Characterizing Long COVID in an International Cohort: 7 Months of Symptoms and Their Impact

Objectives of study

The aim of this study is to better describe the patient experience and recovery process in those with confirmed or suspected COVID-19 illness, with a specific emphasis on the Long COVID experience. The unique approach of this study utilizes patient-driven research in order to establish a foundation of evidence for medical investigation, improvement of care, and advocacy for the Long COVID population. The survey was created by a team of patients with COVID-19 who are members of the Body Politic online COVID-19 support group. The group conducted its first survey in April 2020 and issued a subsequent report in May 2020 [5]. In order to better investigate additional aspects of patient experience, a second survey was developed, emphasizing symptom course and severity over time with an in-depth look into neurological and neuropsychiatric symptoms, recovery, and return to baseline, including work impact. Other topics investigated in the survey will be included in future reports.

Study design

Data were collected using an online survey hosted on the Qualtrics platform. All respondents gave digital informed consent prior to participating. Survey responses contained no personally identifiable information, and email addresses collected for survey distribution were hashed into anonymized ID codes. The study was approved by the UCL Research Ethics Committee [16159.002] (London, UK), and Oregon Health and Science University Institutional Review Board (IRB) (Portland, OR, USA), with UCL serving as the primary site. The Weill Cornell Medical College IRB determined non-engagement.

The survey consisted of 257 questions and required a median time of 69.3 minutes to complete. 61.6% of respondents completed the survey after giving consent. Methods used to distribute the survey did not allow us to determine the number of people who viewed the invitation, so the response rate could not be calculated. To account for post-exertional malaise and brain fog, which are common Long COVID symptoms that limit sustained focus and attention, respondents were encouraged to take breaks while completing the survey. Progress was saved for up to 30 days to allow respondents to return to the survey at a later time.

The survey was created in English and translated into eight additional languages: Spanish, French, Portuguese, Italian, Dutch, Russian, Bahasa Indonesian, and Arabic. Links to the survey were disseminated via email, social media, and online patient support groups listed in Appendix A. Data included in the analysis were collected from September 6 to November 25, 2020.

Patient and Public Involvement

The study was designed and conducted by a team of patients with Long COVID (all authors), who formed the Patient-Led Research Collaborative in May 2020. During the curation of survey questions, inputs from patients on various support groups, including Body Politic COVID-19 Slack support group, and Long COVID Support group on Facebook, were incorporated. In specific, we worked closely with other patients to compile the list of symptoms, design research questions on how the Long Covid condition may affect daily life of the patients, and optimise the questionnaire design in order to reduce survey fatigue.

Participants - Inclusion criteria

The survey “Information Sheet” stated: “You are being invited to participate in this research study because you have had a COVID-19, or suspected COVID-19 infection (still suffering or suffered symptoms) for longer than 1 week and you are 18 years of age or older.” All respondents consented to these criteria. To characterize Long COVID symptoms over an extended period, analysis was limited to respondents with illness lasting longer than 28 days and symptom onset between December 2019 and May 2020.

A total of 7,285 responses were downloaded from the Qualtrics server on November 25, 2020. The following responses were removed from the dataset: incomplete (those not reaching the end of the survey, n=2,367), no illness onset date (n=2), onset date before December 2019 (n=26), 0 days of symptoms (n=1), duplicate users with the same IDs (same email addresses with same demographic information, n=150), symptoms for 28 days or less (n=401), and illness onset after May 2020 (n=576). This resulted in complete data from 3,762 respondents.

SARS-CoV-2 test results

In addition to positively tested subjects [n=1,020, either diagnostic (RT-PCR/antigen, Table 1) or antibody], we included participants with absent (n=1,819) or negative test results (n=923, diagnostic and antibody), since the prevalence of 203 symptoms (out of 205, see Section “Symptoms by test results”), as well as symptom trajectory (Figure 7) and the survival functions (Figure 1a), were not significantly different between diagnostic/antibody positive and negative or untested groups. This strongly suggests SARS-CoV-2 infection in the diagnostic/antibody negative (or untested) group, and we therefore included respondents who did not test positive with a SARS-CoV-2 diagnostic or antibody test in most of our analyses (see Supplemental Figures S7-S8 for separate analyses).

• Download figure
• Open in new tab

Figure 1.

a) Survival function (Kaplan-Meier estimator), characterizing the distribution of disease duration for those who tested Negative (blue) on both diagnostic (RT-PCR/antigen) and antibody tests, those who tested Positive (orange) in either diagnostic or antibody test, and All (green) respondents. The Y axis indicates the probability that symptoms will persist longer than the time specified on the X axis. b) Probability of each symptom severity score over time. c) Average number of reported symptoms over time for those who recovered in less than 90 days (n=154), or otherwise experienced symptoms for more than 90 days (n=3505). a-c) In all plots, times are relative to initial symptom onset. Shaded regions represent 95% simultaneous confidence bands.

View this table:

• View inline
• View popup
• Download powerpoint

Table 1. Testing status

Statistics and data analysis

Variables

In this study we quantified symptom prevalence, probability time-course, severity, count, onset time, temporal clustering. We also measured fatigue using Fatigue Assessment Scale. Return to baseline and working status were also measured.

Prevalence estimation

205 symptoms were investigated by identifying their presence or absence. For 74 of these symptoms, respondents indicated at which points in their illness (weeks 1-4, months 2-7) they experienced the symptom, if at all. For each of the other 131 symptoms, participants indicated whether they had experienced the symptom at any point throughout the duration of their illness (Figures 2, 3). Prevalence estimates were calculated by dividing the number of those who identified experiencing a symptom, either at a given time point (Figure 4) or over the entire sickness period (Figure 2, 3), by the total number of participants to which the symptom applied. Eight symptoms were excluded from analysis, as their measurement required specialized equipment or tests that many participants may not have had access to. Excluded symptoms were high blood pressure, low blood pressure, thrombosis, seizures (confirmed or suspected), low oxygen levels, high blood sugar, and low blood sugar. The remaining 66 symptoms (out of 74) were included in analysis of the timeline of disease progression over 7 months (see below, Figure 4). Each symptom was further categorized into one of ten organ systems (Supplemental material, Appendix A), which were visualized as groups. The respondents for certain symptoms (non-primary language and reproductive/genitourinary symptoms) consisted of the subset of total respondents whom the symptom could apply to (i.e. those who spoke more than one language, cisgender female or non-binary and menstruating, cisgender female or non-binary and above or below 40 years of age, and male). Therefore, the symptom prevalence was calculated within the relevant subsample.

