Vitamin D supplementation to prevent acute respiratory tract infections: Systematic review and meta-analysis of individual participant data

he bmj | BMJ 2017;356:i6583 | doi: 10.1136/bmj.i6583 RESEARCH 1 open access Vi amin D supplemen a ion o p e en acu e espi a o y ac in ec ions: sys ema ic e iew and me a-analysis o indi idual pa icipan da a Ad ian R Ma ineau,1,2 Da id A Jolli e,1 Richa d L Hoope ,1 Lau en G eenbe g,1 John F Aloia,3 Pe e Be gman,4 Gal Dubno -Raz,5 Susanna Esposi o,6 Da aasambuu Ganmaa,7 Adi A Ginde,8 Emma C Goodall,9 Came on C G an ,10 Ch is ophe J G i i hs,1,2,11 Wim Janssens,12 Ilkka Laaksi,13 Semi a Manaseki-Holland,14 Da id Mauge ,15 Da id R Mu doch,16 Rachel Neale,17 Judy R Rees,18 S e e Simpson,J 19 Iwona S elmach,20 Gee a T ilok Kuma ,21 Mi suyoshi U ashima,22 Ca los A Cama go J 23 ABSTRACT Objec i es To assess he o e all e ec o i amin D supplemen a ion on isk o acu e espi a o y ac in ec ion, and o iden i y ac o s modi ying his e ec . Design Sys ema ic e iew and me a-analysis o indi idual pa icipan da a (IPD) om andomised con olled ials. Da a sOu ces Medline, Embase, he Coch ane Cen al Regis e o Con olled T ials, Web o Science, ClinicalT ials.go , and he In e na ional S anda d Randomised Con olled T ials Numbe egis y om incep ion o Decembe 2015. eligibili y c i e ia O s uDy selec iOn Randomised, double blind, placebo con olled ials o supplemen a ion wi h i amin D3 o i amin D2 o any du a ion we e eligible o inclusion i hey had been app o ed by a esea ch e hics commi ee and i da a on incidence o acu e espi a o y ac in ec ion we e collec ed p ospec i ely and p especi ied as an e icacy ou come. esul s 25 eligible andomised con olled ials ( o al 11 321 pa icipan s, aged 0 o 95 yea s) we e iden i ied. IPD we e ob ained o 10 933 (96.6%) pa icipan s. Vi amin D supplemen a ion educed he isk o acu e espi a o y ac in ec ion among all pa icipan s (adjus ed odds a io 0.88, 95% con idence in e al 0.81 o 0.96; P o he e ogenei y <0.001). In subg oup analysis, p o ec i e e ec s we e seen in hose ecei ing daily o weekly i amin D wi hou addi ional bolus doses (adjus ed odds a io 0.81, 0.72 o 0.91) bu no in hose ecei ing one o mo e bolus doses (adjus ed odds a io 0.97, 0.86 o 1.10; P o in e ac ion=0.05). Among hose ecei ing daily o weekly i amin D, p o ec i e e ec s we e s onge in hose wi h baseline 25-hyd oxy i amin D le els <25 nmol/L (adjus ed odds a io 0.30, 0.17 o 0.53) han in hose wi h baseline 25-hyd oxy i amin D le els ≥25 nmol/L (adjus ed odds a io 0.75, 0.60 o 0.95; P o in e ac ion=0.006). Vi amin D did no in luence he p opo ion o pa icipan s expe iencing a leas one se ious ad e se e en (adjus ed odds a io 0.98, 0.80 o 1.20, P=0.83). The body o e idence con ibu ing o hese analyses was assessed as being o high quali y. cOnclusiOns Vi amin D supplemen a ion was sa e and i p o ec ed agains acu e espi a o y ac in ec ion o e all. Pa ien s who we e e y i amin D de icien and hose no ecei ing bolus doses expe ienced he mos bene i . sys ema ic e iew egis a iOn PROSPERO CRD42014013953. In oduc ion Acu e espi a o y ac in ec ions a e a majo cause o global mo bidi y and mo ali y and a e esponsible o 10% o ambula o y and eme gency depa men isi s in he USA1 and an es ima ed 2.65 million dea hs wo ld- wide in 2013.2 Obse a ional s udies epo consis en independen associa ions be ween low se um concen- a ions o 25-hyd oxy i amin D ( he majo ci cula ing i amin D me aboli e) and suscep ibili y o acu e espi- a o y ac in ec ion.3 4 25-hyd oxy i amin D suppo s induc ion o an imic obial pep ides in esponse o bo h i al and bac e ial s imuli,5-7 sugges ing a po en ial mechanism by which i amin D inducible p o ec ion agains espi a o y pa hogens migh be media ed. Vi a- min D me aboli es ha e also been epo ed o induce o he inna e an imic obial e ec o mechanisms, including induc ion o au ophagy and syn hesis o eac i e ni ogen in e media es and eac i e oxygen in e media es.8 These epidemiological and in i o da a Fo numbe ed a ilia ions see end o a icle. Co espondence o: A R Ma ineau [email p o ec ed]k Addi ional ma e ial is published online only. To iew please isi he jou nal online. ci e his as: BMJ 2017;356:i6583 h p://dx.doi.o g/10.1136/bmj.i6583 Accep ed: 01 Decembe 2016 WhAT IS AlReAdy knoWn on ThIS TopIC Randomised con olled ials o i amin D supplemen a ion o he p e en ion o acu e espi a o y ac in ec ion ha e yielded con lic ing esul s Indi idual pa icipan da a (IPD) me a-analysis has he po en ial o iden i y ac o s ha may explain his he e ogenei y, bu his has no p e iously been pe o med