Seek­ing a so­lu­tion for pa­tient record match­ing

Modern Healthcare - - LEGAL - By Joseph Conn

En­sur­ing providers ac­cu­rately match elec­tronic health records with the right pa­tient has emerged as a hot-but­ton is­sue for the health in­for­ma­tion tech­nol­ogy in­dus­try as its provider cus­tomers come un­der in­creas­ing pres­sure to freely share records with other providers.

Many providers’ EHR sys­tems, of­ten pur­chased from dif­fer­ent ven­dors, are in­ca­pable of ac­cu­rately match­ing elec­tronic pa­tient med­i­cal records with­out costly and time-con­sum­ing hu­man in­ter­ven­tion to re­solve con­fu­sion over pa­tients with sim­i­lar names. The re­sult is a fail­ure to max­i­mize the clin­i­cal care and pa­tient-safety ben­e­fits of EHRs. It also lim­its fu­ture re­search op­por­tu­ni­ties, ex­perts say.

Ac­cord­ing to sur­veys and pub­lished re­ports, the er­ror rates for au­to­matic pa­tient match­ing range from 1% to over 50%, de­pend­ing on whether a re­quest­ing provider is query­ing pa­tient records within an or­ga­ni­za­tion, from another provider or from a health in­for­ma­tion ex­change.

For decades, some health IT pro­po­nents have pushed for a na­tional pa­tient iden­ti­fier sys­tem as the best so­lu­tion. But pub­lic op­po­si­tion and a lack of sup­port in Wash­ing­ton, D.C., has kept estab­lish­ment of an NPI on the back­burner un­til re­cently.

“Peo­ple re­al­ize how the lack of an iden­ti­fi­ca­tion so­lu­tion that’s univer­sal is com­pound­ing the prob­lems with elec­tronic health in­for­ma­tion ex­change,” said Rus­sell Branzell, CEO of the Col­lege of Health­care In­for­ma­tion Man­age­ment Ex­ec­u­tives, a pro­fes­sional as­so­ci­a­tion for hos­pi­tal and health­care sys­tem chief in­for­ma­tion of­fi­cers. “It’s prob­a­bly as big an is­sue as we have out there.”

This year’s Health­care In­for­ma­tion and Man­age­ment Sys­tems So­ci­ety’s an­nual con­fer­ence, Feb. 29 through March 4 in Las Ve­gas, will fea­ture at least four for­mal ed­u­ca­tional ses­sions on pa­tient-match­ing tech­nol­ogy and best prac­tices. Pre­sen­ters will in­clude Adam Cul­bert­son, the HIMSS-spon­sored “innovator in res­i­dence” at HHS’ Idea Lab, who will ad­dress match­ing prob­lems, best prac­tices to solve them and why match­ing well is more of an art than a sci­ence.

Tom Leary, vice pres­i­dent of govern­ment re­la­tions for the Chicago-based HIMSS, said fu­ture re­search—par­tic­u­larly in the highly touted field of pre­ci­sion medicine— will de­pend heav­ily on get­ting match­ing right. “If you’re go­ing to have a mil­lion peo­ple par­tic­i­pat­ing in a co­hort, and gain the learn­ing that’s go­ing to oc­cur from that mil­lion-per­son co­hort, you have to be able to fig­ure out where all their data is,” Leary said. But the most im­por­tant rea­sons for fix­ing the match­ing prob­lem in­volve max­i­miz­ing pa­tient safety and im­prov­ing clin­i­cal de­ci­sion­mak­ing in an era of pay­ment re­form. “In terms of value-based care, in­ter­op­er­abil­ity is key—and we’re not go­ing to get to where we need to get to with­out this pa­tient match­ing” be­ing ad­dressed, said Al­bert Oriol, CIO of Rady Chil­dren’s Hos­pi­tal, San Diego.

The record-match­ing chal­lenge at Rady is most acute with new­borns, par­tic­u­larly chil­dren of a mul­ti­ple birth who may need ex­tended or in­ten­sive care. “In many cases, when the baby is born, the birth hos­pi­tal doesn’t even have

a name yet,” Oriol said. “It may be ‘baby girl A’ or ‘baby girl B.’ By the time the par­ents come up with a name, that baby is al­ready in our” neona­tal in­ten­sive-care unit.

Hos­pi­tals han­dle it in­ter­nally by is­su­ing pa­tients an in-house pa­tient iden­ti­fier. But that’s of lit­tle help when that pa­tient’s in­for­ma­tion needs to be matched with records from another or­ga­ni­za­tion later in life. “It fol­lows them from birth,” Oriol said. “They may be John David Smith and John Daniel Smith. So they end up with the same first and last names, same ini­tials, same par­ents, same ad­dress and same data of birth.”

To en­sure ac­cu­rate match­ing, the sys­tem has to have per­son­nel re­view about 1% or 2% of its records, Oriol said. But with a health in­for­ma­tion ex­change, “we’re up to 9% with these things that re­quire peo­ple to in­ter­vene and match. … It’s not ideal.”

