This is a summary, written by members of the CITF Secretariat, of:
Donnici C, Ilincic C, Cao C, Zhang C, Deveaux G, Clifton D, Buckeridge D, Bobrovitz N, Arora RK. Timeliness of reporting of SARS-CoV-2 seroprevalence results and their utility for infectious disease surveillance. Epidemics. Dec 2022. DOI: 10.1016/j.epidem.2022.100645
The results and/or conclusions contained in the research do not necessarily reflect the views of all CITF members.
A study relying on data from CITF-funded SeroTracker, published in Epidemics by Dr. Rahul Arora (University of Calgary), showed that peer-reviewed scientific papers and preprints of COVID-19 seroprevalence studies are published more slowly than those published via other means, thereby diminishing their usefulness to public health decision-makers during a time-sensitive health emergency response. The researchers recommend the development of a global platform or data repository to enable continuous and fast-track dissemination of serosurveillance data for more timely use. Despite the need for evidence to inform best practices in high-risk populations, studies of the general population are published faster than those focusing on specific populations.
- Most seroprevalence studies were released in peer-reviewed journals (59%) or as preprints (24%), while institutional reports were less frequent (8%).
- The median time to publication was 154 days across all types of seroprevalence studies. It was longest for peer-reviewed articles (212 days), followed by preprints (101 days). It was shortest for media articles (12 days) and institutional reports (18 days).
- Conference presentations and media articles were more biasedBias refers to the lack of internal validity due to any systematic error in a study. Bias results in the incorrect assessment of the association between an exposure and an effect in a population. than other types of seroprevalence studies, with only 5% and 10%, respectively, presenting a low-to-moderate risk of bias.
- Researchers identify many studies – particularly some government reports – that were published in a timely fashion and with low risk of bias, which points towards creation of a useful repository of reliable data for surveillance.
- Seroprevalence studies of healthcare workers (hazard ratio: 0.58)Here, the hazard ratio represents the odds for a study focused on the healthcare workers to be published faster than a a study focused on the general population. A hazard ratio of 0.58 could be interpreted as followed: seroprevalence studies of healthcare workers are 42% less likely to be published faster than studies of the general population. , blood donors (hazard ratio: 0.76), other special populations (hazard ratio: 0.58), or multiple populations (hazard ratio: 0.69) took significantly longer to publish than those studying the general population.
The longer delay in publishing peer-reviewed articles and preprints limits the use of academic literature for reporting seroprevalence, especially during a health crisis.
This study relies on the analysis of 1844 seroprevalence studies from the SeroTracker living systematic review database, assembled between January 1, 2020 and December 31, 2021. For each study, timeliness was calculated as the amount of time between the last date of participant sample collection and the first public release of the study results. Validity of the studies was assessed based on the performance of the serological test used and the correction for sampling errors. Representativeness was determined by comparing the sample of participants used in the study with the target population.