Using Media Reports to Track Deaths, Hurricane Sandy, 2012

Brief Summary
Background: Accurate and timely reporting of mortality during disasters is useful and critical; however, timely national mortality surveillance does not exist. During Hurricane Sandy, CDC/Health Studies Branch (HSB) piloted an approach for tracking disaster-related deaths reported by the media for situational awareness. Our objective was to determine the accuracy and timeliness of this data. Methods: From October 29–November 5, 2012, HSB tracked hurricane-related, media-reported deaths on the internet using a keyword search to capture demographics and circumstance of death. From May–October, 2013, we requested death records from states affected by Sandy to evaluate the media data. Frequencies, sensitivity, and positive predictive value (PPV) were calculated. Results: We found 76 hurricane-related, media-reported deaths from New Jersey, Connecticut, North Carolina, Virginia, and New York City (NYC). Of these deaths, 67 (88 %) were found in vital statistics records. NYC identified hurricane-related deaths from vital statistics using keyword search; sensitivity and PPV of the media to detect deaths in NYC were 77% and 83%, respectively. The other four states provided vital statistics records based on our media-based list. Percent agreement varied: demographics (86%), location (86%), circumstance (74%), date (62%), and cause of death (56%). CDC/HSB and NYC reported deaths for situational awareness within 24 hours while none of the other four states conducted active disaster-mortality surveillance during this event. Conclusion:  Media-reported deaths captured by HSB during Hurricane Sandy were moderately sensitive and more timely compared to vital statistics. Tracking media-reported, disaster-related deaths can provide situational awareness and support immediate public health decision-making during the initial stage following disasters.