increased use of primary or specialist care among COVID-19 cases compared to controls. 35
Among COVID-19 cases, risk factors for Long-COVID in individual studies have included female gender (in teenagers),26,27,31,36 poor or very poor pretest physical and mental health, 31 any long-term condition,29,36 hospitalization, 22 intensive care unit admission, 29 and more acute COVID-19 symptoms. 22 Systematic reviews found that risk factors included female gender,15–17 older age,15–17 worse self-rated physical and mental health,15–17 feeling of loneliness pre-infection, 16 severe or more symptoms during acute COVID-19,15,16 and lower study quality. 17
Although tissue damage was hypothesized as a cause of Long-COVID, 10 some authors admitted that this may not be the case,10,15,16,37–39 as summarized in Table 2.
Studies and editorials report that Long-COVID symptoms are not associated with structural pathology (i.e., are “medically unexplained”).
CFS, chronic fatigue syndrome; ME, myalgic encephalomyelitis; PET, positron emission tomography.
In response to one of the reviewers, we searched PubMed to prevent missing systematic reviews relevant to structural organ pathology in Long-COVID. First, we searched “Long COVID” OR “Post COVID condition,” AND “systematic review,” from January 1, 2023 to July 14, 2023. This returned 52 citations, and on review of the abstract and full text to detect data on structural organ pathology, 3 were included.87–89 Second, we searched “Long COVID” OR “Post COVID condition,” AND “pathology,” AND “systematic review” with the same date restriction. This returned nine citations, and on review of abstract and full text to detect data on structural organ pathology, two more publications were included.71,90
This review has an Almetrics score of 14,526, which puts it “in the 99th percentile (ranked 4th) of the 441,563 tracked articles of a similar age in all journals and the 99th percentile (ranked 1st) of the 38 tracked articles of a similar age in Nature Reviews Microbiology” (see: https://www.nature.com/articles/s41579-022-00846-2/metrics).
In adults
There are more studies of Long-COVID in adults than in children. Again, a striking feature is that severe study limitations make definitive statements difficult; almost all studies are of moderate or low quality.
40
Systematic reviews consistently list these common limitations, including attrition bias,
40
severity of Long-COVID symptoms not defined,5,6,40,41 usually no controls or even baseline prevalence of symptoms,6,8,40–42 unmeasured confounder bias (e.g., people who had COVID may be vulnerable in ways that explain both why they got COVID and why they went on to have adverse outcomes, with most studies adjusting for few if any covariates),4,8,40,42–44 selection bias (e.g., due to many included patients having been hospitalized, online recruitment, non-laboratory-confirmed diagnosis of acute infection, and bias to testing those with access to healthcare and with strong health-seeking behavior),5,42,43,45,46 and recall bias due to unblinded self-reporting.7,41 Other biases mentioned in individual studies included ascertainment bias (e.g., being more actively monitored after COVID-19),47,48 unblinded participants (e.g., who may be more likely to report symptoms after a positive test),23,49–51 unblinded investigators (e.g., who may be more likely to document possible post-COVID conditions among cases),
52
misclassification bias (e.g., often acute COVID-19 was not a laboratory-confirmed SARS-CoV-2 infection, with the denominator not based on seroprevalence),48,49,52–56 and possible reverse-causation (i.e., new conditions may be risk factors for more severe acute COVID-19 and may predate the index event).
52
We should not underestimate the effect of these biases that create a self-fulfilling prophecy (i.e., the more doctors look, the more doctors will find). In addition, the plethora of nonspecific, prevalent, often intermittent/fluctuating/relapsing or new usually self-reported symptoms9,40,46,57,58 make it “difficult to operationalize [Long-COVID] in clinical settings”;
59
in other words, it is often unclear whether particular symptoms could be attributed to Long-COVID, given the medical complexity and functional limitations of many patients and absence of specific markers for this condition, which could lead to ongoing monitoring, diagnostic testing, and specialist referral . . . [and] other comorbid conditions [and changes in their treatments] with symptoms that could potentially overlap [with Long-COVID] . . . .
59
For all these reasons, systematic reviews find very high heterogeneity between studies.4–6,9,40,42,45
Long-COVID often has a high incidence in studies without control groups. For example, Chen et al. found a global pooled prevalence of 43% (95% confidence interval (CI) 39, 46) among 31 studies (only one study was mentioned as having had a control group in Table 1), higher among hospitalized (54%) than non-hospitalized (34%) cases. 42 However, this incidence is much lower in studies that had control groups.23,40,49,56,60,61,62,63 For example, the Office of National Statistics in the UK found a difference of 1.6% in incidence compared to controls. 49 A prospective multicenter registry study found persistently poor physical, mental, or social well-being (i.e., moderate to severe impairments across any PROMIS domain) at 3-month follow-up of 39.6% in COVID-19-positive versus 53.5% in COVID-19-negative people, persistently poor mental health (i.e., moderate to severe anxiety or depression) at 3-month follow-up in 21.9% COVID-19-positive versus 27.3% COVID-19-negative people, with more improvement from baseline in the COVID-19-positive participants. 63 Studies using electronic health records have also found low rates. In the UK, (i) general practice consultation rates for symptoms among previously hospitalized COVID-19 patients compared to controls were <1% higher, except for breathlessness (1.4% higher), and among community COVID-19 cases was <1% increased from baseline, such that “healthcare use was similar to that in the negative control group” 47 and (ii) compared to controls, according to any of 32 symptoms, Long-COVID was 1.08% higher, according to any of 62 symptoms, it was 1.92% higher, and individual symptoms had very small differences between cases and controls (<0.05% for 49, 0.05–0.10% for 5, 0.1–0.2% for 6, and >0.2% for the two symptoms of pain at 0.71% difference and fatigue at 0.22% difference). 50 In US healthcare database studies, (i) one new type of clinical sequelae that required medical care was 1.65% higher than in a viral LRTI control group, with non-hospitalized cases having a risk difference well <1%; 48 (ii) non-hospitalized cases had small absolute risk differences compared to controls of <2% for all symptoms except respiratory signs/symptoms at 2.85%, and hospitalized cases usually at <2% for all symptoms except for malaise/fatigue at 3.65% compared to influenza controls; 64 and (iii) non-hospitalized cases had shortness of breath incidence 0.4% higher than controls, with other symptoms higher in test-negative patients, and hospitalized cases had differences from controls ⩽1% for symptoms except for shortness of breath at 5% and fatigue at 2.1%. 30 A UK healthcare worker follow-up study at three hospitals found no difference in cardiovascular findings (cardiac structure, function, magnetic resonance imaging (MRI) tissue characterization, and biomarkers) between COVID-19 cases and uninfected controls, writing that this “provides societal reassurance for the cardiovascular health of working-age individuals with convalescence from mild SARS-CoV-2.” 65
