Validation of a computer-adaptive test to evaluate generic health-related quality of life

RESEARCH Open Access Valida ion o a compu e -adap i e es o e alua e gene ic heal h- ela ed quali y o li e Pablo Rebollo 1* , Ignacio Cas ejón 1 , Jesús Cue o 1 , Guille mo Villa 1 , Edua do Ga cía-Cue o 2 , Helena Díaz-Cue o 1 , Pila C Za daín 1 , José Muñiz 2 , Jo di Alonso 3,4 , he Spanish CAT-Heal h Resea ch G oup Abs ac Backg ound: Heal h Rela ed Quali y o Li e (HRQoL) is a ele an a iable in he e alua ion o heal h ou comes. Ques ionnai es based on Classical Tes Theo y ypically equi e a la ge numbe o i ems o e alua e HRQoL. Compu e Adap i e Tes ing (CAT) can be used o educe es s leng h while main aining and, in some cases, imp o ing accu acy. This s udy aimed a alida ing a CAT based on I em Response Theo y (IRT) o e alua ion o gene ic HRQoL: he CAT-Heal h ins umen . Me hods: C oss-sec ional s udy o subjec s aged o e 18 a ending P ima y Ca e Cen es o any eason. CAT- Heal h was adminis e ed along wi h he SF-12 Heal h Su ey. Age, gende and a checklis o ch onic condi ions we e also collec ed. CAT-Heal h was e alua ed conside ing: 1) easibili y: comple ion ime and es leng h; 2) con en ange co e age, I em Exposu e Ra e (IER) and es p ecision; and 3) cons uc alidi y: di e ences in he CAT-Heal h sco es acco ding o clinical a iables and co ela ions be ween bo h ques ionnai es. Resul s: 396 subjec s answe ed CAT-Heal h and SF-12, 67.2% emales, mean age (SD) 48.6 (17.7) yea s. 36.9% did no epo any ch onic condi ion. Median comple ion ime o CAT-Heal h was 81 seconds (IQ ange = 59-118) and i inc eased wi h age (p < 0.001). The median numbe o i ems adminis e ed was 8 (IQ ange = 6-10). Nei he ceiling no loo e ec s we e ound o he sco e. None o he i ems in he pool had an IER o 100% and i was o e 5% o 27.1% o he i ems. Tes In o ma ion Func ion (TIF) peaked be ween le els -1 and 0 o HRQoL. S a is ically signi ican di e ences we e obse ed in he CAT-Heal h sco es acco ding o he numbe and ype o condi ions. Conclusions: Al hough domain-speci ic CATs exis o a ious a eas o HRQoL, CAT-Heal h is one o he i s IRT- based CATs designed o e alua e gene ic HRQoL and i has p o en easible, alid and e icien , when adminis e ed o a b oad sample o indi iduals a ending p ima y ca e se ings. Backg ound Heal h-Rela ed Quali y o Li e (HRQoL) is among he mos used a iables in Heal h Ou comes Resea ch (HOR) in he academic ield, as well as in clinical ials and pos -au ho isa ion s udies. I e e s o he subjec i e alua ion o he in luence o heal h on he indi iduals’ abili y o ha ing a no mal unc ioning which makes i possible o pe o m all he ac i i ies which a e impo an o hem and which a ec hei well-being [1]. Al hough du ing he las 35 yea s HRQoL assessmen has had an eno mous de elopmen wo ldwide, se e al ba ie s limi i s use in he clinical p ac ice. These ba ie s had been desc ibed by Deyo and Pa ick in 1989 [2] and we e e ised in Spain in 2005 [3]. Classical Tes Theo y (CTT) canno sol e ce ain p ac ical issues, such as he high numbe o ques ions needed o comple e a mul i- dimensional HRQoL ques ionnai e and he lack o accu- acy when dealing wi h he change o indi idual sco es o e ime. I em Response Theo y (IRT) [4,5] o e comes some o he limi a ions ha may a ec ins umen s de eloped unde he CTT. CAT ins umen s based on IRT, clea ly inc ease “measu emen e iciency”( he a io o a measu e’s psychome ic soundness o he esponse bu den he measu es imposes). A g ea e measu emen p ecision can be achie ed h ough he selec ion o a ew i ems om a calib a ed i em pool ha combines high * Co espondence: [email p o ec ed] 1 BAP Heal h Ou comes Resea ch, Calle Azcá aga 12 A, 33010, O iedo, Spain Full lis o au ho in o ma ion is a ailable a he end o he a icle Rebollo e al.Heal h and Quali y o Li e Ou comes 2010, 8:147 h p://www.hqlo.com/con en /8/1/147 © 2010 Rebollo e al; licensee BioMed Cen al L d. This is an Open Access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License (h p://c ea i ecommons.o g/licenses/by/2.0), which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed. quali y i ems om mul iple ins umen s in o a single da a esou ce [6]. Since each i em is independen ly desc ibed by pa ame e s such as di icul y o disc imina- ion [7], hey can be combined as necessa y, he e o e, he e is no ques ionnai e as such bu di e en combina- ions o i ems which p o ide