Researchers at the University of Pennsylvania School of Medicine have discovered that electronic medical records can be used to test drug efficacy.
Richard Tannen, MD, a professor of medicine at the university, was the lead researcher in the study to find out if patient data, as captured by EMR databases, could be used to obtain vital information as effectively as randomized clinical trials when evaluating drug therapies.
"Our findings show that if you do studies using EMR databases and you conduct analyses using new biostatistical methods we developed, we get results that are valid," Tannen said. "That's the real message of our paper - this can work."
Tannen said his group recognized that large EMR databases could potentially give researchers the ability to study groups reflective of the total population, not just those who participate in clinical trials, and circumvent studies too costly or unethical for clinical trials.
"Our study cautiously, yet strongly, suggests that enormous amounts of information within electronic medical records can be used to expand evidence of how we should or shouldn't manage healthcare," he said.
Some critics have argued that these databases contain observational information, which does not offer the same level of control as random trials. To address this criticism, Tannen's group had to determine a way to use EMR databases for insights on therapy efficacy and then prove the results they found were valid.
Tannen's team selected six previously performed random trials with 17 measured outcomes and compared them to study data from an electronic database - the UK general practice research database (GPRD), which included the medical records of roughly 8 million patients. Treatment efficacy was determined by the prevalence of cardiovascular outcomes, such as stroke, heart attack and death.
Researchers used standard biostatistical methods to adjust for differences in the treated and untreated groups in the analysis of the database information and found there were no differences in the database outcomes compared to random clinical trials in nine out of 17 outcomes.
In the other eight outcomes, Tannen's group used an additional new biostatistical approach that controlled for differences between the treated and untreated groups prior to the time the study began. By using the new method instead of the standard approach, the researchers showed there were no differences between the outcomes in the EMR database study compared to the random clinical trials.
Tannen said the ability to use EMR databases from the United States to measure the efficacy of therapies will take more than 10 years of national data, and the results of this study should serve as a catalyst for more researchers to explore the accuracy of the information that can be obtained using EMR database studies.
"An appropriately configured EMR database could offer an invaluable tool, but we need to get to work now on how to configure it properly," Tannen said. "If we don't worry about this issue right now and promote a higher investment in the area of EMR research, we'll lose ... an enormous health opportunity."
