Using Data to Optimize Activation, Staffing, and Duration of Communicable Disease Hotlines

Wednesday, June 17, 2015: 11:05 AM
Back Bay D, Sheraton Hotel
Livia Navon , Centers for Disease Control and Prevention, Chicago, IL
Craig S. Conover , Illinois Department of Public Health, Chicago, IL

BACKGROUND:   In response to the importation of Middle East Respiratory Syndrome (MERS) and Ebola to the United States, the Illinois Department of Public Health worked with the Illinois Poison Center to establish temporary, 24/7-staffed hotlines in May and October, 2014. Hotline activation was based on increased health department call volume and perceived public concern. To optimize activation timing, staffing, and duration of future hotline activations, Google Trends data were analyzed to determine if internet search volume data could be utilized as an objective measure of public concern. To better understand hotline surge patterns, call volume trends were examined.

METHODS:   Google Trends internet search volumes in Illinois were compiled for the terms “MERS” and “Ebola” before and during the times the respective hotlines were open. The relative magnitude of search volumes for “MERS” and “Ebola” were compared. Spearman rank correlation coefficients were calculated to assess the correlation between internet search volume and call volume for each hotline. Lastly, hotline call volume trends were assessed for temporal patterns. 

RESULTS:   The MERS hotline received 36 calls during 19 days of activation. The Ebola hotline received 664 calls during 30 days of activation. Both hotlines experienced highest call volumes (Ebola hotline: 316 calls; MERS hotline: 22 calls) within the first two days of initial activation, followed by a rapid decline. Spikes in call volume were seen in relation to media coverage of key events; however, subsequent call volume never reached initial activation levels. Google Trends search and hotline call volumes were strongly correlated: rs=0.80 (p<0.01) for MERS and rs=0.81 (p<0.01) for Ebola. This indicates that online search volume can be used as a proxy for hotline call volume patterns. Peak online search volume for “Ebola” was 33 times higher than peak search volume for “MERS.”

CONCLUSIONS:   High call volumes within the first two days of hotline activation followed by steep declines were observed. Although Google Trends data for search terms are not available as absolute numbers, comparing search volume between search terms (e.g., “MERS” to “Ebola”) can provide a relative measure of expected call volume and could have prospectively predicted the higher call volume experienced on the Ebola hotline. Google Trends currently has a 1 to 2 day lag in data availability, which limits real-time utility. However, these data provide an objective measure of public interest to consider when deciding about the need for hotline activation, level of staffing required, and timing of hotline closure.