How COVID Could Screw You Worse With Each Reinfection

Authors: David Axe Tue, July 5, 2022

The more times you catch COVID, the sicker you’re likely to get with each reinfection. That’s the worrying conclusion of a new study drawing on data from the U.S. Veterans Administration.

Scientists stressed they need more data before they can say for sure whether, and why, COVID might get worse the second, third, or fourth time around. But with more and more people getting reinfected as the pandemic lurches toward its fourth year, the study hints at some of the possible long-term risks.

To get a handle on the health impact of reinfection, re-reinfection and even re-re-reinfection, three researchers—Ziyad Al-Aly from the Washington University School of Medicine plus Benjamin Bowe and Yan Xie, both from the V.A. St. Louis Health Care System—scrutinized the health records of 5.7 million American veterans.

Some 260,000 had caught COVID just once, and 40,000 had been reinfected at least one more time. The control group included 5.4 million people who never got COVID at all. Al-Aly, Bowe and Xie tracked health outcomes over a six-month period and came to a startling conclusion. “We show that, compared to people with first infection, reinfection contributes additional risks,” they wrote in their study, which hasn’t been peer-reviewed yet but is under consideration for publication in Nature.

Every time you catch COVID, your chance of getting really sick with somethinglikely COVID-related—seems to go up, Al-Aly, Bowe and Xie found. The risk of cardiovascular disorders, problems with blood-clotting, diabetes, fatigue, gastrointestinal and kidney disorders, mental health problems, musculoskeletal disorders and neurologic damage all increase with reinfection—this despite the antibodies that should result from repeat infections.

All of the conditions are directly associated with COVID or have been shown to get worse with COVID. “The constellation of findings show that reinfection adds non-trivial risks,” the researchers warned.

This risk could become a bigger deal as more people get reinfected. Globally, the death rate from COVID is going down, thanks in large part to growing population-wide immunity from past infection and vaccines.

But at the same time, non-fatal reinfections are piling up. Around half a billion people all over the world have caught COVID more than once, according to Al-Aly, Bowe and Xie’s study, citing data from the Johns Hopkins Coronavirus Resource Center. Many more reinfections, including “breakthrough” infections in the fully vaccinated, are likely as new variants and subvariants of COVID evolve to partially evade our antibodies.

The exact increase in risk from reinfection depends on the particular disorder in question—and whether you’ve been vaccinated and boosted. Broadly speaking, however, the likelihood of heart and clotting problems, fatigue and lung damage roughly doubles each time you catch COVID, Al-Aly, Bowe and Xie found.

Ali Mokdad, a professor of health metrics sciences at the University of Washington Institute for Health, offered one important caveat: time. “In general, one would expect that COVID will do more damage with a longer infection,” he told The Daily Beast. A short-lasting COVID infection followed by another short case of COVID should be less damaging than, say, back-to-back long illnesses.

The longer your infections drag on, the greater the stress on your organs. “These are two blows instead of one,” Mokdad said.

But it’s possible the worsening outcomes resulting from reinfection have little or nothing to do with the cumulative stress of successive long illnesses. According to Peter Hotez, an expert in vaccine development at Baylor College, the escalating risk could result from a poorly-understood phenomenon called “immune enhancement.”

A virus undergoes immune enhancement when a person’s immune system, after initial exposure to the pathogen, backfires during reinfection. Someone suffering immune enhancement with regards to a particular disease is likely to get sicker and sicker each time they’re exposed.

Immune enhancement could explain Al-Aly, Bow and Xie’s observation of escalating risk from COVID reinfection. “If the observation is true,” Hotez stressed. But it’s possible the observation is inaccurate. Hotez said he’s “not convinced that reinfection is actually more severe.”

Anthony Alberg, a University of South Carolina epidemiologist, told The Daily Beast he, too, is somewhat skeptical. Just how much more risk you might accumulate with each case of COVID is really hard to predict. And Al-Aly, Bow and Xie’s study is too cursory to totally settle the uncertainty all on its own.

The main problem, Alberg explained, is tied to a classic logical dilemma: causation versus correlation. Just because veterans got sicker with each COVID infection doesn’t necessarily mean COVID is definitely to blame, he pointed out. The vets in the study who came down with COVID more than once maybe tended to belong to groups with overall worse health outcomes whether or not they caught COVID twice, thrice or never.

The Massive Screwup That Could Let COVID Bypass Our Vaccines

“Compared with veterans who were infected once with SARS-CoV-2, those who were infected two times or more were more likely to be older [or] Black people, reside in long-term care, be immunocompromised, have anxiety, depression and dementia and to have had cerebrovascular disease, cardiovascular disease diabetes and lung disease,” Alberg said.

COVID, in other words, might be beside the point. It’s possible the worsening outcomes in Al-Aly, Bow and Xie’s study are due to the fact that the reinfected patients “were on average older and with much poorer health status than those with one infection,” Alberg said, “not because of having been infected more than once.”

Untangling causation and correlation in a study of this scale could be tricky. “More evidence [is] needed on this topic before definitive conclusions can be reached,” Alberg said.

In the meantime, it should be easy for us to mitigate the potential risk. Anyone who comes down with COVID a second time shouldn’t hesitate to take a course of paxlovid or some other antiviral drug that’s approved for the disease. “We should continue to focus on making sure people are aware of the benefits of early treatment,” Jeffrey Klausner, an infectious diseases expert at the University of Southern California Keck School of Medicine, told The Daily Beast.

Better yet, we could focus on developing “strategies for reinfection prevention,” Al-Aly, Bow and Xie wrote.

The top priority, of course, should be vaccinating the unvaccinated. Even the best COVID vaccines aren’t 100-percent effective at preventing infection or reinfection—and they’re getting somewhat worse as SARS-CoV-2 evolves for greater immune-escape.

But even with cleverer viral mutations, the jabs are still pretty effective. You can’t get sicker and sicker with reinfection… if you never get infected in the first place.

Lifestyle risk behaviors among adolescents: a two-year longitudinal study of the impact of the COVID-19 pandemic

Authors: Anne Gardner1, Jennifer Debenham1, Nicola Clare Newton1, Cath Chapman Fiona Elizabeth Wylie2, Bridie Osman1, Maree Teesson, Katrina Elizabeth Champion The British Medical Journal

Abstract

Objective To examine changes in the prevalence of six key chronic disease risk factors (the “Big 6”), from before (2019) to during (2021) the COVID-19 pandemic, among a large and geographically diverse sample of adolescents, and whether differences over time are associated with lockdown status and gender.

Design Prospective cohort study.

Setting Three Australian states (New South Wales, Queensland and Western Australia) spanning over 3000 km.

Participants 983 adolescents (baseline Mage=12.6, SD=0.5, 54.8% girl) drawn from the control group of the Health4Life Study.

Primary outcomes The prevalence of physical inactivity, poor diet (insufficient fruit and vegetable intake, high sugar-sweetened beverage intake, high discretionary food intake), poor sleep, excessive recreational screen time, alcohol use and tobacco use.

Results The prevalence of excessive recreational screen time (prevalence ratios (PR)=1.06, 95% CI=1.03 to 1.11), insufficient fruit intake (PR=1.50, 95% CI=1.26 to 1.79), and alcohol (PR=4.34, 95% CI=2.82 to 6.67) and tobacco use (PR=4.05 95% CI=1.86 to 8.84) increased over the 2-year period, with alcohol use increasing more among girls (PR=2.34, 95% CI=1.19 to 4.62). The prevalence of insufficient sleep declined across the full sample (PR=0.74, 95% CI=0.68 to 0.81); however, increased among girls (PR=1.24, 95% CI=1.10 to 1.41). The prevalence of high sugar-sweetened beverage (PR=0.61, 95% CI=0.64 to 0.83) and discretionary food consumption (PR=0.73, 95% CI=0.64 to 0.83) reduced among those subjected to stay-at-home orders, compared with those not in lockdown.

Conclusion Lifestyle risk behaviors, particularly excessive recreational screen time, poor diet, physical inactivity and poor sleep, are prevalent among adolescents. Young people must be supported to find ways to improve or maintain their health, regardless of the course of the pandemic. Targeted approaches to support groups that may be disproportionately impacted, such as adolescent girls, are needed.

Trial registration number Australian New Zealand Clinical Trials Registry (ACTRN12619000431123)

Data availability statement

Data are available upon reasonable request.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial.

Strengths and limitations of this study

  • A prospective cohort design was used to explore changes in a comprehensive set of health indicators among adolescents, from before (2019) to during (2021) the COVID-19 pandemic, and whether changes varied by gender and lockdown status.
  • The study included a large (n=983) and geographically diverse sample of adolescents across three Australian states (New South Wales, Queensland and Western Australia) spanning over 3000 km.
  • Limitations of the research include the reliance on self-report measures, and while the sample was diverse, it is not representative of the Australian population.

The global spread of COVID-19 and subsequent lockdown measures have presented challenges worldwide. While disease severity, hospital admissions and deaths have typically been lower among adolescents, compared with adults,1 government responses, such as movement restrictions and school closures, present further potential health ramifications due to the related changes in lifestyle behaviours. Critically, despite some studies demonstrating the significant physical and mental health consequences of lockdown measures on adolescents,2–4 research has typically focused on a few select behaviours, rather than a comprehensive set of health indicators. Given the unique presentation of COVID-19 across countries and differing government responses, there is a need to examine health-related changes from a variety of contexts to develop a better understanding of global health.

According to the Oxford COVID-19 Government Response Tracker,5 the strictness of lockdown restrictions since the first confirmed cases in January 2020 to October 2021 was similar in Australia, the USA and the UK, with average stringency indexes of 60/100, 59/100 and 61/100, respectively, despite much lower incidence and mortality rates in Australia.6 However, there can be substantial variation within countries.7 8 In Australia, for example, stringency index values varied between states and territories by as much as 68 during 2020.8 The strictest and most extensive lockdown restrictions have been implemented in Victoria and New South Wales (NSW), two of the most populous states that saw heightened case numbers during the January 2020 to October 2021 period, while other Australian states, such as Queensland (QLD) and Western Australia (WA), experienced far fewer cases and restrictions.9 10 The Australian context may therefore serve as a case study for understanding the impact of various levels of restrictions on adolescent health behaviours.

Previous research has highlighted the importance of six key lifestyle behaviours, including diet, physical activity, sleep, sedentary behaviour (including recreational screen time), alcohol use and smoking—collectively referred to as the ‘Big 6’—for the short-term and long-term health of adolescents.11–14 These behaviours are common among youth worldwide, with more than 80% of adolescents insufficiently physically active15 and screen time rapidly increasing.14 16 The Big 6 contribute significantly to global disease burden and are known predictors of chronic diseases, including cancer, cardiovascular disease and mental disorders.13 17

Research suggests that COVID-19 has impacted the Big 6, and in turn, the health of adolescents. For example, youth in Europe and Palestine have gained weight during the pandemic,18 19 which may be the result of increased consumption of discretionary food and sugar-sweetened beverages (SSB) during lockdown periods.18 20 However, some studies report improvements in dietary behaviours, including less SSB consumption among Colombian adolescents, higher fruit intake among Italian youth and higher vegetable intake among adolescents from Spain, Brazil and Chile.20 21 Among the few existing Australian studies, Munasinghe et al 3 found physical distancing measures implemented in the initial lockdown period (March–April 2020) were associated with a decline in fast food consumption among adolescents, but there were no changes in fruit and vegetable consumption. However, it is unknown whether these changes have been sustained, or whether other dietary behaviours changed.

The pandemic presents particular challenges for movement behaviours, including physical activity, sedentary behaviour and sleep. Typically, lockdowns are associated with lower levels of adolescent physical activity4 18 20 22 23 and increased screen time, both for remote learning and recreation, resulting in sedentary lifestyles.3 4 23 24 However, some research in Australia25 and Germany26 suggests physical activity increased.26 International studies also report an increase in adolescent sleep duration during lockdown periods,18 20 but higher prevalence of sleep problems, particularly among girls.27 Similarly, Australian adolescents perceived an increase in sleep difficulties and had increased sleep disturbance during the first lockdown.25 One study28 reported increased sleep duration among Australian adolescents who were engaged in remote learning; however, another3 found no changes.

Studies investigating the impact of the pandemic on adolescent alcohol and tobacco use have produced mixed findings. For example, alcohol use is reported to have increased among Canadian adolescents,29 reduced among Spanish adolescents,30 while there was no change in alcohol or tobacco use among adolescents from the USA.31 Further, European research suggests a reduction in adolescent tobacco use during the pandemic period,30 32 yet there has been an increase in Uganda.33 To date, changes in alcohol and tobacco use among Australian adolescents have not been examined.

Evidence suggests that the prevalence of the Big 6 varies by gender. For example, adolescent girls are more likely to be physically inactive, whereas adolescent boys are more likely to engage in high levels of recreational screen time, have a poor diet, and use alcohol and tobacco.15 34–38 However, less is known about whether changes in lifestyle behaviours over the pandemic period vary by gender.

To address these gaps in the literature, this study aims to examine changes in the prevalence of the Big 6 among a large, geographically diverse sample of adolescents, from before to during the COVID-19 pandemic, and explore whether differences over time are associated with gender and lockdown status.

Methods

Participants and procedure

The sample comprised participants from three Australian states (NSW, QLD and WA), spanning over 3000 km, who were randomly allocated to the control group of the Health4Life Study.39 Participants who provided written consent and had parental consent (passive, active written or active verbal, depending on approved procedures for the school type and region) completed self-report assessments in a supervised classroom setting. Only students who provided data prior to the beginning of the pandemic (between July and November 2019) and during the pandemic (approximately 24 months after baseline, between July and 10 October 2021) were included in this study. During the 2021 data collection period, Australia had strict border policies, restricting international travel and mandating hotel quarantine, while state-level and territory-level border policies for domestic travel varied.40 Greater Sydney, including the Central Coast, Shellharbour and Wollongong were subjected to lockdown restrictions under the NSW stay-at-home Public Health Order41; the most stringent of which included not being permitted to leave the home unless essential (eg, one person per household to shop for food or 1 hour of exercise per day), movement restricted to a 5 km radius of the home, closure of all non-essential retail (eg, hairdressers), home-based work and schooling requirements, curfews, and mandatory mask wearing, with a high police presence and large fines enforced for non-adherence. QLD, WA and areas outside of Greater Sydney were not subjected to extended stay-at-home lockdown restrictions.

Patient and public involvement

Patients or the public were not involved in the design, or conduct, or reporting, or dissemination plans of our research.

Measures

Sociodemographic characteristics

Participants self-reported their age and gender (male, female, non-binary/gender fluid, missing). A binary ‘lockdown’ variable was created reflecting participants who attended schools in the Greater Sydney region that were subjected to the stay-at-home Public Health Order in 2021 and those who were not.41

Diet

Dietary intake was assessed using items adapted from the NSW School Physical Activity and Nutrition Survey.42 Participants self-reported the number of metric cups of SSB usually consumed per week or day. A binary variable was created to reflect high (≥5–6 cups or more/week) and low consumption (≤4 cups/week). Participants reported how often they consume six discretionary food items (hot chips, French fries, wedges or fried potatoes; potato crisps or other salty snacks; snack foods, for example, sweet and savoury biscuits, cakes, doughnuts or muesli bars; confectionary; ice cream or ice blocks; and takeaway meals or snacks). High discretionary food consumption was defined as eating any of the items ‘2 or more times/day’, or eating at least two of the items ‘3–4 times/week’ or more often. Participants reported the number of serves of fruit and vegetables consumed per day, and in line with the Australian dietary guidelines,43 insufficient fruit and vegetable consumption was defined as <2 serves of fruit and <5 serves of vegetables per day, respectively.

Physical activity

A single item was used to assess the number of days over the past week that participants engaged in moderate-to-vigorous physical activity for at least 60 min.44 As per the Australian health guideline, insufficient physical activity was defined as engaging in <60 min of moderate-to-vigorous physical activity/day.45

Recreational screen time

The International Sedentary Assessment Tool46 was used to evaluate free time spent on a typical weekday and weekend day over the past 7 days watching television/DVDs/streaming services or using an electronic device. In line with the Australian health guideline,45 excessive recreational screen time was defined as >2 hours/day.

Sleep

The Modified Sleep Habits Survey47 was used to assess sleep duration. Total sleep time was calculated by finding the difference between the time participants reported first attempting sleep, and the time they woke up in the morning, minus the reported time taken to fall asleep from first attempt, with a weighted average sleep duration calculated for school and weekend nights. Self-reported bedtime, waketime and sleep duration have been shown to be reliable and valid in adolescent populations.48 49 As per the Australian guidelines, insufficient sleep was defined as an average duration outside of 9–11 hours/night for those aged 11–13 years, or 8–10 hours/night for those aged 14–17 years.45

Alcohol and tobacco use

Alcohol and tobacco use were measured using two dichotomous (yes/no) items drawn from previous large scale trials and population based epidemiological surveys50 51: ‘Have you had a full standard alcoholic drink in the past 6 months?’ and ‘In the past 6 months, have you tried cigarette smoking, even one or two puffs?’

Statistical analysis

Generalised linear mixed models were used to investigate change over time in the Big 6. Owing to the high prevalence of outcomes, we used Robust Poisson methods to generate prevalence ratios (PR) and 95% CIs, to overcome some of the limitations of reporting ORs from logistic regressions, which may appear inflated.52 PR are interpreted as the estimated prevalence of an outcome in one group, compared with another, providing an indication of a change in prevalence, as opposed to risk or odds. All models included a random intercept at the student level and school level, Robust Poisson distribution and a log link function, where time is continuous and represents the prepandemic (2019) and mid-pandemic scores (2021). Group by time interactions were estimated to assess change in the prevalence of the Big 6 over time in relation to gender (female/male, given the low prevalence of the ‘non-binary/gender fluid’ (0.1%) and ‘prefer not to say’ (.5%) subgroups) and the presence of lockdown restrictions during the 2021 survey occasion. All analyses were conducted in Stata V.17.53

Results

Descriptive statistics

The sample included 983 students (baseline Mage=12.6, SD=0.5, 54.8% girl) from 22 schools across NSW, QLD and WA (see table 1 for baseline characteristics). At the 2021 survey occasion, approximately one-third of the sample (32.7%) was under lockdown restrictions. Table 2 presents the prevalence of lifestyle risk behaviours over time.

Table 1

Sample characteristics

Table 2

Prevalence of lifestyle risk behaviours before and during the COVID-19 pandemic

Changes in lifestyle risk behaviours

Change over time in the prevalence of the Big 6 and differences based on lockdown status and gender are illustrated in figure 1, with PR and CIs detailed in online supplemental table 1.

Supplemental material

[bmjopen-2021-060309supp001.pdf]

Figure 1

Figure 1

Change over time in the prevalence of the Big 6 and differences based on lockdown status and gender. a≤13 years old: 9 to 11 hours/night, 14-17years: 8 to 10 hours/night. MVPA, moderate-to-vigorous physical activity; SSB, sugar-sweetened beverage.

Dietary behaviours

SSB consumption

There was no significant change in the prevalence of high SSB consumption over time (PR=0.83 95% CI=0.58 to 1.18). However, the prevalence was 39% lower in individuals under lockdown (PR=0.61, 95% CI=0.64 to 0.83) over time, compared with those not in lockdown.

Discretionary food consumption

There was no significant change in the prevalence of high discretionary food consumption over time (PR=0.97, 95% CI=0.86 to 1.09). However, the prevalence was 27% lower for individuals living under lockdown (PR=0.73, 95% CI=0.64 to 0.83) over time, compared with those not in lockdown.

Fruit and vegetable intake

The prevalence of insufficient fruit intake increased by 50% over time (PR=1.50, 95% CI=1.26 to 1.79). There were no changes in the prevalence of insufficient vegetable intake over time (PR=1.01, 95% CI=0.97 to 1.06), and the presence of lockdown restrictions was not associated with a change in the prevalence of insufficient fruit or vegetable intake over time.

There were no gender-based differences in the prevalence of high SSB consumption, high discretionary food consumption or insufficient fruit/vegetable intake over time.

Sleep

The prevalence of insufficient sleep decreased by 26% over time (PR=0.74 95% CI=0.68 to 0.81). Girls reported a higher prevalence of insufficient sleep over time, compared with boys (PR=1.24, 95% CI=1.10 to 1.41). The presence of lockdown restrictions was not associated with a change in the prevalence of insufficient sleep over time.

Recreational screen time

There was a 6% increase in the prevalence of excessive recreational screen time over time (PR=1.06, 95% CI=1.03 to 1.11). Gender and the presence of lockdown restrictions were not associated with a change in the prevalence of excessive recreational screen time over time.

Physical activity

There was no change in the prevalence of insufficient physical activity over time (PR=1.03, 95% CI=1.00 to 1.07). Neither gender nor the presence of lockdown restrictions was associated with change in the prevalence of insufficient physical activity over time.

Alcohol use

The prevalence of past 6-month alcohol use increased by 334% over time (PR=4.34, 95% CI=2.82 to 6.67). The prevalence of alcohol use increased more in girls compared with boys (PR=2.34, 95% CI=1.19 to 4.62). The presence of lockdown restrictions was not associated with change in the prevalence of past 6-month alcohol use over time.

Tobacco use

The prevalence of past 6-month tobacco use increased by 305% over time (PR=4.05 95% CI=1.86 to 8.84). Neither gender nor the presence of lockdown restrictions was associated with change in the prevalence of past 6-month tobacco use over time.

Discussion

This study was the first to explore changes in all of the Big 6 lifestyle risk behaviours among a large, geographically diverse cohort of adolescents, from before (2019) to during (2021) the COVID-19 pandemic, and whether changes varied by gender and lockdown status. Over the 2-year period, the prevalence of excessive recreational screen time, insufficient fruit intake and alcohol and tobacco use increased, with alcohol use increasing among girls in particular. The prevalence of insufficient sleep reduced in the overall sample; yet, increased among girls. Being in lockdown was associated with improvements in SSB consumption and discretionary food intake.

These findings highlight the varied impact of the pandemic across countries. For example, consistent with other Australian findings,3 but in contrast to international research,18 20 the prevalence of discretionary food intake decreased among those in lockdown. Yet in line with some international findings,21 SSB intake reduced among adolescents in lockdown. This may reflect increased parental monitoring during lockdown and reduced opportunistic exposure to fast food due to not being with friends or commuting to school.54 55 As such, continued parental monitoring beyond the lockdown period and the promotion of healthy food options may be beneficial. However, improvements in healthy dietary behaviours were not observed. In fact, the prevalence of insufficient fruit intake increased among the full sample. This may relate to the higher cost of fresh fruit and vegetables in Australia during the pandemic, caused by labour shortages within the farming, wholesale and retail sectors due to fewer working holiday-makers.56 These findings support calls for governments to consider broader policy-level changes to improve diet, such as taxes and subsidies.57

The finding that sleep duration improved from before to during the pandemic is consistent with some Australian28 and international18 20 studies. This contrasts typical trends over adolescence58 59 and was despite an increase in the prevalence of excessive recreational screen time, which is often considered a primary contributor to poor sleep.60 It is posited that the time usually spent getting ready and commuting to school is instead spent getting additional sleep during periods of lockdown, leading to calls for delayed school start times28; however, we found no differences based on lockdown status to support this. The finding that insufficient sleep increased among girls is consistent with international research reporting increased sleep disorders among girls during the pandemic27 and may reflect the association between girl pubertal maturation and the emergence of insomnia symptoms.61 Targeted intervention approaches to address sleep among girls are needed.

Notably, in contrast to previous international and Australian research attributing increased screen time to lockdown and physical distancing measures,3 4 23 we found no difference in the prevalence of excessive recreational screen time between the lockdown and non-lockdown groups. This increase may instead reflect general trends of increasing screen time among adolescents.16 These findings highlight the value of assessing behaviours among adolescents both in lockdown and not in lockdown in the same period for comparability, and the need for effective interventions targeting screen time among adolescents.39

Contrasting typical trends over adolescence and previous pandemic research,4 23 the prevalence of insufficient physical activity did not change over time, nor did it increase more for those in lockdown. Previous studies have attributed reductions in physical activity during the pandemic to government responses, such as the cancellation of organised sport and closure of gyms and recreation centres.1 23 However, given the current data were from the 2021 lockdown, whereas previous studies focused on the initial lockdown in 2020, it may be that over time, adolescents have learnt to adapt to the rapidly changing situation and find other ways to achieve their physical activity goals (eg, replaced organised sport with outdoor gym sessions and training). It may also be that other forms of physical activity, such as light and incidental physical activity, have been more severely impacted. Future research would benefit from assessing how these different forms of physical activities changed throughout the pandemic.

Finally, although alcohol and tobacco use increased over time, the prevalence of these behaviours at the first timepoint, when participants were aged 12, was very low and remained relatively low 24 months later. This increase is to be expected among adolescents62; however, the greater increase in alcohol use among girls was unexpected. Considering this and the increase in prevalence of insufficient sleep, girls may be disproportionately impacted by the pandemic. This may reflect general patterns of higher prevalence and increasing trends of mental health problems among adolescent girls across the globe,63 64 which are often comorbid with poor sleep and substance use; as well as narrowing of the gender gap in alcohol use among more among recent cohorts.35 65 Links between these factors are complex66 and assessing changes in mental health alongside changes in the Big 6 may be a useful future research direction.

Key strengths of this study include having assessment occasions before and during the pandemic, rather than relying on retrospective accounts of perceived changes in behaviours, and a sample comprised of adolescents both in and not in lockdown at follow-up for comparability. However, we cannot rule out the potential impact of other factors, such as maturation or mental health, that could also be influencing the Big 6. Although the study builds on previous research that has focused on the early pandemic period, claims about behavioural shifts across the early and late pandemic periods need to be interpreted with caution. Other limitations include the reliance on self-report measures, and while the sample was more diverse than other Australian studies, it is limited to three Australian states and is therefore not representative of the entire Australian adolescent population.14

Conclusion

Lifestyle risk behaviours, particularly excessive recreational screen time, poor diet, physical inactivity and poor sleep, are prevalent among adolescents and should be addressed with effective behavior change interventions.39 With the pandemic remaining a continually evolving situation across the world, the impact on health behaviors is also likely to be dynamic and diverse. Supporting young people to improve or maintain their health behaviours, regardless of the course of the pandemic, is important, alongside targeted research and intervention efforts to support groups that may be disproportionately impacted, such as adolescent girls.

