These pages feature some of the mathematical modeling efforts from Penn State’s Center for Infectious Disease Dynamics (CIDD) to forecast hospital needs during the 2020 COVID epidemic in the United States. The forecasts and statistical analyses described below represent a collaboration between the labs of Ephraim Hanks, Justin Pritchard, and Maciej Boni. Please see the About page for contributors and colleagues. The Methods page contains details on the modeling and statistical approaches used, but note that individual results below may have been generated using an adjusted version of these methods, so please read each post carefully. Email us with any questions.
October 7 2020 – Attack Rates through Sep 15 – Analyses of the total population attack rate — i.e. the total number of individuals infected in a population — were completed. Attack rates through July 31 are 6.91% (95% CredInt: 6.28% — 7.73%) in Rhode Island, 5.37% (4.97% — 5.98%) in Massachusetts, and 4.39% (4.17% — 4.77%) in Pennsylvania. Note that our attack-rate estimate for Massachusetts has dropped, due to removal of the ‘new hospitalizations’ data stream from our analysis. Read more.
August 13 2020 – Attack Rates through July 31 – Analyses of the total population attack rate — i.e. the total number of individuals infected in a population — were completed. Attack rates through July 31 are 5.43% (95% CredInt: 4.74% — 6.47%) in Rhode Island, 6.88% (5.55% — 8.38%) in Massachusetts, and 3.81% (2.89% — 5.34%) in Pennsylvania. Read more.
July 18 2020 – Preliminary Analysis – Our first analysis on Pennsylvania data shows that the March-April lockdown reduced transmission-capable population mixing by more than 80% in Pennsylvania. We estimate that about 39% of new cases and 48% of new hospitalizations are reported to the Pennsylvania DOH COVID-19 reporting system. Using these underreporting rates, we estimate that between 2.3% and 6.4% of Pennsylvania residents have been infected with SARS-CoV-2 through July 7. Depending on assumptions made on asymptomatic infection, median attack rate estimates are 3.3% and 4.9%. Read more.
June 30 2020 – Preliminary Analysis, New Model Version – Using the updated model and inference framework from our RI-analysis last week, we show that Massachusetts’s lockdown period in March and April was successful at suppressing new infections and keeping infection levels low through late June; parts of Massachusetts society and economy began to re-open in early May with no observed additional increase in disease burden. Approximately half of new symptomatic COVID-19 cases are reported to the Massachusetts Department of Public Health reporting system, and about 40% of new hospitalized cases are reported, suggesting that some improvements in reporting may be beneficial before the school year starts. We estimate that between 5.4% and 16.1% of Massachusetts residents have been infected with SARS-CoV-2 through June 27. Read more.
June 23 2020 – Preliminary Analysis, New Model Version – An updated model version allows better precision in measuring certain clinical parameters, and a new mobility data allow us to see when population mixing resumed after the lockdown period. Rhode Islanders began leaving home more regularly in early May, but case numbers did not increase, suggesting that poeple were following basic hygiene and distancing guidelines. We estimate that at most 6.6% of Rhode Islanders have been infected by June 16. This is a downward-revised estimate from our previous work, as we have now included more comprehensive data on the rate of asymptomatic infection by age group. Read more.
May 23 2020 – Data Update, Bug Fix – An updated analysis on Rhode Island case data through May 16 gives an estimate of the reporting rate ρ=0.74 (95% HPD: 0.56 – 1.0). We estimate that 5.3% of Rhode Island’s population has been infected through May 22. Re-opening sectors of the economy and society should come with improved hygiene and distancing guidelines in order to avoid an epidemic rebound. Read more.
May 17 2020 – Preliminary Analysis – An analysis of Massachusetts case confirmations and hospitalization data indicate a 50% drop in population mixing and social/physical contact rates between early March and late March, and a 70% drop by mid-April. We estimate that 51% of all symptomatic COVID cases in Massachusetts are captured by the health system. Relaxing social distancing measures in Massachusetts is likely to result in a rebound of case numbers and a nearly full-sized epidemic. Read more.
May 14 2020 – Preliminary Analysis – An analysis of Rhode Island’s case confirmations and hospitalization data indicate a 40% drop in population mixing and social/physical contact rates between early March and late March, and a 60% drop by mid-April. We estimate that 71% of all symptomatic COVID cases in Rhode Island are captured by the health system. We estimate probabilities of hospitalization and ICU admission, which show similar age patterns to estimates obtained on other parts of the US. Read more.
April 27 2020 – Preliminary Analysis – First data fits on Rhode Island’s case confirmations only indicate a >80% drop in population mixing and social/physical contact rates between early March and late March. The contact rate reduction appears to have been sustained through late April. Forecast for May 15 indicates an incidence of 1260 daily reported cases (IQR: 1071–1678). Read more.