WhAT ThIS STudy AddS Me a-analysis o IPD om 10 933 pa icipan s in 25 andomised con olled ials showed an o e all p o ec i e e ec o i amin D supplemen a ion agains acu e espi a o y ac in ec ion (numbe needed o ea (NNT)=33) Bene i was g ea e in hose ecei ing daily o weekly i amin D wi hou addi ional bolus doses (NNT=20), and he p o ec i e e ec s agains acu e espi a o y ac in ec ion in his g oup we e s onges in hose wi h p o ound i amin D de iciency a baseline (NNT=4) These indings suppo he in oduc ion o public heal h measu es such as ood o i ica ion o imp o e i amin D s a us, pa icula ly in se ings whe e p o ound i amin D de iciency is common doi: 10.1136/bmj.i6583 | BMJ 2017;356:i6583 | he bmj RESEARCH 2 ha e p omp ed nume ous andomised con olled ials o de e mine whe he i amin D supplemen a ion can dec ease he isk o acu e espi a o y ac in ec ion. A o al o i e agg ega e da a me a-analyses inco po a ing da a om up o 15 p ima y ials ha e been conduc ed o da e, o which wo epo s a is ically signi ican p o- ec i e e ec s9 10 and h ee epo no s a is ically signi - ican e ec s.11-13 All bu one o hese agg ega e da a me a-analyses11 epo ed s a is ically signi ican he e o- genei y o e ec be ween p ima y ials. This he e ogenei y migh ha e a isen as a esul o a ia ion in pa icipan cha ac e is ics and dosing egi- mens be ween ials, ei he o which may modi y he e ec s o i amin D supplemen a ion on immuni y o espi a o y pa hogens.14 People wi h ch onic obs uc- i e pulmona y disease who ha e lowe baseline i a- min D s a us ha e been epo ed o de i e g ea e clinical bene i om supplemen a ion han hose wi h highe baseline s a us,15 16 and pa icipan cha ac e is- ics such as age and body mass index ha e been epo ed o modi y he 25-hyd oxy i amin D esponse o i amin D supplemen a ion.17 18 T ea men wi h la ge boluses o i amin D has been associa ed wi h educed e icacy o non-classic e ec s,9 and in some cases an inc eased isk o ad e se ou comes.19 While s udy le el ac o s a e amenable o explo a ion h ough agg ega e da a me a-analysis o published da a, po en ial e ec modi ie s ope a ing a an indi idual le el, such as base- line i amin D s a us, can only be explo ed using indi- idual pa icipan da a (IPD) me a-analysis. This is because subg oups a e no consis en ly disagg ega ed in ial epo s, and adjus men s o po en ial con- ounde s canno be applied simila ly ac oss ials.20 To iden i y ac o s ha migh explain he obse ed he e o- genei y o esul s om andomised con olled ials, we unde ook an IPD me a-analysis based on all 25 an- domised con olled ials o i amin D supplemen a ion o p e en ion o acu e espi a o y ac in ec ion ha we e comple ed up o he end o Decembe 2015. Me hods P o ocol and egis a ion The me hods we e p especi ied in a p o ocol ha was eg- is e ed wi h he PROSPERO In e na ional P ospec i e Regis e o Sys ema ic Re iews (www.c d.yo k.ac.uk/ PROSPERO/display_ eco d.asp?ID=CRD42014013953). App o al by a esea ch e hics commi ee o conduc his me a-analysis was no equi ed in he UK; local e hical pe mission o con ibu e deiden i ied IPD om p ima y ials was equi ed and ob ained o s udies by Cama go e al21 ( he e hics e iew commi ee o he Mongolian Min- is y o Heal h), Mu doch e al22 (Sou he n Heal h and Disabili y E hics Commi ee, e e ence URB/09/10/050/ AM02), Rees e  al23 (Commi ee o he P o ec ion o Human Subjec s, Da mou h College, USA; p o ocol No 24381), Tachimo o e al24 (e hics commi ee o he Jikei Uni e si y School o Medicine, e e ence 26-333: 7839), T an e al25 (QIMR Be gho e Medical Resea ch Ins i u e human esea ch e hics commi ee, P1570), and U ashima e al26 27 (e hics commi ee o he Jikei Uni e si y School o Medicine, e e ence 26-333: 7839). Pa ien and public in ol emen Two pa ien and public in ol emen ep esen a i es we e in ol ed in de elopmen o he esea ch ques ions and he choice o ou come measu es speci ied in he s udy p o ocol. They we e no in ol ed in pa ien ec ui men , since his is a me a-analysis o comple ed s udies. Da a ela ing o he bu den o he in e en ion on pa icipan s’ quali y o li e and heal h we e no me a-analysed. Whe e possible, esul s o his sys em- a ic e iew and me a-analysis will be dissemina ed o indi idual pa icipan s h ough he p incipal in es iga- o s o each ial. eligibili y c i e ia Randomised, double blind, placebo con olled ials o supplemen a ion wi h i amin D3 o i amin D2 o any du a ion we e eligible o inclusion i hey had been app o ed by a esea ch e hics commi ee and i da a on incidence o acu e espi a o y ac in ec ion we e col- lec ed p ospec i ely and p especi ied as an e icacy ou - come. The las equi emen was imposed o minimise misclassi ica ion bias (p ospec i ely designed ins u- men s o cap u e acu e espi a o y ac in ec ion e en s we e deemed mo e likely o be sensi i e and speci ic o his ou come). We excluded s udies epo ing esul s o long e m ollow-up o p ima y andomised con olled ials. s udy iden i ica ion and selec ion Two in es iga o s (ARM and DAJ) sea ched Medline, Embase, he Coch ane Cen al Regis e o Con olled T ials (CENTRAL), Web o Science, ClinicalT ials.go , and he In e na ional S anda d Randomized Con- olled T ials Numbe (ISRCTN) egis y using he elec onic sea ch s a egies desc ibed in he supple- men a y ma e ial. Sea ches we e egula ly upda ed up o, and including, 31 Decembe 2015. No language es ic ions we e imposed. These sea ches we e sup- plemen ed by sea ches o e iew a icles and e e - ence lis s o ial publica ions. Collabo a o s we e asked i hey knew o any addi ional ials. Two in es- iga o s (ARM and CAC) de e mined which ials me he eligibili y c i e ia. Da a collec ion p ocesses IPD we e eques ed om he p incipal in es iga o o each eligible ial, and he e ms o collabo a ion we e speci ied in a da a ans e ag eemen , signed by ep e- sen a i es o he da a p o ide and he ecipien (Queen Ma y Uni e si y o London). Da a we e deiden i ied a sou ce be o e ans e by email. On eceip , h ee in es- iga o s (DAJ, RLH, and LG) assessed da a in eg i y by pe o ming in e nal consis ency checks and by a emp - ing o eplica e esul s o he analysis o incidence o acu e espi a o y ac in ec ion whe e his was pub- lished in he ial epo . S udy au ho s we e con ac ed o p o ide missing da a and o esol e que ies a ising om hese in eg i y checks. Once que ies had been esol ed, clean da a we e uploaded o he main s udy da abase, which was held in STATA IC 12 (College S a ion, TX). he bmj | BMJ 2017;356:i6583 | doi: 10.1136/bmj.i6583 RESEARCH 3 Da a ela ing o s udy cha ac e is ics we e ex ac ed o he ollowing a iables: se ing, eligibili y c i e ia, de ails o in e en ion and con ol egimens, s udy du a ion, and case de ini ions o acu e espi a o y ac in ec ion. IPD we e ex ac ed o he ollowing a iables, whe e a ailable: baseline da a we e eques ed o age, sex, clus e iden i ie (clus e an- domised ials only), acial o e hnic o igin, in luenza accina ion s a us, his o y o as hma, his o y o ch onic obs uc i e pulmona y disease, body weigh , heigh (adul s and child en able o s and) o leng h (in an s), se um 25-hyd oxy i amin D concen a ion, s udy alloca ion ( i amin D e sus placebo), and de ails o any s a i ica ion o minimisa ion a iables. Follow-up da a we e eques ed o o al numbe o acu e espi a o y ac in ec ions (uppe o lowe ), uppe espi a o y ac in ec ions, and lowe espi a- o y ac in ec ions expe ienced du ing he ial; ime om i s dose o s udy d ug o i s acu e espi a o y ac in ec ion (uppe o lowe ), uppe espi a o y ac in ec ion, o lowe espi a o y ac in ec ion i applica- ble; o al numbe o cou ses o an ibio ics aken o acu e espi a o y ac in ec ion du ing he ial; o al numbe o days o wo k o school due o symp oms o acu e espi a o y ac in ec ion du ing he ial; se um 25-hyd oxy i amin D concen a ion a inal ollow-up; du a ion o ollow-up; numbe and na u e o se ious ad e se e en s; numbe o po en ial ad e se eac ions (inciden hype calcaemia o enal s ones); and pa ici- pan s a us a end o he ial (comple ed, wi hd ew, los o ollow-up, died). isk o bias assessmen o indi idual s udies We used he Coch ane Collabo a ion isk o bias ool28 o assess sequence gene a ion; alloca ion concealmen ; blinding o pa icipan s, s a , and ou come assesso s; comple eness o ou come da a; and e idence o selec i e ou come epo ing and o he po en ial h ea s o alid- i y. Two in es iga o s (ARM and DAJ) independen ly assessed s udy quali y, excep o he h ee ials by Ma - ineau and colleagues, which we e assessed by CAC. Disc epancies we e esol ed by consensus. De ini ion o ou comes The p ima y ou come o he me a-analysis was inci- dence o acu e espi a o y ac in ec ion, inco po a ing e en s classi ied as uppe espi a o y ac in ec ion, lowe espi a o y ac in ec ion, and acu e espi a o y ac in ec ion o unclassi ied loca ion (ie, in ec ion o he uppe espi a o y ac o lowe espi a o y ac , o bo h). Seconda y ou comes we e incidence o uppe and lowe espi a o y ac in ec ions, analysed sepa- a ely; incidence o eme gency depa men a endance o hospi al admission, o bo h o acu e espi a o y ac in ec ion; use o an imic obials o ea men o acu e espi a o y ac in ec ion; absence om wo k o school due o acu e espi a o y ac in ec ion; inci- dence and na u e o