compa able sco es. Th ough IRT, an I em Cha ac e is ic Cu e (ICC) is cons uc ed o each i em; his cu e e lec s he p ob- abili y o he answe o each i em o each HRQoL le el. Using ICC, he HRQoL le el o a gi en subjec can be es ima ed a e answe ing any numbe o i ems. Fu he - mo e, IRT allows us o es ima e he con ibu ion each i em makes o he assessmen o each le el o he a i- able: he In o ma ion Func ion. Measu emen e o is in e sely linked o he in o ma ion used and hence an e o es ima e is a ailable o each assessmen . Based on his heo y, Compu e Adap i e Tes s (CATs) a ise as a psychome ic assessmen echnique adminis e ed h ough a compu e . Fo each esponden , he selec ion o i ems is adap ed o he p io es ima es o he cons uc being assessed [7]. These es s ha e been success ully used in Educa ion and Psychology ields [8] and hey allow a mo e p ac ical assessmen and a mo e accu a e es ima ion o he a iable being measu ed, in his case, HRQoL. CATs esul in he indi- idual adminis a ion o ques ionnai es, as well as in he collec ion and compu a ion o esponses, p o iding ins an esul s [9]. Since he 90 s, when some au ho s ecommended CAT applica ions o heal h [10], a a ie y o CATs ha e been de eloped in he Heal h ield, as hose o mig aine [11,12], heuma oid a h i is [13], os eoa h i is [14], back pain [15], physical he apy [16,17], anxie y [18], cance [19] and paedia ics [20]. All hese CATs ocus on one speci ic condi ion o HRQoL domain, bu hey canno measu e gene ic HRQoL in heal hy o ill subjec s om he gene al popula ion. As i is well known, HRQoL is essen ially a mul idimensional con- cep . This ac could make i di icul o accomplish he unidimensionali y equi ed o he applica ion o IRT [21]. Despi e his ac , we hink i is possible o de elop a calib a ed i em pool o measu e he unde lying con- s uc o gene ic HRQoL. Pu suing his aim, in a p e- ious s udy [22], an expe panel p oposed a pool o 140 i e-le el Like i ems, chosen among se e al HRQoL ques ionnai es alida ed in Spain. Tha pool was i s adminis e ed o a pilo sample and la e o a gene al sample o pa ien s belonging o 7 P ima y Heal h cen- es. Two adminis a ion op ions we e o e ed o he la e sample: 1) on pape 2) on a ouch panel. I em Response Theo y psychome ic p ope ies (disc imina- ion, eliabili y and alidi y) we e e alua ed by means o a Fac o ial Analysis and o he me hods. The In o ma- ion Func ion was analyzed and an applica ion me hod was es ed by means o simula ion: a minimum o 5 i ems and a maximum o 15 we e shown; he i s i em was andomly selec ed among 13 which deal wi h gene ic heal h aspec s and co e ed a b oad HRQoL ange. These 13 ini ial i ems we e selec ed among he mos in o ma i e i ems by an expe panel. The esul was a calib a ed pool o 96 i ems [22]. This pool o i ems showed a ac o ial s uc u e in one dimension (wi h 45% o he a iance explained and a lowes loading o 0,224) and e idenced high eliabili y (C onbach’s al a = 0.99). This manusc ip p esen s he alida ion o CAT- Heal h: a CAT based on he desc ibed calib a ed i em pool, using he men ioned applica ion me hod and implemen ed wi h a ouch sc een in e ace. This alida- ion s udy, ga he ing in o ma ion om a sample o sub- jec s om he Spanish gene al popula ion, pu sues he goal o ob aining a easible and accu a e ins umen o measu e gene ic HRQoL in he clinical se ing. Me hods A c oss-sec ional s udy o subjec s aged o e 18 a end- ing one o he ou pa icipan P ima y Ca e Cen es (PCC) o any eason was ca ied ou , in o de o assess he alidi y o CAT-Heal h. The s udy hence included subjec s wi h ch onic condi ions o acu e pa hologies bu also heal hy subjec s, o ins ance, pa ien s’heal hy ela i es a he PCC. Be o e hei inclusion, all pa ien s we e in o med and p o ided w i en in o med consen , in acco dance wi h he e hical p inciples o he Decla a- ion o Helsinki and he Good Clinical P ac ice guidelines. Da a we e collec ed du ing h ee consecu i e days in each PCC, be ween Feb ua y and Ma ch 2007, by using wo me hods: a able PC (elec onic pencil equi ed) and a ouch sc een panel (nei he mouse no pencil necessa y). Subjec s comple ed a e y sho ini ial ques- ionnai e abou hei age, sex and whe he hey had su - e ed any o he ch onic condi ions p esen ed in a checklis including: anxie y, dep ession, acu e disease, a e ial hype ension, ca diac disease, diabe es, join pain, mig aine, pulmona y disease and “o he diseases”. The