Systematic reviews have also suggested a high incidence in studies without controls,4–6,8,40,42,45 but small differences when compared to control groups.6,7 Alkodaymi et al. found “only 6 studies with comparator group of COVID-19 negative patients . . . only two were rigorously designed,” 42 and these studies found a higher risk of mood disorder and anxiety,66,67 but a lower incidence of most symptoms (sore throat, cough, body ache/muscular pain, nasal symptoms, headache, abdominal pain, or nausea/vomiting) in the case group, 60 no difference in neurological or cognitive deficits among healthcare worker cases and controls, 67 and no difference in health-related quality of life domains among hospitalized cases and controls. 68 The Global Burden of Diseases Collaborators found at least one cluster of Long-COVID in 6.2% of cases (fatigue cluster 3.2%, cognitive cluster 2.2%, and respiratory cluster 3.7%) compared to controls or to status prior to COVID-19 infection. 7 In a review of UK studies, Thompson et al. found long-COVID among 7.8–17% of cases (without a control group), but when limited to people with symptoms that “limited day-to-day function,” the rate was lower at 1.2–4.8%. 41 Also of note, studies suggest much lower odds of Long-COVID during the Omicron variant wave.51,69,70 Another systematic review found the risk difference (95% prediction intervals) for Long-COVID compared to controls was 13.9% (−16.2, 43.9), compared to controls in community-based samples 10.1% (−12.7, 32.8), and compared to controls in community-based samples assessed as at low risk of bias 4.8% (−13.2, 22.7). 71
Among COVID-19 cases, risk factors for Long-COVID in systematic reviews included female gender,5,7–9,41, 42,45,72 dyspnea in the acute phase or asthma/chronic-lung-disease/chronic-dyspnea,5,41,42,72 previous psychiatric diagnosis (including prepandemic psychological distress),5,8,41,72 severity of acute COVID-19 (including number of acute symptoms 42 and hospitalization,7,42,43,72 especially in intensive care),5,7–9,42,43,72 and underlying comorbidity.8,42,45,72 Individual studies have also emphasized psychiatric diagnoses as risk factors, for example, preexisting fibromyalgia, anxiety, depression, migraine, irritable bowel syndrome (IBS), eating disorder, or back pain; 50 and preexisting diagnosis of depression or anxiety.73,74 Although intensive care admission has been identified as a risk factor, there appeared to be no difference from other intensive care controls. 75 Pre-infection poor sleep health was also an important risk factor. 76 An interesting finding was that in a US healthcare plan study, cases also had an increased risk of new clinical sequelae that required medical care 30 days before the index COVID-19 diagnosis date. 48
As mentioned above, many reviews hypothesized tissue damage as the cause of Long-COVID.4–9,77 A recent review in Scientific American went so far as to claim that Long-COVID is a neurological disease due to some combination of thrombotic events, central nervous system inflammation and autoimmunity, and persistent viral infection/proteins in the brain and that treatments may include IVIG, rituximab, corticosteroids, beta-blockers, and amphetamine/dextroamphetamine. 78 However, many reviews also reported that symptoms largely remain unexplained,4–9,23,43,45,53,64,71,73,77,79–90 as summarized in Table 2.
The most frequent symptoms of Long-COVID have included shortness of breath/dyspnea, fatigue/exhaustion, sleep disorders/insomnia, cough, cognitive problems (with concentration and memory), and psychiatric problems (anxiety and depression).4–6,40,42 These symptoms have been suggested to occur in overlapping clusters of persistent fatigue with bodily pain or mood swings, cognitive problems commonly referred to as “brain fog,” and ongoing respiratory problems.7,50 Despite studies not determining the severity of these symptoms, some have made alarming claims about them. For example, “[fatigue is] unrelenting exhaustion and a constant state of weariness that reduces a person’s energy, motivation, and concentration,” 9 and “millions of economically active people may be disabled by Long-Covid” stating that fatigue was “a feeling of utter exhaustion, energy drain, or bodily dysfunction that is not necessarily triggered by exertion and is not always relieved by rest.” 57 Another review warned of “an alarming picture of an emerging neurological health crisis.” 77 A recent narrative review of Long-COVID claimed that COVID-19 “can severely damage multiple organs, including the nervous system” and that “COVID-19 has significant long-term effects on the nervous system.” 84 A Scientific American paper asserted that patients have “extreme fatigue,” “add up to millions more people affected—and potentially disabled,” and “[Long-COVID] could last many years.” 78 We believe this exaggeration is counter-productive, as will be discussed later.
Long-pandemic effects
Given that a major risk factor for Long-COVID included worse pre-COVID mental health, it is important to emphasize the adverse effect that the pandemic response is having on the population.
In children, the American Academy of Pediatrics called a national emergency in children’s mental health, with an “escalating mental health crisis due to physical isolation, ongoing uncertainty, fear, and grief,” including increased mental health emergencies, suicide attempts, depression, anxiety, trauma, loneliness, and suicidality. 91 In Canada, The Hospital for Sick Children similarly reported higher depression and anxiety in children associated with time spent online learning, less participation in sports and extracurriculars, and increased screen time. 92 The report noted “significant and sustained mental health effects that the public health mitigation strategies and school closures have had on children . . . Kids need school, they need their friends, and they need to have fun.” 92 Systematic reviews have documented high rates of anxiety and depression symptoms,93–97 noting that “school closures and social lockdown . . . were associated with adverse mental health symptoms (such as distress and anxiety) and health behaviors (such as higher screen time and lower physical activity),” 93 and that these rates are double baseline rates of anxiety and depression. 94 These mental health effects were predicted in a prepandemic systematic review in children that found “social isolation and loneliness increased the risk of depression, and possibly anxiety at the time at which loneliness was measured and between 0.25 and 9 years later.” 98 A study confirming these more severe internalizing mental health problems in youth also found that, compared to carefully matched peers before the pandemic, after pandemic shutdowns, there was maladaptive neurodevelopment on MRI with reduced cortical thickness, larger hippocampal and amygdala volumes, and more advanced brain age (findings typical of previous cohorts that had experienced significant adversity during childhood). 99 A systematic review also found that “considerable disruption in the lives and routines of children, adolescents, and families” have led to an approximately 52% increase in screen time. 100 Studies in the United States have found increased rates of mental health conditions including eating disorders, depression, anxiety, and sadness.101–114 Studies from around the world have found markedly increased rates of eating disorders,105–108 likely due to increased anxiety and “a combination of social isolation and school closures [that] has disconnected patients from protective factors . . . reduction of extracurricular activities, school routine and peer relationships have created room for eating disorder cognitions to intensify.” 105 Several studies have also found reduced physical activity in children during the pandemic,92,93,103,109–111 which is a strong risk factor for depression. 112 Not surprisingly, the pandemic has been associated with increasing body mass index,113,114 and reduced cardiorespiratory fitness in children. 115 As editorialists commented, “when we close schools, we close their lives.” 116