• Download figure
• Open in new tab

Figure 2: Symptom prevalence estimates (non-neuropsychiatric symptoms).

Bars represent the percentage of respondents who experienced each symptom at any point in their illness. Symptoms are categorized by the affected organ systems. When all rows in a given panel use the same denominator, the first row, labeled “All,” indicates the percentage of respondents who experienced any symptoms in that category. Error bars are 95% confidence intervals.

Symptom time course estimation

The survey asked respondents to detail their experience of a subset of 66 symptoms over time. Respondents indicated whether each of these symptoms was present during a series of time intervals following the onset of their first symptoms: week 1 (days 1-7), week 2 (days 8-14), week 3 (days 15-21), week 4 (days 22-30), month 2 (days 31-60), month 3 (days 61-90), month 4 (days 91-120), month 5 (days 121-150, month 6 (days 151-180), and month 7 (days 181-210).

The time course of each symptom was defined as the probability of experiencing the symptom in each time interval, given that: 1) recovery had not occurred prior to the end of the interval, and 2) the symptom was applicable (menstruation-related symptoms are presented only for menstruating respondents). Probabilities were estimated for each interval as the fraction of respondents who experienced the symptom, among those who satisfied the two criteria above. The number of qualifying respondents in each time interval is given in Table 2.

View this table:

• View inline
• View popup
• Download powerpoint

Table 2. Number of qualifying respondents in each time interval

Plotted time courses in Fig. 4 are linearly interpolated between the centers of each time interval.

Symptom severity and count

Overall symptom severity for each time interval (weeks 1-4, month 2-7) was measured using Likert scale (no symptom, very mild, mild, moderate, severe, very severe). The probability of each Likert option was calculated as the fraction of qualifying participants (as described above) who selected that option (Figure 1b). Total number of experienced symptoms (from the subset of 66) was measured for each qualifying respondent in each time interval. The mean value of symptom count was then calculated by averaging over all qualifying respondents.

Symptom onset analysis

The heatmap in Figure 5 shows the probability distribution of onset times for each symptom. Continuous, piecewise-constant distributions were fit using maximum likelihood, accounting for interval censoring (onset times for each respondent could only be measured up to the enclosing time intervals, described above). For each symptom, the estimated probability density at time t was given by the fraction of respondents who first experienced the symptom in the interval containing t (among those who experienced it at any point), divided by the duration of the interval. Mean onset time was calculated as the expected value of the estimated distribution.

Symptom time course clustering

Symptom time courses were clustered using spherical k-means, a variant of k-means based on cosine distances [12]. Each time course is a 10-dimensional vector, representing the conditional probability of experiencing the symptom in each of the 10 time bins (as above). The cosine distance is a monotonic function of the angle between vectors, and is insensitive to their magnitudes. Therefore, it is well suited to measuring differences between time course shapes (i.e. changes in relative amplitude over time), while remaining invariant to differences in overall symptom prevalence. We used a variant of Lloyd’s algorithm designed for spherical k-means, with initialization by the k-means++ algorithm, and 100 random restarts to avoid suboptimal local minima. The number of clusters (k=3) was chosen by hand, to provide a reasonable tradeoff between capturing structure in the data and obtaining a parsimonious explanation.

Symptom time course sorting

The heatmaps in Figure 6 and Figure S3 show normalized symptom time courses, sorted such that similarly-shaped time courses appear nearby in the ordering. The sort ordering was computed as follows. Similarity between time courses was measured using the cosine distance, as above. Classical multidimensional scaling (MDS) was then used to embed time courses into a one-dimensional Euclidean space, such that pairwise distances in the embedding space approximated the given cosine distances. Time courses were sorted according to their order in the embedding space.

Confidence intervals

All confidence intervals and confidence bands were estimated using a nonparametric bootstrap approach with 10,000 iterations. Individual confidence intervals and pointwise confidence bands used the bias-corrected, accelerated (BCa) bootstrap [13]. Simultaneous confidence bands used the percentile bootstrap, with the percentile adjusted to give the correct simultaneous coverage probabilities.

Stratification based on the diagnostic test time

Test time, defined as the number of days between first experiencing symptoms and receiving a diagnostic test (RT-PCR or antigen), was significantly shorter in respondents with positive results compared to those with negative results (see Symptoms by test results, and Supplemental Figure S6). To compare the symptom prevalence and time course (Supplemental Figure S7-S8), the two groups were stratified based on the test time by including respondents who were tested within certain time windows after illness onset. For prevalence estimates, time windows were ‘smaller than 10 days’, ‘between 10 and 20 days’, and ‘larger than 20 days’. For time course estimates, we limited the comparison to those who have been tested within 20 days after illness onset. Data stratification based on test time was done only for the diagnostic test.

Text analysis

The survey asked respondents to elaborate on their experience in free text for the following areas: body parts for sensorimotor symptoms, brain fog and memory issues, most debilitating symptoms, other diagnosis post illness, and work status. Deductive thematic analysis was used to tag and extract themes around impact on work [14]. For textual input on participants’ experience of symptoms, such as cognitive dysfunction, a range of quotes was selected to provide a deepened understanding of the diversity of experiences [15]. Identifying data were anonymized and longer sentences were truncated for brevity. For the sensorimotor textual input questions, which asked which body part was affected, natural language processing was used in Python. The text was converted to lowercase, stripped of punctuation and extra whitespace, and stopwords were removed (using the original stopwords list from the NLTK library as well as common non-symptom text inputs [16]). The text was tagged for the parts of speech using a word tokenizer, and only nouns were reserved. The nouns were run through a translation function to convert all non-English nouns to their English counterparts, then counted, and the top four body parts were added to the table. The answers to “most debilitating symptoms’’ followed a similar process, without the parts-of-speech tagging; another function was written to group similar descriptions (e.g. cognition, brain fog, and difficulty concentrating all went under “cognitive dysfunction”).