References

  1. MCRI COVID-19 Governance Group. COVID-19 and child and adolescent health. Victoria, Australia, 2021.Google Scholar
  2. Caroppo E , Mazza M , Sannella A , et al . Will nothing be the same again?: changes in lifestyle during COVID-19 pandemic and consequences on mental health. Int J Environ Res Public Health 2021;18:8433.doi:10.3390/ijerph18168433 Google Scholar
  3. Munasinghe S , Sperandei S , Freebairn L , et al . The impact of physical distancing policies during the COVID-19 pandemic on health and well-being among Australian adolescents. J Adolesc Health 2020;67:65361.doi:10.1016/j.jadohealth.2020.08.008 pmid:http://www.ncbi.nlm.nih.gov/pubmed/33099413 CrossRefPubMedGoogle Scholar
  4. Li SH , Beames JR , Newby JM , et al . The impact of COVID-19 on the lives and mental health of Australian adolescents. Eur Child Adolesc Psychiatry 2021.doi:doi:10.1007/s00787-021-01790-x. [Epub ahead of print: 28 Apr 2021].pmid:http://www.ncbi.nlm.nih.gov/pubmed/33913036 Google Scholar
  5. Hale T , Angrist N , Goldszmidt R , et al . A global panel database of pandemic policies (Oxford COVID-19 government response Tracker). Nat Hum Behav 2021;5:529–38.doi:10.1038/s41562-021-01079-8 PubMedGoogle Scholar
  6. Dong E , Du H , Gardner L . An interactive web-based dashboard to track COVID-19 in real time. Lancet Infect Dis 2020;20:533–4.doi:10.1016/S1473-3099(20)30120-1 pmid:http://www.ncbi.nlm.nih.gov/pubmed/32087114 CrossRefPubMedGoogle Scholar
  7. Hallas L , Hatibie A , Majumdar S . Variation in US states’ responses to COVID-19 2.0. Blavatnik School of Government Working Paper, 2020.Google Scholar
  8. Australian Bureau of Statistics. State economies and the stringency of COVID-19 containment measures, December quarter 2020. Canberra,Australia: ABS, 2021.Google Scholar
  9. COVID-19-AU. COVID-19 in Australia real-time report, 2022. Available: https://covid-19-au.com/ [Accessed 28 Feb 2022].Google Scholar
  10. Australian Bureau of Statistics. National, state and territorypopulation. Canberra: ABS, 2021.Google Scholar
  11. Ding D , Rogers K , van der Ploeg H , et al . Traditional and emerging lifestyle risk behaviors and all-cause mortality in middle-aged and older adults: evidence from a large population-based Australian cohort. PLoSMed 2015;12:e1001917.doi:10.1371/journal.pmed.1001917 pmid:http://www.ncbi.nlm.nih.gov/pubmed/26645683 PubMedGoogle Scholar
  12. Lynch BM , Owen N . Too much sitting and chronic disease risk: steps to move the science forward. Ann Intern Med 2015;162:146–7.doi:10.7326/M14-2552 pmid:http://www.ncbi.nlm.nih.gov/pubmed/25599352 CrossRefPubMedGoogle Scholar
  13. Ezzati M , Riboli E . Behavioral and dietary risk factors for noncommunicable diseases. N Engl J Med Overseas Ed 2013;369:95464.doi:10.1056/NEJMra1203528 Google Scholar
  14. Champion KE , Chapman C , Gardner LA , et al . Lifestyle risks for chronic disease among Australian adolescents: a cross-sectional survey. Med J Aust 2022;216:156–7.doi:10.5694/mja2.51333 pmid:http://www.ncbi.nlm.nih.gov/pubmed/34747039 PubMedGoogle Scholar
  15. Guthold R ,Stevens GA , Riley LM , et al . Global trends in insufficient physical activity among adolescents: a pooled analysis of 298 population-based surveys with 1·6 million participants. Lancet Child Adolesc Health 2020;4:23–35.doi:10.1016/S2352-4642(19)303232 pmid:http://www.ncbi.nlm.nih.gov/pubmed/31761562 PubMedGoogle Scholar
  16. Yang L ,Cao C , Kantor ED , et al . Trends in sedentary behavior among the US population, 20012016. JAMA 2019;321:158797. doi:10.1001/jama.2019.3636  pmid:http://www.ncbi.nlm.nih.gov/pubmed/31012934 PubMedGoogle Scholar
  17. Australian Institute of Health and Welfare. Australia’s health 2018. Australia’s health series no. 16 AUS 221. Canberra: AIHW, 2018.Google Scholar
  18. Allabadi H , Dabis J , Aghabekian V . Impact of COVID-19 lockdown on dietary and lifestyle behaviours among adolescents in Palestine. Dyn Hum Health 2020;2020: 7.Google Scholar
  19. Stavridou A , Kapsali E , Panagouli E , et al . Obesity in children and adolescents during COVID-19 pandemic. Children 2021;8:135. doi:10.3390/children8020135  pmid:http://www.ncbi.nlm.nih.gov/pubmed/33673078 CrossRefPubMedGoogle Scholar
  20. Pietrobelli A , Pecoraro L , Ferruzzi A , et al . Effects of COVID-19 Lockdown on lifestyle behaviors in children with obesity living in Verona, Italy: a longitudinal study. Obesity 2020;28:1382–5.doi:10.1002/oby.22861  pmid:http://www.ncbi.nlm. nih.gov/pubmed/32352652 PubMedGoogle Scholar
  21. Ruiz-Roso MB , de Carvalho Padilha P , Mantilla-Escalante DC , et al . Covid-19 Confinement and Changes of Adolescent’s Dietary Trends in Italy, Spain, Chile, Colombia and Brazil. Nutrients 2020;12:1807.doi:10.3390/nu12061807 Google Scholar
  22. Bates L , Zieff G , Stanford K , et al . COVID-19 impact on behaviors across the 24-hour day in children and adolescents: physical activity, sedentary behavior, and sleep. Children 2020;7:138.doi:10.3390/children7090138 pmid:http://www.ncbi.nlm.nih.gov/pubmed/32947805 PubMedGoogle Scholar
  23. Reece LJ , Owen K , Foley B , et al . Understanding the impact of COVID-19 on children’s physical activity levels in NSW, Australia. Health Promot J Austr 2021;32:365–6.doi:10.1002/hpja.436 pmid:http: //www.ncbi.nlm.nih.gov/ pubmed/33201543 CrossRefPubMedGoogle Scholar
  24. Lu C , Chi X , Liang K , et al . Moving more and sitting less as healthy lifestyle behaviors are protective factors for insomnia, depression, and anxiety among adolescents during the COVID-19 pandemic. Psychol Res Behav Manag 2020;13:1223–33.doi:10.2147/PRBM.S284103 pmid:http://www.ncbi.nlm.nih.gov/pubmed/33364864 PubMedGoogle Scholar
  25. Olive L , Sciberras E , Berkowitz TS . Child and parent physical activity, sleep and screen time during COVID-19 compared to pre-pandemic nationally representative data and associations with mental health 2020.Google Scholar
  26. Schmidt SCE , Anedda B , Burchartz A , et al . Physical activity and screen time of children and adolescents before and during the COVID-19 lockdown in Germany: a natural experiment. Sci Rep 2020;10:21780.doi:10.1038/s41598-020-78438-4 pmid:http://www.ncbi.nlm.nih.gov/pubmed/33311526 PubMedGoogle Scholar
  27. Fidancı İzzet , Aksoy H , Yengil Taci D , et al . Evaluation of the effect of the COVID-19 pandemic on sleep disorders and nutrition in children. Int J Clin Pract 2021;75:e14170.doi:10.1111/ijcp.14170 pmid:http://www.ncbi.nlm.nih.gov/pubmed/33759311 PubMedGoogle Scholar
  28. Stone JE , Phillips AJK , Chachos E , et al . In-person vs home schooling during the COVID-19 pandemic: differences in sleep, circadian timing, and mood in early adolescence. J Pineal Res 2021;71:e12757.doi:10.1111/jpi.12757  pmid:http://www.ncbi.nlm.nih.gov/pubmed/34273194 PubMedGoogle Scholar
  29. Dumas TM , Ellis W , Litt DM . What does adolescent substance use look like during the COVID-19 pandemic? Examining changes in frequency, social contexts, and Pandemic-Related predictors. J AdolescHealth 2020;67: 3541.doi:10.1016/j. jadohealth. 2020.06.018 pmid:http: //www.ncbi.nlm. nih.gov/pubmed/32693983 PubMedGoogle Scholar
  30. Rogés J ,Bosque-Prous M , Colom J , et al . Consumption of alcohol, cannabis, and tobacco in a cohort of adolescents before and during COVID-19 confinement. Int J Environ Res Public Health 2021;18:7849 .doi:10.3390/ijerph18157849  pmid:http://www.ncbi.nlm.nih.gov/pubmed/34360141 PubMedGoogle Scholar
  31. Chaffee BW , Cheng J , Couch ET , et al . Adolescents’ substance use and physical activity before and during the COVID-19 pandemic. JAMA Pediatr 2021;175:715–22.doi:10.1001/jamapediatrics.2021.0541 pmid:http://www.ncbi.nlm.nih.gov/pubmed/33938922 PubMedGoogle Scholar
  32. Benschop A , van Bakkum F , Noijen J . Changing patterns of substance use during the coronavirus pandemic: self-reported use of tobacco, alcohol, cannabis, and other drugs. Front Psychiatry 2021;12:633551. doi:10.3389/fpsyt.2021.633551  pmid:http://www.ncbi.nlm.nih.gov/pubmed/34122170 PubMedGoogle Scholar
  33. Matovu JKB , Kabwama SN , Ssekamatte T , et al . COVID-19 awareness, adoption of COVID-19 preventive measures, and effects of COVID-19 Lockdown among adolescent boys and young men in Kampala, Uganda. J Community Health  2021;46:842–53.doi:10.1007/s10900-021-00961-w pmid:http://www. ncbi.nlm.nih.gov/pubmed/33481156 PubMedGoogle Scholar
  34. Active Healthy Kids Australia. Physical literacy: do our kids have all the tools? the 2016 active healthy kids Australia report card on physical activity for children and young people. Adelaide, South Australia: Active Healthy Kids Australia, 2016.Google Scholar
  35. Inchley J , Currie D , Vieno A . Adolescent alcohol-related behaviours: trends and inequalities in the who European region, 2002–2014: observations from the health behaviour in school-aged children (HBSC) who Collaborative cross-national study. viii. Copenhagen: World Health organization, Regional Office for Europe, 2018: + 83 p.Google Scholar
  36. Bucksch J ,Sigmundova D , Hamrik Z , et al . International trends in adolescent Screen-Time behaviors from 2002 to 2010. J Adolesc Health 2016;58:417–25.doi:10.1016/j.jadohealth.2015.11.014 pmid:http://www.ncbi.nlm.nih.gov/pubmed/26827267 PubMedGoogle Scholar
  37. Centers for Disease Control and Prevention (CDC). 1991-2019 high school youth risk behavior survey data, 2021.Google Scholar
  38. Morley B , Scully M , Niven P . National secondary students’ diet and activity (NaSSDA) survey, 2012-13. Melbourne: Cancer Council Victoria, 2014.Google Scholar
  39. Teesson M ,Champion KE , Newton NC , et al . Study protocol of the Health4Life initiative: a cluster randomised controlled trial of an eHealth school-based program targeting multiple lifestyle risk behaviours among young Australians. BMJ Open 2020;10:e035662.doi:10.1136/bmjopen-2019035662 pmid:http:// www.ncbi.nlm.nih.gov/pubmed/32665344 Abstract/FREE Full TextGoogle Scholar
  40. Bower M , Smout S , Ellsmore S . COVID-19 and Australia’s mental health: An overview of academic literature, policy documents, lived experience accounts, media and community reports. Sydney, NSW: Australia’s Mental Health Think Tank, 2021.Google Scholar
  41. NSW Health. Public health (COVID-19 temporary movement and gathering restrictions) order 2021 under the public health act 2010. NSW, Australia, 2021.Google Scholar
  42. Hardy LL , Mihrshahi S , Drayton BA . Nsw schools physical activity and nutrition survey (spans) 2015: full report. Sydney: NSW Department of Health, 2016.Google Scholar
  43. National Health and Medical Research Council. Australian dietary guidelines. Canberra: National Health and Medical Research Council, 2013.Google Scholar
  44. Active Healthy Kids Australia. Physical literacy: do our kids have all the tools? the 2016 report card on physical activity for children and young people. Adelaide, 2016.Google Scholar
  45. The Australian Government Department of Health. Australian 24-hour movement guidelines for children and young people (5 to 17 years): an integration of physical activity, sedentary behaviour, and sleep. Canberra: Commonwealth of Australia, 2019.Google Scholar
  46. Prince SA , LeBlanc AG , Colley RC , et al . Measurement of sedentary behaviour in population health surveys: a review and recommendations. PeerJ 2017;5: e4130.doi:10.7717/peerj.4130 pmid:http://www.ncbi.nlm.nih.gov/pubmed/29250468 CrossRefPubMedGoogle Scholar
  47. Short MA , Gradisar M , Lack LC , et al . Estimating adolescent sleep patterns: parent reports versus adolescent self-report surveys, sleep diaries, and actigraphy. Nat Sci Sleep 2013;5:23–6.doi:10.2147/NSS.S38369 pmid:http: //www.ncbi.nlm.nih.gov /pubmed/23620690 PubMedGoogle Scholar
  48. Golley RK , Maher CA , Matricciani L , et al . Sleep duration or bedtime? exploring the association between sleep timing behaviour, diet and BMI in children and adolescents. Int J Obes 2013;37:546–51.doi:10.1038/ijo.2012.212 pmid:http: //www.ncbi.nlm.nih.gov /pubmed/23295498 CrossRefPubMedGoogle Scholar
  49. Nascimento-Ferreira MV , Collese TS , de Moraes ACF , et al . Validity and reliability of sleep time questionnaires in children and adolescents: a systematic review and meta-analysis. Sleep Med Rev 2016;30:85–96.doi:10.1016/j.smrv.2015.11.006  pmid:http://www.ncbi.nlm.nih.gov/pubmed/26921735 PubMedGoogle Scholar
  50. CDC CfDCaP. National youth risk behavior survey. U.S. Department of Health and Human Services, 2019.Google Scholar
  51. Australian Institute of Health and Welfare. National drug strategy household survey 2016: detailed findings. Canberra: AIHW, 2017.Google Scholar
  52. Deddens JA , Petersen MR . Approaches for estimating prevalence ratios. Occup Environ Med 2008;65:501–6.doi:10.1136/oem.2007.034777 pmid: http://www.ncbi.nlm.nih.gov/pubmed/18562687 FREE Full TextGoogle Scholar
  53. StataCorp LLC. Stata Statistical Software: Release 17 [program]. College Station: TX: StataCorp LLC, 2021.Google Scholar
  54. Scully M , Morley B , Niven P , et al . Factors associated with frequent consumption of fast food among Australian secondary school students. Public Health Nutr 2020;23:1340–9.doi:10.1017/S1368980019004208 pmid :http://www.ncbi.nlm.nih.gov/pubmed/32172726 PubMedGoogle Scholar
  55. Haszard JJ , Skidmore PML , Williams SM , et al . Associations between parental feeding practices, problem food behaviours and dietary intake in New Zealand overweight children aged 4-8 years. Public Health Nutr 2015;18:1036–43.doi:10.1017/S1368980014001256 pmid:http://www.ncbi.nlm.nih.gov/pubmed/24955942 PubMedGoogle Scholar
  56. ABARES. Agricultural commodities: September quarter 2021. Canberra: Australian Bureau of Agricultural and Resource Economics and Sciences, 2021.Google Scholar
  57. Cobiac LJ , Tam K , Veerman L , et al . Taxes and subsidies for improving diet and population health in Australia: a cost-effectiveness modelling study. PLoS Med 2017;14:e1002232.doi:10.1371/journal.pmed.1002232 pmid:http://www.ncbi.nlm.nih.gov/pubmed/28196089 PubMedGoogle Scholar
  58. Keyes KM , Maslowsky J , Hamilton A , et al . The great sleep recession: changes in sleep duration among US adolescents, 19912012. Pediatrics 2015;135:460.doi:10.1542/peds.2014-2707 pmid:http://www.ncbi.nlm.nih.gov/pubmed/25687142 Abstract/FREE Full TextGoogle Scholar
  59. Olds T , Maher C , Blunden S , et al . Normative data on the sleep habits of Australian children andadolescents. Sleep 2010;33:13818.doi:10.1093/sleep/33.10.1381 pmid:http: //www.ncbi.nlm.nih.gov/pubmed/21061861 PubMedWeb of ScienceGoogle Scholar
  60. Xu F , Adams SK , Cohen SA , et al . Relationship between physical activity, screen time, and sleep quantity and quality in US adolescents aged 16–19. Int J Environ Res Public Health 2019;16:1524.doi:10.3390/ijerph16091524 Google Scholar
  61. Zhang J , Chan NY , Lam SP , et al . Emergence of sex differences in insomnia symptoms in adolescents: a large-scale school-based study. Sleep 2016;39:1563–70.doi:10.5665/sleep.6022 pmid:http://www.ncbi.nlm.nih.gov/pubmed/27091537 PubMedGoogle Scholar
  62. Guerin N , White V . ASSAD 2017 Statistics & Trends: Australian Secondary Students’ Use of Tobacco, Alcohol, Over-the-counter Drugs, and Illicit Substances. Cancer Council Victoria, 2018.Google Scholar
  63. Campbell OLK , Bann D , Patalay P . The gender gap in adolescent mental health: a cross-national investigation of 566,829 adolescents across 73 countries. SSM Popul Health 2021;13:100742.doi:10.1016/j.ssmph.2021.100742 pmid:http://www.ncbi.nlm.nih.gov/pubmed/33748389 PubMedGoogle Scholar
  64. Högberg B , Strandh M , Hagquist C . Gender and secular trends in adolescent mental health over 24 years – The role of school-related stress. Soc Sci Med 2020;250:112890.doi:10.1016/j.socscimed.2020.112890 pmid:http://www.ncbi.nlm.nih.gov/pubmed/32143086 PubMedGoogle Scholar
  65. Slade T , Chapman C , Swift W , et al . Birth cohort trends in the global epidemiology of alcohol use and alcohol-related harms in men and women: systematic review and metaregression. BMJ Open 2016;6:e011827.doi:10.1136/bmjopen-2016-011827 pmid:http://www.ncbi.nlm.nih.gov/pubmed/27797998 CrossRefPubMedGoogle Scholar
  66. Patalay P , Gage SH . Changes in millennial adolescent mental health and health-related behaviours over 10 years: a population cohort comparison study. Int J Epidemiol 2019;48:165064.doi:10.1093/ije/dyz006 pmid:http://www.ncbi.nlm.nih.gov/pubmed/30815691 CrossRefPubMedGoogle Scholar

Risk of new‐onset psychiatric sequelae of COVID‐19 in the early and late post‐acute phase

Authors: Ben Coleman, 1 , 2 Elena Casiraghi, 3 , 4 Hannah Blau, 1 Lauren Chan, 5 Melissa A. Haendel, 6 Bryan Laraway, 6 Tiffany J. Callahan, 6 Rachel R. Deer, 7 Kenneth J. Wilkins, 8 Justin Reese, 9 and Peter N. Robinson 1 , 2 World Psychiatry. 2022 Jun; 21(2): 319–320 2022 May 7. doi: 10.1002/wps.20992 PMCID: PMC9077621 PMID: 35524622

Recent publications have documented that a proportion of COVID‐19 patients develop psychiatric symptoms during or after acute infection 1 . We investigated this risk in the context of the National COVID Cohort Collaborative (N3C) – a centralized, harmonized, high‐granularity electronic health record (EHR) repository 2 – using the largest retrospective cohort reported to date.

Two previous large‐scale EHR studies examined psychiatric sequelae 90 and 180 days after COVID‐19 diagnosis. A cohort of 44,779 individuals with COVID‐19 was propensity score‐matched to control cohorts with conditions such as influenza and other respiratory tract infections (RTI). In the 90 days following the initial presentation, the incidence proportion of new‐onset psychiatric conditions was 5.8% in the COVID‐19 group vs. 2.5% to 3.4% in the control groups 3 . A follow‐up study also included individuals with a prior history of mental illness and similarly showed an increased risk of psychiatric conditions in the six months following initial presentation 4 .

To validate these findings, we leveraged data from N3C, which at our cutoff date of October 20, 2021 had 1,834,913 COVID‐19 positive patients and 5,006,352 comparable controls. Our data set was drawn from 51 distinct clinical organizations. We included patients in the COVID‐19 cohort if they had a confirmed diagnosis of SARS‐CoV‐2 infection by polymerase chain reaction or antigen test after January 1, 2020. Controls were selected from patients with a diagnosis of a RTI other than COVID‐19. We excluded from this analysis patients with a history of any mental illness prior to 21 days after COVID‐19 diagnosis, as well as patients without a medical record extending back a year prior to COVID‐19. There were 245,027 COVID‐19 positive individuals available for propensity matching.

Each COVID‐19 patient was matched with a control patient from the same institution whose age differed by no more than 5 years. Propensity score matching was done on 34 factors using a logistic regression model including main effect terms, resulting in 46,610 matched patient pairs. Multivariable Cox regression was performed to compare the incidence of new‐onset mental illness for all psychiatric conditions, mood disorders and anxiety disorders for 21 to 365 days following initial presentation. We additionally considered dyspnea as a positive control.

We tested the Cox regression proportional hazard assumption for comparisons of COVID‐19 patients and controls 5 . Schoenfeld residual analysis yielded a significant p‐value and led us to reject the null hypothesis of a constant proportional hazard over the full time period of 21‐365 days. We therefore separated the cohort into two time intervals (before and after 120 days) in which the proportional hazard assumption was not violated.

We identified a statistically significant difference in the hazard rate of new‐onset psychiatric sequelae between COVID‐19 and RTI in the early post‐acute phase (from 21 to 120 days), but not in the late post‐acute phase (from 121 to 365 days). The estimated incidence proportion (as modeled on the log‐hazard scale over time) of a new‐onset psychiatric diagnosis in the early post‐acute phase for the COVID‐19 group was 3.8% (95% CI: 3.6‐4.0), significantly higher than the 3.0% (95% CI: 2.8‐3.2) for the RTI group, with a hazard ratio (HR) of 1.3 (95% CI: 1.2‐1.4). The HR for new‐onset mental illness in the late post‐acute phase was not significant in the COVID‐19 compared to the RTI group (HR: 1.0; 95% CI: 0.97‐1.1).

Similar findings were obtained for anxiety disorders, but not for mood disorders. The estimated incidence proportion of a new‐onset anxiety disorder diagnosis was significantly increased for COVID‐19 patients (2.0%; 95% CI: 1.8‐2.1) compared to RTI patients (1.6%; 95% CI: 1.5‐1.7) in the early post‐acute phase (HR: 1.3; 95% CI: 1.1‐1.4). However, the estimated incidence proportion of a new‐onset mood disorder diagnosis in the same period was not significantly increased for COVID‐19 patients (1.2%; 95% CI: 1.1‐1.3) in comparison to RTI patients (1.1%; 95% CI: 1.0‐1.2).

New‐onset anxiety and mood disorders were not significantly increased in the interval of 121‐365 days following initial presentation (HR: 1.0, 95% CI: 0.91‐1.1; and HR: 1.1, 95% CI: 0.97‐1.2, respectively). In contrast, the HR for dyspnea, a known post‐acute COVID‐19 sequela 1 , increased in both time periods (1.4, 95% CI: 1.2‐1.5; and 1.2, 95% CI: 1.0‐1.3, respectively).

We reasoned that patients might be followed more closely after COVID‐19 as compared with other RTIs, and that a higher visit frequency might increase the probability of a mental illness being recorded in the EHR. To assess this, we repeated our analysis but added the frequency of visits 21 days or more after initial presentation as a factor to the Cox regression. The HR for any mental illness in the early post‐acute phase was still significant (p<0.0001), but reduced to 1.2 (95% CI: 1.1‐1.3).

Our results confirm the conclusion of the above‐cited study 3 that patients are at significantly increased risk of psychiatric conditions after a COVID‐19 diagnosis. However, the degree of increased risk documented in our study is substantially lower than previously found.

There are several potential reasons for the differences between our results and those of the above‐mentioned study. The previous study included data from January 20, 2020 (first recorded COVID‐19 case in the US) to August 1, 2020, while our study includes data through October 20, 2021. It is conceivable that perceptions of COVID‐19 by patients have shifted or that clinical practice has changed in the intervening time. It is possible that improved treatment options available later in the pandemic have reduced the risk of psychiatric illness. Finally, COVID‐19 vaccination may reduce rates of anxiety and depression and alleviate symptoms in persons with post‐acute sequelae 6 . Thus, the increasing availability of vaccines might have reduced the rate of mental illness following COVID‐19. The data available in N3C do not include comprehensive information about vaccination status, so we could not test this hypothesis.

Many cohort studies have documented a high prevalence of mental illness in individuals with long COVID. For instance, in our recent analysis, the prevalence of depression was 21.1% (median reported percentage in 25 studies) and that of anxiety was 22.2% (median over 24 studies) 1 . However, it is possible that the reported prevalence of these and other conditions was in­flated by a sampling bias toward long COVID patients who joined support groups or chose to participate in cohort studies 8 . This, and the fact that inclusion criteria for long COVID studies vary, has made it difficult to characterize the natural history of psychiatric manifestations of long COVID. Our study did not fo­cus specifically on long COVID, but instead investigated a cohort of patients following a diagnosis of acute COVID‐19. It is difficult to know what proportion of these patients went on to develop long COVID; the recent introduction of ICD‐10 codes for long COVID 9 may enable studies on this topic in the future.

In summary, we support previously published reports of an increased risk of new‐onset psychiatric illness following acute COVID‐19 infection. In contrast to the nearly doubled risk identified by the earlier study, we found the relative risk to be increased by only about 25% (3.8% vs. 3.0% following other RTI). We did not find a significant difference in risk in the late post‐acute phase, suggesting that the increased risk of new‐onset psychiatric illness is concentrated in the early post‐acute phase.

Our results have important implications for understanding the natural history of psychiatric manifestations of COVID‐19. If confirmed by independent studies, our findings suggest that health services should consider mental health screening efforts early in the post‐COVID clinical course.

Go to:

NOTES

This work was supported by the US National Center for Advancing Translational Sciences (grant no. U24 TR002306).

Go to:

REFERENCES

1. Deer RR, Rock MA, Vasilevsky N et al. EBioMedicine 2021;74:103722. [PMC free article] [PubMed] [Google Scholar]

2. Haendel MA, Chute CG, Bennett TD et al. J Am Med Inform Assoc 2021;28:427‐43. [PMC free article] [PubMed] [Google Scholar]

3. Taquet M, Luciano S, Geddes JR et al. Lancet Psychiatry 2021;8:130‐40. [PMC free article] [PubMed] [Google Scholar]

4. Taquet M, Geddes JR, Husain M et al. Lancet Psychiatry 2021;8:416‐27. [PMC free article] [PubMed] [Google Scholar]

5. Grambsch PM, Therneau TM. Biometrika 1994;81:515. [Google Scholar]

6. Arnold DT, Milne A, Samms E et al. medRxiv 2021;21253225. [Google Scholar]

7. Perez‐Arce F, Angrisani M, Bennett D et al. PLoS One 2021;16:e0256406. [PMC free article] [PubMed] [Google Scholar]

8. Davis HE, Assaf GS, McCorkell L et al. EClinicalMedicine 2021;38:101019. [PMC free article] [PubMed] [Google Scholar]

9. Duerlund LS, Shakar S, Nielsen H et al. Clin Epidemiol 2022;14:141‐8. [PMC free article] [PubMed] [Google Scholar]

University Study Finds Higher Risk Of Psychiatric Diagnoses Among COVID-19 Patients

Authors: Naveen Athrappully via The Epoch Times  June 9,2022

A recent study published by Oregon State University discovered that COVID-19 infected individuals have a higher chance of developing psychiatric disorders within about four months of contracting the virus.

For the study, published in World Psychiatry on May 7, researchers used data from the National COVID Cohort Collaborative (N3C). They matched 46,610 patients infected with COVID-19, which can trigger a respiratory tract infection (RTI), with control patients diagnosed with a different RTI.

This allowed researchers to specifically look into how COVID-19 affected the mental health of infected individuals. No patients with any history of mental illness prior to 21 days after a COVID-19 diagnosis were included in the study. Those with a medical record extending a year prior to their COVID-19 diagnosis were also excluded.

Researchers looked at the rate of psychiatric diagnoses in the 46,610 COVID-19 patients for two time periods—the early post-acute phase between 21 and 120 days from the infection and the late post-acute phase between 121 and 365 days from the infection.

The study discovered that COVID-19 patients had a 3.8 percent rate of developing a psychiatric disorder in the early post-acute phase when compared to just 3 percent for other respiratory tract infections. This amounted to a nearly 25 percent higher risk for COVID-19 patients.

However, the researchers did not find such a “significant difference in risk” when they compared COVID-19 late post‐acute phase patients with individuals with other respiratory tract infections.

When researchers looked at anxiety disorders, they found the incidence proportion of a new‐onset anxiety disorder diagnosis was “significantly higher” for COVID-19 patients when compared to RTI patients. For mood disorders, such significant differences were not observed.

