key: cord-0870777-v76wzj4m authors: Chen, C.; Haupert, S. R.; Zimmermann, L.; Shi, X.; Fritsche, L. G.; Mukherjee, B. title: Global Prevalence of Post-Acute Sequelae of COVID-19 (PASC) or Long COVID: A Meta-Analysis and Systematic Review date: 2021-11-16 journal: nan DOI: 10.1101/2021.11.15.21266377 sha: a6cd445fbca3c26e09eed855676e3056bd887812 doc_id: 870777 cord_uid: v76wzj4m Importance As SARS-CoV-2 pervades worldwide, considerable focus has been placed on the longer lasting health effects of the virus on the human host and on the anticipated healthcare needs. Objective The primary aim of this study is to examine the prevalence of post-acute sequelae of COVID-19 (PASC), commonly known as long COVID, across the world and to assess geographic heterogeneities through a systematic review and meta-analysis. A second aim is to provide prevalence estimates for individual symptoms that have been commonly reported as PASC, based on the existing literature. Data Sources PubMed, Embase, and iSearch for preprints from medRxiv, bioRxiv, SSRN, and others, were searched on July 5, 2021 with verification extending to August 12, 2021. Study Selection Studies written in English that consider PASC (indexed as ailments persisting at least 28 days after diagnosis or recovery for SARS-CoV-2 infection) and that examine corresponding prevalence, risk factors, duration, or associated symptoms were included. A total of 40 studies were included with 9 from North America, 1 from South America, 17 from Europe, 11 from Asia, and 2 from other regions. Data Extraction and Synthesis Data extraction was performed and separately cross-validated on the following data elements: title, journal, authors, date of publication, outcomes, and characteristics related to the study sample and study design. Using a random effects framework for meta-analysis with DerSimonian-Laird pooled inverse-variance weighted estimator, we provide an interval estimate of PASC prevalence, globally, and across regions. This meta-analysis considers variation in PASC prevalence by hospitalization status during the acute phase of infection, duration of symptoms, and specific symptom categories. Main Outcomes and Measures Prevalence of PASC worldwide and stratified by regions. Results Global estimated pooled PASC prevalence derived from the estimates presented in 29 studies was 0.43 (95% confidence interval [CI]: 0.35, 0.63), with a higher pooled PASC prevalence estimate of 0.57 (95% CI: 0.45, 0.68), among those hospitalized during the acute phase of infection. Females were estimated to have higher pooled PASC prevalence than males (0.49 [95% CI: 0.35, 0.63] versus 0.37 [95% CI: 0.24, 0.51], respectively). Regional pooled PASC prevalence estimates in descending order were 0.49 (95% CI: 0.21, 0.42) for Asia, 0.44 (95% CI: 0.30, 0.59) for Europe, and 0.30 (95% CI: 0.32, 0.66) for North America. Global pooled PASC prevalence for 30, 60, 90, and 120 days after index test positive date were estimated to be 0.36 (95% CI: 0.25, 0.48), 0.24 (95% CI: 0.13, 0.39), 0.29 (95% CI: 0.12, 0.57) and 0.51 (95% CI: 0.42, 0.59), respectively. Among commonly reported PASC symptoms, fatigue and dyspnea were reported most frequently, with a prevalence of 0.23 (95% CI: 0.13, 0.38) and 0.13 (95% CI: 0.09, 0.19), respectively. Conclusions and Relevance The findings of this meta-analysis suggest that, worldwide, PASC comprises a significant fraction (0.43 [95% CI: 0.35, 0.63]) of COVID-19 tested positive cases and more than half of hospitalized COVID-19 cases, based on available literature as of August 12, 2021. Geographic differences appear to exist, as lowest to highest PASC prevalence is observed for North America (0.30 [95% CI: 0.32, 0.66]) to Asia (0.49 [95% CI: 0.21, 0.42]). The case-mix across studies, in terms of COVID-19 severity during the acute phase of infection and variation in the clinical definition of PASC, may explain some of these differences. Nonetheless, the health effects of COVID-19 appear to be prolonged and can exert marked stress on the healthcare system, with 237M reported COVID-19 cases worldwide as of October 12, 2021. 5.7, 6.2), 1.5 (95% CI: 1.4, 1.6), and 3.0 (95% CI: 2.7, 3.2) times higher, respectively, in those with a COVID-19 diagnosis as compared to matched controls at a mean follow-up of 140 days. 12 A more recent meta-analysis estimated 80% of those infected with SARS-CoV-2 develop at least one long-term symptom, with the most prevalent symptoms being fatigue, headache, attention disorder, hair loss, and dyspnea. 13 However, as the meta-analysis was conducted in the earlier stage of the pandemic, the review was limited by the inherently smaller sample size of infected individuals underlying the existing studies at that time. Upon this base, further research was forged into the potential factors that see increased PASC prevalence. Time since infection, acute phase severity, geographic region, and select sociodemographic characteristics, such as age and sex, are among the constellation of factors likely to influence PASC prevalence estimates. Although a large proportion of the current evidence focuses on the hospitalized COVID-19 population, a German study found 34.8% of COVID-19 patients, with only a mild acute infection, had PASC at 7 months. 14 To illustrate the geographic heterogeneity seen in PASC prevalence estimates, specific studies from the USA, Italy, and China report prevalence of 28%, 51%, and 76%, respectively. [15] [16] [17] Regarding demographic factors, Sudre et al. found female sex to be associated with developing PASC. 18 Although no existing global reviews (at the time of this report) present age-specific PASC prevalence, Nasserie et al. (2021) offer some evidence that prolonged symptoms are not distinct to older versus younger age groups. 