Dengue fever is entering a new era of clinical risk assessment as researchers seek to identify, as early as possible, which patients are likely to deteriorate. A study by L. Phung Khanh, K.D. Rosenberger, T. Dong Thi Hoai and colleagues, published in Nature Communications in 2026, examines how data from the prospective, multicentre IDAMS study can improve the prediction of adverse dengue outcomes. The work addresses one of the most persistent challenges in dengue medicine: most infections resolve without major complications, but a small and difficult-to-recognise proportion can progress rapidly to shock, severe bleeding, organ dysfunction or death.
Dengue is caused by four closely related viruses known as dengue virus serotypes 1 through 4, all transmitted primarily by Aedes aegypti and Aedes albopictus mosquitoes. Infection commonly begins with fever, headache, muscle and joint pain, nausea and rash. The period when the fever subsides is particularly dangerous, however, because vascular leakage may intensify even as the patient appears to be improving. Fluid can move from the bloodstream into surrounding tissues, reducing circulating blood volume and placing the heart, kidneys and other organs under severe stress. In this narrow clinical window, timely recognition of risk is essential, yet early symptoms often overlap with those of uncomplicated dengue.
The IDAMS, or International Research Consortium on Dengue Risk Assessment, prospective multicentre study was designed to collect clinical information from patients in different settings and follow them over time. Prospective studies are especially valuable for prediction research because information is recorded before the final outcome is known. This allows investigators to test whether early observations—rather than hindsight—can distinguish patients who will remain stable from those who will develop complications. A multicentre design also helps address a major limitation of many clinical prediction tools: a model developed in one hospital or country may perform less reliably when applied to populations with different ages, viral serotypes, coexisting illnesses, healthcare access and patterns of presentation.
The new analysis focuses on improving early prediction for adverse outcomes, a task that requires more than simply identifying patients who are already critically ill. A clinically useful model must operate before severe deterioration becomes obvious, using information available during the first assessment or early follow-up. Potentially informative variables in dengue research include age, duration of fever, warning symptoms, blood pressure, pulse pressure, capillary refill, fluid intake and output, bleeding signs and laboratory measurements such as platelet count, haematocrit and white blood cell levels. The scientific challenge is to determine which combinations carry genuine predictive value, how their importance changes over the course of illness and how to avoid confusing common dengue features with specific signals of danger.
Platelet counts illustrate the difficulty. A falling platelet level is widely associated with dengue progression, but thrombocytopenia alone does not reliably identify patients who will develop shock or organ failure. Haematocrit can provide information about haemoconcentration and plasma leakage, yet its interpretation depends on a patient’s baseline level, hydration status and any bleeding that may be occurring. Similarly, fever duration and the appearance of abdominal pain or persistent vomiting may be informative, but their predictive meaning can vary between patients. Robust risk models must therefore integrate multiple clinical and laboratory signals instead of relying on a single laboratory threshold.
The study’s importance lies in its effort to refine that integrated approach using data gathered across multiple centres. Prediction models can be expressed in several ways, including statistical risk scores, regression-based algorithms or more complex machine-learning systems. Regardless of the method, performance must be assessed carefully. Researchers typically examine discrimination—how well a model separates higher-risk from lower-risk patients—and calibration, which measures whether predicted probabilities correspond to outcomes observed in practice. A model that ranks patients correctly but consistently overestimates risk may lead to unnecessary admissions, while one that underestimates danger could delay monitoring and treatment.
For dengue, such distinctions have immediate implications for health systems. During seasonal epidemics, hospitals may receive more patients than they can safely observe. Over-triage can consume beds, staff time and laboratory capacity, while under-triage can leave vulnerable patients without sufficient monitoring during the critical phase of illness. An improved early-warning system could help clinicians decide who requires frequent reassessment, intravenous-fluid monitoring, laboratory follow-up or transfer to a higher level of care. It could also support safer outpatient management for patients whose early profiles indicate a lower probability of severe complications, provided that they receive clear instructions and rapid access to care if symptoms change.
The use of data from a prospective multicentre cohort is also relevant to the broader movement toward transportable and equitable clinical prediction. A model must be tested not only on the population from which it was derived but also across different demographic and epidemiological contexts. Dengue severity can be influenced by previous infection with another serotype, immune status, age and underlying conditions, including pregnancy and chronic cardiovascular, renal or metabolic disease. Geographic variation may affect the timing of presentation and the prevalence of circulating serotypes. By drawing on multicentre evidence, the IDAMS analysis seeks to make early prediction more representative of the patients encountered in real-world dengue care rather than narrowly optimised for a single institution.
The findings are expected to contribute to a more precise understanding of how early clinical data can be converted into actionable risk information. Their value will ultimately depend on external validation, ease of use and integration into routine care. A prediction tool that requires tests unavailable in rural clinics, or that produces a probability without clear guidance for clinicians, may have limited practical impact. Conversely, a transparent model based on readily obtainable observations could be incorporated into triage protocols, electronic medical records and outbreak-response systems. As dengue expands into new regions under the influence of urbanisation, travel and changing climate conditions, reliable early assessment may become as important as antiviral development in reducing preventable complications.
The IDAMS study therefore represents a step toward treating dengue not as a single uniform illness but as a dynamic infection whose risks change from day to day. By improving the ability to recognise patients on a dangerous trajectory before severe disease is fully established, the research could strengthen clinical surveillance and help direct limited resources to those most likely to need them. The work also highlights a central principle of viral medicine: better outcomes often depend not only on discovering new therapies, but on extracting more precise information from the earliest stages of infection.
Subject of Research: Early prediction of adverse outcomes in dengue using prospective multicentre clinical data
Article Title: Improving early prediction for adverse dengue outcomes using data from the IDAMS prospective multicentre study
Article References: Phung Khanh, L., Rosenberger, K.D., Dong Thi Hoai, T. et al. Improving early prediction for adverse dengue outcomes using data from the IDAMS prospective multicentre study. Nature Communications (2026). https://doi.org/10.1038/s41467-026-76450-2
Image Credits: AI Generated
DOI: 10.1038/s41467-026-76450-2
Keywords: Dengue, dengue virus, severe dengue, early prediction, adverse outcomes, IDAMS, prospective multicentre study, clinical risk assessment, viral disease, infectious disease medicine
