focus Quality Outcomes ad Patiet Safety CPOE Cofiguratio to Reduce Medicatio Errors A Literature Review o the Safety of CPOE Systems ad Desig Recommedatios By Patricia Segstack, DNP, RN-BC, CPHIMS Keywords Adverse drug evets, cliical iformatio systems, CIS, computerized provider order etry, CPOE, cliical decisio support, CDS, huma computer iteractio, HCI, huma factors. medicatio error, usability. Abstract ARRA has allocated $20 billio to stimulate the adoptio of health IT over the ext several years. The largest share of the fudig will be for icetive paymets through CMS to ecourage providers ad hospitals to implemet health IT systems that demostrate meaigful use that icludes the implemetatio of CPOE systems. Ufortuately, few tools exist to provide iformatics specialists best practice cofiguratio guidelies whe desigig or evaluatig CPOE systems to reduce medicatio errors. This article icludes a cosolidated source of the curret literature o the safety of CPOE systems ad offers desig recommedatios based o the literature. A 46-item CPOE desig checklist is provided as a tool that ca be used durig software selectio, desig or evaluatio. Utilizatio of a checklist such as this, coupled with a strog evaluatio program, ca provide orgaizatios with the best chace of reachig their error reductio goals. Less tha 10 percet of the atio s hospitals have implemeted computerized provider order etry (CPOE) as part of their cliical iformatio system. 1 Oe may woder why, give the techological pervasiveess, igeuity ad cotiuous advacemets see i this coutry. It also is iterestig ad perplexig that the first CPOE implemetatio occurred i 1971 almost 40 years ago ad yet adoptio remais low. 2 This issue becomes icreasigly relevat give that ARRA, siged ito law by Presidet Obama i February 2009, allocated $20 billio to stimulate the adoptio of health IT. The largest share of the fudig, approximately $17 billio, will be for icetive paymets through Medicare ad Medicaid reimbursemet systems to ecourage providers ad hospitals to implemet health IT systems that demostrate meaigful use, ad requires the implemetatio of CPOE systems. 3 This ivestmet i health IT calls for a clear uderstadig of the issues surroudig the implemetatio of CPOE systems ad tools to evaluate their effectiveess. Whe lookig for potetial reasos for the slow adoptio of CPOE systems, cost ad complexity appear to be factors, but perhaps aother reaso is that there has bee relatively little 26 jhim
evidece (ad o guaratee) that they reduce errors a primary marketig sloga. As these systems have emerged ad have become more sophisticated over the last decade, several studies have bee coducted that idicate a reductio i medicatio errors. O the other had, there recetly have bee studies providig evidece that these systems ca actually icrease errors, which are attributed to poorly desiged CPOE implemetatio ad system cofiguratio. Both types of studies provide valuable iformatio to orgaizatios as they udertake CPOE implemetatio ad its evaluatio. A sigificat problem appears to be that there is o cosolidated source of evidece that provides orgaizatios with tools to desig the safest CPOE system possible. The purpose of this article is twofold: To offer a compilatio of the CPOE literature as it relates to the reductio of medicatio errors; ad to offer a CPOE desig checklist based o the literature that ca be used by all levels of iformatics specialists to promote error reductio. What the Literature Says For almost a decade, The Istitute of Medicie s To Err is Huma has bee a drivig force for improvemets i patiet safety atiowide. The research quoted clearly idicates the seriousess of the situatio i terms of lives lost ad moey wasted as a result of medicatio errors. 4 Solutios to this problem have bee offered by may authors, orgaizatios ad politicias, but the resoudig theme is the use of techology ad the applicatio of CPOE systems. Searchig peer-reviewed jourals for studies that idetify specific CPOE compoets that lead to the best chace of medicatio error reductio is challegig. Most articles usig the search terms of hospital iformatio systems or CPOE ad medicatio errors were quasi-experimetal studies that either supported CPOE with a outcome demostratig a reductio i medicatio errors, or with a outcome demostratig a icrease i medicatio errors. Both types of studies are reviewed here with the belief that each offers a perspective o CPOE system cofiguratio that has the potetial to add value to a growig body of kowledge. Evidece Supportig CPOE Multiple studies have cocluded that CPOE systems reduce medicatio errors. I a systematic review by Kaushal, Shojaia ad Bates, a total of 12 studies were reviewed. Icluded were two oradomized cotrolled trials, oe radomized cotrolled trial, three observatioal studies with cotrols ad four observatioal studies without cotrols. 5 Five of the 12 studies evaluated CPOE systems with cliical decisio support (CDS), while the remaiig seve were studies evaluatig a CDS system without CPOE (primarily, atibiotic maagemet systems). Overall fidigs reported that CPOE ad CDS systems ca reduce medicatio error rates, but idetified that more research is eeded to evaluate commercial systems give the majority of systems i this review were built i house. The cofiguratio icluded i the CPOE group with CDS was cosidered basic, i.e. drug-allergy checkig ad drug-drug iteractio checkig. I the studies that ivolved CDS systems oly, cofiguratio icluded atibiotic drug advice ad dosig programs to assist the physicia with appropriate dose orderig. Two additioal systematic reviews reported similar fidigs, yet make the appeal for higher quality i study methodology ad reportig. 6,7 A study by Frakli, O Grady, Doyai, Jackli ad Barber foud a sigificatly positive impact o error reductio post implemetatio of a closed-loop electroic prescribig ad admiistratio system o prescribig errors. 