Table 1.

Additional sources of uncertainty used in Figs. 3 and 4.

Parameter	Range	Notes
Lockdown
Timing of lockdown, relative to epidemic	At time of n-th COVID-related death, where n is between 1 and 1000	Lockdown timing can have strong influence on its effect (see Fig. S2, Supplementary Appendix). Here we adopt a wide range of possibilities for numbers of actual deaths before lockdown
Effectiveness of lockdown in reducing transmission	15–85%	Lockdown effectiveness depends on compliance, and on achievable physical distancing in crowded settings. Again adopting wide uncertainty ranges given lack of systematic data
Asymptomatic transmission
Proportion of cases developing symptoms ($p^{\left(\mathit{sym}\right)}$)	66–90%	Parameters relating to the role of asymptomatic infections, for transmission. Scenarios having a strong role for such infections would diminish the impact of isolating symptomatic infections alone
Relative infectiousness of asymptomatic and presymptomatics vs symptomatic infection ($k)$	10–66%
Post-infection immunity
Average duration of immunity ($1/\omega )$	> 6 months	Studies with established human coronaviruses suggest immunity lasting at least a year34; for SARS-CoV-2 we assume a more pessimistic scenario of at least 6 months
Susceptibility to reinfection, relative to susceptibility to primary infection $(g)$	0–10%	0% indicates that immunity is completely protective to reinfection

See Table S1 for further parameter values.