se ious ad e se e en s; incidence o po en ial ad e se eac ions o i amin D (hype cal- caemia o enal s ones); and mo ali y (acu e espi a- o y ac in ec ion ela ed and all cause). syn hesis me hods LG and RLH analysed he da a. Ou IPD me a-analysis app oach ollowed published guidelines.20 Ini ially we eanalysed all s udies sepa a ely; he o iginal au ho s we e asked o con i m he accu acy o his eanalysis whe e i had been pe o med p e iously, and any dis- c epancies we e esol ed. Then we pe o med bo h one s ep and wo s ep IPD me a-analysis o each ou come sepa a ely using a andom e ec s model adjus ed o age, sex, and s udy du a ion o ob ain he pooled in e - en ion e ec wi h a 95% con idence in e al. We did no adjus o o he co a ia es because missing alues o some pa icipan s would ha e led o hei exclusion om s a is ical analyses. In he one s ep app oach, we modelled IPD om all s udies simul aneously while accoun ing o he clus e ing o pa icipan s wi hin s udies. In he wo s ep app oach we i s analysed IPD o each sepa a e s udy independen ly o p oduce an es ima e o he ea men e ec o ha s udy; we hen syn hesised hese da a in a second s ep.20 Fo he one s ep IPD me a-analysis we assessed he e ogenei y by calcula ion o he s anda d de ia ion o andom e ec s; o he wo s ep IPD me a-analysis we summa ised he - e ogenei y using he I2 s a is ic. We calcula ed he num- be needed o ea o p e en one pe son om ha ing any acu e espi a o y ac in ec ion (NNT) using he Visual Rx NNT calcula o (www.nn online.ne / isu- al x/), whe e me a-analysis o dicho omous ou comes e ealed a s a is ically signi ican bene icial e ec o alloca ion o i amin D compa ed wi h placebo. explo a ion o a ia ion in e ec s To explo e he causes o he e ogenei y and iden i y ac- o s modi ying he e ec s o i amin D supplemen a ion, we pe o med p especi ied subg oup analyses by ex end- ing he one s ep me a-analysis amewo k o include ea men -co a ia e in e ac ion e ms. Subg oups we e de ined acco ding o baseline i amin D s a us (se um 25-hyd oxy i amin D <25 ≥25 nmol/L), i amin D dosing egimen (daily o weekly wi hou bolus dosing e sus a egimen including a leas one bolus dose o a leas 30 000 IU i amin D), dose size (daily equi alen <800 IU, 800-1999 IU, ≥2000 IU), age (≤1 yea , 1.1-15.9 yea s, 16-65 yea s, >65 yea s), body mass index (<25 ≥25), and p esence compa ed wi h absence o as hma, ch onic obs uc i e pulmona y disease, and p e ious in luenza accina ion. To ensu e ha epo ed subg oup e ec s we e independen , we adjus ed in e ac ion analyses o po en ial con ounde s (age, sex, and s udy du a ion). The 25 nmol/L cu -o o baseline 25-hyd oxy i amin D concen a ion in subg oup analyses was selec ed on he g ounds ha i is he h eshold o i amin D de iciency de ined by he UK Depa men o Heal h,29 and he le el below which pa icipan s in clinical ials ha e expe i- enced he mos consis en bene i s o supplemen a ion.30 We also pe o med an explo a o y analysis in es iga ing e ec s in subg oups de ined using he 50 nmol/L and 75 nmol/L cu -o s o baseline ci cula ing 25-hyd oxy i a- min D concen a ion, because obse a ional s udies ha e epo ed ha less p o ound s a es o i amin D de iciency may also associa e independen ly wi h an inc eased isk doi: 10.1136/bmj.i6583 | BMJ 2017;356:i6583 | he bmj RESEARCH 4 o acu e espi a o y ac in ec ion.31 32 To minimise he chance o ype 1 e o a ising om mul iple analyses, we in e ed s a is ical signi icance o subg oup analyses only whe e P alues o ea men -co a ia e in e ac ion e ms we e <0.05. Quali y assessmen ac oss s udies Fo he p ima y analysis we in es iga ed he likelihood o publica ion bias h ough he cons uc ion o a con- ou enhanced unnel plo .33 We used he i e GRADE conside a ions (s udy limi a ions, consis ency o e ec , imp ecision, indi ec ness, and publica ion bias)34 o assess he quali y o he body o e idence con ibu ing o analyses o he p ima y e icacy ou come and majo sa e y ou come o ou me a-analysis (see supplemen- a y able S3). addi ional analyses We conduc ed sensi i i y analyses excluding IPD om ials whe e acu e espi a o y ac in ec ion was a sec- onda y ou come (as opposed o a p ima y o co-p ima y ou come), and whe e isk o bias was assessed as being unclea . We also conduc ed a esponde analysis in pa - icipan s andomised o he in e en ion a m o included s udies o whom end s udy da a on 25-hyd oxy i amin D we e a ailable, compa ing isk o acu e espi a o y ac in ec ion in hose who a ained a se um le el o 75 nmol/L o mo e compa ed wi h hose who did no . Resul s s udy selec ion and iPD ob ained Ou sea ch iden i ied 532 unique