In adults, highly prevalent adverse mental health, including stress, anxiety, depression, substance use, suicidal ideation, and loneliness, have been documented in systematic reviews117,118 and individual studies in the United States 119 and the UK. 120 The COVID-19 mental disorders collaborators identified being female as a risk factor, and commented that this may be due to women being more likely to be affected by the social and economic consequences of the pandemic response (e.g., having additional carer and household responsibilities due to school closures or unwell family; more likely to be financially disadvantaged due to lower salaries, less savings, and less secure employment; more likely to be victims of the increased domestic violence). 118 In addition, government covert psychological strategies used to induce fear in the population likely contributed, with people being “bombarded with fear-inducing information,” 121 aiming to increase “the perceived level of personal threat” which “ultimately it [sic] backfired because people became too scared.” 122 Taylor has coined the term “COVID Stress Syndrome” that describes this fear, including fear of becoming infected, of coming into contact with fomites or foreigners, and of socioeconomic consequences, often with compulsive checking and reassurance seeking, and traumatic stress symptoms. 123 Of note, an unhealthy lifestyle had a dose-dependent association with Long-COVID, and this included physical activity, diet quality, and adequate sleep, factors that were adversely affected by the pandemic response. 124
It is worth mentioning that, even prepandemic, stress and adverse mental health were associated with inflammation and neuroinflammation, making the direction of causality difficult to determine.8,125–129 Brusaferri et al. found in pandemic patients without having had COVID-19, “novel evidence of elevated neuro-inflammatory markers in cortical and subcortical regions . . . implicates neuroimmune activation as a possible mechanism underlying the non-virally-mediated symptoms experienced by many during the COVID-19 pandemic.” 128 Reviews have written that if inflammation is found, this is “consistent with decades of psychoneuroimmunology research in patients with anxiety disorders, depression, and traumatic stress-related disorders,” 8 “inflammation and immune dysregulation may link psychological distress with post-COVID-19 conditions [e.g., distress is associated with chronic systemic inflammation, and “mental health disorders are associated with chronic low-grade inflammation and microglia activation” in the central nervous system],” 125 “psychosocial factors are also very important in regulating our immune system [e.g., immune abnormalities are found in chronic fatigue syndrome (CFS), fibromyalgia, chronic pain, depression and other mental health disorders, with increased peripheral inflammation and activation of glial cells with neuroinflammation],” 126 and suggesting that maladaptive behavioral responses are actually causing the abnormal immune findings. 127 Poor sleep health, which increased during the pandemic, is associated with chronic low-grade inflammation and immune abnormalities, COVID-19 risk and severity, and Long-COVID. 76 Anxiety and depression have been found to be associated with more severe acute respiratory infections, including COVID-19 (i.e., hospitalization, intensive care, and mortality), and it is possible this is mediated by the associated immune abnormalities. 130 Psychosocial stress is a strong modifiable risk factor for stroke (with higher locus of control at home or at work being an effect modifier lowering this risk), also possibly mediated by neuroinflammation. 131 Compatible with this is that social disconnection (e.g., social isolation and loneliness) is “a key determinant of health” with an effect on all-cause mortality “comparable in magnitude with that of smoking (15 cigarettes/day) and high levels of alcohol consumption (6 drinks/day).” 132 Another systematic review that found social isolation and/or loneliness associated with increased all-cause mortality suggested a mechanism may be that they “lead to activation of the hypothalamic-pituitary-adrenocortical axis . . . [which] affects a wide range of physiological functions, including . . . metabolism and inflammatory control [with pro-inflammatory immune response] . . . .” 133
Functional somatic disorders (Diagnostic and statistical manual of mental disorders (DSM-V) somatic symptoms and related disorders)
We use the term “functional” to describe these disorders, and the term “medically unexplained” to describe the associated symptoms. These terms are not satisfactory and can be misleading and used pejoratively; however, we use them as commonly used in the medical literature. 134 The term “functional” is misleading because it can imply that all is “functioning normally” in the body so that symptoms are “all in the head,” which mistakenly suggests that other so-called structurally caused symptoms are not also experienced “all in the head” and that not all symptoms are perceived through the same common pathways (“reflecting the [complex interactions between, and] integration of bodily and brain functions and dysfunctions”), as explained later. 134 The term “medically unexplained” is misleading because it “implies that explanations that involve psychosocial or cultural factors are not part of medicine,” and as explained later, these in fact do have an organic brain function explanation. 135
Somatic symptom disorder
SSD describes a patient with persistent physical symptoms (that may include bodily complaints of pain, fatigue, and perceived disturbances of organ functions) that are distressing and/or result in significant disruption in daily life.136–139 These symptoms are accompanied by disproportionate and persistent thoughts (about the seriousness), feelings (high health anxiety), and/or behaviors (excessive energy or time devoted to health concerns) related to them.136–138 These are real experienced symptoms that often cause significant disability.136–138 These are also medically unexplained symptoms (MUS), in that there is no explanatory structural or other pathology.140,141 Both children and adults with somatic symptoms and related disorders have frequent healthcare consumption (sometimes related to under-recognition) and disability.142,143
An SSD is surprisingly common,137–139 with a systematic review finding a general population (using self-report) mean frequency of 12.9% (95% CI 12.5, 13.3), and a non-specialized general medicine setting mean frequency of 35% (95% CI 33.8, 36.2). 137 Reviews have identified predisposing, precipitating, and perpetuating factors for SSD. Predisposing factors include childhood adversity (including parental ill health), stress, female gender, depression, and anxiety (including illness worries, catastrophizing, and fear); of note, these are not required to make the diagnosis.135–141 Precipitating factors (triggers) include acute illness (viral or other infection, physical trauma, or illness) and psychosocial trauma (negative life events, environmental events such as terrorist attacks).135–141,144 Many diverse infections have been associated with so-called unexplained post-acute syndromes with “remarkably consistent” “core symptoms centering on exertion intolerance, disproportionate levels of fatigue, neurocognitive and sensory impairment, flu-like symptoms, unrefreshing sleep, myalgia/arthralgia, and a plethora of nonspecific symptoms.” 144 Perpetuating factors (context amplifiers) include poor physical and social activity, poor sleep hygiene, missed/late diagnosis (often with medical uncertainty, inappropriate treatments, and frustrations with care), unhelpful cognitions (e.g., fear that exercise is damaging, that there is a missed physical cause for symptoms, or symptom focusing), and social reinforcement (e.g., by the wider social circle and media).135–141 Suggested treatment is multidisciplinary and includes explaining the diagnosis of a somatic symptoms disorder through education of the patient and caregivers (including the disorder of brain function explanation to be discussed later), graded physical therapy/rehabilitation, and cognitive behavioral therapy, with an emphasis changed from cure to care and coping, focusing on perpetuating factors that affect overall functioning.136,138 Prevention and treatment would also include managing depression and anxiety (including reducing illness worries and fear).
