Age-specific vaccination coverage estimates for influenza, human papillomavirus and measles containing vaccines from seven population-based healthcare databases from four EU countries : The ADVANCE project

Age-speci ic accina ion co e age es ima es o in luenza, human papilloma i us and measles con aining accines om se en popula ion-based heal hca e da abases om ou EU coun ies – The ADVANCE p ojec Toon B aeye a,b, ⇑ , Hanne-Do he Embo g c , Ana Llo en e-Ga cía d , Consuelo Hue a d , Elisa Ma ín-Me ino d , Tali a Dua e-Salles e , Gio gia Danieli ,g , La a T amon an ,g , Daniel Weibel h , Ch is McGee i , Ma co Villa j , Rosa Gini k , Ma i Leh inen l , Lina Ti ie sky m , Mi iam S u kenboom n,o,p a Sciensano, B ussels, Belgium b Uni e si y o Hassel , Hassel , Belgium c S a ens Se um Ins i u , Copenhagen, Denma k d BIFAP Da abase, Spanish Agency o Medicines and Medical De ices, Mad id, Spain e Fundació Ins i u Uni e si a i pe a la ece ca a l’A enció P imà ia de Salu Jo di Gol i Gu ina (IDIAPJGol), Ba celona, Spain Conso zio A senàl.IT, Vene o Region, I aly g Epidemiological In o ma ion o Clinical Resea ch om an I alian Ne wo k o Family Paedia icians (PEDIANET), Pado a, I aly h E asmus Uni e si y Medical Cen e , Ro e dam, The Ne he lands i Uni e si y o Su ey, Su ey, UK j ATS della Val Padana, C emona, I aly k Agenzia Regionale di Sani a Della Toscana, Flo ence, I aly l Uni e si y o Tampe e, Tampe e, Finland m Epidemiology, P ize , NY, Uni ed S a es n P95, He e lee, Belgium o VACCINE.GRID, Basel, Swi ze land p Uni e si y Medical Cen e U ech , Julius Global Heal h, U ech , The Ne he lands a icle in o A icle his o y: Recei ed 16 July 2019 Recei ed in e ised o m 20 Feb ua y 2020 Accep ed 28 Feb ua y 2020 A ailable online 12 Ma ch 2020 Keywo ds: Vaccina ion co e age In luenza accines Papilloma i us accines Measles accines P obabili y abs ac Backg ound: The Accele a ed De elopmen o VAccine beNe i - isk Collabo a ion in Eu ope (ADVANCE) is a public–p i a e collabo a ion aiming o de elop and es a sys em o apid bene i - isk moni o ing o accines using exis ing heal hca e da abases in Eu ope. We es ima ed accine co e age om elec onic heal hca e da abases as pa o a i - o -pu pose assessmen o accine bene i - isk s udies. Me hods: A e ospec i e dynamic coho s udy was conduc ed h ough a dis ibu ed ne wo k app oach. Co e age wi h measles- accine o bi h yea 2006, human papilloma i us (HPV)- accine o bi h yea s 1990–2000 and in luenza- accine o bi h yea s 1920–1950 was es ima ed using pe iod-p e alence and in e se p obabili y weigh ing me hods. Se en da abases om ou coun ies pa icipa ed: I aly (Pediane , Val Padana), Spain (BIFAP, SIDIAP), UK (RCGP-RSC, THIN), Denma k (SSI/AUH). Da abase access p o ide s ex ac ed he da a, ans o med i in o a common s uc u e and an an R-sc ip locally. The c ea ed ou pu ables we e sha ed and pooled a a cen al se e . Resul s: The o al s udy popula ion comp ised 274,616 pe sons o measles- accine, 2,011,666 pe sons o HPV- accine and 14,904,033 pe sons o in luenza- accine. Measles- accine co e age a ied om 84.3% (Denma k) o 96.5% (I aly, Val Padana) o he i s dose and om 82.8% (I aly, Val Padana) o 90.9% (UK) o he second dose a he age o 7 yea s. The HPV- accine co e age, agg ega ed o e bi h yea s 1997–2000, anged om 60% (UK) o 88.3% (Denma k) a he age o 15 yea s. The in luenza- accine co e age o he in luenza seasons om 2009 o 2015 o pe sons aged 65 yea s and mo e was oughly s able a ound 43% in Denma k and a ound 68% in he UK while a dec ease om 58 o 50% was obse ed in Ca alonia (Spain). h ps://doi.o g/10.1016/j. accine.2020.02.082 0264-410X/Ó2020 The Au ho (s). Published by Else ie L d. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/). ⇑ Co esponding au ho a : Sciensano, Rue Julie e Wy sman 14, 1050 Ixelles, Belgium. E-mail add ess: [email p o ec ed] (T. B aeye). Vaccine 38 (2020) 3243–3254 Con en s lis s a ailable a ScienceDi ec Vaccine jou nal homepage: www.else ie .com/loca e/ accine Conclusions: We ob ained de ailed, age-speci ic co e age es ima es hough a common p ocedu e. We dis- cussed be ween da abase compa abili y and compa abili y o published na ional es ima es. Ó2020 The Au ho (s). Published by Else ie L d. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/). 