WP3-Deliverable: Development of Learning Outcomes [Public Part]
2 Project information Project acronym: PATIENT Project title: Improving the continuity of patient care through teaching and researching novel patient handover processes in Europe Project number: 527620-LLP-1-2012-1-NL-ERASMUS-FEXI Sub-programme or KA: Erasmus Multilateral Projects Project website: http://www.patient-project.eu Report Type Public Reporting period: From 01/04/2013 To 30/09/2013 Report version: 1 Public Date of preparation: 04/10/2013 Contributors Mariona Secanell Carola Orrego, Lina Stieger, Susanne Druener, Sasa Sopka
Authors (Partner) Helen Hynes, Patrick Henn, Bridget Maher, Hendrik Drachsler, Slavi Stoyanov
Contact Person Patrick Henn (p.henn@ucc.ie) Work Package Work Package 3 - Development of Learning Outcomes Project coordinator: Hendrik Drachsler Project coordinator organisation: Open Universiteit Nederland, CELSTEC Project coordinator telephone number: +31 45 576 2218 Project coordinator email address: hendrik.drachsler@ou.nl
This project has been funded with support from the European Commission. This publication [communication] reflects the views only of the author, and the Commission cannot be held responsible for any use, which may be made of the information contained therein.
2008 Copyright Education, Audiovisual & Culture Executive Agency. The document may be freely copied and distributed provided that no modifications are made, that the source is acknowledged and that this copyright notice is included.
3 Executive !"##$%& ("%)*+, 1he alm of Work ackage 3 of Lhe A1lLn1 ro[ecL ls Lo develop, by consulLaLlon, agreed Learnlng CuLcomes for Leachlng of Pandover ln a healLhcare conLexL, conLrlbuLlng Lo sLandardlzaLlon of handover procedures ln Lurope. 1hese learnlng ouLcomes are Lo be generaLed uslng consensus from a panel of experLs. -,./*0 1he meLhod for developlng learnlng ouLcomes lnvolved Lhe parLlclpanLs ln acLlvlLles LhaL mosL professlonals are used Lo: ldea generaLlon, sorLlng of ldeas lnLo groups and raLlng Lhe ldeas on some values (e.g. lmporLance and dlfflculLy Lo achleve). 1he analysls applles mulLl-dlmenslonal scallng and clusLer analysls Lo vlsually deplcL Lhe experLs' shared represenLaLlons on Lhe learnlng ouLcomes as LhemaLlc groups. Cne of Lhe dlsLlngulshlng characLerlsLlcs of CCM ls Lhe vlsuallsaLlon of Lhe resulLs from Lhe analysls. vlsuallsaLlon allows for grasplng aL once Lhe emerglng daLa sLrucLures and Lhelr lnLerrelaLlonshlps Lo supporL declslon-maklng. Croup ConcepL Mapplng produces Lhree maln Lypes of vlsuallsaLlons: concepLual maps, paLLern maLches and go-zones. 1*234"+5*2+ 1hls sLudy provlded noL only an emplrlcal basls for ldenLlfylng Lhe maln learnlng ouLcomes areas, buL also suggesLed how Lo operaLlonally deflne Lhem (Lhrough Lhe sLaLemenLs ln each clusLer). 1hese learnlng ouLcomes are exLenslve and may be more sulLable for lncorporaLlon lnLo Lhe medlcal currlculum as a whole raLher Lhan slmply a speclflc Lralnlng module on handover. Powever Lhose ouLcomes consldered mosL lmporLanL were also among Lhose consldered mosL dlfflculL Lo dellver.