Demographics

This study included 3,762 survey respondents based on the eligibility criteria described above. Detailed demographics are listed in Table 3. The majority of respondents were women (78.9%, significantly more than other genders, p < 0.001, chi-squared test), white (85.3%, p < 0.001, chi-squared test), and between the ages of 30 and 60 (33.7% between ages 40-49, 27.1% ages 50-59, 26.1% ages 30-39). A total of 56 countries were represented in the sample. Most of the respondents resided in the United States (41.2%, p < 0.01, Tukey’s HSD multiple comparisons test). 91.9% of respondents completed the survey in English.

More than half of respondents (56.7%, p < 0.001, chi-squared test) did not seek hospital-based care. 34.9% visited an ER or urgent care clinic but were not admitted to a hospital. 8.43% of respondents were hospitalized. 17.8% of respondents were healthcare workers (see Supplemental Material, Appendix A, for pre-existing conditions).

Symptoms and severity over time

Symptoms experienced at any point

Participants were asked to indicate the presence or absence of 205 symptoms. For 74 of these symptoms, participants indicated at which points in their illness (weeks 1-4, months 2-7) they experienced the symptom, if at all. Eight symptoms were excluded from the main analyses (Methods, Supplemental Figure S5). Symptoms were divided into 10 categories representing the organ system in which they are present (see Appendix A, Table 4). Prevalence estimates for 131 symptoms were estimated as the percentage of respondents who experienced each symptom at any point (Figure 2 for non-neuropsychiatric and Figure 3 for neuropsychiatric symptoms, see Methods). In some cases, such as reproductive symptoms, the denominator used to calculate the percentage varied depending on factors such as respondent age or sex (see Methods).

• Download figure
• Open in new tab

Figure 3: Symptom prevalence estimates for neuropsychiatric symptoms.

Similar to Figure 2, for neuropsychiatric symptoms, divided into nine sub-categories. Each bar represents the percentage of respondents who experienced that symptom. Error bars are bootstrap 95% confidence intervals.

View this table:

• View inline
• View popup
• Download powerpoint

Table 4. Overall System Prevalence Data

Symptoms over time

To characterize the progression of the 66 symptoms over seven months, we estimated symptom time courses—the probability of experiencing each symptom at each time point, given that recovery has not yet occurred (Figure 4, see Supplemental Table S22 for the raw data; Supplemental Figure S9 for male vs. female comparison). We also estimated the distribution of onset times for each symptom (Figure 5).

• Download figure
• Open in new tab

Figure 4: Symptom time courses.

Plotted time courses represent the estimated probability of experiencing each symptom at each time point, given that recovery has not yet occurred (see Methods). Times are relative to initial illness onset. Symptoms are grouped according to the affected organ systems. Shaded regions show 95% simultaneous confidence bands, estimated separately for each symptom.

• Download figure
• Open in new tab

Figure 5. Symptom onset times.

Heatmap shows the estimated probability distribution of the onset time for each symptom. White points and error bars show the mean onset time and 95% pointwise confidence intervals. Symptoms are sorted by mean onset time.

To summarize general patterns in the progression of symptoms over time, we used a clustering algorithm (see Methods) to group the 66 symptoms into three clusters, according to the shapes of their time courses (i.e. changes in relative amplitude over time, ignoring their overall prevalence). Symptoms that clustered together generally had similarly-shaped time courses (Figure 6). Cluster 1 consists of symptoms that are most likely to appear early in the illness (reaching a high point in the first two or three weeks), followed by a decreasing trend in probability over time. Cluster 2 consists of symptoms with a slow decrease, slow increase, or unchanging probability over time. On average, symptoms in this cluster exhibit a slightly increased probability of presenting in the second month of illness. Cluster 3 consists of symptoms that are most likely to ramp up sharply in the first two months. Their probability may hit a plateau (like constipation), decrease slightly (like post-exertional malaise and fatigue), or increase slightly in the later months (like tinnitus, hearing loss, muscle spasms, and tremors). All clusters contained symptoms from multiple organ systems, and cluster 3 contained symptoms from all organ systems (with the exception of pulmonary/respiratory symptoms). A general progression from early to late symptoms can also be seen in the heatmap of normalized time courses (Figure 6, S2), which have been sorted by similarity in shape (see Methods).

• Download figure
• Open in new tab

Figure 6. Symptom clusters, based on temporal similarities.

Plots (top row) show time courses for the symptoms in each cluster (in grey) and their mean (Cluster 1 in blue, Cluster 2 in orange, Cluster 3 in green). Time courses have been scaled separately for each symptom (by root mean squared amplitude) to visually compare their shapes. The table lists symptoms in each cluster, grouped by the affected organ systems. The heatmap (bottom row) shows time courses for all symptoms, sorted such that similarly shaped time courses are adjacent (see Methods). Columns have been scaled by their maximum amplitudes for visual comparison. Symptoms are numbered according to their table entries.

Symptom prevalence plots, together with the onset times and clusters (Figures 2-6) show that experienced symptoms affect multiple organ systems. The mean number of organ systems affected in each respondent was 9.08 (95% confidence interval 9.04 to 9.13; see Symptom Details). Symptoms in the same organ-based category did not necessarily cluster together, and could appear across clusters. This indicates that symptoms affecting the same organ system can have differently shaped time courses and, conversely, symptoms affecting different organ systems can have similarly shaped time courses. Systemic and neurological/cognitive symptoms were the most likely to persist from disease onset to month 7 (see Symptom Details).

Symptoms by test result

Among respondents who received a diagnostic test (RT-PCR or antigen) for SARS-CoV-2 at any point during their illness, 1,730 tested negative and 600 tested positive. The primary difference between these two groups was the time elapsed between symptom onset and testing, with a median of 6 days for those who tested positive and 43 days for those who tested negative (p < 0.001, Mann-Whitney U test) (Supplemental Figure S6). Symptoms were remarkably similar between the two groups. We compared symptom prevalence among positively and negatively tested respondents, stratified by test time. Out of 205 symptoms, 203 showed no significant difference at the 5% level (Fisher test, Bonferroni corrected). The loss of smell and taste were the only exceptions (loss of smell: 22.2% (negative) vs 60.8% (positive), p < 0.0001; 21.5% loss of taste: 21.5% (negative) vs. 54.9% (positive), p < 0.0001; Fisher test, Bonferroni corrected). In addition, 683 participants tested positive for SARS-CoV-2 antibodies (either IgG, IgM, or both). Similarly, the loss of smell and taste were the only significantly different symptoms when comparing prevalence among respondents who tested negative (diagnostic and antibody, 21.6% loss of smell, 25.3% loss of taste) versus positive (diagnostic or antibody, 60.0% loss of smell, 52.5% loss of taste), stratified by test time (p < 0.0001, Fisher test, Bonferroni corrected).