pa ien s illed his ch onic condi ions checklis on hei own. In addi ion, he SF-12 Heal h Su ey was adminis e ed using he same de ices [23-25]. CAT- Heal h and SF-12 comple ion imes we e au oma ically eco ded. The CAT-Heal h sys em The CAT-Heal h sys em e alua es he gene ic HRQoL o heal hy o ill subjec s om he gene al popula ion by showing a a iable numbe o i ems (be ween 5 and 15), ex ac ed om a unidimensional calib a ed pool o 96 i ems which had been p e iously de eloped [22]. All he i ems in he pool ha e 5 esponse ca ego ies. The i s Rebollo e al.Heal h and Quali y o Li e Ou comes 2010, 8:147 h p://www.hqlo.com/con en /8/1/147 Page 2 o 8 i em is andomly selec ed among 13 ini ial i ems which co e a b oad ange o he measu ed cons uc (HRQoL) and which ocus on gene ic heal h aspec s. These 13 ini ial i ems we e selec ed among he mos in o ma i e i ems by an expe panel Based on he esponse o his i s ques ion, he sys em selec s he mos in o ma i e i em o be p esen ed as he ollowing ques ion, i e a i ely. The sys em s ops when ei he : 1) he maximum numbe o i ems (n = 15) has been p e- sen ed o he subjec ; o 2) he minimum numbe o i ems has been achie ed, he es ima ion e o is below he uni y and he pe cen age o educ ion o he e o , wi h espec o he p e ious es ima ion, is below 5%. This applica ion me hod was heo e ically es ed by meanso asimula ions udy[22]andi p o idedan accu a e sco e o mode a ely low o high HRQoL le els (nea o gene al popula ion mean sco e). The sco e cal- cula ed by CAT-Heal h sys em has a heo e ical ange be ween -3.85 and +3.87, bu i was no malized o a 50 ± 10 dis ibu ion, in o de o acili a e i s in e p e a- ion:(CATsco e-mean)/SD*10+50.Thehighe he sco e, he be e he HRQoL. E alua ion o he CAT-Heal h sys em CAT-Heal h was e alua ed conside ing h ee di e en c i e ia: 1) easibili y o he sys em in he clinical p ac- ice, in e ms o comple ion ime and es leng h, paying special a en ion o elde ly subjec s ( ou h qua ile o he dis ibu ion). 2) Psychome ic e alua ion, including con en ange co e age, I em Exposu e Ra e (IER) and es p ecision; and 3) Validi y assessmen o CAT- Heal h. Cons uc alidi y was s udied h ough he ana- lysis o di e ences in CAT-Heal h sco es depending on sex ( emales we e expec ed o ha e wo se sco es han males), age (elde ly people we e expec ed o show wo se sco es) and he p esence o epo ed ch onic condi ions ( he highe he numbe o condi ions, he wo se he sco e). I was also hypo hesized ha subjec s wi h one o he lis ed pa hologies should ha e a wo se sco e han subjec s wi hou hem. Mig aine, acu e pa hology, hype ension and “o he pa hologies”, howe e , we e no conside ed o be associa ed o HRQoL a p io i, because he con en o he i ems was no designed o ake in o accoun acu e o silen pa hologies. Finally, CAT-Heal h con e gen alidi y was e alua ed wi h espec o SF-12 physical and men al componen sco es. A mode a e co - ela ion be ween bo h ques ionnai es was expec ed (co - ela ion coe icien s be ween 0.3 and 0.6), as i is usually ound when HRQoL gene ic ques ionnai es a e com- pa ed [26-28] S a is ical analysis Absolu e and ela i e equencies we e used o desc ibe he sample dis ibu ion wi h espec o he nominal a iables (sex and decla ed pa hologies). Mean and s an- da d de ia ion we e used o he con inuous a iables (age, CAT-Heal h and SF-12 sco es, and numbe o decla ed pa hologies). Qua iles we e used in he case o he CAT-Heal h comple ion imes and es leng h. Di - e ences in he numbe o i ems shown o subjec s, acco ding o age and ch onic condi ions, we e assessed by means o a K uskal-Wallis es . Di e ences acco ding o sex we e e alua ed by means o a Mann-Whi ney es . Con en ange co e age o CAT-Heal h was s udied h ough he analysis o loo and ceiling e ec s. I em Exposu e Ra e (IER) was de ined as he a io o he o al numbe o imes a gi en i em is shown o he numbe o imes CAT-Heal h was adminis e ed. Tes p ecision was s udied h ough he analysis o he Tes In o ma ion Func ion (TIF), which is an agg ega e o he in o ma ion p o ided by each i em. Conside ing he adap i e na u e o he sys em and ha he numbe o i ems shown wi hin each es we e limi ed o 15, a TIF using all he i ems in he pool is no ep esen a i e o a ypical es , so a TIF was cons uc ed by using jus he 15 i ems which p o ided mo e in o ma ion a each le el (gi en his was a con inuous ange, a p ocess o disc e iza ion was necessa y and 800 