8 This study icluded bar coded admiistratio of medicatios, but did ot provide details regardig the cofiguratio of the CPOE system that supports the bar code techology. Shulma, Siger, Goldstoe ad Belliga collected data o medicatio errors from a CPOE system ad compared it to traditioal had-writte medicatio orders. 9 They foud a sigificat reductio i errors with the use of CPOE, eve without cliical decisio support. Compoets of the CPOE cofigured i this study icluded a full audit trail, madatory data fields such as dose, uits, frequecy ad sigatures. They also reported that dose errors were still prevalet with CPOE as a result of physicias choosig a icorrect drug template or selectig from multiple optios i a drop dow list. Oe prospective cohort study i a pediatric ICU foud a sigificat reductio i prescribig errors after the implemetatio of a CPOE system. 10 This homegrow system cotaied several types of cliical decisio support icludig drug-allergy alerts, dose checkig, drug iteractio alerts, order sets ad liks to evidece-based literature sites such as the Natioal Library of Medicie s PubMed site. Cotiuig with studies that support the use of CPOE systems, Raebel et al., coducted a radomized trial icludig more tha 11,000 wome erolled i Kaiser Permaete ceters i the Dever area. 11 They were attemptig to determie if a computerized physicia orderig system with alerts for drugs cotraidicated durig pregacy could reduce the umber of potetially harmful situatios for mother ad fetus. Their fidigs were statistically sigificat i favor of the ew system but it did ot completely stop iappropriate orderig. The recommedatio was made to couple data from the iformatio system with the kowledge ad skill of physicias ad pharmacists. Kim, et al. evaluated safety of chemotherapy admiistratio to pediatric patiets before ad after implemetatio of a CPOE system, lookig at data that spaed over a three year course of time. 12 They icorporated the process of Failure Mode ad Effects Aalysis (FEMA) ito the scree desig for chemotherapy orderig. This icluded limitig choices by pre-defied meus istead of free-text etry, madatory etry of height ad weight data, provisio of alerts for abormal values, eforcemet of protocol specificatios ad automated calculatios. Results idicated that after CPOE deploymet, daily chemotherapy orders were less likely to have improper dosig. I aother study by Vaidya, et al., the safety ad efficiecy of a CPOE system was evaluated for cotiuous medicatio ifusios i a pediatric ICU 13. This was aother homegrow system that was desiged for orderig cotiuous drug ifusios for pediatric ICU patiets. The study was coducted i a simulated test eviromet with scearios to test the safe orderig of IV ifusios. Cofiguratio of this system icluded default stadardized cocetratios based o the patiet s daily height, weight ad www.himss.org volume 24 / umber 4 FALL 2010 jhim 27
fluid itake. All calculatios were performed automatically by the system. They foud a statistically sigificat reductio i errors ad cocluded that this type of system will ehace user acceptace ad rapid adoptio by the physicias. Evidece Citig Issues with CPOE While the studies cited above support the beefits ad ehaced safety features that ca be realized with a CPOE system, there have bee recet articles published that idicate a potetial egative side of CPOE. A popular, cotroversial study that was published i the Joural of the America Medical Associatio i 2005 by Koppel, et al., reported that CPOE systems actually facilitate medicatio errors. 14 This article ra cotrary to what most believed, but idetified, usig quatitative ad qualitative methods, 22 types of medicatio error risks associated with the use of CPOE. These risks were divided ito two categories: iformatio errors: fragmetatio ad systems itegratio failure; ad huma-machie iterface flaws: machie rules that do ot correspod to work orgaizatio or usual behaviors. While may of the 22 types of error risks ivolve workflow process, there were some that were attributed to poor system desig icludig the iability to see all the patiet s curret medicatios listed o oe scree, the ease i which oe ca select the wrog patiet due to a small fot size ad alphabetic sortig of patiet ames. This study clearly poits out that huma factors eed to be cosidered whe seekig to create a CPOE system to reduce errors. I oe qualitative study by Campbell, Sittig, Ash, Guappoe ad Dykstra, researchers attempted to idetify the types of uiteded cosequeces see with the implemetatio of CPOE systems. 15 This study ivolved a expert pael usig a iterative process that took a list of adverse cosequeces of CPOE, ad sorted them ito categories. The category labeled Geeratio of New Kids of Errors icluded orgaizatioal culture ad busiess process or workflow problems, but also idicated that ew kids of errors appear whe CPOE is implemeted. Examples of items i this category iclude juxtapositio errors whe users select a item ext to the iteded choice; a wrog patiet beig selected; desesitizatio to alerts (alert overload); cofusig order optio presetatios; ad system desig issues with poor data orgaizatio ad display. Users get frustrated tryig to fid the right spot to eter a particular data elemet ad ed up eterig orders o geeric screes, bypassig ay rules ad alerts cofigured; all potetially leadig to medicatio errors. This study ad a compaio study by Ash, Sittig, Poo, Guappoe, Campbell ad Dykstra, that ivestigated the extet ad importace of uiteded adverse cosequeces of CPOE, do ot provide statistical aalysis to idicate that these items are the true cause of medicatio errors; rather they provide iformatio based o qualitative methods about categories of potetial issues. 15,16 I a retrospective study by Walsh, et al., researchers attempted to determie the frequecy ad types of pediatric medicatio errors attributable to desig features of CPOE cofiguratio. 