Modeling the potential efficiency of a blood biomarker-based tool to guide pre-hospital thrombolytic therapy in stroke patients

Vol.:(0123456789) 1 3 The Eu opean Jou nal o Heal h Economics (2023) 24:621–632 h ps://doi.o g/10.1007/s10198-022-01495-1 ORIGINAL PAPER Modeling hepo en ial e iciency o ablood bioma ke ‑based ool oguide p e‑hospi al h omboly ic he apy ins oke pa ien s Elizabe hPa ody‑Rua1,2 · Alejand oBus aman e3 · JoanMon ane 4,5 · Ma iaRubio‑Vale a6,7 · Da idSe ano8· SoledadPé ez‑Sánchez5 · AlbaSánchez‑Viñas1 · Césa Gue a a‑Cuella 9 · An oniSe ano‑Blanco7,10,11 Recei ed: 18 Oc obe 2021 / Accep ed: 21 June 2022 / Published online: 27 July 2022 © The Au ho (s) 2022 Abs ac Objec i es S oke ea men wi h in a enous issue- ype plasminogen ac i a o ( PA) is e ec i e and e icien , bu as i s bene i s a e highly ime dependen , i is essen ial o ea he pa ien p omp ly a e symp om onse . This s udy e alua es he cos -e ec i eness o a blood bioma ke es o di e en ia e ischemic and hemo hagic s oke o guide p e-hospi al ea men wi h PA in pa ien s wi h suspec ed s oke, compa ed wi h s anda d hospi al managemen . The s anda d ca e o pa ien s su e ing s oke consis s mainly in diagnosis, ea men , hospi aliza ion and moni o ing. Me hods A Ma ko model was buil wi h ou heal h s a es acco ding o he modi ied Rankin scale, in adul pa ien s wi h suspec ed mode a e o se e e s oke (NIHSS 4-22) wi hin 4.5 hou s a e symp om onse . A Spanish Heal h Sys em pe spec- i e was used. The ime ho izon was 15 yea s.Quali y-adjus ed li e-yea s (QALYs) and li e-yea s gained (LYGs) we e used as a measu e o e ec i eness. Sho - and long- e m di ec heal h cos s we e included. Cos s we e exp essed in Eu os (2022). A discoun a e o 3% was used. P obabilis ic sensi i i y analysis and se e al one-way sensi i i y analyses we e conduc ed. Resul s The use o a blood- es bioma ke compa ed wi h s anda d ca e was associa ed wi h mo e QALYs (4.87 s. 4.77), mo e LYGs (7.18 s. 7.07), and g ea e cos s (12,807€ s. 12,713€). The ICER was 881€/QALY. P obabilis ic sensi i i y analysis showed ha he bioma ke es was cos -e ec i e in 82% o i e a ions using a h eshold o 24,000€/QALY. Conclusions The use o a blood bioma ke es o guide p e-hospi al h ombolysis is cos -e ec i e compa ed wi h s anda d hospi al ca e in pa ien s wi h ischemic s oke. Keywo ds S oke· Bioma ke s· PA· Cos -e ec i eness * An oni Se ano-Blanco an oni.se [email p o ec ed] 1 Teaching, Resea ch andInno a ion Uni , Pa c Sani a i San Joan de Déu, San BoideLlob ega , Spain 2 P ima y Ca e P e en ion andHeal h P omo ion Ne wo k ( edIAPP), Ba celona, Spain 3 S oke Uni , Hospi al Uni e si a i Ge mans T ias i Pujol, Badalona, Spain 4 Neu o ascula Resea ch Labo a o y, Vall d’Heb on Ins i u e o Resea ch (VHIR), Ba celona, Spain 5 Ins i u e de Biomedicine o Se ille, IBiS/Hospi al Uni e si a io Vi gen del Rocío/CSIC/Uni e si y o Se ille andDepa men o Neu ology, Hospi al Uni e si a io Vi gen Maca ena, Se ille, Spain 6 Head o Quali y andPa ien Sa e y, Pa c Sani a i San Joan de Déu. Ins i u de Rece ca San Joan de Déu, San BoideLlob ega , Spain 7 CIBER o Epidemiology andPublic Heal h (CIBERESP), Mad id, Spain 8 F eelance Heal h Economis , Ba celona, Spain 9 Facul ad de Ciencias de la Salud, Uni e sidad Icesi, Cali, Colombia 10 Pa c Sani a i San Joan de Déu. Ins i u de Rece ca San Joan de Déu, Men al Heal h Di ec o a e, C/Camí Vell de la Colònia, 25, 08830San BoideLlob ega , Ba celona, Spain 11 Depa amen de Medicina. Facul a de Medicina i Ciències de la Salu , Uni e si a de Ba celona, Ba celona, Spain 622 E.Pa ody-Rua e al. 1 3 In oduc