“For people that have had COVID, if you’re feeling anxiety, if you’re seeing some changes in how you’re going through life from a psychiatric standpoint, it’s totally appropriate for you to seek some help,” Lauren Chan, co-author of the study, said according to a June 6 news release by Eurekalert.

“And if you’re a care provider, you need to be on the proactive side and start to screen for those psychiatric conditions and then follow up with those patients.”

Chan stressed that not every COVID-19 infected individual is going to have such psychiatric problems. In the context of the health care infrastructure of the United States, an increase in the number of COVID-19 patients seeking psychiatric care could add more strain on the system, she warned.

Multiple other studies have also suggested that a segment of COVID-19 patients might end up facing psychological issues.

Research published in April 2021 found that 34 percent of the 236,379 COVID-19 survivors included in the study developed neurological and mental disorders in the six months after becoming infected, according to WebMD.

Anxiety was the most commonly found disorder, with 17 percent of subjects reporting it. This was followed by mood disorders at 14 percent, substance abuse disorders at 7 percent, and insomnia at 5 percent.

When it came to neurological problems, 0.6 percent reported brain hemorrhage, 2.1 percent reported ischemic strokes, and 0.7 percent reported dementia. Among patients diagnosed as seriously ill with COVID-19, these rates jumped. Of the patients admitted to the intensive care unit, 7 percent experienced a stroke while 2 percent were diagnosed with dementia.

In another study published on Feb. 16 at BMJ, researchers analyzed records of nearly 153,848 COVID-19 patients in the Veterans Health Administration (VHS) system, comparing them with individuals who had not contracted the virus.

Those who got infected were found to be 35 percent more likely to be diagnosed with anxiety following the infection than uninfected people, 38 percent were more likely to be diagnosed with adjustment and stress disorders, 39 percent were more likely to be diagnosed with depression, and 41 percent were more likely to be diagnosed with sleep disorders.

There appears to be a clear excess of mental health diagnoses in the months after Covid,” Paul Harrison, a professor of psychiatry at the University of Oxford who was not involved in the study, told The New York Times.

However, only 4.4 to 5.6 percent of individuals in the study were diagnosed with anxiety, depression, adjustment, and stress disorders.

“It’s not an epidemic of anxiety and depression, fortunately,” Harrison added. “But it’s not trivial.”

Massive 23andMe survey reveals who may be at the highest risk for long COVID

Authors: Amy Graff, SFGATE May 31, 2022

It’s a question everyone wants answered: Who is more likely to develop long COVID-19, the debilitating symptoms that can linger for weeks, months or even longer, after an infection?

new unpublished study from 23andMe is part of a growing body of research shedding light on who experiences post-COVID conditions and why. The survey, which was voluntary and relied on people self-reporting symptoms, had several major findings, including that women were far more likely to experience long-term symptoms, as were people with a prior diagnosis of depression or anxiety. More than half of people who reported a diagnosis of long COVID had a history of cardiometabolic disease, such as heart attacks or diabetes.

The 23andMe survey included 100,000 people who reported a diagnosis of COVID. Of those, 26,000 described experiencing symptoms of COVID at least a month after being infected. In addition, more than 7,000 participants reported an official diagnosis of long COVID. Survey participants were asked about their symptoms at 3, 6 and 12 months. According to Centers for Disease Control and Prevention estimates, about 13.3% of people with COVID will experience symptoms for at least a month, and 2.5% of people will experience symptoms for longer than three months.

Dr. Stella Aslibekyan, a genetic epidemiologist for the consumer genetics company, which is based in South San Francisco, told SFGATE that the 23andMe study is unique because it’s so large and directly surveyed people about their symptoms. Many other studies into long-term symptoms are smaller, and based on data from medical records, as opposed to self-reported experiences. 

“We’re able to paint a more complete picture of the COVID experience than would be possible from just using medical records,” Aslibekyan said.

There are limitations to this approach, too. Participating in the study was voluntary, as opposed to a random sample of patients in a health system. The demographics and characteristics of people represented in these kinds of “self-selecting” studies are often skewed, based on who has enough time and interest to fill out the surveys.

The study, which has not been peer-reviewed or published in a medical journal, is part of an effort to solve the mystery of why so many people experience symptoms such as shortness of breath, cough, fever, fatigue, brain fog or chest pain for weeks to months after a COVID infection. Many of these symptoms align with the symptoms of chronic fatigue syndrome, a poorly understood condition that has often been linked to other viral infections, including flu and Epstein-Barr. 

“One hypothesis positions long COVID as an autoimmune condition, in which the immune system is attacking the body’s own tissues,” Aslibekyan said.

Many studies have found that long-term symptoms are much more likely to occur in the most severe COVID cases, but the U.S. Centers for Disease Control and Prevention says that “anyone who was infected with the virus can experience post-COVID conditions, even people who had mild illness or no symptoms from COVID-19.”

“Researchers found that individuals with COVID who required hospitalization had a more than ten-fold risk of being diagnosed with long COVID compared to those who were not hospitalized when controlling for age, sex, and ethnicity,” 23andMe said. 

Depression and anxiety also seem to be risk factors.

To Aslibekyan, one of the most illuminating findings in the survey was that people diagnosed with long COVID were twice as likely to report experiencing depression or anxiety before they were infected. 

There is extensive research into the relationship between depression and the immune system. People with diseases that cause over-activation of the immune system are more likely to be diagnosed with depression, and vice versa.

“When [depressed people] are hit with that acute COVID-19 infection, those long term symptoms also represent inappropriate malfunction of the immune system, so it makes sense that they’re more vulnerable to those long-term dysfunctions of the immune system,” Aslibekyan said.

Women are more likely than men to have long COVID

Researchers also found that people with two X chromosomes in each cell were more than twice as likely to report a long COVID diagnosis than those with only one. This finding is consistent with other studies that have found women are less likely to die from COVID, but more likely to develop long-term symptoms.

The X chromosome is home to many genes that control immune responses. While most men inherit just one copy from their mothers, women generally inherit two — one from each parent — and, therefore, two sets of those important immune-related genes. (In every individual cell, one or the other X chromosome is switched “on” at random.) That genetic diversity is often a good thing; women are better able to fight off many illnesses than men, for instance. But it’s also linked to a higher rate of autoimmune diseases among women, including multiple sclerosis and Sjogren’s syndrome, Aslibekyan said.

Brain fog was the most commonly self-reported symptom

The symptom most commonly reported by people responding to the survey was brain fog, a term for sluggish thinking, followed by fatigue, shortness of breath and loss of smell. Among people who received long COVID diagnoses, about 19% reported brain fog a year after infection. 

Among participants who received vaccines after catching COVID, more than half (55%) reported no change in their symptoms, 25% reported an improvement in symptoms, and about 13% reported that their symptoms got worse. The rest did not know, weren’t sure or didn’t answer.

23andMe began collecting data in April 2020. That means many of the people in the survey contracted COVID-19 before the vaccine was available. 

“So what does that tell us?” Aslibekyan said. “The vaccine may not be a cure for long COVID. Now what we do know is that the vaccine is very good at preventing both initial infection and hospitalization, yes. And hospitalization was associated with a tenfold increase in risk of long COVID. So the vaccine is great for prevention.”

What You Should Know About Long-Haul COVID

Authors: KELSEY KLOSS PUBLISHED 05/03/22

Here’s what research says about how long COVID lung disease, mental health, brain fog, and more may impact people who are immunocompromised.

Throughout the pandemic, experts have continued to learn more about long COVID (also known as long-haul COVID and post-COVID conditions) — and what it means for risk of lung disease, heart disease, mental health conditions, cognitive issues, and more.  

If you’re immunocompromised and at risk for severe COVID-19, you’ve likely already been taking every step possible to lower your chances of getting infected. However, if you do contract COVID-19, it’s important to know how it affects your risk of other health conditions so you can work with your doctor to monitor symptoms.  

It’s not clear if being immunocompromised alone makes you more likely to experience long COVID. However, you may be at greater risk simply due to your higher likelihood of developing severe COVID-19. 

“I have not seen data to suggest confirming that immunocompromised patients are more likely to develop long COVID than patients who are not immunocompromised,” says Samoon Ahmad, MD, clinical professor of psychiatry at NYU Grossman School of Medicine. “That said, it’s clear that immunocompromised patients are more likely to develop severe COVID if they get it — and research suggests that people who have severe COVID are more likely to develop long COVID.” 

However, more research is needed to confirm this link.  

Meanwhile, factors such as older age, being female, and hospitalization at symptom onset have been found to be significantly associated with an increased risk of developing persistent symptoms, per a July 2021 review in the Journal of the Royal Society of Medicine. 

Needing oxygen therapy, pre-existing hypertension, and chronic lung conditions were also highlighted in the study as being major factors of long-term symptoms.  

So what does this mean for your long-term health and risk of chronic disease? Here are potential complications of long COVID you should know about as an immunocompromised patient. 

Long COVID and Lung Disease

COVID-19 can cause both short-term and long-term complications to your lungs, but the way in which it does has changed over the course of the pandemic.  

You may remember that at the beginning of the pandemic, many people experienced COVID-related pneumonia. This resulted in oxygen levels dropping, feelings of breathlessness, and eventually hospitalization. That’s because the earlier variants had a tendency to infect the lung tissue.  

“With those variants, we saw a lot more scarring happening to the lungs,” says Panagis Galiatsatos, MD, assistant professor of medicine at Johns Hopkins Medicine. “For most people, the scarring kind of came and left. For others, it remained permanent — and then there’s a small subgroup where the scarring actually never ‘shut off’ and they developed post-COVID-19 fibrosis.” 

The newer variants, particularly Omicron, are much more involved in the airways (the tubes that lead to the lungs). That results in much more coughing during infection, but less of a drop in oxygen levels.  

A recent University of Iowa Health Care study of 100 participants revealed that air trapping persisted in eight out of nine participants imaged more than 200 days after COVID-19 diagnosis. Air trapping is a condition in which people cannot empty their lungs when they breathe out, which is indicative of small airways disease, and it leads to side effects such as shortness of breath.  

The researchers found that the percentage of lung affected by air trapping was similar across patients, regardless of how severe their symptoms were.  

What’s more, with the more recent variants, the most common complication patients have post-COVID is a post-viral cough that takes three to six months to go away. During this time, your lungs are essentially trying to “cough out” the affected cells.  

“The lungs are going to clear out the cells that were invaded,” says Dr. Galiatsatos. “This cough is very normal. We can suppress it if you need, but this is your lungs’ way of getting things out.”  

That said, if you’re immunocompromised, you could experience abnormal healing. Touch base with your physician when you’re recovering from COVID-19 so they can monitor your cough or any other lingering symptoms. While a doctor may normally look further into a cough that lasts longer than six months, if you’re immunocompromised, that timeframe may shorten.  

“If an immunosuppressed patient has a cough even just a month after COVID-19, I scan their chest and make sure things are going OK,” says Dr. Galiatsatos. Certain patients, like those who are older or have preexisting pulmonary conditions like asthma, are more vulnerable to developing ongoing pulmonary symptoms.  

In another recent study published in Radiology, researchers assessed lung abnormalities in 91 participants (mean age of 59 years) one year after they had COVID-19 pneumonia. At one year, CT scan abnormalities were found in 54 percent of the participants — 4 percent of which had received outpatient care only, 51 percent of which were treated on a general hospital ward, and 45 percent of which had received intensive care unit treatment.  

What’s more, 63 percent of participants with abnormalities did not show additional improvements after six months. Being older than 60 years, critical COVID-19 severity, and being male were associated with persistent CT abnormalities at one year.  

Long COVID and Diabetes

There may also be a link between long COVID and the development of type 2 diabetes. In a May 2022 study in The Lancet Diabetes & Endocrinology, researchers used the national databases of the U.S. Department of Veterans Affairs to analyze data from more than 8.5 million participants before and during the pandemic.  

They found that people who had been infected with COVID-19 were about 40 percent more likely to develop diabetes up to a year later than those in a control group.  

Almost all cases were type 2 diabetes, in which the body doesn’t produce enough insulin or becomes resistant to it. Patients who were hospitalized or admitted to intensive care had roughly triple the risk compared to control participants who did not have COVID-19 — but even those with mild infections and no previous diabetes risk factors had a higher chance of developing the condition.  

“The mechanism(s) underpinning the association between COVID-19 and risk of diabetes are not entirely clear,” note the researchers.  

However, it’s clear that prevention and monitoring for diabetes should be part of the post-COVID strategies, particularly for those who experienced severe COVID-19.  

“Current evidence suggests that diabetes is a facet of the multifaceted long COVID syndrome,” add the researchers. “Post-acute care strategies of people with COVID-19 should include identification and management of diabetes.” (Acute COVID-19 is the stage of infection that typically lasts four weeks from the onset of symptoms, per a review in Nature Medicine.)  

Long COVID and Mental Health

Long COVID has also been linked to a variety of mental health and cognitive issues, including:  

Depression & Anxiety

COVID-19 may increase your chances of experiencing depression or anxiety. An observational follow-up study in six European countries published in The Lancet Public Health found that COVID-19 survivors who were bedridden for more than seven days had a persistently higher risk for depression (61%) and anxiety (43%) than uninfected participants throughout the study period.  

A 2021 study published in Cardiovascular Diabetology suggests that long COVID is primarily caused by microclots that starve different cells of oxygen. These microclots form around trapped inflammatory markers.  

“I think this mild hypoxia [deprivation of oxygen in tissues] can lead to inflammation and activation of microglia,” says Dr. Ahmad. “These microglia are cells in the brain that release inflammatory signals when activated, which then leads to neuroinflammation.” This could potentially explain the pathology of long COVID and associated issues like anxiety.  

You may find it difficult to differentiate your worries about getting COVID-19 (or fears about experiencing arthritis flares or other symptoms of your underlying condition) with clinical symptoms of anxiety. Of course, the pandemic has been a period of great stress for many — and particularly those who are at high risk for severe COVID-19.  

Continuing to avoid crowds or choosing to work from home doesn’t necessarily mean you have clinical anxiety. However, anxiety does become a clinical problem when it disrupts your life to the point of you avoiding social, occupational, or academic obligations. For instance, if you feel too anxious to pick up phone calls from friends or family.  

“If your anxiety is so severe that it makes it interferes with your ability to live your life, then you definitely want to speak to a doctor,” says Dr. Ahmad. “You may also want to tell your doctor about your anxiety if it is part of a larger cluster of symptoms — including shortness of breath, fatigue, or if you are slow to heal from small cuts or bruises.” 

It’s important to stay in touch with your doctor and keep a symptom journal if necessary. 

“Immunocompromised patients should know that anxiety is one of the symptoms of long COVID and that it can be exacerbated by other symptoms,” says Dr. Ahmad. “For example, many patients with long COVID report sleep problems (“COVID-somnia”). When you don’t sleep well, this can make your anxiety worse.” 

Sleep disturbances are estimated to affect up to 50 to 75 percent of COVID-19 patients, per a 2021 review in the Journal of Personalized Medicine. And of course, if you’re living with another underlying condition, symptoms like insomnia or pain may be commonplace for you.  

“A similar thing can be said of several other common long COVID symptoms, including fatigue and shortness of breath,” says Dr. Ahmad. “When a patient feels severe shortness of breath, this may even trigger a panic attack.” 

Meanwhile, it’s normal to feel sad sometimes (especially during a global pandemic), but if you’re persistently sad, anxious, or in an “empty” mood, it could be a symptom of depression, per the National Institute of Mental Health.  

Other common symptoms of depression include: 

  • Feelings of hopelessness or pessimism 
  • Loss of interest or pleasure in hobbies 
  • Difficulty concentrating 
  • Changes in appetite or unplanned weight changes 
  • Suicide attempts or thoughts of death or suicide 

If you or someone you know is in immediate distress or thinking about hurting themselves, call the National Suicide Prevention Lifeline at 1-800-273-TALK (8255). You also can text the Crisis Text Line (HELLO to 741741). 

Brain Fog

Difficulty with concentration and memory have also been attributed to long COVID. In fact, brain fog — the feeling of slow or sluggish thinking — occurs in an estimated 22 to 32 percent of patients who recover from COVID-19, per Harvard Medical School.  

In a recent study in the journal Natureresearchers analyzed brain changes in 785 participants ages 51-81 whose brains were scanned twice (including 401 people who contracted COVID-19 between their two scans). They found evidence that COVID-19 can cause the brain to shrink by reducing grey matter in regions that control emotion and memory. 

“The participants who were infected with SARS-CoV-2 also showed on average a greater cognitive decline between the two time points,” note the researchers. 

The effects were even seen in those who were not hospitalized with COVID-19. More research is needed to determine if this impact could be partially reversed or if it will persist in the long-term. 

Meanwhile, a January 2022 study published in Brain Communications suggests that some people may have problems with memory and attention after recovering from a mild case of COVID-19, even if they don’t realize it.  

Testing showed that performance on tasks involving attention and memory were poorer in participants who had COVID-19 compared to those that didn’t. However, in this study, both of the effects seemed to improve within six to nine months.  

“My doctor said I was COVID long-hauler after my symptoms continued from June of 2020. Recently, I began noticing brain fog, a symptom I haven’t experienced in quite some time,” says JP Summers, an Advocacy Fellow at the Global Health Living Foundation who lives with migraine, fibromyalgia, rheumatoid arthritis, and heart disease. “My mind goes completely blank. A heavy cloud of confusion sets in and I feel lost on where I am or what I was doing at that moment. It is both incredibly frustrating and terrifying, especially when it happens in a public place.” JP has been actively tracking her symptoms to discuss with her cardiologist at her next appointment.

Long COVID and Heart Disease

Heart health has been a major focus during the COVID-19 pandemic, with several cardiovascular effects appearing to be associated with long COVID. 

In a February 2022 study published in Nature Medicine, researchers analyzed 154,000 U.S. veterans (plus over 10 million patients who served as historical and contemporary control groups). They found that in the year after recovering from COVID-19, patients had increased risks of several cardiovascular issues, including abnormal heart rhythms, heart muscle inflammation, blood clots, strokes, myocardial infarction, and heart failure — even if they weren’t hospitalized with COVID-19.  

The risks were evident regardless of age, race, sex, and other cardiovascular risk factors such as obesity, hypertension, diabetes, chronic kidney disease, and hyperlipidemia (high levels of fat particles in the blood). The risks were also evident in those who did not have any cardiovascular disease before exposure to COVID-19, showing that these risks may manifest even in those at low risk of heart disease.  

“Our results provide evidence that the risk and 1-year burden of cardiovascular disease in survivors of acute COVID-19 are substantial,” note the researchers. “Care pathways of those surviving the acute episode of COVID-19 should include attention to cardiovascular health and disease.”  

Your heart and lungs work together to deliver oxygen-rich blood to your body, but COVID-19 can disrupt both. COVID-19 can cause lung damage, keeping oxygen from reaching the heart muscle, meaning your heart has to work harder to get oxygen to other tissues in the body, per the University of Maryland Medical System.  

COVID-19 can also cause an excess of inflammation, which may further damage the heart and affect the electrical signals that help it beat properly. This can lead to abnormal heart rhythm or exacerbate an existing rhythm problem.  

Work with your doctor to monitor your heart health and practice a heart-healthy lifestyle. This includes staying active, eating a healthy diet, managing your stress, and quitting if you’re a smoker.  

The bottom line: Although research is showing that long-term complications from COVID-19 are prevalent.

Warning to anyone who’s had Covid over ‘irreversible’ damage to the brain

Authors: Vanessa Chalmers, Digital Health Reporter  Apr 11 2022

COVID survivors have been warned that the brain could be irreversibly harmed by the virus.

The major organ has been shown in dozens of studies to be damaged in even the mildest forms of Covid illness.

‘Brain fog’, difficulty concentrating and memory problems have all been reported, with some encouraging studies suggesting most people see improvements in six to nine months.

The new study, by researchers at the University of Oxford, looked at people in the UK over the age of 50 who had a mild case of Covid.

All 785 participants were in the UK Biobank, a large database for medical research, and had two brain scans 38 months aparts.

A total of 401 participants had tested positive for Covid in between the two scans.

The study found a number of effects on the brain, on average 4.5 months following infection.

Covid survivors had a greater reduction in grey matter thickness and tissue damage in regions of the brain associated with smell.

They had a reduction in whole brain size and, after performing a number of tests, showed a drop in cognitive function.

The effects ranged from 0.2 to 2 per cent additional change compared with the participants who had not been infected.

Professor Gwenaëlle Douaud, lead author on the study, said: “Despite the infection being mild for 96 per cent of our participants, we saw a greater loss of grey matter volume, and greater tissue damage in the infected participants.

“They also showed greater decline in their mental abilities to perform complex tasks, and this mental worsening was partly related to these brain abnormalities. 

“All these negative effects were more marked at older ages. 

“A key question for future brain imaging studies is to see if this brain tissue damage resolves over the longer term.”

It is not clear at this stage if the effects on the brain are reversible.

Professor Stephen Smith, senior author on the study, said: “The fact that we have the pre-infection scan helps us distinguish brain changes related to the infection from differences that may have pre-existed in their brains.”

The evidence is stacking up

The study, published in the journal Nature in March, echoes the findings of a number of others.

Researchers at Tulane University reported findings last week based on studying primates, which are used in studies for the likeness to humans. 

They found severe brain swelling and injury linked to reduced blood flow or oxygen to the brain.

They also found evidence of small bleeds, neuron damage and death – even in primates that didn’t have a severe illness.

Lead investigator Dr Tracy Fischer said: “Because the subjects didn’t experience significant respiratory symptoms, no one expected them to have the severity of disease that we found in the brain.

“But the findings were distinct and profound, and undeniably a result of the infection.”

Meanwhile, researchers – including from the universities of Imperial College London and Cambridge – found that Covid can cause a “substantial drop” in intelligence.

The findings came from a series of tests on memory, reasoning, planning and problem solving on more than 81,300 people.

People who had been on a ventilator during their Covid sickness were most likely to see a decline in scores.

In a classic intelligence test, they would have lost the equivalent of seven points in IQ, the team claimed.

The study said: “These results accord with reports of long-Covid, where ‘brain fog’, trouble concentrating and difficulty finding the correct words are common.

“The deficits were of substantial effect size for people who had been hospitalised.”

Another study reassured that “brain fog” shouldn’t persist for more than a year.

Covid patients scored significantly worse in episodic memory and in their ability to sustain attention on a task over time.

However, Professor Masud Husain, of Oxford University, said it was “encouraging” that most people’s attention and memory return “largely to normal in six to nine months”.

He said: “We still do not understand the mechanisms that cause these cognitive deficit.”

Explain this…

A team in the US suggested brain fog symptoms were the result of the organ being starved of oxygen.

After autopsying Covid victims, scientists at Johns Hopkins University School of Medicine found that large cells called megakaryocytes were taking up space and leaving less room for blood to pass through the brain freely.

According to Professor James Goodwin, the Director of Science and Research Impact at the Brain Health Network, it is thought that Covid gets into the brain through tightly sealed blood vessels which surround the organ.

But there is another explanation, he wrote in The Telegraph, and our own immune systems are to blame.

Sometimes the immune system goes into overdrive in response to a virus, releasing too many inflammatory molecules called cytokines.

This phenomenon, known as a cytokine storm, can injure healthy organs, including the brain, as well as the lungs and heart. 

It has led to the death of many Covid victims, and those who survive may have long-term damage.

The cytokine storm is typically more common in people who are unhealthy, have a long-term illness, are older or who have a high viral load, Prof Goodwin said.

COVCOG 1: Factors Predicting Physical, Neurological and Cognitive Symptoms in Long COVID in a Community Sample. A First Publication From the COVID and Cognition Study

Authors: Panyuan Guo1Alvaro Benito Ballesteros1Sabine P. Yeung1Ruby Liu1Arka Saha1Lyn Curtis2Muzaffer Kaser3,4Mark P. Haggard1 and Lucy G. Cheke1*

Since its first emergence in December 2019, coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has evolved into a global pandemic. Whilst often considered a respiratory disease, a large proportion of COVID-19 patients report neurological symptoms, and there is accumulating evidence for neural damage in some individuals, with recent studies suggesting loss of gray matter in multiple regions, particularly in the left hemisphere. There are a number of mechanisms by which COVID-19 infection may lead to neurological symptoms and structural and functional changes in the brain, and it is reasonable to expect that many of these may translate into cognitive problems. Indeed, cognitive problems are one of the most commonly reported symptoms in those experiencing “Long COVID”—the chronic illness following COVID-19 infection that affects between 10 and 25% of patients. The COVID and Cognition Study is a part cross-sectional, part longitudinal, study documenting and aiming to understand the cognitive problems in Long COVID. In this first paper from the study, we document the characteristics of our sample of 181 individuals who had experienced COVID-19 infection, and 185 who had not. We explore which factors may be predictive of ongoing symptoms and their severity, as well as conducting an in-depth analysis of symptom profiles. Finally, we explore which factors predict the presence and severity of cognitive symptoms, both throughout the ongoing illness and at the time of testing. The main finding from this first analysis is that that severity of initial illness is a significant predictor of the presence and severity of ongoing symptoms, and that some symptoms during the initial illness—particularly limb weakness—may be more common in those that have more severe ongoing symptoms. Symptom profiles can be well described in terms of 5 or 6 factors, reflecting the variety of this highly heterogenous condition experienced by the individual. Specifically, we found that neurological/psychiatric and fatigue/mixed symptoms during the initial illness, and that neurological, gastrointestinal, and cardiopulmonary/fatigue symptoms during the ongoing illness, predicted experience of cognitive symptoms.

Introduction

Manifestations of coronavirus 2 (SARS-CoV-2) infection vary in severity ranging from asymptomatic to fatal. In the acute stage, symptomatic patients—at least in the early variants—typically experience respiratory difficulties that can result in hospitalization and require assisted ventilation (Baj et al., 2020Heneka et al., 2020Jain, 2020). While COVID-19 is primarily associated with respiratory and pulmonary challenge, 35% of patients report neurological symptoms including headache and dizziness (e.g., Mao et al., 2020). In severe illness, neurological symptoms can be seen in 50–85% of patients (e.g., Pryce-Roberts et al., 2020Romero-Sánchez et al., 2020). Indeed, alteration in taste or smell (anosmia/dysgeusia) is reported in over 80% of cases (e.g., Lechien et al., 2020), is often the first clinical symptom (Mao et al., 2020Romero-Sánchez et al., 2020) and regularly persists beyond resolution of respiratory illness (Lechien et al., 2020).

Accumulating evidence suggests that many COVID-19 patients experiencing severe illness show evidence of neural damage (Helms et al., 2020Kandemirli et al., 2020) and unusual neural activity (Galanopoulou et al., 2020). There are a number of postulated mechanisms linking COVID-19 infection with neurological problems (Bougakov et al., 2021). For example, based on the behavior of previous SARS viruses, SARS-CoV-2 may attack the brain directly perhaps via the olfactory nerve (Lechien et al., 2020Politi et al., 2020) causing encephalitis. Severe hypoxia from respiratory failure or distress can also induce hypoxic/anoxic-related encephalopathy (Guo et al., 2020). There is considerable evidence that COVID-19 is associated with abnormal blood coagulation, which can increase risk of acute ischemic and hemorrhagic cerebrovascular events (CVAs) (Beyrouti et al., 2020Li et al., 2020Wang et al., 2020Kubánková et al., 2021) leading to more lasting brain lesions. Indeed, ischemic or hemorrhagic lesions have been found in COVID-19 patients in multiple studies (Le Guennec et al., 2020Matschke et al., 2020Moriguchi et al., 2020Poyiadji et al., 2020). A recent study using the United Kingdom Biobank cohort comparing structural and functional brain scans before and after infection with COVID-19 identified significant loss of gray matter in the parahippocampal gyrus, lateral orbitofrontal cortex and insula, notably concentrated in the left hemisphere in patients relative to controls (Douaud et al., 2021).

