Data Science Seminar: Experimental Design in Testing Times: Pandemic Yes, Panic No.11 Jun 2021, by Sponsored events in
11 February 2021 (Online)
Sheila Bird, Medical Research Council (MRC), Biostatistics Unit, Cambridge
Testing times range over in vitro diagnostics (have I got it; am I infectious; have I had it; do I still have antibodies), surveillance (ONS Infection Survey; REACT; blood donors), adherence to and duration of quarantine (self-isolation as mandated by Test & Trace), mass screening of asymptomatic citizens or school-children, testing of vaccines to reduce cases or of pharmaceutical interventions to reduce disease severity of mortality.
Powerful (ie sufficiently large) randomized controlled trials are sine qua non for the licensing of pharmaceutical medicines and vaccines. Study size, randomization and/or random sampling matter in most of the other contexts too.
But too often breached. I review a series of UK examples, from outstandingly successful (RECOVERY Trial) to profoundly concerning (Test & Trace; mass screening of school-children; roll-out of mRNA vaccine with 2nd dose delayed from its RCT-schedule [1/22 days] to 1/85 days as maximum) where appreciation of experimental design has been hard-won.