Background: Seasonal influenza places a substantial burden annually on healthcare services. Policies during the COVID-19 pandemic limited the transmission of seasonal influenza, making the timing and magnitude of a potential resurgence difficult to ascertain and its impact important to forecast. Methods: We have developed a hierarchical generalised additive model (GAM) for the short-term forecasting of hospital admissions with a positive test for the influenza virus sub-regionally across England. The model incorporates a multi-level structure of spatio-temporal splines, weekly cycles in admissions, and spatial correlation. Using multiple performance metrics including interval score, coverage, bias, and median absolute error, the predictive performance is evaluated for the 2022-2023 seasonal wave. Performance is measured against autoregressive integrated moving average (ARIMA) and Prophet time series models. Results: Across the epidemic phases the hierarchical GAM shows improved performance, at all geographic scales relative to the ARIMA and Prophet models. Temporally, the hierarchical GAM has overall an improved performance at 7 and 14 day time horizons. The performance of the GAM is most sensitive to the flexibility of the smoothing function that measures the national epidemic trend. Conclusions: This study introduces an approach to short-term forecasting of hospital admissions for the influenza virus using hierarchical, spatial, and temporal components. The methodology was designed for the real time forecasting of epidemics. This modelling framework was used across the 2022-2023 winter for healthcare operational planning by the UK Health Security Agency and the National Health Service in England.
In this paper, we present work on SARS-CoV-2 transmission in UK higher education settings using multiple approaches to assess the extent of university outbreaks, how much those outbreaks may have led to spillover in the community, and the expected effects of control measures. Firstly, we found that the distribution of outbreaks in universities in late 2020 was consistent with the expected importation of infection from arriving students. Considering outbreaks at one university, larger halls of residence posed higher risks for transmission. The dynamics of transmission from university outbreaks to wider communities is complex, and while sometimes spillover does occur, occasionally even large outbreaks do not give any detectable signal of spillover to the local population. Secondly, we explored proposed control measures for reopening and keeping open universities. We found the proposal of staggering the return of students to university residence is of limited value in terms of reducing transmission. We show that student adherence to testing and self-isolation is likely to be much more important for reducing transmission during term time. Finally, we explored strategies for testing students in the context of a more transmissible variant and found that frequent testing would be necessary to prevent a major outbreak.
12 April 2025 in International Journal of Epidemiology
Background: Seasonal respiratory viruses cause substantial pressure on healthcare systems, particularly over winter. System managers can mitigate the impact on patient care when they anticipate hospital admissions due to these viruses. Hospitalization forecasts were used widely during the SARS-CoV-2 pandemic. Now, resurgent seasonal respiratory pathogens add complexity to system planning. We describe how a suite of forecasts for respiratory pathogens, embedded in national and regional decision-making structures, were used to mitigate the impact on hospital systems and patient care. Methods: We developed forecasting models to predict hospital admissions and bed occupancy 2 weeks ahead for COVID-19, influenza, and respiratory syncytial virus (RSV) in England over winter 2023–4. Bed occupancy forecasts were informed by the ensemble admissions models. Forecasts were delivered in real time at multiple scales. The use of sample-based forecasting allowed effective reconciliation and trend interpretation. Results: Admission forecasts, particularly RSV and influenza, showed high efficacy at regional levels. Bed occupancy forecasts had well-calibrated coverage owing to informative admissions forecasts and slower moving trends. National admissions forecasts had mean absolute percentage errors of 27.3%, 30.9%, and 15.7% for COVID-19, influenza, and RSV, respectively, with corresponding 90% coverages of 0.439, 0.807, and 0.779. Conclusion: These real-time winter infectious disease forecasts produced by the UK Health Security Agency for healthcare system managers played an informative role in mitigating seasonal pressures. The models were delivered regularly and shared widely across the system to key users. This was achieved by producing reliable, fast, and epidemiologically informed ensembles of models, though a higher diversity of model approaches could have improved forecast accuracy.
Background: Norovirus is a leading cause of acute gastroenteritis, adding to strain on healthcare systems. Diagnostic test reporting of norovirus is often delayed, resulting in incomplete data for real-time surveillance. Methods: To nowcast the real-time case burden of norovirus a generalised additive model (GAM), semi-mechanistic Bayesian joint process and delay model “epinowcast”, and Bayesian structural time series (BSTS) model including syndromic surveillance data were developed. These models were evaluated over weekly nowcasts using a probabilistic scoring framework. Results: Using the weighted interval score (WIS) we show a heuristic approach is outperformed by models harnessing time delay corrections, with daily mean WIS = 7.73, 3.03, 2.29 for the baseline, “epinowcast”, and GAM, respectively. Forecasting approaches were reliable in the event of temporally changing reporting values, with WIS = 4.57 for the BSTS model. However, the syndromic surveillance (111 online pathways) did not improve the BSTS model, WIS = 10.28, potentially indicating poor correspondence between surveillance indicators. Interpretation: Analysis of surveillance data enhanced by nowcasting delayed reporting improves understanding over simple model assumptions, important for real-time decision making. The modelling approach needs to be informed by the patterns of the reporting delay and can have large impacts on operational performance and insights produced.
Background: From 2023 to 2024, England had the largest measles outbreak in over a decade. Lags from suspected cases’ symptom onset to the availability of test results mean laboratory-confirmed case data are inherently retrospective rather than real-time. Reporting lags vary by measles prevalence and whether testing was for diagnostic or surveillance purposes. Nowcasting models predict future backfilling of reported cases and can estimate recent trends. Methods: We developed a generalised additive model framework accounting for reporting delays, location, and day-of-week effects in line-list data by onset date. The model was re-fit weekly providing real-time nowcasts and directional trends for national and regional users. Retrospectively, we tested alternative specifications to optimise structure and confirm predictive performance, evaluating with log weighted interval score (WIS) and ranked probability score (RPS). Results: For case count estimates, the operational and retrospective models outperformed the baseline model, with a lower average daily national log WIS by 42% and 41%, respectively, and similar regional improvements. For four-week trend direction, the operational and retrospective models provided better national estimates than the baseline with an average RPS lower by 69% and 6% respectively. Regionally, they also outperformed the baseline model in London, but the baseline model offered better performance for the smoother single-peak West Midlands epidemic. An alternative model indexed by report date instead sometimes outperformed the other nowcasting models for trend direction but also lagged changes in trend. Interpretation: Our work highlights the utility for real-time nowcasting models during outbreaks to inform fast-evolving trends, and the need for early access of accurate reporting delay data to facilitate effective modelling.