Symptoms, Recovery, and Effects of Long COVID Over Time

The survey was coproduced working with public contributors (M.E.O., C.H.), who have lived the experience of Long COVID and provide peer support to others with Long COVID [16]. N.A.A. also had lived the experience of Long COVID. Public contributor members of the Long COVID Support's COVID-19 Research Involvement Group on Facebook provided feedback on early versions of the questionnaire which was amended accordingly. Qualtrics was used as the platform for the follow-up following feedback from the baseline survey about user-friendliness.

Demographic information, baseline health, functional status at start of illness, and preexisting health conditions were captured in the baseline survey. Questions at follow-up included ability to work, current employment status, pattern of illness and impact on health, symptoms that have remained over the longer-term course (symptoms experienced at follow-up), clinical diagnosis of Long COVID and other conditions, and an 8-item patient health questionnaire. Further details on the questions asked to capture the measures listed at follow-up are provided in. Supplementary Box 1

Current employment status was captured using a multiple answer question with detailed options including employed (full-time, part-time, phased return to work, working reduced hours), self-employed (with or without employees), unemployed, volunteering, apprenticeship, student, not looking for work, unable to work, retired, made redundant/took early retirement, and other (with an open-text box to provide details). These responses were used to derive mutually exclusive categories: employed/self-employed (full- and part-time), unable to work, unable to work but employed/self-employed, student/volunteer/at home not looking for work, unemployed and looking for work, and retired/other.

Pattern of illness was captured through several questions. The first question asked about the nature of symptoms with options for constant (experienced at least 1 symptom every day), fluctuating (but symptoms never completely go away compared to pre-COVID health), relapsing and remitting (have symptom-free periods between relapses), constant for 2 weeks at the start of illness but fluctuating since, constant for 2 weeks at the start of illness but relapsing since, constant for 4 weeks at the start of illness but fluctuating since and constant for 4 weeks at the start of illness but fluctuating since, and other (with an open-text box to provide details). Based on input from the public contributors, we did not define number of days for a symptom-free interval for participants reporting relapsing nature of symptoms because of large variation in the patterns experienced by individuals with Long COVID. For participants reporting relapsing or fluctuating nature of symptoms, additional questions captured detail on length of remission or less intense symptoms, pattern of symptoms (triggered by an identifiable factor, set/cyclical pattern with no identifiable trigger, generally set pattern but occasionally triggered by something), and triggers if known. We asked all participants if symptoms have evolved using a multiple-answer question with response options of stayed the same, intensity has reduced, intensity has increased, hard to estimate, and new symptoms have appeared.

Symptoms experienced at follow-up was captured using a list of symptoms developed as part of the co-design of the baseline survey. Common “other” responses to the baseline survey were added as options as well as additional symptoms identified through published research [17] or advocacy/support work by the co-authors.

Clinical diagnosis of Long COVID was captured through a multiple-answer question with response options of: yes—have Long COVID as a diagnosis on my health record, not officially diagnosed but doctors suspect I have Long COVID, not been diagnosed with Long COVID and doctors do not suspect I have Long COVID, not sure if officially diagnosed with Long COVID, and tested positive for COVID-19 but not received a clinical diagnosis of Long COVID. Diagnosis of myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) diagnosis since COVID-19 infection was captured using a binary (yes/no) question. Any other new diagnoses since COVID-19 infection were also captured using a binary (yes/no) with an open-text box to specify details of new diagnosis. This was because patient and public engagement work indicated that people with Long COVID commonly got a diagnosis of ME/CFS instead of Long COVID or received diagnosis based on some of the symptoms experienced.

Recovery was self-defined by the participant and captured using an “Yes, I consider myself fully recovered (feeling as healthy as I did before infection and able to function at the same level of activity)” option to a question on how the participant would currently describe their health. The other response options to this question were: “still experiencing symptoms,” “feel far from recovery,” “feel stable but have lower level of health and activity,” “feel stable and close to baseline health/recovery,” “feel potential for relapse,” and “unsure because symptoms come and go.” Participants who responded unsure were included in all analyses because only a small proportion of participants chose this response alone, with the majority selecting 1 or more of the available response options. Based on responses to other questions in the survey and input from public contributors, we are confident that participants were unsure how to describe their current health but were still experiencing Long COVID. Participants who chose the recovered option were additionally asked “How long they were symptom-free before considering yourself completely recovered” and “Over what period did your Long COVID last.”

Data were downloaded from Qualtrics after the survey was taken offline. Statistical analysis was carried out using Stata. Complete case analysis was carried out as missing data were minimal.

Descriptive percentages and summary statistics were generated for the full sample and stratified by those with and without Long COVID diagnosis. Univariate comparisons between those with and without Long COVID diagnosis were carried out using t-test for continuous variables and chi-squared test for categorical variables.

