Abstract
Dengue, a mosquito-transmitted viral disease, has posed a significant public health challenge in Singapore for decades. In 2020, the nation experienced an unprecedented outbreak coinciding with the COVID-19 pandemic. This paper presents a high-resolution dataset of 16,116 dengue cluster records reported by the National Environment Agency (NEA) from February 15 to July 9, 2020. This period encompasses the "Circuit Breaker" (CB) lockdown (April 7 to June 1), providing a unique opportunity to study how restricted human mobility and shifts in residential occupancy influenced dengue transmission dynamics. The dataset, mapped to fine-grained spatial units (subzones), reveals distinct patterns of viral dispersion and identifies specific subzones that acted as residential hotspots or workplace-related transmission zones.
Introduction: The "Perfect Storm" of 2020
In 2020, Singapore faced a dual public health crisis. While the global COVID-19 pandemic necessitated stringent social distancing measures, the city-state also grappled with its worst-ever dengue outbreak, recording over 32,000 cases. Dengue transmission is intrinsically linked to human mobility; Aedes aegypti mosquitoes typically bite humans where they congregate.
The implementation of the "Circuit Breaker" (CB) lockdown created a massive natural experiment. Most residents were confined to their homes, shifting the primary "blood meal" opportunities for mosquitoes from commercial hubs and workplaces to residential areas. Understanding this shift requires granular, time-stamped spatial data. This study addresses the need for accessible, fine-grained datasets by consolidating and geocoding dengue cluster reports to analyze how lockdown measures inadvertently influenced the epidemic's trajectory.
Methodology
The researchers synthesized data from the National Environment Agency (NEA) and SGCharts, covering the period from February 15 to July 9, 2020. The methodology involved several key steps:
- Geocoding: Infection locations (residential blocks or workplaces) were collected and assigned latitude/longitude coordinates.
- Subzone Mapping: Using ArcGIS, these points were mapped into Singapore’s 323 "subzones"—spatial units defined by the Urban Redevelopment Authority (URA), averaging approximately 1.35 km² each.
- Temporal Classification: The study period was segmented into three phases: Before Lockdown (BL), Circuit Breaker (CB), and After Lockdown (AL).
- Statistical Analysis: The average number of locations per week for each subzone was calculated to identify distinct patterns of outbreak evolution.
The resulting dataset includes 16,116 records, each containing spatial coordinates, subzone IDs, temporal data (week-ending dates), and epidemiological data (case counts per cluster).
Key Insights: Dispersion and the "Blood Meal" Hypothesis
A primary finding of the study is the "dispersion effect." As the lockdown progressed, dengue did not merely intensify in existing hotspots; it spread to previously unaffected areas. By late May, the number of subzones involved in clusters increased significantly. This suggests the virus "leaked" into residential neighborhoods as people spent more time at home, providing a steady supply of hosts for the Aedes population.
The data supports the "Blood Meal" hypothesis: the lockdown increased the frequency of mosquito-human contact in residential settings. This provided the necessary biological resources for the mosquito population to thrive and the virus to circulate more effectively within community settings.
Key Results: Spatial Response Patterns
The researchers identified two distinct types of spatial responses among the subzones:
- CASE 1 (Residential Hotspots): Subzones where cluster locations increased during the lockdown and decreased afterward. These are likely residential areas where increased human presence during the CB period fueled transmission.
- CASE 2 (Workplace/Commercial Zones): Subzones where cluster locations decreased during the lockdown and rose afterward. These likely represent commercial or industrial zones that were largely vacated during the CB period and saw a resurgence in activity (and dengue cases) once restrictions were lifted.
The study identified 4 specific subzones (IDs 148, 235, 269, 274) following the Case 1 pattern and 10 subzones following the Case 2 pattern. Furthermore, the analysis highlighted a "long tail" effect: while 90% of dengue locations were concentrated in a small number of subzones, the remaining 10% were scattered across dozens of different areas, complicating control efforts.
Conclusion and Future Directions
This dataset provides a critical foundation for understanding the intersection of human mobility and vector-borne disease transmission. While the study is limited to "cluster cases" (excluding sporadic infections), it offers high-resolution insights into the 2020 outbreak.
Future research can build upon this work by integrating environmental data (rainfall and temperature), studying the impact of the DENV-3 serotype dominance in 2020, and correlating cluster locations with entomological data, such as Gravitrap indices, to further pinpoint the drivers of residential hotspots.
Summary updated