5 1. Introduction 1he alm of Work ackage 3 of Lhe A1lLn1 ro[ecL ls Lo develop, by consulLaLlon, agreed Learnlng CuLcomes for Leachlng of Pandover ln a healLhcare conLexL, conLrlbuLlng Lo sLandardlzaLlon of handover procedures ln Lurope. 1hese learnlng ouLcomes are Lo be generaLed uslng consensus from a panel of experLs. 1here are Lwo ma[or lssues wlLh onllne experL consulLaLlons: (a) generaLlng a comprehenslve seL of learnlng ouLcomes, and (b) reachlng agreemenL on Lhem. uurlng Lhe preparaLlon of a llsL of learnlng ouLcomes, parLlclpanLs mlghL be focused on Lhe currenL pracLlce of a course already runnlng. 1hls would narrow Lhe scope of Lhe learnlng ouLcomes Lo be deflned. An agreemenL on learnlng ouLcomes mlghL even be more dlfflculL Lo achleve: Lhe parLners represenL dlfferenL healLhcare sysLems and lndlvldually Lhey mlghL have raLher dlfferenL Lhlnklng sLyles. 1he parLlclpanLs mlghL also noL agree on how much emphasls should be puL on each learnlng ouLcome. AddlLlonally, durlng llve meeLlngs Lhere ls always Lhe phenomenon of 'groupLhlnk', or 'peer-pressure', (Lhe negaLlve effecL of Lhe group on Lhe oplnlons of Lhe lndlvldual members). MeLhods for experL consulLaLlon, such as locus groups, AfflnlLy dlagram and Lhe uelphl meLhod are some of Lhe mosL used sLrucLured approaches almed aL achlevlng consensus. Powever, Lhe analysls of focus group daLa lmposes pre-deLermlned classlflcaLlon schemas, whlch can be elLher non-exhausLlve or lmpose blases. ln afflnlLy dlagram sesslons, parLlclpanLs Lyplcally would suggesL dlfferenL clusLerlng soluLlons, boLh ln Lerms of number of clusLers and Lhe conLenL of Lhe clusLers, whlch makes lL dlfflculL for researchers Lo come up wlLh a unlfled vlslon on how besL Lo sLrucLure Lhe lnformaLlon. 1he uelphl meLhod requlres several lLeraLlve rounds before clalmlng consensus ln Lhe group. 1he consensus ls more or less forced and Lhe sub[ecLlve approach ls always Lhere. Cur soluLlon Lo Lhe lssues descrlbed ls Croup ConcepL Mapplng (CCM) (kane & 1rochlm, 2006, 1rochlm, 1989). 1hls research meLhodology, whlle bulldlng on Lhe sLrengLhs of locus groups, AfflnlLy dlagrams and Lhe uelphl-meLhod mlLlgaLes some of Lhelr weaknesses. lL has been successfully applled ln Lhe medlcal domaln ln Lhe pasL by (SLoyanov eL al., 2012) 1he prlmary research quesLlon for Lhe A1lLn1 learnlng ouLcome sLudy Lo answer ls: !"# %&' #( )*++",- +&,-'(,) -" &,,./( &- &' &0,((1 )(- "2 -3( 4567896 !&'1"/(, :"1*;( ;(&,'.'0 "*-%":()< 1he reporL ls sLrucLured as follows: ln Lhe meLhod secLlon, we lnLroduce Lhe deslgn of Lhe sLudy uslng Lhe Croup ConcepL Mapplng meLhodology. ln Lhe resulLs secLlon we presenL parLlclpanL demographlcs and Lhe resulLs from Lhe Croup ConcepL Mapplng on clusLerlng and raLlng of aggregaLed learnlng ouLcome sLaLemenLs. 1he dlscusslon secLlon crlLlcally reflecLs on Lhe ouLcomes. llnally, Lhe concluslons and suggesLlons secLlon presenLs recommendaLlons for furLher developmenL of Lhe A1lLn1 Pandover educaLlonal module.
6 2. Method 2.1 Expert Selection 1he consorLlum members agreed on a selecLlon framework for ldenLlfylng experLs Lo conLrlbuLe Lo Lhe group concepL mapplng (CCM) process for Lhe ldenLlflcaLlon of learnlng ouLcomes for Lhe handover sLudy module for undergraduaLe medlcal sLudenLs. 2.2 Group Concept Mapping Croup ConcepL Mapplng (CCM) ls a sLrucLured, mlxed approach applylng boLh quanLlLaLlve and quallLaLlve measures Lo ob[ecLlvely ldenLlfy an experL group's common undersLandlng abouL a parLlcular lssue, ln our case Lhe A1lLn1 Pandover module learnlng ouLcomes. 