HRQoL le els we e conside ed). No e ha he selec ion o i ems migh change wi hin he HRQoL ange. Di e ences in CAT-Heal h sco es be ween males and emales and be ween subjec s who decla ed ha ing any o he lis ed pa hologies we e assessed by means o a es . Lineal eg ession analysis was pe o med o con- i m he associa ion o each o he pa hologies indepen- den ly wi h he CAT-Heal h sco e. Co ela ion be ween CAT-Heal h sco es and age and he numbe o pa holo- gies was analyzed by means o Pea son and Spea man co ela ion coe icien s, espec i ely. The sample was di ided in o 4 g oups, acco ding o he numbe o decla ed pa hologies (none, 1, 2, 3 o mo e) and also acco ding o he qua iles o he dis ibu ion o age; CAT-Heal h and SF-12 sco es we e compa ed among hese g oups by means o One-Way ANOVA. E ec size was compu ed o CAT-Heal h sco e and he physical and men al componen s o SF-12, when compa ing mean di e ences be ween g oups (age, sex and numbe o pa hologies): Cohen’sdin he case o a es and e a squa e in hecaseo ANOVA.Following heguidelines p oposed by Cohen [29], o es s, an e ec size o 0.1 was conside ed small, 0.3 was conside ed medium and 0.5 was la ge. In he case o ANOVA, 0.01, 0.06 and 0.14 we e conside ed small, medium o la ge, espec i ely. Co ela ions be ween he CAT-Heal h sco e and he SF-12 physical and men al componen s we e assessed by means o Pea son co ela ion coe icien s. Rebollo e al.Heal h and Quali y o Li e Ou comes 2010, 8:147 h p://www.hqlo.com/con en /8/1/147 Page 3 o 8 Resul s The cha ac e is ics o he sample (N = 396) a e desc ibed in Table 1. The mean age (SD) was 48.6 (17.7) yea s and wo hi ds o he subjec s we e emale. Mo eo e , one hi d o he pa icipan s did no epo any ch onic condi ion om he lis . Rega ding he analysis o he p ac ical use o CAT- Heal h, he median comple ion ime o CAT-Heal h was 81 seconds (IQ ange = 59 seconds-118 seconds), inc easing wi h age (p < 0.001) om 66 seconds in he g oup o subjec s aged unde Pc25 o 107 seconds in he g oup o subjec s aged o e Pc75. The median num- be o he CAT-Heal h i ems shown o subjec s was 8 (IQ ange = 6-10). The e we e no s a is ically signi i- can di e ences (p > 0.05) in he numbe o i ems shown o males and emales, o subjec s om di e en age g oups and o subjec s acco ding o he numbe o decla ed pa hologies. The CAT-Heal h non-no malized sco e anged be ween -2.4 and +2.8 ( heo e ical ange be ween -3.8 and +3.8). Nei he ceiling no loo e ec s we e ound. The mean no malized sco e was 50.88 (6.02), wi h a minimum o 34.71 and a maximum o 80.89. The mean Physical Componen Summa y (PCS) sco e o SF-12 was 46.84 (10.1) and i anged be ween 19.12 and 67.06; he mean Men al Componen Summa y (MCS) sco e was 46.88 (10.96), anging be ween 11.18 and 67.83. Figu e 1 shows CAT-Heal h I em Exposu e Ra e (IER). None o he i ems in he pool had an exposu e a e o 100%, while 36 i ems (37.5%) we e no shown a any ime (20 o hem could no be shown wi h he chosen applica ion me hod). IER was o e 5% o 26 i ems (27.1%). The bes -15 TIF is shown in Figu e 2 along wi h he s anda d e o o each le el o he scale. The Tes In o ma ion Func ion (TIF) peaked be ween le el -1 and 0 o HRQoL, which co esponds o he no mal- ized sco es 35.58 and 52.19. In his pa o he scale, he e o was below 0.2. Rega ding cons uc alidi y, in Table 2 he compa i- son o CAT-Heal h and SF-12 sco es is shown wi h espec o sex, age and he numbe o decla ed pa holo- gies. CAT-Heal h and SF-12 MCS sco es we e highe o males han o emales (p < 0.0001). The e ec size (ES) o CAT-Heal h (0.46) was simila o ha o MCS (0.51) and highe han ha o PCS (0.12). The CAT- Heal h sco e showed a nega i e s a is ically signi ican co ela ion wi h age ( = -0.351; p < 0.001) as SF-12 PCS did ( = -0.255; p < 0.001); SF-12 MCS did no show a s a is ically signi ican co ela ion wi h age. By di iding he sample in ou age g oups (unde 34.27 yea s; 34.27-46.26; 46.27-61.19; o e 61.19), ANOVA analysis showed s a is ically signi ican di e ences (p < 0.0001) among age g oups o CAT-Heal h and PCS sco es. The CAT-Heal h sco e also showed a nega i e s a is ically signi ican co ela ion wi h he numbe o pa hologies decla ed by he esponden s ( = -0.548; p < 0.01), like PCS and MCS ( = -0.337 and = -0.262, espec i ely; p < 0.01). The ES o CAT-Heal h (0.12) was highe han hose o PCS (0.06) and MCS (0.01). Figu e 3 shows he CAT-Heal h sco es o subjec s who decla