Related disorders (syndromes)
These are SSD that consist of certain medically unexplained persistent physical symptoms that have been named. Fatigue and pain cluster in syndromes of IBS, CFS, fibromyalgia, and some chronic pain syndromes, and cardiorespiratory symptoms cluster in non-cardiac chest pain or hyperventilation. 140 CFS is associated with debilitating fatigue, malaise, headaches, muscle/joint pain, nausea, disrupted sleep, dizziness, and cognitive difficulties. 145 The same predisposing, precipitating, and perpetuating factors as in SSD have been identified.139–141,145,146 The trigger has often been a recent viral (or other) infection.144,147,148 There are no specific diagnostic physical signs or biomarkers, and no consistent cytokine, cellular, auto-antibody, mitochondrial, or neuroimaging findings. 149 A systematic review found that “the proportion of patients with one unexplained clinical condition meeting criteria for a second unexplained condition was striking” with about 30–80% overlap among CFS, fibromyalgia, IBS, multiple chemical sensitivity, temporomandibular disorder, tension headache, interstitial cystitis, and post-concussion syndrome. 150 Another systematic review found CFS and fibromyalgia coexist in 47.3% of cases. 151 These functional syndromes shared commonalities in symptoms (e.g., fatigue, pain, abdominal distension or bloating or pain, headache), disability out of proportion to physical examination findings, inconsistent demonstration of laboratory abnormalities, and an association with “stress” and psychosocial factors.150,152 Treatment suggestions are similar to SSD in general, focusing on explanation and education, physical therapy/rehabilitation, and cognitive behavioral therapy (aiming to change maladaptive thoughts and behaviors).127,145
An older study is particularly interesting. 153 Of those suffering from environmental hypersensitivity disorder, 90% reported suffering from at least one other so-called “fashionable” (i.e., media popularized) condition, including food allergy, candidiasis hypersensitivity, post-infectious neuro-myasthenia, fibrositis, and temporomandibular joint syndrome. 153 Patients had multiple, vague, nonspecific, ill-defined, and generally common symptoms (e.g., fatigue, nausea, dizziness, respiratory symptoms, poor concentration and memory, headaches). 153 The author suggested that sociocultural context, including “intense media interest and hyperbole” and “illness-behavior role models” influenced the attribution of cause and exaggerated worry over health. 153 In addition, perpetuating factors were suggested to include becoming active advocates of the disease, propensity for self-diagnosis, and achieving a “scientific aura” provided by media, support groups, and healthcare providers, such that the “disorder becomes their identity.” 153 Of note, some have considered chronic Lyme disease a functional syndrome, with risk factors identified to include depression, negative affect, catastrophizing pain, anxiety, and past traumatic psychological events.154–157
Functional neurological disorders
FNDs are common and involve significant, genuinely experienced neurological symptoms without structural pathology that cause considerable distress and disability.158–165 Disorders are commonly motor (including tremor, dystonia, tics, paralysis, abnormal gait, and/or paroxysmal non-epileptic seizures) and can be sensory.159–164 Cardinal features include that symptoms fluctuate in severity, waxing and waning over time, decrease with distraction and increase with body-focused attention, and movements are made with excessive effort and fatigue, with positive neurological signs that demonstrate internal inconsistencies (i.e., FND is not a diagnosis of exclusion).162–164,166 FND symptoms often coexist with other persistent medically unexplained physical symptoms including dizziness, chronic pain, fatigue, sleep disturbance, memory symptoms, and dissociative symptoms.160–162,164 In adults, FNDs also often coexist with other related disorders (syndromes) including IBS, fibromyalgia, chronic pain, and cardiorespiratory symptoms.160,161 Associated anxiety and depression are common.160–162,164,165 Similar (to SSD) predisposing, precipitating, and perpetuating factors have been identified.159,160,162–167 Panic attacks and autonomic hyper-arousal are included as predisposing and precipitating factors for FNDs, and are likely similar to fear and anxiety being included for SSD.160,162 Treatment is recommended to be multidisciplinary and to focus on giving and explaining the functional diagnosis, emphasizing the condition is potentially reversible, providing physical therapy/rehabilitation, cognitive behavioral therapy, and follow-up.158,160,162 In explaining the diagnosis to patients, it is suggested not to simply emphasize normal test results, say “there is no neurological disease,” or focus prematurely on psychiatric comorbidity (i.e., don’t turn a risk factor into the “cause,” and don’t suggest a mind–body dualism).158,160,163 If not explained properly, often the patient hears “this is all in your head,” disrupting patient–clinician trust and further complicating recovery.
The FND variant functional cognitive disorder (FCD) is particularly interesting because patients often describe “cognitive fog” or “brain fog” (terms used well before the COVID-19 pandemic).144,168–170 This common disorder, estimated to comprise at least one-quarter of memory and cognition clinic referrals, is characterized by subjective memory impairment, attention and concentration difficulties, and common co-occurrence of multiple functional symptoms, in the absence of underlying brain structural pathology.168,170,171 The positive diagnostic feature is internal inconsistency, where “functions that remain easy and automatic become difficult when attention is focused towards them.” 170 The cognitive symptoms may be due to a lack of attentional reserve, as evidenced by being susceptible to distraction, with slow information processing during periods of excessive attention toward the body. 169 “Similar cognitive symptoms [forgetfulness, distractibility, word-finding difficulties] . . . [are] correlated with pain in Fibromyalgia and with mental or physical exertion and fatigue in CFS . . . Evidence does not support the existence of separate cognitive disorders in CFS, Fibromyalgia, and FND.” 169 So-called brain fog has been described in association with a wide range of illnesses, drugs, and behaviors, including Long-COVID. 172
Long-COVID as a form of functional disorder
Some studies report findings that suggest Long-COVID may be a functional syndrome. Wang et al. found that preexisting depression, anxiety, worry about COVID-19, perceived stress, and loneliness, in a dose-dependent manner, were stronger risk factors for Long-COVID than other established risk factors. 125 A Norwegian study found that, in non-hospitalized 12- to 25-year-olds, SARS-CoV-2 positivity was not associated with the development of Long-COVID or post-infective fatigue syndrome at 6 months, while baseline symptoms severity, low physical activity, loneliness, and prior negative life events were. 173 Another study found that, among 790 COVID-19 patients who survived hospitalization, life stressors were the strongest independent predictors of prolonged symptoms. 174 Somatic symptoms among Chinese students have correlated with concern regarding the threat to life and health from COVID-19. 175 A single-center study of Long-COVID patients referred to a neurology clinic found that 0/49 (0%) had specific brain MRI findings, 32/50 (64%) met DSM-5 criteria for SSD, and in the remaining 18/50 (36%) patients “SSD was considered possible given the high score on diagnostic scales.” 176 In this study, patients described feeling significant anxiety, social isolation, fear of infecting relatives, and fear of dying during the acute infection, and 38% had a premorbid functional disorder preceding the infection. 176
Willis and Chalder explicitly suggested that Long-COVID may be an SSD. 177 They suggested that pandemic effects “create a ‘perfect storm’ for the development of persistent physical symptoms,” contributing to predisposing (e.g., psychological distress, stress, anxiety, depression, inactivity, social isolation, adverse media exposure), precipitating (e.g., acute COVID-19 symptoms), and perpetuating (e.g., beliefs of a serious prolonged illness conveyed by the term “long-hauler” and medical and media portrayal of serious consequences and prolonged recovery) factors. 177 Another group suggested that “a new paradigm is needed to explain long COVID” and “it is time to break taboos based on a dualistic understanding of physical versus mental illness and bring in existing knowledge about functional somatic symptoms to provide improved explanations and treatments.” 178
Several authors have reported that many Long-COVID patients (between 27% and 45%) meet the criteria for CFS.179–182 A systematic review found Long-COVID descriptions to have similar neurological symptoms to CFS, including problems thinking, remembering, or concentrating, dizziness, fatigue, headaches, muscle or joint pain, and sleep problems. 181 A systematic review of 52 studies found the incidence of CFS in Long-COVID to be 45.2%. 183 In Long-COVID, subjective cognitive complaints were common without abnormal neuropsychological battery scores, and altered dyspnea perception occurred without abnormal lung function. 179 In a single-center study of Long-COVID patients referred to a neurology clinic, 45/50 (90%) met the criteria for CFS, 48/50 (96%) had symptoms compatible with FCD, all had an absence of specific MRI changes, and in the 15 that had a neuropsychological assessment, only mild impairment of attention was found. 176 Another study found that 6/15 (40%) Long-COVID patients met screening criteria for fibromyalgia. 184
Many studies have reported FND during the pandemic. A systematic review found that Long-COVID patients can have similar neurological symptoms to those found in FND, including dizziness, dysphagia, facial pain and spasms, fatigue, headaches or migraines, olfactory symptoms, depression, movement disorders pain, and sleep problems. 181 Increased presentations (up to threefold) of functional motor disorders and other FND have been reported in children and adults, attributed by study authors to pandemic psychological stressors related to social isolation, financial strain, loneliness, anxiety, depression, and mobility restriction.185–187 Several studies have documented a marked increase during the pandemic in functional tick-like behaviors (a form of FND) in children and adolescents, especially females, often associated with other somatic symptoms.188–190 This functional disorder is believed to have occurred with the pandemic-associated surge in social media and digital technology use (i.e., viewing of social media content involving tic-like attacks) combined with increased stress and isolation associated with imposed pandemic restrictions (i.e., lockdowns and mental health deterioration).188,189,191 Those with functional seizures (paroxysmal nonepileptic seizures) reported increased frequency during the pandemic. 192
Some side effects of vaccines for COVID-19 have been recognized as precipitating functional disorders not pathologically due to the vaccine.193–195 The WHO has coined the term “neurological immunization stress-related responses” to describe this phenomenon due to “pandemic stress, feelings of uncertainty about COVID vaccination, normal transient physical symptoms, and discomfort after vaccination.” 193 Acknowledged risks included predisposing (e.g., increased somatic attention from checking for signs of COVID and threat-related hypervigilance, abnormal expectations/beliefs, fear and distrust, widespread information in the media, stress, anxiety), precipitating (e.g., pain/myalgia from vaccination), and perpetuating (e.g., diagnostic delay and incorrect diagnosis and treatments) factors.193,194 Surely similar points, by analogy, can be made about Long-COVID.
