The Hidden Engine of Chikungunya Outbreaks: How Silent Carriers Defy Mosquito Control

Even with mosquito populations slashed, chikungunya outbreaks can still surge—because silent carriers, showing no symptoms, keep the virus alive in the shadows. In 2025, China faced its largest recorded chikungunya outbreak: over 23,000 cases across Guangdong Province, with transmission persisting despite aggressive mosquito control. The standard public health playbook—clear stagnant water, kill adult mosquitoes, prevent bites—was being followed rigorously. Yet in Foshan, cases rebounded after initial decline. In Jiangmen, a second wave exploded weeks later. Why? A new modeling study reveals that asymptomatic infections, often missed by surveillance, are enough to reignite epidemics even when mosquito numbers are low. And critically, the researchers found that screening for these hidden cases isn’t just helpful—it’s game-changing. But not in the way anyone expected. There’s a sweet spot: moderate screening delays the peak of an outbreak more effectively than aggressive screening, buying crucial time for health systems to respond. This counterintuitive insight flips traditional logic on its head and points to a simple upgrade: a “Twelve-character” strategy that adds screening cases early to the existing mosquito-focused approach. For a world grappling with dengue, Zika, and future arboviruses, this could redefine how we stop outbreaks before they explode.

The Science

The team—Bo Zheng, Xingzhi Chen, Jianglong Xiao, Kai Wang, and Jianshe Yu from Guangzhou University, with Huaiping Zhu from York University—built a mathematical model to dissect why chikungunya persists despite mosquito control. Their framework is a vector–host dynamical system, tracking humans and mosquitoes through infection states. Humans are divided into susceptible (Sₕ), asymptomatic infected (Iₕₐ), symptomatic infected (Iₕₛ), and recovered (Rₕ). Mosquitoes are split into susceptible (Sₘ) and infected (Iₘ). Crucially, the model includes a screening rate, denoted by $q$, which represents the proportion of asymptomatic individuals identified and isolated before they can transmit the virus to mosquitoes.

Unlike malaria, where the Ross–Macdonald model long ago established that eliminating vectors breaks transmission, diseases like chikungunya and dengue involve a hidden reservoir: people who carry the virus but feel fine. Studies suggest 15% to 50% of chikungunya infections are asymptomatic, yet these individuals still have enough virus in their blood to infect mosquitoes that bite them

. The model captures this by allowing only unscreened asymptomatic humans—those missed by surveillance—to transmit to mosquitoes, via a transmission efficiency $\beta_{mh}$. The screening rate $q$ thus directly modulates how much hidden transmission persists.

The researchers analyzed three outbreak initiation scenarios: human-seeded (a few infected people arrive, mosquitoes not yet infected), mosquito-seeded (infected mosquitoes spark the outbreak), and joint seeding (both present at the start). They derived approximate analytical expressions for the epidemic peak value and peak time using singular perturbation theory, particularly when $q$ is close to 1 (near-perfect screening). This allowed them to bridge short-term outbreak dynamics—what happens in the first weeks—with long-term outcomes like final epidemic size.

Parameters were grounded in empirical data: $\beta_{hm}$ (mosquito-to-human transmission) ranges from 0.4 to 1.0, $\beta_{mh}$ (human-to-mosquito) from 0.14 to 0.85, recovery rates $\gamma_a$ and $\gamma_s$ between 1/14 and 1/7 per day (7–14 day illness), and mosquito lifespan $1/\mu_m$ between 7 and 30 days. The proportion of symptomatic cases $p$ was set between 50% and 85%, reflecting clinical variability.

What They Found

The simulations revealed three pivotal insights. First, asymptomatic carriers alone can sustain transmission even when mosquito populations are heavily suppressed. In a human-seeded outbreak, where only five asymptomatic individuals start the chain, the virus can still ignite a full-scale epidemic. This happens because each asymptomatic person can infect mosquitoes, which then bite others, creating a feedback loop. Even with aggressive vector control, if asymptomatic cases go undetected, the outbreak continues.

Second, and most surprisingly, the effect of screening on the timing of the epidemic peak is non-monotonic. Increasing $q$ does not simply delay the peak in a straight line. Instead, there’s a critical threshold—around $q = 0.57$—where the delay is maximized. Beyond this point, further screening actually shortens the time to peak.

Screening Rate vs. Peak Time of First Wave

Label	Value
q = 0	25.4
q = 0.5	38.2
q = 0.57	45.8
q = 0.9	36.2

Source: Simulations from Zheng et al. (2025)

This counterintuitive result arises from a trade-off. High screening ($q \approx 1$) quickly removes infectious individuals, reducing transmission and ending the outbreak sooner. But moderate screening ($q \approx 0.5$) slows transmission just enough to delay the peak, without eliminating so many cases that a large pool of susceptible people remains. That susceptible pool is key: if too many people escape infection in the first wave, they become fuel for a second wave when mosquitoes rebound—say, after seasonal rains or when control efforts lapse.

Third, moderate screening can suppress the first wave and prevent a devastating second wave. The model shows that with $q = 0$, the first wave peaks high and fast (day 25.4, 227 infections at peak), and nearly the entire population gets infected. With $q = 0.9$, the peak is lower (121) and later (day 36.2), but a large susceptible pool remains. When mosquito numbers rebound—simulated by a fourfold increase at day 60—a second wave erupts, peaking at 142 infections. But at $q = 0.57$, the first peak is delayed to day 45.8, and the second wave is dramatically smaller, if it occurs at all.