1. In oduc ion The impac o a accine on he bu den o a disease in popula- ions di e s om an assessmen in indi iduals. Because o e ec s like he d immuni y, he popula ion co e age should be included in bene i - isk accine s udies [1]. While se e al sou ces such as su eys, social secu i y da a o heal h insu ance claims da a [2] can be used o ob ain co e age es ima es, elec onic heal hca e eco ds (eHR) a e likely o gain impo ance. Thei g owing impo - ance as sou ce o apidly p o ide e idence on aspec s o accine co e age, bene i s and isks has mul iple easons: I) eHR a e less likely associa ed wi h biases, such as ecall o selec i e sampling bias, ha a e impo an o o he sou ces [3], II) eHR po en ially allow o isk g oup speci ic analysis, a necessa y ea u e since ac- cine ecommenda ions can be isk g oup speci ic and III) eHR allow o age-speci ic, nea eal ime co e age es ima ion wi h a high posi i e p edic i e alue a ela i ely low cos [4–6]. Despi e hese ad an ages, eal wo ld applica ions ha e epo ed issues wi h incomple eness and misclassi ica ion [7–9]. These conce ns and he ac ha he e a e many di e en kinds o eHR-da abases equi e a ‘ i o pu pose’-assessmen p io o hei use in bene i - isk hypo hesis es ing. Mos Eu opean coun ies p oduce egional o na ional accine co e age es ima es, bu da abases and me hods used o es ima e accina ion co e age a y widely and compa abili y is challenging a an in e na ional le el [10–12]. Se e al p ojec s and o ganisa- ions, such as he WHO Cen alized In o ma ion Sys em o In ec- ious Diseases (CISID) [13] and Vaccine Eu opean New In eg a ed Collabo a ion E o (VENICE) [14] aim o imp o e he quali y and compa abili y o accine co e age es ima ion by de ining and implemen ing s anda ds. These p ojec s ha e been in place o se e al yea s and ha e es ablished obus da abases. An al e - na i e app oach o ob ain compa able co e age es ima es om mul iple coun ies is he s anda dized collec ion o da a using su - eys [15,16]. The ADVANCE p ojec di e ed om hese exis ing app oaches as we did no equi e ou pa ne s o p esen co e age es ima es hemsel es no did we collec da a ou sel es. Ins ead, pa icipa ing da abases we e asked o un a sc ip ha ans o med a ailable eHRs in o accina ion co e age es ima es locally. This insu ed ha he s eps o da a cleaning, ans o ma ion and analy- sis we e iden ical be ween da abases. In a p e ious pape , we in es iga ed me hods o deal wi h one sou ce o incomple eness encoun e ed in eHR-da abases; un egis e ed accina ions because o incomple e ollow-up [17]. F om ha pape we selec ed wo me hods ha ake pe son- ime in o accoun when es ima ing co e age. The Accele a ed De elopmen o VAccine beNe i - isk Collabo a- ion in Eu ope (ADVANCE)p ojec was a public p i a ecollabo a ion aiming o de elop and es a sys em o apid bene i - isk moni o - ing o accines using exis ing heal hca e da abases in Eu ope. Fo u he eading on he ADVANCE p ojec and i s code o conduc we e e o h p://www.ad ance- accines.eu/. Se e al s udies we e conduc ed o demons a e he abili y o gene a e e idence on pe - ussis accine co e age, bene i s and isks [18]. The esul s p e- sen ed he e we e pa o a second ound o easibili y assessmen s o c ea e eadiness o s udy addi ional accines beyond pe ussis. In his pape we es ima ed he co e age wi h measles, HPV and in luenza- accines in pa icipa ing eHR-da abases. 2. Ma e ial and me hods 2.1. Design and se ing We conduc ed a e ospec i e dynamic coho s udy using a dis ibu ed ne wo k app oach. 2.2. Desc ip ion o pa icipa ing da abases Se en Eu opean heal hca e da abases pa icipa ed in his s udy (Table 1). The gene al cha ac e is ics o he da abases ha e been desc ibed in mo e de ail be o e [19]. All da abases comp ised elec- onic heal h eco ds. The e was a dis inc ion be ween egional (BIFAP, SIDIAP, PEDIANET, Val Padana) and na ional da abases (THIN, RCGP-RSC, SSI/AUH). The e was also a dis inc ion be ween da abases ha collec ed a subse (BIFAP, SIDIAP, PEDIANET, THIN, RCGP-RSC) o aimed o collec all da a (Val Padana, SSI/AUH) om a gi en a ea. In PEDIANET only child en whose pa en s consen ed linkage o he Vene o egional accina ion egis y we e included. Finally, da a was collec ed mainly om p ima y ca e heal hca e se ices, excep o SSI/AUH and Val Padana whe e linkage be ween egis ies was pe o med. 2.3. Vaccina ions The accines o in e es in his assessmen comp ised measles- con aining accines, HPV- accines and seasonal in luenza accines. Vaccina ion schedules o hese accines di e ed be ween he coun ies included in his s udy: Denma k, I aly, Uni ed Kingdom and Spain (Table 2). Vaccina ion in o ma