Assessing the impact of Delta and Omicron in German intensive care units: a retrospective multistate analysis based on nationwide inpatient billing data (preprint)

Background The spread of several severe acute respiratory syndrome coronavirus type 2 (SARS-CoV-2) variants of concern (VOC) repeatedly led to increasing numbers of coronavirus disease 2019 (COVID-19) patients in German intensive care units (ICUs), resulting in capacity shortages and even transfers of COVID-19 intensive care patients between federal states in late 2021. In this respect, there is limited evidence on the impact of predominant VOC in German ICUs on the population level.Methods A retrospective cohort study was conducted from July 01, 2021, to May 31, 2022, using nationwide inpatient billing data from German hospitals on COVID-19 intensive care patients and SARS-COV-2 sequence data from Germany. A multivariable Poisson regression analysis was performed to estimate incidence rate ratios (IRRs) of transfer (to another hospital during inpatient care), discharge and death of COVID-19 intensive care patients associated with Delta or Omicron, adjusted for age group and sex. Furthermore, a multistate model was used for the clinical trajectories of COVID-19 intensive care patients to estimate their competing risk of transfer, discharge or death associated with Delta or Omicron, while further addressing patient age.Results Poisson regression analysis comparing Omicron versus Delta infection yielded an estimated adjusted IRR of 1.23 (95% CI 1.16–1.30) for transfers, 2.27 (95% CI 2.20–2.34) for discharges and 0.98 (95% CI 0.94–1.02) for deaths. For ICU deaths in particular, the estimated adjusted IRR increased from 0.14 (95% CI 0.08–0.22) for the 0–9 age group to 4.09 (95% CI 3.74–4.47) for those aged 90 and older compared to the reference group of 60-69-year olds. Multistate analysis showed that Omicron infection was associated with a higher estimated risk of discharge for COVID-19 intensive care patients across all ages, while Delta infection was associated with a higher estimated risk transfer and death.Conclusions Retrospective, nationwide comparison of transfers, discharges and deaths of COVID-19 intensive care patients during Delta- and Omicron-dominated periods in Germany suggested overall less severe clinical trajectories with Omicron. Age confirmed as an important determinant for fatal clinical outcomes in COVID-19 intensive care patients, necessitating close therapeutic care for the elderly and appropriate public health control measures.

A nationwide multistate analysis estimating the rates and risks of transferring critically ill COVID-19 patients during the Delta and Omicron waves in Germany (preprint)

Background The spread of several SARS-CoV-2 variants of concern (VOC) led to increasing numbers of patients with coronavirus disease 2019 (COVID-19) in German intensive care units (ICU), resulting in capacity shortages and even transfers of COVID-19 ICU patients between federal states in late 2021. Comprehensive evidence on the impact of predominant VOC, in this case Delta and Omicron, on inter-hospital transfers of COVID-19 ICU patients remains scarce. Methods A retrospective cohort study was conducted from July 01, 2021 until May 31, 2022 using nationwide reimbursement inpatient count data of COVID-19 ICU patients and weekly sequence data of VOC in Germany. A multivariable Poisson regression analysis was performed to estimate incidence rates and incidence rate ratios (IRR) for competing events of transfer, discharge and death, adjusted for VOC infection, age group and sex. For corresponding risk estimation, a multistate model for the clinical trajectory in ICU was applied. Results Omicron versus Delta infection yielded estimated adjusted IRR of 1.23 (95% CI, 1.16 - 1.30) for transfer, 2.27 (95% CI, 2.20 - 2.34), for discharge and 0.98 (95% CI, 0.94 - 1.02) for death. For death in ICU, estimated adjusted IRR increased progressively with age up to 4.09 (95% CI, 3.74 - 4.47) for those 90 years and older. COVID-19 ICU patients with Omicron infection were at comparatively higher estimated risk of discharge, whereas the estimated risk of transfer and death were higher for those with Delta infection. Conclusions Inter-hospital transfers and discharges occurred more frequently in COVID-19 ICU patients with Omicron infection than in those with Delta infection, who in turn had a higher estimated risk of death. Age emerges as a relevant determinant for fatal clinical trajectories in COVID-19 ICU patients and imposes close therapeutic care.

