SARS-CoV-2

Amsterdam Long COVID muscle-biopsy PEM cohort (Appelman)

Case-control · N=46 · control Healthy convenience · 1 obs · max follow-up n/r mo

Measures: Skeletal muscle pathology (biopsy)

Small longitudinal case-control study (25 long COVID patients with PEM vs 21 healthy controls) and the only cohort in this database with tissue-level objective measures: a two-day cardiopulmonary exercise test to induce post-exertional malaise, with skeletal muscle biopsies before and after. It documented exercise-induced myopathy, metabolic disturbance, and amyloid-containing deposits that worsened after PEM. Included for the objective-measures and PEM-assessment gap matrices rather than for symptom prevalence; not denominator-defined for prevalence estimation.

Bergen COVID-19 home-isolated cohort (Blomberg)

Prospective (inception) · N=312 · control None · 7 obs · max follow-up 6 mo

Measures: Any persistent symptom, Cognitive dysfunction, Dyspnoea, Fatigue, Loss of smell/taste, Memory impairment

Notable because it captured a near-complete community denominator (82% of local cases) and included predominantly mild, home-isolated patients, showing that persistent symptoms at 6 months affect even young adults with mild acute disease. No unexposed comparator.

Jinyintan (Wuhan) COVID-19 discharge cohort (Huang)

Prospective (non-inception) · N=1733 · control Unexposed (matched) · 2 obs · max follow-up 24 mo

Measures: Fatigue

Large prospective survivor cohort with matched community controls at later timepoints. Integrity caveat: the 6-month paper received a Lancet Expression of Concern (2022) and was retracted and republished (2023) after some variables were found to have been disordered; the principal conclusions were upheld in the republished version and in the 1- and 2-year follow-ups. Retained here with this replication note precisely because the database records integrity status rather than hiding it.

Long COVID Host Genetics Initiative GWAS

Registry linkage · N=6450 · control Unexposed (unmatched) · 0 obs · max follow-up n/r mo

Measures: no observations yet

The first genome-wide significant host-genetic association for long COVID: the FOXP4 locus (rs9367106, OR ~1.65), independent of FOXP4's known association with severe acute COVID-19 and replicated in an independent sample. FOXP4 acts in lung morphogenesis and respiratory-epithelial immunity. A consortium meta-analysis rather than a single followed cohort; genetic factors are temporality-valid by construction.

NIH RECOVER-Adult Cohort

Prospective (non-inception) · N=15000 · control Unexposed (unmatched) · 3 obs · max follow-up 48 mo

Measures: Any persistent symptom, Fatigue

The flagship US Long COVID study: the largest prospective, controlled Long COVID cohort globally. Enrols SARS-CoV-2 positive adults across the full spectrum (asymptomatic through severe acute COVID) plus SARS-CoV-2 negative controls from the same communities. Standardised protocol across >200 sites with harmonised data collection, imaging, and biospecimen banking. Acute-phase and convalescent specimens banked — this is the most important specimen-banking Long COVID cohort. Multiple substudies are publishing as the cohort matures; add publications and observations iteratively as substudies report. The design is prospective but non-inception (many participants enrolled post-acutely with retrospective infection date); this is recorded in the design field. The denominator is defined (all enrolled with confirmed SARS-CoV-2 status) but the source population denominator is the catchment areas of the 200+ sites, not a closed population.

Patient-Led Long COVID international cohort (Davis 2021)

Survey · N=3762 · control None · 7 obs · max follow-up 7 mo

Measures: Cognitive dysfunction, Fatigue, Not working due to illness, Post-exertional malaise

Deliberately included as a methodologically weak but historically pivotal cohort: a large (n=3762, 56 countries) self-selected online convenience sample recruited through patient support communities, with no control group, no laboratory confirmation for most respondents, and self-reported outcomes. Its breadth of symptom characterisation (203 symptoms across 10 organ systems) is unmatched, but the design cannot estimate population prevalence. Useful in the gap matrix to contrast against denominator-defined cohorts.

PHOSP-COVID (UK post-hospitalisation COVID-19)

Prospective (non-inception) · N=1965 · control None · 2 obs · max follow-up 12 mo

Measures: Felt fully recovered

Large multicentre prospective cohort of hospitalised COVID-19 survivors with an active biobank and deep phenotyping/inflammation profiling. Full recovery was reported by only about a quarter of participants and did not improve meaningfully between 5 months and 1 year. No unexposed comparison group.