ion S oke is one o he leading causes o dea h and disabil- i y wo ldwide [1, 2]. Addi ionally, s oke a ec s pa ien s’ quali y o li e [3] and in ol es a conside able economic bu den du ing hospi aliza ion and subsequen discha ge [3–5], mainly due o hospi al s ays and ehabili a ion [5]. Th ombolysis wi h ecombinan issue plasminogen ac i- a o ( PA) o ischemic s oke (IS) is e ec i e [6, 7] and, acco ding o a li e a u e e iew, is an e icien ea men [8]. Howe e , i has a educed he apeu ic window (4.5h a e s oke onse ) and i s bene i s a e highly ime depend- en . The e o e, i is essen ial o con i m IS diagnosis and o ea pa ien s p omp ly a e symp om onse . Delays in diagnosis con i ma ion, which depend on neu oimaging echniques conduc ed on admission, a e among he ba - ie s o imely acu e s oke ea men [9] and a e associ- a ed wi h low a es o PA ea men and poo e s oke ou comes [10]. Howe e , symp om ecogni ion and ime o anspo o hospi al/accessibili y o expe ise would appea o g ea e ba ie s [11]. To educe ea men delays, s udies ha e assessed p e- hospi al h ombolysis. S udies in mobile s oke uni s (MSUs) sugges ha p e-hospi al commencemen o in a- enous h ombolysis can imp o e unc ional ou comes in s oke pa ien s [12–14]. MSUs a e equipped wi h a compu e omog aphy scanne , poin -o -ca e labo a o y, elemedicine [15, 16], s oke iden i ica ion algo i hm a dispa che le el, and a p e-hospi al s oke eam [13]. How- e e , al hough MSUs ha e shown o be cos -e ec i e in some s udies, hei g ea cos a ound 1M€/yea is una - o dable o mos heal hca e sys ems [17]. Using blood bioma ke s in he ambulance o di e en ia e s oke ypes wi hou special equipmen o medical pe sonnel would be a mo e p ac ical and cheape al e na i e o MSUs [14, 18]. Howe e , he e iciency o hese bioma ke s is s ill unknown. In a ecen s udy, ou g oup showed he easibili y o his me hod by using highly speci ic blood es s o de ec an IS pa ien subg oup ha migh bene i om p e-hospi al in e en ions, such as in a enous h ombolysis, e en wi h- ou neu oimaging [19]. A wo-bioma ke panel including e inol-binding p o ein and N- e minal p o-B- ype na iu- e ic pep ide iden i ied IS pa ien s wi h 100% speci ici y and sensi i i y a es a ound 20%. Mo eo e , a h ee-bioma ke panel also including glial ib illa y acid p o ein, measu ed wi h a high-sensi i i y assay, imp o ed sensi i i y o 50%, main aining 100% speci ici y. These esul s a e cu en ly unde e alua ion in he BIOFAST s udy (Bioma ke s o Ini ia ing Onsi e and Fas e Ambulance S oke The apies, ClinicalT ials.go iden i ie : NCT04612218 s udy), which measu es hese bioma ke s in he p e-hospi al scena io wi h poin -o -ca e es de ices. This s udy de eloped a hypo he ical economic model o e alua e he cos -e ec i eness o hese blood bioma ke es s o guide p e-hospi al PA ea men s. s anda d ca e ( h ombolysis a he hospi al) in pa ien s wi h suspec ed IS. Me hods App oach A cos -e ec i eness s udy based on a Ma ko model was conduc ed in pa ien s wi h suspec ed s oke wi h < 4.5h om symp om onse and wi h a Na ional Ins i u es o Heal h S oke Scale sco e o 4–22 (mode a e o se e e s oke). A Spanish Heal h Sys em pe spec i e was adop ed. The Span- ish public heal hca e sys em p o ides uni e sal co e age o ci izens and o eign na ionals and is mainly unded h ough axes. I is a decen alized sys em and he 17 Spanish egions con ol heal h planning, public heal h, and heal h se ice managemen [20]. Na ional and in e na ional economic e alua ion me hodological guidelines we e ollowed [21, 22]. The 15-yea ime ho izon was based on he mean age o s oke diagnosis (70yea s) and