A key candidate mechanism is dysfunctional or excessive immune response to infection. For example, excessive cytokine release (“cytokine storm”) and immune-mediated peripheral neuropathy (e.g., Guillain-Barre syndrome) are both linked with neurological and sensory-motor issues (Alberti et al., 2020Das et al., 2020Poyiadji et al., 2020Whittaker et al., 2020Zhao et al., 2020). In addition to acute effects, chronic inflammation has also been associated with neural and cognitive dysfunction, particularly in the hippocampus—a key area responsible for memory (Ekdahl et al., 2003Monje et al., 2003Jakubs et al., 2008Belarbi et al., 2012). Considerable rodent evidence links inflammatory cytokines with cognitive impairments (e.g., IL-1β: Thirumangalakudi et al., 2008Beilharz et al., 20142018Che et al., 2018Mirzaei et al., 2018; TNF-α: Thirumangalakudi et al., 2008Beilharz et al., 2014Almeida-Suhett et al., 2017). These findings are broadly reflected in human studies, wherein circulating cytokines have been associated with reduced episodic memory (e.g., Kheirouri and Alizadeh, 2019) and chronic neuroinflammation has been heavily implicated in the pathophysiology of neurodegenerative diseases (McGeer and McGeer, 2010Zotova et al., 2010Chen et al., 2016Bossù et al., 2020). Given the volume of reports of excessive immune response to COVID-19 infection (Mehta et al., 2020Tay et al., 2020), and evidence for neuroinflammation from postmortem reports (Matschke et al., 2020) research into cognitive sequalae is highly implicated.

Given the evidence for widespread neural symptoms and demonstrable neural damage, it could be expected that COVID-19 infection would be associated with cognitive deficits. Indeed, there is some early evidence linking neural changes following COVID-19 and cognitive deficits. Hosp et al. (2021) found that evidence of frontoparietal hypometabolism in older patients presenting with post-COVID-19 neurological symptoms via positron emission tomography (PET) was associated with lower neuropsychological scores, particularly in tests of verbal memory and executive functions.

Many forms of neuropathology would be unlikely to be present uniquely as cognitive deficits, but would be associated with a range of related symptoms. Some of these symptoms may be neurological (e.g., disorientation, headache, numbness) while others may reflect systemic/multisystem involvement (e.g., reflecting the symptom profile of chronic inflammatory or autoimmune diseases). It may therefore be possible to gain information as to the mechanism of neurological involvement via investigation of symptomatology. If it is possible to identify groups of symptoms (such as neurological, respiratory, systemic) during either the acute or post-acute phase of illness that predict cognitive problems, this may aid in the identification of patients that are at risk of developing cognitive deficits. In a highly heterogenous condition, in which up to 200 symptoms have been suggested (Davis et al., 2021), reduction of dimensionality is essential to allow meaningful associations to be drawn between experienced symptoms and relevant outcomes.

The United Kingdom Office for National Statistics [ONS] (2021) has estimated that around 21% of those experiencing COVID-19 infection still have symptoms at 5 weeks, and that 10% still have these symptoms at 12 weeks from onset. These figures may not tell the full story, being based on a list of 12 physical symptoms which does not include neurological or cognitive manifestations (e.g., Alwan and Johnson, 2021Ziauddeen et al., 2021). Other calculations suggest that around 1 in 3 non-hospitalized COVID-19 patients have physical or neurological symptoms after 2–6 weeks from disease onset (Sudre et al., 2020Tenforde et al., 2020Nehme et al., 2021) and that 11–24% still have persisting physical, neurological or cognitive symptoms 3 months after disease onset (Cirulli et al., 2020Ding et al., 2020). A community-based study reported that around 38% symptomatic people experienced at least one physical or neurological symptom lasting 12 weeks or more from onset and around 15% experienced three or more of these symptoms (Whitaker et al., 2021). Ongoing symptoms seem to occur regardless of the severity of the initial infection, with even asymptomatic patients sometimes going on to develop secondary illness (FAIR Health, 2021Nehme et al., 2021), however, initial severity may impact severity of ongoing issues (e.g., Whitaker et al., 2021).

The National Institute for Health and Care Excellence (NICE) guidelines describe “post-COVID-19 syndrome” as “Signs or symptoms that develop during or after infection consistent with COVID-19, continue for more than 12 weeks and are not explained by an alternative diagnosis” (National Institute for Health and Care Excellence [NICE], 2020). One difficulty with this definition is that the “signs or symptoms” that qualify for the diagnosis are not specified (e.g., Alwan and Johnson, 2021Ziauddeen et al., 2021) thus many patients could go uncounted and unrecognized clinically, or conversely over-liberal inclusion may lead to overcounting. The patient-created term “Long COVID” has increasingly been used as an umbrella term to describe the highly heterogenous condition experienced by many people following COVID-19 infection (Callard and Perego, 2021).

Emerging evidence suggests that Long COVID is a debilitating multisystem illness that affects multiple organ systems and there have been some attempts to characterize “phenotypes.” An online survey involved in 2,550 non-hospitalized participants detected two clusters within both initial and ongoing symptoms. Initial symptoms showed a majority cluster with cardiopulmonary symptoms predominant, and a minority cluster with multisystem symptoms that did not align specifically with any one organ system. Similarly, ongoing symptoms were clustered into a majority cluster with cardiopulmonary, cognitive symptoms and exhaustion, and a minority cluster with multisystem symptoms. Those with more related symptoms in the initial major cluster were more likely to move into ongoing multisystem cluster, and this movement can be predicted by gender and age, with higher risk in women, those younger than 60, and those that took less rest during the initial illness (Ziauddeen et al., 2021).

“Long COVID” research has repeatedly identified cognitive dysfunction as one of the most common persistent symptoms (after fatigue), occurring in around 70% of patients (Cirulli et al., 2020Bliddal et al., 2021Davis et al., 2021Ziauddeen et al., 2021). Indeed, brain fog and difficulty concentrating are more common than cough is at many points in the Long COVID time course (Assaf et al., 2020). Ziauddeen et al. (2021) report nearly 40% of participants endorsing at least one cognitive symptom during the initial 2 weeks of illness, with this persisting in the long term. However around 30% of participants also reported developing cognitive symptoms—particularly brain fog and memory problems—later. Indeed, Davis et al. (2021) demonstrate that brain fog, memory problems and speech and language problems were more commonly reported at week 8 and beyond than they were during initial infection. Furthermore, strenuous cognitive activity was found to be one of the most common triggers leading to relapse/exacerbation of existing symptoms (Davis et al., 2021Ziauddeen et al., 2021). Crucially, 86% of participants indicated that cognitive dysfunction and/or memory impairment was impacting their ability to work, with nearly 30% reporting being “severely unable to work” and only 27% working as many hours as they had pre-COVID-19 (Davis et al., 2021). These figures suggest that the cognitive sequelae of COVID-19 have the potential for long-term consequences not just for individuals but also—given the prevalence of Long COVID—for the economy and wider society.

Here we report on the first stage of a mixed cross-sectional/longitudinal investigation—The COVID and Cognition Study (COVCOG)—aimed at understanding cognition in post-acute COVID-19. The aims of this current paper are threefold: First, to provide a detailed demographic profile of our sample, comparing those who had experienced COVID-19 infection to those who had not, and those who recovered to those who continued to experience COVID-19 symptoms after acute phase of illness. Second, we aim to contribute to the understanding of phenotypes of Long COVID by using a rigorous factor analytic approach to identify groups of symptoms that tend to co-occur. We investigate symptom profiles both during and following initial infection in those that had experienced COVID-19. This allows investigation of symptoms during initial illness that may be predictive of ongoing symptoms, as well as exploring the nature of those ongoing symptoms themselves. These phenotypes may, through future studies, be directly linked to disease profiles and mechanisms. In an application of this second aim, a third objective is to use the symptom factors extracted (such as those incorporating neurological symptoms) to investigate predictors of self-reported cognitive deficits. Due to the novel character of both the virus and the subsequent ongoing illness at the time of study creation, this study was designed not to test specific hypotheses but to map the terrain, generating hypotheses for future, more targeted investigation.

Materials and Methods

Participants

A total of 421 participants aged 18 and over were recruited through word of mouth, student societies and online/social media platforms such as the Facebook Long COVID Support Group (over 40K members). Of these, 163 participants were recruited through the Prolific recruitment site, targeting participants with demographic profiles otherwise underrepresented in our sample. Specifically, recruitment through Prolific was limited to those with low socioeconomic status and levels of education below a bachelor’s degree. As the study was conducted in English, participants were recruited from majority English speaking countries (the United Kingdom, Ireland, United States, Canada, Australia, New Zealand, or South Africa). Informed consent to use of anonymized data was obtained prior to starting.

Data collection for this stage of the study took place between October 2020 and March 2021, and recorded data on infections that occurred between March 2020 and February 2021. As such, all participants with experience of COVID-19 infection were likely to have been infected with either Wild-Type or Alpha-variant SARS-CoV-2, as the later-emerging variants (e.g., Delta, Omicron) were not common in the study countries at that time. Study recruitment started before the roll out of vaccinations, thus we do not have confirmed vaccination status for all participants. Once vaccination became available, the questionnaire was revised to ask about vaccination status. Of the 33 participants who were tested after this point, 11 (2 in the No COVID group, 9 in the COVID group) reported being vaccinated. Among them, 8 had received the first dose and 3 had had two doses. The majority (over 80%) had the vaccine within the last 7 days to last month. All received Pfizer (BNT162b2) except 1 (COVID group) who received AstraZeneca (AZD1222).

Procedure

The study was reviewed by University of Cambridge Department of Psychology ethics committee (PRE.2020.106, 8/9/2020). The current paper is part of a larger, mixed cross-sectional/longitudinal online study (“COVCOG”) conducted using the online assessment platform Gorilla.1 The COVCOG study consists of a baseline assessment of characteristics and cognition in samples of individuals who had or had not experienced COVID-19 infection. Both groups completed questionnaire and a range of cognitive tasks and were then followed up at regular intervals. The results reported here are for the questionnaire section of the baseline session only. The questionnaire covered demographics, previous health and experience of COVID-19.

Participants answered questions relating to their age, sex, education level, country of permanent residence, ethnicity, and profession. They were then asked a series of questions relating to their medical history and health-related behaviors. These included self-reporting their height and weight—which were used to calculate body mass index (BMI), and their usual diet intake, use of tobacco and alcohol, and physical activity (before the illness if infected) on a 6-point frequency scale from “Never” to “Several times daily.” Following this, they were asked for details of their experience of COVID-19. Because many of the participants in this study contracted COVID-19 before confirmatory testing of infection state was widely available, both those with (“Confirmed”) and without test confirmation (“Unconfirmed”) were included in the “COVID” group. Those that didn’t think they had had COVID-19 but had experienced an illness that could have been COVID-19 were assigned an “Unknown” infection status. Those that confirmed that they had not had COVID-19, nor any illness that might have been COVID-19, were included in the “No COVID” group. The procedure for grouping and progression through the baseline session is detailed in Figure 1.FIGURE 1

Figure 1. Study procedural flow.

Participants in the “COVID” group indicated the number of weeks since infection on a drop-down menu. Those that reported being within the first 3 weeks of infection proceeded straight to debriefing and were followed up 2 weeks later, once the initial infection was passed. Apart from this delay, they proceeded with the experiment in the same way as the rest of the COVID group. Participants then answered questions on the severity of the initial illness and whether they were experiencing ongoing symptoms. Finally, participants were asked to give details on a large number of individual symptoms during three time periods: initial illness (first 3 weeks), ongoing illness (“since then,” i.e., the time since initial infection), and currently (past 1–2 days). When reporting on initial symptoms, participants gave an indication of severity on a scale of 1–3 from “Not at all” to “Very severe.” When reporting symptoms over the period “since then” they reported on both severity and regularity of symptoms on a scale of 1–5 from “Not at all” to “Very severe and often.” When reporting on symptoms in the past 1–2 days, they reported the presence or absence of the symptoms dichotomously (i.e., check the box of the symptom if present). These symptom lists were developed based on currently available medical literature reporting symptoms experienced by COVID-19 patients and through consulting medical doctors and COVID-19 patients from the Long COVID Support Group. Participants in the “No COVID” Group were not asked their experience of COVID-19.

Data Processing and Analysis

Analyses were conducted using IBM SPSS Statistics for Windows, Version 23.0. We describe quantitative variables using means and standard deviations, and numbers and percentages for qualitative variables. Sidak’s correction for multiple comparisons was employed. All p-values are reported uncorrected, and the Sidak-corrected alpha is quoted where appropriate.

We investigated differences in the first group of variables: sociodemographic, medical history, and health behaviors, concerning two COVID group classifications. First dividing the sample into two groups (COVID/No COVID), second subdividing the COVID group by symptom longevity and severity (Recovered, Ongoing mild infection, and Ongoing severe infection). Where parametric analysis was not appropriate, we employed the Pearson’s chi-square (χ2) test for categorical variables and the Mann-Whitney and Kruskal-Wallis test for continuous variables depending on the number of COVID groups. To investigate differences between groups (COVID/No COVID; Recovered/Ongoing mild/Ongoing severe), we employed Mann-Whitney and ANOVA/Kruskal-Wallis. To examine whether these variables and initial symptoms predicted degrees of ongoing illness, we ran independent multinomial logistic regression, using forward stepwise method to identify what items within these variables were significant predictors while controlling for demographics including sex, age, education, and country of residence. Next, to determine suitable groups of symptoms, we employed exploratory principal component analysis (PCA) with varimax rotation. Based on our high number of items (Nunnally, 1978) and the novelty of the subject (Henson and Roberts, 2006), we performed two PCAs, one for the initial symptoms and another one symptoms experienced since the initial phase. We then used the high-loading items on the “since then” symptom factors to calculate profiles for currently experienced symptoms. To explore what symptom factors were associated with infection or ongoing symptoms, we employed various independent multinomial logistic regression with backward elimination of variables p > 0.05 to identify the best fitted models. Data analyzed in relation to our study aims are depicted in Figure 2.FIGURE 2

Figure 2. Data analyzed in relation to our study aims.

Results

Sample Characteristics

No COVID (NCn = 185) vs. COVID (Cn = 181)

Distributions of demographics including sex, age, education level, country, and ethnicity of the two groups (NC/C) are shown in Table 1. The majority of participants were from the United Kingdom and were of White (Northern European) ethnicity (over 70% in both groups). Pearson’s chi-square tests showed that the groups did not significantly differ in sex, but differed in age [χ2(5) = 19.08, p = 0.002, V = 0.228] and level of education [χ2(5) = 56.86, p < 0.001, V = 0.394], with the COVID group tending to fall into the older age ranges and higher education level more than the No COVID group.TABLE 1

Table 1. Distribution of demographics in No COVID and COVID groups.

Employment

Supplementary Table 1 shows the distributions of pre-pandemic profession and employment status. To adjust for multiple comparisons, Sidak corrections were applied and alpha levels were adjusted to 0.003 for profession and 0.007 for employment status. The COVID group had significantly more people working in healthcare [χ2(1) = 12.77, p < 0.001, V = 0.187] and engaging in full-time work before the pandemic [χ2(1) = 21.19, p < 0.001, V = 0.241]. In contrast, the No COVID group were more likely not to be in paid work [Profession “Not in paid work” χ2(1) = 27.72, p < 0.001, V = 0.275; Employment status “Not Working” χ2(1) = 13.18, p < 0.001, V = 0.190], and they were more likely to be students [χ2(1) = 8.91, p = 0.003, V = 0.156].

Health and Medical History

Supplementary Table 2 compares medical history and health behaviors across the COVID and No COVID groups, which may be informative as to vulnerabilities. Sidak correction adjusted the alpha level to 0.003 for medical history and 0.008 for health behaviors. Pearson’s chi-square tests showed that inflammatory or autoimmune diseases [χ2(1) = 9.81, p = 0.002, V = 0.164] were found more commonly in the COVID group than the No COVID group. Mann-Whitney U-tests showed that the COVID group consumed more fruit and vegetables (U = 13,525, p = 0.001) and had higher level of physical activity (U = 13,752, p = 0.002) than the No COVID group, while the No COVID group consumed sugary (U = 14168.5, p = 0.008) food more than the COVID group. ANOVA showed that the COVID group (M = 26.71, SD = 7.26) had higher BMI than the No COVID group (M = 25.15, SD = 5.64), [F(1, 361) = 5.24, p = 0.023]. However this effect was not significant after controlling for sex, age, education and country [F(1, 357) = 1.57, p = 0.211].

Characteristics of Those Experiencing Ongoing Symptoms

To understand the potential association between the progression of COVID-19 and various potential risk factors at baseline, including demographics, medical history and health behaviors, and the severity of initial illness and initial symptoms, we further divided the COVID group into three duration subgroups: (i) those who, at the time of test, had recovered from COVID-19 (“Recovered group,” Rn = 42), (ii) those who continued to experience mild or moderate ongoing symptoms [“Ongoing (Mild/Moderate) group,” C + ; n = 53], and (iii) those who experienced severe ongoing symptoms [“Ongoing (Severe) group,” C + + ; n = 66]. Those who were still at their first 3 weeks of COVID-19 infection (n = 17) or those who reported “it is too soon” to comment on their ongoing symptoms (n = 3) were not included in the following analyses. Participants in all groups ranged between 3 and 31 + weeks since symptom-onset, and a majority (81.5%) of those with ongoing symptoms reporting after more than 6 months since infection.

Figure 3 shows the distribution of demographic variables across the COVID-19 duration subgroups (further details available in Supplementary Table 3). In each, more than half of the participants were from the United Kingdom (54.8–92.4%) and were of White (Northern European) ethnicity (69–93.9%). Pearson’s chi-square tests suggested that age [χ2(10) = 53.41, p < 0.001, V = 0.407] and education level [χ2(10) = 20.03, p = 0.029, V = 0.249], but not sex, significantly differed between subgroups. In terms of age, the R subgroup tended to fall more in the younger age ranges (see Figure 3A). In terms of education level, the R subgroup tended to have lower education level (GCSE or below and A level), but the C + + (Severe) subgroup clustered more in higher education level (bachelor’s degree) (see Figure 3B). The subgroups also differed in the time elapsed since infection at the time of completing the study [χ2(6) = 19.64, p = 0.003, V = 0.247]. The R subgroup were more likely to be in their first 10 weeks of infection, while the C + + (Severe) subgroup were more likely to be at their 31 weeks or above (Figure 3C).FIGURE 3

Figure 3. Distributions of (A) age, (B) education level, (C) weeks since infection, and (D) severity of initial illness in Recovered, Ongoing (Mild/Moderate) and Ongoing (Severe) subgroups.

A multinomial logistic regression indicated that only age, but not sex or education, was significantly associated with COVID-19 progression [χ2(10) = 43.6, p < 0.001]. People in the age ranges of 18–20 and 21–30 years were more likely to recover from COVID-19 than to progress into mild/moderate (ps = 0.02–0.03) or severe (p = 0.002) ongoing symptoms.

We examined whether medical history and health behaviors were different between COVID-19 duration subgroups. Table 2 shows the descriptive statistics of these factors in RC +, and C + + subgroups for medical history and pre-pandemic health behaviors. None of the listed health conditions significantly differed between subgroups (against Sidak α = 0.003). There were, however, significant group differences (Sidak α = 0.008) in fruit and vegetables consumption [H(2) = 15.92, p < 0.001] and fatty food consumption [H(2) = 36.54, p < 0.001]. Both ongoing symptom subgroups ate more fruit and vegetables (C + + : U = 810, p < 0.001; C + : U = 808, p = 0.016) and less fatty food (C + : U = 773.5, p = 0.005; C + + : U = 552.5, p < 0.001) than the R subgroup. The C + (Mild/Moderate) subgroup also consumed more fatty food than the C + + (Severe) subgroup (U = 1142, p < 0.001). The subgroups did not significantly differ in BMI [F(2, 157) = 0.085, p = 0.919].TABLE 2

Table 2. Distribution of medical history and health behaviors (1 = Never–6 = Several times daily; higher scores indicating higher frequency) in COVID subgroups: Recovered (R), Ongoing (Mild/Moderate) (C+) and Ongoing (Severe) (C++).

After controlling for sex, age, education, and country, a forward stepwise multinomial logistic regression indicated that no medical history variables were associated with COVID-19 progression, however, health behaviors including fatty food consumption [χ2(2) = 23.25, p < 0.001], physical activity [χ2(2) = 10.31, p = 0.006], and alcohol consumption [χ2(2) = 8.18, p = 0.017] were all significantly associated with COVID-19 progression. In our sample, people consuming more fatty food had a higher chance of having recovered from COVID-19 (p < 0.001) or having developed mild/moderate ongoing symptoms (p < 0.001) than progressing into severe ongoing symptoms. Higher levels of physical activity were associated with reduced chance of recovery relative to progression onto mild/moderate (p = 0.002) or severe ongoing symptoms (p = 0.034). Those drinking alcohol more frequently were more likely to recover from COVID-19 than to develop severe ongoing symptoms (p = 0.007).

Severity of Initial Illness

The severity of illness in the first 3 weeks of infection was associated with subsequent symptom longevity. Multinomial logistic regression showed that severity of initial illness was significantly associated with COVID-19 progression [χ2(2) = 24.44, p < 0.001], with higher initial severity associated with more severe subsequent ongoing symptoms (ps < 0.001–0.02). This effect was maintained after controlling for sex, age, education, and country [χ2(2) = 12.28, p = 0.002; C + + > C + : p = 0.048; C + + > Rp = 0.001]. Those with severe ongoing symptoms experienced more severe initial illness than those whose ongoing symptoms were mild/moderate (U = 1,258, p = 0.005, Figure 3D) and those who were fully recovered (U = 658.5, p < 0.001). The severity difference between the C + (Mild/Moderate) subgroup and the R subgroup was also significant (U = 842, p = 0.034).

Supplementary Table 4 shows the relative frequencies of particular diagnoses received during the initial illness. Of the 109 participants who sought medical assistance, the most common diagnoses received were hypoxia (14.7%), blood clots (5.5%), and inflammation (4.6%).

Symptoms During Initial Illness

Symptoms that appeared in less than 10% of participants were excluded. Kruskal-Wallis H-tests (Sidak α = 0.001) showed significant duration-group differences in 11/33 symptoms in terms of the severity experienced (see Figure 4, more information in Supplementary Table 5). In post hoc analysis (Sidak α = 0.017), muscle/body pains, breathing issues and limb weakness showed gradation, with the C + + (Severe) subgroup having experienced the most severe symptoms, followed by the C + (Mild/Moderate) subgroup, and the R subgroup experiencing the least (p ranges < 0.001–0.012). Some symptoms did not show gradation with severity of ongoing symptoms, but were reliably higher in those with ongoing symptoms. Both the ongoing symptoms subgroups reported more severe symptoms of fatigue, brain fog and chest pain/tightness during the initial illness than those that recovered (ps < / = 0.001) but did not differ from one another. Those with severe ongoing symptoms experienced more severe nausea and blurred vision than those with mild/moderate or who recovered (p ranges < 0.001–0.009). Finally, the C + + (Severe) subgroup experienced more abdominal pain, altered consciousness and confusion during the initial illness than the R subgroup (ps < / = 0.001).FIGURE 4

Figure 4. Severity of different symptoms during the initial (left) and ongoing (right) illness among those who recovered or had ongoing mild or severe illness. Higher scores indicate higher severity.

After controlling for sex, age, education, and country, a forward stepwise multinomial logistic regression suggested that six initial symptoms were significantly associated with COVID-19 progression. These were: limb weakness [χ2(2) = 25.92, p < 0.001], brain fog [χ2(2) = 13.82, p = 0.001], chest pain or tightness [χ2(2) = 10.81, p = 0.005], dizziness [χ2(2) = 7.82, p = 0.02], cough [χ2(2) = 7.74, p = 0.021], and breathing difficulties [χ2(2) = 6.98, p = 0.031]. People initially experiencing more severe limb weakness were more likely to experience severe ongoing symptoms than to recover (p < 0.001) or develop mild/moderate ongoing symptoms (p < 0.001). More severe initial breathing issues (p = 0.014) and dizziness (p = 0.037) were associated with greater likelihood of severe than mild/moderate ongoing symptoms, but people with more severe initial dizziness (p = 0.02) and cough (p = 0.009) were more likely to recover rather than to develop mild/moderate ongoing symptoms. More severe initial brain fog and chest pain/tightness were associated with more progression into mild/moderate than either severe ongoing symptoms (brain fog: p = 0.029; chest pain: p = 0.026) or recovery (brain fog: p = 0.001; chest pain: p = 0.007).

Symptoms During Ongoing Illness

Excluding those who reported being totally asymptomatic throughout or feeling completely better very quickly after initial illness (who did not report on ongoing symptoms, n = 15), the COVID subgroups were asked to report on their ongoing experience of a list of 52 symptoms. Symptoms that appeared in less than 10% of participants were excluded. The duration-groups differed significantly in 27/47 symptoms (Sidak α = 0.001; see Figure 4 and Supplementary Table 6). Post hoc tests (Sidak α = 0.017) showed that the C + + (Severe) subgroup reported higher levels of severity than the R subgroup in all 27 symptoms (ps < 0.001–0.017) and then the C + (Mild/Moderate) subgroup in all except two (altered consciousness and eye-soreness; ps < 0.001–0.017). The C + (Mild/Moderate) subgroup also reported experiencing higher severity in 16 symptoms (including fatigue, difficulty concentrating, brain fog, and forgetfulness) than the R subgroup (ps < 0.001–0.016; see Figure 4 and Supplementary Table 6; see also Supplementary Table 7 for similar analysis of current symptoms).

Symptoms in Those With Confirmed or Suspected COVID-19 vs. “Other” Illnesses

As much of our sample experienced infection early in the pandemic before widespread testing was available, not all cases included in our COVID group were confirmed by a polymerase chain reaction (PCR) test (infection statuses: “Confirmed” COVID, “Unconfirmed” COVID). Meanwhile, a significant minority of participants had an illness during the pandemic period that they did not think was COVID-19 (infection status: “Unknown”) (see Figure 1). We compared symptom prevalence across these three groups (Unknown, n = 55; Unconfirmed, n = 96; Confirmed, n = 65) for both the initial 3 weeks of illness, and the time since then. Those who were still at their first 3 weeks of COVID-19 infection (n = 17) and who reported “it is too soon” to comment on their ongoing symptoms (n = 3) were not included in this analysis.

The groups significantly differed in 14 out of 31 symptoms during the initial illness (Sidak α = 0.0016; Supplementary Table 8). Both Confirmed and Unconfirmed groups reported higher severity than the Unknown group on 13 symptoms (including fatigue, muscle/body pains and loss of smell/taste; p ranges < 0.001–0.014; Sidak α = 0.017). Additionally, the Unconfirmed group reported more severe blurred vision than the Unknown group (p < 0.001), and the Unknown group reported more severe sore throat/hoarseness than the Confirmed group (p < 0.001). As for the differences within those with COVID-19, the Confirmed group experienced greater loss of smell/taste than the Unconfirmed group (p = 0.002), while the Unconfirmed group reported higher levels of breathing issues, chest pain/tightness, sore throat/hoarseness, and blurred vision than the Confirmed group (ps = 0.004–0.015).

Of these participants, 177 (Unknown group: n = 31; Unconfirmed group: n = 88; Confirmed group: n = 58) reported experiencing ongoing symptoms after the 3 weeks of illness. Significant group differences were found in 11/47 ongoing symptoms (Sidak α = 001; see Figure 5 and Supplementary Table 9). Post hoc tests (Sidak α = 0.017) showed that, compared with the Unknown group, both the Confirmed and Unconfirmed groups reported higher levels of fatigue, difficulty concentrating, brain fog, tip-of-the-tongue (ToT) problems, muscle/body pains, fast/irregular pulse, semantic disfluency, chest pain/tightness, limb weakness, and loss of smell/taste (ps < / = 0.001). The Unconfirmed group also experienced higher level of night waking (p = 0.001) than the Unknown group. There were no significant differences in ongoing symptoms between the Confirmed and the Unconfirmed groups.FIGURE 5

Figure 5. Experience of ongoing symptoms in Unknown, Unconfirmed COVID, and Confirmed COVID groups.