17 The rate of idetified computer related errors was 10 errors per 1,000 patiet days, ad the rate of serious computer related errors was 3.6 errors per 1,000 patiet days. The mai types of errors idetified icluded duplicate medicatio orders, drop dow meu selectio errors, icorrect abbreviatios ad order set errors (orders selected from the order set that were ot cliically appropriate for the patiet). No compariso was made betwee the data obtaied ad the previous state of errors. They cocluded that serious pediatric computer-related errors were ucommo, but preset ad recommeded the refiemet of CPOE system desig features. I aother retrospective chart audit at a 658-bed academic hospital, 6,019 medicatio admiistratio evets were studied. This study utilized descriptive statistics ad o sigificace was reported, yet the coclusio was that medicatio admiistratios do ot cosistetly occur as ordered with a electroic CPOE system. Recommedatios icluded the avoidace of duplicate medicatio orders, developig methods to alert urses whe medicatio orders have bee etered by the physicia ad developig a safe meas for urses to shift medicatio dosig schedules. 18 Additioally, Berger & Kichak, dowplay the ofte quoted studies cited i the IOM report, To Err is Huma. 19 They report that the methodology of these studies has bee challeged ad cite the fact that they were coducted i the 1980s, makig them ot relevat i today s more techologically advaced world. This article coteds that there are relatively few cotrolled studies that provide evidece that these types of systems decrease serious adverse drug evets ad poits out that medical facilities cosiderig the implemetatio of CPOE should realize that the literature o the medical ad ecoomic effects are still evolvig. Evidece from Experts or Opiio Leaders o CPOE Additioal iformatio surroudig CPOE ad medicatio errors ca be foud i the literature i the form of expert advice ad opiio. I a article by Kuperma, et al., the role of CDS combied with CPOE is recommeded to reduce medicatio errors ad improve patiet safety. 20 This article while iformative ad providig good backgroud iformatio does ot provide evidece that ay of the recommedatios actually reduce medicatio errors. They suggest that cliical decisio support should iclude the basic cofiguratio of drug-allergy checkig, basic dosig guidace, formulary decisio support, duplicate therapy checkig ad drug-drug iteractio checkig, as well as advaced cofiguratio of drug-pregacy checkig, drug-disease cotraidicatio checkig ad guidace for medicatio related laboratory testig. These cocepts of advaced CDS cofiguratio are echoed i a article by Bobb, et al., i additio to recommedig pharmacist ivolvemet i the medicatio orderig process. 21 I 2001, 13 experts o CPOE from aroud the world gathered at a two-day coferece to develop a cosesus statemet o successful CPOE implemetatio. This qualitative research approach was used to geerate ad validate a list of categories ad cosideratios to guide CPOE implemetatios. 22 I additio to workflow, traiig ad techology cosideratios, a list of CDS cofiguratio items was reported to improve patiet care quality. These iclude: drug-drug iteractio, drug-allergy iteractio, duplicate medicatio checkig, drug level moitorig, order sets ad active surveillace i place. 28 jhim
Aother CPOE opiio leader, The Leapfrog Group, represets a coalitio of healthcare purchasers that has bee a drivig force i the improvemet of healthcare quality ad advocate use of CPOE systems. They have developed a CPOE stadard icludig a requiremet that orgaizatios operatig CPOE systems should demostrate (via testig scearios) to esure that their ipatiet CPOE system ca alert physicias to at least 50 percet of commo serious prescribig errors. I a article by Kilbridge, Welebob ad Classe, the developmet of these stadards are described icludig the descriptio of a framework with 12 differet categories of CPOE-based decisio support that have the potetial to prevet a prescribig error. 23 Orgaizatios are asked to coduct this test i a developmet or practice database that is similar to their productio CPOE system. Cofiguratio elemets of the system that are tested iclude duplicate orderig, sigle- ad cumulative-dose levels, allergy checkig, drug-drug iteractio, cotraidicatios based o patiet diagosis, cotraidicatios based o relevat laboratory values ad dose levels based o radiology studies. This test is Web-based ad self-admiistered, but ca oly be take if the orgaizatio also participates i Leapfrog s geeral hospital survey. Evidece from the Field of Huma Computer Iteractio ad CPOE Several articles have highlighted the cocept of huma computer iteractio (HCI) or huma factors desig as a major cosideratio whe lookig at causes of errors. 24 HCI is a applied sciece that is focused o miimizig huma errors ad optimizig performace where huma beigs iterface with a device. 25 Gaier, Pacheri ad Zhag metio a heuristic evaluatio as a techique to idetify usability problems by idetifyig violatios of well-established ed user cofiguratio that good desig should follow. 26 This symposium proceedig idetified that a evaluatio of a commercial CPOE system foud umerous usability problems liked to user satisfactio ad patiet safety, although o study data were icluded. I a case study reported by Horsky, Kaufma ad Patel, usability studies were performed usig CPOE scearios ad foud that there are commercial systems that itroduce uecessary cogitive complexity leadig to errors. 27 They purport that poor cofiguratio ca lead to placig heavy cogitive demads o users ad they urge system desigers to utilize guidig priciples ad desig solutios whe they are developig complex iteractive systems i a effort to reduce errors. While the authors admit that what costitutes optimal cofiguratio may be a elusive cocept, they offer a coceptual methodology to evaluate cofiguratio of the user iterface that emphasizes clarity. Critical Review of Data Sources ad Methods I these studies reviewed, the measuremet of medicatio errors as they relate to the use of CPOE systems has bee less tha optimal; i fact, o two studies have used the same defiitio of what costitutes a medicatio error. Some studies collected data o actual adverse drug evets, while others collected data o medicatio errors that may or may ot have led to a adverse drug evet. I a study by Gadhi, et al., medicatio errors were defied as, Ay error that occurred i the medicatio use process icludig orderig/prescribig, dispesig, adherece, ad moitorig. 