li e expec ancy in Spain (85yea s). A discoun a e o 3% was used o bo h cos s and ou comes [21]. In e en ion The in e en ion was a bioma ke es om he s udy by Bus aman e e al. [19], consis ing o he combina ion o e i- nol-binding p o ein > 52µg/mL and N- e minal p o-B- ype na iu e ic pep ide > 4062pg/mL, p o iding 20% sensi i i y and 100% speci ici y o IS (Table1S Supplemen al Ma e- ial). This bioma ke combina ion was used as base-case in he economic model. Sensi i i y analyses we e conduc ed wi h a h ee-bioma ke panel including glial ib illa y acid p o ein, desc ibed in he same s udy wi h 51.5% sensi i i y and 100% speci ici y o IS. Compa a o The s anda d ca e o pa ien s su e ing s oke consis s mainly in diagnosis, ea men , hospi aliza ion, and moni- o ing. Ini ially, a neu oimaging es is conduc ed o ule ou hemo hage. In addi ion, analy ical es s and complemen a y es s a e done. I he s oke is ischemic and he pa ien mee s he h ombolysis c i e ia, PA IV is adminis e ed. Heal h ou comes The main heal h ou come was measu ed in quali y-adjus ed li e-yea s (QALYs) gained. This e ec i eness measu e was chosen because s oke pa ien s’ unc ional s a us ypically 623 Modeling hepo en ial e iciency o ablood bioma ke ‑based ool oguide p e‑hospi al… 1 3 wo sens conside ably, which educes heal h- ela ed qual- i y o li e. Deg ee o disabili y o dependence in he daily ac i i ies o people who ha e su e ed a s oke was meas- u ed using he modi ied Rankin scale (mRS) o neu ologic disabili y. The p obabili ies o mRS o IS we e ob ained om pooled andomized con olled ials [7] ha e alua ed he use o PA s. placebo. In ace eb al hemo hagic (ICH) mRS p obabili ies came om a clinical ial [23]. Rega d- ing he bioma ke es , mRS alues we e es ima ed om an MSUs s udy [15]. This s udy compa ed 3-mon h unc ional ou comes a e in a enous h ombolysis in pa ien s wi h acu e ischemic s oke who had ecei ed eme gency mobile ca e o con en ional ca e. P e iously published da a on u ili- ies we e used o de e mine he QALY o each mRS le el [24, 25]. All-cause mo ali y p obabili ies we e de e mined by gende and age and applied o pa ien s in he long- e m phase one yea a e s oke, acco ding o Spanish mo ali y ables, om he Na ional S a is ics Ins i u e [26], adjus ed o he ela i e isk o ha ing a p e ious s oke. The Na ional S a is ics Ins i u e a es we e con e ed in o p obabili ies. Addi ionally, we adjus ed mo ali y p obabili ies by p opo - ion o men (0.58) and women (0.42) su e ing s oke. This in o ma ion was ob ained om RENISEN (Na ional S oke Regis y o he Spanish Socie y o Neu ology) [27, 28], a s oke egis y om 42 cen e s in Spain. We also used LYGs as an e ec i eness measu e. Decision model The model was designed based on p e ious published model-based economic e alua ions in s oke [29, 30]. We combine a decision ee wi h a Ma ko model. The model was alida ed by clinical expe s. Ma ko s a es ep esen ou come ca ego ies acco ding o mRS. Fou mRS ca ego- ies we e used acco ding o pa ien s’ measu emen equi e- men s: mRS 0–1 (non-disabled pa ien , equi ing only sec- onda y p e en ion s a egies), mRS 2–3 (disabled pa ien s also equi ing ehabili a ion), mRS 4–5 (se e ely disabled pa ien s, equi ing long- e m nu sing ca e), and mRS 6 (s oke dea h). Figu e1 shows model s uc u e and possible ansi ions be ween heal h s a es. The di e en heal h s a es we e de ined o be clinically and economically ele an . The model has a wo-phase s uc u e: acu e (sho un ep- esen ed by a decision ee model) and long e m (Ma ko model). The acu e phase consis s o pa ien managemen and ou comes om s oke onse o 