Dengue epidemic alert thresholds for surveillance and decision-making in Puerto Rico: development and prospective application of an early warning system using routine surveillance data

Objectives The Puerto Rico Department of Health (PRDH) seeks to identify dengue epidemics as early as possible with high specificity. Design Development and prospective application of an early warning system for dengue epidemics using routine historical surveillance data. A weekly intercept-only negative binomial regression model was fitted using historical probable and confirmed dengue data. A range of threshold definitions was explored using three model-estimated percentiles of weekly dengue case counts. Setting Dengue is endemic in Puerto Rico with irregular occurrence of large epidemics with substantial impact on health burden and health systems. Probable and confirmed dengue data are routinely collected from all hospitals and private clinics. Participants A total of 86 282 confirmed or probable dengue virus cases were reported from 1 January 1986 to 30 June 2024, with an annual mean of 2212 cases (median: 1533; range: 40–10 356). Primary and secondary outcome measures The model was fitted retrospectively to mimic real-time epidemic detection and assessed based on sensitivity and specificity of epidemic detection. Results The 75th percentile threshold aligned best with historical epidemic classifications, balancing false alarms and missed detections. This model provides a robust method for defining thresholds, accounting for skewed data, using all historical data and improving on traditional methods like endemic channels. Conclusions In March 2024, PRDH declared a public health emergency due to an early, out-of-season surge in cases that exceeded the epidemic alert threshold developed in this study. This real-time application highlights the value of these thresholds to support dengue epidemic detection and public health response. Integrating thresholds with other tools and strategies can enhance epidemic preparedness and management.