Beyond Odwalla

Epidemic Investigation in an On-line World

Patrick O'Carroll

 On October 21, 1996, a physician at Children's Hospital and Medical Center notified the Seattle-King County Health Department that two children had been hospitalized with infections from Escherichia coli 0157:H7. An investigation was begun with the Washington State Department of Health to determine whether the cases were related to a common source. Public health officials had investigated several previous clusters of E. coli 0157:H7-associated disease, including the major epidemic of early 1993. However, this would be the first such investigation since the deployment of Washington's Information Network for Public Health Officials (INPHO).

Public Health Practice Web Sites for Washington Through the Washington INPHO project, a statewide public health "intranet" now links Washington's 34 local health jurisdictions to each other, to the Washington State Department of Health, and to the Internet. The network provides public health practitioners throughout Washington with full access to the World Wide Web (WWW), e-mail and e-mail-mediated discussion "lists," and secure, remote access to certain computer applications and databases at the state health department (e.g., the automated birth certificate system and the hospital discharge database). INPHO also supports the production and maintenance of a public access Web site that provides public health news, an on-line event calendar, and organized links to a wealth of practice-relevant scientific information on the WWW.

How was this new information technology used in the 1996 E. coli 0157:H7 epidemic investigation and response? And did it make a difference?

Putting INPHO to Work

World Wide Web access and e-mail messaging were recently acquired communications tools for many local health jurisdictions in October 1996. In this first major test for INPHO, health officials used a remarkably wide array of INPHO-supported information technology to investigate and respond to theE. coli epidemic. Within seven days of the first evidence of a cluster, epidemiologists at the Seattle-King County Department of Health and Washington State Department of Health were able to announce that the E. coli 015 7: H 7 infections were epidemiologically associated with drinking Odwalla brand unpasteurized apple juice. Subsequently, investigators found E. coli 015 7: H7 bacteria in a previously unopened container of Odwalla apple juice, and genetic "fingerprinting" at the UW School of Public Health and Community Medicine laboratories confirmed that this E. coli was of the same strain that had caused the recent illnesses.

Did information technology make a difference?

The investigation proceeded rapidly; only about a week elapsed from first hint of a cluster to concerted, statewide public health action. The time frame from disease recognition to control measures was only slightly longer in the 1993 epidemic. Should we conclude that the new information technology provided no substantial benefits?

I believe this conclusion is not warranted for several reasons. First, the fundamental ingredients of successful epidemic investigation - well-trained epidemiologists, good field investigation, teamwork among epidemiologists, laboratorians, and community clinicians - are unaltered by the advent of modern information technology. Much of the efficiency of the 1996 investigation undoubtedly reflects the sophisticated epidemiologic capacity of Seattle-King County Department of Health and the Washington State Department of Health, a strong laboratory infrastructure with cutting-edge diagnostic techniques, and experience in managing several past clusters of E. coli 0157:H7. Health departments with less advanced epidemiologic and laboratory capacity would probably benefit more directly and substantially from information technology than would experienced epidemiologists in a large urban center.

Second, for the 1993 epidemic and the 1996 cluster, the similar time frame from outbreak detection to control belies the different levels of effort required to achieve comparable results. For example, in 1993 broad, timely communication with numerous partners involved in the investigation was achieved via facsimile and telephone. This effort required many hours and extraordinary efforts by health department staff members. The same communication during the Odwalla outbreak required dramatically fewer personnel resources, and thus freed staff to attend to higher-level responsibilities. This finding was also noted in the response to Iowa's statewide flood in 1993. During that emergency, daily faxes and telephone calls from each of 99 county health departments, plus manual data entry by several state health department staff, formed a functioning and effective surveillance system for flood-associated illness and injury The extraordinary personnel demands of this system could not be maintained, however, and information technology was successfully used to keep the surveillance going beyond the initial weeks.

Third, the extensive communications used in the Odwalla investigation show how information technology can increase the scientific and experiential information base on which decisions are made. For example, although it may have made no difference in this case, the Odwalla investigation was monitored simultaneously by epidemiologists as far away as Sydney, Australia. Comments from distant colleagues might prove critical in responding to future epidemics caused by pathogens with which we have less experience.

Finally, and most importantly, the information technology used in the Odwalla outbreak does not in any sense represent a thorough exploration of the potential for this technology to promote the practice of public health. Rather, apart from experimental techniques used for rapid genetic "fingerprinting," the information technology advances established epidemic investigation practices only incrementally - primarily through more efficient communications. Despite Washington's recognized leadership in applying information technology to public health practice, we have only just begun to tap its potential. Consider this scenario.

A State-of-the-Art Scenario for the Not Too Distant Future

At 8: 15 am, a local health department administrator in Eastern Washington arrives at her office. She passes her identification card through the reader on her computer and is greeted the computer's voice saying, "Good morning, Ms. Smith. Please enter your password." After she does so, an "Alert!" icon flashes on her screen, and the computer says: "Please note: the state laboratory surveillance network has automatically detected a statistically significant excess of E. coli 0157:H7 cases in Washington during the past five days. Some cases are fro this health jurisdiction. Also note: you have three e-mail messages marked 'urgent.' How would you like to proceed?"