s udies ha we e assessed o eligibili y; o hese, 25 s udies wi h a o al o 11 321 andomised pa icipan s ul illed he eligibili y c i e ia ( ig 1). IPD we e sough and ob ained o all 25 s udies. Ou come da a o he p ima y analysis o p o- po ion o pa icipan s expe iencing a leas one acu e espi a o y ac in ec ion we e ob ained o 10 933 (96.6%) o he andomised pa icipan s. s udy and pa icipan cha ac e is ics Table 1 p esen s he cha ac e is ics o eligible s udies and hei pa icipan s. T ials we e conduc ed in 14 coun ies on ou con inen s and en olled pa icipan s o bo h sexes om bi h o 95 yea s o age. Baseline se um 25-hyd oxy i amin D concen a ions we e de e - mined in 19/25 ials: mean baseline concen a ion anged om 18.9 o 88.9 nmol/L. Baseline cha ac e is- ics o pa icipan s andomised o in e en ion and con- ol we e simila (see supplemen a y able S1). All s udies adminis e ed o al i amin D3 o pa icipan s in he in e en ion a m: his was gi en as bolus doses e e y mon h o e e y h ee mon hs in se en s udies, weekly doses in h ee s udies, a daily dose in 12 s udies, and a combina ion o bolus and daily doses in h ee s udies. S udy du a ion anged om se en weeks o 1.5 yea s. Incidence o acu e espi a o y ac in ec ion was he p ima y o co-p ima y ou come o 14 s udies and a seconda y ou come o 11 s udies. IPD in eg i y was con i med by eplica ion o p ima y analyses in published pape s whe e applicable. The p ocess o checking IPD iden i ied h ee ypog aphical e o s in published epo s. Fo he 2012 ial by Manaseki-Holland e al,35 he co ec numbe o epea episodes o ches adiog aphy con i med pneumonia was 134, a he han 138 as epo ed. Fo he ial by Dubno -Raz e al,36 he numbe o pa ien s andomised o he in e en ion a m was 27, a he han 28 as epo ed. Fo he ial by Laaksi e al,37 he p opo ion o men andomised o placebo who did no expe ience any acu e espi a o y ac in ec ion was 30/84, a he han 30/80 as epo ed. isk o bias wi hin s udies Supplemen a y able S2 p o ides de ails o he isk o bias assessmen . All bu wo ials we e assessed as being a low isk o bias o all aspec s assessed. Two ials we e assessed as being a unclea isk o bias owing o high a es o loss o ollow-up. In he ial by Dubno -Raz e al,36 52% o pa icipan s did no com- ple e all symp om ques ionnai es. In he ial by Laaksi e al,37 37% o andomised pa icipan s we e los o ol- low-up. incidence o acu e espi a o y ac in ec ion O e all esul s Table 2 p esen s he esul s o he one s ep IPD me a-analysis es ing he e ec s o i amin D on he p opo ion o all pa icipan s expe iencing a leas one acu e espi a o y ac in ec ion, adjus ing o age, sex, and s udy du a ion. Vi amin D supplemen a ion esul ed in a s a is ically signi ican educ ion in he p opo ion o pa icipan s expe iencing a leas one acu e espi a o y ac in ec ion (adjus ed odds a io 0.88, 95% con idence in e al 0.81 o 0.96, P=0.003; P o he e ogenei y <0.001; NNT=33, 95% con idence in e al 20 o 101; 10 933 pa icipan s in 25 s udies; see Ca es plo , supplemen a y igu e S1). S a is ically Addi ional s udies iden i ied h ough o he sou ces, including con ac wi h esea che s (n=3) S udies iden i ied h ough da abase sea ches (n=717): Medline (n=261) Coch ane CENTRAL (n=146) Embase (n=52) Web o Science (n=258) A ailable da a: IPD ob ained o eligible s udies (n=25) Randomised pa icipan s wi h ou come da a o p ima y analysis (n=10 933) Randomised pa icipan s wi h missing ou come da a o p ima y analysis (n=388) Analysis, p opo ion expe iencing ≥1 acu e espi a o y ac in ec ions: One s ep: da a om 10 933 pa icipan s in 25 s udies analysed Two s ep: da a om 10 899 pa icipan s in 24 s udies analysed (34 pa icipan s in one s udy excluded – ea men e ec no es imable) Unique s udies a e duplica es emo ed (n=532) S udies wi h o al o 11 321 andomised pa icipan s eligible; IPD sough o all (n=25) Excluded (no ele an , e iew a icle, no andomised con olled ials, acu e espi a o y ac in ec ion no p especi ied as e icacy ou come) (n=507) ig 1 | low o s udy selec ion. iPD=indi idual pa icipan da a he bmj | BMJ 2017;356:i6583 | doi: 10.1136/bmj.i6583 RESEARCH 5 able 1 | cha ac e is ics o he 25 eligible ials and hei pa icipan s e e ence se ing (s udy du a ion) Pa icipan s (male: emale) mean (sD) age, yea s ( ange) 25(OH)D no in in e en ion: con ol g oup O al dose o i amin D3 a i no en e ing p ima y analysis/no andomised (%) assay, eQa scheme mean (sD) baseline le el, nmol/l ( ange) baseline le el <25 nmol/l (%) De ini ion Ou come ype Li-Ng 200941 USA (3 mon hs) Heal hy adul s (34:128) 57.9 (13.6) (21.4-80.6) RIA (DiaSo in), DEQAS 63.7 (25.5) (16.0-156.0) 