Risk assessment and perspectives of local transmission of chikungunya and dengue in Italy, a European forerunner

Abstract

To address the growing frequency, extension, and size of local arboviral outbreaks in Europe we retrospectively analyzed dengue and chikungunya transmission in Italy from 2006 to 2023. We applied generalized additive models to the records of travel-related cases to highlight the spatiotemporal patterns of disease importation, calculated reproduction numbers for six local outbreaks based on autochthonous case data and mapped current transmission risks by applying a computational model that integrates human density, entomological, and climate data. Outbreak locations appear driven by case importation, which is notably higher for dengue - especially from June to October - rather than local transmission risks. Although reporting delays and favorable temperatures allowed onward transmission for several generations from mid-August to mid-November, upon outbreak detection control of transmission was achieved within 15 days. In high-risk areas, significantly longer epidemic risks were found for chikungunya (over 4 months). However, considering observed importation trends, increasingly frequent local dengue outbreaks are expected. Case detection should be prioritized focusing on areas, and in times, where environmental and climate conditions are permissive, regardless of prior outbreaks.

Fig. 1. Disease case importation in Italy.

A Map and B temporal distribution of travel-related DENV cases from 2006 to 2023. C Annual trend in the number of identified travel-related DENV cases, shown as the estimated multiplicative change (solid line) relative to the average number of cases observed over the 2006–2023 period; the shaded area represents the 95% confidence interval (CI) of model estimates. The inset displays the corresponding monthly trend, shown as the estimated multiplicative change (solid line) relative to the average number of cases observed over the year. D Annual trend in reporting delays for travel-related DENV cases, shown as the estimated multiplicative change (solid line) relative to the average over the 2006–2023 period; the shaded area represents the 95% confidence interval (CI) of model estimates. E–H Same as (A–D), but for CHIKV.

Fig. 2. Transmission from surveillance of human cases.

Mean (solid line) and 95%CI (shaded area) estimates of the net reproduction number R_t, based on the time series of notified cases by date of symptom onset across the six outbreaks of DENV (top) and CHIKV (bottom) occurred in Italy between 2006 and 2023. Gray bars represent the weekly number of notified cases by date of symptom onset.

Fig. 3. Risk of onward transmission from entomological data.

Estimated mean basic reproduction number (R₀) for DENV (top) and CHIKV (bottom) in Italy at the beginning of August during the outbreak years, assuming Ae. albopictus as the primary vector species. Estimates are provided at a 100 m × 100 m spatial resolution and are shown only for areas with a population density of at least ten inhabitants per hectare and where the estimated R₀ exceeds 0.1. Values are displayed as proportional to the population size to highlight potential epidemic risks and represent the average estimate of 500 model runs. Maps also show the locations and sizes of historically observed outbreaks.

Fig. 4. Comparison of transmission estimates from human and entomological data.

Mean (solid line) and 95%CI (shaded area) estimates of DENV and CHIKV R₀ obtained by applying the risk model for onward transmission across various outbreak locations during the outbreak years. The red boxplot represents the estimated posterior distribution (2.5th, 25th, 50th, 75th, and 97.5th percentiles) of the peak R_t value estimated from the corresponding time series of human confirmed cases by date of symptom onset.

Fig. 5. Temporal trends in reproduction number.

A Estimated variations in the mean DENV R₀ on August 1st across the outbreak years, considering all patches with a population density of at least ten inhabitants per hectare. Estimates are stratified by four Italian macro-areas and normalized to the average values observed across the four years. Points represent the mean difference in the macro-area, while lines show the 2.5th and 97.5th percentiles of the distribution across the patches within each macro-area. North-East (South Tyrol, Trentino, Emilia-Romagna, Friuli Venezia Giulia, Veneto): 370,736 patches; North-West (Liguria, Lombardy, Piedmont, Aosta Valley): 455,061 patches; Centre (Lazio, Marche, Tuscany, Umbria): 286,690 patches; South (Abruzzo, Apulia, Basilicata, Calabria, Campania, Molise, Sardinia, Sicily): 438,542 patches. B Same as (A) but for CHIKV.

Fig. 6. Likelihood of onward transmission and duration of epidemic risk in Italy.

A Estimated likelihood of onward DENV transmission in Italy on August 1st, 2023, shown only for areas at risk of autochthonous transmission (R₀ ≥ 1). Patches are displayed as proportional to population size to highlight potential epidemic risks and represent the average estimate of 500 model runs. B Same as (A) but for CHIKV. C Estimated duration of epidemic risk for DENV, defined as the number of consecutive days during which the estimated R₀ exceeds the epidemic threshold of one. Estimates are shown only for areas at risk of autochthonous transmission for at least one day and represent the average estimate of 500 model runs. D Same as (C) but for CHIKV.