Furthermore, respondents experienced similar variation in symptoms over time, despite differences in testing status. For 65 out of 66 symptoms, time courses overlapped substantially between participants with confirmed COVID-19 (n=1,020, positive RT-PCR, antigen, or antibody test at any point) and participants with no positive test result (n=2,742, Figure 7). As above, change in smell/taste was the lone exception. Similar overlap was observed when separately comparing positively tested participants to negatively tested and untested participants (Supplemental Figures S7 and S8).

• Download figure
• Open in new tab

Figure 7. Symptom time courses for participants with COVID-19 confirmed via testing vs. rest of the population.

Plots show symptom time courses (similar to Fig. 4) for respondents who were confirmed COVID-positive via diagnostic or antibody testing (orange) vs those without a positive confirmation (untested, or tested negative, blue). Shaded regions show simultaneous 95% confidence bands. Symptom names are colored according to the affected organ systems.

Symptom Details

Neuropsychiatric

Brain fog/Cognitive dysfunction and memory impairment

85.1% (95% confidence interval 83.9% to 86.2%) of respondents (3203) reported experiencing brain fog and cognitive dysfunction (Figure 3d, Supplemental Table S13). The most common symptoms were poor attention or concentration 74.8% (73.4% to 76.2%), difficulty thinking 64.9% (63.4% to 66.4%), difficulty with executive functioning (planning, organizing, figuring out the sequence of actions, abstracting) 57.6% (56.0% to 59.1%), difficulty problem-solving or decision-making 54.1% (52.4% to 55.6%), and slowed thoughts (49.1%, 40.2% to 43.4%). For 31.2% (29.7% to 32.7%) of respondents, onset of brain fog/cognitive dysfunction occurred in their first week of symptoms. Reports of cognitive dysfunction increased over the first three months, peaking at 66.7% (65.1% to 68.2%), then decreased slightly in the following months. 55.5% (52.5% to 58.8%) of month 7 respondents experienced cognitive dysfunction during month 7 (Figure 4j).

72.8% (71.4% to 74.2%) of all respondents (2739) experienced memory impairments (Figure 3e, Supplemental Table S14). Of those, 64.8% (63.3% to 66.4%) experienced short-term memory loss, 36.12% (34.6% to 37.6%) experienced long-term memory loss, 12.0% (11.0% to 13.1%) forgot how to do routine tasks, and 7.3% (6.5% to 8.2%) were unable to make new long-term memories. The likelihood of experiencing memory symptoms increased the first few months, with 55.9% (54.3% to 57.5%) reporting memory symptoms in month 4. 50.5% (47.3% to 53.6%) of respondents with symptoms for over 6 months experienced memory symptoms in month 7 (also Figure 4j).

Of those who experienced memory and/or cognitive dysfunction symptoms and had a brain MRI, 87% of the brain MRIs (n=345, of 397 who were tested) came back without abnormalities.

Impact of cognitive dysfunction/memory on daily abilities and impact by age

88.0% of the total respondents (3310) experienced either cognitive dysfunction or memory loss (Figure 8). The greatest area of impact reported was on work, with 86.2% (95% confidence interval 84.4 to 88.0%) of working respondents feeling mildly to severely unable to work - 29.1% (26.7% to 31.6%) severely. This is reflected in the working status of respondents, discussed in the Impact on Work section below. Other areas of impact included making serious decisions 85.3% (80.7% to 89.8%), communicating thoughts and needs 74.8% (72.5% to 77.1%), having conversations with others 68.3% (65.8% to 70.8%), maintaining medication schedules 62.5% (59.8% to 65.1%), following simple instructions 54.4% (51.6% to 57.2%), and driving 53.2% (50.5% to 56.0%). See Figure 8d for the full list.

• Download figure
• Open in new tab

Figure 8. Memory and cognitive dysfunction.

a) Percentage of respondents in six age groups who experienced different types of memory impairments. b) Same as (a) for cognitive dysfunction. c) Impact of memory and cognitive dysfunction on work (for those who work), for different age groups. Participants were asked to rate the impact by choosing one of the four options “Able, Mildly unable, Moderately unable, and Severely unable”. d) Overall impact of memory and cognitive dysfunction on daily life. Participants to whom the question was not applicable were excluded. Error bars show 95% confidence intervals.

Memory symptoms, cognitive dysfunction, and the impact of these on daily life were experienced at the same frequency across all age groups (Figure 8a-c). Selected quotes from respondents who described specific instances of memory loss or brain fog can be found in Appendix B.

Speech and language

Speech and language issues occurred in 48.6% (95% confidence interval 47.0% to 50.2%, Supplemental Table S15) of respondents (Figure 3h). The most common speech/language symptom was word retrieval, with 46.3% (44.8% to 47.9%) of respondents having difficulty finding words while speaking or writing. 29.2% (27.8% to 30.7%) of respondents had difficulty communicating verbally, 24.8% (23.3% to 26.1%) had difficulty reading/processing written text and 23.8% (22.5% to 25.2%) had difficulty processing/understanding others. 28.9% (27.1% to 23.6%) of those who spoke two or more languages had changes to their non-primary language. Speech and language symptoms occurred in 13.0% (12.0% to 14.1%) of respondents in the first week, increasing to 40.1% (38.5% to 41.7%) experiencing these issues in month 4. 38.0% (34.5% to 41.0%) of respondents with symptoms for over 6 months reported speech and language symptoms in month 7 (Figure 4k).

Sensorimotor symptoms

Sensorimotor symptoms encompass a collection of symptoms: tremors, “vibrating sensation”, numbness, coldness in a body part, tingling/pins and needles, “electric zap,” facial paralysis, facial pressure/numbness, and weakness (Figure 3b). These were experienced by 80.5% (95% confidence interval 79.3% to 81.8%) of respondents, occurring in 32.0% (30.5% to 33.4%) of week 1 respondents and increasing to 59% (57.5% to 60.7%) of month 4 respondents (Figure 4k). Tingling, prickling, and/or pins and needles was the most common at 49% (47.7% to 50.8%) of respondents. See Supplemental Table S3, and Table S16 for the most commonly affected anatomical locations.