19 That is, although the bulk of PASC exhibiting individuals across the included studies were older (median age near 60 years), younger age groups were also found to comprise a non-negligible number of those with persistent symptoms. 19 Existing research suggests older age to be associated with a moderate increased risk of persisting symptoms (for ten-year increments past age of 40, estimated odds ratio (OR) is 1.10 [95% CI: 1.01-1.19]). 20 Moreover, the existing inequities by race/ethnicity as it pertains to PASC remain largely unexplored, 21 despite the same having been shown for COVID-19. Select comorbidities have been identified . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint as being associated with PASC in the existent literature (e.g., increased risk of PASC among individuals with asthma, 18 though these findings are generally in early stages). Following these collective efforts, we too emphasize that PASC must be well-defined and wellunderstood to enable data to inform clinical decision-making and guidance, and thereby, aid the millions of affected individuals worldwide. At this juncture of being nearly two years into the COVID-19 pandemic, numerous large, high-quality studies on PASC, with substantial follow-up time, have been conducted. Expanding on previous meta-analyses hampered by smaller sample sizes and shorter follow-up times, this systematic review and meta-analysis aims to provide a comprehensive synthesis of information on prevalence and symptoms of PASC todate. Based on previous research, we hypothesize that PASC is common across geographic and demographic groups, with respiratory, neurological, cardiac, and psychological symptoms having the highest prevalence. We close with some avenues for future research considerations, as highlighted by the findings of this review. We employed PICO and PRISMA frameworks to guide our entire research process (eTable 1). 22 The literature databases, PubMed and Embase for published articles, as well as iSearch for preprint articles from bioRxiv, medRxiv, SSRN, Research Square, and preprints.org, were searched on July 5, 2021, and search verification was extended through August 12, 2021. The search aimed to capture papers relating to PASC and that examine prevalence, risk factors, and/or duration, published during the years 2020-2021, and written in English. We adapted some search components from a public resource made available by Yale University Libraries. 23 The full search strategy, including filters for each database, is presented in eMethods 1. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint A two-step approach to screening was used with an initial title/abstract screening, followed by a full-text screening, and an ultimate discussion and re-examination to resolve conflicting marks. Screeners 1 and 2 performed both phases of the screening independently (i.e., were blinded). Rayyan, a web-based application, was used as a tool to help expedite literature screening for systematic reviews. 24 Our inclusion criteria were as follows: (1) human study population with confirmed COVID-19 diagnosis through PCR test, antibody test, or diagnosis, (2) index date of first test/diagnosis, date of hospitalization, discharge date, or date of clinical recovery/negative test, (3) primary outcome must include prevalence, risk factors, duration, or symptoms of PASC, and (4) the follow-up time is at least 28 days after the index date. We excluded case studies, reviews, studies with imaging or molecular/cellular testing as primary results, and studies with only healthcare workers or residents of nursing homes/long-term care facilities. We also excluded studies that did not meet the sample size threshold of 323, pre-calculated herein. The reason for this is to ensure the included studies were adequately powered to achieve a margin of error of at least 0.05 on the provided PASC prevalence estimate. The sample size threshold was calculated with an estimated prevalence of 30% and for a 95% Wald-type confidence interval for binomial proportion; see eMethods 2 for further details. After studies were selected, the following relevant data elements were manually extracted separately by both screeners 1 and 2: article title, authors, date of publication, study purpose, study design, population, setting, country, sample size, method of COVID-19 confirmation, index date, follow-up time, demographic variables (i.e., age and sex), and outcomes examined. In the . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint instance of multiple study versions with the same underlying population, we used the most recently published article. The primary outcome was the prevalence of PASC and symptoms at least 28 days after the index date. We defined PASC as having any symptoms, or at least one new or persisting symptom during the follow-up time. The follow-up time of COVID-19 patients across studies was divided into the following four groups: PASC persisting at 28-30 days (labeled as 30 days), 60 days, 90 days, and 120 days after the index date. We combined similar symptoms into a broader concept. For example, we joined together dyspnea, shortness of breath, and problem of breathing reported in different studies into a broader symptom concept of dyspnea (see eTable 2). Studies were classified into the following three groups based on the study population of PASC: (1) studies with non-hospitalized COVID-19 positive individuals, (2) studies with hospitalized COVID-19 positive individuals, (3) studies with all COVID-19 positive individuals (i.e., a case-mix with hospitalized and non-hospitalized individuals). In addition to prevalence, we were also interested in the risk factors for PASC as secondary outcomes. Meta-analysis with random effects and generic inverse variance weighting was performed to estimate the prevalence of PASC and symptoms, for outcomes reported in at least five studies. Of further note is that upon examining the distribution of PASC prevalence, we apply a logit transformation to the proportion. The confidence interval was calculated incorporating betweenstudy variance obtained by the DerSimonian-Laird (DL) estimator (eMethods 3). Heterogeneity among studies was reflected by the 2 statistic, where 2 between 75% and 100% indicates considerable heterogeneity. We further stratified our analysis by (1) study population type (hospitalized versus mixed hospitalized and non-hospitalized), (2) sex (female versus male), (3) . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint follow-up time, (4) region (Asia, Europe, and USA). Another stratified analysis is presented in the supplement (see eFigure 1) wherein pooled PASC prevalence is estimated (A) among studies defining PASC to be persisting symptoms (i.e., extended beyond a pre-specified number of days) and (B) among studies defining PASC to be at least one symptom or not recovered from COVID-19. All analyses were conducted in R (version 4.0.2) using packages meta 25,26 and metafor. 27 For critical appraisal, we used a checklist-based tool from Joanna Briggs Institute (JBI), corresponding to prevalence studies and hence, enabling assessment of risk of bias among the included study designs. 28 Assessment of publication bias was carried out visually by generating funnel plot and formally by conducting Egger's and Begg's tests for funnel plot asymmetry (for further details, see eMethods 4 and eFigure 2). In our main literature search, we identified 4,438 unique citations of which 270 had titles or abstracts that passed our criteria for a full-text assessment. After the full-text screen, we deemed 40 studies eligible for a qualitative synthesis, of which we further meta-analyzed reported measures from 34 with compatible outcomes. See the PRISMA flow diagram ( Figure 1 ) and eTable 3 for details concerning study inclusion/exclusion criteria. In efforts to further verify the search results, we performed a second literature search one month after the first screen, although no additional eligible studies were identified (eMethods 1 and eFigure 3). . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint While only studies with at least 4 weeks follow-up were selected, several studies had data on substantially longer follow-up times: for at least 8 weeks (36 studies), 12 weeks (33 studies), and 6 months (17 studies). Figure 1 lists additional study characteristics. Among the 34 included studies in the quantitative synthesis, we meta-analyzed the 29 studies reporting an overall prevalence of PASC. Pooled global PASC prevalence was estimated to be 0.43 (95% CI: 0.35, 0.63) ( Table 2) . Substantial heterogeneity was observed among the included studies ( 2 =100%, P < 0.001). Estimates ranged widely from 0.09 to 0.81 which may in part be driven by differences in terms of sex, region, COVID-19 study population, and followup time. For example, the studies that included only hospitalized cases tended to show higher PASC prevalence than non-hospitalized or the mix of hospitalized and non-hospitalized patients ( Figure 2 ). To better understand the interplay of these factors with PASC prevalence estimates, we performed additional stratified meta-analyses ( Next, when focusing on sex, we estimated a pooled PASC prevalence in females of 0.49 (95% CI: 0.35, 0.63), which was higher than that in males of 0.37 (95% CI: 0.24, 0.51). Considering the same studies underly both strata, this imbalance was unlikely attributable to differences in the contributing studies (eFigure 4A). Examining region-specific prevalences, pooled estimated prevalence of PASC was lower in the USA at 0.30 (95% CI: 0.21, 0.42) than in Europe at 0.43 (95% CI: 0.29, 0.58), while the highest estimated prevalence was in Asia at 0.49 (95% CI: 0.32, 0.66). Considerable within-region variation was observed among the included studies in that the corresponding ranges of prevalence of PASC were generally wide, with Europe exhibiting the largest range of 0.09 -0.81. Overall, we did not identify any patterns with respect to particular countries that could explain the heterogeneity within each of the meta-analyzed regions (P < 0.001; eFigure 4C). Some studies that measured PASC prevalence at multiple time points experienced a similar phenomenon. 