28 Shulma, et al. defied a medicatio error as havig occurred whe a prescribig decisio or prescriptio writig process resulted i either a uitetioal sigificat reductio i the probability of treatmet beig timely ad effective or a uitetioal sigificat icrease i the risk of harm whe compared with geerally accepted practice. 9 The defiitio provided by Potts, et al., stated that a medicatio error occurs whe a order is foud to be icomplete, icorrect or iappropriate at the time of physicia orderig. 10 The Natioal Coordiatig Coucil for Medicatio Error Reportig ad Prevetio, a idustry leader i error prevetio, provides the followig defiitio, A medicatio error is ay prevetable evet that may cause or lead to iappropriate medicatio use or patiet harm while the medicatio is i the cotrol of the healthcare professioal, patiet or cosumer. 29 Each study ad each professioal orgaizatio reviewed had a differet defiitio of what comprised a medicatio error. Istead of providig a global defiitio of a medicatio error, some studies provided defiitios of types of medicatio errors. Koppel, et al., categorized errors as beig oe of two types as previously metioed. 14 Shulma, et al. categorized medicatio errors ito 12 types icludig icorrect drug, drug eeded but ot give, dose error, icorrect route ad frequecy omitted. 9 These icosistecies ad lack of stadardizatio i measuremet hold true for the cocepts of medicatio error severity rakig ad measuremet of outcomes of adverse evets related to medicatio errors. May studies reviewed customized their defiitios of medicatio error, type, severity, cause ad outcome to reflect their particular orgaizatio s curret practices ad study methodology. All used varyig scales, scores ad tables of measuremet. No two studies appear to have used a similar methodology for data collectio or measuremet. It is ow clear why there have bee o meta-aalysis studies coducted related to medicatio errors ad CPOE. A few of the studies cite as a limitatio to their work, the fact that there are o stadards for measuremets or a commo taxoomy whe coductig ad reportig o medicatio errors ad recommed further study. Most studies clearly idicate that their results are ot geeralizable. May of the studies simply reported that they compared a CPOE system to a hadwritte system, omittig the details regardig the specifics of CPOE cofiguratio. May icluded the fact that they used various compoets of cliical decisio support, but the failed to metio how they were cofigured. The fact that a system is reported to have drug-allergy checkig cofigured does ot idicate whether or ot the alert will hard stop a user whe attemptig to order a medicatio that the patiet has a reported aaphylactic reactio. The fact that a system is reported to have duplicate order checkig cofigured does ot idicate whether the alert will fire if the same medicatio is ordered withi a two-hour widow, a 12-hour widow or a 24-hour widow. A typical descriptio of a orgaizatio s CDS compoets reported i studies reviewed is foud i a article by Cordero, Kueh, Kumar ad Mekhjia that states, Numerous cliical www.himss.org volume 24 / umber 4 FALL 2010 jhim 29
Table 1: CPOE Desig Checklist. Category Descriptio/Commets Source Cliical Decisio Support Allergies ad cross allergies. Duplicate order checkig. Display alert whe a allergy has bee documeted or a allergy to aother drug i same category is documeted. Provide alert of potetial allergy at time of order etry, ot order submissio. Display alert whe the same medicatio is ordered ad whe separate doses of the same medicatio are to be give withi a closely spaced time. [14, 23, 29, 31, 32, 33] [17, 18, 21, 23, 30, 33, 35, 36, 37] Sigle ad cumulative dose limits. Display alert whe ordered dose exceeds recommeded dose rages for both sigle ad cumulative doses. [14, 23, 30, 36] Cotraidicated route of admiistratio. Drug-drug iteractio. Drug-food iteractio. Display alert whe order specifyig a route of admiistratio that is ot appropriate for the ordered medicatio. (e.g. Atifugal topical cream ordered with route of IV) Display alert whe there is a potetially dagerous iteractio whe the medicatio is admiistered with aother medicatio. Display alert whe there is a potetially dagerous iteractio whe the medicatio is admiistered with a particular food group. [23, 30. 36] [23, 30, 32, 38] [23, 32, 34, 38] Cotraidicatio/drug - based o patiet diagosis. Display alert whe drug is cotraidicated based o the patiet s diagosis. [21, 23, 32, 33, 34] Cotraidicatio/dose limits - based o patiet diagosis. Display alert whe dose ordered is a cotraidicatio based o the patiet s diagosis or the diagosis affects recommeded dosig. [21, 23, 32, 33, 34] Cotraidicatios/dose limits -based o patiet weight. Display alert for over dosig or uder dosig based o patiet weight. [10, 17, 23, 30, 33] Iaccurate weight or height data etry (used to calculate medicatio dosages). Cotraidicatio /dose limits -based o laboratory studies. Cotraidicatio/dose limits -based o radiology studies. Potassium order appropriateess checkig. Display alert that wars cliicia that patiet weight or height just etered is 10% < or > the previous value etered. Cliicias ca accidetly add or delete a digit whe eterig a ew patiet weight (or eter lbs istead of kilos) ad the is subsequetly used for drug calculatios. Display alert whe a drug is ordered for which laboratory studies must be cosidered prior to dosig. [12, 23, 32, 33, 34, 37, 40] Display alert whe drug is ordered which may have a potetial iteractio with cotrast medium (i ordered radiology study). Ex, Metphorme Display alert iformig users