EHRs A “Necessary Step” In Moving Towards Personalized Medicine
Electronic health records are poised to play a critical role in health care’s move to more personalized medicine. Widespread adoption and use of EHRs is “a necessary step” to getting to personalized medicine, Abel Kho, Associate Director of the Medical Informatics Program at Northwestern University’s Feinberg School of Medicine, said in an interview. “We will never get there on paper,” he said. Kho was lead author of a study report on the utility of EHRs in genomics research. The study, conducted by the Electronic Medical Records and Genomics Network consortium, looked at EHRs at five sites and whether the data captured could be used to accurately identify patients with particular diseases – dementia, cataracts, peripheral arterial disease, type 2 diabetes, and cardiac conduction defects – for inclusion in genome-wide association studies, which involve scanning markers across the complete sets of DNA of many people to find genetic variations associated with diseases. Results of the eMERGE consortium’s work, published April 20 in the American Association for the Advancement of Science’s Science Translational Medicine publication, show that “even though the electronic medical records were different types and did not all use natural language processing (NLP) to extract information from records, they were able to obtain robust positive and negative values for identifying patients with these diseases with sufficient accuracy for use in GWAS. They conclude that widespread adoption of electronic medical records will provide real-world clinical data that will be valuable for GWAS and other types of genetic research.” eMERGE looked at EHRs from Group Health Cooperative, Marshfield Clinic Research Foundation, the Mayo Clinic, Northwestern University and Vanderbilt University. Each had a unique EHR platform, and there was a mix of internally developed programs and external vendor-supplied packages. All meet or exceed current meaningful use requirements as defined by CMS, though there was variation in what was captured in a structured format. For example, only one vendor-developed EHR captured family history as a structured data point, while the others included that in the free-text comment sections that providers could populate with information. The study sites also had separate DNA biorepositories to house biological samples for genotyping. “Despite variations in categories and completeness of data capture across sites, four of the five study sites achieved [positive predictive value] of close to 100% for use of [electronic medical record] data alone to identify their primary disease phenotype,” the report says. “One site achieved a lower PPV of 73% using EMR data to identify cases with dementia.” Effectively Mining Non-Structured Data A key finding of the study was that data was not only mined effectively from different kinds of EHRs, but that current tools for extracting information from non-structured physician’s notes were able to demonstrate the utility of using EHRs for genetic research. “We performed a comparison of the number of cases identified using structured data alone compared with that using structured data and NLP at one site,” the authors wrote. “At this site, the use of NLP tools identified 129% more cases” using qualitative measures related to cardiac conduction defects “than did the use of structured data and string matching only, while maintaining a PPV of 97%.” “The observation that NLP tools allowed identification of 129% more cases … only emphasizes the value of information captured in free text,” the report says. Researchers have been looking at electronic health records as a key source of information, particularly in the comparative effectiveness research arena, and have suggested EHRs should be designed with more structured data elements to increase their utility in conducting research (Also see "Meaningful Electronic Health Records Needed For Comparative Effectiveness" - Pink Sheet, 11 May, 2009.). However, Kho, a practicing clinician, noted that as EHRs evolve, it will be important to not include too many requests for structured information. “I think that although we would want [EHRs] to capture structured data where it makes sense, we also don’t want to burden clinicians with too much of having to try to click boxes,” Kho said. “They should be taking care of the patients, and it’s really up to us to come up with better tools to be able to get data out from electronic health records.” Kho highlighted the potential for getting this information from EHRs to help improve health outcomes. “One thing that’s interesting about this kind of trial where we’re linking phenotypes and genotypes is that pharmacogenomics is a particularly interesting aspect of this type of genome-wide association study, [because] that’s where we see there being some real potential on clinical care,” he said. “Right now, when we pick medications, we don’t have that much information to go on. But if we had information about what kind of patients respond to medications versus which ones didn’t, based on their genetic profile, that would have tremendous implications.” Despite the potential for using EHRs in personalized medicine, privacy concerns remain (Also see "Privacy Laws Remain Hurdle To Utilizing Personalized Medicine Through HIT" - Pink Sheet, 7 Feb, 2011.). Kho noted that the study was conducted following Health Insurance Portability and Accountability Act patient privacy requirements. |