The study, published in Frontiers in Epidemiology, exploited a natural experiment that few countries can offer. Ecuador packs Pacific coastal lowlands, Amazonian plains, inter-Andean valleys between 1,500 and 3,000 metres, and high-altitude paramo above 3,500 metres into a small area, allowing researchers to track disease across a dramatic altitudinal gradient. The team, led by Jaime Angamarca-Iguago and Daniel Simancas-Racines of the Universidad Tecnológica Indoamérica, drew on Ecuador’s national hospital discharge registry administered by the National Institute of Statistics and Censuses (INEC), identifying dengue admissions through the ICD-10 codes A90, A91 and A97.

The headline numbers are striking. The national dengue hospitalisation rate rose from 29.7 per 100,000 people in 2011 to 77.0 per 100,000 in 2024, the highest level of the entire study period and well above the previous epidemic peak of 64.4 per 100,000 recorded in 2012. In Andean cantons above 1,500 metres, the rate climbed roughly seven-fold, from 1.9 to 12.8 per 100,000. The steepest rise occurred in the low Andean band between 1,500 and 2,199 metres, where hospitalisation rates increased more than ten-fold, from 9.4 to 95.7 per 100,000 person-years.

Dengue hospitalisations were documented in 68 distinct cantons above 1,500 metres, including one at 3,212 metres above sea level in Colta, Chimborazo province, first recorded in 2013. Highland provincial capitals including Quito at 2,850 metres, Cuenca at 2,560 metres, Riobamba at 2,754 metres, Ambato at 2,577 metres and Ibarra at 2,225 metres all logged their first dengue hospitalisations in 2011, coinciding with a national epidemic peak. Notably, the number of cantons reporting at least one dengue hospitalisation grew only modestly, from 133 in 2011 to 145 in 2024, indicating that the rising burden reflects intensifying transmission within affected areas rather than simple geographic spread.

To link these hospitalisations to climate, the researchers combined the hospital records with satellite-derived environmental data from NASA’s Earth Observing System, using monthly land surface temperatures from the MODIS instruments, precipitation estimates from the Global Precipitation Measurement mission, and vegetation indices to characterise each canton’s conditions. The centrepiece of the analysis was a distributed lag non-linear model, a statistical framework that captures how an exposure such as temperature affects an outcome both immediately and over a delayed window. The team set a maximum lag of three months, chosen to span the roughly 8 to 12 days the dengue virus needs to develop inside the Aedes aegypti mosquito plus the 4 to 10 days from symptom onset to hospitalisation.

The model identified a minimum hospitalisation temperature of 18.4 degrees Celsius nationally and 18.9 degrees Celsius in Andean cantons, values closely consistent with the experimentally derived lower thermal limit of roughly 18 degrees for sustained Aedes aegypti transmission. Below that threshold, the mosquito’s development, its vector competence and the speed at which the virus completes its extrinsic incubation period all falter. Until the early 2010s, scientists assumed this thermal floor confined the vector to elevations below about 1,500 to 1,700 metres, but field surveys have since documented Aedes aegypti at 2,130 metres in Mexico, up to 2,000 metres in central Nepal and 2,100 metres in Colombian municipalities.

The study’s most distinctive contribution, however, was its test of whether temperature anomalies, rather than absolute temperatures, drive highland dengue. For each canton and calendar month, the researchers calculated how much the observed temperature deviated from that canton’s own 2007 to 2015 baseline, then fitted a quasi-Poisson regression restricted to Andean cantons with fixed effects for canton, calendar month and year. This design is statistically powerful: by absorbing every time-invariant characteristic of each canton, from urbanisation to diagnostic capacity, the model isolates the effect of temperature swings within each place over time. A one-degree positive anomaly was associated with a 1.07-fold increase in hospitalisation rates, with a 95 percent confidence interval of 1.02 to 1.13 and a p-value of 0.012.

That estimate comes with important caveats the authors themselves emphasise. Without canton fixed effects, the same contrast yielded a larger relative risk of 1.21, meaning roughly two-thirds of the crude association reflects systematic differences between cantons rather than within-canton variation. When the 2024 epidemic year, which coincided with an intense El Niño warm phase and a record dengue epidemic across the Americas, was excluded, the association weakened to 1.04, hovering at the margin of conventional significance. The authors interpret this pattern as evidence that Andean dengue expansion may be paced by episodic interannual climatic shocks superimposed on slower chronic warming, though they stress this remains a hypothesis, since no El Niño-Southern Oscillation index was directly included in the models.

The biological logic behind the anomaly finding is coherent. In coastal cantons, temperatures already exceed the thermal optimum for Aedes aegypti, so small year-to-year deviations matter little and extreme warming may even suppress transmission. In Andean cantons, by contrast, baseline temperatures hover near or below the lower threshold for sustained transmission, so modest positive anomalies of one to two degrees can push monthly conditions across the critical line, shortening the virus’s extrinsic incubation period and enabling adult mosquito survival. Laboratory work has shown that even modest temperature fluctuations around a low mean accelerate dengue transmission, and the same mechanism has been implicated in the climate-driven upslope expansion of malaria in the Ethiopian highlands.

The study also flagged a clinical warning sign: in-hospital case fatality in low Andean cantons was 0.62 percent, nearly twice the coastal rate of 0.33 percent, though based on only five deaths and not statistically significant. The authors suggest this may reflect delayed clinical recognition or the absence of prior population immunity in newly affected zones, and they argue that training highland clinicians in dengue recognition and management is a low-regret measure regardless. They also caution that hospital discharge records identify patients by residence rather than place of infection, that some highland cases may be travel-associated, and that the ecological design cannot establish individual-level causation. Still, the researchers argue their approach, built entirely on freely available satellite products and public administrative data, offers a reproducible template for other Andean countries facing the same quiet upward march of a disease once thought to belong only to the lowlands.

Subject of Research: The association between local thermal anomalies and the altitudinal rise of dengue hospitalisations in Ecuador

Article Title: Local thermal anomalies and the altitudinal rise of dengue hospitalisation rates in Ecuador, 2011–2024: a national ecological study

Article References: Local thermal anomalies and the altitudinal rise of dengue hospitalisation rates in Ecuador, 2011–2024: a national ecological study. (n.d.). https://doi.org/10.3389/fepid.2026.1906738

Image Credits: AI Generated

DOI: 10.3389/fepid.2026.1906738

Keywords: dengue, Ecuador, Andes, climate change, Aedes aegypti, temperature anomaly, hospitalisations, vector-borne disease, distributed lag non-linear model, altitudinal expansion, public health, epidemiology