Introduction: Symptom based clustering suggests that long COVID is a heterogenous condition with distinct phenotypes. We aimed to assess whether blood biomarkers and biological age differ across these phenotypes.

Methods: A retrospective cohort study was done using English primary care data in the Clinical Practice Research Datalink Aurum database. We included 384,786 non-hospitalised patients with confirmed SARS-CoV-2 infection. They were classified using latent class analysis as having either a broad spectrum, respiratory, or mental health/cognitive symptoms. Blood biomarkers were extracted at 12 weeks after infection to estimate biological age (PhenoAge). Differences in biomarkers and biological age across the phenotypes and asymptomatic controls were tested using ANOVA. Chronological age was subtracted from biological age to measure excess ageing.  The association between long COVID phenotypes and excess ageing was assessed using multivariable linear regression.

Results: Patients from the respiratory phenotype (n=2866) were distinct and reported lower levels of albumin, and higher levels of creatinine, glucose, C-reactive protein, erythrocyte sedimentation rate, and white cell count. Patients within the broad spectrum and respiratory phenotypes were on average 0.67 and 2.25 years biologically older than their chronological age, while patients within the mental health phenotype were 0.11 years younger (p<0.001). After adjusting for confounders, this remained statistically significant among patients within the broad symptom spectrum cluster (?: 0.50, 95% CI 0.15-0.84; p=0.005).

Conclusion: Long COVID phenotypes correspond to differences in blood biomarkers and biological ageing.