More­over, some­times by the time the match­ing ques­tion is re­solved, the physi­cian has al­ready seen the pa­tient and lost the op­por­tu­nity to in­cor­po­rate the ex­ter­nal in­for­ma­tion into their de­ci­sion­mak­ing. “We’d like to have that in­for­ma­tion flow­ing like when you open a tap at your house,” he said.

Katherine Lusk, chief health in­for­ma­tion man­age­ment and ex­change of­fi­cer at Chil­dren’s Health, Dal­las, said her hos­pi­tal has had a qual­ity as­sur­ance and data gov­er­nance pro­gram for pa­tient match­ing for many years. Chil­dren’s Health is one of 11 or­ga­ni­za­tions par­tic­i­pat­ing in a pi­lot by the Amer­i­can Health In­for­ma­tion Man­age­ment As­so­ci­a­tion to cre­ate a na­tional in­for­ma­tion gov­er­nance model, which will in­clude guid­ance on pa­tient match­ing.

The model, sched­uled for pub­lic re­lease this year, will in­clude a tool to iden­tify gaps in an or­ga­ni­za­tion’s pa­tient-match­ing ap­proach, such as in­ad­e­quate train­ing, Lusk said.

Cost is also another rea­son to fix the match­ing prob­lem, said Marc Probst, vice pres­i­dent and CIO at In­ter­moun­tain Health­care, Salt Lake City. The in­te­grated de­liv­ery sys­tem stud­ied it a few years ago and es­ti­mated it could save be­tween $4 mil­lion to $5 mil­lion a year sim­ply by do­ing a bet­ter job of match­ing records.

In­ter­moun­tain re­cently an­nounced re­sults of a pa­tient-match­ing im­prove­ment project with its statewide HIE, the Utah Health In­for­ma­tion Net­work, which in­cludes Univer­sity of Utah Health Care, its chief ri­val. “The Univer­sity of Utah came up with a pretty good al­go­rithm that got us closer,” Probst said. But much of the im­prove­ments came by virtue of par­tic­i­pants reach­ing con­sen­sus on do­ing ba­sic things the same way, such as agree­ing on a larger than nor­mal num­ber of data fields to use for match­ing.

It in­cluded not only the first and last names, sex and dates of birth, but also So­cial Se­cu­rity num­bers, home phone num­bers, race and home ad­dresses. The project stan­dard­ized how those at­tributes are recorded, which im­proved the rate of au­to­matic match­ing across the ex­change to 95% from 10%.

Dan Chavez, ex­ec­u­tive di­rec­tor of San Diego Health Con­nect, that South­ern Cal­i­for­nia com­mu­nity’s lo­cal health in­for­ma­tion ex­change, is sched­uled to talk at HIMSS about pa­tient match­ing and how his HIE over­came “the largest ob­sta­cle to EHR ex­change.”

Chavez es­ti­mates that 30% of EHRs have ba­sic data on pa­tient iden­ti­ties such as names, ad­dresses or So­cial Se­cu­rity num­bers that are old, in­com­plete or in­cor­rect and there­fore can’t be matched across providers with­out man­ual in­ter­ven­tion. The San Diego HIE cut its man­ual ef­fort by 75% and dou­bled its match­ing ac­cu­racy us­ing third-party data.

Ear­lier this month, CHIME kicked off a Na­tional Pa­tient ID Chal­lenge with HeroX, an on­line plat­form to pro­mote chal­lenges linked to tech in­no­va­tions. They’re of­fer­ing $1 mil­lion in prize money—to be raised by CHIME and other spon­sors—to en­cour­age de­vel­op­ers to ad­dress pa­tient match­ing on a na­tional scale with 100% ac­cu­racy.

Physi­cian in­for­mati­cist Dr. Barry Heib said his not­for-profit or­ga­ni­za­tion, Global Pa­tient Iden­ti­fiers, based in Tuc­son, Ariz., will be a CHIME prize com­peti­tor. The GPI ap­proach is to gen­er­ate and store unique pa­tient iden­ti­fiers, but not store pa­tient med­i­cal records them­selves. A GPI data­base will keep track of those providers that have records for each stored iden­ti­fier. Users of the sys­tem will be charged a few cents each time they query it for the where­abouts of a pa­tient’s records.

Im­prov­ing pa­tient match­ing will take far more than a tech­nol­ogy up­grade, im­por­tant as that may be, said Dr. Charles Jaffe, CEO of Health Level 7, a health­care stan­dards de­vel­op­ment or­ga­ni­za­tion.

“How­ever good pa­tient-match­ing al­go­rithms are, I would never ar­gue that you have the per­fect al­go­rithm,” Jaffe said. “It’s re­ally con­tin­gent on the qual­ity of the de­mo­graphic data you’ve col­lected and that tends to be er­ror-filled. Even if we do get a pa­tient iden­ti­fier, we’ll still have prob­lem with data en­try in the iden­ti­fier.”

Hos­pi­tals han­dle it in­ter­nally by is­su­ing pa­tients an in-house pa­tient iden­ti­fier. But that’s of lit­tle help when that pa­tient’s in­for­ma­tion needs to be matched with records from another or­ga­ni­za­tion later in life.

GETTY IM­AGES

Newspapers in English

Newspapers from USA

© PressReader. All rights reserved.