Are influenza statistics based on diagnoses by signs and symptoms, by testing of patients presenting with apparent influenza symptoms, or by random serological testing of the population?

Unless it's the second one, what value does directly comparing influenza statistics to covid statistics have?