The long-term consequences of COVID-19, often referred to as Long COVID or post-COVID syndrome, remain a significant area of concern. Characterized by a wide range of persistent symptoms lasting weeks or months, the syndrome's prevalence and duration are still not well understood, hindering the development of effective treatments and management strategies. This uncertainty stems from the heterogeneity of study designs and definitions used to characterize Long COVID. While studies have examined hospitalized patients, there's a need for community-based data to provide a broader understanding of the prevalence and risk factors associated with persistent symptoms. This study addresses this gap by leveraging data from the Real-Time Assessment of Community Transmission-2 (REACT-2) study conducted in England between September 2020 and May 2021, focusing on individuals experiencing symptoms lasting 12 weeks or more after a suspected or confirmed COVID-19 infection. The study aims to estimate the prevalence of persistent symptoms, investigate their co-occurrence, and assess risk factors for their persistence.

Existing literature on Long COVID exhibits substantial variability in prevalence and persistence estimates, largely attributed to differences in study designs and definitions of the syndrome. Some studies have suggested that Long COVID may represent a collection of distinct conditions, including post-viral symptoms and long-term disruptions to healthcare. The severity of initial COVID-19 infection has been linked to the likelihood of developing Long COVID, with hospitalized patients showing a higher risk. Other factors like older age and male sex have also been implicated. However, many studies have focused on hospitalized patients, necessitating research on broader community samples to capture a more representative picture of Long COVID's impact.

This study utilizes data from rounds 3-6 of the REACT-2 study, a large-scale representative community survey of adults in England. Rounds 3-5 (September 2020 - February 2021) included 508,707 participants, while round 6 (May 2021) comprised 97,727 participants. Participants reported on demographics, lifestyle factors, pre-existing conditions, and symptoms experienced following a suspected or confirmed COVID-19 infection. The primary outcome was the prevalence of one or more symptoms lasting 12 weeks or more after symptom onset. Data analysis involved weighted prevalence estimations to account for sampling design and response rates. Logistic regression models were used to assess the independent effects of sociodemographic and lifestyle factors on the risk of persistent symptoms. Generalized additive models (GAMs) explored the relationship between age, sex, and symptom persistence. Unsupervised learning techniques, specifically clustering analysis using the CLARA algorithm, were employed to identify symptom clusters among participants with persistent symptoms. Sensitivity analyses were conducted using different clustering methods and subsets of the data to validate the findings.

At 12 weeks post-symptom onset in rounds 3-5, 37.7% of those with COVID-19 reported at least one persistent symptom, decreasing to 21.6% in round 6. This translates to a weighted population prevalence of 5.8% for one or more persistent symptoms and 2.2% for three or more in rounds 3-5, and 3.1% and 1.6% respectively in round 6. The most prevalent persistent symptom was tiredness (16.8% in rounds 3-5, 8.0% in round 6). Risk factor analysis (rounds 3-5) revealed significant associations between persistent symptoms and female sex, increasing age, obesity, smoking, COVID-19 hospitalization, deprivation, low household income, and healthcare/care home work. Asian ethnicity showed an association with lower risk. Clustering analysis identified two main symptom clusters: one characterized by tiredness and related symptoms, and another dominated by respiratory symptoms. The replication analysis in round 6 showed similar results, although the association of some risk factors varied slightly due to reduced statistical power. Sensitivity analyses using different clustering methods and subsets of the data supported the main findings regarding the prevalence and clustering of symptoms.

This large-scale community-based study provides valuable insights into the prevalence and predictors of persistent COVID-19 symptoms in England. The findings highlight the significant burden of Long COVID in the population, with a substantial proportion of individuals experiencing persistent symptoms 12 weeks or more after infection. The identified risk factors align with previous research, emphasizing the role of biological, socioeconomic, and occupational factors in the development of Long COVID. The identification of distinct symptom clusters suggests potential heterogeneity in the underlying pathophysiology of Long COVID. These findings underscore the importance of developing targeted interventions and support services for affected individuals, particularly those from economically disadvantaged backgrounds or in deprived areas. Further research is needed to fully understand the underlying mechanisms of these symptom clusters and to develop effective long-term management strategies.

This study demonstrates a substantial prevalence of persistent COVID-19 symptoms in the English community, with significant disparities across demographic and socioeconomic groups. The identification of distinct symptom clusters highlights the need for individualized approaches to treatment and management. Future research should focus on longitudinal studies to track symptom trajectories and the long-term health outcomes of individuals with Long COVID, as well as further investigating the underlying mechanisms and effective interventions for these distinct symptom clusters.

The study's retrospective design and reliance on self-reported data introduce potential recall bias. The cross-sectional nature of the study limits inferences about causality. The specific list of symptoms may not capture the full spectrum of Long COVID manifestations. While the large sample size strengthens the study, variations in response rates and potential selection bias should be considered when interpreting the findings. Finally, the study's focus on England limits the generalizability of the results to other populations and healthcare systems.