Predictive performance of multi-model ensemble forecasts of COVID-19 across European nations (preprint)

Background: Short-term forecasts of infectious disease burden can contribute to situational awareness and aid capacity planning. Based on best practice in other fields and recent insights in infectious disease epidemiology, one can maximise the predictive performance of such forecasts if multiple models are combined into an ensemble. Here we report on the performance of ensembles in predicting COVID-19 cases and deaths across Europe between 08 March 2021 and 07 March 2022. Methods: We used open-source tools to develop a public European COVID-19 Forecast Hub. We invited groups globally to contribute weekly forecasts for COVID-19 cases and deaths reported from a standardised source over the next one to four weeks. Teams submitted forecasts from March 2021 using standardised quantiles of the predictive distribution. Each week we created an ensemble forecast, where each predictive quantile was calculated as the equally-weighted average (initially the mean and then from 26th July the median) of all individual models predictive quantiles. We measured the performance of each model using the relative Weighted Interval Score (WIS), comparing models forecast accuracy relative to all other models. We retrospectively explored alternative methods for ensemble forecasts, including weighted averages based on models past predictive performance. Results: Over 52 weeks we collected and combined up to 28 forecast models for 32 countries. We found a weekly ensemble had a consistently strong performance across countries over time. Across all horizons and locations, the ensemble performed better on relative WIS than 84% of participating models forecasts of incident cases (with a total N=862), and 92% of participating models forecasts of deaths (N=746). Across a one to four week time horizon, ensemble performance declined with longer forecast periods when forecasting cases, but remained stable over four weeks for incident death forecasts. In every forecast across 32 countries, the ensemble outperformed most contributing models when forecasting either cases or deaths, frequently outperforming all of its individual component models. Among several choices of ensemble methods we found that the most influential and best choice was to use a median average of models instead of using the mean, regardless of methods of weighting component forecast models. Conclusions: Our results support the use of combining forecasts from individual models into an ensemble in order to improve predictive performance across epidemiological targets and populations during infectious disease epidemics. Our findings further suggest that median ensemble methods yield better predictive performance more than ones based on means. Our findings also highlight that forecast consumers should place more weight on incident death forecasts than incident case forecasts at forecast horizons greater than two weeks.

BNT162b2 post-exposure-prophylaxis against COVID-19 (preprint)

Background: During the COVID-19 pandemic, post-exposure-prophylaxis is not a practice. Following exposure, only patient isolation is imposed. Moreover, no therapeutic prevention approach is applied. We asked whether evidence exists for reduced mortality rate following post-exposure-prophylaxis. Methods: To estimate the effectiveness of post-exposure-prophylaxis, we obtained data from the Israeli Ministry of Health (MoH) registry. The study population consisted of Israeli residents aged 12 years and older, identified for the first time as PCR-positive for SARS-CoV-2, between December 20th, 2020 (the beginning of the vaccination campaign) and October 7th, 2021. We compared 'recently injected' patients - that proved PCR-positive on the same day or on one of the five consecutive days after first vaccination (representing an unintended post-exposure-prophylaxis), to unvaccinated control group. Results: Among Israeli residents identified PCR-positive for SARS-CoV-2, 11,690 were found positive on the day they received their first vaccine injection (BNT162b2) or on one of the 5 days thereafter. In patients over 65 years, 143 deaths occurred among 1413 recently injected (10.12%) compared to 280 deaths among the 1413 unvaccinated (19.82%), odd ratio (OR) 0.46 (95% confidence interval (CI), 0.36 to 0.57; P<0.001). The most significant reduction in the death toll was observed among the 55 to 64 age group, with 8 deaths occurring among the 1322 recently injected (0.61%) compared to 43 deaths among the 1322 unvaccinated control (3.25%), OR 0.18 (95% CI, 0.07 to 0.39; P<0.001). Conclusion: Post-exposure-prophylaxis is effective against death in COVID-19 infection. Israeli MoH Registry Number: HMO-0372-20

Estimating COVID-19-induced Excess Mortality in Lombardy (preprint)

We compare the expected all-cause mortality with the observed one for different age classes during the pandemic in Lombardy, which was the epicenter of the epidemic in Italy and still is the region most affected by the pandemic. A generalized linear mixed model is introduced to model weekly mortality from 2011 to 2019, taking into account seasonal patterns and year-specific trends. Based on the 2019 year-specific conditional best linear unbiased predictions, a significant excess of mortality is estimated in 2020, leading to approximately 35000 more deaths than expected, mainly arising during the first wave. In 2021, instead, the excess mortality is not significantly different from zero, for the 85+ and 15-64 age classes, and significant reductions with respect to the 2020 estimated excess mortality are estimated for other age classes.

Spatial modelling of COVID-19 incident cases using Richards' curve: an application to the Italian regions (preprint)

We introduce an extended generalised logistic growth model for discrete outcomes, in which a network structure can be specified to deal with spatial dependence and time dependence is dealt with using an Auto-Regressive approach. A major challenge concerns the specification of the network structure, crucial to consistently estimate the canonical parameters of the generalised logistic curve, e.g. peak time and height. Parameters are estimated under the Bayesian framework, using the {\texttt{ Stan}} probabilistic programming language. The proposed approach is motivated by the analysis of the first and second wave of COVID-19 in Italy, i.e. from February 2020 to July 2020 and from July 2020 to December 2020, respectively. We analyse data at the regional level and, interestingly enough, prove that substantial spatial and temporal dependence occurred in both waves, although strong restrictive measures were implemented during the first wave. Accurate predictions are obtained, improving those of the model where independence across regions is assumed.

Nowcasting COVID-19 incidence indicators during the Italian first outbreak (preprint)

A novel parametric regression model is proposed to fit incidence data typically collected during epidemics. The proposal is motivated by real-time monitoring and short-term forecasting of the main epidemiological indicators within the first outbreak of COVID-19 in Italy. Accurate short-term predictions, including the potential effect of exogenous or external variables are provided; this ensures to accurately predict important characteristics of the epidemic (e.g., peak time and height), allowing for a better allocation of health resources over time. Parameters estimation is carried out in a maximum likelihood framework. All computational details required to reproduce the approach and replicate the results are provided.

An ensemble approach to short-term forecast of COVID-19 intensive care occupancy in Italian Regions (preprint)

The availability of intensive care beds during the Covid-19 epidemic is crucial to guarantee the best possible treatment to severely affected patients. In this work we show a simple strategy for short-term prediction of Covid-19 ICU beds, that has proved very effective during the Italian outbreak in February to May 2020. Our approach is based on an optimal ensemble of two simple methods: a generalized linear mixed regression model which pools information over different areas, and an area-specific non-stationary integer autoregressive methodology. Optimal weights are estimated using a leave-last-out rationale. The approach has been set up and validated during the epidemic in Italy. A report of its performance for predicting ICU occupancy at Regional level is included.