ABSTRACT
Importance: Persistent symptoms are reported in patients who survive the initial stage of COVID-19, often referred to as ''long COVID'' or ''post-acute sequelae of SARS-CoV-2 infection'' (PASC); however, evidence on incidence is still lacking, and symptoms relevant to pain are yet to be assessed. Objective: To determine long-term symptoms in COVID-19 survivors after infection. Data Sources: A literature search was performed using the electronic databases PubMed, EMBASE, Scopus, and CHINAL and preprint servers MedR{chi}iv and BioR{chi}iv through January 15, 2021. Study Selection: Eligible studies were those reporting patients with a confirmed diagnosis of SARS-CoV-2 and who showed any symptoms persisting beyond the acute phase. Data Extraction and Synthesis: Incidence rate of symptoms were pooled using inverse variance methods with a DerSimonian-Laird random-effects model. Main Outcomes and Measures: The primary outcome was pain-related symptoms such as headache or myalgia. Secondary outcomes were symptoms relevant to pain (depression or muscle weakness) and symptoms frequently reported (anosmia and dyspnea). Heterogeneity among studies and publication bias for each symptom were estimated. The source of heterogeneity was explored using meta-regression, with follow-up period, age and sex as covariates. Results: In total, 35 studies including 18,711 patients were eligible. Eight pain-related symptoms and 26 other symptoms were identified. The highest pooled incidence among pain-related symptoms was chest pain (17%, 95% CI, 12%-25%), followed by headache (16%, 95% CI, 9%-27%), arthralgia (13%, 95% CI, 7%-24%), neuralgia (12%, 95% CI, 3%-38%) and abdominal pain (11%, 95% CI, 7%-16%). The highest pooled incidence among other symptoms was fatigue (45%, 95% CI, 32%-59%), followed by insomnia (26%, 95% CI, 9%-57%), dyspnea (25%, 95% CI, 15%-38%), weakness (25%, 95% CI, 8%-56%) and anosmia (19%, 95% CI, 13%-27%). Substantial heterogeneity was identified (I2, 50-100%). Meta-regression analyses partially accounted for the source of heterogeneity, and yet, 53% of the symptoms remained unexplained. Conclusions and Relevance: The current meta-analysis may provide a complete picture of incidence in PASC. It remains unclear, however, whether post-COVID symptoms progress or regress over time or to what extent PASC are associated with age or sex.