3/150 (2.0) 84:78 50 µg daily, placebo URTI: ≥2 URTI symp oms in absence o alle gy symp oms P ima y 157/162 (96.9) U ashima 201027 Japan (4 mon hs) Schoolchild en (242:188) 10.2 (2.3) (6.0-15.0) -- ND -- 217:213 30 µg daily, placebo URTI: in luenza A/B diagnosed by RIDT o RIDT-nega i e ILI P ima y 334/430 (77.7) Manaseki- Holland 201042 A ghanis an (3 mon hs) P eschool child en wi h pneumonia (257:196) 1.1 (0.8) (0.1-3.3) -- ND -- 224:229 2.5 mg bolus once, placebo LRTI: epea episode o pneumonia—age- speci ic achypnoea wi hou wheeze Seconda y 453/453 (100.0) Laaksi 201037 Finland (6 mon hs) Mili a y consc ip s (164:0) 19.1 (0.6) (18.0-21.0) EIA (IDS OCTEIA) 75.9 (18.7) (41.9-129.0) 0/73 (0.0) 80:84 10 µg daily, placebo ARTI: medical eco d diagnosis P ima y 164/164 (100.0) Majak 201143 Poland (6mon hs) Child en wi h as hma (32:16) 10.9 (3.3) (6.0-17.0) RIA (BioSou ce Eu ope), RIQAS 88.9 (38.2) (31.5-184.7) 0/48 (0.0) 24:24 12.5 µg daily, placebo ARTI: sel epo Seconda y 48/48 (100.0) T ilok-Kuma 201144 India (6 mon hs) Low bi hweigh in an s (970:1109) 0.1 (0.0) (0.0-0.3) -- ND ND 1039:1040 35 µg weekly, placebo ARTI: medical eco d diagnosis o e en s esul ing in hospi al admission Seconda y 2064/2079 (99.3) Lehouck 201215 Belgium (1 yea ) Adul s wi h COPD (145:37) 67.9 (8.3) (48.0- 86.0) RIA (Diaso in), DEQAS 49.8 (29.2) (9.0-159.7) 31/182 (17.0) 91:91 2.5 mg bolus mon hly, placebo URTI: sel epo Seconda y 175/182 (96.2) Manaseki- Holland 201235 A ghanis an (1.5 yea s) In an s (1591:1455) 0.5 (0.3) (0.0-1.0) -- ND ND 1524:1522 2.5 mg bolus 3-mon hly, placebo LRTI: pneumonia con i med by ches adiog aphy P ima y 3011/3046 (98.9) Cama go 201221 Mongolia (7weeks) 3 d/4 h g ade schoolchild en (129:118) 10.0 (0.9) (7.0-12.7) LC-MS/MS, DEQAS 18.9 (9.7) (3.3-61.2) 192/245 (78.4) 143:104 7.5 µg daily, placebo ARTI: pa en epo ed “ches in ec ions o colds” Seconda y 244/247 (98.8) Mu doch 201222 New Zealand (1.5 yea s) Heal hy adul s (81:241) 48.1 (9.7) (18.0-67.6) LC-MS/MS, DEQAS 72.1 (22.1) (13.0-142.0) 5/322 (1.6) 161:161 2×5 mg bolus mon hly hen 2.5 mg bolus mon hly, placebo URTI: assessed wi h symp om sco e P ima y 322/322 (100.0) Be gman 201245 Sweden (1 yea ) Adul s wi h inc eased suscep ibili y o ARTI (38:102) 53.1 (13.1) (20.0-77.0) CLA (DiaSo in), DEQAS 49.3 (23.2) (8.0-135.0) 15/131 (11.45) 70:70 100 µg daily, placebo URTI: assessed wi h symp om sco e Seconda y 124/140 (88.6) Ma chisio 201346 I aly (6mon hs) Child en wi h ecu en acu e o i is media (64:52) 2.8 (1.0) (1.3-4.8) CLA (DiaSo in), ISO9001 65.3 (17.3) (24.7-120.6) 2/116 (1.7) 58:58 25 µg daily, placebo URTI: doc o diagnosed acu e o i is media P ima y 116/116 (100.0) Rees 201323 USA (13 mon hs, a e age) Adul s wi h p e ious colo ec al adenoma (438:321*) 61.2 (6.6) (47.1-7 7.9) RIA (IDS), DEQAS 62.5 (21.3) (30.2-171.6) 0/759 (0.0) 399:360 25 µg daily, placebo URTI: assessed om daily symp om dia y Seconda y 759/759 (100.0) T an 201425 Aus alia (1 yea ) Heal hy olde adul s (343:301) 71.7 (6.9) (60.3-85.2) CLA (DiaSo in), DEQAS 41.7 (13.5) (12.6-105.0) 66/643 (10.3) 430:214 0.75 mg bolus 1.5 mg bolus mon hly, placebo URTI: sel epo ed cold Seconda y 594/644 (92.2) Goodall 201447 Canada (8 weeks) Heal hy uni e si y s uden s (218:382) 19.6 (2.2) (17.0-33.0) -- ND -- 300:300 0.25 mg weekly ( ac o ial wi h ga gling), placebo URTI: sel epo ed cold P ima y 492/600 (82.0) (Con inued) doi: 10.1136/bmj.i6583 | BMJ 2017;356:i6583 | he bmj RESEARCH 6 able 1 | cha ac e is ics o he 25 eligible ials and hei pa icipan s e e ence se ing (s udy du a ion) Pa icipan s (male: emale) mean (sD) age, yea s ( ange) 25(OH)D no in in e en ion: con ol g oup O al dose o i amin D3 a i no en e ing p ima y analysis/no andomised (%) assay, eQa scheme mean (sD) baseline le el, nmol/l ( ange) baseline le el <25 nmol/l (%) De ini ion Ou come ype U ashima 201426 Japan (2 mon hs) High school s uden s (162:85) 16.5 (1.0) (15.0-18.0) -- ND -- 148:99 50 µg daily, placebo URTI: in luenza A diagnosed by RIDT o RIDT nega i e ILI P ima y 247/247 (100.0) G an 201448 New Zealand (9 mon hs: 3mon hs in p egnancy + 6 mon hs in in ancy) P egnan women and o sp ing (0:260 (mo he s) 121:128 (o sp ing)) unbo n LC-MS/MS, DEQAS 54.8 (25.8) (8.0-128.0) 30/200 (15.0) 173:87 (mo he s) 164:85 (o sp ing) Mo he s: 25 µg 50 µg daily In an s: 10 µg 20 µg daily, placebo ARTI: doc o diagnosed ARTI p ecipi a ing p ima y ca e consul a ion Seconda y 236/260 (90.8) Ma ineau 2015a16 (ViDiCO) UK (1 yea ) Adul s wi h COPD (144:96) 64.7 (8.5) (40.0- 85.0) LC-MS/MS, DEQAS 46.1 (25.7) (0.0-160.0) 50/240 (20.8) 122:118 3 mg bolus 2-mon hly, placebo URTI: assessed om daily symp om dia y Cop ima y 240/240 (100.0) Ma ineau 2015b49 (ViDiAs) UK (1 yea ) Adul s wi h as hma (109:141) 47.9 (14.4) (16.0-78.0) LC-MS/MS, DEQAS 49.6 (24.7) (0.0-139.0) 36/250 (14.4) 125:125 3 mg bolus 2-mon hly, placebo URTI: assessed om daily symp om dia y Cop ima y 250/250 (100.0) Ma ineau 2015c50 (ViDiFlu) UK (1 yea ) Olde adul s and hei ca e s (82:158) 67.1 (13.0) (21.4-94.0) LC-MS/MS, DEQAS 42.9 (23.0) (0.0-128.0) 60/240 (25.0) 137:103 Olde adul s: 2.4 mg bolus 2-mon hly+10 µg daily. Ca e s: 3 mg 2-mon hly, olde adul s: placebo+10 µg daily. Ca e s: placebo URTI and LRTI, bo h assessed om daily symp om dia y Cop ima y 240/240 (100.0) Simpson 201551 Aus alia (17weeks) Heal hy adul s (14:20) 32.2 (12.2) (18.0-52.0) LC-MS/MS, DEQAS 67.9 (23.0) (32.0-132.0) 0/33 (0.0) 18:16 0.5 mg weekly, placebo ARTI assessed wi h symp om sco e P ima y 34/34 (100.0) Dubno -Raz 201536 Is ael (12weeks) Adolescen swimme s wi h i amin D insu iciency (34:20) 15.2 (1.6) (12.9-18.6) RIA (DiaSo in), DEQAS 60.4 (11.9) (28.0-74.6) 0/54 (0.0) 27:27 50 µg daily, placebo URTI assessed wi h symp om sco e P ima y 25/54 (46.3) Denlinge 201652 USA (28weeks) Adul s wi h as hma (130:278) 39.2 (12.9) (18.0-85.0) CLA (DiaSo in), VDSP 47.0 (16.9) (10.0-74.6) 55/408 (13.5) 201:207 2.5 mg bolus hen 100 µg daily, placebo URTI assessed wi h symp om sco e Seconda y 408/408 (100.0) Tachimo o 201624 Japan (6mon hs) Child en wi h as hma (50:39) 9.9 (2.3) (6.0-15.0) RIA (DiaSo in), CAP 74.9 (24.6) (20.0-187.2) 1/89 (1.1) 54:35 20 µg daily, i s 2 mon hs, placebo URTI: assessed wi h symp om sco e Seconda y 89/89 (100.0) Ginde, 201653 USA (1 yea ) Olde ca e home esiden s (45:62) 80.7 (9.9) (60.0- 95.0) LC-MS/MS, VDSP 57.3 (22.7) (11.7-106.1) 12/107 (11.2) 55:52 2.5 mg bolus mon hly+≤25 µg pe day equi alen , placebo+10-25 µg pe day equi alen ARTI: medical eco d diagnosis P ima y 107/107 (100.0) 25(OH)D=25-hyd oxy i amin D; RIDT= apid in luenza diagnos ic es ; COPD=ch onic obs uc i e pulmona y disease; D3, i amin D3 (cholecalci e ol); ARTI=acu e espi a o y ac in ec ion; CAP=College o Ame ican Pa hologis s; CLA=chemiluminescen assay; DEQAS=Vi amin D Ex e nal Quali y Assessmen Scheme; EIA=enzyme immunoassay; EQA=ex e nal quali y assessmen ; LC-MS/MS=liquid ch oma og aphy andem-mass spec ome y; RIA= adioimmunoassay; URTI=uppe espi a o y ac in ec ion; LRTI=lowe espi a o y ac in ec ion; ILI=in luenza-like illness; RIQAS=Randox In e na ional Quali y Assessmen Scheme; VDSP=Vi amin D S anda disa ion P og am o he O ice o Die a y Supplemen s, Na ional Ins i u es o Heal h, USA. 1 µg i amin D3=40 in e na ional uni s (IU); 25(OH)D concen a ions epo ed in ng/mL we e con e ed o nmol/L (mul iplying by 2.496) *Sex missing o wo pa icipan s andomised o in e en ion a m and subsequen ly excluded om analysis owing o lack o ou come da a. he bmj | BMJ 2017;356:i6583 | doi: 10.1136/bmj.i6583 RESEARCH 7 signi ican p o ec i e e ec s o i amin D we e also seen o one s ep analyses o acu e espi a o y ac in ec ion a e (adjus ed incidence a e a io 0.96, 95% con idence in e al 0.92 o 0.997, P=0.04; P o he e ogenei y <0.001; 10 703 pa icipan s in 25 s udies) bu no o analysis o ime o i s acu e espi a o y ac in ec ion (adjus ed haza d a io 0.95, 95% con idence in e al 0.89 o 1.01, P=0.09; P o he e ogenei y <0.001; 9108 pa icipan s in 18 s udies). Two s ep analyses also showed consis en e ec s o he p opo ion o pa ici- pan s expe iencing a leas one acu e espi a o y ac in ec ion (adjus ed odds a io 0.80, 0.69 o 0.93, P=0.004; P o he e ogenei y 0.001; 10 899 pa icipan s in 24 s udies; ig 2), acu e espi a o y ac in ec ion a e (adjus ed incidence a e a io 0.91, 0.84 o 0.98, P=0.018; P o he e ogenei y <0.001; 10 703 pa icipan s in 25 s udies), and ime o i s acu e espi a o y ac in ec ion (adjus ed haza d a io 0.92, 0.85 o 1.00, P=0.051; P o he e ogenei y 0.14; 9108 pa icipan s in 18 s udies). This e idence was assessed as being o high quali y (see supplemen a y able S3). Subg oup analyses To explo e easons o he e ogenei y, we conduc ed subg oup analyses o in es iga e whe he e ec s o i amin D supplemen a ion on isk o acu e espi a o y ac in ec ion di e ed acco ding o baseline i amin D s a us, dosing equency, dose size, age, body mass index, he p esence o absence o como bidi y (as hma o ch onic obs uc i e pulmona y disease), and in lu- enza accina ion s a us. Race o e hnici y was no in es iga ed as a po en ial e ec modi ie , as da a o his a iable we e missing o 3680/10 933 (34%) pa ic- ipan s and powe o