Dengue spatiotemporal patterns in Minas Gerais, Brazil, 2014-2023: regional epidemic forces dominate over the environmental impact of the Brumadinho dam collapse

This study concludes that while dengue incidence in Minas Gerais, Brazil, exhibited strong spatiotemporal synchrony driven by regional epidemic forces between 2014 and 2023, the 2019 Brumadinho dam collapse had no significant independent effect on dengue transmission patterns in the affected Paraopeba River basin.

The Gist

The Big Picture: A Flood vs. A Wave

Imagine the state of Minas Gerais in Brazil as a giant, bustling city. For years, this city has been dealing with a recurring problem: Dengue fever, a mosquito-borne illness. Think of Dengue outbreaks like massive ocean waves that crash over the whole city at the same time, hitting everyone regardless of where they live.

In January 2019, a specific disaster happened in one part of the city: the Brumadinho dam collapsed. This released a massive amount of mud and waste into the local river. Scientists wondered: Did this local mudslide create a new, separate "tsunami" of Dengue specifically in the towns right next to the river, or did the regular "ocean waves" of Dengue just wash over them like everyone else?

This study set out to answer that question by looking at data from 2014 to 2023.

How They Investigated (The Detective Work)

The researchers acted like detectives comparing two groups of towns:

1. The "Mud" Group: Towns directly in the path of the dam collapse (the Paraopeba River basin).
2. The "Control" Group: Towns nearby that were in the same health regions but not hit by the mud.

They looked at the number of Dengue cases in these towns year by year. To make a fair comparison, they didn't just count the raw numbers (because some towns are bigger than others); they looked at the pattern of the outbreaks. It's like comparing the rhythm of music rather than just the volume.

What They Found

The results were surprisingly clear: The dam collapse did not create a unique Dengue pattern.

Here is the breakdown of their findings using simple metaphors:

• The "Regional Rhythm" is King: The biggest factor determining when Dengue hits is the Health Region (a large administrative area), not the specific river. It's like how a whole neighborhood might hear a siren at the same time because of a city-wide event, rather than because one house had a fire. The towns near the dam and the towns far away all danced to the same beat.
• The "Big Waves" Drowned Out the "Local Ripple": The study found huge spikes in Dengue in 2016, 2019, and 2023 across the entire state. The 2019 spike happened right when the dam collapsed, but it was a massive state-wide event, not a local one caused by the mud.
  • The Analogy: Imagine you are trying to hear a whisper (the effect of the dam) while a rock concert is playing next door (the state-wide Dengue epidemic). The whisper is impossible to hear because the concert is so loud. The researchers found that the "concert" (the state-wide epidemic) was the only thing driving the numbers.
• No Special "Mud Zone": When they grouped the towns based on their Dengue patterns, the towns near the dam did not cluster together. They mixed right in with the other towns. There was no special "Dengue signature" unique to the disaster area.

Why Didn't the Dam Change Things?

You might think, "But the dam ruined the environment! Shouldn't that make more mosquitoes?"

The researchers suggest a few reasons why this didn't show up in the data:

1. The "State-Wave" was too strong: The national epidemic in 2019 was so huge that it likely overwhelmed any small, local changes the dam might have caused.
2. The Terrain: Many of the towns near the river are rural and spread out. Dengue usually explodes in crowded cities where mosquitoes and people are packed together. In these rural areas, the "crowd" wasn't dense enough for the dam's environmental damage to create a massive local outbreak.
3. The Data Limitations: The study relied on official reports of sick people. Sometimes, when a disaster happens, hospitals get overwhelmed or people can't get to them, so cases go unreported. This might have hidden a small local effect, but based on the data they did have, no effect was visible.

The Bottom Line

The study concludes that Dengue in Minas Gerais is driven by big, state-wide forces (like weather patterns and virus cycles), not by local environmental disasters like the Brumadinho dam collapse.

While the dam was a tragedy that caused massive environmental damage, it did not leave a detectable mark on the timing or pattern of Dengue outbreaks in the affected towns. The "waves" of Dengue that hit those towns were the same waves that hit the rest of the state, driven by factors much larger than the river valley itself.

