ABSTRACT
Objective To measure the association between self-reported signs and symptoms and SARS-CoV-2 seropositivity. Design Cross sectional study of three key worker groups. Setting Six acute NHS hospitals and two Police and Fire and Rescue sites in England. Participants Individuals were recruited from three streams: (A) Police and Fire and Rescue services (n=1147), (B) healthcare workers (n=1546) and (C) healthcare workers with previously positive virus detection (n=154). Main outcome measures Detection of anti-SARS-CoV-2 antibodies in plasma. Results 943 of the 2847 participants (33%) reported belief they had had COVID-19, having experienced compatible symptoms (including 152 from Stream C). Among individuals reporting COVID-19 compatible symptoms, 466 (49%) were seronegative on both Nucleoprotein (Roche) and Spike-protein (EUROIMMUN) antibody assays. However, among the 268 individuals with prior positive SARS-CoV-2 tests, of whom 96% reported symptoms with onset a median of 63 days (IQR 52 to 75 days) prior to venesection, Roche and EUROIMMUN assays had 96.6% (95% CI 93.7% to 98.2%) and 93.3% (95% CI 89.6% to 95.7%) sensitivity respectively. Symptomatic but seronegative individuals had significantly earlier symptom onset dates than the symptomatic seropositive individuals, shorter illness duration and a much lower anosmia reporting frequency. Conclusions Self-reported belief of COVID-19 was common among our frontline worker cohort. About half of these individuals were seronegative, despite a high sensitivity of serology in this cohort, at least in individuals with previous positive PCR results. This is compatible with non-COVID-19 respiratory disease during the COVID-19 outbreak having been commonly mistaken for COVID-19 within the key worker cohort studied.