Mass sociogenic illness
We will include the unfortunate term “mass hysteria” when directly quoting from some sources. The term is unfortunate as it can be interpreted pejoratively, and suggests mental illness. As will become clear below, we intend neither.
Past exposures
Mass functional illness has been described for centuries across varied cultures.189,196–200 Symptoms have included breathlessness (e.g., hyperventilation, shortness of breath, tight chest, cough), nausea and vomiting, headache, dizziness and light-headedness, weakness, fatigue, diffuse musculoskeletal pain, sleep disturbance, and neurological symptoms (e.g., concentration and memory complaints).196–198,200 The precipitating event is belief in an environmental cause, with a dramatic emergency response, creating extreme stress and fear of a perceived often unpredictable (or inescapable) threat; this threat often reflects the dominant sociocultural concerns of the time, such as environmental toxicity or infectious diseases.189,196–200 Predisposing factors have been identified to include female gender, anxiety (e.g., checking behavior, engaging in preventive and avoidant behaviors), depression, psychological distress (e.g., fear induced by repeated extensive negative media coverage), and unhealthy behaviors (e.g., unhealthy eating, lack of exercise, disordered sleep, lack of socialization).196–198 Perpetuating factors involve social contagion, including physical or visual proximity to others who are ill (especially involving reuniting with the group affected), the excitement induced by the emergency and media response (e.g., collective anxiety, stress, and fear), and clinically labeling the illness and providing preferential medical care for it.196–198,200 The perceived threat of terrorism or exposure to a poison/toxin or infection has been a common precipitant in modern times.189,196–201 Many outbreaks of mass sociogenic illness in the past have been confined to a small group of people, but social (and other global) media “breaks the geographical barriers that typically confine such symptoms.” 189 Mass sociogenic illness is “exacerbated and self-reinforcing when the negative information comes from an authoritative source, when the media are politicized, and social networks make the negative information omnipresent.” 200 It has been suggested that the contribution of digital media and the internet to anxiety and emotional contagion may contribute to a global form of mass sociogenic illness as, “what are temporarily, locally limited, isolated outbreaks of mass hysteria, the state may convert into a global mass hysteria for an extended period of time.” 200
Some authors we believe gave prescient advice about these syndromes (see Table 3), and have suggested that de-escalating fear with clear information is paramount for treatment.197,198,202
Some quotations that merit emphasis, from papers describing mass sociogenic illness prior to the COVID-19 pandemic.
Past pandemics
The “Russian Influenza” pandemic of 1889–1892 was associated with widespread dread (especially of feared respiratory complications) and media focus, such that the “dread could itself become a ‘nervous’ symptom of the disease.” 203 “Post-influenza varied from lethargy to lassitude, to more serious conditions such as depression and neurasthenia [notoriously vague and amorphous],” such that “somatopsychic aspects of the disease tended to blur the lines between organic physiological processes and psychogenic categories.” 203 Symptoms were diverse and unpredictable, and convalescents were “plagued with mysterious and erratic symptoms and chronic illnesses” that were given many names including neurasthenia, nerve exhaustion, and prostration. 204
There is evidence that other influenza pandemics had similar post-infectious outcomes. The 1918–1919 pandemic was followed by symptoms including “loss of muscular energy” and debilitating lethargy, “nervous complications” and “apathy and depression,” and restlessness or sleeplessness. 205 Spanish flu survivors “reported sleep disturbances, depression, mental distraction, dizziness, and difficulties coping at work.” 206 A study from Norway during the 2009 influenza H1N1 pandemic found being diagnosed with influenza infection to have an adjusted HR 2.04 (1.78, 2.33) for CFS. 148 There are 12 reports of outbreaks of CFS since 1934, with a systemic syndrome of excessive fatigue, myalgias, headaches, low-grade fever, other constitutional symptoms, and neuropsychological changes (including forgetfulness, difficulty thinking, inability to concentrate). 207
War syndromes
Post-combat disorders (apart from Post-Traumatic Stress Disorder) are common and have been labeled variously as Soldier’s Heart, Irritable Heart, Disordered Action of the Heart, Rheumatism, Shell Shock, Effort Syndrome, Non-Ulcer Dyspepsia, Toxic Neurasthenia, and Gulf War Syndrome. 208 These “war syndromes” have all included overlapping clusters of common nonspecific multisystem MUSs of fatigue, weakness, sleep difficulties, headache, muscle ache, joint pain, problems with memory, attention and concentration, nausea and other gastrointestinal symptoms, anxiety, depression, irritability, palpitations, shortness of breath, dizziness, sore throat, and dry mouth,208,209 and often overlapped with other named functional syndromes including CFS, fibromyalgia, multiple chemical sensitivity, and IBS. 209 Precipitating factors included the traumatic experience of war, described as “man’s reaction to adversity.” 201 Predisposing factors identified have included anxiety, depression, and “popular health fears and limitations of medical science [thus conveying “a sense of seriousness”].”208,209 Perpetuating factors have included having many investigations, media stories (including the internet), and the secrecy associated with the military (likely inducing more fear).208,209
We give what we consider some prescient descriptions in Table 3.208,209
The COVID-19 pandemic
The increase in FND during this pandemic has been suggested to reflect mass sociogenic illness by several authors.148,185,189,190 The reach of these disorders has been exacerbated by social media (e.g., so-called TikTok Tics).185,190,210 The pandemic response affected the mental health of many people creating widespread anxiety, fear (with anticipated negative impacts), depression, helplessness, adverse experiences, and social isolation without access to supports, again suggested to exacerbate the global reach of these disorders.189,190,210
Bagus et al. alluded to Long-COVID as a mass sociogenic illness, and suggested that with “the digital age of global mass and social media, the possibility of global mass hysteria exists . . . [and] can cause real symptoms in a self-fulfilling prophecy . . . anxiety and fear contribute to this process . . . both media and the state may actively contribute to the contagion of fear [e.g., stressing breathing problems, the possibility of unknown long-term irreversible health damage] . . . .” 200 This suggests that treatment would include reducing stress and fear, and encouraging exercise, socializing, and distractions. 200 Importantly, they suggested that “negative information which is spread through mass media repetitively can affect public health negatively in the form of nocebo effects and mass hysteria.” 200
Nocebo effects
Nocebo effects occur when cognitive expectancy of an anticipated negative future outcome causes that very physiologic negative outcome to occur. Nocebo effects are “powerful, pervasive, and common in clinical practice” and include phenomena such as side effects associated with placebo treatment (known to occur in around one-quarter of people). 211 Effects include any MUSs such as pain, dyspnea, and even measures of inflammation. 211 Some symptoms turn out to be preexisting symptoms that were previously ignored or dismissed. 211 Nocebo effects are exacerbated by anxiety, psychological distress, verbal suggestion (e.g., how a medication is framed), learning (e.g., anticipation based on the experience of others, modeling, reports in mass media and lay press, and social observation), and relationship with the clinician (e.g., worrisome information, pessimistic expectations, social messaging, and therapeutic milieu). 211 A randomized study found that patients with MUSs in the positive frame (i.e., those given a firm diagnosis and told confidently they would be better in a few days) had much better outcomes on follow-up than the negative frame (i.e., those told the doctor cannot be certain what is the matter with them) group, suggesting that “[the doctor] is the placebo [or nocebo] and his/her influence is felt to a greater or lesser extent at every consultation.” 212