Characterizing Symptom Profiles

While data on individual symptoms are useful in identifying highly specific predictors, these are too numerous for more systematic analysis, which require data-reduction. A stated aim of this study was to identify symptom profiles that may be informative as to underlying pathology.

Initial Symptom Factors

To group the initial symptoms, we included 34 symptoms in the PCA after excluding paralysis and seizures (experienced by less than 10% of the participants). A total of 164 participants reported on their symptoms during the first 3 weeks of illness (the factor analysis coded here as 1 = Very severe, 3 = Not at all). The Kaiser-Meyer-Olkin (KMO) test (value 0.861) and Bartlett’s test of sphericity [χ2(528) = 2,250, p < 0.001] showed the data were suitable for factor analysis. We employed the varimax rotation. Initially, nine factors were obtained with eigenvalue > 1.0, which was reduced to five via Cattell’s Scree test (Kline, 2013). Assessments were conducted of 4, 5, and 6 factor solutions for interpretability and robustness. The ratio of rotated eigenvalue to unrotated eigenvalue was higher for the 5-factor solution than for the 4- or 6-factor solutions, and this structure was also the most interpretable. We thus proceeded with a 5-factor solution, which explained 50.59% of item variance with last rotated eigenvalue of 1.998.

We labeled the new components as “F1: Neurological/Psychiatric,” “F2: Fatigue/Mixed,” “F3: Gastrointestinal,” “F4: Respiratory/Infectious,” and “F5: Dermatological” (see Table 3 for factor loadings). We computed the factor scores using the regression method (see Supplementary Table 10 for factor scores).TABLE 3

Table 3. Factors and loadings from the “Initial Symptoms” PCA.

People who went on to experience ongoing symptoms showed higher factor scores in the Fatigue/Mixed symptom factor during the initial illness [F(2, 158) = 23.577, p < 0.001], but did not differ in any other initial symptom factor. Pairwise analysis revealed that those who recovered were significantly less likely to experience Fatigue/Mixed symptoms than those with mild/moderate (p < 0.001) or severe (p < 0.001) ongoing symptoms (Figure 6).FIGURE 6

Figure 6. Severity of Fatigue/Mixed symptom factor during initial illness among those who went on to full recover, or have ongoing mild or severe symptoms.

Ongoing Symptom Factors

We performed a second PCA using the symptoms experienced since the initial phase (after the first 3 weeks), including 45 symptoms. Paralysis and seizures were excluded (experienced by less than 10% of the participants). A total of 149 participants reported on their symptoms over the time since the first 3 weeks of illness (the factor analysis coded here as 1 = Very severe and often, 5 = Not at all). The KMO test (value 0.871) and Bartlett’s test of sphericity [χ2(861) = 3,302, p < 0.001] showed suitability for factor analysis. We employed the varimax rotation. PCA showed 11 components with eigenvalues > 1.0, and this was reduced to 6 via inspection of the eigenvalue gradient (scree plot). The ratio of rotated eigenvalue to unrotated eigenvalue was higher for the 7-factor solution, followed by the 6-factor. The 6- and 7-factor solutions were differentiated by subdivision of the second factor, reducing the degree of cross-loading. However, the 7-factor solution was less interpretable and less robust to removal to cross-loaders (the presence of which can be accepted from a pathology perspective, given that multiple mechanisms can produce the same symptom). As such, we proceeded with the 6-factor solution, which explained 54.17% of item variance and had a last rotated eigenvalue of 2.227.

We labeled the new components as “F1: Neurological,” “F2: Gastrointestinal/Autoimmune,” “F3: Cardiopulmonary/Fatigue,” “F4: Dermatological/Fever,” “F5: Appetite Loss,” and “F6: Mood” (see Table 4 for factor loadings). We computed the factor scores using the regression method (see Supplementary Table 11 for factor scores).TABLE 4

Table 4. Factors and loadings from the exploratory factor analysis of ongoing “since then” symptoms PCA.

In order for cognitive symptoms [brain fog, forgetfulness, tip-of-the-tongue (ToT) problems, semantic disfluency and difficulty concentrating] to be used as a dependent variable, these were isolated and a PCA run separately. A single component emerged, with all the cognitive symptoms loading homogeneously highly (see Supplementary Table 12). The KMO test (value 0.886) and Bartlett’s test of sphericity [χ2(10) = 564, p < 0.001] indicated suitability for factor analysis, and the single 5-item factor explained 76.86% of variance.

Current Symptoms

The current symptoms assessed were the same as the ongoing symptoms, but rated dichotomously as either currently present or absent. To estimate the degree to which current symptoms aligned with the factors established for the ongoing period, we generated a quasi-continuously distributed variable according to how many of the high loading (> / = 0.5) items from the ongoing factors were recorded as present currently. Using this sum scores by factor method (Tabachnick et al., 2007Hair, 2009), each score was subsequently divided by the number of items in that factor producing quasi “factor scores” that were comparable and indicative of “degree of alignment” of current symptoms to established factors.

To assess the stability and specificity of symptom profiles between these periods, serial correlations were conducted for corresponding and non-corresponding factors. Correlations of the same factor across time points were materially higher (> 0.2) from the next highest correlation among the 5 non-corresponding factors, with Williams tests (Steiger, 1980) giving the narrowest gap at p = 0.003 (Neurological: r = 0.676, t = 5.712; Gastrointestinal/Autoimmune: r = 0.531, t = 3.778; Cardiopulmonary/Fatigue: r = 0.678, t = 7.272; Dermatological/Fever: r = 0.523, t = 3.364; Appetite Loss: r = 0.591, t = 5.017; Mood: r = 0.490, t = 4.803). This consistency suggests that while particular symptoms may fluctuate, the profile of symptoms—once grouped into an adequately supported factor—is moderately stable for individuals, and can be relatively well represented by a “snapshot” of current symptoms. For completeness, an additional factor analysis was conducted on the current symptoms, which are reported in Supplementary Table 13.

One symptom factor showed change over time since infection, suggesting higher severity in those who had been ill for longer: Number of weeks since infection (positive test/first symptoms) was positive correlated with severity of ongoing severity of Cardiopulmonary/Fatigue symptoms [r(147) = 0.271, p < 0.001; Figure 7] and, to a weaker extent, current alignment with the same factor [r(147) = 0.206, p = 0.012], however, only the former association survived correction for multiple comparisons (Sidak α = 0.0085).FIGURE 7

Figure 7. Association between number of weeks since infection and severity of (top) Cardiopulmonary/Fatigue Symptoms and (bottom) cognitive symptoms in the entire period since the initial infection (left) and the past 1–2 days (right). Higher scores indicate higher symptom severity.

Cognitive Symptoms

Within those currently experiencing symptoms (n = 126), 77.8% reported difficulty concentrating, 69% reported brain fog, 67.5% reported forgetfulness, 59.5% reported tip-of-the-tongue (ToT) word finding problems and 43.7% reported semantic disfluency (saying or typing the wrong word).

Symptoms experienced during the initial illness significantly predicted both ongoing and current cognitive symptoms (Figure 8). A linear regression with backward elimination found that the best model contained the Neurological/Psychiatric, Fatigue/Mixed, Gastrointestinal, and Respiratory/Infectious symptom factors and explained 20% of variance (Radj2 = 0.2, p < 0.001). Table 5 shows that the Fatigue/Mixed symptoms factor (η′p2 = 0.129) was the better predictor followed by the Neurological/Psychiatric symptom factor (η′p2 = 0.092). For current cognitive symptoms, the best model contained both the Neurological/Psychiatric and Fatigue/Mixed symptom factors, together explaining 13.9% of variance (p < 0.001). Of the two, the Fatigue/Mixed factor was the better predictor (η′p2 = 0.110). No interactions between factors contributed significantly and were thus not included in the final models.FIGURE 8

Figure 8. Association between combined regression model predicted value for (A) initial symptom factors and ongoing cognitive symptoms; (B) initial symptom factors and current cognitive symptoms; (C) ongoing symptom factors and ongoing cognitive symptoms; (D) ongoing symptom factors and current cognitive symptoms; and (E) current symptom factors and current cognitive symptoms.TABLE 5

Table 5. Regression models predicting variation in the cognitive symptom factor (ongoing and current) from non-cognitive symptom factors (initial, ongoing, and current).

A similar, but much stronger, pattern emerged when considering the predictive value of ongoing (non-cognitive) symptoms (Figure 8). Using backward elimination to factors with significance (p < 0.05), all factors except Dermatological/Fever remained in the model, which explained over 55% of variance (Radj2 = 0.558, p < 0.001). The effect size (η′p2) for each factor is given in Table 5. The Gastrointestinal/Autoimmune and Cardiopulmonary/Fatigue factors were the biggest contributors to the model. Indeed, in an extreme elimination model in which contributing factors were limited to two or fewer, these two factors alone explained 38% of variance retaining strong significance (p < 0.001). No interactions between factors contributed significantly and were thus not included in the final models. Ongoing symptoms also predicted current cognitive symptoms. The best model explained 36% of the variance (p < 0.001) and included the Neurological, Gastrointestinal/Autoimmune and Cardiopulmonary/Fatigue factors and an interaction between the Gastrointestinal/Autoimmune and Cardiopulmonary/Fatigue factors. Of these, Cardiopulmonary/Fatigue symptoms were the strongest predictor (η′p2 = 0.208), with Neurological (η′p2 = 0.118) and Gastrointestinal/Autoimmune (η′p2 = 0.115) being relatively equal.

Current symptom factors also strongly predicted current cognitive symptoms (Figure 8). The backward elimination model left three contributing factors: Neurological, Cardiopulmonary/Fatigue and Appetite Loss. Together these explained around 50% of variance (Radj2 = 0.494). Of these, Cardiopulmonary/Fatigue was the stronger predictor (η′p2 = 0.306). Indeed, when the model was limited to just this factor, this model still explained 43% of the variance.

There was a significant association between degree of cognitive symptoms and duration of illness. Those who had been ill for longer were more likely to report having had cognitive symptoms throughout the ongoing illness [r(147) = 0.262, p = 0.001] and to be experiencing them at the time of test [r(147) = 0.179, p = 0.03] (Figure 7).

Experiences and Impact of Long COVID

Here we limited analysis to all those who reported some degree or period of ongoing symptoms following COVID-19 [i.e., excluding those who reported being totally asymptomatic throughout or feeling completely better very quickly after initial illness (n = 15)]. Of the remaining 146 participants, 108 (74%) self-identified as experiencing or having experienced “Long COVID.”

We examined the impact and experiences of ongoing illness (Table 6). In most cases, the nature and degree of negative experience of ongoing symptoms scaled with perceived severity. The change in symptoms over time differed between severity subgroups [χ2(6) = 37.52, p < 0.001, V = 0.367]. The C + + (Severe) subgroup were more likely to report that symptoms were consistent over time, while those with mild/moderate ongoing symptoms were more likely to report improvement in symptoms. As might be expected, the R subgroup were alone in reporting complete resolution of symptoms after recovery from the initial illness (Supplementary Table 14).TABLE 6

Table 6. Experiences and impact of Long COVID in different ongoing symptom severity groups.

Long COVID has significant impact on individuals’ lives. Over 54.6% of those with ongoing symptoms had experienced long periods unable to work and 34.5% had lost their job due to illness, 63.9% reported difficulty coping with day-to-day activities, 49.6% had had difficulty getting medical professionals to take their symptoms seriously, and 43.7% felt that they had experienced a trauma, while 17.6% had experienced financial difficulty as a result of illness. These impacts scaled with symptom severity. Those with severe ongoing symptoms were more likely to report being unable to work for a long period due to illness [χ2(2) = 46.42, p < 0.001, V = 0.564], having difficulty coping with day-to-day requirements [χ2(2) = 20.23, p < 0.001, V = 0.372], having difficulty getting medical professionals to take their symptoms seriously [χ2(2) = 23.05, p < 0.001, V = 0.397], and losing their job due to illness [χ2(2) = 24.39, p < 0.001, V = 0.409]. In contrast, the R subgroup tended to report experiencing none of the above [χ2(2) = 52.73, p < 0.001, V = 0.601].

We further compared job-loss with the No COVID group (n = 185). Those with ongoing symptoms were more likely to have lost their job than those who had not experienced COVID-19 [χ2(1) = 26.74, p < 0.001, V = 0.297]. The most common reason for job-loss among those with ongoing symptoms was illness [χ2(1) = 56.85, p < 0.001, V = 0.432], while the most common reason in the No COVID group was economy [χ2(1) = 7.67, p = 0.006, V = 0.159].

Discussion

Nature of Illness and Symptom Profiles

Here we report the initial findings from a cross-sectional/longitudinal study investigating cognition post-COVID-19. One aim of this first publication was to characterize the “COVID and Cognition Study” (COVCOG) sample. Within the COVID group, we recruited specifically to get good representation of those who were experiencing or had experienced ongoing symptoms. Indeed, 74% identified with the term “Long COVID.” Our final sample had a relatively even spread of those that had fully recovered at the time of test (42), or had mild/moderate (53) or severe (66) ongoing symptoms. Medical history did not differ between those experiencing ongoing symptoms and those who recovered. However, in terms of health behaviors, those with ongoing symptoms were in general “healthier,” being more likely to have previously been consuming less fatty food and more fruits and vegetables. This result is counterintuitive and may reflect insufficient controls for confounding demographic variables relating to socio-economic status. Nonetheless potential links between lifestyle and nutrition and COVID-19 recovery warrant further investigation.

The nature of the initial illness was found to have a significant impact on the likelihood and severity of ongoing symptoms. Despite this sample almost entirely comprised of non-hospitalized patients, those with more severe initial illness were more likely to have ongoing symptoms, and for those symptoms to be more severe. This suggests even in “community” cases, initial infection severity is a predictor of vulnerability to Long COVID. In an analysis of all symptoms experienced during the initial illness, there were several that were predictive of presence or severity of ongoing symptoms. In particular, individuals with severe ongoing symptoms were significantly more likely to have experienced limb weakness during the initial illness than those that recovered. However, some differences in severity ratings between ongoing subgroups were small despite being statistically significant, which warrant caution in interpreting the results.

We asked participants to retrospectively report on symptoms over three time periods: initial illness, ongoing illness, and currently experienced. Given the highly heterogenous nature of Long COVID, we used principal component analysis (PCA) with the aim to ascertain whether there may be different phenotypes of the condition within our sample—that is to say, that there may be certain types of symptoms that tend to (or not to) co-occur. For both the initial and ongoing illness, the symptom factors resemble those found in previous studies (e.g., Davis et al., 2021Whitaker et al., 2021Ziauddeen et al., 2021), with some quite coherent cardiopulmonary clusters, and other less specific “multisystem” profiles which may reflect more systemic issues such as inflammation, circulation, or endocrine function.

Predictors of Cognitive Difficulties

A large proportion of our sample reported cognitive difficulties. We isolated the cognitive symptoms for the ongoing and current illness and computed a single factor including only these. Using this, we investigated which (non-cognitive) symptom factors during both the initial and ongoing illness explained significant variance in severity of cognitive symptoms.

Together, the Fatigue/Mixed, Neurological/Psychiatric, Gastrointestinal and Respiratory/Infectious symptom factors during the initial illness explained around 20% of variance in ongoing (“since then”) cognitive symptoms, and a similar model (containing only Neurological/Psychiatric and Fatigue/Mixed symptom factors) explained nearly 14% of variance in current cognitive symptoms. These findings strongly suggest that experience of neurological symptoms during the initial illness are significant predictors of self-reported cognitive impairment. While only one factor is named “Neurological” both this and the Fatigue/Mixed factor contain clear elements of neurological involvement. Indeed, headache, dizziness, and brain fog all loaded more highly on the Fatigue/Mixed factor than on the Neurological/Psychiatric factor (which was more characterized by disorientation, visual disturbances, delirium, and altered consciousness). This suggests different types of neurological involvement, potentially reflecting neuroinflammation (the Fatigue/Mixed factor) and encephalitis (the Neurological/Psychiatric factor), respectively. It is of note then that both these factors independently predicted subjective cognitive problems. Both inflammation and encephalitis have been proposed as mechanisms through which COVID-19 may impact the brain (Bougakov et al., 2021) and the presence of indications of neuro-inflammation have been found in post-mortem studies (Matschke et al., 2020). It will be an important next step in the investigation to explore whether the neurological and (possible) inflammatory symptom factors explain variance in performance in cognitive tests.

Participants’ experience of ongoing Neurological, Cardiopulmonary/Fatigue, Gastrointestinal/Autoimmune, Mood and Appetite Loss symptom factors all predicted current cognitive symptoms, together explaining around over 55% of variance. Unlike the initial symptom factors, the vast majority of neurological symptoms were contained within the Neurological factor for ongoing symptoms, with only headache and dizziness loading more strongly into the Gastrointestinal/Autoimmune factor. This latter factor was instead more characterized by symptoms associated with systemic illness—potentially endocrine, or reflecting thyroid disruption—including diarrhea, hot flushes and body pains. An additional predictor here was Cardiopulmonary/Fatigue symptoms, a factor which was quite narrowly characterized by symptoms associated with breathing difficulties. Alone, the Gastrointestinal/Autoimmune and Cardiopulmonary/Fatigue factors explained a large proportion of the variance (36%), suggesting these were the biggest contributor to individual differences in cognitive symptoms. These findings suggest that the symptoms linked with cognitive issues are not so specifically neurological as during the initial illness, but may also incorporate problems with heart and lung function (potentially implying hypoxia, which can induce hypoxic/anoxic-related encephalopathy; Guo et al., 2020) and with other ongoing ill health that is harder to label (resembling symptoms of the menopause, Crohn’s disease, hypothyroidism, and a number of other conditions), but may imply systemic inflammation. Again, these associations align with previous findings, in which cardiopulmonary and cognitive systems clustered in the same factor (Ziauddeen et al., 2021).

In terms of current symptoms, the Cardiopulmonary/Fatigue factor again emerged as a significant predictor, this time paired with Neurological and Appetite Loss symptom factors and explaining nearly 50% of variance. It is potentially notable that both the cognitive and Cardiopulmonary/Fatigue factors showed positive correlation with length of illness, suggesting either that the same disease process underpinning both increases in severity over time, or that the relationship between the two may be the result of both being symptoms more commonly still experienced in those with longer-lasting illness. Longitudinal investigation within individuals would be necessary to disambiguate this.

Impact of Long COVID

Of those experiencing Long COVID, more than half (and 75% of those with severe symptoms) reported long periods unable to work due to illness. These findings chime with evidence from other studies on Long COVID (e.g., Davis et al., 2021Ziauddeen et al., 2021). Notably, Davis et al. (2021) found that in their sample 86% of participants reported that it was the cognitive dysfunction in particular that was impacting their work (30% severely so). The reported experiences of those with Long COVID—many of whom were at least 6 months into their illness at the time of completing the study—suggest that in addition to broader economic challenges associated with the pandemic, society will face a long “tail” of workforce morbidity. It is thus of great importance—not just for individuals but for society—to be able to prevent, predict, identify and treat issues associated with Long COVID, and including treatment for cognitive symptoms as part of this policy.

A major roadblock to progress in management and treatment of Long COVID is that clinicians do not have the appropriate information or experience. A significant number (over 50% of those with severe symptoms) of our sample reported struggling to get medical professionals to take their symptoms seriously. Part of this issue will be the nature of the symptoms experienced. Patients whose symptoms cannot be, or are not routinely, clinically measured (such as cognitive symptoms; Kaduszkiewicz et al., 2010) are at greater risk of “testimonial injustice”—that is, having their illness dismissed by medical professionals (De Jesus et al., 2021). The novel and heterogenous nature of Long COVID also provides a particular challenge for clinicians dealing with complex and undifferentiated presentations and “medically unexplained symptoms” (Davidson and Menkes, 2021). The data presented here demonstrate that cognitive difficulties reported by patients can be predicted by severity and pattern of symptoms during the initial stages of infection, and during the ongoing illness. These findings should provide the foundation for clinicians to assess the risk of long-term (6 months +) cognitive difficulties, as well as for researchers to investigate the underlying mechanism driving these deficits. In our next paper, we will explore the association between general and cognitive symptoms and performance on cognitive tasks, with the aim of establishing whether self-reported cognitive issues translate into “objective” deficits on cognitive evaluations.

Some have argued that cognitive changes following COVID-19 infection may reflect changes related to experience of lockdown or social isolation (perhaps via development of depression or anxiety). There is indeed some evidence that pandemic-related changes in lifestyle impact cognition (e.g., Fiorenzato et al., 2021Okely et al., 2021). However, many of these studies did not record COVID-19 infection history (Okely et al., 2021Smirni et al., 2021) so it is difficult to ascertain to what degree these findings may have been related to COVID-19 infection. One study that did control for this (Fiorenzato et al., 2021) identified significant declines in self-reported attention and executive function, however, showed reduced reports of forgetfulness compared with pre-lockdown. Our results show that, compared to individuals who experienced a (probable) non-COVID-19 illness during the pandemic, those with suspected or confirmed COVID-19 infection experienced greater levels of fatigue, difficulty concentrating, brain fog, tip-of-the-tongue (ToT) word finding problems and semantic disfluency, but did not differ in levels of anxiety and depression. Meanwhile there was little difference between those that did and did not have biological confirmation of their COVID-19 infection. This strongly suggests that self-reported cognitive deficits reported in our sample are associated with COVID-19 infection, rather than the experience of illness, or pandemic more generally.

Limitations and Future Research

While the findings of this study are notable, there are a number of limitations in design and execution which warrant caution in interpreting the results.

Being unable to bring participants into the lab for clinical assessment, this study relied on online retrospective self-report of symptoms sometimes experienced some months previously. We thus must be cognizant of potential issues of misremembering and that questionnaires may not have been completed in an environment conducive to concentration and reflection. The manner of reporting symptoms differed between different reporting times, with a longer list and more reporting options (reflecting both severity and regularity) for the “ongoing” period. In particular, our binary present/absent reporting approach for currently experienced symptoms was not able to reflect current severity and did not lend itself to factor analysis. Using the sum scores by factor method (Tabachnick et al., 2007Hair, 2009) to calculate alignment of currently experienced symptoms with the symptom factors got around some of these issues, future studies should keep lists consistent to allow for direct comparison of symptom profiles at the different time points. A similar issue is that symptoms information was not collected for the “No COVID” group, or (in terms of current symptoms) for those that reported having recovered. This would have been highly useful in order to establish the degree to which symptoms (particularly those which might be expected to be exacerbated by lockdowns, such as depression, anxiety, fatigue) were more common in those that had previously experienced COVID-19 than those that had not. It would also be useful to ask both the COVID and No COVID groups about their living situation at the time of completing the study, such as whether lockdown or any social restrictions were taking place and how much these measures were affecting their physical and psychological health. It would also have been useful to assess whether people who reported having “recovered” showed symptomatology similar to the “No COVID” group, or remained distinct.

Due to the intensive performance focus of the current investigation, our study had a relatively smaller sample size than is feasible in an epidemiological cohort. Characterizing the sample, we found that those who had experienced COVID-19 infection—and within these, those with more severe ongoing symptoms—tended to be older and more educated. We do not believe that these features reflect vulnerabilities toward COVID-19 or Long COVID, but rather the biases in our recruitment and target populations. Our sample was recruited from English speaking countries (the United Kingdom, Ireland, United States, Canada, Australia, New Zealand, or South Africa) and the majority were from the United Kingdom, which may not be representative of people from other parts of the world. Where possible, we controlled for age, sex, education, and country of residence, which should mitigate some of these biases, however, these sampling discrepancies should be kept in mind. We furthermore specifically targeted our recruitment to those self-identifying as experiencing Long COVID, and we advertised the study as investigating memory and cognition in this group. Our sample may thus have been biased toward those individuals with more severe symptoms and cognitive symptoms in particular (as these individuals may be more motivated to take part). Overrepresentation of Long COVID sufferers is not a serious issue outside of prevalence studies, however, our reported rates of cognitive symptoms within the Long COVID cohort should be treated with caution. It is reassuring, however, that the figures for these symptoms within our cohort are comparable to those seen in much larger studies not explicitly investigating cognition (e.g., Davis et al., 2021Ziauddeen et al., 2021).

Finally, much of the analysis in this study was necessarily exploratory, as too little was known at the time of study design to form many clear hypotheses. To handle this, multiple comparisons were conducted, for which the alpha adjustments entailed that only the very strongest effects survived at conventional statistical thresholds. This high type 2 error rate means that it is likely that more than just these findings would be confirmed on replication, and because a stated aim of this study was to generate hypotheses that could be tested in later, more targeted research, we have additionally reported the uncorrected results. Similarly, in terms of investigating symptom profiles, we did not aim to present a “definitive” set of factors, but to provide stratifiers and covariates for future analysis, particularly of cognitive test performance, and changes over time. While this study is not able to identify a specific mechanism, it may be able to lay the groundwork with sufficient breadth and detail to inform future mechanistic investigation.

Conclusion

The COVID and Cognition study is a cross-sectional/longitudinal study assessing symptoms, experiences and cognition in those that have experience COVID-19 infection. Here we present the first analysis in this cohort, characterizing the sample and investigating symptom profiles and cognitive symptoms in particular. We find that particular symptom-profiles—particularly neurological symptoms—during both the initial infection and ongoing illness were predictive of experience of cognitive dysfunction. The symptoms and experiences reported by our sample appear to closely resemble those reported in previous work on Long COVID (e.g., Davis et al., 2021Ziauddeen et al., 2021) which suggests that our, smaller, sample might be generally representative of the larger Long COVID patient community. The participants in this study are being followed up over the course of the next 1–2 years, and it is hoped that future publications with this sample will provide valuable information as to the time-course of this illness.

The severity of the impact of “Long COVID” on everyday function and employment reported in our sample appear to reflect previous studies (e.g., Davis et al., 2021) and is notable, particularly given the large proportion of healthcare and education staff in our sample. All of these issues should be of interest to policy makers, particularly when considering the extent to which large case numbers should be a concern in the context of reduced hospitalizations and deaths due to vaccination. While we do not yet know the impact of vaccination on Long COVID numbers, there are reasons to believe that high levels of infection among relatively young, otherwise healthy individuals may translate into considerable long-term workforce morbidity.

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

References

Alberti, P., Beretta, S., Piatti, M., Karantzoulis, A., Piatti, M. L., and Santoro, P. (2020). Guillain-Barré syndrome related to COVID-19 infection. Neurol. Neuroimmunol. Neuroinflam. 7:741.

Google Scholar

Almeida-Suhett, C. P., Graham, A., Chen, Y., and Deuster, P. (2017). Behavioral changes in male mice fed a high-fat diet are associated with IL-1β expression in specific brain regions. Physiol. Behav. 169, 130–140. doi: 10.1016/j.physbeh.2016.11.016

PubMed Abstract | CrossRef Full Text | Google Scholar

Alwan, N. A., and Johnson, L. (2021). Defining long COVID: Going back to the start. Med 2, 501–504. doi: 10.1016/j.medj.2021.03.003

PubMed Abstract | CrossRef Full Text | Google Scholar

Assaf, G., Davis, H., McCorkell, L., Wei, H., Brooke, O., Akrami, A., et al. (2020). An Analysis of the Prolonged COVID-19 Symptoms Survey by Patient-led Research Team. Available online at: https://patientresearchCOVID19.com/research/report-1/ (accessed 15 November 2020).