14 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint Significant levels of heterogeneity being present within each stratified meta-analysis corroborates that no single factor alone may account for the variation in PASC prevalence, but that rather a combination of factors should be considered. Concerning a perceived ordering of influence, the factors that seem to have the largest bearing on the PASC prevalence (as ordered from highest to lowest) were study population, region, and follow-up time. Furthermore, inconsistent PASC definition is a source of heterogeneity. As detailed in eFigure 1, studies measuring PASC as having at least one persistent symptom had lower heterogeneity (as measured by a chi-square test statistic) compared to those measuring PASC as at least one symptom (with symptoms not necessarily starting during the acute phase). Noting that the prevalence of each symptom varied, effect size of PASC prevalence estimates may differ in part due to the underlying symptoms assessed therein. An additional meta-analysis of studies with at least 120 days follow-up stratified by COVID-19 population resulted in similar observed heterogeneity (eFigure 5). Ultimately, these findings suggest that such variation may be indelible, as key considerations, such as the definition of PASC itself, as well as other clinical and methodological subcomponents, remain largely in flux. 62 Considering a unified definition of PASC remains under investigation (as discussed in the Introduction section), it was important to understand the prevalence of specific symptoms after COVID-19. In total, we assessed 23 symptoms reported across 30 studies (Table 2, Figure 3 ). The five most prevalent symptoms were the following, with corresponding estimated pooled symptom-specific prevalence: fatigue at 0.23 (95% CI: 0.13, 0.38), dyspnea at 0.13 (95% CI: 0.09, 0.19), insomnia at 0.13 (95% CI: 0.06, 0.28), joint pain at 0.13 (95% CI: 0.05, 0.29), and memory problems at 0.13 (95% CI: 0.10, 0.18). Forest plots for symptom-specific prevalence estimates are presented in eFigure 6. We note that a study by Orrū et al. from Italy tended to fall toward the higher end of the observed range for several symptom categories, and as such . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint Although all included studies were screened for reported PASC risk factors, sex and preexisting asthma were the only risk factors that were estimated in multiple studies and thus metaanalyzed. Female sex and pre-existing asthma had higher odds of having PASC with pooled estimated odds ratios (OR) of 1.57 (95% CI: 1.09, 2.26) and 2.15 (95% CI: 1.14, 4.05), respectively. Both meta-analyzed ORs were based on less than 5 studies and should thus be interpreted with caution. Among the studies that were not meta-analyzed, several found that individuals with more severe COVID-19 during the acute phase had higher risk of developing PASC. 37,45,50,52 Additionally, two studies found older age to be associated with PASC. 20 Six studies were not included in the meta-analysis since they did not report a composite binary endpoint as prevalence (Figure 1) . Three studies used incidence rate or incidence density to measure PASC. Chevinsky et al. reported a 7% incidence rate of at least one of the five most common new conditions during days 31 to 120 for inpatients and a 7.7% incidence rate for at least one of 10 new conditions. 57 A UK study found breathlessness (85 and 536 events per 100,000 person-years in non-hospitalized and hospitalized patients) and joint pain (168 and 295 events per 100,000 person-years in non-hospitalized and hospitalized patients) to be the most common sequelae at 2 months. 52 Another UK study found the rates of respiratory disease and . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint 12 The other three studies investigated PASC with a focus on the psychiatric and neurological illness. Damiano et al. focused on psychiatric and cognitive sequela and reported a prevalence of 0.08 (95% CI: 0.06, 0.11), 0.14 (95% CI: 0.11, 0.18), and 0.16 (95% CI: 0.12, 19) for depression, generalized anxiety disorder, and mixed anxietydepression. 56 Another study by Taquet et al. also concentrated on psychiatric disorder and measured incidence and hazard ratio of psychiatric disorder, dementia and insomnia with 90 days follow up. 61 The estimated probability of having new psychiatric illness 90 days after COVID-19 diagnosis was 5.8% (95% CI: 5.2, 6.4) Huang et al.'s study from China showed psychosocial problems (57.7%), worse depression (35%), and worse dyspnea (32.6%) to be among the most common complaints 4-6 months after discharge. 29 Two articles described the duration of PASC and persistent symptoms. According to Sudre et al., the median duration of PASC with persisting symptoms was 41 days. For persisting symptoms that occurred at least 28 days after COVID-19 diagnosis, the median duration of persisting fatigue, headache, dyspnea, and myalgia was 33 days, 22 days, 24 days, and 7 days, respectively. 18 However, the median duration of persisting symptoms was longer in another India study. 35 A summary table of duration was reported in eTable 5. We screened nearly 4.5 thousand articles and synthesized information from 40 large studies including almost one million individuals worldwide. The empirical findings suggest a global PASC prevalence of approximately 43%. Based on a WHO estimate of 237 million worldwide COVID-19 infections, this global pooled PASC estimate indicates that around 100 million individuals currently experience or have previously experienced long-term health-related . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint consequences of COVID-19. Individuals who were hospitalized during acute COVID-19 infection had higher PASC prevalence at 57%. Female adults had both higher prevalence and risk of having PASC than male adults (49% vs 37%). The prevalence of PASC in Asia, Europe, and USA are approximately 49%, 43%, and 30%, respectively. Next, we contextualize our results among findings from other PASC-related reviews. Our global PASC estimate of 43% is considerably lower than the 80% figure provided by Lopez-Leon et al. 63 Their most prevalent sequela was fatigue at 58% which is concordant with fatigue being the most prevalent sequela at 23% in this study. In general, empirical symptom-specific prevalence estimates are lower in this study, although multiple estimates (e.g., for insomnia, memory problems, anxiety, depression) generally reconcile with the Lopez-Leon et al. review. 