Effectiveness of covid-19 vaccines against post covid-19 condition/long covid: systematic review and meta-analysis

Background Persons infected with SARS-CoV-2 can develop long-term symptoms known as post-COVID-19-condition (PCC; symptoms ≥three months after infection) or long-COVID (LC; symptoms ≥one month after infection). Vaccination against COVID-19 might prevent PCC/LC, but the extent of protection is unclear. Objective Aim of this systematic review was to evaluate vaccine efficacy/effectiveness (VE) of COVID-19 vaccines given prior to SARS-CoV-2-infection in preventing PCC or LC. Methods Data Sources Studies were identified in Embase, MEDLINE, PreView, COVID-19 L.OVE repository and Cochrane Library up to August 1, 2024. Study eligibility criteria, Participants, Intervention Randomized controlled trials (RCTs) and non-randomized studies of interventions (NRSI) that investigated immunization with a COVID-19 vaccine before SARS-CoV-2-infection were eligible, irrespective of participant age and sex. Assessment of risk of bias Risk of bias was assessed using ROBINS-I. Methods of data synthesis Primary outcome was PCC, secondary outcomes were LC, time until reconvalescence, limitations in every day activity, and quality of life. Meta-analyses were primarily conducted using the random-effects model. Results 6423 records were screened and 65 non-randomized studies of interventions (NRSI) reporting adjusted estimates were included, comprising >5.7 mio. participants. VE for ≥one vaccine dose against PCC was 41.0% (95% confidence interval (CI) 27.8%; 51.7%; 22 NRSI, certainty of evidence: low). VE after one, two or three doses versus unvaccinated was 19.1% (-119.4%; 70.2%, three NRSI), 43.2% (4.5%; 66.2%; four NRSI) and 70.0% (30.0%; 87.0%; one NRSI), respectively. In <18-years-olds, VE against PCC was 26% for ≥one dose (-4%; 48%, one NRSI) and in >60-years-olds 41% (17%; 59%, one NRSI). VE after pre-Omicron-SARS-CoV-2 infection was 32.1% (-54.3%; 70.1%, three NRSI) and 20.9% (-10.1%; 43.3%, two NRSI) after Omicron-infection. Sensitivity analyses indicated no influence of risk of bias and effect measure. Conclusions ...