Google Scholar

Baj, J., Karakuła-Juchnowicz, H., Teresiński, G., Buszewicz, G., Ciesielka, M., Sitarz, E., et al. (2020). COVID-19: specific and non-specific clinical manifestations and symptoms: the current state of knowledge. J. Clin. Med. 9:1753. doi: 10.3390/jcm9061753

PubMed Abstract | CrossRef Full Text | Google Scholar

Beilharz, J., Kaakoush, N., Maniam, J., and Morris, M. (2018). Cafeteria diet and probiotic therapy: cross talk among memory, neuroplasticity, serotonin receptors and gut microbiota in the rat. Mol. Psychiatry 23, 351–361. doi: 10.1038/mp.2017.38

PubMed Abstract | CrossRef Full Text | Google Scholar

Beilharz, J. E., Maniam, J., and Morris, M. J. (2014). Short exposure to a diet rich in both fat and sugar or sugar alone impairs place, but not object recognition memory in rats. Brain Behav. Immun. 37, 134–141. doi: 10.1016/j.bbi.2013.11.016

PubMed Abstract | CrossRef Full Text | Google Scholar

Belarbi, K., Arellano, C., Ferguson, R., Jopson, T., and Rosi, S. (2012). Chronic neuroinflammation impacts the recruitment of adult-born neurons into behaviorally relevant hippocampal networks. Brain Behav. Immun. 26, 18–23. doi: 10.1016/j.bbi.2011.07.225

PubMed Abstract | CrossRef Full Text | Google Scholar

Beyrouti, R., Adams, M. E., Benjamin, L., Cohen, H., Farmer, S. F., Goh, Y. Y., et al. (2020). Characteristics of ischaemic stroke associated with COVID-19. J. Neurol. Neurosurg. Psychiatry 91, 889–891. doi: 10.1136/jnnp-2020-323586

PubMed Abstract | CrossRef Full Text | Google Scholar

Bliddal, S., Banasik, K., Pedersen, O. B., Nissen, J., Cantwell, L., Schwinn, M., et al. (2021). Acute and persistent symptoms in non-hospitalized PCR-confirmed COVID-19 patients. Sci. Rep. 11, 1–11. doi: 10.1038/s41598-021-92045-x

PubMed Abstract | CrossRef Full Text | Google Scholar

Bossù, P., Toppi, E., Sterbini, V., and Spalletta, G. (2020). Implication of aging related chronic neuroinflammation on COVID-19 pandemic. J. Personal. Med. 10:102. doi: 10.3390/jpm10030102

PubMed Abstract | CrossRef Full Text | Google Scholar

Bougakov, D., Podell, K., and Goldberg, E. (2021). Multiple neuroinvasive pathways in COVID-19. Mol. Neurobiol. 58, 564–575. doi: 10.1007/s12035-020-02152-5

PubMed Abstract | CrossRef Full Text | Google Scholar

Callard, F., and Perego, E. (2021). How and why patients made Long Covid. Soc. Sci. Med. 268:113426.

Google Scholar

Che, Y.-Y., Xia, X.-J., He, B.-P., Gao, Y.-Y., Ren, W.-B., Liu, H.-T., et al. (2018). A corn straw-based diet increases release of inflammatory cytokines in peripheral blood mononuclear cells of dairy cows. J. Zhejiang Univ. Sci. B 19, 796–806. doi: 10.1631/jzus.B1700571

PubMed Abstract | CrossRef Full Text | Google Scholar

Chen, W. W., Zhang, X., and Huang, W. J. (2016). Role of neuroinflammation in neurodegenerative diseases. Mol. Med. Rep. 13, 3391–3396. doi: 10.3892/mmr.2016.4948

PubMed Abstract | CrossRef Full Text | Google Scholar

Cirulli, E., Barrett, K. M. S., Riffle, S., Bolze, A., Neveux, I., Dabe, S., et al. (2020). Long-term COVID-19 symptoms in a large unselected population. medrxiv doi: 10.1101/2020.10.07.20208702

CrossRef Full Text | Google Scholar

Das, G., Mukherjee, N., and Ghosh, S. (2020). Neurological insights of COVID-19 pandemic. ACS Chem. Neurosci. 11, 1206–1209.

Google Scholar

Davidson, S. L., and Menkes, D. B. (2021). Long covid: reshaping conversations about medically unexplained symptoms. BMJ 374:1857. doi: 10.1136/bmj.n1859

PubMed Abstract | CrossRef Full Text | Google Scholar

Davis, H. E., Assaf, G. S., Mccorkell, L., Wei, H., Low, R. J., Re’em, Y., et al. (2021). Characterizing long COVID in an international cohort: 7 months of symptoms and their impact. EClinicalMedicine 38:101019 doi: 10.1016/j.eclinm.2021.101019

PubMed Abstract | CrossRef Full Text | Google Scholar

De Jesus, V. V. A., Alwan, N., Callard, F., and Berger, Z. (2021). Listening to Long COVID: Epistemic Injustice and COVID-19 morbidity. medrxiv doi: 10.31219/osf.io/tfbnd

CrossRef Full Text | Google Scholar

Ding, X., Xu, J., Zhou, J., and Long, Q. (2020). Chest CT findings of COVID-19 pneumonia by duration of symptoms. Eur. J. Radiol. 127:109009. doi: 10.1016/j.ejrad.2020.109009

PubMed Abstract | CrossRef Full Text | Google Scholar

Douaud, G., Lee, S., Alfaro-Almagro, F., Arthofer, C., Wang, C., Lange, F., et al. (2021). Brain imaging before and after COVID-19 in UK Biobank. medRxiv doi: 10.1101/2021.06.11.21258690

PubMed Abstract | CrossRef Full Text | Google Scholar

Ekdahl, C. T., Claasen, J.-H., Bonde, S., Kokaia, Z., and Lindvall, O. (2003). Inflammation is detrimental for neurogenesis in adult brain. Proc. Nat. Acad. Sci. 100, 13632–13637. doi: 10.1073/pnas.2234031100

PubMed Abstract | CrossRef Full Text | Google Scholar

FAIR Health (2021). A Detailed Study of Patients with Long-Haul COVID. Available online at: https://s3.amazonaws.com/media2.fairhealth.org/whitepaper/asset/A%20Detailed%20Study%20of%20Patients%20with%20Long-Haul%20COVID–An%20Analysis%20of%20Private%20Healthcare%20Claims–A%20FAIR%20Health%20White%20Paper.pdf

Google Scholar

Fiorenzato, E., Zabberoni, S., Costa, A., and Cona, G. (2021). Cognitive and mental health changes and their vulnerability factors related to COVID-19 lockdown in Italy. PLoS One 16:e0246204. doi: 10.1371/journal.pone.0246204

PubMed Abstract | CrossRef Full Text | Google Scholar

Galanopoulou, A. S., Ferastraoaru, V., Correa, D. J., Cherian, K., Duberstein, S., Gursky, J., et al. (2020). EEG findings in acutely ill patients investigated for SARS-CoV-2/COVID-19: a small case series preliminary report. Epilepsia Open 5, 314–324. doi: 10.1002/epi4.12399

PubMed Abstract | CrossRef Full Text | Google Scholar

Guo, Y.-R., Cao, Q.-D., Hong, Z.-S., Tan, Y.-Y., Chen, S.-D., Jin, H.-J., et al. (2020). The origin, transmission and clinical therapies on coronavirus disease 2019 (COVID-19) outbreak–an update on the status. Milit. Med. Res. 7, 1–10. doi: 10.1186/s40779-020-00240-0

PubMed Abstract | CrossRef Full Text | Google Scholar

Hair, J. F. (2009). Multivariate data analysis. New York, NY: Springer. doi: 10.1002/9781118887486.ch1

CrossRef Full Text | Google Scholar

Helms, J., Kremer, S., Merdji, H., Clere-Jehl, R., Schenck, M., Kummerlen, C., et al. (2020). Neurologic features in severe SARS-CoV-2 infection. N. Engl. J. Med. 382, 2268–2270. doi: 10.1056/NEJMc2008597

PubMed Abstract | CrossRef Full Text | Google Scholar

Heneka, M. T., Golenbock, D., Latz, E., Morgan, D., and Brown, R. (2020). Immediate and long-term consequences of COVID-19 infections for the development of neurological disease. Alzheimer’s Res. Ther. 12, 1–3. doi: 10.1186/s13195-020-00640-3

PubMed Abstract | CrossRef Full Text | Google Scholar

Henson, R. K., and Roberts, J. K. (2006). Use of exploratory factor analysis in published research: Common errors and some comment on improved practice. Edu. Psychol. Measure. 66, 393–416.

Google Scholar

Hosp, J. A., Dressing, A., Blazhenets, G., Bormann, T., Rau, A., Schwabenland, M., et al. (2021). Cognitive impairment and altered cerebral glucose metabolism in the subacute stage of COVID-19. Brain 144, 1263–1276. doi: 10.1093/brain/awab009

PubMed Abstract | CrossRef Full Text | Google Scholar

Jain, R. (2020). Evolving neuroimaging findings during COVID-19. Am. Soc. Neuroradiol. 41, 1355–1356. doi: 10.3174/ajnr.A6658

PubMed Abstract | CrossRef Full Text | Google Scholar

Jakubs, K., Bonde, S., Iosif, R. E., Ekdahl, C. T., Kokaia, Z., Kokaia, M., et al. (2008). Inflammation regulates functional integration of neurons born in adult brain. J. Neurosci. 28, 12477–12488. doi: 10.1523/JNEUROSCI.3240-08.2008

PubMed Abstract | CrossRef Full Text | Google Scholar

Kaduszkiewicz, H., Zimmermann, T., Van Den Bussche, H., Bachmann, C., Wiese, B., Bickel, H., et al. (2010). Do general practitioners recognize mild cognitive impairment in their patients? J. Nutrit. Health Aging 14, 697–702. doi: 10.1007/s12603-010-0038-5

PubMed Abstract | CrossRef Full Text | Google Scholar

Kandemirli, S. G., Dogan, L., Sarikaya, Z. T., Kara, S., Akinci, C., Kaya, D., et al. (2020). Brain MRI findings in patients in the intensive care unit with COVID-19 infection. Radiology 297, E232–E235. doi: 10.1148/radiol.2020201697

PubMed Abstract | CrossRef Full Text | Google Scholar

Kheirouri, S., and Alizadeh, M. (2019). Dietary inflammatory potential and the risk of neurodegenerative diseases in adults. Epidemiol. Rev. 41, 109–120. doi: 10.1093/epirev/mxz005

PubMed Abstract | CrossRef Full Text | Google Scholar

Kline, P. (2013). Handbook of psychological testing. Milton Park: Routledge.

Google Scholar

Kubánková, M., Hohberger, B., Hoffmanns, J., Fürst, J., Herrmann, M., Guck, J., et al. (2021). Physical phenotype of blood cells is altered in COVID-19. Biophys. J. 120, 2838–2847. doi: 10.1016/j.bpj.2021.05.025

PubMed Abstract | CrossRef Full Text | Google Scholar

Le Guennec, L., Devianne, J., Jalin, L., Cao, A., Galanaud, D., Navarro, V., et al. (2020). Orbitofrontal involvement in a neuroCOVID-19 patient. Epilepsia 61, e90–e94. doi: 10.1111/epi.16612

PubMed Abstract | CrossRef Full Text | Google Scholar

Lechien, J. R., Chiesa-Estomba, C. M., De Siati, D. R., Horoi, M., Le Bon, S. D., Rodriguez, A., et al. (2020). Olfactory and gustatory dysfunctions as a clinical presentation of mild-to-moderate forms of the coronavirus disease (COVID-19): a multicenter European study. Eur. Arch. Oto-Rhino-Laryngol. 277, 2251–2261. doi: 10.1007/s00405-020-05965-1

PubMed Abstract | CrossRef Full Text | Google Scholar

Li, Y., Li, M., Wang, M., Zhou, Y., Chang, J., Xian, Y., et al. (2020). Acute cerebrovascular disease following COVID-19: a single center, retrospective, observational study. Stroke Vascul. Neurol. 5:431. doi: 10.1136/svn-2020-000431

PubMed Abstract | CrossRef Full Text | Google Scholar

Mao, L., Jin, H., Wang, M., Hu, Y., Chen, S., He, Q., et al. (2020). Neurologic manifestations of hospitalized patients with coronavirus disease 2019 in Wuhan. China JAMA Neurol. 77, 683–690. doi: 10.1001/jamaneurol.2020.1127

PubMed Abstract | CrossRef Full Text | Google Scholar

Matschke, J., Lütgehetmann, M., Hagel, C., Sperhake, J. P., Schröder, A. S., Edler, C., et al. (2020). Neuropathology of patients with COVID-19 in Germany: a post-mortem case series. Lancet Neurol. 19, 919–929. doi: 10.1016/S1474-4422(20)30308-2

PubMed Abstract | CrossRef Full Text | Google Scholar

McGeer, E. G., and McGeer, P. L. (2010). Neuroinflammation in Alzheimer’s disease and mild cognitive impairment: a field in its infancy. J. Alzheimer’s Dis. 19, 355–361. doi: 10.3233/JAD-2010-1219

PubMed Abstract | CrossRef Full Text | Google Scholar

Mehta, P., Mcauley, D. F., Brown, M., Sanchez, E., Tattersall, R. S., and Manson, J. J. (2020). COVID-19: consider cytokine storm syndromes and immunosuppression. Lancet 395, 1033–1034. doi: 10.1016/S0140-6736(20)30628-0

PubMed Abstract | CrossRef Full Text | Google Scholar

Mirzaei, F., Khazaei, M., Komaki, A., Amiri, I., and Jalili, C. (2018). Virgin coconut oil (VCO) by normalizing NLRP3 inflammasome showed potential neuroprotective effects in Amyloid-β induced toxicity and high-fat diet fed rat. Food Chem. Toxicol. 118, 68–83. doi: 10.1016/j.fct.2018.04.064

PubMed Abstract | CrossRef Full Text | Google Scholar

Monje, M. L., Toda, H., and Palmer, T. D. (2003). Inflammatory blockade restores adult hippocampal neurogenesis. Science 302, 1760–1765. doi: 10.1126/science.1088417

PubMed Abstract | CrossRef Full Text | Google Scholar

Moriguchi, T., Harii, N., Goto, J., Harada, D., Sugawara, H., Takamino, J., et al. (2020). A first case of meningitis/encephalitis associated with SARS-Coronavirus-2. Int. J. Infect. Dis. 94, 55–58. doi: 10.1016/j.ijid.2020.03.062

PubMed Abstract | CrossRef Full Text | Google Scholar

National Institute for Health and Care Excellence [NICE] (2020). COVID-19 Rapid Guideline: Managing the Long-Term Effects of COVID-19. London: National Institute for Health and Care Excellence.

Google Scholar

Nehme, M., Braillard, O., Alcoba, G., Aebischer Perone, S., Courvoisier, D., Chappuis, F., et al. (2021). COVID-19 symptoms: longitudinal evolution and persistence in outpatient settings. Ann. Int. Med. 174, 723–725. doi: 10.7326/M20-5926

PubMed Abstract | CrossRef Full Text | Google Scholar

Nunnally, J. C. (1978). Psychometric Theory, 2nd Edn. New York, NY: McGraw-Hill.

Google Scholar

Office for National Statistics [ONS] (2021). Prevalence of Ongoing Symptoms Following Coronavirus (COVID-19) Infection in the UK: 1 April 2021. Available online at: https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/bulletins/prevalenceofongoingsymptomsfollowingcoronavirusCOVID19infectionintheuk/1april2021 (accessed April 1, 2021).

Google Scholar

Okely, J. A., Corley, J., Welstead, M., Taylor, A. M., Page, D., Skarabela, B., et al. (2021). Change in physical activity, sleep quality, and psychosocial variables during COVID-19 lockdown: Evidence from the Lothian Birth Cohort 1936. Int. J. Environ. Res. Public Health 18:210. doi: 10.3390/ijerph18010210

PubMed Abstract | CrossRef Full Text | Google Scholar

Politi, L. S., Salsano, E., and Grimaldi, M. (2020). Magnetic resonance imaging alteration of the brain in a patient with coronavirus disease 2019 (COVID-19) and anosmia. JAMA Neurol. 77, 1028–1029. doi: 10.1001/jamaneurol.2020.2125

PubMed Abstract | CrossRef Full Text | Google Scholar

Poyiadji, N., Shahin, G., Noujaim, D., Stone, M., Patel, S., and Griffith, B. (2020). COVID-19–associated acute hemorrhagic necrotizing encephalopathy: imaging features. Radiology 296, E119–E120. doi: 10.1148/radiol.2020201187

PubMed Abstract | CrossRef Full Text | Google Scholar

Pryce-Roberts, A., Talaei, M., and Robertson, N. (2020). Neurological complications of COVID-19: a preliminary review. J. Neurol. 267, 1870–1873. doi: 10.1007/s00415-020-09941-x

PubMed Abstract | CrossRef Full Text | Google Scholar

Romero-Sánchez, C. M., Díaz-Maroto, I., Fernández-Díaz, E., Sánchez-Larsen, A., Layos-Romero, A., García-García, J., et al. (2020). Neurologic manifestations in hospitalized patients with COVID-19: the ALBACOVID registry. Neurology 95, e1060–e1070. doi: 10.1212/WNL.0000000000009937

PubMed Abstract | CrossRef Full Text | Google Scholar

Smirni, D., Garufo, E., Di Falco, L., and Lavanco, G. (2021). The playing brain. the impact of video games on cognition and behavior in pediatric age at the time of lockdown: a systematic review. Pediatric Rep. 13, 401–415. doi: 10.3390/pediatric13030047

PubMed Abstract | CrossRef Full Text | Google Scholar

Steiger, J. H. (1980). Tests for comparing elements of a correlation matrix. Psychol. Bull. 87:245.

Google Scholar

Sudre, C. H., Murray, B., Varsavsky, T., Graham, M. S., Penfold, R. S., Bowyer, R. C., et al. (2020). Attributes and predictors of Long-COVID: analysis of COVID cases and their symptoms collected by the Covid Symptoms Study App. Medrxiv doi: 10.1101/2020.10.19.20214494

CrossRef Full Text | Google Scholar

Tabachnick, B. G., Fidell, L. S., and Ullman, J. B. (2007). Using multivariate statistics. Boston, MA: Pearson.

Google Scholar

Tay, M. Z., Poh, C. M., Rénia, L., Macary, P. A., and Ng, L. F. (2020). The trinity of COVID-19: immunity, inflammation and intervention. Nat. Rev. Immunol. 20, 363–374. doi: 10.1038/s41577-020-0311-8

PubMed Abstract | CrossRef Full Text | Google Scholar

Tenforde, M. W., Kim, S. S., Lindsell, C. J., Rose, E. B., Shapiro, N. I., Files, D. C., et al. (2020). Symptom duration and risk factors for delayed return to usual health among outpatients with COVID-19 in a multistate health care systems network—United States. Morb. Mortal. Week. Rep. 69:993.

Google Scholar

Thirumangalakudi, L., Prakasam, A., Zhang, R., BimonteNelson, H., Sambamurti, K., Kindy, M. S., et al. (2008). High cholesterol-induced neuroinflammation and amyloid precursor protein processing correlate with loss of working memory in mice. J. Neurochem. 106, 475–485. doi: 10.1111/j.1471-4159.2008.05415.x

PubMed Abstract | CrossRef Full Text | Google Scholar

Wang, Z., Yang, Y., Liang, X., Gao, B., Liu, M., Li, W., et al. (2020). COVID-19 associated ischemic stroke and hemorrhagic stroke: incidence, potential pathological mechanism, and management. Front. Neurol. 11:1152. doi: 10.3389/fneur.2020.571996

PubMed Abstract | CrossRef Full Text | Google Scholar

Whitaker, M., Elliott, J., Chadeau-Hyam, M., Riley, S., Darzi, A., Cooke, G., et al. (2021). Persistent symptoms following SARS-CoV-2 infection in a random community sample of 508,707 people. medRxiv doi: 10.1101/2021.06.28.21259452

CrossRef Full Text | Google Scholar

Whittaker, A., Anson, M., and Harky, A. (2020). Neurological manifestations of COVID-19: a systematic review and current update. Acta Neurol. Scand. 142, 14–22. doi: 10.1111/ane.13266

PubMed Abstract | CrossRef Full Text | Google Scholar

Zhao, H., Shen, D., Zhou, H., Liu, J., and Chen, S. (2020). Guillain-Barré syndrome associated with SARS-CoV-2 infection: causality or coincidence? Lancet Neurol. 19, 383–384.

Google Scholar

Ziauddeen, N., Gurdasani, D., O’hara, M. E., Hastie, C., Roderick, P., Yao, G., et al. (2021). Characteristics of Long Covid: findings from a social media survey. medRxiv doi: 10.1101/2021.03.21.21253968

CrossRef Full Text | Google Scholar

Zotova, E., Nicoll, J. A., Kalaria, R., Holmes, C., and Boche, D. (2010). Inflammation in Alzheimer’s disease: relevance to pathogenesis and therapy. Alzheimers Res. Ther. 2, 1–9. doi: 10.1186/alzrt24

PubMed Abstract | CrossRef Full Text | Google Scholar

7 in 10 long COVID patients are dealing with memory and concentration problems

Authors: Study Finds MARCH 17, 2022

The vast majority of people dealing with “long COVID” are experiencing memory and concentration problems — months after their actual coronavirus infection, a new study warns. Researchers at the University of Cambridge say seven in 10 people experiencing the lingering effects of COVID are now struggling mentally.

The study finds long COVID patients are also performing worse on cognitive exams. Moreover, three in four people with a severe case of long COVID say they have been unable to work because of it.

The team also found a link between the severity of symptoms and how much fatigue, dizziness, and headache pain patients experienced during their initial bout with the virus. Worryingly, half of long COVID sufferers claim they’ve struggled to get doctors to take their condition seriously.

Long COVID has received very little attention politically or medically. It urgently needs to be taken more seriously, and cognitive issues are an important part of this. When politicians talk about ‘Living with COVID’ – that is, unmitigated infection, this is something they ignore. The impact on the working population could be huge,” says study senior author Dr. Lucy Cheke in a university release.

“People think that long COVID is ‘just’ fatigue or a cough, but cognitive issues are the second most common symptom – and our data suggest this is because there is a significant impact on the ability to remember.”

Long COVID patients dealing with brain fog, forgetfulness

Researchers say there is growing evidence that COVID-19 impacts the brain, with multiple studies likening its impact to Alzheimer’s disease.

“Infection with the virus that causes COVID-19 can lead to inflammation in the body, and this inflammation can affect behavior and cognitive performance in ways we still don’t fully understand, but we think are related to an early excessive immune response,” says Dr. Muzaffer Kaser.

“It’s important that people seek help if they’re concerned about any persistent symptoms after COVID infection. COVID can affect multiple systems and further assessment is available in long COVID clinics across the UK, following a GP referral.”

Of the 181 people who took part in the study, 78 percent reported difficulty concentrating, 69 percent said they experienced “brain fog,” 68 percent had moments of forgetfulness, and three in five had problems finding the right words while speaking. These self-reported symptoms were confirmed by the significantly lower ability among long COVID sufferers to remember words and pictures in cognitive tests.

Severe cases of COVID leading to more cognitive issues

During the study, participants took part in several tasks to assess their decision-making abilities and memory. These included remembering words in a list and remembering which two images appeared together. Results revealed a consistent pattern of ongoing memory problems in those who previously suffered a coronavirus infection.

Study authors say these problems were more pronounced in people whose overall ongoing symptoms were more severe. The researchers investigated other symptoms that could have a link to long COVID to help them pinpoint their causes.

They found people who experienced fatigue and neurological symptoms, such as dizziness and headache, during their initial illness were more likely to have cognitive symptoms later on. They also found that those who were still experiencing neurological symptoms particularly struggled on cognitive tests.

Results show that, even among people who did not need to go to the hospital, those with worse initial symptoms of COVID-19 were more likely to have a variety of ongoing long COVID symptoms including nausea, abdominal pain, chest tightness, and breathing issues weeks and months later. Those symptoms were likely to be more severe than in people whose initial illness was mild.

‘A huge impact on my life’

Study authors also found that people over 30 were more likely to have severe ongoing symptoms than younger COVID patients. The findings are of particular concern given the prevalence of long COVID, which health experts estimate could affect between 10 and 25 percent of people who test positive for COVID.

“Having been fit and active all my life, after catching COVID-19 during the first wave, my son (then 13) and I didn’t seem to recover. We were left with debilitating fatigue and a confusing mix of strange and life changing symptoms. I was also left with significant neurological symptoms, including speech and language issues, which had a huge impact on my life,” explains long COVID patient Lyn Curtis.

“My other children also experienced significant ongoing symptoms every time we were re-infected, such as changes to periods, fatigue, insomnia, changes in mood, nausea, vomiting, diarrhea, and nose bleeds,” Curtis continues. “The acknowledgement of long COVID and a greater understanding of the associated symptoms is essential both for identifying treatments and the management of existing symptoms. The work into the effects on cognition are especially important to me, as this is the ongoing symptom that impacts the most on my quality of life and ability to work.”

The researchers add long COVID is causing and will continue to cause high rates of workplace absences and disruptions to society. They say it is important not just for sufferers themselves but for society as a whole to understand what causes the condition and how to treat it.

The findings are published in the journal Frontiers in Aging Neuroscience.