1he meLhod lnvolved Lhe parLlclpanLs ln a few acLlvlLles LhaL mosL of professlonals are used Lo: ldea generaLlon, sorLlng of ldeas lnLo groups and raLlng Lhe ldeas on some values (e.g. lmporLance and dlfflculLy Lo achleve). 1he parLlclpanLs work lndlvldually buL lL ls Lhe advanced sLaLlsLlcal Lechnlques of mulLldlmenslonal scallng and hlerarchlcal clusLer analysls LhaL quanLlLaLlvely aggregaLes lndlvldual lnpuLs of Lhe parLlclpanLs Lo reveal ob[ecLlve paLLerns ln Lhe daLa. Cne of Lhe dlsLlngulshlng characLerlsLlcs of CCM ls vlsuallsaLlon, whlch ls a subsLanLlal parL of Lhe analysls. vlsuallsaLlon allows for grasplng aL once Lhe emerglng daLa sLrucLures, Lhelr lnLerrelaLlonshlps, and Lhelr lnLerpreLaLlon Lo supporL declslon-maklng. Croup ConcepL Mapplng produces Lhree maln Lypes of vlsuallsaLlons: concepLual maps, paLLern maLches and go-zones. ln conLrasL Lo Lhe uelphl meLhod, ln CCM, Lhere ls only one round of daLa sLrucLurlng as Lhe parLlclpanLs work lndependenLly and anonymously of each oLher Lo llmlL Lhe posslblllLy of 'groupLhlnk' or 'peer-pressure'. unllke lnLervlews and focus groups, CCM does noL rely on pre- deLermlned classlflcaLlon schemas. 1he meLhod does noL need lnLer-coder dlscusslon Lo come up wlLh an agreemenL. When sorLlng Lhe sLaLemenLs lnLo groups, Lhe parLlclpanLs, ln facL, 'code' Lhe LexL Lhemselves. 1hen mulLlvarlaLe sLaLlsLlcal analysls aggregaLes Lhe lndlvldual codlng schemas across Lhe parLlclpanLs. Consensus ls noL forced, buL emerges from Lhe daLa. Croup ConcepL Mapplng supporLs Lhe researcher ln deallng wlLh dlverse lnformaLlon, sLrucLured ln varlous ways, whlch ls a problem ln AfflnlLy dlagram sesslons. 1he CCM procedure conslsLed of flve phases: (1) ldea generaLlon (bralnsLorm) and ldea prunlng, (2) sorLlng of ldeas lnLo groups, (3) raLlng on Lwo values (lmporLance and dlfflculLy Lo achleve), (4) analysls of Lhe daLa and (3) lnLerpreLaLlon of Lhe resulLs. All pro[ecL members were lnvlLed Lo parLlclpaLe ln Lhe learnlng ouLcomes sLudy Lhrough Lhe pro[ecL's onllne managemenL sysLem. See Appendlx A for leLLer of lnvlLaLlon. An lnvlLaLlon Lo some exLernal experLs was senL by emall. All parLlclpanLs were fully lnformed abouL Lhe purpose, Lhe procedure, and Lhe Llme needed for compleLlng Lhe acLlvlLles. 1he parLlclpanLs were provlded wlLh a llnk Lo Lhe bralnsLormlng page of a web-based Lool for daLa collecLlon and analysls (ConcepL SysLem Clobal, 2013). 1hey could vlslL Lhe web slLe, as many Llmes Lhey needed uslng Lhelr own username and password. 1he parLlclpanLs were asked Lo generaLe ldeas compleLlng Lhe followlng Lrlgger sLaLemenL: Cne speclflc learnlng ouLcome of Lhe Pandover module ls." 1he ldeas should be a shorL phrases or sLaLemenLs expresslng one LhoughL.
7 We purposely dld noL ask Lhe parLlclpanLs Lo follow sLandard formaLs for deflnlng learnlng ouLcomes. lnLroduclng such a formaL would be counLerproducLlve, as lL would resLrlcL Lhe free flow of ldeas. arLlclpanLs were lnsLrucLed LhaL Lhe ldeas generaLed should Lake Lhe form of shorL phrases or sLaLemenLs, each expresslng a slngle LhoughL. 1he parLlclpanLs were glven Lwo weeks Lo compleLe Lhe ldea generaLlon Lask. AfLer compleLlng Lhe ldea generaLlon phase, a small group of professlonals from Lhe consorLlum was lnsLrucLed Lo check, edlL and lf needed reduce Lhe ldeas Lo a manageable seL (abouL 100) for Lhe nexL sLages of sorLlng and raLlng. 