Subphenotypes of self-reported symptoms and outcomes in long COVID: a prospective cohort study with latent class analysis

BMJ Open. 2024 Mar 14;14(3):e077869. doi: 10.1136/bmjopen-2023-077869.

Abstract

Objective: To characterise subphenotypes of self-reported symptoms and outcomes (SRSOs) in postacute sequelae of COVID-19 (PASC).

Design: Prospective, observational cohort study of subjects with PASC.

Setting: Academic tertiary centre from five clinical referral sources.

Participants: Adults with COVID-19 ≥20 days before enrolment and presence of any new self-reported symptoms following COVID-19.

Exposures: We collected data on clinical variables and SRSOs via structured telephone interviews and performed standardised assessments with validated clinical numerical scales to capture psychological symptoms, neurocognitive functioning and cardiopulmonary function. We collected saliva and stool samples for quantification of SARS-CoV-2 RNA via quantitative PCR.

Outcomes measures: Description of PASC SRSOs burden and duration, derivation of distinct PASC subphenotypes via latent class analysis (LCA) and relationship with viral load.

Results: We analysed baseline data for 214 individuals with a study visit at a median of 197.5 days after COVID-19 diagnosis. Participants reported ever having a median of 9/16 symptoms (IQR 6-11) after acute COVID-19, with muscle-aches, dyspnoea and headache being the most common. Fatigue, cognitive impairment and dyspnoea were experienced for a longer time. Participants had a lower burden of active symptoms (median 3 (1-6)) than those ever experienced (p<0.001). Unsupervised LCA of symptoms revealed three clinically active PASC subphenotypes: a high burden constitutional symptoms (21.9%), a persistent loss/change of smell and taste (20.6%) and a minimal residual symptoms subphenotype (57.5%). Subphenotype assignments were strongly associated with self-assessments of global health, recovery and PASC impact on employment (p<0.001) as well as referral source for enrolment. Viral persistence (5.6% saliva and 1% stool samples positive) did not explain SRSOs or subphenotypes.

Conclusions: We identified three distinct PASC subphenotypes. We highlight that although most symptoms progressively resolve, specific PASC subpopulations are impacted by either high burden of constitutional symptoms or persistent olfactory/gustatory dysfunction, requiring prospective identification and targeted preventive or therapeutic interventions.

Keywords: COVID-19; respiratory infections; virology.

Publication types

  • Observational Study

MeSH terms

  • Adult
  • COVID-19 Testing
  • COVID-19* / epidemiology
  • Disease Progression
  • Dyspnea
  • Humans
  • Latent Class Analysis
  • Post-Acute COVID-19 Syndrome*
  • Prospective Studies
  • RNA, Viral
  • SARS-CoV-2
  • Self Report

Substances

  • RNA, Viral