Screening Rate vs. Peak Infections in First Wave

Label	Value
q = 0	227.2
q = 0.5	189.4
q = 0.57	178.6
q = 0.9	121

Source: Simulations from Zheng et al. (2025)

The model also derived analytical approximations for peak time and value. For near-complete screening ($q = 1 - \epsilon$, $\epsilon$ small), the peak time is:

$$t_p(ϵ)=t_p^{(0)}-ϵ{\frac{d{I}_h^{(1)}}{dt}|}_{t=t_p^{(0)}}{\left({\frac{d^2{I}_h^{(0)}}{d{t}^2}|}_{t=t_p^{(0)}}\right)}^{-1}+O({ϵ}^2)$$

And the peak value:

$$I_{h,p}(ϵ)=I_h^{(0)}(t_p^{(0)})+ϵI_h^{(1)}(t_p^{(0)})+O({ϵ}^2)$$

These expressions, validated against numerical simulations with less than 1% relative error, confirm that screening doesn’t just reduce cases—it reshapes the entire epidemic curve in a predictable, quantifiable way.

Why This Changes Things

For decades, mosquito-borne disease control has focused almost entirely on the vector. The logic is sound: no mosquitoes, no transmission. But chikungunya, dengue, and Zika don’t play by malaria’s rules. They have a stealth mode. And that changes everything.

The 2025 Guangdong outbreak mirrors what’s happening globally. Dengue cases have increased eightfold over the last two decades; chikungunya has spread to over 100 countries. Climate change is expanding the range of Aedes mosquitoes. Urbanization creates ideal breeding grounds. And global travel means a single infected traveler can seed an outbreak thousands of miles away.

Yet surveillance systems are still built for symptoms. In Guangdong, asymptomatic cases weren’t even reported until October 2025—months after the outbreak began. That blind spot allowed silent transmission to continue unchecked.

This study shows that even with perfect mosquito control, asymptomatic carriers can sustain $R_0 > 1$, the threshold for sustained transmission. The basic reproduction number in their model is:

$$R_0=\sqrt{\frac{{\beta }_{hm}{\beta }_{mh}(1-q){\Lambda }_{m}p{N}_h}{{\mu }_m{\gamma }_s{N}_m}}$$

Notice the $ (1-q) $ term: screening directly reduces $R_0$. But more importantly, $R_0$ alone doesn’t tell the whole story. Two outbreaks with the same $R_0$ can have wildly different peak times and final sizes depending on $q$. That means public health decisions can’t rely solely on whether $R_0 < 1$. They must also consider when the peak hits.

The non-monotonic effect of screening is a paradigm shift. It means that maximizing screening isn’t always optimal. In resource-limited settings, chasing 90% screening might backfire—delaying the peak less than a 60% effort, while consuming disproportionate resources. The goal isn’t just to reduce transmission, but to pace it: slow the first wave enough to avoid overwhelming hospitals, but fast enough to build herd immunity and prevent a second surge.

This has direct implications for policy. The current “Nine-character” strategy—clear stagnant water, kill adult mosquitoes, prevent bites—is necessary but insufficient. The authors propose upgrading it to a “Twelve-character” strategy: adding screening cases early. This isn’t about mass testing everyone. It’s about targeted surveillance: testing in high-risk areas, monitoring febrile illness even when symptoms are mild, and using pooled sampling to detect virus in asymptomatic populations.

What’s Next

The model makes simplifying assumptions. It assumes symptomatic individuals are fully isolated—a best-case scenario. In reality, many symptomatic people continue to be bitten by mosquitoes before diagnosis, contributing to transmission. Relaxing this assumption would likely make asymptomatic screening even more critical.

The model also assumes constant mosquito recruitment ($\Lambda_m$). In reality, mosquito populations fluctuate with rainfall, temperature, and control efforts. Future versions could incorporate seasonality or stochasticity to better reflect real-world dynamics.

And while the study focuses on chikungunya, the implications extend to dengue and Zika, which share the same vectors and asymptomatic transmission pathways. A similar analysis for dengue—where up to 80% of infections are asymptomatic—could reveal even stronger effects.

The biggest challenge is operational: how to implement screening at scale. Rapid antigen tests for chikungunya are not yet widely available. But multiplex panels that detect dengue, chikungunya, and Zika simultaneously are in development. Pairing these with mobile clinics or school-based testing could make early detection feasible.

There’s also a behavioral dimension. If people believe they’re safe because mosquito spraying is happening, they may ignore personal protection or avoid testing. Public messaging must emphasize that both vector control and human surveillance are essential.

Finally, the concept of an optimal screening threshold opens a new frontier in epidemic control. It suggests that for many infectious diseases, the goal isn’t always to suppress transmission as much as possible, as fast as possible. Sometimes, the smarter strategy is to manage the pace of spread—buying time, protecting the vulnerable, and avoiding the boom-bust cycles that exhaust health systems.

In a world of emerging pathogens and climate-driven outbreaks, we can’t just react. We need strategies that are not only effective, but intelligent—ones that account for human behavior, ecological complexity, and the hidden dynamics of silent transmission. The “Twelve-character” strategy is a small phrase with a big idea: that stopping an outbreak isn’t just about killing mosquitoes. It’s about seeing the unseen.