Us­ing soft­ware to avoid mis­di­ag­noses

Modern Healthcare - - BEST PRACTICES - By Sabriya Rice

In 2012, a pre­teen en­tered the emer­gency depart­ment at Chil­dren’s Na­tional Med­i­cal Cen­ter in Washington, D.C., with a high fever, weight loss and di­ar­rhea. The at­tend­ing pe­di­a­tri­cian sus­pected a rare, travel-re­lated in­fec­tious dis­ease, as the child had just been in South­east Asia.

Spe­cial­ists combed through the Cen­ters for Dis­ease Con­trol and Preven­tion web­site for health alerts from the re­gion. They ran tests. All came back neg­a­tive.

Only af­ter two days did the hos­pi­tal turn to a di­ag­nos­tic de­ci­sion-sup­port soft­ware pro­gram called Is­abel. Once a clin­i­cian typed in the child’s symp­toms, a list of po­ten­tial con­di­tions popped up within sec­onds.

The spe­cial­ists quickly re­al­ized their ini­tial mis­take. “Bi­ased by the pa­tient’s travel history, we didn’t con­sider a pretty straight­for­ward di­ag­no­sis,” re­called Dr. Paul Man­i­cone, as­so­ciate chief of the hospi­tal­ists di­vi­sion for Chil­dren’s Na­tional Health Sys­tem. Is­abel di­rected the team’s at­ten­tion to hy­pothy­roidism, a con­di­tion they had over­looked.

Soft­ware pro­grams such as Is­abel have been avail­able for decades. “And they have been un­der­uti­lized for decades,” said Dr. Mark Graber, a mem­ber of the Na­tional Academy of Medicine’s Com­mit­tee on Di­ag­nos­tic Er­ror in Healthcare. “That’s the real shame.”

In Septem­ber, the com­mit­tee called di­ag­nos­tic mis­takes a per­sis­tent blind spot amid ef­forts to im­prove qual­ity and safety in healthcare set­tings. The re­port rec­om­mended re­search into com­put­eras­sisted di­ag­no­sis tech­nol­ogy.

Di­ag­no­sis-as­sist pro­grams are par­tic­u­larly help­ful in dif­fi­cult or rare cases. Users en­ter symp­toms, such as fever, ab­dom­i­nal pain and skin rash, and pa­tient char­ac­ter­is­tics, such as med­i­cal history and gen­der. The soft­ware gen­er­ates a list of di­ag­noses and ranks them in or­der of like­li­hood.

Stud­ies sug­gest the pro­grams have been suc­cess­ful in boost­ing physi­cians’ di­ag­nos­tic con­fi­dence, re­duc­ing costs and im­prov­ing pa­tient out­comes. At Chil­dren’s Na­tional, check- lists gen­er­ated by Is­abel pointed to the cor­rect di­ag­no­sis in 43% of pe­di­atric ail­ments, ac­cord­ing to pre­lim­i­nary find­ings shared with Mod­ern Healthcare.

In 2000, DX­plain, a sim­i­lar tool, was of­fered to gen­eral medicine res­i­dents at the Mayo Clinic’s St. Marys Hos­pi­tal in Rochester, Minn. Clin­i­cians were en­cour­aged to use it for “di­ag­nos­ti­cally chal­leng­ing” cases.

A sub­se­quent anal­y­sis found DX­plain helped re­duce costs for Medi­care Part A pa­tients by an av­er­age of $900.

If used na­tion­wide, the sav­ings for Medi­care pa­tients alone could ap­proach $100 mil­lion an­nu­ally, ac­cord­ing to a 2010 study in the In­ter­na­tional Jour­nal of Med­i­cal In­for­mat­ics.

De­spite such find­ings, “it’s been a hard sell,” ac­knowl­edged Dr. Ed­ward Hof­fer of Mas­sachusetts Gen­eral Hos­pi­tal’s Com­puter Science Lab. He’s part of the team that launched DX­plain in 1985. “The main prob­lem we face is try­ing to con­vince over­con­fi­dent physi­cians that they ac­tu­ally need to use it.”

Physi­cian over­con­fi­dence is a ma­jor stum­bling block to wider use. A 2013 study in JAMA In­ter­nal Medicine found over­con­fi­dence may pre­vent physi­cians from re-ex­am­in­ing dif­fi­cult cases in which their di­ag­no­sis is more likely to be in­cor­rect. Un­war­ranted cer­ti­tude “is the big chal­lenge we face,” Hof­fer said.

Even the Mayo hos­pi­tal that had suc­cess with DX­plain no longer uses it. But Hof­fer said about 30 U.S. med­i­cal schools as well as 15 hos­pi­tals, clin­ics and healthcare sys­tems li­cense the prod­uct, which ranges in cost from about $500 a year for a small prac­tice to $12,000 for a large sys­tem.

Don­ald Bau­man, CEO of Is­abel Healthcare, which launched the tool in 2004, says it is used by about 130 U.S. or­ga­ni­za­tions, in­clud­ing physi­cian prac­tices, med­i­cal schools, hos­pi­tals and health sys­tems. A sub­scrip­tion for Is­abel can range from $299 a year for an in­di­vid­ual to as much as $15,000 for a large prac­tice of 75 providers or more.

Doc­tors at Chil­dren’s Na­tional ini­tially re­sisted Is­abel, but em­braced it af­ter the fa­cil­ity fo­cused on clin­i­cal ar­eas with the high­est risk of di­ag­nos­tic mis­takes, such as in the in­ten­sive-care unit, the emer­gency depart­ment and among hospi­tal­ists. “These are ar­eas where pa­tients get large workups and fo­cus is on the di­ag­nos­tic process,” Man­i­cone said. “That’s dif­fer­ent from ar­eas like car­di­ol­ogy, where the di­ag­no­sis has al­ready been made and the fo­cus is on man­ag­ing the dis­ease.”

Physi­cians ar­gue they don’t have time, but us­ing the pro­gram takes only a few min­utes, and can have a big im­pact on pa­tient care, said Graber, an en­thu­si­as­tic pro­po­nent of the soft­ware. Just be­cause a clin­i­cian is 100% sure of a di­ag­no­sis “doesn’t mean we’re right.”

Newspapers in English

Newspapers from USA

© PressReader. All rights reserved.