ion was ex ac ed om he eHR-da abases locally and ans o med in o a common acci- na ion ile which comp ised he ollowing a iables: a pa ien iden i ie , a accine ype (coded as he disease agains which he accine p o ec ed), Ana omical The apeu ic Chemical (ATC-code, acco ding o WHO), a b and name, a eco ded dose numbe and/ o a de i ed dose numbe (de i a ion based on age and sequence) and a da e o accina ion. The b and and ATC-code could be le emp y. Val Padana and RCGP-RSC p o ided a de i ed dose (de i ed om he sequence o obse ed doses). The o he da abases p o- ided a eco ded dose. To de e mine he seasonal in luenza accine co e age, we looked a he da e o accina ion. Fo example, an in luenza accina ion egis e ed be ween 1 July 2010 and 30 June 2011 was included in he 2010/11 seasonal co e age es ima ion. 2.4. Sou ce and s udy popula ion The sou ce popula ion o his assessmen comp ised all sub- jec s egis e ed in he da abases lis ed abo e. Each da abase p o- ided a s a - and end-da e o ollow-up o each pe son. Re- en y in o he da abase, p o iding se e al s a - o end-da es pe pe son, was no pe mi ed. O he a iables a ailable a he indi id- ual le el in he local da a we e gende and bi h da e. P io popu- la ion cha ac e iza ion has shown ha he popula ion cap u ed by he da abases e lec ed he age and gende dis ibu ion in he coun ies [19]. Fo each accine, we de ined s udy coho s by bi h yea s. The age a he s a o ollow-up was an addi ional inclusion c i e ion o measles and HPV- accine co e age es ima ion. This inclusion 3244 T. B aeye e al. / Vaccine 38 (2020) 3243–3254 c i e ion de e mined he amoun o le censo ing. Le censo ing was de ined as un egis e ed accina ions because ollow-up had no s a ed a he ime o accina ion. We es ima ed age-speci ic measles- accine co e age o child en bo n in 2006. Only child en who s a ed ollow-up be o e he age o 100 weeks we e included in he measles analysis. Fo he es ima ion o co e age wi h HPV- accine, he s udy popula ion comp ised all emale child en bo n be ween 1990 and 2000 who s a ed ollow-up be o e he age o nine yea s. To es ima e in luenza- accine co e age in he elde ly (65 yea s and olde ) o he in luenza seasons be ween 2009 and 2015, he s udy popula ion comp ised all pe sons bo n be ween 1920 and 1950. 2.5. Dis ibu ed da a p ocessing In he dis ibu ed app oach, da a access p o ide s ex ac ed and ans o med he da a in o he common da a model (CDM) comp is- ing he indi idual-le el accina ion and popula ion iles desc ibed abo e. An R-sc ip was sen o he da a access p o ide s. The R- sc ip conduc ed cleaning, applica ion o inclusion and exclusion c i e ia, da a collec ion o he a i ion ables, ans o ma ion and co e age es ima ion. Cleaning o he s udy popula ion con- sis ed o emo ing incomple e eco ds, inco ec eco ds (e.g. a nega i e pe iod o ollow-up) and duplica es. Duplica ed accina- ion eco ds we e de ined as ha ing he ‘same dose, same accine componen ’ and ‘same da e, same accine componen ’. A e link- ing he popula ion ile o he accina ion ile, accina ions occu - ing ou side o an indi idual’s s a - and end-da e o ollow-up we e censo ed. The a i ion ables p o ided in o ma ion on: he numbe o pe sons by bi h yea and hei ime in ollow-up, he numbe o pe sons excluded because hei age a he s a o ollow-up exceeded he p ede ined limi , he numbe o accina- ion eco ds. The ables ha he R-sc ip gene a ed con ained age-speci ic co e age es ima es by me hod, week o age, bi h yea , accine componen and dose. Boo s apped 95% con idence in e als we e ob ained o he measles- accine by sampling om he inpu - eco ds andomly wi h eplacemen . A o al o 1000 samples we e aken o he calcula ion o each boo s apped con idence in e al. Tables wi h agg ega ed da a we e sha ed o a emo e esea ch en i onmen h ough a secu e ile ans e p o ocol. Cen al o ou app oach is ha a single R-sc ip was p o ided o all da abases. The sc ip ex ac ed and analysed he da a a ailable o he p e iously desc ibed s udy popula ions. No all da abases we e able o p o ide all es ima es. Pediane could no p o ide in o ma ion on HPV- accine, since only bi h yea s 2006 and 2007 we e linked o accina ion egis ies, no on in luenza in adul s, as his is a paedia ic da abase. Val Padana did no con- ibu e o he in luenza- accine co e age es ima ion because he bi h yea s unde in es iga ion we e no p esen in he da abase. Se e al da abases could no p o ide HPV- accine co e age es i- ma es o he bi h yea s 