Spatiotemporal Cluster Detection for COVID-19 Outbreak Surveillance: Descriptive Analysis Study.

Area of Science:

• Epidemiology
• Public Health Surveillance
• Statistical Modeling

Background:

• During winter 2020-2021, Washington State faced numerous COVID-19 outbreaks, primarily in workplaces, community settings, and schools.
• Existing detection methods, like automated address matching for healthcare facilities, were insufficient for systematic statewide outbreak identification, leading to delays and incomplete data.

Purpose of the Study:

• To implement and assess a systematic, statewide cluster detection model for timely identification of COVID-19 clusters.
• The model aimed to aid local health jurisdictions (LHJs) in investigations and resource prioritization.

Main Methods:

• A pilot study involving 6 LHJs tested parameters like analysis type, geographic aggregation, cluster radius, and data lag.
• A weekly, LHJ-specific retrospective space-time permutation model was implemented statewide from July 17 to December 17, 2021.
• Detected clusters were analyzed by LHJ population and disease incidence, and compared with reported outbreaks.

Main Results:

• The model identified 2874 new COVID-19 clusters statewide.
• Over 58% of identified clusters were timely, occurring within one week of analysis, allowing for early intervention.
• While 363 reported outbreaks matched SaTScan clusters, the model identified approximately one-third of these before they were recognized in the surveillance system.

Conclusions:

• The space-time permutation model successfully identified timely and actionable COVID-19 clusters statewide, irrespective of population or incidence.
• The model demonstrated effectiveness in detecting clusters in high-priority settings like schools and community settings, matching reported outbreaks.
• Results suggest the model may offer earlier outbreak detection in workplaces compared to existing methods.