Sleep

78.6% (95% confidence interval 84.0% to 79.9%) of respondents experienced difficulty with sleep (Figure 3c, Supplemental Table S17). Table 5 lists each type of sleep symptom, as well as the percentage of respondents with that symptom who also listed it as pre-existing (before COVID-19 infection).

View this table:

• View inline
• View popup
• Download powerpoint

Table 5. Prevalence of sleeping issues before and during illness

Headaches

Headaches were reported by 77.0% of participants (95% confidence interval 75.4% to 78.0%, Supplemental Table S18), with the most common manifestations being ocular 40.9% (38.6 to 41.7%), diffuse 35.0% (33.6% to 36.7%), and temporal 34.0% (32.4% to 35.5%) (Figure 3f). 24.0% (22.5% to 25.2%) of respondents reported headaches after thinking/mental exertion and 23.0% (21.9% to 24.6%) experienced migraines. Of those experiencing migraines, 56.4% did not list migraines as a pre-existing condition. 46% of all respondents reported headaches during week 1, 54% of respondents experiencing symptoms in month 4 reported headaches in month 4, and 50% of respondents experiencing symptoms in month 7 reported headaches in month 7 (Figure 4l).

Emotion and mood

Changes to emotion and mood were reported by 88.3% (95% confidence interval 87.2% to 89.3%Supplemental Table S19) of participants (Figure 3a). Anxiety was the most common psychological symptom reported at 57.9% (56.4% to 59.5%), followed by irritability at 51.0% (49.5% to 52.7%). Depression was reported by 47.3% (45.7% to 48.9%) with 39.2% (37.6% to 40.7%) experiencing apathy. Mood lability, assessed by “mood swings” and “difficulty controlling emotions,” was reported by 46.3% (37.6% to 40.7%). Suicidality was reported by 11.6% (10.6% to 12.6%), and mania and hypomania were reported at 2.6% (2.1% to 3.1%) and 3.4% (2.8% to 4.0%), respectively. Of those who reported anxiety, 61.4% (59.4% to 63.4%) had no anxiety disorder prior to COVID. Of those who reported depression, 53.12% (50.8% to 55.4%) had no depressive disorder prior to COVID.

Taste and smell

Changes to taste and smell (Figure 3g) were reported by 57.6% (95% confidence interval 56.0% to 59.2%, Supplemental Table S20), with no significant difference seen in loss of smell (35.9%, 34.4% to 37.5%) vs. loss of taste (33.7%, 32.2% to 35.2%, p > 0.1, chi-squared test). Altered sense of taste was experienced by 25.1% (23.7% to 26.4%) of respondents, phantom smells (i.e. olfactory hallucinations or phantosmia) by 23.2% (21.9% to 24.6%) of respondents, and altered sense of smell by19.8% (18.5% to 21.1%) of respondents. Phantom smells were accompanied by a write-in question asking for a description of the smells, in which the most common words were “smoke,” “burning,” “cigarette,” and “meat.”

Changes to smell and taste were more likely to occur earlier in the illness course, with 33.2% occuring in week 1. 25.2% (22.5% to 28%) of respondents with symptoms for over 6 months experienced changes to taste and smell in month 7 (Figure 4k).

Hallucinations

The most common hallucination reported was olfactory hallucinations 23.2% (21.9% to 24.6%, Supplemental Table S21), mentioned above (Figure 3i). Visual hallucinations were reported by 10.4% (9.5% to 11.4%) of respondents, auditory hallucinations by 6.5% (5.7% to 7.3%), and tactile hallucinations by 3.1% (2.6% to 3.7%).

Post-exertional malaise

The survey asked participants whether they have experienced “worsening or relapse of symptoms after physical or mental activity during COVID-19 recovery” [11]. Borrowing from Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) terminology [20], this is referred to as post-exertional malaise (PEM). 89.1% of participants (95% confidence interval 88.0% to 90.0%) reported experiencing either physical or mental PEM.

Of the respondents who experience PEM triggered by physical exertion, 49.6% (48.0% to 51.3%) experience it the following day, 42.5% (40.8% to 44.2%) experience it the same day, and 28.7% (28.3% to 31.3%) experience PEM immediately after (Figure 9). Of the respondents who experience PEM triggered by mental exertion, 42.2% (40.5% to 43.8%) experience it the same day, and 31.4% (29.9% to 33.0%) experience it immediately after. For some respondents the time PEM started varied. A high number of the respondents with PEM (68.3%, 66.4% to 69.6%) indicated that the PEM lasted for a few days. For physical exertion, the mean severity rating was 7.71, and for mental exertion, the mean severity rating was 5.47.

• Download figure
• Open in new tab

Figure 9. Worsening or relapse of symptoms after physical or mental activity (post-exertional malaise).

Worsening of symptoms or relapse after physical or mental exertion. When does it start (a), how long does it last (b), and how severe is it? (c) (all patients who experienced PEM, n=3350). d-e: What are the triggers for relapses/worsening of symptoms and the experience of symptoms over time and relapses?

Relapses: triggers & experience

Patients with Long COVID can experience relapsing-remitting symptoms [5]. Minimum of 85.9% (84.8% to 87.0%) of respondents reported experiencing relapses. Respondents characterized their relapses as occurring in an irregular pattern (52.8%, 95% confidence interval 51.2% to 54.4%) and in response to a specific trigger (52.4%, 50.8% to 54.0%). The most common triggers of relapses, or of general worsening of symptoms, that respondents reported were physical activity (70.7%, 69.2% to 72.1%), stress (58.9%, 57.3% to 60.5%), exercise (54.39%, 52.8% to 56.0%), and mental activity (46.2%, 44.7% to 47.8%). More than a third of menstruating participants experienced relapses during (34.3%, 32.0% to 36.5%) or before menstruation (35.2%, 33.0% to 37.3%). Heat and alcohol were other triggers of relapse.

