Information bias for case definitions and mortality between and within studies over time undermines our understanding of COVID-19 transmission and disease severity

Global internet connectivity and rapid data extraction have now allowed for case numbers and deaths from SARS-CoV-2 to be monitored in real-time [1,2]. However, to make reliable inferences about the risk factors for infection, hospitalization, and mortality from COVID-19, we must examine important biases that may lead to inaccurate conclusions, based on the use of routinely collected observational data. In particular, the problem of information bias relating to case and mortality definitions for COVID-19 that can vary between and sometimes even within studies has been inadequately addressed.