Rapid assessment of the impact of lockdown on the COVID-19 epidemic in Portugal

Background: Portugal took early action to control the COVID19 epidemic, imposing a lockdown on March 16 when it recorded only 62 cases of COVID-19 per million inhabitants and no reported deaths. The Portuguese people complied quickly, reducing their overall mobility by 80% . We estimate the impact of the lockdown in Portugal in terms of reducing burden on the health service. Methods: Using publicly available official data from the Portuguese Directorate-General of Health we forecasted epidemic curves for: Cases, hospital inpatients (overall and in ICU), and deaths without lockdown, assuming that the impact of containment measures would start 14 days after lockdown was implemented. We used exponential smoothing models for deaths, intensive care (ICU) and hospitalizations and an ARIMA model for number of cases. Models were selected considering fitness to the observed data to the 31st of March 2020. We then compared observed(with intervention) and forecasted curves( without intervention). Results: Between April 1 and April 15, there were 146 fewer deaths(-25%), 5568 fewer cases (-23%) and, as of April 15, there were 519 fewer ICU inpatients(-69%) and 508 fewer overall hospital inpatients(-28%) than forecasted without lockdown. On April 15 the number of ICU inpatients could have reached 748, three times higher than the observed value (229) if the intervention had been delayed. Conclusion: If the lockdown had not been implemented in mid-March, Portugal ICU capacity (528 ICU beds) would likely have been breached in the first half of April. The lockdown seems to have been effective in reducing transmission of SARS-Cov-2, serious Covid-19 illness and associated mortality, thereby decreasing demand on health services. Early action allowed time for the National Health Service to acquire protective equipment, to increase capacity to test and cope with the surge in hospital and ICU demand caused by the pandemic.