FND has been suggested to share mechanisms with the nocebo effect, including a functional mechanism (i.e., no organ pathology, with inconsistent waxing and waning of symptoms), “maladaptive prior expectations that are reinforced via attention, stress and anxiety,” and “prior beliefs, negative expectations, heightened attentional focus” sometimes exacerbated by “negative interactions with their doctors, perceived poor treatment, and sensations of feeling abandoned.” 213 Such is the power of negative suggestion, with social influences and learning, and dissemination of negative information creating self-fulfilling mechanisms for symptoms. 213
Several studies of Long-COVID suggest that nocebo effects are occurring.23,24,214,215 Matta et al. found that belief in having been infected (i.e., self-reported infection) had odds ratio (OR) ranging from 1.39 to 16.37 for persistent symptoms—that is, belief was associated with persistent symptoms to a similar extent among participants with negative and positive serology results. 214 Having had confirmed infection by laboratory testing was associated only with persistent anosmia, suggesting “symptoms may not emanate from SARS-CoV-2 infection per se.” 214 Some self-identified Long-COVID support group surveys have also found that symptoms are similar in those having had confirmed and unconfirmed infections, except for loss of smell and taste.54,55 Similarly, Rouquette et al. found that COVID-19-like symptoms were associated with long-term depression and anxiety after illness, while seropositivity for SARS-CoV-2 was not. 216 Liu et al. found that perception of (subjective) cognitive deficit during acute COVID-19 was associated with later Long-COVID, suggesting “an affective component to Post-COVID-19-condition in some patients.” 217 Haddad et al. found that “the number of moderate or severe persistent symptoms reported by individuals (in both an exposed uninfected group and an infected group) was associated with the number of moderate or severe persistent symptoms reported by their household members [i.e., prolonged symptoms tended to cluster within families],” and that “parents who reported their own health status at T1 was worse or much worse than before the pandemic were around 3-times more likely to report that their child had symptoms that persisted until T2.” 23 The authors noted that in several other conditions, including chronic pain, CFS, and fatigue, “symptom measures in children are associated with parents’ symptoms, stress, and/or parenting behavior.” 23 Bertran et al. found that in both SARS-CoV-2 PCR-positive and PCR-negative adolescents, parents having Long-COVID increased the risk of the adolescent having Long-COVID by an adjusted OR 1.74 (95% CI 1.54, 2.01; absolute risk 10.7% higher) and an adjusted OR 1.92 (95% CI 1.50, 2.46; absolute risk 10.7% higher), respectively. 215 The authors hypothesized that parental Long-COVID “increases the focus of attention on symptoms and results in increased frequency of reporting by children of parents with ongoing COVID-19 problems.” 215 Sorg et al. found that clustered CFS symptoms or substantial fatigue among children and adolescents unaware of their previous seropositive infection status were no different from seronegative controls, although this was not the case when including those who were aware of their previous infection. 24 One group suggested that the pandemic has created a “perfect storm in which nocebo effects may be flourishing,” including a flood of negative information from the media, and the fear and anxiety of negative expectations. 218
The effect of vaccination on reducing the risk of long-COVID has been estimated at 15% or higher.72,219,220 Although some have hypothesized, in our view implausibly, that this reflects “autoimmune processing being ‘reset’ by vaccination . . . [or] any residual viral reservoir may be destroyed by antibody response,” one “cannot rule out the possibility of a change in reported symptoms after vaccination being due to a placebo effect [i.e., reversal of the nocebo effect].” 221 One study directly supported this theory by finding the efficacy of vaccination in preventing symptoms typical of Long-COVID to be the same as or better in test-negative controls compared to test-positive Omicron cases.70,222 Often, vaccination allows some freedom from social restrictions (e.g., being allowed to attend classes and participate in social interactions), which may improve symptoms of anxiety and depression. 223
Mechanisms of functional disorders
The BPCM for perception
Instead of passively awaiting sensory input, the brain is an active inferential machine. The brain hierarchically integrates sensory information (bottom-up input) and internal predictions about the expected information (top-down prior beliefs), each weighted according to their precision (mediated by top-down synaptic gain), to reach a posterior inference about what has happened (i.e., the percept).160–162,213,221,224–230 In Bayesian terms, this hierarchical process involves top-down prior predictions (represented by expectations), bottom-up likelihood (the Bayes Factor, represented by sensory input), and bottom-up prediction error (represented by the difference between the likelihood and prior, weighted by the precision of these two signals), to reach the posterior inference (represented by the percept) (see Figures 1 and 2).160–162,213,221,224–230 In this model, all experienced symptoms occur along a continuum of objectivity, more or less accurately representing what has happened to the body, and sometimes mistaking noise (normal bodily processes) for symptoms.224,226,229,230 In functional symptoms and syndromes, inaccurate inferences (so-called somatovisceral illusions or false perceptions) are made about the state of the body due to overly precise prior expectations overriding any bottom-up sensory data.160,162,213,221,224–226,228–230 In other words, the multi-network inferential functioning (i.e., functional connectivity) of the brain is (reversibly) abnormal, making “the brain’s best-guess [hypothesis] about the world [perception]” inaccurate, without structural pathological changes in either the body or the brain. 226 Very importantly, the functional disturbance at lower levels of symptom perception is unconscious and beyond naïve dualistic mind–body models.161,224,225,231 At higher levels, the functional disturbance involves too much confidence in prior predictions of concern, leading to the attenuation of disconfirming evidence by negative cognitive reappraisal. 231 This model can also explain placebo and nocebo effects that occur from abnormally precise prior expectations.213,225–227
Bayesian predictive coding model of symptom perception. The prior (expectations based on previous symptom experience episodes) is compared with the afferent sensory input (observation) leading to a prediction error. To minimize error, a posterior inference (symptom experience) is generated that best matches the prior and prediction error. This posterior inference then becomes the prior input in a new symptom perception episode. Moderators can cause less precise/accurate afferent sensory input/processing, which leaves more room for the prior expectation to determine perception (see Figure 2).
The Bayesian predictive coding model of inference by the brain. In panel (a) with a low precision prior, new information (observation) has a large impact on the formation of a posterior interpretation. In panel (b) with a high precision prior, new information (observation) has limited impact on the formation of a posterior interpretation. In panel (c) again with a high precision prior, low precision new information (observation) has little impact on the formation of a posterior interpretation (percept of a symptom).