6-month neurological and psychiatric outcomes in 236,379 survivors of COVID-19: a retrospective cohort study using electronic health records

Authors: Maxime Taquet, John R Geddes, Masud Husain, Sierra Luciano, Paul J Harrison

Summary

Background Neurological and psychiatric sequelae of COVID-19 have been reported, but more data are needed to adequately assess the effects of COVID-19 on brain health. We aimed to provide robust estimates of incidence rates and relative risks of neurological and psychiatric diagnoses in patients in the 6 months following a COVID-19 diagnosis. Methods For this retrospective cohort study and time-to-event analysis, we used data obtained from the TriNetX electronic health records network (with over 81 million patients). Our primary cohort comprised patients who had a COVID-19 diagnosis; one matched control cohort included patients diagnosed with influenza, and the other matched control cohort included patients diagnosed with any respiratory tract infection including influenza in the same period. Patients with a diagnosis of COVID-19 or a positive test for SARS-CoV-2 were excluded from the control cohorts. All cohorts included patients older than 10 years who had an index event on or after Jan 20, 2020, and who were still alive on Dec 13, 2020. We estimated the incidence of 14 neurological and psychiatric outcomes in the 6 months after a confirmed diagnosis of COVID-19: intracranial haemorrhage; ischaemic stroke; parkinsonism; Guillain-Barré syndrome; nerve, nerve root, and plexus disorders; myoneural junction and muscle disease; encephalitis; dementia; psychotic, mood, and anxiety disorders (grouped and separately); substance use disorder; and insomnia. Using a Cox model, we compared incidences with those in propensity score-matched cohorts of patients with influenza or other respiratory tract infections. We investigated how these estimates were affected by COVID-19 severity, as proxied by hospitalisation, intensive therapy unit (ITU) admission, and encephalopathy (delirium and related disorders). We assessed the robustness of the differences in outcomes between cohorts by repeating the analysis in different scenarios. To provide benchmarking for the incidence and risk of neurological and psychiatric sequelae, we compared our primary cohort with four cohorts of patients diagnosed in the same period with additional index events: skin infection, urolithiasis, fracture of a large bone, and pulmonary embolism. Findings Among 236 379 patients diagnosed with COVID-19, the estimated incidence of a neurological or psychiatric diagnosis in the following 6 months was 33·62% (95% CI 33·17–34·07), with 12·84% (12·36–13·33) receiving their first such diagnosis. For patients who had been admitted to an ITU, the estimated incidence of a diagnosis was 46·42% (44·78–48·09) and for a first diagnosis was 25·79% (23·50–28·25). Regarding individual diagnoses of the study outcomes, the whole COVID-19 cohort had estimated incidences of 0·56% (0·50–0·63) for intracranial haemorrhage, 2·10% (1·97–2·23) for ischaemic stroke, 0·11% (0·08–0·14) for parkinsonism, 0·67% (0·59–0·75) for dementia, 17·39% (17·04–17·74) for anxiety disorder, and 1·40% (1·30–1·51) for psychotic disorder, among others. In the group with ITU admission, estimated incidences were 2·66% (2·24–3·16) for intracranial haemorrhage, 6·92% (6·17–7·76) for ischaemic stroke, 0·26% (0·15–0·45) for parkinsonism, 1·74% (1·31–2·30) for dementia, 19·15% (17·90–20·48) for anxiety disorder, and 2·77% (2·31–3·33) for psychotic disorder. Most diagnostic categories were more common in patients who had COVID-19 than in those who had influenza (hazard ratio [HR] 1·44, 95% CI 1·40–1·47, for any diagnosis; 1·78, 1·68–1·89, for any first diagnosis) and those who had other respiratory tract infections (1·16, 1·14–1·17, for any diagnosis; 1·32, 1·27–1·36, for any first diagnosis). As with incidences, HRs were higher in patients who had more severe COVID-19 (eg, those admitted to ITU compared with those who were not: 1·58, 1·50–1·67, for any diagnosis; 2·87, 2·45–3·35, for any first diagnosis). Results were robust to various sensitivity analyses and benchmarking against the four additional index health events. Interpretation Our study provides evidence for substantial neurological and psychiatric morbidity in the 6 months after COVID-19 infection. Risks were greatest in, but not limited to, patients who had severe COVID-19. This information could help in service planning and identification of research priorities. Complementary study designs, including prospective cohorts, are needed to corroborate and explain these findings. Funding National Institute for Health Research (NIHR) Oxford Health Biomedical Research Centre. Copyright © 2021 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license. Articles www.thelancet.com/psychiatry Vol 8 May 2021 417 Introduction Since the COVID-19 pandemic began on March 11, 2020, there has been concern that survivors might be at an increased risk of neurological disorders. This concern, initially based on findings from other coronaviruses,1 was followed rapidly by case series,2–4 emerging evidence of COVID-19 CNS involvement,5–7 and the identification of mechanisms by which this could occur.8–11 Similar concerns have been raised regarding psychiatric sequelae of COVID-19,12,13 with evidence showing that survivors are indeed at increased risk of mood and anxiety disorders in the 3 months after infection.14 However, we need large scale, robust, and longer term data to properly identify and quantify the consequences of the COVID-19 pandemic on brain health. Such information is required both to plan services and identify research priorities. In this study, we used an electronic health records network to investigate the incidence of neurological and psychiatric diagnoses in survivors in the 6 months after documented clinical COVID-19 infection, and we compared the associated risks with those following other health conditions. We explored whether the severity of COVID-19 infection, as proxied by hospitalization, intensive therapy unit (ITU) admission, and encephalopathy, affects these risks. We also assessed the trajectory of hazard ratios (HRs) across the 6-month period. Methods Study design and data collection For this retrospective cohort study, we used The TriNetX Analytics Network, a federated network recording anonymized data from electronic health records in 62 health-care organizations, primarily in the USA, comprising 81 million patients. Available data include demographics, diagnoses (using codes from ICD-10), medications, procedures, and measurements (eg, blood pressure and body-mass index). The health-care organizations are a mixture of hospitals, primary care, and specialist providers, contributing data from uninsured and insured patients. These organizations warrant that they have all necessary rights, consents, approvals, and authority to provide the data to TriNetX, so long as their name remains anonymous as a data source and their data are used for research purposes. By use of the TriNetX user interface, cohorts can be created on the basis of inclusion and exclusion criteria, matched for confounding variables with a built-in propensity score-matching algorithm, and compared for outcomes of interest over specified time periods. Additional details about TriNetX, its data, provenance, and functionalities, are presented in the appendix (pp 1–2). Cohorts The primary cohort was defined as all patients who had a confirmed diagnosis of COVID-19 (ICD-10 code U07.1). We also constructed two matched control cohorts: patients diagnosed with influenza (ICD-10 codes J09–11) and patients diagnosed with any respiratory tract infection including influenza (ICD-10 codes J00–06, J09–18, or J20–22). We excluded patients with a diagnosis of COVID-19 or a positive test for SARS-CoV-2 from the control cohorts. We refer to the diagnosis of COVID-19 (in the primary cohort) and influenza or other respiratory See Online for appendix For the TriNetX Analytics Network see www.trinetx.com Research in context Evidence before this study We searched Web of Science and Medline on Aug 1 and Dec 31, 2020, for studies in English, with the terms “(COVID-19 OR SARS-CoV2 OR SARS-CoV-2) AND (psychiatry* or neurology*) AND (incidence OR epidemiology* OR ‘systematic review’ or ‘meta-analysis’)”. We found case series and reviews of series reporting neurological and neuropsychiatric disorders during acute COVID-19 illness. We found one large electronic health records study of the psychiatric sequelae in the 3 months after a COVID-19 diagnosis. It reported an increased risk for anxiety and mood disorders and dementia after COVID-19 compared with a range of other health events; the study also reported the incidence of each disorder. We are not aware of any large-scale data regarding the incidence or relative risks of neurological diagnoses in patients who had recovered from COVID-19. Added value of this study To our knowledge, we provide the first meaningful estimates of the risks of major neurological and psychiatric conditions in the 6 months after a COVID-19 diagnosis, using the electronic health records of over 236000 patients with COVID-19. We report their incidence and hazard ratios compared with patients who had had influenza or other respiratory tract infections. We show that both incidence and hazard ratios were greater in patients who required hospitalization or admission to the intensive therapy unit (ITU), and in those who had encephalopathy (delirium and other altered mental states) during the illness compared with those who did not. Implications of all the available evidence COVID-19 was robustly associated with an increased risk of neurological and psychiatric disorders in the 6 months after a diagnosis. Given the size of the pandemic and the chronicity of many of the diagnoses and their consequences (eg, dementia, stroke, and intracranial hemorrhage), substantial effects on health and social care systems are likely to occur. Our data provide important evidence indicating the scale and nature of services that might be required. The findings also highlight the need for enhanced neurological follow-up of patients who were admitted to ITU or had encephalopathy during their COVID-19 illness. Articles 418 www.thelancet.com/psychiatry Vol 8 May 2021 tract infections (in the control cohorts) as index events. The cohorts included all patients older than 10 years who had an index event on or after Jan 20, 2020 (the date of the first recorded COVID-19 case in the USA), and who were still alive at the time of the main analysis (Dec 13, 2020). Additional details on cohorts are provided in the appendix (pp 2–3). Covariates We used a set of established and suspected risk factors for COVID-19 and for more severe COVID-19 illness:15,16 age, sex, race, ethnicity, obesity, hypertension, diabetes, chronic kidney disease, asthma, chronic lower respiratory diseases, nicotine dependence, substance use disorder, ischemic heart disease and other forms of heart disease, socioeconomic deprivation, cancer (and hematological cancer in particular), chronic liver disease, stroke, dementia, organ transplant, rheumatoid arthritis, lupus, psoriasis, and disorders involving an immune mechanism. To capture these risk factors in patients’ health records, we used 55 variables. More details, including ICD-10 codes, are provided in the appendix (pp 3–4). Cohorts were matched for all these variables, as described in the following subsections. Outcomes We investigated neurological and psychiatric sequelae of COVID-19 in terms of 14 outcomes occurring 1–180 days after the index event: intracranial hemorrhage (ICD-10 codes I60–62); ischemic stroke (I63); Parkinson’s disease and parkinsonism (G20–21); Guillain-Barré syndrome (G61.0); nerve, nerve root, and plexus disorders (G50–59); myoneural junction and muscle disease (neuromuscular disorders; G70–73); encephalitis (G04, G05, A86, or A85.8); dementia (F01–03, G30, G31.0, or G31.83); psychotic, mood, and anxiety disorders (F20–48), as well as each category separately; substance use disorder (F10–19), and insomnia (F51.0 or G47.0). For outcomes that are chronic illnesses (eg, dementia or Parkinson’s disease), we excluded patients who had the diagnosis before the index event. For outcomes that All patients Patients without hospitalization Patients with hospitalization Patients with ITU admission Patients with encephalopathy Cohort size 236379 (100·0%) 190077 (100·0%) 46302 (100·0%) 8945 (100·0%) 6229 (100·0%) Demographics Age, years 46 (19·7) 43·3 (19·0) 57 (18·7) 59·1 (17·3) 66·7 (17·0) Sex Male 104015 (44·0%) 81 512 (42·9%) 22 503 (48·6%) 5196 (58·1%) 3307 (53·1%) Female 131460 (55·6%) 107 730 (56·7%) 23 730 (51·3%) 3743 (41·8%) 2909 (46·7%) Other 904 (0·4%) 835 (0·4%) 69 (0·1%) 10 (0·1%) 13 (0·2%) Race White 135 143 (57·2%) 109635 (57·7%) 25 508 (55·1%) 4918 (55·0%) 3331 (53·5%) Black or African American 44459 (18·8%) 33868 (17·8%) 10591 (22·9%) 2184 (24·4%) 1552 (24·9%) Unknown 48085 (20·3%) 39841 (21·0%) 8244 (17·8%) 1457 (16·3%) 1071 (17·2%) Ethnicity Hispanic or Latino 37 772 (16·0%) 29155 (15·3%) 8617 (18·6%) 2248 (25·1%) 895 (14·4%) Not Hispanic or Latino 134075 (56·7%) 106844 (56·2%) 27 231 (58·8%) 5041 (56·4%) 3873 (62·2%) Unknown 64532 (27·3%) 54078 (28·5%) 10454 (22·6%) 1656 (18·5%) 1461 (23·5%) Comorbidities Overweight and obesity 42871 (18·1%) 30198 (15·9%) 12673 (27·4%) 3062 (34·2%) 1838 (29·5%) Hypertensive disease 71014 (30·0%) 47 516 (25·0%) 23498 (50·7%) 5569 (62·3%) 4591 (73·7%) Type 2 diabetes 36696 (15·5%) 22 518 (11·8%) 14178 (30·6%) 3787 (42·3%) 2890 (46·4%) Asthma 25 104 (10·6%) 19834 (10·4%) 5270 (11·4%) 1132 (12·7%) 755 (12·1%) Nicotine dependence 17 105 (7·2%) 12639 (6·6%) 4466 (9·6%) 1042 (11·6%) 803 (12·9%) Substance use disorder 24870 (10·5%) 18173 (9·6%) 6697 (14·5%) 1620 (18·1%) 1316 (21·1%) Ischemic heart diseases 21082 (8·9%) 11815 (6·2%) 9267 (20·0%) 2460 (27·5%) 2200 (35·3%) Other forms of heart disease 42431 (18·0%) 26066 (13·7%) 16365 (35·3%) 4678 (52·3%) 3694 (59·3%) Chronic kidney disease 15908 (6·7%) 8345 (4·4%) 7563 (16·3%) 1941 (21·7%) 1892 (30·4%) Neoplasms 45 255 (19·1%) 34362 (18·1%) 10893 (23·5%) 2339 (26·1%) 1793 (28·8%) Data are n (%) or mean (SD). Only characteristics with a prevalence higher than 5% in the whole population are displayed. Additional baseline characteristics are presented in the appendix (pp 25–27). ITU=intensive therapy unit. Table 1: Baseline characteristics for the whole COVID-19 cohort and for the non-hospitalization, hospitalization, ITU admission, and encephalopathy cohorts during the illness Articles www.thelancet.com/psychiatry Vol 8 May 2021 419 tend to recur or relapse (eg, ischaemic strokes or psychiatric diagnoses), we estimated separately the incidence of first diagnoses (ie, excluding those who had a diagnosis before the index event) and the incidence of any diagnosis (ie, including patients who had a diagnosis at some point before the index event). For other outcomes (eg, Guillain-Barré syndrome), we estimated the incidence of any diagnosis. More details, and a full list of ICD-10 codes, are provided in the appendix (pp 4–5). Finally, to assess the overall risk of neurological and psychiatric outcomes after COVID-19, we estimated the incidence of any of the 14 outcomes, and the incidence of a first diagnosis of any of the outcomes. This is lower than the sum of incidences of each outcome because some patients had more than one diagnosis. Secondary analyses We investigated whether the neurological and psychiatric sequelae of COVID-19 were affected by the severity of the illness. The incidence of outcomes was estimated separately in four subgroups: first, in those who had required hospitalization within a time window from 4 days before their COVID-19 diagnosis (taken to be the time it might take between clinical presentation and confirmation) to 2 weeks afterwards; second, in those who had not required hospitalization during that window; third, in those who had been admitted to an intensive therapy unit (ITU) during that window; and fourth, in those who were diagnosed with delirium or other forms of altered mental status during that window; we use the term encephalopathy to describe this group of patients (appendix p 5).17,18 Differences in outcome incidence between these subgroups might reflect differences in their baseline characteristics. Therefore, for each outcome, we estimated the HR between patients requiring hospitalization (or ITU) and a matched cohort of patients not requiring hospitalization (or ITU), and between patients with encephalopathy and a matched cohort of patients without encephalopathy. Finally, HRs were calculated for patients who had not required hospitalization for COVID-19, influenza, or other respiratory tract infections. To provide benchmarks for the incidence and risk of neurological and psychiatric sequelae, patients after COVID-19 were compared with those in four additional matched cohorts of patients diagnosed with health events selected to represent a range of acute presentations during the same time period. These additional four index events were skin infection, urolithiasis, fracture of a large bone, and pulmonary embolism. More details are presented in the appendix (pp 5–6). We assessed the robustness of the differences in outcomes between cohorts by repeating the analysis in three scenarios: one including patients who had died by All patients Patients without hospitalization Patients with hospitalization Patients with ITU admission Patients with encephalopathy Intracranial hemorrhage (any) 0·56% (0·50–0·63) 0·31% (0·25–0·39) 1·31% (1·14–1·52) 2·66% (2·24–3·16) 3·61% (2·97–4·39) Intracranial hemorrhage (first) 0·28% (0·23–0·33) 0·14% (0·10–0·20) 0·63% (0·50–0·80) 1·05% (0·79–1·40) 1·19% (0·82–1·70) Ischemic stroke (any) 2·10% (1·97–2·23) 1·33% (1·22–1·46) 4·38% (4·05–4·74) 6·92% (6·17–7·76) 9·35% (8·23–10·62) Ischemic stroke (first) 0·76% (0·68–0·85) 0·43% (0·36–0·52) 1·60% (1·37–1·86) 2·82% (2·29–3·47) 3·28% (2·51–4·27) Parkinsonism 0·11% (0·08–0·14) 0·07% (0·05–0·12) 0·20% (0·15–0·28) 0·26% (0·15–0·45) 0·46% (0·28–0·78) Guillain-Barré syndrome 0·08% (0·06–0·11) 0·05% (0·03–0·07) 0·22% (0·15–0·32) 0·33% (0·21–0·54) 0·48% (0·20–1·14) Nerve, nerve root, or plexus disorders 2·85% (2·69–3·03) 2·69% (2·51–2·89) 3·35% (3·02–3·72) 4·24% (3·58–5·03) 4·69% (3·81–5·77) Myoneural junction or muscle disease 0·45% (0·40–0·52) 0·16% (0·12–0·20) 1·24% (1·05–1·46) 3·35% (2·76–4·05) 3·27% (2·54–4·21) Encephalitis 0·10% (0·08–0·13) 0·05% (0·03–0·08) 0·24% (0·17–0·33) 0·35% (0·19–0·64) 0·64% (0·39–1·07) Dementia 0·67% (0·59–0·75) 0·35% (0·29–0·43) 1·46% (1·26–1·71) 1·74% (1·31–2·30) 4·72% (3·80–5·85) Mood, anxiety, or psychotic disorder (any) 23·98% (23·58–24·38) 23·59% (23·12–24·07) 24·50% (23·76–25·26) 27·78% (26·33–29·29) 36·25% (34·16–38·43) Mood, anxiety, or psychotic disorder (first) 8·63% (8·28–8·98) 8·15% (7·75–8·57) 8·85% (8·22–9·52) 12·68% (11·28–14·24) 12·96% (11·13–15·07) Mood disorder (any) 13·66% (13·35–13·99) 13·10% (12·73–13·47) 14·69% (14·09–15·32) 15·43% (14·27–16·68) 22·52% (20·71–24·47) Mood disorder (first) 4·22% (3·99–4·47) 3·86% (3·60–4·14) 4·49% (4·05–4·99) 5·82% (4·86–6·97) 8·07% (6·56–9·90) Anxiety disorder (any) 17·39% (17·04–17·74) 17·51% (17·09–17·93) 16·40% (15·76–17·06) 19·15% (17·90–20·48) 22·43% (20·65–24·34) Anxiety disorder (first) 7·11% (6·82–7·41) 6·81% (6·47–7·16) 6·91% (6·38–7·47) 9·79% (8·65–11·06) 9·24% (7·70–11·07) Psychotic disorder (any) 1·40% (1·30–1·51) 0·93% (0·83–1·04) 2·89% (2·62–3·18) 2·77% (2·31–3·33) 7·00% (6·01–8·14) Psychotic disorder (first) 0·42% (0·36–0·49) 0·25% (0·19–0·33) 0·89% (0·72–1·09) 0·70% (0·46–1·06) 2·12% (1·53–2·94) Substance use disorder (any) 6·58% (6·36–6·80) 5·87% (5·63–6·13) 8·56% (8·10–9·04) 10·14% (9·25–11·10) 11·85% (10·55–13·31) Substance use disorder (first) 1·92% (1·77–2·07) 1·74% (1·58–1·91) 2·09% (1·82–2·40) 3·15% (2·60–3·82) 2·58% (1·91–3·47) Insomnia (any) 5·42% (5·20–5·64) 5·16% (4·91–5·42) 5·95% (5·53–6·39) 7·50% (6·66–8·44) 9·82% (8·57–11·24) Insomnia (first) 2·53% (2·37–2·71) 2·23% (2·05–2·43) 3·14% (2·81–3·51) 4·24% (3·55–5·07) 5·05% (4·10–6·20) Any outcome 33·62% (33·17–34·07) 31·74% (31·22–32·27) 38·73% (37·87–39·60) 46·42% (44·78–48·09) 62·34% (60·14–64·55) Any first outcome 12·84% (12·36–13·33) 11·51% (10·98–12·07) 15·29% (14·32–16·33) 25·79% (23·50–28·25) 31·13% (27·29–35·36) Data are percentage at 6 months (95% CI). Additional outcomes are presented in the appendix (pp 27–28). ITU=intensive therapy unit. Table 2: Major outcomes for the whole COVID-19 cohort, and for the non-hospitalization, hospitalization, ITU admission, and encephalopathy cohorts during the illness Articles 420 www.thelancet.com/psychiatry Vol 8 May 2021 the time of the analysis, another restricting the COVID-19 diagnoses to patients who had a positive RNA or antigen test (and using antigen test as an index event), and another comparing the rates of sequelae of patients with COVID-19 with those observed in patients with influenza before the pandemic (ie, in 2019 or 2018). Details of these analyses are provided in the appendix (p 6). Finally, to test whether differences in sequelae between cohorts could be accounted for by differences in extent of follow-up, we counted the average number of health visits that each cohort had during the follow-up period. Statistical analysis We used propensity score matching19 to create cohorts with matched baseline characteristics, done within the TriNetX network. Propensity score with 1:1 matching used a greedy nearest neighbor matching approach with a caliper distance of 0·1 pooled SDs of the logit of the propensity score. Any characteristic with a standardized mean difference between cohorts lower than 0·1 was considered well matched.20 The incidence of each outcome was estimated by use of the KaplanMeier estimator. Comparisons between cohorts were made with a log-rank test. We calculated HRs with 95% CIs using a proportional hazard model wherein the cohort to which the patient belonged was used as the independent variable. The proportional hazard assumption was tested with the generalized Schoenfeld approach. When the assumption was violated, the time varying HR was assessed with natural cubic splines fitted to the log cumulative hazard.21 Additional details are presented in the appendix (p 6). Statistical analyses were done in R, version 3.4.3, except for the log-rank tests, which were done within TriNetX. Statistical significance was set at two-sided p-value <0⋅05. Our study was reported according to the Reporting of studies Conducted using Observational Routinely collected health Data (RECORD, appendix pp 55–60). Role of the funding source The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the manuscript. Results Our primary cohort comprised 236 379 patients diagnosed with COVID-19, and our two propensity-score matched control cohorts comprised 105579 patients diagnosed with influenza and 236 038 patients diagnosed with any respiratory tract infection including influenza. The COVID-19 cohort was divided into subgroups of patients who were not hospitalized (190077 patients), those who were hospitalized (46 302 patients), those who required ITU admission (8945 patients), and those who received a diagnosis of encephalopathy (6229 patients). The main demographic features and comorbidities of the COVID-19 cohort are summarized in table 1, with additional demographic details presented in the appendix (pp 25–27). Matched baseline characteristics of the two control cohorts are also presented in the appendix (pp 29–30 for patients with influenza, and pp 31–32 for patients with other respiratory tract infections). Adequate propensity-score matching (standardized mean dif­ference <0·1) was achieved for all comparisons and baseline characteristics. We estimated the diagnostic incidence of the neurological and psychiatric outcomes of the primary cohort in the 6 months after a COVID-19 diagnosis. In the whole cohort, 33·62% (95% CI 33·17–34·07) of patients received a diagnosis (table 2). For the cohort subgroups, these estimates were 38·73% (37·87–39·60) for patients who were hospitalized, 46·42% (44·78–48·09) for those admitted to ITU, and 62·34% (60·14–64·55) for those diagnosed with COVID-19 vs influenza (N=105 579)* COVID-19 vs other RTI (N=236038)* HR (95% CI) p value HR (95% CI) p value Intracranial hemorrhage (any) 2·44 (1·89–3·16) <0·0001 1·26 (1·11–1·43) 0·0003 Intracranial hemorrhage (first) 2·53 (1·68–3·79) <0·0001 1·56 (1·27–1·92) <0·0001 Ischemic stroke (any) 1·62 (1·43–1·83) <0·0001 1·45 (1·36–1·55) <0·0001 Ischemic stroke (first) 1·97 (1·57–2·47) <0·0001 1·63 (1·44–1·85) <0·0001 Parkinsonism 1·42 (0·75–2·67) 0·19 1·45 (1·05–2·00) 0·020 Guillain-Barré syndrome 1·21 (0·72–2·04) 0·41 2·06 (1·43–2·96) <0·0001 Nerve, nerve root, or plexus disorders 1·64 (1·50–1·81) <0·0001 1·27 (1·19–1·35) <0·0001 Myoneural junction or muscle disease 5·28 (3·71–7·53) <0·0001 4·52 (3·65–5·59) <0·0001 Encephalitis 1·70 (1·04–2·78) 0·028 1·41 (1·03–1·92) 0·028 Dementia 2·33 (1·77–3·07) <0·0001 1·71 (1·50–1·95) <0·0001 Mood, anxiety, or psychotic disorder (any) 1·46 (1·43–1·50) <0·0001 1·20 (1·18–1·23) <0·0001 Mood, anxiety, or psychotic disorder (first) 1·81 (1·69–1·94) <0·0001 1·48 (1·42–1·55) <0·0001 Mood disorder (any) 1·47 (1·42–1·53) <0·0001 1·23 (1·20–1·26) <0·0001 Mood disorder (first) 1·79 (1·64–1·95) <0·0001 1·41 (1·33–1·50) <0·0001 Anxiety disorder (any) 1·45 (1·40–1·49) <0·0001 1·17 (1·15–1·20) <0·0001 Anxiety disorder (first) 1·78 (1·66–1·91) <0·0001 1·48 (1·42–1·55) <0·0001 Psychotic disorder (any) 2·03 (1·78–2·31) <0·0001 1·66 (1·53–1·81) <0·0001 Psychotic disorder (first) 2·16 (1·62–2·88) <0·0001 1·82 (1·53–2·16) <0·0001 Substance use disorder (any) 1·27 (1·22–1·33) <0·0001 1·09 (1·05–1·12) <0·0001 Substance use disorder (first) 1·22 (1·09–1·37) 0·0006 0·92 (0·86–0·99) 0·033 Insomnia (any) 1·48 (1·38–1·57) <0·0001 1·15 (1·10–1·20) <0·0001 Insomnia (first) 1·92 (1·72–2·15) <0·0001 1·43 (1·34–1·54) <0·0001 Any outcome 1·44 (1·40–1·47) <0·0001 1·16 (1·14–1·17) <0·0001 Any first outcome 1·78 (1·68–1·89) <0·0001 1·32 (1·27–1·36) <0·0001 Additional details on cohort characteristics and diagnostic subcategories are presented in the appendix (pp 29–33). HR=hazard ratio. RTI=respiratory tract infection. *Matched cohorts. Table 3: HRs for the major outcomes in patients after COVID-19 compared with those after influenza and other RTIs Articles www.thelancet.com/psychiatry Vol 8 May 2021 421 encephalopathy. A similar, but more marked, increasing trend was observed for patients receiving their first recorded neurological or psychiatric diagnosis (table 2). Results according to sex, race, and age are shown in the appendix (p 28). The baseline characteristics of the COVID-19 cohort divided into those who did versus those who did not have a neurological or psychiatric outcome are also shown in the appendix (p 7). We assessed the probability of the major neurological and psychiatric outcomes in patients diagnosed with COVID-19 compared with the matched cohorts diagnosed with other respiratory tract infections and with influenza (table 3; figure 1, appendix pp 8–10). Most diagnostic categories were more common in patients who had COVID-19 than in those who had influenza (HR 1·44, 95% CI 1·40–1·47 for any diagnosis; 1·78, 1·68–1·89 for any first diagnosis) and those who had other respiratory tract infections (1·16, 1·14–1·17 for any diagnosis; 1·32, 1·27–1·36 for any first diagnosis). Hazard rates were also higher in patients who were admitted to ITU than in those who were not (1·58, 1·50–1·67 for any diagnosis; 2·87, 2·45–3·35 for any first diagnosis). HRs were significantly greater than 1 for all diagnoses for patients who had COVID-19 compared with those who had influenza, except for parkinsonism and Guillain-Barré syndrome, and significantly greater than 1 for all diagnoses compared with patients who had respiratory tract infections (table 3). Similar results were observed when patients who had COVID-19 were compared with those who had Figure 1: Kaplan-Meier estimates for the incidence of major outcomes after COVID-19 compared with other RTIs Shaded areas are 95% CIs. For incidences of first diagnoses, the number in brackets corresponds to all patients who did not have the outcome before the follow-up period. For diagnostic subcategories, see appendix (pp 8–10). RTI=respiratory tract infection. Number at risk COVID-19 Other RTI 0 50 100 150 200 92579 131885 67102 116315 Intracranial haemorrhage (any) 50172 103261 32705 90066 20679 77005 12775 65909 0 0·2 0·6 0·4 0·8 Outcome probability (%) 30 60 90 120 150 180 0 50 100 150 200 91998 131352 66499 115264 48528 102599 32265 89412 20361 76367 11415 63334 0 0·5 2·0 1·5 2·5 30 60 90 120 150 180 0 50 100 150 200 92193 131363 66587 115073 48488 102233 32186 88929 19962 75806 11585 62702 0 1·0 3·0 2·0 4·0 30 60 90 120 150 180 COVID-19 (n=236038) Other RTI (n=236038) COVID-19 (n=236038) Other RTI (n=236038) COVID-19 (n=236038) Other RTI (n=236038) Ischaemic stroke (any) Nerve, nerve root, or plexus disorder Number at risk COVID-19 Other RTI 0 50 100 Time since index event (days) Time since index event (days) Time since index event (days) 150 200 91646 133203 66346 115207 Myoneural junction or muscle disease 50653 102653 34259 90454 21522 76919 11895 63909 0 0·2 0·1 0·5 0·4 0·3 0·6 Outcome probability (%) 30 60 90 120 150 180 0 50 100 150 200 89958 128680 65186 113623 47578 101313 32182 88082 19593 75359 12242 62553 0 0·2 0·6 0·4 0·8 30 60 90 120 150 180 0 50 100 150 200 84435 122790 58504 103824 41026 89662 26310 75998 15885 63173 8741 51033 0 10 5 20 15 25 30 60 90 120 150 180 Dementia Mood, anxiety, or psychotic disorder COVID-19 (n=234527) Other RTI (n=234810) COVID-19 (n=230151) Other RTI (n=230495) COVID-19 (n=236038) Other RTI (n=236038) Articles 422 www.thelancet.com/psychiatry Vol 8 May 2021 one of the four other index events (appendix pp 11–14, 34), except when an outcome had a predicted relationship with the comparator condition (eg, intracranial hemorrhage was more common in association with fracture of a large bone). HRs for diagnostic subcategories are presented in the appendix (p 33). There were no violations of the proportional hazards assumption for most of the neurological outcomes over the 6 months of follow-up (appendix pp 15, 35). The only exception was for intracranial hemorrhage and ischemic stroke in patients who had COVID-19 when compared with patients who had other respiratory tract infections (p=0·012 for intracranial hemorrhage and p=0·032 for ischemic stroke). For the overall psychiatric disorder category (ICD-10 F20–48), the HR did vary with time, declining but remaining significantly higher than 1, indicating that the risk was attenuated but maintained 6 months after COVID-19 diagnosis (appendix p 9). HRs for COVID-19 diagnosis compared with the additional four index events showed more variation with time, partly reflecting the natural history of the comparator condition (appendix, pp 16–19, 36). We explored the effect of COVID-19 severity in four ways. First, we restricted analyses to matched cohorts of patients who had not required hospitalization (matched baseline characteristics in the appendix, pp 37–40). HRs remained significantly greater than 1 in this subgroup, with an overall HR for any diagnosis of 1·47 (95% CI 1·44–1·51) for patients who had COVID-19 compared with patients who had influenza, and 1·16 (1·14–1·17) compared with those who had other respiratory tract infections (table 4, appendix pp 20–21). For a first diagnosis, the HRs were 1·83 (1·71–1·96) versus patients who had influenza and 1·28 (1·23–1·33) versus those who had other respiratory tract infections. Second, we calculated HRs for the matched cohorts of patients with COVID-19 requiring hospitalization versus those who did not require hospitalization (44 927 matched patients; matched baseline characteristics are presented in the appendix, pp 41–42). This comparison showed greater hazard rates for all outcomes in the hospitalized group than in the non-hospitalized group, except for nerve, nerve root, or plexus disorders (table 5, figure 2), with an overall HR of 1·33 (1·29–1·37) for any diagnosis and 1·70 (1·56–1·86) for any first diagnosis. Third, we calculated HRs for the matched cohorts of patients with COVID-19 requiring ITU admission versus those not COVID-19 vs influenza in patients without hospitalization (N=96803)* COVID-19 vs other RTI in patients without hospitalization (N=183 731)* HR (95% CI) p value HR (95% CI) p value Intracranial hemorrhage (any) 1·87 (1·25–2·78) 0·0013 1·38 (1·11–1·73) 0·0034 Intracranial hemorrhage (first) 1·66 (0·88–3·14) 0·082 1·63 (1·11–2·40) 0·010 Ischemic stroke (any) 1·80 (1·54–2·10) <0·0001 1·61 (1·45–1·78) <0·0001 Ischemic stroke (first) 1·71 (1·26–2·33) 0·0003 1·69 (1·38–2·08) <0·0001 Parkinsonism 2·22 (0·98–5·06) 0·028 1·20 (0·73–1·96) 0·42 Guillain-Barré syndrome 0·90 (0·44–1·84) 0·99 1·44 (0·85–2·45) 0·10 Nerve, nerve root, or plexus disorders 1·69 (1·53–1·88) <0·0001 1·23 (1·15–1·33) <0·0001 Myoneural junction or muscle disease 3·46 (2·11–5·67) <0·0001 2·69 (1·91–3·79) <0·0001 Encephalitis 1·77 (0·86–3·66) 0·095 2·29 (1·28–4·10) 0·0046 Dementia 1·88 (1·27–2·77) 0·0008 1·95 (1·55–2·45) <0·0001 Mood, anxiety, or psychotic disorder (any) 1·49 (1·45–1·54) <0·0001 1·18 (1·15–1·21) <0·0001 Mood, anxiety, or psychotic disorder (first) 1·85 (1·72–1·99) <0·0001 1·40 (1·32–1·48) <0·0001 Mood disorder (any) 1·49 (1·43–1·55) <0·0001 1·22 (1·19–1·26) <0·0001 Mood disorder (first) 1·78 (1·61–1·96) <0·0001 1·37 (1·27–1·47) <0·0001 Anxiety disorder (any) 1·48 (1·43–1·54) <0·0001 1·16 (1·13–1·19) <0·0001 Anxiety disorder (first) 1·80 (1·67–1·94) <0·0001 1·37 (1·30–1·45) <0·0001 Psychotic disorder (any) 1·93 (1·63–2·28) <0·0001 1·44 (1·27–1·62) <0·0001 Psychotic disorder (first) 2·27 (1·56–3·30) <0·0001 1·49 (1·15–1·93) 0·0016 Substance use disorder (any) 1·26 (1·19–1·33) <0·0001 1·11 (1·07–1·17) <0·0001 Substance use disorder (first) 1·21 (1·05–1·38) 0·0054 0·89 (0·81–0·97) 0·013 Insomnia (any) 1·52 (1·42–1·63) <0·0001 1·18 (1·12–1·24) <0·0001 Insomnia (first) 2·06 (1·82–2·33) <0·0001 1·51 (1·38–1·66) <0·0001 Any outcome 1·47 (1·44–1·51) <0·0001 1·16 (1·14–1·17) <0·0001 Any first outcome 1·83 (1·71–1·96) <0·0001 1·28 (1·23–1·33) <0·0001 Details on cohort characteristics are presented in the appendix (pp 37–40). HR=hazard ratio. RTI=respiratory tract infection. *Matched cohorts. Table 4: HRs for the major outcomes in patients without hospitalization after COVID-19 compared with those after influenza or other RTIs Articles www.thelancet.com/psychiatry Vol 8 May 2021 423 requiring ITU admission (8942 patients; matched baseline characteristics presented in the appendix, pp 43–44), with a HR of 1·58 (1·50–1·67) for any diagnosis and 2·87 (2·45–3·35) for any first diagnosis (table 5, appendix p 22). Fourth, we calculated HRs for the matched cohorts of patients with COVID-19 who had encephalopathy diagnosed during acute illness versus those who did not (6221 patients; matched baseline characteristics presented in the appendix, pp 45–46). HRs for all diagnoses were greater for the group who had encephalopathy than for the matched cohort who did not, with an overall HR of 1·85 (1·73–1·98) for any diagnosis and 3·19 (2·54–4·00) for any first diagnosis (table 5, figure 2). We inspected other factors that might influence the findings. The results regarding hospitalization, ITU admission, or encephalopathy (which we had defined as occurring up to 14 days after diagnosis) could be confounded by admissions due to an early complication of COVID-19 rather than to COVID-19 itself. This was explored by excluding outcomes during this period, with the findings remaining similar, albeit with many HRs being reduced (appendix pp 47–49). Additionally, COVID-19 survivors had fewer health-care visits during the 6-month period compared with the other cohorts (appendix p 50). Hence the higher incidence of many diagnoses was not simply due to having had more diagnostic opportunities. The increased rates of neurological and psychiatric sequelae were robust in all three sensitivity analyses: when patients who had died by the time of the analysis were included (appendix p 51), when the COVID-19 diagnosis was confirmed by use of an RNA or antigen test (appendix p 52), and when the sequelae were compared with those observed in patients who had influenza in 2019 or 2018 (appendix pp 53). Discussion Various adverse neurological and psychiatric outcomes occurring after COVID-19 have been predicted and COVID-19 with vs without hospitalization (N=45167) COVID-19 with vs without ITU admission (N=8942) COVID-19 with vs without encephalopathy (N=6221) HR (95% CI) p value HR (95% CI) p value HR (95% CI) p value Intracranial hemorrhage (any) 3·09 (2·43–3·94) <0·0001 5·06 (3·43–7·47) <0·0001 4·73 (3·15–7·11) <0·0001 Intracranial hemorrhage (first) 3·75 (2·49–5·64) <0·0001 5·12 (2·68–9·77) <0·0001 5·00 (2·33–10·70) <0·0001 Ischemic stroke (any) 1·65 (1·48–1·85) <0·0001 1·93 (1·62–2·31) <0·0001 1·65 (1·38–1·97) <0·0001 Ischemic stroke (first) 2·82 (2·22–3·57) <0·0001 3·51 (2·39–5·15) <0·0001 3·39 (2·17–5·29) <0·0001 Parkinsonism 2·63 (1·45–4·77) 0·0016 3·90 (1·29–11·79) 0·024 1·64 (0·75–3·58) 0·24 Guillain-Barré syndrome 2·94 (1·60–5·42) 0·00094 11·01 (2·55–47·61) 0·0007 2·27 (0·76–6·73) 0·24 Nerve, nerve root, or plexus disorders 0·94 (0·83–1·06) 0·29 1·16 (0·92–1·45) 0·21 1·41 (1·07–1·87) 0·018 Myoneural junction or muscle disease 7·76 (5·15–11·69) <0·0001 11·53 (6·38–20·83) <0·0001 5·40 (3·21–9·07) <0·0001 Encephalitis 3·26 (1·75–6·06) 0·0002 1·78 (0·75–4·20) 0·22 9·98 (2·98–33·43) <0·0001 Dementia 2·28 (1·80–2·88) <0·0001 1·66 (1·12–2·46) 0·018 4·25 (2·79–6·47) <0·0001 Mood, anxiety, or psychotic disorder (any) 1·23 (1·18–1·28) <0·0001 1·34 (1·24–1·46) <0·0001 1·73 (1·58–1·90) <0·0001 Mood, anxiety, or psychotic disorder (first) 1·55 (1·40–1·71) <0·0001 2·27 (1·87–2·74) <0·0001 2·28 (1·80–2·89) <0·0001 Mood disorder (any) 1·21 (1·15–1·28) <0·0001 1·15 (1·03–1·27) 0·010 1·51 (1·35–1·70) <0·0001 Mood disorder (first) 1·53 (1·33–1·75) <0·0001 2·06 (1·57–2·71) <0·0001 2·09 (1·55–2·80) <0·0001 Anxiety disorder (any) 1·16 (1·10–1·22) <0·0001 1·39 (1·26–1·53) <0·0001 1·64 (1·45–1·84) <0·0001 Anxiety disorder (first) 1·49 (1·34–1·65) <0·0001 2·22 (1·82–2·71) <0·0001 1·91 (1·48–2·45) <0·0001 Psychotic disorder (any) 2·22 (1·92–2·57) <0·0001 1·48 (1·14–1·92) 0·0028 3·84 (2·90–5·10) <0·0001 Psychotic disorder (first) 2·77 (1·99–3·85) <0·0001 1·77 (0·98–3·20) 0·072 5·62 (2·93–10·77) <0·0001 Substance use disorder (any) 1·53 (1·42–1·64) <0·0001 1·62 (1·41–1·85) <0·0001 1·45 (1·24–1·70) <0·0001 Substance use disorder (first) 1·68 (1·40–2·01) <0·0001 2·53 (1·83–3·50) <0·0001 2·03 (1·32–3·11) 0·0015 Insomnia (any) 1·08 (0·99–1·18) 0·088 1·40 (1·19–1·66) <0·0001 1·73 (1·42–2·11) <0·0001 Insomnia (first) 1·49 (1·28–1·74) <0·0001 1·93 (1·46–2·55) <0·0001 3·44 (2·35–5·04) <0·0001 Any outcome 1·33 (1·29–1·37) <0·0001 1·58 (1·50–1·67) <0·0001 1·85 (1·73–1·98) <0·0001 Any first outcome 1·70 (1·56–1·86) <0·0001 2·87 (2·45–3·35) <0·0001 3·19 (2·54–4·00) <0·0001 Details on cohort characteristics are presented in the appendix (pp 41–46). HR=hazard ratio. ITU=intensive therapy unit. *Matched cohorts. Table 5: HRs for the major outcomes after COVID-19 for patients with vs those without hospitalization, patients with vs without ITU admission, and patients with vs without encephalopathy Articles 424 www.thelancet.com/psychiatry Vol 8 May 2021 Encephalopathy Matched cohort without encephalopathy Number at risk Encephalopathy Matched cohort without encephalopathy Hospitalization Matched cohort without hospitalization 0 50 100 150 200 Intracranial hemorrhage (any) 0 1 4 3 2 Outcome probability (%) 30 60 90 120 150 180 Ischemic stroke (any) Total 6221 6221 45167 45167 3214 3424 20486 20010 2296 2372 14717 14696 1746 2372 11818 11185 1269 1244 7766 7344 1032 1244 5232 4799 733 642 4030 3598 0 50 100 150 200 0 2·5 10·0 7·5 5·0 Total 30 60 90 120 150 180 6221 6221 45167 45167 3133 2989 20218 19587 2201 2187 14429 14515 1639 1641 10786 10792 1177 1221 7566 7464 821 758 5083 4714 634 758 3551 3133 Number at risk Encephalopathy Matched cohort without encephalopathy Hospitalization Matched cohort without hospitalization 0 50 100 150 200 Nerve, nerve root, or plexus disorder 0 1 5 4 3 2 Outcome probability (%) 30 60 90 120 150 180 Myoneural junction or muscle disease Total 0 50 100 150 200 0 1 4 3 2 Total 30 60 90 120 150 180 6221 6221 45167 45167 3277 3125 20453 19636 2317 2225 14614 14537 1701 1701 10902 10775 1231 1314 7737 7447 881 825 5062 4549 602 596 3378 2606 5906 6109 44481 44788 2996 3067 20044 20069 2127 2236 14345 16550 1836 1916 10853 13067 1292 1916 8206 10185 790 1139 5270 10185 604 635 3320 4092 Number at risk Encephalopathy Matched cohort without encephalopathy Hospitalization Matched cohort without hospitalization 0 50 100 Time since index event (days) Time since index event (days) 150 200 Dementia 0 1 5 4 3 2 Outcome probability (%) 30 60 90 120 150 180 Mood, anxiety, or psychotic disorder Total 0 50 100 150 200 0 10 40 30 20 Total 30 60 90 120 150 180 4704 5094 42434 42877 2627 2562 19428 18719 1929 2010 13970 13904 1425 1419 10567 10329 1036 986 7611 7017 717 986 5427 5174 583 986 3616 2697 6221 6221 45167 45167 2640 2729 18072 18092 1734 1910 12352 12874 1217 1417 8933 9170 888 955 6068 5925 575 633 3976 3653 351 437 2501 2001 Hospitalization Matched cohort without hospitalization Articles www.thelancet.com/psychiatry Vol 8 May 2021 425 reported.1–5,14 The data presented in this study, from a large electronic health records network, support these predictions and provide estimates of the incidence and risk of these outcomes in patients who had COVID-19 compared with matched cohorts of patients with other health conditions occurring contemporaneously with the COVID-19 pandemic (tables 2, 3, figure 1). The severity of COVID-19 had a clear effect on subsequent neurological diagnoses (tables 4, 5, figure 2). Overall, COVID-19 was associated with increased risk of neurological and psychiatric outcomes, but the incidences and HRs of these were greater in patients who had required hospitalization, and markedly so in those who had required ITU admission or had developed encephalopathy, even after extensive propensity score matching for other factors (eg, age or previous cerebrovascular disease). Potential mechanisms for this association include viral invasion of the CNS,10,11 hypercoagulable states,22 and neural effects of the immune response.9 However, the incidence and relative risk of neurological and psychiatric diagnoses were also increased even in patients with COVID-19 who did not require hospitalization. Some specific neurological diagnoses merit individual mention. Consistent with several other reports,23,24 the risk of cerebrovascular events (ischemic stroke and intracranial hemorrhage) was elevated after COVID-19, with the incidence of ischemic stroke rising to almost one in ten (or three in 100 for a first stroke) in patients with encephalopathy. A similarly increased risk of stroke in patients who had COVID-19 compared with those who had influenza has been reported.25 Our previous study reported preliminary evidence for an association between COVID-19 and dementia.14 The data in this study support this association. Although the estimated incidence was modest in the whole COVID-19 cohort (table 2), 2·66% of patients older than 65 years (appendix p 28) and 4·72% who had encephalopathy (table 2), received a first diagnosis of dementia within 6 months of having COVID-19. The associations between COVID-19 and cerebrovascular and neurodegenerative diagnoses are concerning, and information about the severity and subsequent course of these diseases is required. Whether COVID-19 is associated with Guillain-Barré syndrome remains unclear;26 our data were also equivocal, with HRs increased with COVID-19 compared with other respiratory tract infections but not with influenza (table 3), and increased compared with three of the four other index health events (appendix p 34). Concerns have also been raised about post-COVID-19 parkinsonian syndromes, driven by the encephalitis lethargica epidemic that followed the 1918 influenza pandemic.27 Our data provide some support for this possibility, although the incidence was low and not all HRs were significant. Parkinsonism might be a delayed outcome, in which case a clearer signal might emerge with a longer follow-up. The findings regarding anxiety and mood disorders were broadly consistent with 3-month outcome data from a study done in a smaller number of cases than our cohort, using the same network,14 and showed that the HR remained elevated, although decreasing, at the 6-month period. Unlike the earlier study, and in line with previous suggestions,28 we also observed a significantly increased risk of psychotic disorders, probably reflecting the larger sample size and longer duration of follow-up reported here. Substance use disorders and insomnia were also more common in COVID-19 survivors than in those who had influenza or other respiratory tract infections (except for the incidence of a first diagnosis of substance use disorder after COVID-19 compared with other respiratory tract infections). Therefore, as with the neurological outcomes, the psychiatric sequelae of COVID-19 appear widespread and to persist up to, and probably beyond, 6 months. Compared with neurological disorders, common psychiatric disorders (mood and anxiety disorders) showed a weaker relationship with the markers of COVID-19 severity in terms of incidence (table 2) or HRs (table 5). This might indicate that their occurrence reflects, at least partly, the psychological and other implications of a COVID-19 diagnosis rather than being a direct manifestation of the illness. HRs for most neurological outcomes were constant, and hence the risks associated with COVID-19 persisted up to the 6-month timepoint. Longer-term studies are needed to ascertain the duration of risk and the trajectory for individual diagnoses. Our findings are robust given the sample size, the propensity score matching, and the results of the sensitivity and secondary analyses. Nevertheless, they have weaknesses inherent to an electronic health records study,29 such as the unknown completeness of records, no validation of diagnoses, and sparse information on socioeconomic and lifestyle factors. These issues primarily affect the incidence estimates, but the choice of cohorts against which to compare COVID-19 outcomes influenced the magnitude of the HRs (table 3, appendix p 34). The analyses regarding encephalopathy (delirium and related conditions) deserve a note of caution. Even among patients who were hospitalized, only about 11% received this Figure 2: Kaplan-Meier estimates for the incidence of major outcomes after COVID-19 comparing patients requiring hospitalization with matched patients not requiring hospitalization, and comparing those who had encephalopathy with matched patients who did not have encephalopathy 95% CIs are omitted for clarity but are shown in the appendix (p 23). For incidences of first diagnoses, the total number corresponds to all patients who did not have the outcome before the follow-up period. The equivalent figure showing the comparison between patients with intensive therapy unit admission versus those without is presented in the appendix (p 22). Articles 426 www.thelancet.com/psychiatry Vol 8 May 2021 diagnosis, whereas much higher rates would be expected.18,30 Under-recording of delirium during acute illness is well known and probably means that the diagnosed cases had prominent or sustained features; as such, results for this group should not be generalized to all patients with COVID-19 who experience delirium. We also note that encephalopathy is not just a severity marker but a diagnosis in itself, which might predispose to, or be an early sign of, other neuropsychiatric or neurodegenerative outcomes observed during follow-up. The timing of index events was such that most infections with influenza and many of the other respiratory tract infections occurred earlier on during the pandemic, whereas the incidence of COVID-19 diagnoses increased over time (appendix p 24). The effect of these timing differences on observed rates of sequelae is unclear but, if anything, they are likely to make the HRs an underestimate because COVID-19 cases were diagnosed at a time when all other diagnoses were made at a lower rate in the population (appendix p 24). Some patients in the comparison cohorts are likely to have had undiagnosed COVID-19; this would also tend to make our HRs an underestimate. Finally, a study of this kind can only show associations; efforts to identify mechanisms and assess causality will require prospective cohort studies and additional study designs. In summary, the present data show that COVID-19 is followed by significant rates of neurological and psychiatric diagnoses over the subsequent 6 months. Services need to be configured, and resourced, to deal with this anticipated need. Contributors PJH and MT were granted unrestricted access to the TriNetX Analytics network for the purposes of research, and with no constraints on the analyses done or the decision to publish; they designed the study and directly accessed the TriNetX Analytics web interface to do it. MT, JRG, MH, and PJH defined cohort inclusion and exclusion criteria, and the outcome criteria and analytical approaches. MT did the data analyses, assisted by SL and PJH. All authors contributed to data interpretation. MT and PJH wrote the paper with input from JRG, MH, and SL. MT and PJH verified the data. PJH is the guarantor. PJH and MT had full access to all the data in the study, and the corresponding author had final responsibility for the decision to submit for publication. Declaration of interests SL is an employee of TriNetX. All other authors declare no competing interests. Data sharing The TriNetX system returned the results of these analyses as .csv files, which were downloaded and archived. Data presented in this paper can be freely accessed online. Additionally, TriNetX will grant access to researchers if they have a specific concern (through a third-party agreement option).