orderig potassium whe there has ot bee a serum potassium value recorded i the past 12 hours or if the most recet potassium value is greater tha 4.0.This would reduce the likelihood of orderig potassium whe the patiet is hyperkalemic. Provide automated mechaism to calculate dosages. Do ot rely o maual calculatio of dosages. [33] [39] [23, 33] [35] Create a alert iformig users orderig potassium whe there has ot bee a serum potassium value recorded i the past 12 hours or if the most recet potassium value is greater tha 4.0. This would reduce the likelihood of orderig potassium whe the patiet is hyperkalemic. [35] Alerts should clearly idicate the problem ad offer a solutio. Alerts from Cliical Decisio Support should fire at the time of potetial error ad allow the user to recover without the loss of data etered. Error messages should be expressed i plai laguage (o codes), precisely idicate the problem, ad costructively suggest a solutio. Poor coceptual graphical represetatio of alerts makes it difficult for CPOE users to fid certai iformatio leadig to iefficiet searches for iformatio provided by the system. Loss of orders etered prior to alert require users to start over leadig to frustratio ad potetial loss ad iaccuracy durig repeat order etry. Order Form Cofiguratio Madatory fields for drug, dose, frequecy ad route. Caot submit order without completig these fields. [12, 32, 37, 43] All orderig screes should be desiged i a similar fashio. Fields for drug, dose, route, frequecy, etc should be i the same place o all screes. Do ot use field labels that require a egative aswer for a positive respose (e.g. Is IV cotrast cotraidicated?). Evaluate field labels for potetial user-desiger mismatch. All data elemets eeded for a order should have a etry field that is easy to fid ad screes should be cosistet ad easy to avigate. Iclude display of routie or default times for orders writte to be admiistered q4h, q6h, q8h, etc. If a drug dosage is calculated by the system, display the calculatio used to derive the dose. There should be cosistecy of scree cotrol, behavior, clarity of meaig ad data labels for fields. Order etry screes for IV bolus ijectios should be cofigured similar to those for IV fluids with medicatio additives. This is cofusig. Esure clear uderstadig of all data field labels. [39] Esure cosistet meaig of data labels ad make chages if ot clear (e.g. Total Volume of what?, Date for what?, etc). If a busy cliicia does ot immediately see where to eter orderig data, they may eter iformatio ito the commets or miscellaeous sectio. This creates problems ad potetial errors i categorizig, cross checkig ad applyig cliical decisio support. If ot clearly displayed, providers may ot be aware whe first dose of a medicatio will be admiistered ad may iappropriately order a STAT or Now dose. Although automated calculatios ca assist users i decidig upo a drug dose, without showig the calculatio used cliicias are sometimes proe to recalculate maually to check the computers work. [41, 42] [42] [35, 41] [35] [15] [18, 39, 43] [42] Cotiued o page 32 30 jhim
I light of the fact that sigificat ARRA fudig is beig allocated for the adoptio of CPOE systems, it is imperative that iformatics specialists at all levels, from ovice to expert, have quality tools to implemet ad evaluate these systems. decisio support tools are itegrated ito the orderig pathways. These iclude, but are ot limited to, drug allergy, drug-drug iteractios, order duplicatio, corollary orders, weight-based dosage, maximum dosage, ad drug route restrictio. 30 Aother study described their cofiguratio as the icorporatio of CPOE with a moderate level of CDS. 30 Three systematic reviews that i total aalyzed 50 CPOE studies all reflect o the heterogeeity of the methodologies ad systems uder study. 5-7 With the best way to cofigure a CPOE system yet to be defied, it becomes clear that we are at a phase i the evolutio of CPOE that still cotais much trial ad error work at the orgaizatioal level. It becomes clear that tools are eeded based o what has bee leared ad documeted i the literature; tools that ca be used to evaluate CPOE systems ad provide guidace to iformatics specialists as the iterative process of improvemet cotiues to cycle. The importace of this becomes paramout as govermetal icetives require orgaizatios to adopt CPOE systems. The CPOE Desig Checklist As hospitals ad care provider groups move forward i their jourey to adopt electroic health records with CPOE, they will eed resources ad tools to esure that systems are created to meet meaigful use criteria. Simply istallig a vedor s versio of a CPOE system or buildig oe i-house requires kowledge of best practices ad evidece from the literature that support the cofiguratio of a quality system. Meaigful CPOE cofiguratio eeds to be added to the list of required criteria alog with evidece of a ogoig evaluatio pla. The checklist offered i this article, has pulled quality evidece together i oe tool. It is iteded to be a istrumet for iformatics specialists at all levels that ca be used to coduct a assessmet of a CPOE system s cofiguratio from a error reductio stadpoit. It is easy to use ad based o available evidece. The tool is called the CPOE desig checklist ad ca be used by orgaizatios ivolved i the iitial selectio or desig of their system as well as by orgaizatios who wish to evaluate their curret system. The checklist is orgaized ito fuctioal sectios that iclude: cliical decisio support, order form cofiguratio, huma computer iteractio factors ad workflow cofiguratio. Each item o the checklist is listed with a descriptio ad icludes its source or referece from the literature (quatitative studies, qualitative/observatioal studies ad documetatio from opiio leaders ad healthcare iformatics goverig bodies). A few of the items o the list are CPOE desig cofiguratios that have bee applied at the author s orgaizatio, the Natioal Istitutes of Health, Cliical Ceter i