COVID-19 UK Lockdown Forecasts and R 0

Frontiers in Public Health

article

Greg Dropkin¹ 

[1] Independent Researcher, United Kingdom

关键词: COVID-19;    UK;    NHS;    modelling;    forecast;    Bayesian;    SEIR;    R 0;   

DOI  :  10.3389/fpubh.2020.00256

学科分类：社会科学、人文和艺术（综合）

来源: Frontiers

【 摘 要 】

Introduction: The first reported UK case of COVID-19 occurred on 30 January 2020. A lockdown from 24 March was partially relaxed on 10 May. One model to forecast disease spread depends on clinical parameters and transmission rates. Output includes the basic reproduction number R 0 and the log growth rate r in the exponential phase. Methods: Office for National Statistics data on deaths in England and Wales is used to estimate r . A likelihood for the transmission parameters is defined from a gaussian density for r using the mean and standard error of the estimate. Parameter samples from the Metropolis-Hastings algorithm lead to an estimate and credible interval for R 0 and forecasts for cases and deaths. Results: The UK initial log growth rate is r = 0.254 with s.e. 0.004. R 0 = 6.94 with 95% CI (6.52, 7.39). In a 12 week lockdown from 24 March with transmission parameters reduced throughout to 5% of their previous values, peaks of around 90,000 severely and 25,000 critically ill patients, and 44,000 cumulative deaths are expected by 16 June. With transmission rising from 5% in mid-April to reach 30%, 50,000 deaths and 475,000 active cases are expected in mid-June. Had such a lockdown begun on 17 March, around 30,000 (28,000, 32,000) fewer cumulative deaths would be expected by 9 June. Discussion: The R 0 estimate is compatible with some international estimates but over twice the value quoted by the UK government. An earlier lockdown could have saved many thousands of lives.

【 授权许可】

CC BY