90days. The model hen ollows a long- e m phase wi h wo di e en cycle leng hs: 91days o a yea a e s oke and om one yea o 15yea s wi h an annual cycle leng h; he e o e, a hal -cycle co ec- ion was made. The model was es ed o cons uc alida ion, e i i- ca ion, and c oss- alida ion by clinical expe s (JM, AB) and e alua ed using a i s -o de Mon e Ca lo simula ion in Mic oso Excel 2016. Model assump ions we e as ollows: (1) he p opo ion o pa ien s unde going s anda d ca e s. es bioma ke was 50%; (2) when he es bioma ke is posi i e o IS, heo e i- cally, PA could be applied in he ambulance i he pa ien mee s he c i e ia o h ombolysis. Subsequen ly, he pa ien would be ans e ed o he hospi al o con inue wi h s and- a d ca e; (3) p obabili ies o heal h-s a e ansi ion we e he same o bo h g oups; (4) a e he i s yea o s oke, pa ien s emained in he same mRS unless hey had a ecu - en s oke o died; (5) in case o ecu ence, he same ype o s oke was assigned; and (6) a conse a i e isk op ion o annual ecu ence was adop ed and we assumed ha ecu - en s oke a es we e equal ac oss all ca ego ies o mRS, in acco dance wi h p e ious s udies [24, 25]. Table1 shows he model pa ame e s and dis ibu ions. Use o  esou ces andcos s In line wi h he s udy pe spec i e, di ec heal h cos s we e included using a mic o-cos ing s a egy. Fi e aspec s we e conside ed o cos es ima ions: (1) Compa ison o he al e na i es. The bioma ke es was he di e en ia ing cos . All o he esou ces we e conside ed o be he same, excep o he p opo ion o esou ces used; (2) Type: IS o ICH; (3) Func ional s a us acco ding o mRS; (4) Cos - alloca ion ime: acu e phase, discha ge a e h ee mon hs, h ee mon hs du ing he i s yea and i s yea , long- e m; and (5) Des ina ion a discha ge, o alloca e co esponding cos s o ins i u ionalized pa ien s. Clinical and economic e alua ion s udies on s oke we e used as a basis o esou ce use acco ding o he i e a o emen ioned aspec s. An elec onic su ey was used o de e mine he use o clinical esou ces in di e en Span- ish hospi als. Twen y-eigh neu ologis s om eigh egions pa icipa ed, o whom 13 answe ed all he ques ions. Finally, ou s oke neu ologis s alida ed he in o ma ion on esou ce use. Acu e phase cos s we e as ollows: neu oimaging, labo a- o y es s, leng h-o -s ay including inpa ien ca e, pha ma- cological ea men ( PA and o he d ugs acco ding o s oke ype), and ehabili a ion (physio he apy, speech he apy and occupa ional he apy, he p opo ion o pa ien s, and he numbe o sessions). We included he cos o ea men o symp oma ic hemo hagic ans o ma ion as an ad e se e en om PA adminis a ion (2% o pa ien s, acco ding o da a om Ca alonia’s Da abase-CICAT). Cos s included om discha ge o 90days and om 90days o one yea we e: ollow-up ou pa ien isi s wi h he neu ologis and/o p ima y ca e physician, ehabili- a ion, addi ional o hopedic esou ces co e ed by he heal h sys em (wheelchai , walke , o cane), ollow-up 624 E.Pa ody-Rua e al. 1 3 neu oimaging, and labo a o y es s. Fo cos s om one yea onwa d, we conside ed he cos o s oke ecu ence and ollow-up ou pa ien isi s whe e necessa y, acco ding o su ey esul s and alida ion by s oke physicians/clini- cal expe s. I was assumed ha pa ien s a e p esc ibed an annual blood es . Leng h-o -s ay cos s we e included o pa ien s whose des ina ion a discha ge was a con alescen acili y o long- e m s ay. Pa ien s in a long- e m ca e acili y one yea a e he s oke we e assumed o emain he e o he es o hei li es. To al cos was calcula ed by mul iplying he p opo - ion o esou ces used by hei uni cos and by he uni s equi ed o each esou ce. Public heal h se ice