Triggers that were written in by respondents included food with sugar and high histamines (reported by 70 respondents); lack of sleep or rest (64 respondents); cold air (39 respondents); overworking or schoolwork (28 respondents); smoke, pollution, and chemical odors (24 respondents);

Approximately half (51.7%, 50.1% to 53.3%) of respondents indicated that their symptoms have slowly improved over time, while 8.9% (7.9% to 9.8%) indicated that their symptoms have gradually worsened and 10.8% (9.9% to 11.8%) have had symptoms rapidly worsen over time.

Remaining symptoms after 6 months

Only 164 out of 3762 participants (4.4%) experienced a temporary break in symptoms (Figure S4). The remaining participants reported symptoms continuously, until symptom resolution or up to taking the survey. A total of 2454 (65.2%) respondents were experiencing symptoms for at least 6 months. For this population, the top remaining symptoms after 6 months were primarily a combination of systemic and neurological symptoms (Figure 10). Over 50% experienced the following symptoms: fatigue (80.0%, 95% confidence interval 78.5% to 81.6%), post-exertional malaise (73.3%, 71.5% to 75.1%), cognitive dysfunction (58.4%, 56.5% to 60.2%), sensorimotor symptoms (55.7%, 53.7% to 57.6%), headaches (53.6%, 51.5% to 55.5%), and memory issues (51.0%, 49.1% to 53.0%). In addition, between 30%-50% of respondents were experiencing the following symptoms after 6 months of symptoms: insomnia, heart palpitations, muscle aches, shortness of breath, dizziness and balance issues, speech and language issues, joint pain, tachycardia, and other sleep issues.

• Download figure
• Open in new tab

Figure 10. Remaining symptoms after 6 months.

a) Symptoms remaining after 6 months. b) Symptoms remaining after 6 months for respondents still experiencing PEM after 6 months (orange), respondents not experiencing PEM after 6 months (green), and respondents who never experienced PEM (blue). c) Average number of symptoms over time for each group in (b).

Respondents were also asked if they had been diagnosed with any number of conditions post-acute COVID-19 infection. Nearly half of respondents (43.4%) responded with at least one common diagnosis and/or elaborated on their diagnosis in free text (see Table S2 Appendix A).

Fatigue assessment

Participants answered the Fatigue Assessment Scale (FAS) questionnaire [21, 22], which includes ten questions that assess both physical and mental fatigue. FAS scores were calculated based upon participants’ subjective report during the “past one week.” Figure 11.a shows the distribution of FAS scores for the recovered (blue) and unrecovered (orange) participants. The scores were then summarized into three categories (Figure 11.b): no fatigue (scores of 10-21), fatigue (22-34), and extreme fatigue (≥35).

• Download figure
• Open in new tab

Figure 11. Return to baseline and work impact.

a) Distribution of Fatigue Assessment Scale scores for recovered (n=257, blue) and unrecovered (n=3505, yellow) population. The vertical dashed lines indicate the range for “No fatigue” (10-21), “Fatigue” (22-34), and “Extreme” (>=35). Mean values for each distribution are also marked. b) Percentage of participants in each of the three categories. c) Distribution of scores in response to “return to pre-COVID” health baseline, where 0 indicates worst (most different from baseline) and 100 indicates best (most similar to baseline). d) Working status due to COVID-19. Error bars show 95% simultaneous confidence interval.

We contrasted the FAS scored of unrecovered and recovered participants. Of the total of 257 who recovered, respondents experienced symptoms for a mean of 91 (IQR 49-121) and a maximum of 250 days. Those not recovered (n=3505) had experienced 144 days of symptoms on average at the time of the survey (IQR 176-213). On average, unrecovered participants had higher FAS scores than unrecovered participants (31.8 vs 22.2, P < 0.001, Mann-Whitney U-test). 55.3% (95% confidence interval 49.4% to 61.5%) of recovered participants were classified as having no fatigue. This is significantly more than the 13.2% (12.2% to 14.4%, P < 0.001, Mann-Whitney U-test) of unrecovered participants who experienced no fatigue at the time of survey. 40.7% (39.9% to 42.3%) of unrecovered participants were classified as experiencing extreme levels of fatigue, significantly higher than the 8.9% (5.8% to 12.8%) of recovered participants in this category (P < 0.001, Mann-Whitney U-test). 35.8% (30.0% to 41.6%) of recovered and 46.0% (44.4% to 47.7%) of unrecovered participants were classified as experiencing non-extreme fatigue (P < 0.001, Mann-Whitney U-test). FAS scores for the respondents who tested positive for COVID-19 and those who tested negative (with either a diagnostic or antibody test) are similar (P = 0.92, Mann-Whitney U-test, data not shown).

Return to baseline

Respondents were asked, “How would you rate how you feel today, on a scale of 0-100% (with 100% being your pre-COVID baseline)?” (Figure 11c). Unrecovered participants reported a mean score of 59.2, while recovered participants reported a mean score of 86.5 (p<0.001, Mann-Whitney U-test). “Pacing” was the treatment with the highest percentage of respondents considering it ‘significantly helpful’ (23.1% out of 1788 who tried it). 18.8% found it ‘slightly helpful’.

Impact on work

Of unrecovered respondents who worked before becoming ill, only 27.3% (95% confidence interval 25.3% to 29.4%) were working as many hours as they were prior to becoming ill at the time of survey, compared to 49.3% (40.8% to 57.9%) of recovered respondents (see Figure 11d). Nearly half 45.6% (43.2% to 48.0%) of unrecovered respondents were working reduced hours at the time of the survey, and 23.3% (21.3% to 25.4%) were not working at the time of the survey as a direct result of their illness. This included being on sick leave, disability leave, being fired, quitting, and being unable to find a job that would accommodate them. The remaining respondents retired, were volunteers, or did not provide enough information to determine their working status. Overall, 45.2% (42.9% to 47.2%) of respondents reported requiring a reduced work schedule compared to pre-illness. 22.3% (20.5% to 24.3%) were not working at the time of survey due to their health conditions.

Respondents had the option to elaborate on their situation in free text, and these responses captured the precariousness of working with Long COVID (see selected quotes in Appendix B). Several themes emerged through thematic analysis: At least 45% of working respondents were working remotely at the time of the survey, and it was noted how critical this was to respondents’ continued ability to work. Teleworking enabled respondents to take breaks when necessary and saved them the physical exertion of commuting to work. Respondents mentioned asking for other accommodations at work like flextime or moving to a role with lower physical or mental strain. Even with telecommuting, phased returns, and other accommodations, respondents commented on how difficult it was for them to work full or part-time, but described their financial need to do so.