ed su e ing om any o he lis ed pa holo- gies and o hose who did no . Di e ences we e all s a- is ically signi ican (p < 0.05), excep ha o acu e pa hology and mig aine. These di e ences emained Table 1 Sample desc ip ion (N = 396) Sample desc ip ion All (N = 396) Male (N = 130) Female (N = 266) Mean age (SD) 48.61 (17.67) 49.03 (18.45) 48.40 (17.30) F equency o decla ed pa hologies: n (%) Anxie y 70 (17.7%) 11 (8.5%) 59 (22.2%) Dep ession 56 (14.1%) 5 (3.8%) 51 (19.2%) Acu e pa hology 98 (24.7%) 31 (23.8%) 67 (25.2%) A e ial hype ension 66 (16.7%) 23 (17.7%) 43 (16.2%) Ca diac disease 32 (8.1%) 13 (10%) 19 (7.1%) Diabe es 37 (9.3%) 19 (14.6%) 18 (6.8%) Join pain 125 (31.6%) 32 (24.6%) 93 (35%) Mig aine 36 (9.1%) 5 (3.8%) 31 (11.7%) Pulmona y disease 31 (7.8%) 13 (10%) 18 (6.8%) O he pa hology 85 (21.5%) 34 (26.2%) 51 (19.2%) Numbe o sel - epo ed condi ions 036.9% 43.1% 33.8% 133.8% 36.2% 32.7% 216.4% 12.3% 18.4% 3 o mo e 12.9% 8.4% 15.1% Rebollo e al.Heal h and Quali y o Li e Ou comes 2010, 8:147 h p://www.hqlo.com/con en /8/1/147 Page 4 o 8 s a is ically signi ican when analysing each o he pa hologies adjus ing by he es o hem in a eg ession analysis. SF-12 PCS did no de ec di e ences be ween pa ien s wi h o wi hou anxie y, dep ession, acu e dis- ease, pulmona y disease and mig aine; nei he did SF-12 MCS de ec di e ences in he case o diabe es, acu e disease, ca diac disease, pulmona y disease, a e ial hype ension and “o he disease”. The co ela ion coe icien s be ween CAT-Heal h sco es and SF-12 PCS ( = 0.547) and MCS ( = 0.346) we e mode a e and s a is ically signi ican (p < 0.001). Discussion Se e al compu e -adap i e es s (CAT) e alua ing heal h ou comes ha e been de eloped and alida ed in ecen yea s, bu , o ou knowledge, he CAT-Heal h sys em is one o he i s CATs designed o e alua e gene ic HRQoL.The esul so hep esen s udyshow ha CAT-Heal h is easible, alid and e icien o HRQoL e alua ion: i s psychome ic p ope ies we e sa is ac o y when e alua ing HRQoL in a wide ange o subjec s a ending p ima y ca e se ings. The use o HRQoL as a heal h ou come measu e is becoming mo e impo an in he e alua ion o pa ien ca e and heal h se ices. I is usually e alua ed using ques ionnai es based on CTT, adminis e ed h ough pencil and pape . Once he answe s o a subjec a e collec ed, he ques ionnai e has o be coded and sco ed, and hen he esul s ha e o be in e p e ed. This is a ime consuming p ocess and, he e o e, i is expensi e, especially o ollow-ups in clinical p ac- ice. These p oblems cons i u e a ba ie ha p e en s he e alua ion o HRQoL in he clinical se ing [3]. CAT ins umen s clea ly inc ease “measu emen e i- ciency”( he a io o a measu e’s psychome ic Figu e 1 The CAT-Heal h I em Exposu e Ra e (IER). The IER is he a io o he o al numbe o imes one i em is p esen ed o he numbe o imes CAT-Heal h is adminis e ed. Figu e 2 The CAT-Heal h bes -15 Tes In o ma ion Func ion (TIF). The TIF is used o e alua e he es p ecision o di e en HRQoL le els. Rebollo e al.Heal h and Quali y o Li e Ou comes 2010, 8:147 h p://www.hqlo.com/con en /8/1/147 Page 5 o 8 soundness o he esponse bu den he measu es imposes) [7]. CAT ins umen s dec ease he esponse bu den, diminishing he numbe o ques ions o be answe ed by he subjec ; hey ha e small loo and ceiling e ec s, when hey use an ex ensi e i em bank; hey educe he e o o measu emen and so hey measu e much mo e accu a ely; hey a e lexible, adap ing hemsel es o he ai le el o he espon- den and also o speci ic measu emen con ex s [11-20]. Table 2 Compa ison o he CAT-Heal h and SF-12 sco es acco ding o sex, age and numbe o sel - epo ed condi ions Compa ison o he CAT-Heal h and SF-12 sco es acco ding o sex, age and numbe o decla ed pa hologies CAT-Heal h SF-12 PCS SF-12 MCS Sex Male (N = 130) 52.68 (6.51) 47.65 (9.43) 50.89 (9.45) Female (N = 266) 49.99 (5.56) 46.44 (10.40) 44.92 (12.57) p( es ) < 0.0001 0.262 < 0.0001 E ec size (Cohen’sd) 0.455 0.120 0.513 Age < 34.27 yea s (N = 100) 53.96 (5.69) 50.46 (7.98) 47.59 (11.19) 34.27-46.26 yea s (N = 98) 51.33 (5.53) 47.82 (10.42) 47.24 (11.65) 46.27-61.19 yea s (N = 99) 49.66 (5.29) 44.83 (10.32) 45.36 (12.30) > 61.19 yea s (N = 99) 48.54 (6.18) 44.20 (10.36) 47.34 (12.70) p (ANOVA) < 0.0001 < 0.0001 0.537 E ec size (e a squa e) 0.116 0.062 0.006 Numbe o sel - epo ed condi ions None (N = 146) 54.23 (5.72) 50.56 (7.84) 50.80 (9.04) 1 pa hology (N = 134) 50.89 (4.86) 46.74 (10.23) 46.52 (12.40) 2 pa hologies (N = 65) 47.75 (7.42) 44.14 (10.81) 42.78 (12.70) 3 o mo e pa hologies (N = 51) 45.24 (4.66) 39.85 (9.97) 41.86 (13.59) p (ANOVA) < 0.0001 < 0.0001 < 0.0001 E ec Size (e a squa e) 0.272 0.124 0.082 es : * ( p<0.05 ) ** ( p<0.001 ) Figu e 3 Di e ences in he CAT-Heal h sco es acco ding o sel - epo ed condi ions (means and 95% con idence in e als). Rebollo e al.Heal h and Quali y o Li e Ou comes 2010, 8:147 h p://www.hqlo.com/con en /8/1/147 Page 6 o 8 His o ically, HRQoL has been conside ed a mul i- dimensional concep . Some au ho s a gue, howe e , ha he e is an unde lying cons uc ha a ec s all hem and can be di ec ly measu ed [30]. In a p e ious pape , we p esen ed he de elopmen and calib a ion o a gen- e ic HRQoL i em bank, so ha he assessmen o HRQoL wi h a single dimension was easible [22]. Using ha i em bank, we de eloped one o he i s IRT-based CATs o he e alua ion o gene ic HRQoL, he alida- ion o which we ha e p esen ed in his manusc ip . The e alua ion o CAT-Heal h has shown ha CAT- Heal h sys em median comple ion ime is eally sho (unde 1minu eandahal ).E en houghweha e shown ha he comple ion ime was associa ed wi h he age o he esponden s (i mus be aken in o accoun ha elde ly people a e no amilia wi h he in e aces used o ill in he ques ionnai e), he median comple ion ime o he g oup o indi iduals in he las qua ile o age (o e 61.19 yea s) was s ill well benea h 2 minu es. Also, he e alua ion o gene ic HRQoL wi h CAT-heal h equi ed a median o 8 i ems, a es leng h simila o ha o published CATs o speci ic g oups o pa holo- gies: AM-PAC-CAT [16], o pos -acu e ca e, showed a mean o 6.6 i ems; each o he 5 domains o he CAT- 5D-QOL [15], o back pain, 4.4 o 6.6 i ems; and he Anxie y-CAT [18] 6 o 8 i ems. In his alida ion s udy, 60 i ems we e shown a leas once, which ep esen s he 79% o he i ems in he pool ha could be shown, acco ding o he chosen applica- ion me hod (a leas 20 i ems will ne e appea , as hey a e no amongs he 15 mos in o ma i e o any HRQoL le el). The ac ha none o he i ems had an IER o 100% and ha he numbe o di e en i ems used a some poin was high indica es ha he sys em ac ually adap ed he i ems p esen ed o he indi iduals in he sample, poin ing o he adequacy o a CAT o HRQoL measu emen . Wi h espec o he con en ange co e age o CAT- Heal h, he equency dis ibu ions o he non-no mal- ized sco es o he subjec s unde s udy we e no mally dis ibu ed, wi h no loo no ceiling e ec s, as expec ed o a gene ic ins umen in his sample. On he con a y, he abo emen ioned domain-speci ic CATs p esen ed ceiling e ec s: 10% o AM-PAC- CAT [16] ( ha p esen ed a oughly no mally dis ibu- ed sco e) and 0% o 6.1% o he di e en CAT-5D- QOL scales [15]. No malized CAT-Heal h sco es (50 ± 10) had a simila ange o hose o SF-12 PCS and sligh ly na owe han hose o MCS. The pe o - mance o CAT-Heal h in HRQoL assessmen in he s udied sample, which co e ed a b oad spec um o indi iduals a ending p ima y ca e se ings, indica es a good po en ial o he e alua ion o he gene al popula ion. The analysis o he TIF showed ha CAT-Heal h is a e y disc imina i e measu emen ool in he ange o sco es be ween 35 and 52, which co esponds o a no - mal o sligh ly de e io a ed HRQoL, he mos equen s a us in people who demand heal h ca e a a p ima y ca e cen e. Fu u e esea ch is needed in o de o add new i ems o he pool which would allow CAT-Heal h o co e a b oade ange o HRQoL. The analysis o alidi y demons a ed ha he CAT- Heal h sco e is a alid gene ic measu e o HRQoL. CAT-Heal h adequa ely de ec s he hypo hesized di e - ences be ween male and emale subjec s, as well as be ween di e en age g oups and among g oups by he numbe o decla ed pa hologies. Recen ly, he Adap i e Measu emen o Change (AMC) has been p oposed as a easible and e ec i e me hod o measu ing indi idual change using CATs [31]. The a ailabili y o an e o es ima e o each sub- jec , in his ype o measu emen s, u ns high he p eci- sion o CATs, like CAT-Heal h, in o use ul ins umen s o moni o ing HRQoL. The sensi i i y o change o CAT-Heal h will be add essed in he u u e by means o a longi udinal s udy. The alida ion o he sys em in a b oade andom sample o he gene al popula ion and he adap a ion o i s use in English a e also planned. These s udies will include a la ge numbe o clinical a iables o allow o a de ailed e alua ion o he con- s