The trusted source for
healthcare information and
Critical components of EHR database in the UK
All treatment data should be tracked
Physicians in the United Kingdom have been using electronic medical records since the late 1980s, creating a large database that's available for research. An investigator in the United States who has used this database in research says there are a number of lessons Americans can learn from the UK experience.
UK doctors were the ones who pushed for electronic health records (EHRs) because they found it to be more efficient, says Richard Tannen, MD, a professor of medicine at the University of Pennsylvania School of Medicine in Philadelphia, PA.
"When this started to happen more widely, some people in the UK decided this information could be coalesced into a large database that would be useful for research purposes," Tannen says. "The database now has some characteristics that I believe are crucial."
For instance, the UK's National Health System requires that any providers other than primary care doctors who treat patients must report their findings to the patients' primary care physicians, Tannen explains.
"So when you go to a consultant that consultant is obligated to send back to the primary care doctor the results of his or her observation," he adds. "If a patient goes to the hospital then the hospital must send the primary care doctor the results of that encounter."
This means that patient medical records capture all of the major health events that occur in patients' lives, Tannen says.
"It might have things coded wrong, but it's all in that record, and that's very important when you're doing research looking at those outcomes," he says.
As the United States switches to universal EHRs, it's important that they be designed to capture everything, similarly to the UK model, Tannen says.
"The second thing that makes the UK database enormously powerful is the way prescriptions are handled in the UK," he says. "They have a central group that dispenses prescriptions for everybody in the health system."
Virtually all prescriptions patients take over any period of time are written by the primary care physician, Tannen says.
"A consultant can write a prescription for a week or two, but he can't write for long-standing medical needs," he adds. "So, therefore, since the primary care record is linked electronically to this national prescription data, we have in the database virtually every drug prescribed for every patient."
If the United States makes medical research a priority then someone needs to establish standards and a mechanism for capturing medical data for a universal database, he suggests.
"The British database we used had eight million patients in it," Tannen says. "Our estimates, based on work we've done to date, is to have an ideal-sized database in the U.S., you'd need about 50 million patients."
Tannen and co-investigators developed a sense of how many patients they could find who'd meet entry criteria for randomized, controlled trials when they used the UK database.
To achieve statistically meaningful results, they needed more than a 5% sample, he adds.
"We needed five-to-six times that," Tannen says.
Other types of research also require large databases, including these, Tannen says:
1. Monitoring adverse events (AEs): "When you're looking for adverse events with new drugs, those events aren't very frequent, so you need a large database if you want to detect whether there's a sufficient number of events to analyze in a meaningful way," Tannen says. "Like with Vioxx [rofecoxib] and Avandia [rosiglitazone], the diabetes drug, look at the number of outcome events those observational studies had — they were relatively few."
So it takes a much larger cohort to make valid analyses of AEs, he adds.
2. Analyzing subsets of populations: "One of the potential powers of this is to look at whether particular subsets of patients with particular characteristics respond more effectively and are more likely to have complications from a given treatment," Tannen says.
"You might look at different ages, different diseases, gender, and a host of things," he explains.
The UK database uses random code numbers in place of identifying information, and it doesn't contain actual dates of birth — only years of birth, he notes.
"And there are no racial characteristics," Tannen says. "It's a political issue whether or not to have racial characteristics."
An argument for having the database include racial characteristics is that people of different ethnic backgrounds may respond differently to various therapies, and if investigators want to analyze the response of people with different ethnic backgrounds to a treatment, then they'll need to see data with racial characteristics, Tannen explains.
3. Post-marketing studies: "Unequivocally, a national database will have enormous impact on post-marketing studies and could in fact be the way to do post-marketing studies," Tannen says.
"Once a drug is on the market you could examine its use for treatments other than what the initial randomized, controlled trials were designed to test," Tannen says. "Then you could get a lot of information that would help you do a better design of a randomized, controlled trial."
This data could inform the medical community of new indications, adverse complications, and other post-marketing issues, he adds.