Impact of the COVID-19 pandemic and its associated public health and social measures on emergency department utilization in Germany between March 2017 and June 2021 (preprint)

Background: During the COVID-19 pandemic and associated public health and social measures, decreasing patient numbers have been described in various healthcare settings in Germany, including emergency care. This could be explained by changes in disease burden, e.g. due to contact restrictions, but could also be a result of changes in utilization behaviour of the population. To better understand those dynamics, we analysed routine data from emergency departments to quantify changes in consultation numbers, age distribution, disease acuity and day and hour of the day during different phases of the COVID-19 pandemic. Methods: We used interrupted time series analyses to estimate relative changes for consultation numbers of 20 emergency departments spread throughout Germany. For the study period of 06-03-2017 to 13-06-2021 four different phases of the COVID-19 pandemic were defined as interruption points. Results: The most pronounced decreases were visible in the first and second wave of the pandemic, with changes of -30.0% (95%CI: -32.2%; -27.7%) and -25.7% (95%CI: -27.4%; -23.9%) for overall consultations, respectively. The decrease was even stronger for the age group of 0-19 years, with -39.4% in the first and -35.0% in the second wave. Regarding acuity levels, consultations assessed as urgent, standard and non-urgent showed the largest decrease, while the most severe cases showed the smallest decrease.Conclusions: The number of emergency department consultations decreased rapidly during the COVID-19 pandemic, without extensive variation in the distribution of patient characteristics. Smallest changes were observed for the most severe consultations and older age groups, which is especially reassuring regarding concerns of possible long-term complications due to patients avoiding urgent emergency care during the pandemic.

Using routine emergency department data for syndromic surveillance of acute respiratory illness in Germany, week 10-2017 to 10-2021 (preprint)

Background The Coronavirus disease 2019 (COVID-19) pandemic expanded the need for timely information on acute respiratory illness on the population level. Aim We explored the potential of routine emergency department data for syndromic surveillance of acute respiratory illness in Germany. Methods We included routine attendance data from emergency departments who continuously transferred data between week 10-2017 and 10-2021, with ICD-10 codes available for >75% of the attendances. Case definitions for acute respiratory illness (ARI), severe ARI (SARI), influenza-like illness (ILI), respiratory syncytial virus disease (RSV) and COVID-19 were based on a combination of ICD-10 codes, and/or chief complaints, sometimes combined with information on hospitalisation and age. Results We included 1,372,958 attendances from eight emergency departments. The number of attendances dropped in March 2020, increased during summer, and declined again during the resurge of COVID-19 cases in autumn and winter of 2020/2021. A pattern of seasonality of acute respiratory infections could be observed. By using different case definitions (i.e. for ARI, SARI, ILI, RSV) both the annual influenza seasons in the years 2017-2020 and the dynamics of the COVID-19 pandemic in 2020-2021 were apparent. The absence of the 2020/2021 flu season was visible, parallel to the resurge of COVID-19 cases. The percentage SARI among ARI cases peaked in April-May 2020 (17%) and November 2020-January 2021 (14%). Conclusion Syndromic surveillance using routine emergency department data has the potential to monitor the trends, timing, duration, magnitude and severity of illness caused by respiratory viruses, including both influenza and SARS-CoV-2.

Data-Driven Prediction of COVID-19 Cases in Germany for Decision Making (preprint)

Background: The COVID-19 pandemic has led to a high interest in mathematical models describing and predicting the diverse aspects and implications of the virus outbreak. Model results represent an important part of the information base for the decision process on different administrative levels. The Robert-Koch-Institute (RKI) initiated a project whose main goal is to predict COVID-19-specific occupation of beds in intensive care units: Steuerungs-Prognose von Intensivmedizinischen COVID-19 Kapazitaten (SPoCK). The incidence of COVID-19 cases is a crucial predictor for this occupation. Methods: We developed a model based on ordinary differential equations for the COVID-19 spread with a time-dependent infection rate described by a spline. Furthermore, the model explicitly accounts for weekday-specific reporting and adjusts for reporting delay. The model is calibrated in a purely data-driven manner by a maximum likelihood approach. Uncertainties are evaluated using the profile likelihood method. The uncertainty about the appropriate modeling assumptions can be accounted for by including and merging results of different modelling approaches. Results: The model is calibrated based on incident cases on a daily basis and provides daily predictions of incident COVID-19 cases for the upcoming three weeks including uncertainty estimates for Germany and its subregions. Derived quantities such as cumulative counts and 7-day incidences with corresponding uncertainties can be computed. The estimation of the time-dependent infection rate leads to an estimated reproduction factor that is oscillating around one. Data-driven estimation of the dark figure purely from incident cases is not feasible. Conclusions: We successfully implemented a procedure to forecast near future COVID-19 incidences for diverse subregions in Germany which are made available to various decision makers via an interactive web application. Results of the incidence modeling are also used as a predictor for forecasting the need of intensive care units.

