Basic prediction methodology for covid-19: estimation and sensitivity considerations

Tom Britton

We now include preventive measures. We still assume R 0 = 3 and a doubling time of d = 3 days without preventions. Then preventive measures are put in place, resulting in a prolonged doubling time of case fatalities starting (around) s d = 21 days later. Suppose that the new observed doubling time equals d E = 5 days. These two doubling times correspond to the exponential rates r = 0.23 and r E = ln(2)/5 = 0.14. If we first assume relation (1) to hold, this reduced exponential rate gives a lower bound the magnitude ρ given by ρ 1 = (r − r E )g 1 /R 0 = 0.27. If we instead consider relation (2) we obtain the upper bound of the magnitude as ρ 2 = 1 − e (r−r E )g 2 = 0.35. It hence follows that the magnitude ρ of the preventive measure giving rise to an increased doubling time from d = 3 to d E = 5 days is estimated to lie somewhere in the interval (0.27, 0.35). The effective reproduction R E is hence estimated to lie between R 0 (1 − 0.35) = 1.93 and R 0 (1 − 0.27) = 2.19, so a substantial reduction.