Acknowledgments

This work was supported by the NIHR Oxford Health Biomedical Research Centre (grant BRC-1215–20005). MT is an NIHR Academic Clinical Fellow. MH is supported by a Wellcome Trust Principal Research Fellowship and the NIHR Oxford Biomedical Research Centre. The views expressed are those of the authors and not necessarily those of the UK National Health Service, NIHR, or the UK Department of Health.

References

1 Rogers JP, Chesney E, Oliver D, et al. Psychiatric and neuropsychiatric presentations associated with severe coronavirus infections: a systematic review and meta-analysis with comparison to the COVID-19 pandemic. Lancet Psychiatry 2020; 7: 611–27.

2 Ellul MA, Benjamin L, Singh B, et al. Neurological associations of COVID-19. Lancet Neurol 2020; 19: 767–83.

3 Varatharaj A, Thomas N, Ellul MA, et al. Neurological and neuropsychiatric complications of COVID-19 in 153 patients: a UK-wide surveillance study. Lancet Psychiatry 2020; 7: 875–82.

4 Paterson RW, Brown RL, Benjamin L, et al. The emerging spectrum of COVID-19 neurology: clinical, radiological and laboratory findings. Brain 2020; 143: 3104–20.

5 Kremer S, Lersy F, Anheim M, et al. Neurologic and neuroimaging findings in patients with COVID-19: a retrospective multicenter study. Neurology 2020; 95: e1868–82.

6 Pezzini A, Padovani A. Lifting the mask on neurological manifestations of COVID-19. Nat Rev Neurol 2020; 16: 636–44.

7 Raman B, Cassar MP, Tunnicliffe EM, et al. Medium-term effects of SARS-CoV-2 infection on multiple vital organs, exercise capacity, cognition, quality of life and mental health, post-hospital discharge. EClinicalMedicine 2021; 31: 100683.

8 Iadecola C, Anrather J, Kamel H. Effects of COVID-19 on the nervous system. Cell 2020; 183: 16–27.

9 Kreye J, Reincke SM, Prüss H. Do cross-reactive antibodies cause neuropathology in COVID-19? Nat Rev Immunol 2020; 20: 645–46.

10 Meinhardt J, Radke J, Dittmayer C, et al. Olfactory transmucosal SARS-CoV-2 invasion as a port of central nervous system entry in individuals with COVID-19. Nat Neurosci 2021; 24: 168–75.

11 Rhea EM, Logsdon AF, Hansen KM, et al. The S1 protein of SARS-CoV-2 crosses the blood-brain barrier in mice. Nat Neurosci 2021; 24: 368–78.

12 Holmes EA, O’Connor RC, Perry VH, et al. Multidisciplinary research priorities for the COVID-19 pandemic: a call for action for mental health science. Lancet Psychiatry 2020; 7: 547–60.

13 Vindegaard N, Benros ME. COVID-19 pandemic and mental health consequences: systematic review of the current evidence. Brain Behav Immun 2020; 89: 531–42.

14 Taquet M, Luciano S, Geddes JR, Harrison PJ. Bidirectional associations between COVID-19 and psychiatric disorder: retrospective cohort studies of 62 354 COVID-19 cases in the USA. Lancet Psychiatry 2021; 8: 130–40.

15 de Lusignan S, Dorward J, Correa A, et al. Risk factors for SARS-CoV-2 among patients in the Oxford Royal College of General Practitioners Research and Surveillance Centre primary care network: a cross-sectional study. Lancet Infect Dis 2020; 20: 1034–42.

16 Williamson EJ, Walker AJ, Bhaskaran K, et al. Factors associated with COVID-19-related death using OpenSAFELY. Nature 2020; 584: 430–36.

17 Slooter AJC, Otte WM, Devlin JW, et al. Updated nomenclature of delirium and acute encephalopathy: statement of ten Societies. Intensive Care Med 2020; 46: 1020–22.

18 Oldham MA, Slooter AJC, Cunningham C, et al. Characterising neuropsychiatric disorders in patients with COVID-19. Lancet Psychiatry 2020; 7: 932–33.

19 Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011; 46: 399–424.

20 Haukoos JS, Lewis RJ. The propensity score. JAMA 2015; 314: 1637–38. 21 Royston P, Parmar MKB. Flexible parametric proportional-hazards and proportional-odds models for censored survival data, with application to prognostic modelling and estimation of treatment effects. Stat Med 2002;

21: 2175–97.

22 Panigada M, Bottino N, Tagliabue P, et al. Hypercoagulability of COVID-19 patients in intensive care unit: a report of thromboelastography findings and other parameters of hemostasis. J Thromb Haemost 2020; 18: 1738–42.

23 Siow I, Lee KS, Zhang JJY, Saffari SE, Ng A, Young B. Stroke as a neurological complication of COVID-19: a systematic review and meta-analysis of incidence, outcomes and predictors. J Stroke Cerebrovasc Dis 2021; 30: 105549. For the study data see https:// osf.io/7tzvy Articles www.thelancet.com/psychiatry Vol 8 May 2021 427

24 Hernandez-Fernandez F, Valencia HS, Barbella-Aponte R, et al. Cerebrovascular disease in patients with COVID-19: neuroimaging, histological and clinical description. Brain 2020; 143: 3089–103.

25 Xie Y, Bowe B, Maddukuri G, Al-Aly Z. Comparative evaluation of clinical manifestations and risk of death in patients admitted to hospital with COVID-19 and seasonal influenza: cohort study. BMJ 2020; 371: m4677.

26 Keddie S, Pakpoor J, Mousele C, et al. Epidemiological and cohort study finds no association between COVID-19 and Guillain-Barré syndrome. Brain 2020; published online Dec 14. https://doi. org/10.1093/brain/awaa433.

27 Hoffman LA, Vilensky JA. Encephalitis lethargica: 100 years after the epidemic. Brain 2017; 140: 2246–51.

28 Watson CJ, Thomas RH, Solomon T, Michael BD, Nicholson TR, Pollak TA. COVID-19 and psychosis risk: real or delusional concern? Neurosci Lett 2021; 741: 135491.

29 Casey JA, Schwartz BS, Stewart WF, Adler NE. Using electronic health records for population health research: a review of methods and applications. Annu Rev Public Health 2016; 37: 61–81.

30 Docherty AB, Harrison EM, Green CA, et al. Features of 20133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study. BMJ 2020; 369: m1985.