Temporal and Spatial Analysis for Severe Weather Related Health Concerns *

BACKGROUND:   In New Jersey (NJ), real-time emergency department (ED) chief complaint data are provided from 78 of 80 acute care and satellite EDs to EpiCenter, NJ's syndromic surveillance data management system. In 2014, NJDOH implemented a severe weather classifier in EpiCenter to classify chief complaint data to track both health and mental health concerns related to adverse weather conditions. This new classifier can assist health departments in tracking emerging health hazards during an event, and also to prepare for public health response after the event. 

METHODS:   Daily trends for severe weather-related classifications during two weeks before and after Superstorm Sandy were plotted and analyzed using the Cumulative Sum with Exponential Moving Average (CuSum EMA) method in EpiCenter. In addition to time trend analysis, cluster analysis including both space-time parameters was conducted to evaluate the geographical clusters.  SaTScan v9.3 is used to provide cluster analysis for severe weather classifications using time-space permutation method. Cases with residency zip code outside of New Jersey or without zip code were excluded.  ArcGIS 10.2.2 for Desktop was used to provide visualization of clusters identified. 

RESULTS:   ED visits related to immediate health concerns including injury (carbon monoxide poisoning, tree related injury), interrupted medical care needs (dialysis, medicine refills, oxygen needs), upper respiratory illness, asthma, mental health (anxiety/adjustment disorders), and methadone/opiate/heroin use increased for one to several days after the storm. CuSum EMA generated threshold levels for each classification. Scan statistics at zip code level indicated six clusters among cases with carbon monoxide poisoning concerns during the first two weeks of Superstorm Sandy using cluster radius limitation of 5 miles.  Clusters were identified in northeast counties of New Jersey and along the shore. One small cluster in Camden County was identified with medicine refill needs which indicated a significant increase in that specific area in comparison with other areas in New Jersey.   

CONCLUSIONS: Identified clusters of immediate health concerns (carbon monoxide poisoning and medicine refill needs) can assist health department to alert the public for health concerns and to prepare for public health needs. The major limitation of this analysis is that 7 of 10 hospitals in Monmouth and Ocean counties along the shore did not participate in EpiCenter prior to Sandy. However, as participation in EpiCenter has increased to 98% in New Jersey, the severe weather classifier will be a valuable tool to provide alert information for future events.