a i s a e published in Regional Go e nmen O icial Bulle ins, upda ed o 2022 using he speci ic egional heal hca e con- sume p ice index. D ug cos s we e ob ained om Boo - Plus 2022. Cos s we e exp essed in eu os (2022). Table2 shows he cos s pe heal h s a e. De e minis ic andp obabilis ic sensi i i y analysis We conduc ed an analysis o de e mine he inc emen al cos - e ec i eness a io (ICER). ICER is exp essed ma hema ically in he ollowing exp ession: In his s udy, B is he in e en ion and A he compa a o . De e minis ic one-way sensi i i y analyses we e pe - o med o iden i y a iables signi ican ly in luencing model ou comes: (1) es sensi i i y and speci ici y, (2) using a h ee-bioma ke panel which includes glial ib illa y acid p o ein, (3) es cos , and (4) changing h ombolysis ea - men ( PA o enec eplase—TNK). This en ailed modi ica- ions in mRS p obabili ies and cos s in he economic model. TNK is non-in e io o PA as a ea men o s oke pa ien s ICER = Inc emen alcos Inc emen alQALYs = Cos B− Cos A QALY B −QALY A . Fig. 1 Model S uc u e. A Sho - e m decision analy ic ee s uc u e o clinical ial ou comes. Ou comes we e: non-disabled pa ien , only equi ing seconda y p e en ion s a egies (modi ied Rankin scale [mRS 0–1]), disabled pa ien s also equi ing ehabili a ion (mRS 2–3), se e ely disabled pa ien s, equi ing long- e m nu sing ca e (mRS 4–5) and dea h (mRS 6) om s oke onse o 90days. Pa ien s en e he model wi h suspec ed s oke wi h < 4.5 h om symp om onse and wi h a Na ional Ins i u es o Heal h S oke Scale sco e be ween 4 and 22 (mode a e o se e e s oke). B Long- e m Ma ko model used o simula e li e ime pa ien ou comes. Pa ien s ansi ion be ween he di e en heal h s a es as indica ed by he a ows 625 Modeling hepo en ial e iciency o ablood bioma ke ‑based ool oguide p e‑hospi al… 1 3 Table 1 Model inpu pa ame e s wi h base-case alues and dis ibu ion p obabili ies used in he sensi i i y analysis Pa ame e Base-case alue Type o dis i- bu ion Pa ame e 1 o dis ibu ionaPa ame e 2 o dis ibu ionbSou ce(s) S anda d ca e p ob- abili ies IS a 3mon hs mRS 0–1 0.42 β 194 269 [7] mRS 2–3 0.21 β 97 366 mRS 4–5 0.19 β 88 375 mRS 6 0.18 β 83 380 Bioma ke es al e na i e p ob- abili ies IS a 3mon hs mRS 0–1 0.50 β 220 218 [15] mRS 2–3 0.25 β 110 328 mRS 4–5 0.14 β 61 377 mRS 6 0.11 β 48 390 Bo h al e na- i es p ob- abili ies ICH a 3mon hs mRS 0–1 0.10 β 9 84 [23] mRS 2–3 0.34 β 32 61 mRS 4–5 0.33 β 31 62 mRS 6 0.23 β 21 72 Annual p ob- abili y o ecu en s oke 0.05 β 5 95 [32] P obabili ies o ansi ion in he i s yea P obabili y om mRS 0–1 o: mRS 0–1 0.70 β 66 28 [33] mRS 2–3 0.07 β 7 87 mRS 4–5 0.09 β 8 86 mRS 6 0.14 β 13 81 P obabili y om mRS 2–3 o: mRS 2–3 0.33 β 24 48 [33] mRS 4–5 0.13 β 9 63 mRS 6 0.14 β 10 62 mRS 0–1 0.40 β 29 43 P obabili y om mRS 4–5 o: mRS 4–5 0.55 β 31 25 [33] mRS 6 0.30 β 17 39 mRS 0–1 0.02 β 1 55 mRS 2–3 0.13 β 7 49 U ili ies 626 E.Pa ody-Rua e al. 1 3 [35] wi h simila sa e y p o iles [38]. We used mRS p ob- abili ies wi h TNK om a me a-analysis [37] compa ed wi h s anda d ca e. Fo he bioma ke es s a egy, we es ima ed he p obabili ies om Kunz e al. [15]. A p obabilis ic sensi i i y analysis (PAS) wi h second- o de Mon e Ca lo simula ion wi h 1000 i e a ions was pe - o med. A willingness- o-pay h eshold o 24,000€/QALY was conside ed consis en wi h na ional ecommenda ions [39]. The e iciency h eshold in he PAS has been changed, using alues o 22, 25, 30, and 60 housand €/QALY [39, 40]. Resul s Table3 shows he de e minis ic cos -e ec i eness analysis and one-way sensi i i y analysis. The blood es bioma ke used o guide p e-hospi