Study population and setting
The objective of the study was to model COVID-19 growth trajectories in Germany, and use that to determine the impact of sequentially implemented physical distancing interventions on the spread of COVID-19. The authors specified an epidemic model using confirmed COVID-19 cases in Germany through April 21, 2020, to predict the rate of new infections in correspondence with three time points when the following physical distancing interventions were implemented: 1) suspension of large (1,000+) public gatherings (March 9), 2) business and school closures (March 16), and 3) a nationwide contact ban, including closure of all non-essential businesses (March 23). The authors also simulated different implementation scenarios, including the magnitude and timing of interventions, to assess their relative impact on the spread of COVID-19.
Summary of Main Findings
The growth rate of COVID-19 case rate decreased from 0.43 to 0.25 following the suspension of large public gatherings. The growth rate subsequently decreased again from 0.25 to 0.15 following widespread business and school closures on March 16. The implementation of the contact ban ultimately resulted in a negative growth [-3% (95%CI: -5%, -2%)]; resulting in declining case counts. Scenario-based models, which varied the timing of implementation of physical interventions to five days before and after March 16, revealed strong temporal effects of these interventions, with earlier and later implementation resulting in major differences in cumulative COVID-19 cases.
The primary epidemic model in the study contained a limited number of parameters, which facilitated epidemic growth estimations using a limited number of data points (i.e., COVID-19 cases). The authors also uniquely specified change points into their model parameters to flexibly account for differential rates of exponential growth/decline in COVID-19 cases after implementation of various physical distancing interventions at different points in time. Models fit observed case count data well, and numerous sensitivity analyses were performed.
Due to the limited number of parameters specified in the study’s final model, other relevant epidemic parameters (e.g., testing capacity, geographic differences in epidemic trajectories) were not included, which may be essential for more robust, accurate forecasting of COVID-19 spread and impact of physical distancing measures in other settings. Data was from just one country and the models required a number of assumptions, making the broader implications unclear.
This is among the first studies to specify change points aligned with the implementation of various non-pharmaceutical interventions in an epidemic model for estimating their relative impact in curtailing COVID-19 transmission.
This review was posted on: 4 July 2020