Questions on employment status, symptom pattern, job loss, and income loss resulting from COVID-19 illness were included in both baseline and follow-up surveys, and variables were derived to characterize the change based on the responses given at each survey. Logistic regression was used to examine the association between having received a Long COVID diagnosis at follow-up and symptom pattern, work status, and post-COVID-19 functional status at baseline [18]. Initial univariable analysis was followed by multivariable models adjusting for age, gender, ethnicity, highest educational attainment, smoking status at baseline survey, health before COVID-19, preexisting health condition, household income (model 1), and the other exposures considered (eg, symptom pattern was adjusted for employment status at baseline and post-COVID-19 functional status score at 6 weeks from start of illness).

More than two thirds (68.4%, n = 792) of participants reported still experiencing Long COVID symptoms. Of the 792 participants, 145 only selected the still experiencing symptoms option, 129 also reported feeling far from recovery, and 83 reported experiencing symptoms and feeling stable at a lower level of health and activity than pre-COVID (). Eighty-four participants (7.6%) reported feeling stable and close to the pre-COVID level of health and activity, 38 participants reported feeling unsure about their health as symptoms come and go and a further 184 participants reported feeling unsure in combination with other response options. Supplementary Figure 1

Forty-five percent reported a constant pattern of illness, with the proportion higher in those with a Long COVID diagnosis (55.4%) than those without (35.7%) (Table 1). A higher proportion of participants without a Long COVID diagnosis (13.4%) reported relapsing/remitting symptom pattern than those with a diagnosis (5.9%). Of 637 participants reporting fluctuating or relapsing pattern, 40% reported that their illness (change in symptom intensity or relapse) was usually triggered by an identifiable factor. A further 30% reported that their illness (change in symptom intensity or relapse) generally followed a set/cyclical pattern but was occasionally triggered by something, and 10% reported that they had been unable to identify a trigger. Common triggers for change in symptom intensity or relapse were physical activity (44%), stress (40%), cognitive effort (27%), social effort (25%), and work (21%).

Forty percent of participants reported decrease in symptom intensity over the course of the illness (since infection), with a further 19% also reporting decrease in symptom intensity alongside the appearance of new symptoms. A total of 73% of participants with and 48.5% of those without a Long COVID diagnosis reported never having had a symptom-free day. Most common symptoms at follow-up were exhaustion (67.8%), postexertional symptom exacerbation (62.5%), and cognitive dysfunction (brain fog 62.3%, poor concentration 51.1%, memory problems 49.2%, and difficulty processing information 48.6%) (Supplementary Table 2). Postexertional symptom exacerbation and difficulty processing information were not collected at baseline but exhaustion and cognitive dysfunction were also the most common symptoms at baseline and in line with existing evidence and international case definitions (eg, WHO [1]) for Long COVID.

Among participants reporting experiencing a constant symptom pattern at baseline survey (n = 285), more than half (61.1%) reported continuing to experience a constant pattern of illness, 28.1% experiencing fluctuating, and 10.9% experiencing a relapsing pattern at follow-up survey(Figure 1A). Among those reporting experiencing a relapsing symptom pattern at baseline survey (n = 157), 35.0% continued to experience a relapsing pattern, 35.7% experienced fluctuating, and 29.3% experienced a constant symptom pattern at follow-up survey.

Being ill affected leisure activities (79.4%), social activities (72.2%), domestic chores (67.2%), job (61.1%), mental health (59.0%), and personal relationships (51.4%) at the follow-up survey (Table 2). Data on illness affecting personal relationships were not collected in the baseline survey but the activities most commonly affected by illness from the baseline survey were the same as in the follow-up survey. A higher proportion reported illness affecting domestic chores (86.7% at baseline, 67.2% at follow-up) and work (76.1% at baseline, 61.1% at follow-up) at baseline.

Participants experiencing a relapsing symptom pattern at baseline were less likely to report having a Long COVID diagnosis at follow-up (adjusted odds ratio [aOR] 0.43; 95% confidence interval [CI], 0.28-0.67) compared to those experiencing a constant pattern of illness (Table 3). Compared to participants experiencing none or negligible functional limitations at 6 weeks from start of illness, participants experiencing moderate (aOR 2.97; 95% CI, 1.75-5.05) and severe (aOR 3.82; 95% CI, 2.23-6.54) functional limitations were more likely to have a Long COVID diagnosis.

An equal proportion reported being unable to work at baseline (20.4%, n = 235) and follow-up (20.6%, n = 237) (Table 2). However, a higher proportion reported being made redundant or taking early retirement at follow-up (8.9%, n = 102) than at baseline (2.2%, n = 25). A further 59 participants (5.1%) reported being employed but unable to work (on paid or unpaid sick leave) at follow-up.