How Bayesian predictive coding can lead to perceptual illusions
Several factors modulate symptom perception in this model (see Figures 1 and 2). First, the type of sensory input. When the input is less intense, more systemic and widespread, with poor on/off boundaries (e.g., fatigue or malaise), or when there is interoceptive dysfunction with low signal to noise (e.g., during chronic stress with cytokine and stress-axis activation), this decreases precision of sensory input, leaving more room for prior expectations to determine perception.224,225 Conversely, when input is associated with strong cues (e.g., prior viral infection symptoms or panic attacks), these cues can later activate strong prior expectations that override incoming sensory noise.221,224,225 Second, the focus of attention. Attention can be thought of as modulating the balance of precision weights of sensory input and prior expectations. Body focus can lead to a self-fulfilling cycle, with bodily scrutiny (especially in the context of cues suggesting a possible health threat, or recent illness, injury, or emotional arousal) and confirmation bias (making relatively weak everyday interoceptive stimulation represented as stronger and more precise, especially when there is also negative affect) leading to updated priors that are excessively precise (i.e., reinforcing the learning of stronger illness beliefs), until the priors are precise enough to have noise represented as signal (i.e., as posterior percept).213,221,225,229 Third, genetic influences. For example, females appear to be more sensitive to contextual cues that influence prior expectations. 225 Innate suggestibility has some genetic component and is increased under conditions of stress and trauma. 213 Fourth, trait anxiety, especially in the face of anticipated threat and negative affect (with harm avoidance, catastrophizing, and fear). Chronic worry and stress can create a vicious circle, with elevated threat and salience detection (e.g., activating priors that predict threat, and that are categorically precise), and leading to active inference (e.g., activation of the autonomic nervous system, endocrine and immune systems, to produce low-precision input that conforms to the prior predictions) that reinforces the prior belief.213,224–226,229 Fifth, top-down prior beliefs and expectations. Self-fulfilling health-related expectations can lead to pathologically precise prior beliefs. 213 This includes cultural expectations; for example, whiplash injury is rare in countries where the concept is not known, and a campaign to change expectations of consequences of minor injury led to a reduction in the population of chronic back pain. 221 Worrisome ideas (about cause, significance, and prognosis of symptoms) and expectations about illness can originate from the social context, including from health scares in the media or from physicians, or illness in family or friends, and can lead to selective body-focused monitoring.221,224,225,229
All of these factors can lead to three incorrect inferences made during hierarchical Bayesian brain processing213,221,224: (1) “Autonomous emergence of a percept or belief that is held with undue certainty (precision) following top-down attentional modulation of synaptic gain.” 221 (2) This percept “is falsely inferred to be a symptom to explain why its content was not predicted by the [higher level] source of attentional modulation.” 221 (3) Active inference, where interoactions lead to sensory input that conforms to predictions and reinforces the precision of abnormal prior expectations.213,224,227 In the end, “symptoms unfold increasingly independent of actual physiological changes over the course of somatic symptom and related disorders . . . [with] symptom report decoupled from sensory input.” 230 Cardinal features of these amplified symptoms include disproportionate findings of physical distress (e.g., exceptionally noxious and disruptive symptoms, and multiplicity of symptoms), negative cognitions (e.g., “unnecessarily negative expectations, unduly alarming suspicions, troubling interpretations, worrisome beliefs about their significance and cause,” including the conviction that an undiagnosed disease is present), health-related anxiety and disease fear, impairment of function (e.g., illness and sick-role behaviors, such as excessive use of medical care, information seeking, reassurance seeking, and avoidance of activity suspected of worsening symptoms), pervasiveness (e.g., preoccupied with symptoms that become part of self-identity), and dissatisfaction with medical care. 229
This unifying theoretical account strengthens the case that Long-COVID is usually a functional syndrome that was predictable given the common pandemic responses described below.
Implications for management of Long-COVID
Reduce modifiable predisposing factors for functional syndromes
This would entail reducing health anxiety (including worry and catastrophizing), depression, fear (including anticipated threat and unduly negative expectations), negative media coverage (including by medical “experts”), social isolation, and physical inactivity. We have several suggestions to consider.
First, provide accurate information about risk to reduce anxiety and fear. The median infection fatality rate (IFR) from SARS-CoV-2 infection, prior to vaccines, was 0.034% for those aged 0–59 years, and 0.095% for those aged 0–69 years. 232 The median IFR by age group was a median 0.0003% at 0–19 years, 0.002% at 20–29 years, 0.011% at 30–39 years, 0.035% at 40–49 years, 0.123% at 50–59 years, and 0.506% at 60–69 years. 232 For those <70 years, this is 0.33%/0.095% = 3.5× lower than the case fatality rate in Canada in March 2021; correcting for this difference between case and infection outcome rates, the infection hospitalization and ICU admission rates for those <70 years in Canada were 3.0%/3.5 = 0.86% and 0.7%/3.5 = 0.2%, respectively, in May 2021 233 (and with Omicron variants is now likely 3.5× lower).234,235 From a public health view, serious outcomes from SARS-CoV-2 are rare in these age groups, particularly in children. In adults aged >70 years living in the community, the median IFR before vaccines and Omicron variants was 2.2%, and focused protection, especially in those with multiple comorbidities, can be offered. 236 Explaining that Long-COVID is usually not due to irreversible tissue damage, can be expected to be reversible, and is not much more common after COVID-19 infection than in non-infected controls, also can reduce fear and negative expectations. Second, lockdowns have led to increased anxiety, depression, social isolation, and physical inactivity, yet did not reduce COVID-19 rates or mortality in the population.233,237–239 Although physical activity decreased markedly during lockdown, physically active people were less likely, when infected, to report prolonged symptoms (OR 0.24; 95% CI 0.10, 0.55). 240 Similarly, closing schools led to severe adverse effects on children’s mental health and learning, yet were not effective at reducing the COVID-19 burden.233,237,241–251 Not implementing lockdowns (including not closing schools), and explaining that they have a very negative cost–benefit ratio, is important.233,241 Third, negative media coverage must be replaced with accurate information. This requires repeated messaging by trusted leaders who explain the cost–benefit trade-offs of interventions and explain risk in the context of other risks we have always tolerated in order to have a democratic and free society.233,238 Fourth, ensuring that health leaders create surge capacity in hospitals for patients with and without COVID-19 during endemic respiratory viral seasons, instead of creating fear that hospitals are overwhelmed (and implementing cruel visitation policies), is also important. 233 In Canada, hospitals have often been well above capacity in prepandemic years,252–254 and we suggest that it would be better to fix this problem instead of diverting attention by inducing fear and anxiety.255,256 Finally, masking signals to others (and reinforces) a fear of viruses including SARS-CoV-2, and this affective problem “is a contagious one: fear spreads among the public, leading to intensification of risk management.” 257 The best evidence from before the pandemic,258–260 the community randomized trials during the pandemic,261–264 updated evidence after the pandemic, 265 and observational school masking studies during the pandemic266–268 support that universal masking (especially) outside of hospitals is not effective in reducing transmission. One way to signal that it is time to move away from fear may be to abandon talk of community mask mandates.
Reduce modifiable perpetuating factors for functional syndromes
This would entail reducing physical inactivity, social isolation, health anxiety (with unhelpful cognitions and fears), depression, social reinforcement, and contagion by (sometimes dramatic) exaggeration in conventional and social media, and late or missed diagnosis. We have several suggestions to consider.