It is important to note that the survey captured only a moment in time. Respondents described taking months of leave before going back to work either full-time or at reduced hours. Further, there were respondents who indicated that they tried to go back to work for several weeks but then relapsed or were unable to complete their work satisfactorily.

Principal findings

Results from this international online survey of 3,762 individuals with suspected or confirmed COVID-19 illness suggest that Long COVID is composed of heterogeneous post-acute infection sequelae that often affect multiple organ systems, with impact on functioning and quality of life ranging from mild to severe. A unique patient-led approach allowed for a thorough and systematic identification of possible symptoms based upon reports in online support networks. To our knowledge, this represents the largest collection of symptoms identified in the Long COVID population to date. While several others have investigated Long COVID symptoms [6, 10], our approach also allowed the first representation of individual symptom trajectory over time. The cohort was composed predominantly of individuals with continued symptoms at 6 months. Of the symptoms for which time course data were collected, the most likely early symptoms were fatigue, dry cough, shortness of breath, headaches, muscle aches, chest tightness, and sore throat. Importantly, while presence of fever has widely been used for screening purposes [23–25], we found only 30% of participants reported a fever, consistent with previous reports [24], while the majority did experience some combination of mild elevation in temperature (98.6 - 100.4 F), diaphoresis, temperature lability, and chills.

In this cohort, the most likely symptoms to persist after month 6 were fatigue, post-exertional malaise, cognitive dysfunction (“brain fog”), neurologic sensations (neuralgias, weakness, coldness, electric shock sensations, facial paralysis/pressure/numbness), headaches, memory issues, insomnia, muscle aches, palpitations, shortness of breath, dizziness/balance issues, and speech and language issues. Some symptoms, like bone aches, tinnitus, and other ear symptoms, increased in likelihood during and after month 6. Prolonged symptoms were most likely to be reported as “moderate” (36.6%, 95% confidence interval 32.2% to 40.9%). Notably, the probability of having “severe” or “very severe” symptoms after month 6 was more than 21% (severe: 14.5%, 11.3% to 18.5%; very severe: 5.2%, 3.1% to 7.1%). Respondents indicated that fatigue, breathing issues, and cognitive dysfunction were the most debilitating of symptoms. Those not recovered within three months experienced an average of 13 symptoms during week 1, increasing to 17 symptoms during month 2. They continue to experience an average of 14 symptoms after 6 months.

We propose clusters of symptoms in three groups, each with different morphologies over time. Importantly, the clusters of symptoms that persist longest include a combination of the neurological/cognitive and systemic symptoms. This indicates the need for a multidisciplinary approach to workup and care of the Long COVID population.

Dysautonomia, in part manifesting as Postural Orthostatic Tachycardia Syndrome (POTS), and Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) appear as highly possible diagnoses for this population [26]. By the time respondents took the survey, 155 had received a diagnosis of POTS, and 118 had received a diagnosis of ME/CFS. Based on the reported symptoms, higher percentages of respondents might meet the criteria for these diagnoses than have been clinically diagnosed. 33.9% of respondents who reported tachycardia measured an increase of at least 30 BPM when standing, suggesting a possible POTS diagnosis [27]. Given these findings, we suggest that all patients who present with any signs or symptoms of POTS, including tachycardia, dizziness, brain fog, or fatigue, be screened for POTS [19].

To investigate the possible overlap with ME/CFS in this population, we asked participants to identify whether they experienced worsening of symptoms after physical or mental exertion. Post-exertional malaise (PEM) is one of the three required symptoms for ME/CFS diagnosis, along with unrefreshing sleep and a reduction in ability to engage in pre-illness levels of activity [28]. Similar to cognitive dysfunction, we found PEM to be highly represented in this cohort (89.1% at any time during the course of illness, 72.2% at month 7). Intriguingly, among those still experiencing symptoms at month 6 with no PEM (n = 707, 28.8%), fatigue was still the most common symptom. This suggests that while a subset of the Long COVID population may meet ME/CFS diagnostic criteria, there remains a subpopulation with significant fatigue who do not meet the criteria, and therefore the mechanisms of fatigue and the degree to which ME/CFS adequately explains it require further investigation.

Participants also experienced symptoms that are not commonly mentioned in public discussion of Long COVID [29, 30], and may benefit from further attention. These include but are not limited to: anaphylaxis and new allergies, seizures, suicidality, changes in sensitivity to medication, vision loss, hearing loss, and facial paralysis. Several of these symptoms, as well as the more commonly reported Long COVID symptoms, overlap with symptoms of Mast Cell Activation Syndrome (MCAS), possibly warranting further exploration into the role of mast cells in Long COVID [31].

This work also highlights the wide range of neurologic symptoms experienced by patients with Long COVID. While respiratory and some cardiovascular symptoms have been widely reported, neurological symptoms remain unclear [32]. Prior studies have identified evidence of cognitive dysfunction induced by COVID-19 illness, with few studies in the non-hospitalized population [26, 33]. Memory and cognitive dysfunction, experienced by over 85% of respondents, were the most pervasive and persisting neurologic symptoms in this cohort, equally common across all ages, and with substantial impact on work. Headaches, insomnia, vertigo, neuralgia, neuropsychiatric changes, tremors, sensitivity to noise and light, hallucinations (olfactory and other), tinnitus, and other sensorimotor symptoms were also all common among respondents, and may point to larger neurological issues involving both the central and peripheral nervous system. This area is particularly important to study, as others have found neurological symptoms may be more common in nonhospitalized patients, and that those with neurological symptoms may have impaired seroconversion [33].

The reduced work capacity because of cognitive dysfunction, in addition to other debilitating symptoms, translated into the loss of hours, jobs, and ability to work relative to pre-illness levels. 68.9% of unrecovered respondents reported reduced work hours or not working at all as a direct result of their COVID-19 illness, and on average the unrecovered group felt they were less than 60% returned to their pre-illness baseline. More than half of recovered respondents, however, still reported being unable to work their pre-COVID hours, as well as being on average only 86.5% back to their pre-illness baseline. Also, only 55.3% of recovered respondents had Fatigue Assessment Scores ranked as “no fatigue”. This could suggest that some respondents who reported that they were no longer experiencing symptoms felt that any lingering effects were part of their new baseline, despite not fully recovering health- or work-wise.