Technical Summary

Technical Summary: Regional Drivers of Dengue in Minas Gerais (2014–2023)

Problem Statement
Dengue fever represents a critical public health challenge in Brazil, with the state of Minas Gerais bearing a significant burden of cases. The transmission dynamics of dengue are complex, driven by interactions between climatic, environmental, demographic, and socioeconomic factors. In January 2019, the catastrophic collapse of the Brumadinho dam released approximately 12 million cubic meters of iron ore tailings into the Paraopeba River basin. This event caused substantial environmental disturbance, potentially altering aquatic breeding habitats for Aedes aegypti and disrupting water supply systems, which could plausibly modify local dengue transmission dynamics. However, distinguishing a localized epidemiological signal from such a disaster against the backdrop of concurrent large-scale national outbreaks (notably in 2019 and 2023) and systemic surveillance limitations remains an empirical challenge. This study investigates whether the Brumadinho dam collapse induced a detectable, independent change in the spatiotemporal patterns of dengue in affected municipalities compared to regional controls.

Methodology
The authors conducted an ecological spatiotemporal analysis of dengue notifications in Minas Gerais from 2014 to 2023 using data from the Brazilian Notifiable Diseases Information System (SINAN).

• Study Population: The analysis included 67 municipalities: 26 designated as "Paraopeba basin" (directly/indirectly affected by the dam collapse) and 41 "regional controls" (unaffected municipalities within the same administrative health regions). A third group of state-level controls was used for descriptive context.
• Data Processing: Annual incidence rates were calculated per 1,000 inhabitants and standardized using municipality-specific Z-scores to compare temporal patterns independent of absolute magnitude.
• Analytical Approach:
  • Pattern Recognition: Temporal similarity was assessed using Pearson correlation-based hierarchical clustering and non-metric multidimensional scaling (NMDS).
  • Multivariate Analysis: Permutational multivariate analysis of variance (PERMANOVA) and Principal Component Analysis (PCA) were employed to quantify sources of variation (e.g., health region vs. basin status) and identify years driving variability.
  • Statistical Modeling: A linear mixed-effects model with municipality as a random effect was fitted to test for changes in incidence after 2019. The model included year, Paraopeba basin status, health region, and their interactions as fixed effects. Pre/post contrasts were estimated using marginal means.

Key Results

• Temporal Synchrony: Dengue incidence exhibited strong temporal synchrony across the state, with major epidemic peaks occurring simultaneously in 2015–2016, 2019, and 2023. These peaks were observed across all municipalities regardless of exposure status.
• Drivers of Variation: Health region explained 31.5% of the variation in temporal incidence profiles ($p = 0.001$), whereas Paraopeba basin status explained no significant variation ($p = 0.998$). Hierarchical clustering identified five distinct groups of municipalities, none of which were significantly enriched for Paraopeba basin municipalities after multiple-testing correction.
• Multivariate Drivers: PCA indicated that interannual variability was primarily driven by epidemic years, specifically 2023, 2019, and 2016.
• Impact of the Disaster: In the linear mixed-effects model, the year was a significant predictor ($p < 0.001$), but Paraopeba basin status and its interaction with time were not.
  • Incidence increased significantly after 2019 in non-exposed (regional control) municipalities ($p < 0.001$).
  • No statistically significant change was detected in the Paraopeba basin municipalities post-2019 ($p = 0.088$).
  • The interaction between Year and Paraopeba basin status was non-significant ($p = 0.197$).

Key Contributions
The study provides a robust, comparative assessment of the epidemiological impact of a major environmental disaster on vector-borne disease transmission. By utilizing a mixed-effects framework that controls for health-region-level determinants and employs a contemporaneous control group, the authors isolate the specific signal of the dam collapse from broader state-wide epidemic forces. The findings challenge the hypothesis that the Brumadinho disaster independently altered local dengue dynamics, suggesting instead that regional epidemic processes dominate over local environmental perturbations in this context.

Significance and Claims
The authors conclude that dengue dynamics in Minas Gerais during the study period were driven primarily by regional and state-wide epidemic processes, with no statistically significant independent effect attributable to the Brumadinho dam collapse on notified dengue patterns.

• Dominance of Regional Drivers: The results reinforce that health regions serve as meaningful epidemiological units where shared determinants (climate, vector ecology, population connectivity) modulate dengue risk more strongly than localized environmental disruptions.
• Confounding by Large-Scale Outbreaks: The study highlights the difficulty of detecting localized disaster effects when they coincide with massive, synchronous national outbreaks (e.g., the 2019 and 2023 epidemics). The temporal overlap between the disaster and a major national surge acts as a critical confounder.
• Public Health Implications: The findings underscore the necessity of using robust comparative designs to assess the consequences of environmental disasters. They suggest that while localized effects on vector habitats may exist, they may be too fine-grained to be detected at the annual, municipality-level resolution against a background of large-scale outbreaks. The authors note that limitations such as passive surveillance underreporting (exacerbated by the COVID-19 pandemic) and the lack of entomological data may mask subtle local effects, warranting future studies with higher spatial and temporal resolution.