The administrator views the choices then presented on her screen: view e-mail; view infectious diseases surveillance data; and, view regular desktop applications. The administrator says, "View surveillance data," and is presented with a spreadsheet linked to charts and maps that show the excess number of E. coli cases, and list dates of onset, hospital admissions and pertinent demographic data about the cases. She notes that all but one of the E. coli case are either in her health jurisdiction or in adjacent counties. Clicking an on-screen button labeled "Further information" provides the administrator with a wealth of background information, including the basic epidemiology of E. coli-associated disease, common disease presentations, pertinent prevention guidelines, summaries of past outbreak investigations in the region, questionnaires used in the previous investigations, a list of names, telephone numbers and e-mail addresses of regional contacts with experience in managing past E. coli epidemics and an option to launch one of several "distance learning" modules for E. coli 0157:H7.

After briefly reviewing this information, the administrator says, "View e-mail." The computer responds by listing her new messages, urgent ones first. One message is from the state laboratory director (who had been paged by the computerized surveillance program during the preceding evening). He notes that a manual check of the data assembled by the compute of the hospital-based infectious disease surveillance network confirms the automated report of excess cases. A message from the state epidemiologist notes the advisability of a timely response to the apparent cluster and offers assistance in the investigation.

The administrator then sends a brief e-mail message to a list of all practicing primary care physicians in her health jurisdiction. The message notes the possibility of an epidemic of E. coli 0157:H7 disease and alerts them to report suspect cases to the health department without delay Then, after inviting her infectious disease control nurse to join her in her office, she clicks the on-screen "television" icon to launch a real-time video conference with the state epidemiologist at the Department of Health in Olympia, the state laboratory director and infectious disease epidemiologists in Seattle, and health officials in the other affected jurisdictions, to discuss next steps.

Realizing the Vision

Obviously, certain assumptions are implicit in the above scenario: pervasive deployment of state-of-the-art information technology throughout the entire public health community; a public health work force fully trained to use this technology; an automated, state-wide laboratory-based network for public health disease surveillance (a network that includes private and hospital-based laboratories as well as public health laboratories); and a continuing, systematic effort to capture, organize, and assemble computer archives of public health practice information resources (e.g., questionnaires used in past investigations, or lists of persons with relevant expertise).

It may not be obvious, however, that all the technology employed in this scenario is available today. Video conferencing, software "agents" that automatically check data for unusual patterns, network-based data exchange across public and private entities, even voice-directed computers could all be deployed today to enhance the effectiveness of public health practice. Some of this technology is still maturing, and will certainly be better (and cheaper) in the future. Nevertheless, the scenario outlined above is entirely possible using current off-the-shelf computer technology. And the benefits and potential uses of such technology for public health obviously go far beyond, infectious disease control.

Of course, substantial barriers making such a scenario still exist. Cost is one such barrier; although, thanks to projects such as, INPHO and numerous content development projects (e.g., Health Education Resource Exchange in Washington and the Web site for the Breast and Cervical Health Program in Washington State), it is less of a barrier now than in the past. The lack of informatics training is another extremely important barrier to widespread use of modern information technologies.

However, the most important obstacles we face in moving from today's disparate, isolated systems to the kind illustrated in our scenario are not technological. The critical obstacles are organizational and institutional. We will need to work in a far more integrated and cooperative manner than we do now, across health jurisdictions, among hospitals and managed care organizations, and across the public health/private health care divide. Establishing such integration and cooperation will require hundreds of complex decisions and necessitate new relationships and roles for many of those involved. For example, the infectious diseases information system of our scenario would employ instantaneous, automated. electronic data exchange among public and private hospital and laboratory computer systems throughout Washington. Developing just this one system would certainly require thousands of hours of work, and probably involve a small army of professionals from various fields (health care, hospital administrators, data managers, computer scientists, lawyers, etc.). It is a daunting prospect.

Still, with or without public health input, Millions of private and public dollars will be invested during the next decade in the continued development of computerized information systems for hospital and laboratories in Washington. These complex systems will be developed for various reasons (e.g., to manage hospital stays more efficiently), but their development might be guided in part by the public health sector if we can articulate a consistent vision of information technology in service to public health. With health care reform, the need for better and more timely public health information is one of the most critical challenges we face. In the coming years, we need to extend the reach of INPHO beyond the traditional public health sector, to include hospitals, clinics, laboratories, and other institutions that hold the information we need to monitor and protect the public's health.

Acknowledgments

 I thank Dr. John Kobayashi, Washington State Department of Health; Dr. Russ Alexander, Seattle-King County Department of Health; and Dr. Lisa Cairns, Centers for Disease Control and Prevention, for their comments on this article.

References

 Centers for Disease Control and Prevention: Outbreak of Escherichia coli 0157:H7 infections associated with drinking unpasteurized commercial apple juice - British Columbia, California, Colorado, and Washington, October 1996. MMUT 1996; 45(44):975.

 Centers for Disease Control and Prevention: Update: Multistate outbreak of Escherichia coli 0157:H7 infections from hamburgers - Western United States, 1992-1993. MMWR 1993,42(14):258-263.

O'Carroll PW, Friede AM, Noji EK, Lillibridge SR, Fries DJ, Atchison CG: The rapid implementation of a statewide emergency health information system during a flood disaster: Iowa, 1993. Am J Public Health 1995; 85:564-567.

 Author

Patrick O'Carroll, M.D., M.P.H., is special assistant to the director, Public Health Practice Program Office, Centers for Disease Control and Prevention, based at the UW School of Public Health and Community Medicine.

***Use of trade names and commercial sources is for identification purpose only and does not imply endorsement by the Public Health Service or by the U.S. Department of Health and Human Services.
 

Return to the Table of Contents
Return to Northwest Public Health
Created: 5/6/98  Updated: 7/15/99