Forecasting intensive care unit demand during the COVID-19 pandemic: A spatial age-structured microsimulation model (preprint)

Background The COVID-19 pandemic poses the risk of overburdening health care systems, and in particular intensive care units (ICUs). Non-pharmaceutical interventions (NPIs), ranging from wearing masks to (partial) lockdowns have been implemented as mitigation measures around the globe. However, especially severe NPIs are used with great caution due to their negative effects on the economy, social life and mental well-being. Thus, understanding the impact of the pandemic on ICU demand under alternative scenarios reflecting different levels of NPIs is vital for political decision-making on NPIs. Objective The aim is to support political decision-making by forecasting COVID-19-related ICU demand under alternative scenarios of COVID-19 progression reflecting different levels of NPIs. Substantial sub-national variation in COVID-19-related ICU demand requires a spatially disaggregated approach. This should not only take sub-national variation in ICU-relevant disease dynamics into account, but also variation in the population at risk including COVID-19-relevant risk characteristics (e.g. age), and factors mitigating the pandemic. The forecast provides indications for policy makers and health care stakeholders as to whether mitigation measures have to be maintained or even strengthened to prevent ICU demand from exceeding supply, or whether there is leeway to relax them. Methods We implement a spatial age-structured microsimulation model of the COVID-19 pandemic by extending the Susceptible-Exposed-Infectious-Recovered (SEIR) framework. The model accounts for regional variation in population age structure and in spatial diffusion pathways. In a first step, we calibrate the model by applying a genetic optimization algorithm against hospital data on ICU patients with COVID-19. In a second step, we forecast COVID-19-related ICU demand under alternative scenarios of COVID 19 progression reflecting different levels of NPIs. We apply the model to Germany and provide state-level forecasts over a 2-month period, which can be updated daily based on latest data on the progression of the pandemic. Results To illustrate the merits of our model, we present here “forecasts” of ICU demand for different stages of the pandemic during 2020. Our forecasts for a quiet summer phase with low infection rates identified quite some variation in potential for relaxing NPIs across the federal states. By contrast, our forecasts during a phase of quickly rising infection numbers in autumn (second wave) suggested that all federal states should implement additional NPIs. However, the identified needs for additional NPIs varied again across federal states. In addition, our model suggests that during large infection waves ICU demand would quickly exceed supply, if there were no NPIs in place to contain the virus. Conclusion Our results provide evidence for substantial spatial variation in (1) the effect of the pandemic on ICU demand, and (2) the potential and need for NPI adjustments at different stages of the pandemic. Forecasts with our spatial age-structured microsimulation model allow to take this spatial variation into account. The model is programmed in R and can be applied to other countries, provided that reliable data on the number of ICU patients infected with COVID-19 are available at sub-national level.

Using Differentiable Programming for Flexible Statistical Modeling (preprint)

Differentiable programming has recently received much interest as a paradigm that facilitates taking gradients of computer programs. While the corresponding flexible gradient-based optimization approaches so far have been used predominantly for deep learning or enriching the latter with modeling components, we want to demonstrate that they can also be useful for statistical modeling per se, e.g., for quick prototyping when classical maximum likelihood approaches are challenging or not feasible. In an application from a COVID-19 setting, we utilize differentiable programming to quickly build and optimize a flexible prediction model adapted to the data quality challenges at hand. Specifically, we develop a regression model, inspired by delay differential equations, that can bridge temporal gaps of observations in the central German registry of COVID-19 intensive care cases for predicting future demand. With this exemplary modeling challenge, we illustrate how differentiable programming can enable simple gradient-based optimization of the model by automatic differentiation. This allowed us to quickly prototype a model under time pressure that outperforms simpler benchmark models. We thus exemplify the potential of differentiable programming also outside deep learning applications, to provide more options for flexible applied statistical modeling.