A higher proportion of participants with a Long COVID diagnosis reported being unable to work at follow-up (22.8%), which was a decrease from baseline (28.3%). The pattern was the opposite in participants without a Long COVID diagnosis, with 13.4% reporting being unable to work at baseline increasing to 19.5% at follow-up. A higher proportion of participants with a Long COVID diagnosis reported being employed and unable to work (7.9%) than those without a diagnosis (2.7%). Compared to participants employed full-time at baseline survey, those reporting being unable to work (aOR 2.56; 95% CI, 1.72-3.52) were more likely to report having a Long COVID diagnosis (Table 3). Participants reporting working reduced hours (aOR 1.61; 95% CI, 1.04-2.50) were more likely to report a Long COVID diagnosis, but this was attenuated on adjusting for symptom pattern and functional status.

A total of 209 (18.1%) participants reported losing, resigning from, or leaving their job because of Long COVID at follow-up compared with 170 (14.8%) participants at baseline. A higher proportion of participants with a Long COVID diagnosis reported resigning from or leaving their job (13.4%) than those without a diagnosis (10.5%).

A total of 307 (26.6%) participants reported not taking time off sick due to Long COVID at baseline, which decreased to 122 (10.6%, 4.6% in those with a Long COVID diagnosis and 15.4% in those without) at follow-up. A total of 11.5% reported taking unpaid leave. Of the 656 individuals reporting length of time off sick, 354 (54%) were off sick for more than 3 months, with 113 (17.2%) being off sick for more than 12 months at follow-up. A total of 169 participants with a Long COVID diagnosis reported being off sick for 6 months or more (48.7%), more than double the proportion in those with a diagnosis (21.0%, n = 59). Nearly half (47%, n = 538) reported a loss in income, increasing from 37.6% (n = 433) at baseline.

More than half (53.2%) of participants that reported being unable to work at baseline were still unable to work at follow-up with 17.1% reporting working reduced hours (Figure 1B). A total of 26.6% participants who reported being unable to work at baseline reported being employed full- or part-time at follow-up. A high proportion of participants that reported being employed at baseline were employed at follow-up but 11.3% of those employed full-time and 15.5% employed part-time at baseline reported being unable to work at follow-up. A total of 10.9% of participants employed full-time and 8.7% employed part-time at baseline reported working reduced hours at follow-up. Nearly one third (30.8%) of participants who reported working reduced hours or a phased return to work at baseline were still working reduced hours or a phased return at follow-up. A total of 17.5% of participants who were furloughed at baseline reported losing (3.2%) or resigning from or leaving their job (14.3%) at follow-up (Figure 1C). Twenty-one percent of participants who reported no loss of income because of COVID-19 illness at baseline reported a loss of income at follow-up (Figure 1D).

Of the 54 participants reporting full recovery in this sample, 13.3% (n = 7) reported experiencing Long COVID symptoms for 1–3 months before recovering (Figure 2). Thirteen participants (24.5%) reported experiencing Long COVID for ≥12 months before recovery. Most participants were symptom-free for at least 1–2 months (44.4%, n = 24) before considering themselves recovered, with 6 participants (11.1%) reporting being symptom-free for more than 6 months before considering themselves recovered.

This is a nonrepresentative sample recruited through online support groups and generally through social media using convenience nonprobability sampling. This is likely a highly self-selecting group and could overrepresent those who are more severely affected or more engaged in research. The study sample was recruited at a time when research into Long COVID was still in its infancy. Participants were predominantly White, female, and with higher educational attainment; findings therefore cannot be generalized to groups not represented among participants and cannot be used to calculate the prevalence of severity levels among people with Long COVID. The data were collected through online questionnaires, and we attempted to keep both surveys as short as possible to be manageable for participants. There is a possibility of recall bias in the baseline survey as the data about the acute stage were collected retrospectively; however, ongoing symptoms/experiences in both baseline and follow-up surveys were reported at the time point of data collection. Individuals with more symptoms or more severe symptoms may have been more likely to respond to the follow-up survey. The follow-up survey was available to complete for a 4-week period, and a 52% follow-up rate was achieved.

A key strength of this survey is that both baseline and follow-up surveys were co-produced with people with Long COVID. The initial idea for the survey came from people with Long COVID, and they were involved throughout the research. We additionally implemented feedback in an iterative manner from people with Long COVID. They were invited to give feedback from a post in the COVID-19 Research Involvement Group, and group members tested initial versions of both surveys. We changed survey platforms from the first to the second survey so that participants had the option of returning to complete the survey at a later date based on feedback that this made it more feasible for participants to participate in the study making it more inclusive. We captured lived experience with our analysis demonstrating that many people are still struggling to get recognition and diagnosis of Long COVID.

This research demonstrates the continued impact of Long COVID on daily activities and work in a sample of predominantly healthy adults before infection. Further research in a representative population sample is needed to characterize the effect on working patterns in people with Long COVID, particularly in those who may be less able to take time off to recover because of less flexible or more physically demanding occupations, and the effect of clinical recognition of Long COVID and workplace accommodations.