1990–1993. THIN did no p o ide es i- ma es o he second dose measles accine. THIN and BIFAP did no p o ide all seasonal in luenza- accines es ima es. 2.6. Co e age es ima ion Co e age was es ima ed by bi h yea o e age in weeks. Fo example o pe sons bo n in 2006, we es ima ed he co e age wi h he i s dose measles- accine a 1 week o age, 2 weeks o age, e c. Age in weeks was calcula ed as bi hda es udyda e 7 ounded down. The numbe o pe sons in ollow-up o a leas one day du ing an age in weeks was coun ed. We u he coun ed he numbe o pe - sons who ecei ed a accina ion du ing ha week, and hose who had a egis e ed accina ion p io o ha age in weeks. F om hese coun s we calcula ed wo co e age es ima es: a pe iod p e alence (PP: u i ) es ima e and an in e se p obabili y weigh ed (IPW) es ima e. The PP: u i o age in weeks iis es ima ed as: PP: u i ¼ N accina ed&inFU i NinFU i Wi h PP: u i : he co e age es ima ed by he PP: u-me hod du ing week i,N accina ed&inFU i he numbe o pe sons in ollow-up du - ing week iwho ha e been accina ed p io o o du ing week and NinFU i : he o al numbe o pe sons in ollow-up du ing week i. Table 1 Cha ac e is ics o he da abases included in he i - o -pu pose. Coun y Denma k Spain I aly Uni ed Kingdom Name SSI/AUH BIFAP SIDIAP PEDIANET Val Padana THIN RCGP-RSC Type o o ganisa ion p o iding access Di e en public da a holde s Spanish egula o y agency Public esea ch o ganisa ion P i a e o ganisa ion; accines om public heal h Local public heal h agency Academic License holde (E asmus MC) Cha i y O igin o da a Hospi al discha ge diagnoses linked o popula ion and accina ion egis ies. Na ional heal h ca e Family paedia icians and gene al p ac i ione s medical eco ds Family paedia icians and gene al p ac i ione s medical eco ds Family paedia icians medical eco ds linked o egional accina ion egis e Hospi alisa ion discha ge diagnoses linked o popula ion and accina ion egis ies Gene al p ac i ione s medical eco ds Gene al p ac i ione s medical eco ds Geog aphic sp ead Na ional Mul i egional:9 ou o 17 egions Ca alonia Region Sample om Vene o Region Regional, p o ince Na ional sample Na ional sample Table 2 O e iew o he ecommended age o accina ion by coun y and accine, da a ob ained om eCDC [20]. Coun y Da abases 1 e Measles 2 e Measles 1 e HPV In luenza ecommended o pe sons > 65 yea s Denma k SSI/AUH 15 m 4y 12y Yes I aly PEDIANET – Val Padana 13–15 m 5-6y 12y Yes Spain BIFAP – SIDIAP 12 m 3-4y 11-14y Yes UK RCGP-RSC – THIN 12 m 3y 12-13y Yes T. B aeye e al. / Vaccine 38 (2020) 3243–3254 3245 FU p opo ion;i ¼ NinFU i NinFU Wi h FU p opo ion;i : he p opo ion o pe sons in ollow-up du ing week i,NinFU i : he o al numbe o pe sons in ollow-up du ing week iand NinFU: he o al numbe o pe sons in he bi h coho . Vacc IPW;i ¼ Vacc obse ed;i FU p opo ion;i Wi h Vacc IPW;i : he es ima ed numbe o accina ion du ing week i, Vacc obse ed;i : he obse ed numbe o accina ions du ing week iand FU p opo ion;i : he p opo ion o pe sons in ollow-up du ing week i. IPW i ¼P 0!i Vacc IPW;i NinFU Wi h IPW i : he co e age es ima ed by he IPW-me hod du ing week i,Vacc IPW;i : he es ima ed numbe o accina ion du ing week iand NinFU: he o al numbe o pe sons in he bi h coho . A simple example o illus a e he me hod; when 50% o he s udy coho was in ollow-up du ing age week 3 and 100 accina- ions we e egis e ed du ing his week, we assumed ha 200 accina ions we e adminis e ed a 3 weeks o age (100 egis- e ed + 100 censo ed). The PP. u-me hod elies on he assump ion ha he age-speci ic co e age es ima ed om he pa o he popula ion in ollow-up a any age in weeks ep esen he age-speci ic co e age o he popu- la ion. The IPW-me hod elies on he assump ion ha he p opo - ion o pe sons in ollow-up ecei ing a accine du ing a ce ain age in weeks equals he p opo ion o pe sons no in ollow-up ecei ing a accine. Since his assump ion is likely iola ed o he in luenza- accine s udy popula ion, as in olde age g oups dea h is a common cause o loss o ollow-up, he IPW-me hod was no applied o in luenza- accine. A gene al summa y o hese assump ions is ha wi h he PP. u-me hod we assumed ha he obse ed co e age equalled he s udy popula ion co e age, while wi h he IPW-me hod we es ima ed he co e age. Bo h me hods will deli e biased es ima es when he p obabili y o accina ion di e s in and ou o ollow-up. 3. Resul s 3.1. Measles- accine co e age Fo measles- accine, he se en da abases con ibu ed a o al o 362,063 pe sons bo n in 