Write a comment

Summary:

Couper et al. studied the risk of dengue transmission in California, a non-endemic region that saw its first local cases in 2023–2024. This research is vital for public health, as climate change and globalization drive mosquito-borne viruses into temperate areas.

Quantifying risk is challenging because transmission relies on local environmental conditions and the importation of the virus by infected travelers. To address this, the authors developed a sophisticated risk model integrating three key components: vector presence, temperature suitability, and viral introductions.

The analysis estimates that 46% of California residents (18.2 million people) currently live in areas where peak monthly risk exceeds this threshold, mainly in the Central Valley and Southern California metropolitan areas. Under moderate future projections, an additional 4.1 million residents may be at risk by mid-century, with the largest increase in September.

Significantly, the study identified different regional limitations to transmission. They found that climate suitability limits the Bay Area, while viral introductions limit the Central Valley. This distinction informs targeted surveillance and control strategies for effectively managing the growing dengue threat.

Strengths:

This study’s strengths lie in its integrative, multi-factor approach, utilizing climate data, vector presence, and human mobility to provide a comprehensive assessment of dengue transmission risk rather than relying on a single factor. Researchers also utilize high-resolution data, allowing for precise identification of risk hotspots and seasonal variations, which enhances targeted intervention planning.

The model's ability to distinguish locations and periods of actual reported cases from controls demonstrates its reliability and predictive validity. Focusing on the US, especially areas with emerging or reemerging dengue risk, fills a critical gap where limited existing modeling efforts have been made in similar contexts. By identifying specific hotspots and seasonal windows of elevated risk, the framework supports proactive vector control, surveillance, and public health strategies.

Major Revisions:

1. The modeling of travel-associated cases relies on assumptions about travel volume based on demographics and does not account for regional seasonality or potential changes over time. Clarifying the impact of these limitations in the discussion, and possibly exploring sensitivity analyses, would improve the robustness of projections.

2. Furthermore, while multiplicative integration of three different risk components as defined in the Methods and Figure 1, requires further theoretical justification as no explorations of alternative methods or sensitivity analyses were provided. The multiplicative form assumes proportional effects across full ranges yet dengue transmission exhibits threshold dynamics where minimum vector densities, temperature bounds and importation rates interact non-linearly. The authors should provide mechanistic justification for multiplicative form by deriving it from or relating it to established vector-borne disease transmission models or test alternative functional modeling forms and include them in the methods of the paper.

3. Using monthly averaged temperatures (mean of maximum and minimum) may introduce systematic bias by overlooking daily thermal fluctuations and nonlinear trait responses. For traits with accelerating responses near thermal limits, monthly averaging may overestimate suitability in inland regions frequently exceeding 34-35°C while underestimating it in coastal areas with moderate but variable temperatures. Given that mosquito generation times and viral incubation periods span 7-15 days, within-month variation could substantially affect transmission probability. Authors should conduct an analysis using daily temperature data, calculate the R0(T) separately and then average the resulting values to correct for nonlinearities. They could also conduct a sensitivity analysis and report this in their results as possible error.

4. In the methods section, authors state that future projections rely exclusively on the MIROC climate model from CMIP5, severely understating structural uncertainty in regional climate projections.The Methods provide no justification for MIROC selection nor assessment of its California climate biases, particularly for summer temperatures and precipitation seasonality critical for mosquito life history. The reported confidence intervals (e.g., 3.7-4.6 million additional at mid-century) reflect only bootstrap threshold uncertainty, not climate model disagreement. Authors should btain climate projections from a wider range of CMIP5 models, calculate ensemble mean and inter-model spread for temperature and precipitation, projecting vector presence and R0(T) for each model separately, reporting this change in the methods and results sections.

Minor Revisions:

1. The manuscript would benefit from standardized color schemes for related risk metrics to facilitate visual comparison. Figure 3 employs a white-to-dark-red gradient for eco-epidemiological risk, while Figure 5's top panel uses blue-to-red for ecological risk. Although these metrics differ slightly (the former includes travel-associated cases), they both represent transmission suitability, making the divergent color palettes, particularly the different low-risk colors (white versus blue), complicate direct visual comparison between current and future risk patterns. Adopting a single color scheme or clarifying color meanings in the figure captions could be helpful.

2. Figure 1’s legend is a bit confusing; the legend is labeled "# cases" yet displays values ranging 0-0.09, which can not represent integer counts of individual cases. Authors should revise the panel title to "Expected monthly travel-associated cases" or "Travel-associated case rate" to clarify the metric's nature and modify the legend label from "# cases" to "Expected cases/month" or "Monthly case probability" depending on the actual interpretation.

Conclusion:

Overall, this paper addresses a critical topic, utilizing a compelling modeling methodology to estimate dengue risk in California. With some major revisions and clarifications to the methodology, it would be an important paper to publish.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they used generative AI to come up with new ideas for their review.