A Cox-based multistate model for disease progression using large-scale population-based registry data: Application to COVID-19
This study demonstrates the effectiveness of a Cox-based multistate model applied to large-scale population registry data in analyzing COVID-19 disease progression, revealing how sociodemographic factors and vaccination influenced transition risks while highlighting evolving healthcare system dynamics over the pandemic.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
To understand how a disease moves through a population, scientists often look at more than just the moment a person gets sick or the moment they die. They are interested in the journey in between: the time spent in a hospital, the transfer to an intensive care unit, the recovery, or the tragic outcome. This approach, known as a multistate model, treats a disease not as a single event but as a series of steps or states that a person can move through over time. It allows researchers to see how different factors, such as age, health history, or social circumstances, influence the likelihood of moving from one step to the next. During the global pandemic caused by the SARS-CoV-2 virus, understanding these pathways became critical for hospitals and governments trying to manage resources and save lives. While many models existed to predict the spread of the virus, fewer were designed to track the complex, individual journeys of patients using real-world data from entire populations.
A team of researchers in Spain set out to map these journeys using a massive collection of health records from the Basque Country. They analyzed data from 377,285 adults who were diagnosed with the virus between March 2020 and January 2022. This period covered the entire early phase of the pandemic, including the strict lockdowns, the arrival of vaccines, and the emergence of new viral variants. Instead of looking at the patients as a single group, the researchers built a detailed map with eight distinct states. A person started at the moment of diagnosis. From there, they could move to a hospital ward, be transferred to an intensive care unit, be discharged back to a regular ward, or be discharged directly home. The map also included two final destinations: dying while still in the hospital or dying after leaving the hospital. By tracking 377,000 individuals through these eleven possible transitions, the team could calculate the exact probability of a patient moving from one state to another and see how that probability changed based on who the patient was and when they got sick.
The results revealed a clear picture of who was most vulnerable and how the healthcare system evolved. The data showed that men, older adults, people living in nursing homes, and those with significant social deprivation faced a much higher risk of being hospitalized and dying. The number of chronic health conditions a person had also played a major role; those with more underlying illnesses were far more likely to end up in the hospital or die. Interestingly, the study found that people taking many different medications at the start of the pandemic were more likely to be hospitalized, yet they were less likely to die once they were there, suggesting that their medical teams were already closely monitoring them. Conversely, being vaccinated acted as a powerful shield. As the vaccination campaign rolled out, the risk of hospitalization dropped significantly, and vaccinated patients who did get sick were much more likely to recover and go home without needing intensive care.
The researchers also observed that the nature of the pandemic changed over time. In the very first period, when the virus was new and hospitals were overwhelmed, the risk of dying was high for everyone. However, as time passed and the healthcare system adapted, the pattern shifted. Fewer people were being admitted to the hospital overall, but those who did get admitted were more likely to require intensive care. This suggests that the system became better at keeping mild cases out of the hospital, reserving its most critical resources for the sickest patients, who were then more likely to survive. The study also highlighted that the risk of dying after leaving the hospital was higher in the early days of the pandemic than in later periods, indicating that post-discharge care and monitoring improved as the crisis continued.
By using this dynamic map of patient journeys, the researchers demonstrated that it is possible to turn a massive, complex database of real-world records into a clear story about disease progression. They showed that factors like social inequality and living in a nursing home were just as important as age or medical history in determining a patient's fate. The study did not just count deaths; it explained the path that led to them. This approach offers a way for public health officials to see the full picture of an epidemic, helping them decide where to send resources and how to protect the most vulnerable. The methods used in this research are not limited to the pandemic; they can be applied to other diseases that move through stages, providing a robust tool for understanding how illnesses affect populations and how healthcare systems respond to them.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.