EHRs Drive Bet­ter Care

HCA’s Jonathan Per­lin, MD, feels com­pelled to have the com­pany’s record sys­tem sup­port bet­ter care for pa­tients.

Health Data Management - - CONTENTS - BY MAG­GIE VAN DYKE

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. □

Newspapers in English

Newspapers from USA

© PressReader. All rights reserved.