1he guldellnes were as follows: look for sLaLemenLs LhaL conLaln more Lhan one ldea and lf needed spllL Lhem, remove ldenLlcal ldeas, check wheLher Lhe ldeas address Lhe focus prompL, make sure LhaL each unlque ldea ls lncluded ln Lhe flnal llsL, and make sure LhaL Lhe ldea ls clear, conclse, undersLandable (a crlLerlon ls how easy or dlfflculL an ldea can be raLed on Lhe Lwo values: lmporLance and dlfflculLy). 1he flnal llsL, randomlsed, was made Lhen avallable Lo Lhe parLlclpanLs, flrsLly for Lhe sorLlng of ldeas lnLo groups (based on slmllarlLy ln meanlng), glvlng names Lo Lhe groups, and secondly for Lhe raLlng of Lhe ldeas on Lwo values - lmporLance and dlfflculLy. lor Lhe sorLlng, we advlsed Lhe parLlclpanL Lo noL puL all sLaLemenLs ln one group and LhaL for such sorL of sLudles beLween 3 and 20 groups works well, also noL Lo make a 'resldual' group, oLher" or , lL ls beLLer Lo puL a sLaLemenL ln lLs own group lf dlfflculL Lo flnd place for lL. lor Lhe raLlng we lnsLrucLed Lhe parLlclpanLs Lo raLe each sLaLemenL on a 3 polnL LlkerL scale, and expllclLly asked Lhem Lo use Lhe full range of raLlng. 1he parLlclpanLs were glven Lhree weeks Lo compleLe boLh sorLlng and raLlng. A remlnder senL afLer Lwo weeks. As ln Lhe bralnsLormlng phase, Lhe parLlclpanLs could save Lhelr work and reLurn laLer Lo conLlnue. 1he analysls lncludes mulLldlmenslonal scallng (MuS) and hlerarchlcal clusLer analysls (PCA) for sorLlng Lhe daLa, and means, sLandard devlaLlons, correlaLlons and slgnlflcance LesLs for Lhe raLlng daLa. MuS analysls Lakes flrsL Lhe LoLal slmllarlLy maLrlx, whlch aggregaLes all lndlvldual maLrlces of Lhe parLlclpanLs. 1he sorLlng of every parLlclpanL ls sLored as an lndlvldual maLrlx An lndlvldual maLrlx conslsLs of 1s and 0s. 1 ls puL aL a cross-secLlonal cell lf Lwo sLaLemenLs have been grouped LogeLher, oLherwlse Lhe value glven ls 0. A cell ln Lhe LoLal slmllarlLy maLrlx shows how many people have grouped any Lwo sLaLemenLs. lL can range from 0 (no one has clusLered Lhem) Lo Lhe maxlmal number of parLlclpanLs (all sorLed Lhe Lwo sLaLemenLs). 1he LoLal slmllarlLy maLrlx can be represenLed as a coordlnaLe maLrlx. lrom Lhe coordlnaLes Lhe dlsLances beLween all palrs of polnLs can be compuLed and furLher presenLed as a maLrlx of dlsLances beLween polnLs (1rochlm, 1989). Croup ConcepL Mapplng uses Lhe Ward agglomeraLlve hlerarchlcal clusLer analysls. lL sLarLs wlLh Lhe assumpLlon LhaL all ldeas are lndlvldual clusLers, SLlll, human experLs need Lo look aL Lhe soluLlon proposed and declde upon Lhe number of clusLers LhaL represenLs Lhe daLa ln Lhe besL posslble way and reflecLs Lhe conLexL of Lhe sLudy.
8 3. Results 3.1 Demographics 61 parLlclpanLs reglsLered lnlLlally Lo Lhe sysLem for onllne daLa collecLlon, supporLlng Lhe CCM approach, (ConcepL SysLem Clobal, 2013) creaLlng a username and password. 1hey gave Lhelr lnformed consenL Lo parLlclpaLe. Cf Lhem 43 conLrlbuLed effecLlvely Lo Lhe bralnsLormlng sesslon and 22 compleLed Lhe sorLlng and raLlng phases. 3.2 Statements 1he experLs produced sLaLemenLs durlng Lhe ldea generaLlon phase. AfLer compleLlng Lhe ldea generaLlon phase, a small group of professlonals from Lhe consorLlum was ldenLlfled for Lhe prunlng of ldeas. 1hey were lnsLrucLed Lo check, edlL and lf needed reduce Lhe ldeas Lo a manageable seL of 100 or so ldeas for Lhe nexL sLages of sorLlng and raLlng. 3.3 Clustering Results ClusLerlng analysls ldenLlfled clusLers and raLed Lhem on lmporLance and dlfflculL Lo achleve, see flgure 1 as an example of a 16 clusLer soluLlon and flgure 2 as an example of merglng lL Lo a 3 clusLer soluLlon
llgure 1. 16-clusLer soluLlon.
