Impact of Public Health Interventions to Curb SARS-CoV-2 Spread by an Evidence-Educated Delphi-Panel and Tailored SEIR Model

Abstract

Background: With the absence of immunization, public health interventions are the basis for curbing the spread of the SARS-CoV-2 virus. Evidence of impact of non-pharmaceutical interventions (NPI) on SARS-CoV-2 spread in Germany is scarce. The objectives of this study were to use a Delphi-panel based assessment of the effectiveness of different COVID-19 specific prevention measures in order to retrospectively approximate and to prospectively predict the SARS-CoV-2 pandemic progression via a SEIR model (SEIR: Susceptible - Exposed - Infectious - Removed). The SEIR model will be made available with a modifiable user interface.Methods: We applied an evidence-educated Delphi-panel approach to elicit the impact of NPIs being discussed in Germany on the SARS-CoV-2 transmission rate R0. Initial estimates were discussed and agreed upon in two Delphi stages resulting in a final set of average efficacy and compliance estimates for each NPI. Effectiveness was defined as the product of efficacy and compliance. A discrete, deterministic SEIR model with time step of 1 day, a latency period of 1.·8 days, duration of infectiousness of 5 days, and a share of the total population of 15% assumed to be protected by immunity was developed in order to estimate the impact of selected NPI measures on pandemic course. The model was populated with the Delphi-panel results and varied in sensitivity analyses.Results: Efficacy and compliance estimates were obtained as follows (ranked by effectiveness): test and isolate 49% (efficacy)/78% (compliance), keeping distance 42%/74%, personal protection masks (cloth masks or other face masks) 33%/79%, ban of large public events 26%/96%, contact reduction 33%/59%, closure of non-essential stores 10%/97%, closure of schools 10%/100%, working from home 12%/66%, closure of restaurants 8%/96%, and improved hand hygiene 7%/54%. Applying these NPI effectiveness estimates to the SEIR model resulted in a valid replication of reported occurrence of the German SARS-CoV-2 pandemic.Conclusion: Employing an evidence-educated Delphi-panel approach for generating NPI effectiveness estimates is feasible and could help to generate model simulations with close replication of reported infected cases in Germany. Future curbing scenarios require a combination of NPIs. A Delphi-panel based NPI assessment and modelling might support public health policy decision making by informing sequence and number of needed public health measures.Funding Statement: None.Declaration of Interests: None declared.Ethics Approval Statement: Not applicable.