uc alidi y o he sys em. Conclusions Al hough domain speci ic CATs exis o a ious a eas o HRQoL, CAT-Heal h is one o he i s IRT-based CAT designed o e alua e gene ic HRQoL and i has p o en easible, alid and e icien , when adminis e ed o a b oad sample o indi iduals a ending p ima y ca e se ings. The educed numbe o i ems equi ed o HRQoL e alua ion and he esul ing sho ened comple ion ime, oge he wi h he cha ac e is ics inhe en o compu e - ized ins umen s, such as au oma ic sco ing and in e - p e a ion o esul s, make o CAT-Heal h a p ac ical ins umen o clinical se ings, as P ima y Ca e Cen es. These wo ac s, along wi h i s sound psychome ic p ope ies, which open he possibili y o e alua ing HRQoL changes a he indi idual le el, a e impo an ad an ages o he CAT-Heal h sys em o e o he gen- e ic ques ionnai es based on CTT. No e The Spanish C.A.T-Heal h g oup is o med by esea ch- e s om: BAP Heal h Ou comes Resea ch, O iedo, Spain; 4 h A ea P ima y Ca e Cen e s, P incipali y o As u ias Heal h Se ice, O iedo, Spain; Uni e sidad de O iedo, Psychology Depa men , O iedo, Spain; Uni e sidad Rebollo e al.Heal h and Quali y o Li e Ou comes 2010, 8:147 h p://www.hqlo.com/con en /8/1/147 Page 7 o 8 Complu ense de Mad id, Me hodology Depa men , Mad id, Spain; Ins i u Municipal d’In es igació Mèdica (IMIM-Hospi al del Ma ), Heal h Resea ch Uni Se - ices, Ba celona, Spain. Acknowledgemen s The au ho s would like o hank he Heal h Di ec o o he 4 h A ea P ima y Ca e Can e s o As u ias and all he GPs and polls e s o he P ima y Ca e Cen es ha pa icipa ed in his s udy: “El C is o”,“Collo o”,“Paulino P ie o”, “Puma ín”,“Tea inos”,“Vallobín-Concinos”and “Ven anielles”. Thanks, also, o all he pa ien s who kindly spen some minu es o hei ime o comple e he ques ionnai e and o Ad iana Pé ez o he p oo eading o his a icle. Au ho de ails 1 BAP Heal h Ou comes Resea ch, Calle Azcá aga 12 A, 33010, O iedo, Spain. 2 Uni e sidad de O iedo, Psychology Depa men , Plaza Feijoo s/n, 33003, O iedo, Spain. 3 Ins i u Municipal d’In es igació Mèdica (IMIM-Hospi al del Ma ), Doc o Aiguade 88, 08003, Ba celona, Spain. 4 CIBER en Epidemiología y Salud Pública (CIBERESP), Doc o Aiguade 88, 08003, Spain. Au ho s’con ibu ions All au ho s a e esponsible o he epo ed manusc ip and ha e pa icipa ed in i s concep and design, analysis and in e p e a ion o da a and, inally, in i s d a ing and e iew. Compe ing in e es s The au ho s he eby decla e ha he e is no con lic o in e es s, inancial ag eemen o o he in ol emen wi h any company whose p oduc igu es in he submi ed wo k. Pablo Rebollo, Ignacio Cas ejón, Jesús Cue o, Guille mo Villa and Helena Díaz-Cue o wo k a BAP Heal h Ou comes Resea ch, which is he Applican o a Spanish pa en applica ion (P200701072) and a Eu opean pa en applica ion (EPO8013312) ela ed o he CAT-Heal h sys em. Recei ed: 4 June 2010 Accep ed: 3 Decembe 2010 Published: 3 Decembe 2010 Re e ences 1. McHo ney CA: Heal h s a us assessmen me hods o adul s: pas accomplishmen s and u u e challenges. Annu Re Public Heal h 1999, 20:309-335. 2. Deyo RA, Pa ick DL: Ba ie s o he use o heal h s a us measu es in clinical in es iga ion, pa ien ca e, and policy esea ch. Med Ca e 1989, 27:S254-S268. 3. Rebollo P, Valde as JM, O ega F: [P og ess in Spain o he desc ibed ba ie s o he use o pe cei ed heal h s a us measu es in he clinical p ac ice]. Med Clin (Ba c) 2005, 125:703-705. 4. Hamble on RK, Swmina han J, Roge s HJ: Fundamen als o i em esponse heo y Newbu y Pa k, CA: Sage Publica ions; 1991. 5. Hamble on RK: I em Response Theo y: in oduc ion and bibliog aphy. Psico hema 1990, 1:97-107. 6. Hamble on RK: Applica ions o i em esponse heo y o imp o e heal h ou comes assessmen : de eloping i em banks, linking ins umen s, and compu e -adap i e es ing. In Ou comes assessmen in cance . Edi ed by: Lipscomb J, Go ay CC, Snyde C. Camb idge, U.K.: Camb idge Uni e si y P ess; 2005:445-464. 7. Cook KF, O’Malley KJ, Roddey TS: Dynamic assessmen o heal h ou comes: ime o le he CAT ou o he bag? Heal h Se Res 2005, 40:1694-1711. 8. Wa e JE J : Concep ualiza ion and measu emen o heal h- ela ed quali y o li e: commen s on an e ol ing ield. A ch Phys Med Rehabil 2003, 84: S43-S51. 9. Ha niss M, Am mann D, Cook D, Johnson K: Conside a ions o de eloping in e aces o collec ing pa ien - epo ed ou comes ha allow he inclusion o indi iduals wi h disabili ies. Med Ca e 2007, 45:S48-S54. 