2006, o which 274,616 we e eligible o analysis. The exclusion o pe sons was due o en e ing he da abase a an age olde han 100 weeks (N = 87,447). A o al o 401,094 ac- cine doses (dose 1 and dose 2) we e eco ded o his s udy popu- la ion. The numbe s by da abase can be ound in he a i ion ables in he supplemen a y ile. Fig. 1 shows he p opo ion o pe sons in he s udy popula ion in ollow-up om bi h o se en yea s o age. The s udy popula ion in Pediane ha had consen ed o be linked o he accine egis y had comple e ollow-up. The p opo ion in ollow-up in he na ional SSI/AUH-da abases emained high. In he egional Val Padana and p ima y ca e da abases he p opo ion o pe sons in ollow-up was mo e dynamic; he p opo ion o pe sons in ollow-up inc eased un il he age-limi (100 weeks o age) se as addi ional inclusion c i e ion. A e which, he p opo ion o pe - sons in ollow-up dec eased. The dec ease was subs an ially o he BIFAP, RCGP-RSC and THIN-da abases. Fig. 2 shows he age-speci ic co e age es ima es o he i s and second dose o measles- accines, using he PP. u-me hod and he IPW-me hod. Table 3 shows he es ima es a se en yea s o age. The s a o up ake o he i s dose o measles- accine anged om 0.6 yea s (SIDIAP) o 1.2 yea s o age (SSI/AUH). A e he ini ial s eep inc ease in co e age a one yea o age, THIN, RCGP- RSC and SSI/AUH epo ed a con inuing inc ease o e ime, ep e- sen ing accina ions a e he ecommended age o accina ion. A second inc ease was seen a he age o i e yea s in he Val Padana- da abase. A se en yea s, he IPW-es ima e o he i s dose anged om 96.5% (Val Padana) o 84.3% (SSI/AUH). Fo he second dose o he measles- accine only RCGP-RSC eached a co e age o 90% (IPW-es ima e). The inc ease in co e - age o e age was less s eep, s epwise o some da abases, and he a ained co e age o he second dose was below he co e age a ained o he i s dose. The minimum di e ence be ween he i s and second dose co e age was 1.5% (BIFAP), he maximum di - e ence was 13.7% (Val Padana). The e we e di e ences be ween coun ies wi h espec o he age a which he co e age wi h he second dose s a ed inc easing. Fo SIDIAP, a second dose co e age o a ound 30% was eached a he age o wo yea s. A se en yea s he IPW-es ima e o he second dose anged om 90.9% (RCGP- RSC) o 80.0% (SSI/AUH). 3.2. HPV- accine co e age The HPV s udy popula ion o emales bo n be ween 1990 and 2000 wi h ollow-up be o e he age o nine yea s comp ised 2,011,666 pe sons. The o al numbe o i s dose eco ded HPV- accine egis a ions o he s udy popula ion was 838,823. Se e al o he da abases did no con ibu e o he bi h coho s om 1990 o 1993. Fo he 1990 bi h coho s o be included, he da abases needed o ha e s a ed ollow-up by 1999 (because he ‘age a s a o ollow-up’-inclusion c i e ion was se a nine yea s) and his in o ma ion needed o be p esen in he inpu - iles. An o e iew o he s udy popula ion by da abase can be ound in he a i ion ables in he supplemen a y ile. The age a he s a o HPV- accina ion di e ed be ween da a- bases as did he co e age a ained a he age o 15 yea s. E.g. in he UK (THIN & RCGP-RSC) HPV- accina ion s a ed one bi h yea la e han in Denma k (SSI/AUH) and Spain (BIFAP). SSI/AUH a ained a HPV- accine co e age o >70% o bi h yea 1994 a e which he co e age con inued o inc ease. The o he da abases, excep o RCGP-RSC and THIN, also epo ed an inc easing co e - age o e he bi h yea s included in his s udy. The a ained HPV- accine i s dose co e age a he age o 15 yea s o emales bo n be ween 1997 and 2000 es ima ed wi h he IPW-me hod anged be ween 60% (RCGP-RSC/THIN) and 88.3% (SSI/AUH) (Table 4) (Fig. 3). A igu e wi h PP. u-es ima es is p o ided in he supple- men a y ile. 3.3. In luenza- accine co e age Fi e da abases cap u ed in o ma ion on in luenza- accines in he elde ly. The numbe o pe sons included in he es ima ion inc eased om 3,686,640 pe sons o he in luenza season 2009– 2010 o 4,932,582 o he in luenza season 2014–2015. A igu e o he p opo ion o pe sons in ollow-up pe da abase pe season and he a i ion ables a e p o ided in he supplemen a y ile. The e we e la ge be ween da abases di e ences in he a ained co e age. SSI/AUH epo ed co e ages be ween 40 and 45% o all seasons, while THIN and RCGP-RSC epo ed co e ages a ound 70%. The e was a downwa d end in co e age in ecen seasons in he SIDIAP-da abase ( om 58% in 2009/10 o 50% co e age in 2013/14 seasons). The o he da abases showed no clea ime- end in es ima ed co e age om he 2009–2010 season o he 2014–2015 season (Fig. 4). In luenza- accine co e age inc eased wi h age o pe sons aged 65 yea s and mo e. The inc ease was ollowed by a decline in he oldes age g oups. In he UK (RCGP-RSC and THIN) he highes co - e age was eached in pe sons aged 77–80 yea s. In Spain (BIFAP 3246 T. B aeye e al. / Vaccine 38 (2020) 3243–3254 and SIDIAP) and Denma k (SSI/AUH) he highes co e age was eached a a la e age, a ound 85–87 yea s. This pa e n was seen o e all s udied in luenza seasons. Due o he low numbe o pe - sons in ollow-up in he oldes age g oups, he PP. u-es ima e became uns able (Fig. 5). 