The investigated cohort had a relatively high socioeconomic status which may have skewed results - respondents may have been more likely than the average Long COVID patient to have jobs with sufficient sick leave, have enough savings to sustain them through a period of no or low income, and/or have jobs that were able to offer accommodations. The write-in responses revealed that there were respondents who were working full-time or at reduced hours at the time of the survey, but had taken several weeks, if not months, off of work. For those who returned to their job, respondents reported experiencing relapses triggered by the mental exertion and stress of work, often needing to go back on leave. This emphasizes the importance of all patients having adequate time off to recover, being able to qualify for disability benefits if long-term assistance is needed, and receiving accommodations at work including telecommuting, flexible hours, and phased returns. Lower wage earners may find it especially challenging to access these accommodations and benefits, particularly in locations without robust legal protections for disabled workers, yet they are in need of these protections the most to ensure financial stability [34].

Overall, these findings suggest that the morbidity of COVID-19 illness has been greatly underappreciated. Patients experience multisystem symptoms for over 7 months, resulting in significant impact to patients’ lives and livelihoods.

SARS-CoV-2 testing

Our analysis confirms prior studies’ results which show that, with the exception of change to smell and taste, symptoms are not significantly different between those who test positive for SARS-CoV-2 and those who test negative (or have not been tested) but who otherwise show strongly suggestive symptoms [5, 35]. The reason for this is not known, though the sensitivity of diagnostic tests may be different depending on the primer/probe sets [36, 37]. Further, the likelihood of false negatives increases after day 3 of symptom onset, when the false negative rate is 20%, reaching 66% by day 21 [38]. This reinforces the need for early testing in patients with suspicion of SARS-CoV-2 infection given that up to 54% of patients could have an initial RT-PCR false-negative result [39]. The importance of early testing was reflected in this cohort as well: the median number of days between first experiencing symptoms and being tested was 6 days for those who tested positive and 43 days for those who tested negative. Access to adequate diagnostic tests in the early stages of the pandemic was notably limited, which likely contributed to respondents in this cohort being unable to be tested and/or being tested later in their illness [40]. The site of sample collection can also play an important role in testing accuracy. Compared to nasopharyngeal swab sampling, sputum testing resulted in significantly higher rates of SARS-CoV-2 RNA detection than oropharyngeal swab testing [41]. Similarly, viral particles may be detected from stool specimens, while respiratory tract specimens are negative by RT-PCR [42]. Regarding antibody testing, it has been reported that antibody levels decrease with time [43, 44]. In a recent study, 96% of participants had reduced antibodies and 28% no longer tested positive for antibodies at the 2 month follow up [43]. There are evidences suggesting that patients with neurological symptoms but minimal respiratory symptoms may fail to seroconvert [33]. These findings indicate that absent or negative SARS-CoV-2 diagnostic and antibody tests should not be used as an indicator to rule out Long COVID in patients who otherwise have suggestive symptoms. Therefore, we included suspected COVID-19 cases in our analysis, as have several other studies [35,45–47]. Further investigations are needed to understand why some Long COVID patients test positive and others do not despite having similar symptom courses.

Limitations

While the majority of participants did not report receiving a positive SARS-CoV-2 diagnostic or antibody test result, our analysis confirms that this is not a limitation of our study; rather, it is a limitation of the availability and accuracy of SARS-CoV-2 tests.

However, there are several limitations to this study. First, the retrospective nature of the study exposes the possibility of recall bias. Second, as the survey was distributed in online support groups, there exists a sampling bias toward Long COVID patients who joined support groups and were active participants of the groups at the time the survey was published. Additionally, despite eight translations and inclusive outreach efforts, the demographics were strongly skewed towards English speaking (91.9%), white (85.3%), and higher socioeconomic status (see Figure S1). In future studies, more outreach and partnerships with diverse support groups, low-income communities, and communities of color can be established to counter sampling bias. Moreover, the study required respondents to have stable internet and email addresses, which may have excluded participants who lacked access and/or had low digital literacy. Lastly, the effort to complete the survey may have deterred some respondents who experienced cognitive dysfunction, or were no longer ill and did not have enough incentives to participate.

Due to these limitations, we suggest that the results laid forth be considered only in this context, and caution that extrapolation to the entire Long COVID population requires caution.

Implications

Research by the United Kingdom’s Office of National Statistics estimates that 21% of people who were infected with SARS-CoV-2 still experience symptoms at five weeks, a number which includes asymptomatic patients and will likely be higher for symptomatic patients [48]. Given the millions of cases of COVID-19 worldwide, the prevalence of Long COVID is likely to be substantial, and will only increase as the virus continues to spread. This research demonstrates how expansive and debilitating this prolonged illness can be, with profound impacts to people’s livelihoods and ability to care for themselves and their loved ones. This research demonstrates the importance of slowing the spread of COVID-19 through validated public health measures and vaccinations; a robust safety net including sick leave, disability benefits, and workplace protections and flexibilities for patients, as well as adequate family leave for caregivers; and continued research into Long COVID, ME/CFS, POTS, MCAS, and similar illnesses to determine etiologies and find treatment options. Since Long COVID often impacts multiple organ systems and causes extensive disruption to daily functioning, it is important for medical professionals to consider a multidisciplinary approach to treat and care for patients.

Future research

This paper is a description of Long COVID from a patient’s perspective - the symptoms; disease course; and impact on daily life, work, and return to baseline health. Valuable future research would emphasize the pathophysiology of Long COVID, answering questions such as: What are the biological underpinnings of Long COVID and what are potential treatments? How is Long COVID impacting the day-to-day life of people, taking into account different socio-economic backgrounds? What can be done to support those with Long COVID, both medically and through policy, until treatments are identified?

Our future work will focus on investigating several emerging topics in Long COVID: mental health outcomes, diagnostic and antibody testing, symptom clustering, patient perspectives on medical attention, and socioeconomic impact from the illness. We look forward to partnering with other research teams for diagnostic data, and driving further collaborations between patient communities, scientists, and clinicians via patient-centered research. We welcome inquiries for analysis by Long COVID patients, particularly patients and community groups of color.