COVID-19 lockdown: if, when and how

Background: With the lack of an effective SARS-CoV-2 vaccine, mathematical modeling has stepped up in the COVID-19 management to guide non-pharmaceutical intervention (NPI) policies. Complete lockdown has been characterized as the most powerful strategy for the epidemic; anyhow, it is associated with undeniable negative consequences. Not aware that global panic could make countries adopt premature and lengthy lockdowns, previous studies only warned about the inefficacy of late quarantine sets. Therefore, we proposed ourselves to find the optimal timing and lasting for COVID-19 suppressive measures. Methods: We used our previously elaborated compartmental SEIR (Susceptible-Exposed-Infected-Recovered) model to scan different timings for lockdown set and various lockdown lengths under different reproduction number (R0) scenarios. We explored healthcare parameters focusing on ICU occupation and deaths since they condition the sanitary system and reflect the severity of the epidemic. Results: The timing for the lockdown trigger varies according to the original R0 and has great impact on ICU usage and fatalities. The less the R0 the later the lockdown should be for it to be effective. The lockdown length is also something to consider. Too short lockdowns (~15 days) have minimal effect on healthcare parameters, but too long quarantines (>45 days) do not benefit healthcare parameters proportionally when compared to more reasonable 30 to 45-day lockdowns. We explored the outcome of the combination of a 45-day lockdown followed by strict mitigation measures sustained in time, and interestingly, it outperformed the lengthy quarantine. Additionally, we show that if strict mitigation actions were to be installed from the very beginning of the epidemic, lockdown would not benefit substantially regarding healthcare parameters. Conclusion: Lockdown set timing and lasting are non-trivial variables to COVID-19 management.