EHRs Drive Better Care
HCA’s Jonathan Perlin, MD, feels compelled to have the company’s record system support better care for patients.
HCA’s Jonathan Perlin: Data collected from care records informs and improves patient interactions and system operations.
When U.S. veterans in New Orleans were displaced by Hurricane Katrina, their medical records remained safe and accessible to all Veterans Affairs health providers, wherever the veterans ended up moving. That was quite an accomplishment in 2005, when integrated EHRs were rare. But the pride that former VA executive leader Jonathan Perlin, MD, felt in his health system was disrupted by an IT leader’s comment: “That’s great, but how are you learning?” Thirteen years later, that question still haunts Perlin, who advocates for using EHRs not only to store and share patient information but to drive improvement. “At HCA, we didn’t just want to install a new electronic health record, we wanted to be a learning organization, which means using data captured from patient care to continually inform and improve every patient interaction and system operation.”
On learning organizations
Computerization and the standardization of data elements under meaningful use is allowing us to learn at a much faster pace than before. A good example was our Reduce MRSA [methicillin-resistant Staphylococcus aureus] study. We had 43 hospitals over 18 months compare three potentially equal strategies for reducing MRSA among 75,000 ICU patients. The best approach turned out to be giving ICU patients a sponge bath with an antiseptic and applying an antibiotic in the nose. This cut MRSA by 37 percent and all potentially life-threatening infections by 44 percent, resulting in a new, worldwide standard of care. We went on to look at reducing infections on non-ICUs, following 524,000 patients at 53 HCA hospitals. We found the antiseptic sponge bath and nasal antibiotic uniquely effective in patients with implanted vascular access lines and other devices, the group in which most infections occur.
On what data to keep
Our philosophy is that we don’t know what data might be important to have in the future. This is particularly important with molecular and genomic medicine. We think we understand certain single gene mutations but we don’t know, for example, the relationship of those mutations to environmental factors or internal factors, such as the bacteria in our guts. So we’re trying to use data science to look for patterns that aren’t necessarily intuitive relationships.
On unstructured data
We’re excited to be beginning to work with unstructured data, such as text and images, which make up 70 to 80 percent of the patient chart. We are using natural language processing software to read unstructured data in pathology and radiology reports. This is allowing our cancer patient navigators to work through lab reports 23 times faster. Patients often have to wait longer than they should to get results of a biopsy to rule out or confirm cancer. After the radiologist or pathologist issues a report, another human typically has to read that report. The program we’re using can quickly scan reports and alert navigators of the results so they can inform patients sooner. □