Early and accurate prediction of COVID-19 hospitalization risk - methodological framework for pandemic preparedness

When a virus like SARS-CoV-2 spreads through a population, the most urgent question for doctors and public health officials is not just who is infected, but who will become dangerously ill. In the early days of the pandemic, hospitals were overwhelmed because there was no reliable way to tell which patients would need intensive care and which would recover at home. Current guidelines often rely on broad factors like a person's age or body weight to decide who gets early treatment, but these clues are imperfect. A better approach would be to look for specific signals inside the body that appear very soon after infection, acting as an early warning system. These signals, known as biomarkers, are measurable substances in the blood that reflect how the immune system is reacting to the virus. If doctors could identify these signals quickly, they could triage patients more effectively, ensuring that those at highest risk receive life-saving care immediately while others are monitored safely at home.

A team of researchers from Germany and France set out to find these early signals by studying a group of two hundred and one people who had never been vaccinated against the virus and had no prior infection. The scientists recruited these individuals either on the day they tested positive for SARS-CoV-2 or within the first week after diagnosis. Over several weeks, the team took blood samples from each person to measure a wide range of chemical markers, including standard blood test results and specific immune proteins called cytokines. They tracked these levels as the patients' conditions evolved, comparing the blood profiles of those who eventually required hospitalization against those who stayed at home. The goal was to see if a specific combination of these blood markers could predict the course of the disease before it became severe.

The study found that looking at a person's age or body mass index alone was not a strong enough predictor to separate those who would need hospital care from those who would not. While older age and higher body weight were indeed more common in the hospitalized group, the blood chemistry provided a much clearer picture. The researchers discovered that two specific markers stood out: a protein called lactate dehydrogenase, which indicates cell stress, and a signaling protein called interleukin-6, which is part of the body's inflammatory response. When either of these markers was found at high levels in the blood shortly after diagnosis, it was a powerful sign that the patient was at risk of needing hospitalization. In fact, the combination of these two markers was so accurate that it correctly identified every single patient who eventually went to the hospital, while also correctly identifying nearly all those who did not.

Beyond predicting who would need a hospital bed, the team also wanted to understand how to distinguish between people who would have mild symptoms and those who would develop a moderate illness, even if they did not end up in the hospital. For this, they focused on a different set of immune signals. They found that a specific pattern involving four different immune proteins—interferon-alpha, interferon-gamma, interleukin-6, and interleukin-8—could reliably separate the mild cases from the moderate ones. This pattern worked well for patients who were tested within the first six days of their symptoms. The researchers noted that these findings were consistent regardless of the specific variant of the virus the patient had, including the Omicron variant, and the predictive power held true even when accounting for factors like gender or whether the patient had diabetes.

The researchers also tracked how these markers changed over time. They observed that in patients who ended up in the hospital, the levels of stress and inflammation markers remained high for weeks, whereas in those with milder cases, these levels dropped off quickly. This suggests that the body's initial reaction to the virus is a critical window for intervention. If a patient shows high levels of these specific markers right after diagnosis, it indicates a trajectory toward severe disease that might be halted with early treatment. The study concludes that by measuring just a few of these blood markers at the point of care, doctors could make highly accurate predictions about a patient's future health needs. This approach offers a way to move beyond guesswork and demographic assumptions, providing a concrete, biological basis for deciding who needs urgent medical attention during a pandemic.