A Mouthful of Info
Electronic patient records are a data gold mine for a new generation of oral health research
Where some dentists might see a patient’s oral health record as an important diagnostic tool, dental researchers see something else: golden nuggets of data just waiting to be mined.
As dental schools have switched from paper to electronic medical records, the potential for using patients’ case notes, prescription histories, X-rays and test results as research data has opened new doors for scientists. That means every time patients visit the dentist, they could be contributing to the next big breakthrough in the relationship between gum disease and heart disease or helping prevent implant failures.
Already, Tufts has partnered with three other dental schools—Harvard, the University of California at San Francisco and the University of Houston at Texas—to create a warehouse of patient information that is being used for research. In just two years, the schools have gathered and stored data from 1.2 million patients.
Paul Stark, director of advanced and graduate education at Tufts School of Dental Medicine and a creator of the warehouse, envisions this data gold mine eventually shedding light on problems that commonly vex clinicians, such as why 5 to 10 percent of restorations fail every year. “What are the factors that contributed to the failure? Was it operator error? Was it patient compliance? Was it behavioral?” he asks.
The data can be used retrospectively. In August, Stark and colleagues published a study in the Journal of the American Dental Association based on electronic health records that found adults with intellectual and developmental disabilities aren’t getting enough specialized dental care to meet their needs.
Scientists can also use the database to efficiently recruit participants for intervention studies because it can easily identify patients who fit a certain profile. It can be especially useful for investigating rare diseases, such as Sjögren’s syndrome, an autoimmune illness that causes dry mouth. “Even if only 2 percent of the patients have it, you have 500 or 1,000 patients in your database,” said Stark.
To build the data warehouse, “patient confidentiality is paramount,” Stark said. When any treatment is done in the dental school clinics, patients have to give their consent, and a part of the consent form states that their information might be used for research.
Patient data is stripped of identifying information, and each school holds a decryption key that is separate from the database. Researchers who wish to identify patients, to recruit them for a clinical trial, for example, must seek approval from an institutional review board, as is the case with any other study involving human subjects.
As soon as electronic health records systems became available to dental schools in the early 1990s, Professor Emeritus Robert Chapman, A63, D67, DG74, then chair of prosthodontics and operative dentistry, became intrigued by their promise for research. Currently most dental researchers rely on data from insurance companies. But less than half of American adults have dental insurance. “He definitely saw the value in patient care, but he also saw the wealth of data that was there and the utility of the data,” Stark said.
Soon after Tufts moved to an electronic records system in 2004, the dental school hired Stark, a biostatistician who had been trained at Cornell and the Harvard School of Public Health, and most recently worked at Tufts Medical Center, researching cancer rates.
Using electronic health records for research
may become an economic imperative.
Chapman and Stark came up with the idea of bringing dental schools together to share data. In June 2007 they invited representatives from interested schools to come to Boston for two days of meetings. By the end, representatives from 14 dental schools in the United States, Canada and Europe had outlined the beginnings of the Consortium of Oral Health Research and Informatics, or COHRI. The consortium now has 52 members from 20 dental schools, all of which intend to contribute to the data warehouse in the coming years.
COHRI members—who include clinicians, researchers, epidemiologists, statisticians and informaticians (who deal with the collection, cataloging and dissemination of digitized data)—meet twice annually and on conference calls throughout the year. They work on developing standardized medical and dental history forms, constructing uniform dental diagnostic terms and establishing research objectives.
“Everyone checks their egos at the door when they come to these meetings,” said Stark, who will chair the consortium in the coming year. “That helps us accomplish as much as we have.”
At the COHRI national meeting in July, hosted by Tufts, members were thinking a lot about what they could learn from hospitals—or more specifically, what missteps they could avoid. Medicine’s countless specialties and tests—from bloodwork to MRIs to CAT scans—have made extracting data from electronic health records fairly complicated.
Ross Koppel, an adjunct professor at the University of Pennsylvania who studies health-care information technology, estimates that on average, a hospital has between 100 and 400 information technology systems of various sorts, and none of them talks to each other. “They are really a Tower of Babel,” he said. This lack of interoperability makes data collection for research extremely difficult.
“As dental IT becomes more popular, you are going to face the same monstrosities of this data wild west,” he told the group, while applauding them for having the forethought to form a consortium early on. “You have an opportunity to help define the debate, and that’s what I’m begging you to do. Don’t let it go the way of our cousins in health care.”
Dental schools are in a much better position, Koppel said. Most U.S. dental schools are already using digital records, and 85 percent of them use the same electronic health records software system, which makes sharing data much easier.
Using electronic health records for research may become an economic imperative. Shawn Murphy, an associate professor of neurology at Harvard Medical School and medical director of research computing for Partners Healthcare, has been working with medical information technology since 1995.
“Especially with genomic studies, we need larger and larger cohorts of patients to show significance,” he told the COHRI group. “The problem is that doing them the old way—which is recruiting patients for clinical trials, signing them up and getting their blood and so forth—is just too expensive.” He noted that the U.S. Department of Health and Human Services estimates a study involving 500,000 to 1 million participants would cost $3 billion.
But not every physician and dentist is rushing to embrace this new method of research. “Inevitably, I’ll give this talk, and there will be people in the room who just hate this whole concept,” Murphy said. For starters, doctors worry that asking extra questions and filling in boxes on a computer screen may keep them from even looking up at the patient. “The time it would take to collect five minutes of research, when you only have a 10-minute urgent-care slot in the first place, isn’t going to work,” Murphy said. He has found that doctors need to be assured that every click of the mouse has a clinical purpose, or “they just won’t to do it.”
Some researchers also balk at the idea of substituting electronic health-record data for actual clinical trials, where patients are enrolled, questions are standardized and forms are completed in full.
“The [electronic medical record] is not like that at all,” admits Murphy. “Patients are coming in and out of the system. Some data is being collected, but some data is not.” To balance that, researchers who use health-record data must do a much more rigorous statistical analysis of their capture rate and potential biases of the data. (For example, did several patients at one clinic receive the same medication because that clinic was visited by a persuasive drug rep?)
To demonstrate that dental records are a viable source of data, Stark and his team are comparing COHRI data to that of the National Health and Nutrition Examination Survey (NHANES), a large population survey that is the cornerstone for many biomedical research studies. “It’s proof of concept to show that the same relationships that are being found in these other data sources can be found in the data we have,” he said.
Robert Gellin, professor and chair of the stomatology department at the Medical University of South Carolina College of Dental Medicine, said that as a periodontist, he hopes the COHRI database will one day elucidate the numerous links between gum disease and systemic illnesses—including heart disease and diabetes.
“There are a lot of correlations, but cause and effect has not really been established,” he said. “The studies we have right now are maybe 100 to 500 people, which isn’t millions of people. So by pooling the data from all these schools, we get really big numbers—and then it becomes meaningful.”
Julie Flaherty, a senior health sciences writer in Tufts’ Office of Publications, can be reached at firstname.lastname@example.org.