Study: Modeling COVID-19 care capacity in a major health system. Image Credit: Eakdesign/Shutterstock

Modeling COVID-19 hospital capacity


The coronavirus disease 2019 (COVID-19) pandemic has represented a major burden on global health services. At the start of the pandemic, very few treatments were known, increasing the burden even further. This led many countries to enact harsh restrictions to lower transmission rates, such as mandatory face masks in public, the closing of office buildings and non-essential businesses, and even full lockdowns/stay-at-home orders. As cases rise again, researchers from Yale School Of Public Health have revealed a model for predicted hospital occupancy during the pandemic.


Study: Modeling COVID-19 care capacity in a major health system. Image Credit: Eakdesign/ShutterstockStudy: Modeling COVID-19 care capacity in a major health system. Image Credit: Eakdesign/Shutterstock


A preprint version of the group’s study is available on the medRxiv* server while the article undergoes peer review.


The study


The researchers described the flow of patients infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) through a hospital system using a system of differential equations. They modeled transitions between eight different compartments, splitting the hospital into multiple areas.


The immediate entryway, where triage occurs was assigned as P, floor beds were assigned as F, ICU beds as C, any individual discharged from the emergency department with mild symptoms was in state MS, patients recovering post-discharge from the hospital were in state R, any patients in a queue for floor beds when they were not available were in state WF, patients in the same situation but for ICU beds were in state WC, and the final state was death.


The researchers designed the model to respond to scenarios entered by a user. The user must select the number of COVID-19 positive patients presenting to the health system during a certain time frame, the number of COVID-19 positive patients presenting on day 0, and the expected number of presentations during the time of the simulation. For the type of increase in patients, users can choose exponential, linear, saturated, and no increase.


The users can also specify the number of available beds, possible policy responses, and how this could be implemented. They can modify the model by adjusting parameters to more accurately reflect the patient population in the aforementioned individual areas and tailor the model more accurately to the healthcare facility. The user can change age distribution, the average length of stay of patients, and the likelihood of death in certain areas of the hospital.


The model outputs information to help inform the healthcare providers. It provides information on the number of days to overflow, how many extra beds will be needed for COVID-19 patients, the number of deaths likely in each hospital area, and the predicted fatality rate.


To collect the data they needed to calibrate the model, the researchers extracted information from the Yale-New Haven Hospital System, consisting of five hospitals, between March and July 2020. They examined individual-level records of both COVID-19 and regular patients, as well as hospital-level summaries of capacity. This data was used to reconstruct the trajectory of patients through the hospitals and the census of patients in floor and ICU beds.


Survival analysis with competing risks was used to estimate parameters describing rates of departure from the emergency department, from there to discharge, admission to floor beds, and admission to the ICU. Inpatient data was used to estimate the rate of transition from the floor to the ICU, from the floor to discharge, from ICU to the floor, and deaths in both the ICU and on the floor. These parameters were estimated differently for three different age groups.


To evaluate the success of their model, it was examined against the largest of the five hospitals examined. The researchers input the conditions the hospital saw on March 8th and estimated parameters using gamma-distributed time to depart from a department. They found that the model accurately predicted occupancy, with the ICU occupancy closely following observed occupancy in the following months, including the peak of infections and the decrease in occupancy. The model also showed reasonable accuracy in predicting floor occupancy and deaths.


Conclusion


The authors highlight the value of their model in helping to inform healthcare providers and administrators. While the floor occupancy was slightly underpredicted, ICU occupancy was much closer to fact. ICU occupancy is the most important factor for severely ill COVID-19 patients and is often the limiting factor in treating them.


While in many developed countries, mass vaccination schemes have shown enormous success, more and more vaccine breakthrough infections are being reported, to the extent that some European countries are re-introducing lockdowns. This model could help to save lives by accurately predicting occupancy in hospitals.


*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



https://www.news-medical.net/news/20211123/Modeling-COVID-19-hospital-capacity.aspx