4. Discussion This s udy aimed o assess pa o he sui abili y o eHR- da abases o pa icipa e in dis ibu ed accine s udies. We also aimed o es he b oad use o a common da a model and a com- mon analysis sc ip . We ound ha , gi en ha he da a we e a ail- able, age-speci ic, bu also calenda yea o season-speci ic co e age es ima es could be ob ained h ough his common p oce- du e. Limi a ions howe e exis , bo h wi h espec o me hods o co e age es ima ion and he es ima es ob ained. 4.1. Me hods o co e age es ima ion P e ious esea ch has poin ed o he impo ance o bo h he o al up ake and he imeliness [21]. We he e o e selec ed wo me hods ha allowed o age-speci ic es ima es and could ep e- sen accina ions be o e and a e he ecommended age a acci- na ion. Since le and igh censo ing because o incomple e ollow- up we e p esen in di e en p opo ions in se e al o he da a- bases, he me hods also had o be able o accoun o censo ed e en s. Censo ing was especially p esen in da abases linked o gene al p ac ice. The PP. u- and IPW-me hod accoun ed o censo - ing in di e en ways. PP. u and IPW-es ima es we e compa able when incomple eness in ollow-up was limi ed. As he e was no loss o ollow-up in he Pediane -da abase, he IPW and PP. u-es ima es we e equal. The absence o incomple e ollow-up in he Pediane da abase is an a e ac o he need o ask o con- sen o be linked o he immuniza ion egis y. Consen was only ob ained o pe sons s ill egis e ed in 2015 and bo n in 2006 and 2007. A limi a ion o he PP. u-me hod was he inabili y o accoun o le censo ing; en e ing he da abase a e accina ion caused unde es ima ion [22,23]. As we had se he inclusion c i e ion o he i s dose o measles- accine a a s a ing age o ollow-up o 100 weeks (an age a which mos child en had al eady ecei ed a i s dose o measles- accine) and because ollow-up was dynamic in some da abases, le censo ing was likely. As a esul , he PP. u- es ima es o he i s dose o measles- accine we e lowe han he IPW-es ima es o some da abases. As incomple eness inc eased, he di e ence be ween IPW and PP. u-es ima es became la ge . I was la ges o he THIN and BIFAP-da abase. I he ‘age a he s a o ollow-up’-inclusion c i e ion was se be o e he ecommended age o accina ion, i could a oid mos le -censo ing and he e o e unde es ima ion by he PP. u-me hod. This happened o bo h he second dose o measles- accine and o he i s dose o HPV- accine. O e all, an ‘age a he s a o ollow-up’-inclusion c i e- ion can educe le censo ing, bu i will also educe he s udy popula ion o a smalle sample and ha sample migh no longe be ep esen a i e o he o al popula ion. This c i e ion and he Fig. 1. P opo ion o pe sons in ollow-up, wi h s a o ollow-up be o e he age o 100 weeks and bi h yea 2006 o e age in yea s, by da abase. T. B aeye e al. / Vaccine 38 (2020) 3243–3254 3247 easons o en e ing o lea ing ollow-up he e o e would be in e - es ing opics o u he esea ch. The IPW-me hod accoun s o bo h le and igh censo ing o accina ions, bu can p oduce uns able es ima es when weigh s a e e y small o la ge and bias can accumula e as he me hod sums o e he weekly es ima ed numbe o accina ions. In he BIFAP-da abase he la ge di e ence be ween he IPW (86.9%) and PP. u (78.8%)-es ima es o he i s HPV- accine dose was likely caused by he high p opo ion o incomple e ollow-up (<50%) o e a longe ime pe iod. In such ins ances he PP. u-es ima es migh be p e e ed. We ecommend o always p esen and desc ibe he incomple e- ness o he da abases and o apply se e al me hods. La ge di e - ences be ween me hods can indica e le censo ing, uns able es ima ion because o li le da a o a iola ion o he assump ion o an equal age-speci ic p obabili y o accina ion in and ou o ollow-up. An unequal accina ion p obabili y in and ou o ollow-up will bias bo h me hods, bu i will esul in a bias ha Fig. 2. Co e age wi h measles- accine by