Bethesda, MD, as a result of lessos leared ad correctios made. (See Table 1.) The CPOE desig checklist cotais 46 items. Sixtee items are categorized as cliical decisio support, such as allergy checkig, duplicate order checkig ad sigle/cumulative dose checkig. Each cofiguratio item is described i the checklist, yet it is realized that graular details o each oe is ot icluded, or is it icluded i most of the CPOE literature. Each CDS item ca be cofigured i ay umber of ways typically based o a eed idetified by the orgaizatio. Noe of the studies provided details o how a specific CDS compoet was cofigured makig it difficult to do aythig other tha add it as a geeral CDS compoet o the checklist. Numerous evaluatio opportuities exist to determie the effectiveess of specific compoets of CDS alog with their uique cofiguratio. There are 22 items o the checklist i the category of order form cofiguratio icludig madatory fields for drug, dose, frequecy ad route; iclusio of relevat lab values; cosistecy of form desig; displayig of calculatios; ad use of tall ma letterig. This category provides more specific iformatio ad detailed cofiguratio features tha the CDS category. These items were typically icluded i the literature as a result of errors previously made ad a electroic solutio offered. Agai, the evaluatio of these compoets oce implemeted will ultimately aswer the questio did it help? There are four items each i the categories of huma computer iteractio ad workflow process cofiguratio. Oe could probably make the case that all items o the checklist are related to HCI i oe way or aother, but the focus for the items listed are believed to highlight the iteractio betwee huma ad machie. Items that are listed to support workflow iclude providig a way to alert caregivers to ew orders ad providig access to medicatio referece iformatio that is coveiet ad does ot disrupt workflow. The importace of supportig workflow to reduce errors is also see as a area of further study as CPOE systems evolve. Reviewig limitatios of the checklist, the literature review was i all probability ot exhaustive. While the terms ad subject headigs used i the search were felt to be iclusive of ay articles that pertaied to CPOE ad medicatio errors, articles were subsequetly foud that did ot fall ito these headigs, but were foud usig referece lists from other articles or sys- www.himss.org volume 24 / umber 4 FALL 2010 jhim 31
Cotiued from page 30 Category Descriptio/Commets Source Provide the ability to search from medicatio lists which use Tall Ma letters. Oly allow cliicias to order the maximum split size dose allowed by hospital policy, as multiple idividual doses. For multiple potassium chloride ifusios, provide a graphical display of the patiet s serum potassium ad creatiie values of the last week. Tall Ma letters help to differetiate similar lookig medicatios (NIFEdipie vs. icardipie or predisone vs. predisolone). Example if a 80 meq dose of potassium were ordered, pharmacy would eed to split it ito four 20 meq dose IV bags. The system should oly allow users to order oe bag of 20 meqs per order. Could also be show a list of those medicatios that the patiet is takig that might affect potassium excretio (e.g. spiroolactoe or ACE ihibitors). Cosider the system requirig orderig of stat serum potassium levels betwee each ifusio. Make stop after field madatory. If forgotte, patiet may receive more doses tha required. [39] Provide flag or otificatio to provider whe medicatio orders have bee auto discotiued by the system. Oly eter default values for drug, dose, frequecy ad route if will always be correct. Do ot require cliicias to determie the base solutio o order forms for IV ifusios of medicatios. Use of White Space ot too busy. Do ot make prescribers eter orders based o formulary. Allow them to put i actual dose. For IV ifusios allow ability to eter the order i dose per time istead of drops per time. Pull i relevat iformatio oto order forms ad iclude the date ad time of specime collectio if it s a lab value. Should be the most recet lab value available displayed. Miimize free-text etry. Do ot put similar terms o top of oe aother i drop dow lists. Avoid abbreviatios i drop dow lists. Legth of items i drop dow list should be maageable. If greater tha 12 items, the group the most commoly used items at the top.. Huma Factors Cofiguratio Use alterate lie colors betwee patiets to help visual separatio of ames. Providers may ot be aware that atibiotics or arcotics were discotiued by the system at a predetermied time based o policy. If the default is correct 80% of the time, the chaces are the other 20% will use the defaults icorrectly. House staff relies o the system to idicate usual doses. If dose listed is based o pharmacy warehousig or purchasig decisios, ot cliical guidelies could cause problems. Ex if usual dosages are 20 30 mg ad the pharmacy oly stocks 10 mg doses, the 10 mg dose should ot be the default. House staff is ofte uaware of iappropriate combiatios ad ca iadvertetly choose the wrog oe. [14] Keep desig simplistic. Too much iformatio o oe scree ca be distractig ad cause users to miss importat iformatio. Medicatio should be put i as the total dose, for example: If pharmacy stocks 10mg tablets of Predisoe, order should be Predisoe 80mg p.o. ad ot Predisoe 10mg, give eight 10mg tabs. This is cofusig for may people ad ca lead to wrog dosages. If drops per time is required, the you have to kow what the quatity of IV solutio the medicatio is goig to be mixed i the you have to maually calculate the drip rate. Cofusig. Examples: Curret vetilator settigs, lab values, height/weight, Potassium levels whe furosemide is ordered.the user should ot have to remember iformatio from oe part of the system while workig i aother. Care providers should be able to choose most iformatio eeded i a order from a list of limited choices. Free text etry of ay of the five medicatio order rights ca lead to errors. Also, free