al PA adminis a ion compa ed wi h s anda d ca e was associa ed wi h mo e QALYs (4.87 s. 4.77), mo e LYGs (7.18 s. 7.07), and g ea e cos s (12,807€ s. 12,713€). The ICER was 881€/QALY in he base-case scena io. The uni a ia e sensi i i y analysis e ealed ha when es sensi i i y inc eased (e.g., o 30%), QALY and LYG Table 1 (con inued) Pa ame e Base-case alue Type o dis i- bu ion Pa ame e 1 o dis ibu ionaPa ame e 2 o dis ibu ionbSou ce(s) mRS 0–1 0.84 β 85 15 [25] mRS 2–3 0.72cβ 73 27 [24] mRS 4–5 0.45cβ 41 58 [24] Cos sd mRS 0–1 IS 8,101. 51 γ 81,015.10 0.1 Regional Go e n- men O icial Bulle ins; Ca alonia hos- pi al; Andalucia hospi al; Bo Plus 2019; [34]; [35] mRS 2–3 IS 11,099.88 γ 110,991.80 0.1 mRS 4–5 IS 16,257.44 γ 162,574.40 0.1 mRS 6 ISe6,787.84 γ 67,878.40 0.1 mRS 0–1 ICH 7,992.68 γ 79,926.80 0.1 Regional Go e n- men O icial Bulle ins, Ca alonia hos- pi al; Andalucia hospi al; Bo Plus 2019; [35]; [36] mRS 2–3 ICH 14,048.47 γ 140,484.70 0.1 mRS 4–5 ICH 16,758.86 γ 167,588.60 0.1 mRS 6 ICHe10,172.30 γ 101,723.00 0.1 Bioma ke blood es 101.04 γ 1,010.40 0.1 BIOFAST IS ischemic s oke, ICH in ace eb al hemo hagic a Pa ame e 1 is: alpha (be a and gamma dis ibu ion), lowe alue (uni o m dis ibu ion) b Pa ame e 2 is: be a (be a and gamma dis ibu ion), uppe alue (uni o m dis ibu ion) c U ili y by ime ade-o (TTO) me hod d Eu os 2022 e Only hospi aliza ion cos s Table 2 Cos s o ea men (in 2022 eu os) pe s oke sub ype and modi ied Rankin scale (mRS) mRS/ esou ce Ischemic s oke In ace eb al hemo hagic mRS 0–1 Acu e phase 6651.39 6410.48 Discha ge o 90days 759.88 821.94 90days o one yea 495.78 558.78 A e one yea 194.46 201.48 S oke ecu ence 4500.60 5739.49 mRS 2–3 Acu e phase 8019.64 10,438.20 Discha ge o 90days 2139.85 2637.23 90days o one yea 694.88 715.20 A e one yea 244.81 257.84 S oke ecu ence 6053.03 9824.67 mRS 4–5 Acu e phase 10,067.72 10,589.38 Discha ge o 90days 4433.52 4624.48 90days o one yea 941.03 1090.78 A e one yea 444.69 454.22 S oke ecu ence 8168.39 9925.05 627 Modeling hepo en ial e iciency o ablood bioma ke ‑based ool oguide p e‑hospi al… 1 3 also inc eased. The es was cos -e ec i e in all scena ios, excep when es sensi i i y was 56.5% and es speci ici y was 68%. In his case, he bioma ke es had an ICER o 33,045€/QALYs and was a domina ed s a egy (mo e cos ly and ewe LYGs compa ed wi h s anda d ca e). Using h ee bioma ke s (52% sensi i i y and 100% speci ici y), he ICER was 820€/QALYs. Figu e2 ep esen s he cos -e ec i eness accep abili y cu e. The es bioma ke p esen ed a g ea e p obabili y o being cos -e ec i e o willingness- o-pay h esholds highe han 10,000€/QALY. Fo a willingness- o-pay o 24,000€/ QALY, he es bioma ke p esen ed an 82.5% p obabili y o being cos -e ec i e. We an se e al models applying di e en blood es sen- si i i y and speci ici y. The highe he sensi i i y o he es , he lowe he ICER (Fig.1S Supplemen al Ma e ial) and QALYs anged om 0.06 o 0.52 wi h es sensi i i ies o 10% and 99%, espec i ely (Fig.2S Supplemen al Ma e- ial). When he cos o he es was changed using di e en sensi i i y and speci ici y alues (acco ding o Bus aman e’s s udy [19] and hypo he ical alues), he es was cos -e ec- i e when i s uni p ice was 10,000€ and had a sensi i i y o 99% and speci ici y o 100%. As he sensi i i y dec eases, he ICER is no longe cos -e ec i e despi e lowe ing he cos o he es ; o ins ance, a es wi h a sensi i i y o 10% ha cos s 2000€ would no longe be e ec i e (Table4). Table2S (Supplemen al Ma e ial) shows he maximum indi idual cos o he es o make i a cos -e ec i e al e na- i e; i is obse ed ha his cos a ies om 