age. A: i s dose, PP. u-es ima e, B: i s dose, IPW-es ima e, C: second dose, PP. u-es ima e, D: second dose, IPW-es ima e o e age in yea s by da abase o measles- accine, bi h yea 2006. Table 3 Measles co e age es ima es o he i s and second dose a se en yea s o age and boo s apped 95% con idence in e al. Coun y Da abase Fi s dose measles a age 7 yea s Second dose measles a age 7 yea s PP. u-es ima e IPW-es ima e PP. u-es ima e IPW-es ima e I aly Val Padana 96.1% (95.6–97.2%) 96.5% (95.8–97.9%) 83.8% (83.3–85.3%) 82.8% (82.3–84.3%) Pediane 92.4% (92–93.2%) 92.4% (92–93.2%) 86% (85.3–87%) 86% (85.3–87%) Spain BIFAP 90% (89.5–90.5%) 89.3% (88.9–89.5%) 91.9% (91.6–92.1%) 87.8% (87.5–88%) SIDIAP 92.1% (91.7–92.3%) 94% (94–95.2%) 90.1% (89.5–90.6%) 88.3% (84.6–85.4%) UK RCGP-RSC 92.3% (91.6–93%) 95.2% (94.5–95.9%) 93.1% (92.3–93.9%) 90.9% (90.1–91.7%) THIN 91.5% (91.2–91.7%) 94.1% (93.8–94.3%) / / Denma k SSI/AUH 85.2% (85–85.5%) 84.3% (84.1–84.5%) 80.4% (80.1–80.6%) 80.0% (79.7–80.2%) Table 4 HPV- accine, i s dose co e age es ima es (IPW & PP. u-me hod) a he age o 15 yea s agg ega ed o e bi h yea s 1997–2000, emales. Coun y Da abase PP. u-es ima e IPW-es ima e I aly Val Padana 73.7% 75.8% Spain BIFAP 78.8% 86.9% SIDIAP 77.5% 77.0% UK RCGP-RSC 56.2% 60.0% THIN 57.8% 60.0% Denma k SSI/AUH 88.6% 88.3% 3248 T. B aeye e al. / Vaccine 38 (2020) 3243–3254 Fig. 3. A: p opo ion o emales in ollow-up a he age o 12 yea s o e bi h yea s by da abase, B: HPV- accine, i s dose co e age es ima es a 15 yea s (IPW-me hod) o e bi h yea s (1990–2000) by da abase, C: age-speci ic i s dose HPV- accine co e age es ima es (IPW-me hod) o e age by da abase, bi h yea s 1997–2000. T. B aeye e al. / Vaccine 38 (2020) 3243–3254 3249 accumula es o e he age-speci ic es ima es ob ained wi h he IPW-me hod (as he es ima ed numbe o accines a ‘age in weeks’ 1, 2,... is used o es ima ing he co e age a ‘age in week’ ), while i will bias he PP. u-es ima es o each ‘age in week’-es ima e independen ly. I is he e o e likely o esul in a la ge bias o he IPW-me hod ( his can be u he explo ed in ou co esponding simula ion s udy [17]). We only accoun ed o censo ing because o incomple e ollow- up. We assumed no inco ec o missing egis a ion du ing ollow- up. P e ious s udies ha e shown ha he p esence o a accina ion eco d in an elec onic da abase eliably indica es immuniza ion while he absence o such a eco d can be inaccu a e [24]. This he e o e was a s ong assump ion; e.g. accina ions may be adminis e ed by non- adi ional p o ide s who we e no a ilia ed wi h he da abase [4,25]. 4.2. Compa ison o published co e age es ima es Resea che s ha e been conce ned wi h he alidi y and compa- abili y o co e age es ima es o e di e en s udies, a eas and ime [35]. As me hodological di e ences and di e ences in da a sou ce ha e p o en o be ele an [3,36], hey should be aken in o con- side a ion when compa ing ou es ima es o p e iously published es ima es. We i s conside ed he epo ed ou come; ou es i- ma es we e epo ed as age-speci ic es ima es by bi h yea . O he esea ch migh epo poin es ima es o age g oups by calenda yea . In addi ion o he he e ogenei y in ou comes, we an icipa ed o he me hodological di e ences. Ou me hodology was designed speci ically o accoun o he dynamic ollow-up o popula ions in eHR-da abases. While su i al me hods ha e been applied in co e age es ima ion o in es iga e imeliness o accine up ake and igh -censo ing [37], accoun ing o bo h le - and igh - censo ing is, o ou knowledge, unique o his wo k. Finally, di e - ences in co e age es ima ion can be linked o he sou ce o he da a. Fo example, he da abases linked o p ima y ca e om ce - ain egions migh no be ep esen a i e o he na ional le el. In addi ion, we added an inclusion c i e ia o some o ou examples, possibly u he limi ing he compa abili y o ou es ima es o p e- iously published es ima es. The e o e, while we an icipa ed some di e ences, we also expec ed hem o ha e mul iple causes and i was unclea wha he clinical signi icance (i.e., he magni ude o impac ) o hese di e ences was [36]. An indi idual assessmen o all accines by da abase was necessa y. In ecen yea s se e al measles ou b eaks ha e been epo ed in he Eu opean Region associa ed o immuni y gaps; pocke s o un accina ed child en and adul s [38]. F om he ou coun ies in his s udy, Spain was he only coun y wi h a co e age o e 95% o he i s dose, none ob ained a co e age o e 95% o he second