text etry is typically ot part of cliical decisio support cofiguratio ad importat alerts may ot fuctio properly. If list items are close together ad similar i ame, there is a icreased risk of clickig o the wrog item. [17] Havig similar choices right o top of oe aother i a drop dow list such as IV ad IP for route ca create errors. Lists that are too log make it difficult to view all the choices at oce ad hard to use ad require scrollig dow to see other items. [33] [18] [18] [14, 36, 37] [14, 16, 17, 37] [41] [39, 42] [16] [16, 17, 32, 35, 41, 42] Helps keep rows separated visually to reduce wrog patiet choice [14, 42] Do ot put patiet lists i alphabetical order. This places similar patiet ames ext to each other makig it easier to choose the wrog patiet [14, 17, 42] [12] [17] [42] Provide method to verify correct patiet prior to order etry (e.g. display the selected ame i large letters o the scree). Esure that patiet ame appears o all order etry screes. Prior to submittig orders provide validatio that the patiet is the right oe [15, 16] To reduce wrog patiet choice [14] Workflow Process Cofiguratio Provide way to alert caregivers to ew orders. Provide access to medicatio iformatio that is coveiet ad does ot disrupt workflow. Provide method to access view of all patiet s medicatios (icludig dose ad frequecy) o oe scree. Do ot use test patiets with cute ames i the productio system. Develop ad use uique ad cosistet amig covetio for test patiets. Without face to face iteractio, urses may ot be aware that a ew medicatio order has bee etered. Provide flag or otificatio that ew order is pedig. Medicatio kowledge deficiecy was oted i oe study to cotribute to over half of the prescribig errors oted that could have potetially bee prevetable. [16, 43] Screes that list active medicatio orders also should list IV drip orders [14, 35] They may be mistake for real patiets ad orders etered will ot be carried out o the real patiet. [15] CCNIHTEST, TESTPATIENT is more uique tha Tester, Joe [39] [21] 32 jhim
tematic reviews. A additioal limitatio to the checklist is the fact that it does ot represet a comprehesive list of cofiguratio items that ca lead to the reductio of medicatio errors. There are certaily others that are ot published. It is hoped that this is the first iteratio of a tool that will evolve ad improve over time as evaluative studies add to the list of items. The checklist is also ot detailed eough to provide iformatics specialists the ability to replicate each cofiguratio item. This is a more systemic issue as the literature lacks the detail to iclude i the checklist. It is however a good startig poit for iformatics specialists to assess their CPOE systems ad idetify areas where improvemets ca result i the reductio of medicatio errors. Coclusio CPOE systems have the potetial to reduce medicatio errors. It is depedet, however, o how each idividual orgaizatio cofigures their system. Most published studies o error reductio ad CPOE do ot meet rigorous quality stadards, resultig i poor geeralizability. They have ot added sigificatly to CPOE s body of error reductio kowledge. I may cases, it is difficult to ascertai what is really beig evaluated. This review has exposed the fact that the evidece is limited i its geeralizability ad orgaizatios should stregthe their evaluatio programs ad use themselves as a cotrol or baselie from which to measure success. The result of the developmet of the CPOE desig checklist has led to a clear uderstadig that more evaluative research is eeded; research that is coducted usig stadard tools recogized by the health IT idustry; research that icludes a clear uderstadig of how a CPOE system has bee desiged ad; research that is published usig stadard guidelies that result i quality reportig, ultimately leadig to improvemets i patiet safety. I light of the fact that sigificat ARRA fudig is beig allocated for the adoptio of CPOE systems, it is imperative that iformatics specialists at all levels, from ovice to expert, have quality tools to implemet ad evaluate these systems. JHIM Patricia Segstack, DNP, RN-BC, CPHIMS, is the Deputy CIO, Cliical Iformatics at the Natioal Istitutes of Health, Cliical Ceter ad a recet graduate of the Doctor of Nursig Practice program at Vaderbilt Uiversity. Refereces 1. HIMSS Aalytics. EMR adoptio model. 2010. Available at: www.himssaalytics.org. 2. Goolsby K. (2002). CPOE odyssey: the story of evolvig the world s first computerized physicia order etry system ad implicatios for today s cpoe decisio makers. Available at: www.outsourcig-iformatio-techology.com/ cpoe.html 3. U.S. Departmet of Health & Huma Services. 2010. Health iformatio techology for the future of health ad care. Available at: http://healthit.hhs.gov/ portal/server.pt?ope=512&objid=1204&paretame=commuitypage&paretid =8&mode=2&i_hi_userid=10741&cached=true. 4. Koh LT, Corriga JM, Doaldso MS. (Eds.). 2000. To Err is Huma: Buildig a Safer Health System. Washigto D.C.: Natioal Academy Press. 5. Kaushal R, Shojaia KG, Bates DW. 2003. Effects of computerized physicia order etry ad cliical decisio support systems o medicatio safety. Arch Iter Med. 163, 1409-1416. 6. Ammewerth E, Schell-Iderst P, Macha C, Sievert U. 2008. The effect of electroic prescribig o medicatio errors ad adverse drug evets: A systematic review. JAMIA. 15(5), 585-600. 7. Reckma MH, Westbrook JI, Koh Y, Lo C, Day RO. 2009. Does computerized provider order etry reduce prescribig errors for hospital ipatiets? A systematic review. JAMIA. 16(5), 613-622. 8. Frakli BD, O Grady K, Doyai P, Jackli G, Barber N. The impact of a closed-loop electroic prescribig ad admiistratio system o prescribig errors, admiistratio errrors ad staff time: a before ad after study. Quality ad Safety of Health Care. 2007; 16:279-284. 9. Shulma R, Siger M, Goldstoe J, Belliga G. Medicatio errors: A prospective cohort study of had writte ad computerised physicia order etry i the itesive care uit. Critical Care. 2005;9, R516-R521. 