874 o 13,986€, alues ha depend on he diagnos ic alidi y used. The inc emen al cos -e ec i eness plane is p o ided in Supplemen a y Ma e ials (Fig.3S). The Mon e Ca lo simula ion shows ha mos o he simula ions a e wi hin he con idence ellipse, showing ha he esul s a e qui e obus o changes in he alues o he di e en a iables simul aneously. The bioma ke es was cos -e ec i e in 82.4%, 82.5%, 82.8%, and 83.1% o in e ac ions when he willingness- o- pay h eshold was changed by 22,000, 25,000, 30,000, and 60,000 €/QALY, espec i ely. Table 3 Cos -e ec i eness analyses o he bioma ke es o guide p e-hospi al PA: base-case scena io and sensi i i y analyses S sensi i i y, E speci ici y, QALY quali y adjus ed-li e yea , ICER inc emen al cos -e ec i eness a io, LYG li e-yea gained a P e alence o ischemic s oke was 81.9% and o hemo hagic s oke was 18.1%. Speci ici y = 100% b P e alence o ischemic s oke was 49.3% and o hemo hagic s oke was 50.7% Al e na i es Cos s (€) Inc emen al Cos QALY Inc emen al QALY ICER (€/QALY) LYG Inc emen al LYG ICER (€/LYG) Base-case scena io (S = 20%)a S anda d ca e 12,713 4.77 7.07 Bioma ke es 12,807 94 4.87 0.11 881 7.18 0.11 822 Tes sensi i i y = 30% a S anda d ca e 12,713 4.77 7.07 Bioma ke es 12,818 105 4.92 0.16 662 7.24 0.17 618 Tes Sensi i i y = 10% a (hypo he ical wo s scene y) S anda d ca e 12,713 4.77 7.07 Bioma ke es 12,794 81 4.83 0.06 1369 7.13 0.06 1286 Tes sensi i i y = 60% a (hypo he ical bes scene y) S anda d ca e 12,713 4.77 7.07 Bioma ke es 12,862 139 5.09 0.32 433 7.41 0.34 404 Tes sensi i i y = 56.5% and Speci ici y = 68%b S anda d ca e 12,713 4.77 7.07 Bioma ke es 14,692 1979 4.83 0.06 33,045 6.99 -0.07 Domina ed Using h ee bioma ke s (Sensi i i y = 52%; Speci ici y = 100%) S anda d ca e 13,611 4.45 6.86 Bioma ke es 13,742 130 4.61 0.16 820 7.04 0.17 760 Doubling he cos o he es (200€)a S anda d ca e 12,713 4.77 7.07 Bioma ke es 12,893 180 4.87 0.11 1576 7.18 0.11 1689 Changing al eplase o enec eplase (TNK)a S anda d ca e 13,089 5.66 7.99 Bioma ke es 13,156 67 5.73 0.07 956 8.05 0.07 1027 628 E.Pa ody-Rua e al. 1 3 Discussion Ou s udy shows ha using a p e-hospi al blood es o guide PA may be a cos -e ec i e s a egy in a hypo he ical coho o pa ien s wi h suspec ed mode a e o se e e s oke (NIHSS 4-22) wi hin 4.5h a e symp om onse om he Spanish Heal h Sys em pe spec i e. This is among he i s economic e alua ions o bio- ma ke es s o guiding PA in he ambulance in pa ien s wi h IS. Howe e , p e ious s udies showed he e iciency o p e-hospi al ea men o acu e s oke. An economic model s udy ound ha he MSUs s a egy could be cos - e ec i e in di e en scena ios, al hough e iciency was ela ed o popula ion densi y [41]. The MSUs s a egy led o a highe equency o h ombolysis and a highe p opo ion o pa ien s in he ea ly ime in e al (wi hin 90min: 48.1% s. 37.4%; 91–180min: 37.4% s. 50%; 181–270min: 14.5% s. 12.8%), which esul ed in 32,456€/QALY [42]. In he MSUs s a egy, he ambulance was equipped wi h a scanne , a poin -o -ca e labo a o y, and elemedicine capabili ies, making he cos p ohibi i e o ou ine clinical implemen a ion in many coun ies [42]. In ou economic model, he ambulance cos is he same as in usual ca e, and he bioma ke cos (a ound 100€/uni ) is easily a o dable. Adminis a ion o PA in ambulances is a usual p ac ice in he d ip-and-ship model [43], wi h he excep ion ha PA is ini ia ed a p ima y s oke cen e s, while he 1-h in a enous in usion is con inued a he ambulance. The e o e, pa a- medics a e amilia wi h PA in usion and able o ecognize PA- ela ed complica ions, such as bleedings o angioedema. Howe e , PA ini ia ion a con en ional ambulances migh equi e an ex a in e sion in educa ional aspec s. This educa- ion would also imp o e one o he g ea e ba ie s o ea ly ea men s, which is s oke symp oms’ ecogni ion. In he sensi i i y analysis, when he es uni p ice was doubled, i was s ill an e icien al e na i e. In he base-case and all one-way sensi i i y analyses in his s udy, QALYs we e highe o he bioma ke s a egy s. s anda d ca e, mainly due o sho ening o PA adminis a- ion ime, esul ing in mo e pa ien s wi h a a o able mRS [16]. Acco dingly, a s udy showed ha in acu e IS pa ien s who we e ea ed wi h PA, sho e doo - o-needle imes (wi hin 45min) we e associa ed wi h lowe all-cause mo - ali y and lowe all-cause eadmission a 1yea [44]. The economic model used p elimina y sensi i i y (20%) and speci ici y (100%) esul s om he bioma ke s udy [20]. When he sensi i i y is low, pa ien s who a e no diagnosed in he ambulance a e no comp omised, as hey ollow he same cu en s oke pa hways and would be ans e ed o a hospi al whe e s oke diagnosis is made by neu oimaging. The es (which akes abou 15min) needs no addi ional ime as i is pe o med du ing he ans e . On he o he hand, when speci ici y alues we e dec eased, he numbe o alse posi i es inc eases; in hese cases, pa ien s wi h IH a e Fig. 2 Cos -e ec i eness accep abili y cu e 629 Modeling hepo en ial e iciency o ablood bioma ke ‑based ool oguide p e‑hospi al… 1 3 ea ed wi h PA wi h he isk o wo sening. Lowe speci ic- i y alues we e included in he model in sensi i i y analysis. I is impo an o no e ha in addi ion o clinical en o cemen and economic impac , e hical dilemmas would a ise. Rega d- ing he sensi i i y analysis wi h he h ee-bioma ke panel, esul s should be ca e ully in e p e ed, as his subs udy was Table 4 Inc emen al cos -e ec i eness a io (ICER) acco ding o cos , sensi i i y, and speci ici y o he bioma ke es Tes cos (eu os) 50 100 200 500 1000 2000 3000 4000 5000 6000 7000 8000 9000 10,000 Speci ici y 100% ICER Sensi i i y 99% 254 339 510 1021 1873 3577 5281 6985 8689 10,392 12,096 13,800 15,504 17,208 Sensi i i y 60% 294 431 703 1521 2884 5610 8335 11,061 13,787 16,513 19,238 21,964 24,690 Sensi i i y 50% 301 464 789 1766 3394 6650 9905 13,161 16,417 19,673 22,928 26,184 Sensi i i y 40% 324 525 926 2131 4139 8154 12,170 16,185 20,201 24,216 Sensi i i y 30% 385 665 1195 2814 5512 10,910 16,307 21,705 27,102 Sensi i i y 20% 472 870 1666 4055 8036 15,998 23,960 31,923 Sensi i i y 10% 636 1349 2775 7055 14,188 28,454 Speci ici y 97% ICER Sensi i i y 99% 569 685 916 1608 2763 5072 7381 9690 11,999 14,308 16,617 18,926 21,235 23,544 Sensi i i y 60% 792 981 1359 2494 4384 8166 11,948 15,729 19,511 23,293 27,074 Sensi i i y 50% 951 1183 1646 3035 5351 9982 14,613 19,244 23,875 60,582 Sensi i i y 40% 1103 1390 1963 3684 6550 12,284 18,018 23,751 5849,2 Tes cos (eu os) 50 100 200 500 1000 2000 3000 4000 5000 6000 7000 8000 900010,000 Sensi i i y 30% 1460 1858 2655 5044 9027 16,993 24,958 Sensi i i y 20% 2207 2839 4104 789914,222 26,870 Sensi i i y 10% 5158 6715 9828 19,169 34,737 Speci ici y 94% Sensi i i y 99% 907 1055 1351 2239 3720 66819642 12,602 15,563 18,52421,485 24,446 Sensi i i y 60% 1465 1722 2236 3776 6344 11,480 16,615 21,751 7886,2 Sensi i i y 50% 1713 2025 2650 4525 7651 13,901 20,152 3046,2 Sensi i i y 40% 2166 2572 3382 5812 986217,963 26,064 Sensi i i y 30% 3094 3689 4877 8442 14,384 26,267 Sensi i i y 20% 5735 6844 9061 15,714 26,801 Sensi i i y 10% 31,190 Speci ici y 68% ICER Sensi i i y 56.5% 29,740 Values in ed a e no cos -e ec i e ICER: inc emen al cos -e ec i eness a io