dose. The IPW-es ima es o he i s dose o measles- accine we e bo h below (2–8% Denma k and Spain) and abo e (2–9% UK and I aly) p e iously published na ional es i- ma es (Table 5). Fo he second dose, ou es ima es we e gene - ally below (3–5%) p e iously published na ional es ima es. We obse ed some issues wi h he dose o measles- accines in Val Padana. A sudden inc ease in i s dose co e age some ime a e he ecommended age o accina ion p obably e lec ed w ongly coding he second dose as he i s . Val Padana showed he high- es co e age wi h he i s dose o measles and he lowes wi h he second dose. The i s dose co e age inc eased suddenly a age 5, which is he ecommended age o he second dose. The Val Padana da abase did no egis e he dose a adminis a ion bu de i ed i a e wa ds om he sequence o obse ed accina- Fig. 4. In luenza- accine co e age es ima es (PP. u-me hod) o hose aged o e 65 yea s o e in luenza season by da abase, bi h yea s 1920–1950, in luenza seasons 2010– 2015. 3250 T. B aeye e al. / Vaccine 38 (2020) 3243–3254 ions. We obse ed di e ences be ween da abases om he same coun y: o he measles- accine we obse ed a 1–2% di e ence be ween SIDIAP and BIFAP (Spain) and 3–4% be ween Pediane and Val Padana (I aly). The di e ences we e small gi en ha hey collec ed da a om di e en egions. In 2006 an unp eceden ed ou b eak occu ed in Ca alonia, Spain, a ec ing mos ly young child en (<16 mon hs) [39]. This esul ed in an ea ly adminis a- ion o he i s dose o measles accine, be o e he age o one yea , ollowed by an ea ly adminis a ion o he second measles dose, by he age o 12–15 mon hs. This explained he 30% second dose co e age a a ound 15 mon hs ollowed by he, an icipa ed, second inc ease in co e age a 4 yea s. Since 2007 (HPV- accine was licensed in 2006), HPV- accina ion has been implemen ed in Wes e n-Eu ope. The a - ge ed popula ion ypically consis s o gi ls aged 10–14 yea s. Se - e al modelling s udies ha e sugges ed co e age le els a which HPV se o ypes can be e adica ed anging om 30 o 66% o 86– 94% depending on he se o ype [40]. In 2014, he o e all co e age o he i s dose o HPV in he Eu opean egion was es ima ed a 49.6% (95% CI 40–54%) o emales aged 10–20 [12]. The ou coun- ies included in his s udy had p e iously epo ed high co e ages (74–91%) [26–28], bu es ima es a e known o a y o e di e en bi h coho s in he same coun y. Fo example, in Denma k a sig- ni ican dec ease has been epo ed: a i s dose co e age o 81% was epo ed o gi ls bo n in 2002, he co e age dec eased o 54% o gi ls bo n in 2003 and con inued o dec ease o la e bi h yea s [30]. Ou es ima es a e in gene al in ag eemen (+-5%) wi h p e iously published es ima es excep o he UK es ima es. O i- cial UK HPV- accine co e age es ima es o he i s dose ha e been a ound 85% since 2006. We howe e ound a lowe es ima e o 60% ha was consis en o e ime o bo h da abases. The mos likely explana ion is ha HPV- accina ion is no always p o ided by he GP, since i is ou inely o e ed in schools o gi ls aged 12 o 13 yea s. The BIFAP da abase ecen ly conduc ed a s udy o al- ida e hei HPV- accina ion da a quali y and con i med ha BIFAP was ‘a po en ial da a sou ce o HPV accine esea ch’ [41]. They epo ed egion-speci ic co e age be ween 70.6 and 99.8% o bi h coho s bo n be ween 2000 and 2002, which included he 78.8% ob ained in his s udy [42]. P e ious esea ch epo ed ha in luenza- accine co e age has been low in pe sons aged 65 yea s and mo e. Only a ew coun ies, among which he UK, had achie ed 75% co e age in he pe iod 2009–2011 [33]. In luenza- accine co e age was es ima ed a 58.6% in he 2010–2011 season and 62.7% in he 2009–2010 season in pe sons aged 65 yea s and mo e in Na a e, Spain. In he 2012– 2013 season he co e age in Mad id was es ima ed a 56.6% in pe - sons aged 65 yea s and mo e [43]. Ou in luenza es ima es by sea- son we e below (2–7%) hese p e iously published es ima es o all da abases, bu showed compa able ime ends. Ou esul s on in luenza co e age by age ag eed wi h p e ious esea ch and p e- sen ed he age-g oup om 77 o 94 yea s as he age-g oup wi h he highes co e age. The e a e se e al limi a ions associa ed wi h ou app oach. Fi s he c ea ion o he inpu - iles was a da abase-speci ic p ocess. The Fig. 5. In luenza- accine co e age (PP. u-me hod) o e age in yea s by season, (A) RCGP-RSC, (B) SSI/AUH, (C) SIDIAP, bi h yea s 1920–1950, in luenza seasons in 2009–2015. T. B aeye e al. / Vaccine 38 (2020) 3243–3254 3251