The $3 million, five-year award will be a joint effort between the Department of Ambulatory Care and Prevention (of Harvard Medical School and Harvard Pilgrim Health Care), Brigham and Women's Hospital, Massachusetts Department of Public Health, Harvard Vanguard Medical Associates, Harvard School of Public Health, Kaiser Permanente Northern California Division of Research, and the Argentine National Institute of Infectious Diseases.

The goal of this project is to develop powerful mathematical and statistical modeling techniques to quickly detect and monitor infectious disease outbreaks, whether they occur naturally or are released intentionally in a bioterrorist attack.

"This new support will improve our ability to use electronic healthcare data to enhance public health capabilities in detecting outbreaks of serious infectious diseases. We are particularly interested in recognizing unusual patterns of illness at the earliest possible time," says grant co-principal investigator Richard Platt, professor and chair of the Department of Ambulatory Care and Prevention (of Harvard Medical School and Harvard Pilgrim Health Care).

Martin Kulldorff, the other co-principal investigator, and an associate professor and biostatistician in the same department notes, "This grant will allow us to learn how best to use and combine a variety of electronic health data to detect and monitor infectious disease outbreaks such as pandemic influenza; what fraction of the population must be observed to have a reasonable chance to detect important problems; and how we should deal with reporting delays and missing data. We anticipate addressing these gaps with the MIDAS grant and combining this approach with those of other MIDAS research groups."

"Richard Platt and his research group have tremendous expertise in studying trends in large data sets," said NIGMS director Jeremy M. Berg, PhD. "By integrating these details into computer models and then testing the models in very complex health systems, his research team will contribute information critical to identifying an outbreak in the very early stages of development."

This project will develop mathematical and statistical models to enable the use of routinely collected computerized medical information for disease outbreak detection and surveillance. The MIDAS research team will determine optimal ways to use and combine different kinds of health care information to identify hospital-based outbreaks of antimicrobial-resistant infections that may be difficult to detect, and to identify the spread of pathogens within and between communities. This information will allow health officials to better detect and monitor infectious disease outbreaks.

The MIDAS researchers have three areas of modeling they are focusing on: model specification, which describes the number of people utilizing health services of interest under normal conditions when there are no outbreaks, adjusting for natural seasonal and day-of-week effects; model building, which includes implementing detection models that will generate signals when potential outbreaks occur; and model evaluation, which uses historical data with actual disease outbreaks as well as simulated data based on infectious disease transmission models.

The team will develop models for a range of data sets at different geographical scales, from individual hospital wards to an entire country, and for different disease manifestations ranging from non-specific symptoms, such as influenza-like illness, to microbial strains with unusual resistance profiles.

As a test bed, Platt and colleagues will use existing data from two health plans (Harvard Pilgrim Health Care in Massachusetts and Kaiser Permanente Northern California) with more than four million members, from Harvard Vanguard Medical Associates, from a large referral hospital (Brigham and Women's Hospital), from a statewide (Massachusetts) registry of methicillin-resistant Staphylococcus aureus, and from a national antibiotic resistance monitoring consortium (55 hospitals in Argentina).

Co-investigators at Brigham and Women's Hospital's (BWH) Infection Control Unit, led by Platt and Deborah Yokoe, HMS assistant professor of medicine at BWH, will use historical data to see if hospital-based infection clusters could have been better identified, either by recognizing clusters that were not seen at the time (for instance, because they were dispersed across different areas of the hospital), identifying the clusters earlier.

BWH Microbiology Laboratory investigators Thomas F. O'Brien, HMS associate professor of medicine at BWH, and John Stelling, HMS instructor in medicine at BWH, also developed WHONET, a microbiology information system for monitoring antimicrobial resistance and is used by microbiology laboratories in more than 80 countries. WHONET is the source of much of the microbiology data the MIDAS team will use. WHONET is available at http://www.who.int/drugresistance/whonetsoftware/en/.

Kaiser Permanente Northern California Division of Research's John Hsu will work with the Boston-based MIDAS researchers to analyze that health plan's large population and diverse range of data (for instance, radiology and laboratory data) so they can test complex detection algorithms. At the Harvard School of Public Health, biostatistics professor Louise Ryan brings complementary modeling skills. At the Argentine National Institute of Infectious Diseases, Dr. Mauricio Galas will coordinate the team's work with 55 hospitals across Argentina in looking for outbreaks of resistant pathogens that may be regional or national in scope.

Previously, this research team has worked with the Centers for Disease Control and Prevention (CDC), the Massachusetts Department of Public Health, and with other state and local health departments to use such methods to perform real-time analyses of non-specific diagnoses, such as fever and cough, that are assigned during medical visits. Platt is currently the principal investigator of a CDC grant establishing a Center of Excellence in Public Health Informatics at the Department of Ambulatory Care and Prevention (of HMS and HPHC).

Related Links
Kaiser Permanente Northern California Division Of Research
Massachusetts Departmenet Of Public Health
Argentine National Institute Of Infectoius Diseases
Harvard Medical School
Harvard Pilgrim Health Care
Brigham And Women's Hospital
Harvard Vanguard Medical Associates
National Insitute Of General Medical Sciences

Harvard School Of Public Health