When dengue cases among US travelers more than doubled their previous peak in 2024 — reaching 3,742 reported cases against a prior high of 1,474 in 2019 — the limitations of existing warning systems came into sharp focus. Many of the countries fueling that surge lacked robust local surveillance infrastructure, meaning official outbreak alerts arrived late, incompletely, or not at all. A new study published in the February 2026 issue of Emerging Infectious Diseases proposes a way to fill that gap by turning the travelers themselves into a sentinel surveillance network.
The study, conducted by Kristyna Rysava of the University of Warwick and colleagues at the US Centers for Disease Control and Prevention, demonstrates that routine data from ArboNET — CDC’s national arboviral surveillance system — can be used to detect periods of elevated dengue transmission in other countries before formal outbreak alerts are issued. By modeling historical trends in travel-associated dengue cases and applying a dual-criterion alert framework, the researchers identified sustained high-risk periods in multiple countries that went unflagged by existing CDC Travel Health Notices (THNs).
Modeling Travelers to Map Global Risk
The team analyzed 10,530 travel-associated dengue cases reported to ArboNET between January 2010 and April 2024, spanning exposure locations across 128 countries. Using country-specific regression models, they accounted for seasonal variation and differences in state-level reporting to establish expected monthly case counts. Countries were then classified as low, medium, or high risk depending on whether observed case counts exceeded modeled thresholds set at the 75th, 80th, or 90th percentile. To reduce false alerts — particularly in destinations with low traveler volume — the researchers added a case-count criterion requiring more than ten travel-associated cases within any three-month window. This dual-criteria approach cut the number of flagged country-months by nearly 53 percent compared to a threshold-only model, from 714 to 337, substantially improving specificity without sacrificing sensitivity for sustained transmission events.
The 80th percentile threshold paired with the case-count criterion emerged as the most operationally useful combination, balancing early detection against alert fatigue. Applied to real-world data, the framework identified extended high-risk periods across several major dengue source countries that were not covered by contemporaneous THNs. Brazil and Costa Rica each met both alert criteria for six to ten consecutive months in the 2023–2024 period, yet neither country received a corresponding THN during that window. Cuba presented a particularly striking case: traveler data revealed nearly continuous elevated transmission from late 2018 through early 2024 — a pattern largely invisible in official reporting — including a single-month peak of 241 US traveler cases in August 2022.
Reporting Delays Are a Known Obstacle — and a Manageable One
A critical practical question for any real-time surveillance tool is how well it performs with incomplete data. The study found that real-time ArboNET data missed a median of 25 percent of retrospective high-risk months across countries, with the median delay between symptom onset and case reporting standing at 1.3 months across the full study period. That lag was notably longer during 2020–2021, reaching 2.8 months, reflecting pandemic-era disruptions to health systems and surveillance infrastructure.
Despite these delays, the researchers found that real-time and retrospective data produced broadly similar outbreak signals, suggesting the framework is reasonably robust to moderate reporting lags. Where differences did emerge — particularly for Brazil, the Dominican Republic, Haiti, India, and the Philippines — real-time data tended to underestimate the duration of high-risk periods, breaking what were actually continuous outbreaks into apparent shorter episodes. The authors note that improvements in coordination among clinicians, laboratories, and state and federal public health agencies would further sharpen real-time performance.
What Traveler Data Can and Cannot Do
Traditional dengue THNs have historically depended on official outbreak declarations or granular national surveillance data — resources that are unevenly distributed globally and frequently delayed. Countries with limited diagnostic capacity or inconsistent reporting infrastructure may not generate the data needed to trigger formal alerts, leaving travelers and clinicians without timely guidance precisely where risk is highest. A traveler-based surveillance layer addresses this asymmetry directly: because US travelers seek care and are tested in a relatively consistent domestic healthcare environment, their case data can serve as a proxy signal for transmission intensity abroad, even when local systems are silent.
The framework is explicitly designed as a complement to, not a replacement for, in-country surveillance and official outbreak notifications. The authors acknowledge that traveler data cannot account for where within a country transmission is occurring, how travel volume fluctuates, or whether concentrated tourism in low-risk areas skews apparent risk downward. The approach is also less sensitive for destinations with low US traveler volume, such as Kenya and Singapore, where case counts are too sparse to generate reliable signals. Nevertheless, embedding automated traveler-based alerts into CDC’s THN review cycle — or integrating them into public health dashboards — could give decision-makers an additional, near-real-time data stream to cross-check against official reports, particularly for high-traffic dengue-endemic destinations in the Caribbean, Latin America, and South and Southeast Asia.
Sources and further reading:
Rysava K, Madewell ZJ, Thayer MB, et al. Using Routine Surveillance Data to Assess Dengue Virus Transmission Risk in Travelers Returning to the United States. Emerging Infectious Diseases, February 15, 2026.