13 Similarly, when comparing to the Iqbal et al. 64 meta-analyzed PASC-related symptom prevalence findings, the estimates herein are lower. A potential reasoning for this is the sample size threshold that we employed may have led to select studies being excluded that were conducted in early 2020 with smaller samples and focused mainly on sicker patients. Additional notable studies have been published after the date of this systematic search (August 12, 2021), and as such are not captured in the empirical estimates presented herein. As examples, another study by Taquet et al., using the TriNetX Analytics EHR network, estimated 36.55% of COVID-19 patients to have at least one PASC-related symptom 3-6 months after diagnosis. 65 also provided an update on their cohort from Jin Yin-tan Hospital in Wuhan. 66 Their 6-month study (which is included in this review) estimated 6-month PASC to be 76%. Their 12-month update found that, among patients who attended both the 6-month and 12-month follow-up, PASC prevalence decreased from 68% at 6 months to 49% at 12 months. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Our meta-analysis showed that female sex and pre-existing asthma correspond with higher proportions of PASC development. Outside of meta-analysis, we also found age, acute phase symptoms and severity, hypothyroidism, obesity, hypertension, and other pre-existing conditions to be risk factors for PASC. Protective factors for PASC may also exist, as a recent study suggested vaccines may offer protection. 68 However, a large hospital-based study suggests the opposite. 69 As such, the interplay between COVID-19 vaccines and PASC is at-large yet to be determined. Multiple other risk factors for PASC have been detected, and, although encompassed among select included studies, such factors were not meta-analyzed because they did not reach the threshold of at least 5 studies. Increased number of acute-phase symptoms is associated with PASC; however, one study reported 32% of individuals with PASC were asymptomatic during the acute phase in a non-hospitalized population. 55 Similarly, few studies examined the duration of PASC. Future research needs to further explore risk factors and duration for PASC, as these are generally critical components for clinicians in screening patients for increased risk of developing PASC, and in devising an appropriate treatment protocol accordingly. This leads to the several limitations of this systematic review and metaanalysis. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. positive cohort. 72 In other words, patients without access to testing, patients without strong health-seeking behavior, and asymptomatic individuals are not blanketly reflected in the empirical findings. Additionally, included studies conducted in early 2020 may tend to be older and higher risk individuals, as testing among these groups was prioritized at that time. Fifth, our sample size criteria may have curtailed inclusion of early-pandemic studies, as sample sizes were generally smaller at that time, and thus favored studies examining acute-phase manifestations over studies focusing on PASC. Lastly, while our review included studies across 17+ countries, data from multiple regions are largely absent (notably Africa and Australia). Existing inequities in healthcare access may hamper underserved populations being adequately reflected herein. Moreover, we emphasize that stratifying PASC by race-ethnicity is a noteworthy gap in the literature. With respect to the age composition of the included articles, few children were included in the underlying sample. Future investigators may seek to further examine differences in PASC prevalence among such demographic subgroups. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint Findings from this study provide insight into the empirical estimates of prevalence, symptoms, risk factors, and duration of PASC, with an examination of differences by several factors including geography. We recommend continued attention be focused on identifying patients atrisk of developing PASC and on quantifying duration of PASC to aid in the clinical advancements globally for alleviating the long-lasting health effects of COVID-19. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Notes: Prevalence estimates and 95% CIs are provided for each study with a relevant measure, and for the meta-analysis of all such studies. For individual studies, the horizontal line represents the estimate, whiskers represent the confidence interval, the size of the box represents the weight assigned to the study, and the color shading reflects the hospitalization status of the study population, as noted in the legend. For the pooled estimate, the width of the diamond represents the confidence interval. Meta-analyzed prevalence and 95% CIs are calculated using random-effects models with inverse variance weighting as described in the methods. Measures of heterogeneity of prevalence estimates are provided. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint Figure 3 . Forest plot for PASC prevalence by hospitalization status, region, follow-up time, and sex, as well as symptom-specific prevalence. Notes: Pooled estimates and 95% CIs calculated from random-effect models with inverse variance weighting as described in methods. Pooled estimates with confidence intervals are provided on the left, and visualization of the intervals on the right. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this this version posted November 16, 2021. ; eMethods 1. Systematic review procedure To perform the systematic review presented herein, we collected both publications and preprints systematically that concern prevalence, risk factors, and/or duration of PASC in any country worldwide. In identifying relevant articles, we searched the following three databases: PubMed, Embase, and iSearch for preprints (encompassing bioRxiv, medRxiv, preprints.org, Research Square, SSRN). The search was conducted on July 5, 2021. Hence, the resulting captured studies reflect those available from January 1, 2020 to July 5, 2021. A second search was conducted on August 12, 2021, to ensure we would not fail to capture recent, important studies. In this search, we used the same search terms and filters as our July 5 search and restricted our attention to studies published in well-established medical journals (i.e., JAMA, Lancet, NEJM, Nature, BMJ, and PloS). Details of study inclusion from the August 12 search can be found in eFigure 3. Upon securing citations from the search engines into Mendeley 1 , the reference manager, the resulting body of citations were deduplicated, and subsequently imported into the online tool Rayyan 2 for screening. Search blocks and filters for PubMed, Embase, and iSearch are detailed below. 1 AND 2 AND 3 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. ("long COVID" OR "long covid-19" OR "long-covid" OR "long-covid-19" OR "long haul" OR "long hauler" OR "long haulers" OR "long-haul" OR "long-hauler" OR "long-haulers" OR "chronic COVID" OR "chronic covid-19" OR "post-acute COVID" OR "post-acute covid-19" OR "post acute COVID" OR "post acute covid-19" OR "persistent COVID" OR "persistent covid-19" OR "post-COVID" OR "post-covid-19" OR "post COVID" OR "post covid-19" OR "sequela" OR "sequelae" OR "long-term" OR "long term" OR "covid syndrome" OR "covid-19 syndrome" OR "persistent symptom" OR "persistent symptoms" OR "PASC" OR "PACS" OR "PPCS" OR "post-acute" OR "post acute") 2. ("prevalent" OR "prevalence" OR "occurrence" OR "occurrences" OR "duration" OR "durations" OR "length" OR "lengths" OR "risk factor" OR "risk factors" OR ("risk" AND "factor") OR ("risk" AND "factors") OR "predict" OR "prediction" OR "predictions" OR "predicting" OR "predictive" OR "predictor" OR "predictors" OR "symptom" OR "symptoms" OR "define" OR "defining" OR "definition" OR "definitions" OR "follow up" OR "follow-up" OR "followed up") 1 AND 2 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint eMethods 3. Meta-analysis framework We used random effects model with logit transformation and the DerSimonian-Laird (DL) estimator for 2 . Thus, the pooled estimated prevalence of PASC ( � ) is calculated as: where i represents the i-th study, = 1,2, … , ; � is the logit transformed prevalence in study i so that � = ( ) = ln � 1− � , and � is the estimated weight for study i using the inverse-variance method. The total variance of study i is the sum of within-study variability denoted by ̂, and between-study variability calculated by the DL estimator ̂2. Therefore, � equals to the inverse of the variance, i.e., The DL estimator is calculated, as is given below: CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint eMethods 4. Risk of bias assessment across included articles Using the Joanna Briggs Institute (JBI) tool 3 , studies were evaluated across 9 common sources of bias in observational studies and given a score out of 9 to reflect how well each study handled these biases. The vast majority (31/40) included studies scored 6/9 or 7/9. Only one study received a perfect score, and our lowest score was 4/9. Supplementary file Supplementary_RiskofBias.xlsx contains full results from the risk of bias assessment across the included studies. Next, we discuss the types of bias that can be introduced in cross-sectional and cohort studies, which make up the majority of the study designs in this review. First, several studies gave COVID-19 cases optional access to a post-COVID clinic or follow-up, which may have resulted in self-selection of sicker individuals. Reporting of symptoms was at times documented through self-report, which has been evidenced to differ from doctor-diagnosed symptoms. 2 Among studies whose target population was a mixture of hospitalized and non-hospitalized individuals, proportions of hospitalized to non-hospitalized patients varied, possibly biasing the results for this group. Misclassification bias may also be of concern. Patients admitted to the ICU are known to sometimes experience so-called Post-Intensive Care Syndrome (PICS), whose symptomatology is somewhat similar to that of PASC . 3 Further, some studies suggest that as many as 85% of PASC patients experience symptom resolution, only relapse at a later date, which could obscure the true proportion of individuals experiencing chronic PASC. 4 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint Selection process 8 Specify the methods used to decide whether a study met the inclusion criteria of the review, including how many reviewers screened each record and each report retrieved, whether they worked independently, and if applicable, details of automation tools used in the process. Data collection process 9 Specify the methods used to collect data from reports, including how many reviewers collected data from each report, whether they worked independently, any processes for obtaining or confirming data from study investigators, and if applicable, details of automation tools used in the process. 