Perspective: Foreign Travel Classification

In the event of an outbreak, public health routinely strives to accurately identify all potential cases of disease. This process includes differentiating between the “worried well,” who may reference an infectious disease by name simply because it is receiving widespread media coverage, and patients who cite genuine causes for concern, like contact with an infected person or travel to a region where an outbreak is peaking. To help address these concerns, EpiCenter users in public health have requested the implementation of a foreign travel classification.

Specifically, this request was made during the peak of concern over Ebola, when recent travel to West Africa was of particular concern and a “Foreign Travel Syndrome Query” was being shared within the BioSense User Group (BUG) ISDS Community Forum. The timeliness of the request demonstrates the diligence and proactiveness of EpiCenter users. The public health representatives who utilize EpiCenter consistently make an impressive effort to use all available technology to maximize their preparedness.

When travel to Mexico was of interest during 2009 swine flu pandemic, Health Monitoring Systems thoroughly researched the efficacy of tracking foreign travel via syndromic surveillance. During the 2014 Ebola outbreak, we revisited that research. In addition, we thoroughly reviewed the BUG Foreign Travel Syndrome Query while evaluating the effectiveness of implementing a similar foreign travel classification in EpiCenter.

Our research revealed a number of problems that arise when using a foreign classification with traditional syndromic surveillance methods. We observed the following issues:

  • False Negatives: The primary difficulty lies in the inherently limiting definition of a chief complaint. Because chief complaints are intended to capture the single, primary reason for a patient’s ED visit, false negatives are created by incorrect assumptions that no mention of foreign travel indicates its absence.
  • False Positives: The potential for high rates of false positives is equally problematic. There is nothing to prevent foreign travel from being referenced in a chief complaint, even if it is wholly unrelated to a patient’s symptoms. This has the potential to generate a great deal of “noise.”
  • Lack of Situational Awareness: Enumerating all possible places of travel that could theoretically be of interest in a generic future outbreak is virtually impossible. The BUG Foreign Travel Syndrome Query utilized a standardized list of nation-states that included country names such as “Terres Australes et Antarctiques Françaises,” but failed to include any city names (eg. “Taipei”), a comprehensive list of regions (eg. “Southeast Asia”), or a comprehensive list of continent names (eg. “South America”). During 2014’s Ebola outbreak, US cities such as “Dallas” and “Cleveland” were also of interest. These would obviously not be captured by a “foreign travel” classification.

As a result of these findings, we have determined that the efficacy of a generic foreign travel classification would be extremely limited. In the event of an outbreak, what is needed most is the ability of a syndromic surveillance platform to adapt and respond to changing situations. EpiCenter has several tools designed to be helpful toward this end:

  • Medical Notes: Our primary recommendation for improving situational awareness about relevant travel during an outbreak is for public health EpiCenter users to aggressively pursue obtaining medical and triage note data from facilities participating in syndromic surveillance. During the Ebola 2014 outbreak, the ability to review the full text of medical notes in which patients were asked about their travel history using EpiCenter’s Triage Notes Report provided invaluable context. A number of patient interactions with commonly seen chief complaints–like “Cough” or “Fever”–were linked to triage notes that included references to Ebola, travel to a region of concern like Dallas or West Africa, and other indicators of potential exposure. In one case, a patient revealed in a triage note that they were concerned about potential Ebola exposure from having been on the same aircraft (but not the same flight) as a confirmed Ebola patient.
  • Custom Classifications: In the event of an outbreak, the Custom Classification tool can be used to quickly establish a means of surveilling chief complaints for mention of recent travel to specific geographic locations of interest. For example, during the 2014 Ebola outbreak, Ohio quickly published a custom classification for Ebola to surveil for keywords such as “Africa,” “Guinea,” “Sierra Leone,” and “Nigeria.”
  • Novel Keyword Surveillance: In an emergency outbreak situation (or for certain types of special event surveillance), HMS can enable the Novel Keyword surveillance tool. The tool surveils chief complaints for uncommonly seen keywords—a way of being vigilant even when you may not yet know exactly what it is you’re looking for.

For more information about any of the tools mentioned above, please contact us directly at inquiries[at]