First, the same suggestions discussed above to reduce predisposing factors will also reduce perpetuating factors. Second, de-escalating social reinforcement (and contagion). The focus from leaders, “experts,” social media influencers, and conventional media should be on providing accurate and therefore reassuring information about Long-COVID, without unduly exaggerated claims that perpetuate fear. Clear explanations (including by treating clinicians) should emphasize the functional nature of the condition (e.g., “a software not hardware problem” in perception), the lack of structural damage to the brain or organs, and the reversibility of the syndrome.158,160 Explanation should avoid problematic statements such as “all the tests were normal, so there is no disease,” “this is a psychiatric condition,” or “this is all in the mind”; these are not only inaccurate but also reduce trust and increase problematic cognitions.138,158,160 Third, at the individual level, therapy should be multidisciplinary (e.g., involving somatic rehabilitation clinics that existed prior to the pandemic), and focus on delivering and explaining the functional diagnosis, exercise and physical rehabilitation, and cognitive behavioral therapy.136,138,158,160,162 Follow-up is important to prevent the patient from feeling abandoned or ignored.136,138,158 Cognitive behavioral therapy aims to change maladaptive thoughts (e.g., symptom focusing, believing symptoms are a sign of damage, catastrophizing) and behaviors (e.g., avoidance of social interaction or physical activity). 127 Improving sleep hygiene, and treating comorbid anxiety and depression can also be helpful.136,140,162
Some important clarifications
First, the SARS-CoV-2 pandemic response did not follow previous pandemic plans nor the emergency management process, leading to the predisposing, precipitating, and perpetuating factors described above. What a better response would look like, led by emergency management experts following the emergency management process, is beyond the scope of this review, and is described elsewhere.233,256,269–271 Second, in functional syndromes, the symptoms and disability are real and genuinely experienced, and not feigned or faked. Third, clinicians must take the symptoms and disability seriously, and explain clearly that there is a diagnosis, this diagnosis is common, the mechanism is functional, and therefore the condition is potentially reversible and treatment can help. Fourth, we acknowledge that some Long-COVID cases will be due to pathological structural disease, and some investigations will be important and necessary to rule these out in individual cases according to physician judgment. This may be particularly true for the most severe cases of acute COVID-19, especially those with post-intensive-care syndrome. Indeed, in a population-based cohort study in Ontario, Canada, the risk of incident cardiovascular, neurological, and mental health conditions and rheumatoid arthritis more than 30 days after hospitalization for COVID-19 was comparable with other acute infectious illnesses (e.g., influenza and sepsis), suggesting these disorders “may be related to the severity of infectious illness necessitating hospitalization, rather than being direct consequences of infection with SARS-CoV-2.” 272 Fifth, although having COVID-19 viral illness can be a precipitating factor for functional syndromes, we believe this is far less common than people have been led to believe, and the more common precipitating factors may be psychosocial trauma and belief in the threat of Long-COVID (along with the dramatic public health and media response, creating stress and fear).
Sixth, our hypothesis avoids the accusation of “medical gaslighting,” by attempting to maintain epistemic humility and avoid ontological politics.11,57 Specifically, we explicitly suggest not dismissing patient descriptions of their symptoms; accepting patient reports as describing a genuine illness associated with significant suffering; providing supportive, empathetic, and timely medical diagnosis; and offering timely multidisciplinary treatment including CBT and graded exercise. We also argued that symptoms cannot be dismissed as simply the “product of anxiety” (e.g., don’t turn a predisposing factor into the “cause”) or “a mental problem” (e.g., all symptom perception depends on the same brain mechanisms). In addition, we offered a physiological explanation for the syndrome, that is, the brain’s Bayesian perceptual processing. Accordingly, it would be a mistake to think we suggest that Long-COVID is “only a psychological problem”; this would be a belief that mistakenly perpetuates mind–body dualism, and misunderstands the mechanisms of brain functioning and perception.
Limitations
First, this was not a systematic review, and we may have missed important studies contrary to our hypothesis. We refer to many systematic reviews to support our argument from analogy. These reviews consistently found that, although there are many hypothesized mechanisms, the symptoms of Long-COVID remained largely “medically unexplained” (Table 2). Second, there may be other pathophysiologic mechanisms for Long-COVID. We refer to hypotheses of organ damage, viral persistence, autoimmunity, and neuroinflammation, and argued that findings are inconsistent at best (Table 2) and that biomarkers of inflammation sometimes found in Long-COVID are nonspecific and of unclear cause–effect relationship.
This is not the view taken by many authors; however, those authors did not consider the mechanism we suggest, as demonstrated in a systematic review of functional neurologic symptoms in Long-COVID. 273 For example, CFS (also occurring in Long-COVID) has been thought due to mitochondrial pathology or abnormal inflammation, and CBT and graded exercise therapy have been suggested to worsen outcomes.86,274–276 Yet, recent systematic reviews of CFS found that “it is difficult to establish the role of mitochondria in the pathomechanisms of ME/CFS/SEID due to inconsistencies across studies,” 277 that “there are few consistent [immunological] findings and there is almost a complete lack of longitudinal studies,” 278 that the quality of studies for or against CBT or graded exercise therapy was often low, 276 and that CBT, graded exercise therapy, and pacing are effective therapies.279–282 The finding in large cohort studies that “patients with mild Covid-19 are at risk for a small number of health outcomes, most of which are resolved within a year from diagnosis,” 283 and that workers’ compensation claims have “fallen sharply over time” with “approximately 18% of claimants with Long Covid . . . unable to return to work for more than one year,” 284 seem contrary to views that Long-COVID is a long-term disease that does not respond to therapy. 285 The only randomized controlled trial (RCT) that we are aware of studying CBT for severe fatigue in Long-COVID, found that CBT was effective in reducing fatigue, with positive effect sustained at 6-month follow-up. 286 Similarly, the only RCT of exercise for long-COVID that we are aware of found that a supervised exercise intervention at low and moderate intensity was a “more effective, safe, and well-tolerated intervention in post-COVID-19 conditions” than usual care. 287
Third, our hypothesis has not been tested. Long-COVID studies have rarely considered functional diagnoses, nor systematically looked for positive features of functional disorders (e.g., inconsistency over time, modulation by attention, distractibility) including FCD. 273 Whether long-COVID can be prevented or treated in the ways we suggested requires future research that we believe is extremely important. Fourth, whether long-COVID is one or many different conditions is unclear, and whether all or even most are explained by our hypothesis can only be determined by future studies. We make this hypothesis so that it can be tested in future studies, thus aiming for scientific progress in explaining and treating Long-COVID.
Conclusion
We used an argument by analogy, reviewing what is known about Long-COVID, pandemic response effects on mental and physical health, functional syndromes including mass sociogenic illness, and the unifying mechanism among these (with common predisposing, precipitating, and perpetuating factors, Table 4), to offer an alternative perspective—that the majority of Long-COVID is a functional disorder. We discussed the implication that current pandemic response strategies have been causing predisposing, precipitating, and perpetuating factors for Long-COVID. Perhaps a better term for the syndrome than “Long-COVID” or “Long-Pandemic” would be “Pandemic-Response Syndrome,” to better reflect the etiologic factors we propose, and to serve as a lesson learned, not to be repeated in the future. Ultimately, we aimed to help the many people suffering from Pandemic-Response Syndrome.
Predisposing, precipitating, and perpetuating factors that are remarkably similar among Long-COVID and functional disorders, and how the Bayesian predictive coding model for perception can explain these factors.
For more details and many references, please see the text.
Footnotes
Acknowledgements
A previous version of this manuscript is available as a preprint at: SSRN: https://ssrn.com/abstract=4424007, 
.
Author Contributions
ARJ and AE made substantial contributions to the concept and design of the work, interpretation of the data, revised the article critically for important intellectual content, approved the version to be published, and participated sufficiently in the work to take public responsibility for appropriate portions of the content. ARJ wrote the first draft of the manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Ethics approval
Not applicable
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Informed consent
Not applicable