9-10, eMethods 1 Data items 10a List and define all outcomes for which data were sought. Specify whether all results that were compatible with each outcome domain in 9-10, eTable 2 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint 10b List and define all other variables for which data were sought (e.g., participant and intervention characteristics, funding sources). Describe any assumptions made about any missing or unclear information. Study risk of bias assessment 11 Specify the methods used to assess risk of bias in the included studies, including details of the tool(s) used, how many reviewers assessed each study and whether they worked independently, and if applicable, details of automation tools used in the process. Effect measures 12 Specify for each outcome the effect measure(s) (e.g., risk ratio, mean difference) used in the synthesis or presentation of results. 10 Synthesis methods 13a Describe the processes used to decide which studies were eligible for each synthesis (e.g. tabulating the study intervention characteristics and comparing against the planned groups for each synthesis (item #5)). 13b Describe any methods required to prepare the data for presentation or synthesis, such as handling of missing summary statistics, or data conversions. 10, eTable 2 13c Describe any methods used to tabulate or visually display results of individual studies and syntheses. 13d Describe any methods used to synthesize results and provide a rationale for the choice(s). If meta-analysis was performed, describe the model(s), method(s) to identify the presence and extent of statistical heterogeneity, and software package(s) used. 10-11, eMethods 3 13e Describe any methods used to explore possible causes of heterogeneity among study results (e.g., subgroup analysis, metaregression). 10-11, eMethods 3 13f Describe any sensitivity analyses conducted to assess robustness of the synthesized results. Reporting bias assessment 14 Describe any methods used to assess risk of bias due to missing results in a synthesis (arising from reporting biases). Certainty assessment 15 Describe any methods used to assess certainty (or confidence) in the body of evidence for an outcome. 16a Describe the results of the search and selection process, from the number of records identified in the search to the number of studies included in the review, ideally using a flow diagram. 11, Figure 1 16b Cite studies that might appear to meet the inclusion criteria, but which were excluded, and explain why they were excluded. Study characteristics 17 Cite each included study and present its characteristics. 11-12, Figure 1 , Table 1 Risk of bias in studies 18 Present assessments of risk of bias for each included study. eMethods 4 19 For all outcomes, present, for each study: (a) summary statistics for each group (where appropriate) and (b) an effect estimate and its precision (e.g. confidence/credible interval), ideally using structured . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint 20a For each synthesis, briefly summarise the characteristics and risk of bias among contributing studies. 20b Present results of all statistical syntheses conducted. If meta-analysis was done, present for each the summary estimate and its precision (e.g. confidence/credible interval) and measures of statistical heterogeneity. If comparing groups, describe the direction of the effect. 12-16, Figure 2 -3, Table 2 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint Confusion, lack of concentration, attention disorders, concentration, concentration problems, confusion/lack of concentration, impairment (brain fog, loss of concentration), inability to concentrate, problems concentrating and thinking, brain fog Appetite/eating disorder appetite, anorexia, decreased appetite, decreased or lack of appetite, eating disorders, loss of appetite Tachycardia tachycardia, tachycardia-palpitations, palpitations Symptoms names were inconsistent across studies. Therefore, we grouped symptoms into groups according to similarity as detailed in eTable 2. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint eTable 4 presents a list of non-meta-analyzed risk factors and their associated odds ratios and confidence intervals. Only factors with a significant positive association with PASC were included (i.e., OR > 1 and CI does not cross 1). If both odds ratios and adjusted odds ratios were provided for a risk factor, we used the adjusted odds ratio. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; eTable 5. Summary of duration of PASC and symptoms Sudre et al. 11 Naik et al. 6 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint eFigure 5. Meta-analysis of studies with 120 follow-up days stratified by acute-phase hospitalization status of study population a) Mixed of hospitalized and non-hospitalized individuals b) Hospitalized individuals . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint eFigure 6 Cont'd eFigure 6. Supplementary meta-analysis of PASC symptom-specific prevalence . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint eFigure 6 Cont'd . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint eFigure 6 Cont'd . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint eFigure 6 Cont'd . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint eFigure 6 Cont'd eFigure 6 reports the forest plot of pooled estimated PASC symptoms from the meta-analysis. The legend is consistent across all subfigures where blue, orange, red represents studies with non-hospitalized population, mixed of hospitalized and non-hospitalized population, and hospitalized population respectively. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint The copyright holder for this this version posted November 16, 2021. ; https://doi.org/10.1101/2021.11.15.21266377 doi: medRxiv preprint CDC. Coronavirus Disease 2019 (COVID-19) -Symptoms Systemic and organ-specific immune-related manifestations of COVID-19 COVID-19) Dashboard Disparities in COVID-19 Testing and Positivity in New York City SARS-CoV-2 infection fatality rates in India: systematic review, meta-analysis and modelbased estimation | medRxiv Behavioral Dynamics of COVID-19: Estimating Under-Reporting, Multiple Waves, and Adherence Fatigue Across 92 Nations. Social Science Research Network CoV-2 infection: A systematic review and meta-analysis | PNAS Post-COVID syndrome in non-hospitalised patients with COVID-19: a longitudinal prospective cohort study Characterizing Post-Acute Sequelae of SARS-CoV-2 Infection across Claims and Electronic Health Record Databases Surviving COVID-19 in Bergamo province: a postacute outpatient re-evaluation 6-month consequences of COVID-19 in patients discharged from hospital: a cohort study Attributes and predictors of long COVID Assessment of the Frequency and Variety of Persistent Symptoms Among Patients With COVID-19: A Systematic Review Persistence of symptoms up to 10 months following acute COVID-19 illness The Role of Primary Care The PRISMA 2020 statement: an updated guideline for reporting systematic reviews Rayyan-a web and mobile app for systematic reviews General Package for Meta-Analysis.; 2021 How to Conduct a Meta-Analysis of Proportions in R: A COVID Symptoms, Symptom Clusters, and Predictors for Becoming a Long-Hauler: Looking for Clarity in the Haze of the Pandemic Psychiatric and Cognitive Morbidity Following Moderate and Severe SARS-CoV-2 Infection: A Cohort Study Late conditions diagnosed 1-4 months following an initial COVID-19 encounter: a matched cohort study using inpatient and outpatient administrative data -United States Longitudinal Surveillance of Post-Acute Sequelae of SARS-CoV-2 (PASC) Among Long Beach City Residents Symptom Cluster Analysis of long COVID-19 in Patients Discharged from the Temporary COVID-19 Hospital in Mexico City: A Longitudinal Study Characterising the long-term clinical outcomes of 1190 hospitalised patients with COVID-19 in New York City: a retrospective case series 6-month neurological and psychiatric outcomes in 236 379 survivors of COVID-19: a retrospective cohort study using electronic health records Measuring inconsistency in metaanalyses More than 50 Long-Term Effects of COVID-19: A Systematic Review and Meta-Analysis. Social Science Research Network Characteristics and predictors of acute and chronic post-COVID syndrome: A systematic review and metaanalysis Incidence, cooccurrence, and evolution of long-COVID features: A 6-month retrospective cohort study of 273,618 survivors of COVID-19 Risk factors and disease profile of post-vaccination SARS-CoV-2 infection in UK users of the COVID Symptom Study app: a prospective, community-based, nested, case-control study Six-month sequelae of post-vaccination SARS-CoV-2 infection: a retrospective cohort study of 10,024 breakthrough infections | medRxiv Publication bias: What are the challenges and can they be overcome? Post-acute effects of SARS-CoV-2 infection in individuals not requiring hospital admission: a Danish population-based cohort study coronavirinae'/exp OR 'coronavirus disease 2019'/exp OR covid:ti,ab,de,tn 2. 'long covid-19':ti,ab,de,tn OR 'long covid':ti,ab,de,tn OR 'long covid 19':ti,ab,de,tn OR 'long haul':ti,ab,de,tn OR 'long hauler':ti,ab,de,tn OR 'long haulers':ti,ab,de,tn OR 'chronic covid':ti,ab,de,tn OR 'chronic covid-19':ti,ab,de,tn OR 'post-acute covid':ti,ab,de,tn OR 'postacute covid-19':ti,ab,de,tn OR 'post acute covid':ti,ab,de,tn OR 'post acute covid-19':ti,ab,de,tn OR 'persistent covid':ti,ab,de,tn OR 'persistent covid-19':ti,ab,de,tn OR 'post covid 19':ti,ab,de,tn OR 'post covid':ti,ab,de,tn OR 'post covid-19':ti,ab,de,tn OR sequela:ti,ab,de,tn OR sequelae:ti,ab,de,tn OR 'long term':ti,ab,de,tn OR 'covid syndrome':ti,ab,de,tn OR 'covid-19 syndrome':ti,ab,de,tn OR 'persistent symptom':ti,ab,de,tn OR 'persistent symptoms Assessing the Amount of Heterogeneity in Random-Effects Meta-Analysis The gap between self-reported and objective measures of disease status in India Post-Intensive care syndrome and COVID-19: crisis after a crisis? Heart Lung Characterizing Long COVID in an International Cohort: 7 Months of Symptoms and Their Impact Post-COVID syndrome in non-hospitalised patients with COVID-19: a longitudinal prospective cohort study Post COVID-19 Sequelae: A Prospective Observational Study from Northern India Burden of Post-COVID-19 Syndrome and Implications for Healthcare Service Planning: A Population-based Cohort Study Persistence of symptoms up to 10 months following acute COVID-19 illness Symptom Cluster Analysis of long COVID-19 in Patients Discharged from the Temporary COVID-19 Hospital in Mexico City: A Longitudinal Study Longitudinal Surveillance of Post-Acute Sequelae of SARS-CoV-2 (PASC) Among Long Beach City Residents Attributes and predictors of long COVID Post-COVID-19 syndrome in outpatients: a cohort study The authors thank the librarians from the University of Michigan Taubman Health Sciences Library, and in particular, Judith E. Smith, for their guidance on constructing the search strategy for this systematic review.