10. Potts AL, Barr FE, Gregory DF, Wright L, Patel NR. Computerized physicia order etry ad medicatio errors i a pediatric critical care uit. Pediatrics. 2004;113(1), 59-63. 11. Raebel MA, Carrlo NM, Kelleher JA, Chester EA, Berga S, Magid DJ. Radomized trial to improve prescribig safety durig pregacy. JAMIA. 2007;14, 440-450. 12. Kim GR, Che AR, Arceci RJ, Mitchell SH, Kokoszka KM, Daiel D, Lehma CU. Error reductio i pediatric chemotherapy. Arch Ped Adolesc Med. 2006;160, 495-498. 13. Vaidya V, Sowa AK, Mills ME, Socke K, Gaffoor M, Hilmas E. Evaluatig the safety ad efficiecy of a CPOE system for cotiuous medicatio ifusios i a pediatric ICU. America Medical Iformatics Associatio 2006 Proceedigs, USA, 1128. 14. Koppel R, Metley JP, Cohe A, Abaluck B, Localia AR, Dimmel SE, Strom BL. Role of computerized physicia order etry systems i facilitatig medicatio errors. JAMA. 2005;293, 1197-1203. 15. Campbell EM, Sittig DF, Ash JS, Guappoe KP, Dykstra RH. Types of uiteded cosequeces related to computerized provider order etry. JAMIA. 2006;13, 547-556. www.himss.org volume 24 / umber 4 FALL 2010 jhim 33
16. Ash JS, Sittig DF, Poo EG, Guappoe K, Campbell E, Dykstra RH. The extet ad importace of uiteded cosequeces related to computerized provider order etry. JAMIA. 2007;14, 415-423. 17. Walsh KE, Adams WG, Baucher H, Vici RJ, Chessare JB, Cooper MR, Herbert PM, Schaiker EG, Ladriga CP. Medicatio errors related to computerized order etry for childre. Pediatrics. 2006;118, 1872-1879. 18. FitzHery F, Peterso JF, Arrieta M, Waitma LR, Schildcrout JD, Miller RA. Medicatio admiistratio descrepacies persist despite electroic orderig. JAMIA. 2007;14, 756-764. 19. Berger RG, Kichak JP. Computerized physicia order etry: helpful or harmful? JAMIA. 2004;11, 100-103. 20. Kuperma GJ, Bobb A, Paye TH, Avery AJ, Gadhi TK, Burs G, Classe DC, Bates DW. Medicatio-related cliical decisio support i computerized provider order etry systems: a review. JAMIA. 2007;14, 29-40. 21. Bobb A, Gleaso K, Husch M, Feiglass J, Yarold P, Noski G. The epidemology of prescribig errors: The potetial impact of computerized prescriber order etry. JAMA. 2004;164, 785-792. 22. Ash JS, Stavri P, Kuperma GJ. A cosesus statemet o cosideratios for a successful CPOE implemetatio. JAMIA. 2003;10, 229-234. 23. Kilbridge PM, Welebob EM, Classe DW. Developmet of the Leapfrog methodology for evaluatiog hospital implemeted iapatiet computerized physicia order etry systems. Quality ad Safety i Health Care. 2006;15, 81-84. 24. Meyers B, Hollad J, Cruz I. Strategic directios i huma computer iteractio. ACM Computig Surveys. 1996;28(4). Available at: www2.cs.cmu. edu/~bam/sfworkshop/hcireport.html. 25. Hasler RA. Huma factors desig: What is it ad how ca it affect you? J Itrav Nurs. 1996;19(35): S5-S8. 26. Gaier A, Pacheri K, Zhag J. Improvig the huma computer iterface desig for a physicia order etry system. America Medical Iformatics Associatio 2003 Proceedigs, USA, 847. 27. Horsky J, Kaufma DR, Patel VL. The cogitive complexity of a provider order etry iterface. America Medical Iformatics Associatio 2003 Proceedigs, USA, 294-298. 28. Gadhi T, Weigart SN, Seger AC, Borus J, Burdick E, Poo EG, Leape LL, Bates DW. (2005). Outpatiet prescribig errors ad the impact of computerized prescribig. J Ge It Med. 2005;20, 837-841. 29. Natioal Coordiatig Coucil for Medicatio Error Reportig ad Prevetio (NCC MERP). 2008. Available at: www.ccmerp.org. 30. Cordero L, Kueh L, Kumar RR, Mekhjia HS. Impact of computerized physicia order etry o cliical practice i a ewbor itesive care uit. J Periatology. 2006;24, 88-95. 31. Colpaert K, Claus B, Somers A, Vadewoude K, Robays H, Decruyeaere J. Impact of computerized physicia order etry o medicatio prescriptio errors i the itesive care uit: a cotrolled cross-sectio trial. Critical Care. 2005;10, R21. Available at: http//:ccforum,com/cotet/10/1/r21. 32. Bates DW, Leape LL, Culle DJ, Laird N, Peterse LA, Teich JM, et al. Effect of computerized physicia order etry ad a team itervetio o prevetio of serious medicatio errors. JAMA. 1998;280, 1311-1316. 33. Certificatio Commissio for Healthcare Iformatio Techology. Ipatiet EHR Certificatio. Accessed April 9, 2009. Available at: www.cchit.org. 34. Olive A, Michalake I, Zalma D, Dorma E, Yeshuru D, Odeh M. Prevetio of prescriptio errors by computerized, o-lie surveillace of drug order etry. It J Med Iform. 2005;74, 377-386. 35. Horsky J, Kuperma GJ, Patel VL. Comprehesive aalysis of a medicatio dosig error related to CPOE. J Med Iform Soc. 2005;12, 377-382. 36. Evas DK, Beham SW, Garrard CS. A compariso of hadwritte ad computer-assisted prescriptios i a itesive care uit. Critical Care. 1998;2(73). Available at: http://ccforum.com.cotet/2/2/73. 37. Mahoey CD, Berard-Collis CM, Colema R, Amaral JF, Cotter CM. Effects of a itegrated cliical iformatio system o medicatio safety i a multihospital settig. Am J Health Syst Pharm. 2007;64, 1969-1977. 38. Campbell EM, Sittig DF, Guappoe KP, Dykstra RH, Ash JS. Overdepedece o techology: A uiteded adverse cosequece of computerized provider order etry. America Medical Iformatics Associatio 2007 Proceedigs, USA, 94-98. 39. Natioal Istitutes of Health, Cliical Ceter. Departmet of Cliical Research Iformatics (2008). CPOE cofiguratio as a result of lessos leared. Author s persoal experiece. 40. Voeffray M, Paatier A, Stupp R, Fucia N, Leyvraz S, Wasserfalle J. Effect of computerizatio o the quality ad safety of chemotherapy prescriptio. Qual Saf Healthcare. 2006;15, 418-421. 41. Nielse J. 2005. Te usability heuristics. Available at: www.useit.com/papers/ heuristic/heuristic_list.html. 42. Khajouei R, Jaspers MW. CPOE system desig aspects ad their qualitative effect o usability. Stud Health Tech Iform. 2008;136, 309-314. 43. Ha YY, et al. Uexpected icreased mortality after implemetatio of a commercially sold computerized physicia order etry system. Pediatrics. 2005;116, 1506-1512 34 jhim