9
llgure 2. Merglng 16-clusLer soluLlon Lo 3-clusLer soluLlon 1he nexL sLep ln maklng sense of Lhe daLa was Lo aLLach meanlngful labels Lo Lhe clusLers. 1here are Lhree meLhods avallable for labellng. 1he flrsL meLhod ls Lo check whaL Lhe sysLem suggesLs. 1he sysLem suggesLs a label for a group of sLaLemenLs, based on Lhe label glven by a parLlclpanL, whose cenLrold ls Lhe closesL Lo Lhe cenLrold of Lhe clusLer formed by Lhe aggregaLlon of Lhe daLa from all Lhe parLlclpanLs. 1he second meLhod ls Lo look aL Lhe brldglng values of Lhe sLaLemenLs composlng Lhe clusLer. 1he sLaLemenLs wlLh lower brldglng values beLLer represenL a clusLer. 1he Lhlrd meLhod ls Lo read Lhrough all Lhe sLaLemenLs ln a clusLer and Lo deflne ln a label whaL ls Lhe sLory behlnd Lhe sLaLemenLs (whaL ls lL LhaL Lhe clusLer wanLs Lo Lell us). 1o deflne Lhe clusLer labels (e.g., collecLlve Lheme of Lhe sLaLemenLs, or caLegory) we comblned all Lhree meLhods. 4. Discussion 43 parLlclpanLs conLrlbuLed effecLlvely Lo Lhe bralnsLormlng sesslon and 22 parLlclpanLs from 4 Luropean counLrles compleLed Lhe sorLlng and raLlng phases. 1he prlnclple LhaL dlsLances beLween lndlvldual ldeas ln Croup ConcepL Mapplng maLLer, applles Lo dlsLances beLween Lhe clusLers as well. 1he closer Lhe clusLers are Lo each oLher, Lhe closer Lhey are concepLually. 1he clusLer analysls ldenLlfled 10 clusLers for conslderaLlon as learnlng ouLcomes for a handover Lralnlng module for medlcal sLudenLs. lL seems LhaL Lhe A1lLn1 CCM sLudy ldenLlfled learnlng ouLcomes noL only for a slngle module on handover buL also for a whole currlculum on handover LhaL gradually can be developed ln Lhe fuLure. urachsler eL al. (urachsler eL al., 2012) already provlde rlch educaLlonal maLerlals on Lhe Pandover Loolbox LhaL can be Laken lnLo accounL for Lhe deslgn of Lhe selecLed learnlng ouLcomes. Maher eL al. provldes a sulLable moblle app Lo Lraln
10 accuraLe and compleLe handovers wlLh medlcal docLors (Maher eL al., 2013, Maher, urachsler, kalz, & SpechL, 2012). 1he groups of learnlng ouLcomes Lhrough Lhe sLaLemenLs generaLed by Lhe experL parLlclpanLs provlde an emplrlcal basls for deflnlng learnlng ouLcomes of Lhe A1lLn1 Pandover module applylng Lhe sLandard of deflnlng learnlng ouLcomes. 1he learnlng ouLcomes of Lhe A1lLn1 pro[ecL have ldenLlfled learnlng ouLcomes from whlch a handover Lralnlng module for medlcal sLudenLs mlghL be deslgned. 5. References
ConcepL SysLems Clobal [CompuLer SofLware]. ConcepL SysLems, lnc. lLhaca, n.?, 2012. Drachsler, H., Kicken, W., van der Klink, M., Stoyanov, S., Boshuizen, H. P. A., & Barach, P. (2012). The Handover Toolbox: a knowledge exchange and training platform for improving patient care. BMJ quality & safety, 21, 11141120. doi:10.1136/bmjqs-2012-001176 Kane, M., & Trochim, W. M. K. (2006). Concept mapping for planning and evaluation. Concept mapping for planning and evaluation. (p. 216). Sage Publications. Maher, B., Drachsler, H., Kalz, M., & Specht, M. (2012). The CLAS App - A mobile training tool to improve handover procedures between hospital interface and family doctors. In S. Marcus, S. Mike, & M. Jari (Eds.), Proceedings of the 11th International Conference on Mobile and Contextual Learning (pp. 3845). Retrieved from http://ceur-ws.org/Vol- 955/papers/paper_29.pdf Maher, B., Kalz, M., Specht, M., Lezcano, L., Drachsler, H., & Henn, P. (2013). Use of Mobile Applications for Hospital Discharge Letters - Improving Handover at Point of Practice. International Journal of Mobile and Blended Learning, 5(3), 129. Stoyanov, S., Boshuizen, H., Groene, O., van der Klink, M., Kicken, W., Drachsler, H., & Barach, P. (2012). Mapping and assessing clinical handover training interventions. BMJ Quality & Safety, 21, 150157. doi:10.1136/bmjqs-2012-001169 Trochim, W. M. K. (1989). An introduction to concept mapping for planning and evaluation. (W. M. Trochim, Ed.)Evaluation and Program Planning, 12(1), 116. doi:10.1016/0149- 7189(89)90016-5 Ward, !. P., !r. (1963), "Plerarchlcal Crouplng Lo CpLlmlze an Cb[ecLlve luncLlon", !ournal of Lhe Amerlcan SLaLlsLlcal AssoclaLlon, 38, 236-244