Predicting local surges of COVID-19 hospitalizations in 2022

A recent study predicts whether coronavirus disease 2019 (COVID-19) hospitalization may exceed the local hospitalization capacity within a four or eight-week period – using simple decision rules and when no additional mitigating strategies are implemented during this time. The researchers in the study provided simple decision rules that are visual and easy to practice by local decision-makers without the need to use numerical computations.

Using real-time data related to the hospital occupancy and new hospitalizations that are associated with the COVID-19, and also the genomic surveillance of SARS-CoV-2, the researchers demonstrated reasonable accuracy, sensitivity, and specificity (all ≥80%) in predicting the local surges in hospitalizations. They showed the predictions under numerous simulated scenarios, capturing uncertainties over the future trajectories of COVID-19 during the winter and spring of 2022.

The research is published on the preprint server, medRxiv*.

Introduction

Low rates of vaccination, the emergence of novel variants of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2, the causative agent of COVID-19), and seasonal changes in the viral transmission cause increase in infection cases and populate the hospital wards. Surges in the ongoing COVID-19 pandemic have resulted in overburdening ofhealthcare support and strained hospital capacity.

The overall COVID-19 hospitalization behavior differs according to the different communities, the vaccination coverage there, the cumulative incidence of infection, and the extent of risk mitigations adopted. Understanding such trajectories is useful to policy-decision makers in strategizing and managing the situation.

Models such as COVID-19 Forecast Hub or the IHME COVID-19 Forecasting Model predict the trajectories of cases, hospitalizations, and deaths associated with COVID-19. However, presently there are only national-level and state-level predictive models of COVID-19 hospitalizations. Therefore, a tool to provide early warning of COVID-19 hospitalizations at the ‘local level’ is urgently needed.

Researchers can translate data from hospital occupancy censuses, as well as the rate of new  COVID-19 hospital admissions, and vaccination coverage, to monitor the local spread of SARS-CoV-2 and trends in COVID-19 hospitalizations.

Findings

In the present study, the researchers have identified simple and easy-to-communicate decision rules to provide early warnings – when a pre-specified threshold of hospital capacity will be likely exceeded within a four or eight-week period.

The COVID-19 trajectory in a local domain may be impacted by factors such as the proportion of the population with infection- or vaccine-induced immunity, the duration of infection- and vaccine-induced immunity, uptake and effectiveness of vaccine boosters, the transmissibility and virulence of novel variants (such as the omicron variant), the effectiveness of vaccines against prevalent strains including novel variants, and population behavior and adherence to mitigating strategies.

In their model, the researchers incorporated the complexities and uncertainties in the biology of the SARS-CoV-2 and also the above-mentioned diverse factors that may determine the course of COVID-19 during the winter and spring of 2022. With these, the researchers evaluated the predictive accuracy of the simple decision rules they have used to simulate the trajectory of COVID-19 hospitalizations and demands.

Observations related to COVID-19, such as, the timing and the effectiveness of mitigating strategies, the characteristics of novel variants, and the coverage of vaccination among different age groups will change over the coming months. Therefore, the researchers developed decision trees that are robust against changes in the data, including future uncertainties. Their simulation model is structured to incorporate all of those that impact the local size of the COVID-19 hospitalizations during the winter and spring of 2022.

Based on their analysis, they employed decision rules that use data on current hospital occupancy and the weekly rate of new hospitalizations due to COVID-19.

The Decision Tree A in this study uses surveillance data that is ‘related to the hospital occupancy, the weekly rate of new hospitalizations, and the vaccination coverage,’ – to predict whether the hospital occupancy due to COVID-19 would surpass the threshold of 15 per 100,000 population within the next eight weeks.

While the Decision Tree B uses data from ‘the genomic surveillance systems’ – to predict whether the hospital occupancy due to COVID-19 would surpass the threshold of 15 per 100,000 population within the next eight weeks.

Notably, a major challenge in this kind of study is that the true values of the parameters can only be observed or estimated with high uncertainty.

Despite this, the researchers observed that the structure of the decision trees strongly predicts the current hospital occupancy and weekly rate of new hospitalizations in short- and mid-term surges to determine the local trajectories of COVID-19.

Significance

This study identified simple, easy-to-communicate decision rules using surveillance data to alert local United States policymakers on the expected surge in COVID-19 related hospitalizations during the winter and spring of 2022, surpassing the local health care capacity within the next four or eight weeks, if no additional mitigating measures are implemented.

This prediction is important with factors such as seasonal changes, low vaccination coverage, and the emergence of new variants of SARS-CoV-2, such as the omicron variant. It would help the local policymakers to respond proactively in a prepared manner to mitigate future surges in COVID-19 hospitalization.

*Important notice

medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.

Written by

Dr. Ramya Dwivedi

Ramya has a Ph.D. in Biotechnology from the National Chemical Laboratories (CSIR-NCL), in Pune. Her work consisted of functionalizing nanoparticles with different molecules of biological interest, studying the reaction system and establishing useful applications.