10. Re icki DA, Cella DF: Heal h s a us assessmen o he wen y- i s cen u y: i em esponse heo y, i em banking and compu e adap i e es ing. Qual Li e Res 1997, 6:595-600. 11. Bjo ne JB, Kosinski M, Wa e JE J : Using i em esponse heo y o calib a e he Headache Impac Tes (HIT) o he me ic o adi ional headache scales. Qual Li e Res 2003, 12:981-1002. 12. Wa e JE, Kosinski M, Bjo ne JB, Bayliss MS, Ba enho s A, Dahlo CG, e al: Applica ions o compu e ized adap i e es ing (CAT) o he assessmen o headache impac . Qual Li e Res 2003, 12:935-952. 13. Ma in M, Kosinski M, Bjo ne JB, Wa e JE, Maclean R, Li T: I em esponse heo y me hods can imp o e he measu emen o physical unc ion by combining he modi ied heal h assessmen ques ionnai e and he SF-36 physical unc ion scale. Qual Li e Res 2007, 16:647-660. 14. Kosinski M, Bjo ne JB, Wa e JE, Sulli an E, S aus WL: An e alua ion o a pa ien - epo ed ou comes ound compu e ized adap i e es ing was e icien in assessing os eoa h i is impac . J Clin Epidemiol 2006, 59:715-723. 15. Kopec JA, Badii M, McKenna M, Lima VD, Say e EC, D o ak M: Compu e ized adap i e es ing in back pain: alida ion o he CAT-5D- QOL. Spine 2008, 33:1384-1390. 16. Je e AM, Haley SM, Tao W, Ni P, Moed R, Meye s D, e al:P ospec i e e alua ion o he AM-PAC-CAT in ou pa ien ehabili a ion se ings. Phys The 2007, 87:385-398. 17. Haley SM, Gandek B, Siebens H, Black-Scha e RM, Sinclai SJ, Tao W, e al: Compu e ized adap i e es ing o ollow-up a e discha ge om inpa ien ehabili a ion: II. Pa icipa ion ou comes. A ch Phys Med Rehabil 2008, 89:275-283. 18. Wal e OB, Becke J, Bjo ne JB, Fliege H, Klapp BF, Rose M: De elopmen and e alua ion o a compu e adap i e es o ‘Anxie y’(Anxie y-CAT). Qual Li e Res 2007, 16(Suppl 1):143-155. 19. Pe e sen MA, G oen old M, Aa onson N, Faye s P, Sp ange s M, Bjo ne JB: Mul idimensional compu e ized adap i e es ing o he EORTC QLQ-C30: basic de elopmen s and e alua ions. Qual Li e Res 2006, 15:315-329. 20. Allen DD, Ni P, Haley SM: E iciency and sensi i i y o mul idimensional compu e ized adap i e es ing o pedia ic physical unc ioning. Disabil Rehabil 2008, 30:479-484. 21. Reckase MD: Uni ac o la en ai models applied o mul i ac o es : esul s and implica ions. Jou nal o S a is ics 1979, 4:207-230. 22. Rebollo P, Ga cía-Cue o E, Za dain PC, Cue o J, Ma ínez I, Alonso J, e al: [De elopmen o he CAT-Heal h, he i s Adap i e Compu e ized Tes (CAT) o he e alua ion o Heal h-Rela ed Quali y o Li e in Spain. Med Clin (Ba celona) 2009, 133:241-251. 23. Vilagu G, Valde as JM, Fe e M, Ga in O, Lopez-Ga cia E, Alonso J: [In e p e a ion o SF-36 and SF-12 ques ionnai es in Spain: physical and men al componen s]. Med Clin (Ba c) 2008, 130:726-735. 24. Gandek B, Wa e JE, Aa onson NK, Apolone G, Bjo ne JB, B azie JE, e al: C oss- alida ion o i em selec ion and sco ing o he SF-12 Heal h Su ey in nine coun ies: esul s om he IQOLA P ojec . In e na ional Quali y o Li e Assessmen . J Clin Epidemiol 1998, 51:1171-1178. 25. Vilagu G, Fe e M, Rajmil L, Rebollo P, Pe manye -Mi alda G, Quin ana JM, e al:[The Spanish e sion o he Sho Fo m 36 Heal h Su ey: a decade o expe ience and new de elopmen s.]25. Gac Sani 2005, 19:135-150. 26. Badia X, Alonso J: Validi y and ep oducibili y o he Spanish Ve sion o he Sickness Impac P o ile. J Clin Epidemiol 1996, 49:359-365. 27. Sch ag A, Selai C, Jahanshahi M, Quinn NP: The EQ-5D–a gene ic quali y o li e measu e-is a use ul ins umen o measu e quali y o li e in pa ien s wi h Pa kinson’s disease. J Neu ol Neu osu g Psychia y 2000, 69:67-73. 28. B azie J, Robe s J, Tsuchiya A, Busschbach J: A compa ison o he EQ-5 D and SF-6 D ac oss se en pa ien g oups. Heal h Econ 2004, 13:873-884. 29. Cohen J: S a is ical Powe Analysis o he Beha io al Sciences. second edi ion. Law ence E lbaum Associa es; 1988. 30. F yback DG, Pal a M, Che epano D, Bol D, Kim JS: Compa ison o 5 heal h- ela ed quali y-o -li e indexes using i em esponse heo y analysis. Med Decis Making 2010, 30:5-15. 31. Kim-Kang G, Weiss DJ: Adap i e Measu emen o Indi idual Change. Jou nal o Psychology 2008, 216:49-58. doi:10.1186/1477-7525-8-147 Ci e his a icle as: Rebollo e al.: Valida ion o a compu e -adap i e es o e alua e gene ic heal h- ela ed quali y o li e. Heal h and Quali y o Li e Ou comes 2010 8:147. Rebollo e al.Heal h and Quali y o Li e Ou comes 2010, 8:147 h p://www.hqlo.com/con en /8/1/147 Page 8 o 8