Electronic medical records may be the greatest public health innovation of our time. This outcome will be as remarkable as it is welcome, because public health is not the reason the United States is preparing to spend billions of dollars on these records.
Public health systems have always depended on clinicians to identify and report important problems. Some reporting is required (such as cases of tuberculosis, sexually transmitted diseases, and measles/mumps/rubella) and some is optional (such as counts of cases of influenza-like illness). But so many reports are never submitted, or arrive late, or lack essential information. As a result, public health action never occurs, or occurs late, or is misdirected. So few physicians report influenza-like illness that large segments of the US population go without timely information.
Public health agencies currently use several kinds of electronic health information that originate in the private delivery system. Electronic laboratory reporting identifies conditions for which laboratory reports are sufficient, such as culture-positive tuberculosis. The CDC's BioSense program uses electronic health information from hospitals to look for early evidence of disease clusters. As useful as these approaches are, they fail to capture critical information, ranging from the address of a patient with a communicable condition (laboratories rarely have patients’ addresses), to historical information (prior positive serologies), to diagnoses made or treatments administered in the ambulatory setting.
There is much to learn from electronic health records in ambulatory care settings, as this issue's article by Jick et al makes clear.1 Looking back, they were able to observe the origin and spread of a multiyear national outbreak of mumps in the United Kingdom. Since this information was available in real time in electronic medical records, in principle it could have been used in real time to help guide public health response.
The CDC's National Center for Public Health Informatics has funded demonstration projects to support real-time surveillance using electronic records in ambulatory practices. These demonstrations include both notifiable disease surveillance, which requires individual reports (www.cdc.gov/mmwr/preview/mmwrhtml/mm5714a4.htm), and syndromic surveillance, which collects counts—for example, to monitor influenza-like illness or for early detection of bioterrorism events.2 Notably, such systems can use distributed analysis methods to allow the clinicians (or the covered entities that collect protected health information) to retain all but specific data elements that serve an essential public health purpose. For instance, it is not necessary to report all positive hepatitis B virus antigen results, since only a minority of infected individuals have acute infection and should be reported. Nor is it necessary to report each person with influenza-like illness when counts by age and sex are sufficient.
Another appealing feature of these systems is the capability of the same infrastructure to support other needs, including detection of noninfectious conditions of public health importance, such as diabetes and prediabetes. Many of these measures are also of interest to clinicians, who can use the data to target practice management activities. These functions are well within the capabilities of electronic medical records.
What is needed, then, to make electronic medical records an effective public health resource? First, the organizations that set standards for clinical data exchange and the public health community must ensure that standard data formats are readily usable for public health purposes. In principle, any standard should be acceptable. In practice, some standards will be better for this purpose than others.
Second, the public health community will need to adopt definitions that can use electronic medical record information and then keep these definitions current. Broadly accepted definitions that are suitable for electronic medical record data are only now beginning to be developed.
Third, public health functions must be kept separate from other clinical information activities, both to relieve electronic medical record vendors of the obligation to satisfy ever-changing requirements, and to reduce the operational burden on clinical systems. This can be accomplished by bundling the public health functions into companion software that operates on exported data. For large practices and health information exchanges this may require stand-alone hardware. For small practices, the public health software could coexist on the same hardware platform as the electronic medical record. It will be ideal if vendors could bundle public health software with their products, but this is not necessary.
Finally, public health funding will be required to build, disseminate, and sustain these activities. The needed planning and investment will be small in comparison with the expenditure we are about to make in the United States as electronic medical records make their way into routine clinical practice. It is a small price to pay to create a new foundation for public health.
ABOUT THE AUTHOR
RICHARD PLATT, is Professor and Chair of the Harvard Medical School Department of Ambulatory Care and Prevention at Harvard Pilgrim Health Care. He is Principal Investigator of a CDC Center of Excellence in Public Health Informatics.
1. Jick H, Chamberlin DP, Hagberg KW. The origin and spread of a mumps epidemic: United Kingdom, 2003–2006. Epidemiology
2. Yih WK, Caldwell B, Harmon R, et al. National Bioterrorism Syndromic Surveillance Demonstration Program. Morb Mortal Wkly Rep