subg oup analyses was limi ed by small numbe s in many acial o e hnic subg oups ha could no be meaning ully combined. Table 2 p esen s he esul s. Subg oup analysis e ealed a s ong p o ec- i e e ec o i amin D supplemen a ion among hose wi h baseline ci cula ing 25-hyd oxy i amin D le els less han 25 nmol/L (adjus ed odds a io 0.58, 0.40 o0.82, NNT=8, 5 o 21; 538 pa icipan s in 14 s ud- ies; wi hin subg oup P=0.002; see Ca es plo , supplemen a y igu e S1) and no s a is ically signi i- can e ec among hose wi h baseline le els o 25 o mo e nmol/L (adjus ed odds a io 0.89, 0.77 o 1.04; 3634 pa icipan s in 19 s udies; wi hin subg oup P=0.15; P o in e ac ion 0.01). This e idence was assessed as being o high quali y (see supplemen a y able S3). An explo a o y analysis es ing he e ec s o i amin D supplemen a ion in hose wi h baseline 25-hyd oxy i amin D concen a ions in he anges 25-49.9 nmol/L, 50-74.9 nmol/L, and 75 o mo e nmol/L able 2 | One s ep indi idual pa icipan da a me a-analysis, p opo ion o pa icipan s expe iencing a leas one acu e espi a o y ac in ec ion (a i): o e all and by subg oup a iables no o ials* P opo ion wi h ≥1 a i, con ol g oup (%) P opo ion wi h ≥1 a i, in e en ion g oup (%) adjus ed odds a io (95% ci)† P alue P alue o in e ac ion O e all 25 2204/5225 (42.2) 2303/5708 (40.3) 0.88 (0.81 o 0.96) 0.003 -- Baseline 25(OH)D (nmol/L): <25 14 137/249 (55.0) 117/289 (40.5) 0.58 (0.40 o 0.82) 0.002 0.01 ≥25 19 1027/1639 (62.7) 1179/1995 (59.1) 0.89 (0.77 o 1.04) 0.15 Dosing egimen ype: Bolus dose ≥30 000 IU gi en 10 994/2786 (35.7) 1097/3014 (36.4) 0.97 (0.86 o 1.10) 0.67 0.05 Bolus dose no gi en 15 1210/2439 (49.6) 1206/2694 (44.8) 0.81 (0.72 o 0.91) <0.001 Daily dose equi alen (µg): <20 5629/1321 (47.6) 619/1435 (43.1) 0.80 (0.68 o 0.94) 0.006 0.12 20-50 9945/2796 (33.8) 1023/3077 (33.2) 0.90 (0.79 o 1.01) 0.08 ≥50 11 630/1108 (56.9) 661/1196 (55.3) 0.98 (0.81 o 1.18) 0.84 Age (yea s): ≤1 4 832/2744 (30.3) 854/2827 (30.2) 0.94 (0.83 o 1.06) 0.33 0.61 1.1-15.9 8241/513 (47.0) 194/566 (34.3) 0.60 (0.46 o 0.77) <0.001 16-65 17 854/1459 (58.5) 885/1592 (55.6) 0.93 (0.79 o 1.10) 0.41 >65 11 277/509 (54.4) 370/723 (51.2) 0.86 (0.67 o 1.09) 0.21 Body mass index (kg/m2): <25 19 972/1943 (50.0) 956/2074 (46.1) 0.85 (0.74 o 0.97) 0.02 0.29 ≥25 17 659/1039 (63.4) 754/1235 (61.1) 0.95 (0.79 o 1.14) 0.58 As hma: No 11 518/1008 (51.4) 520/1101 (47.2) 0.82 (0.68 o 0.99) 0.04 0.48 Yes 11 296/534 (55.4) 285/542 (52.6) 0.95 (0.73 o 1.25) 0.73 COPD: No 7477/763 (62.5) 493/791 (62.3) 1.00 (0.80 o 1.26) 0.98 0.38 Yes 6122/230 (53.0) 120/238 (50.4) 0.84 (0.57 o 1.24) 0.38 In luenza accina ion: No 10 255/373 (68.4) 253/407 (62.2) 0.74 (0.52 o 1.03) 0.08 0.51 Yes 10 564/779 (72.4) 577/826 (69.9) 0.86 (0.68 o 1.09) 0.22 25(OH)D=25-hyd oxy i amin D; COPD=ch onic obs uc i e pulmona y disease; 1 µg i amin D3=40 in e na ional uni s (IU). *Some ials did no con ibu e da a o a gi en subg oup, ei he because indi iduals wi hin ha subg oup we e no ep esen ed o because da a ela ing o he po en ial e ec modi ie we e no eco ded; acco dingly he numbe o ials ep esen ed a ies be ween subg oups. †Adjus ed o age, sex, and s udy du a ion. doi: 10.1136/bmj.i6583 | BMJ 2017;356:i6583 | he bmj RESEARCH 8 did no e eal e idence o a s a is ically signi ican in e ac ion (see supplemen a y able S4). Me a-analysis o da a om ials in which i amin D was adminis e ed using a daily o weekly egimen wi h- ou addi ional bolus doses e ealed a p o ec i e e ec agains acu e espi a o y ac in ec ion (adjus ed odds a io 0.81, 0.72 o 0.91, NNT=20, 13 o 43; 5133 pa ici- pan s in 15 s udies; wi hin subg oup P<0.001; see Ca es plo , supplemen a y igu e S1). No such p o ec i e e ec was seen among pa icipan s in ials whe e a leas one bolus dose o i amin D was adminis e ed (adjus ed odds a io 0.97, 0.86 o 1.10; 5800 pa icipan s in 10 s udies; wi hin subg oup P=0.67; P o in e ac ion 0.05). This e idence was assessed as being o high quali y (see supplemen a y able S3). P alues o in e ac ion we e mo e han 0.05 o all o he po en ial e ec modi ie s in es iga ed. Fo bo h o hese subg oup analyses, b oadly consis en e ec s we e obse ed o e en a e analysis (see supplemen a y able S5) and su i al analysis (see supplemen a y able S6). Ha ing iden i ied wo po en ial ac o s ha modi ied he in luence o i amin D supplemen a ion on isk o acu e espi a o y ac in ec ion (ie, baseline i amin D s a us and dosing equency), we hen p oceeded o in es iga e whe he hese ac o s we e ac ing as independen e ec modi ie s, o whe he hey we e con ounded by each o he o by ano he po en ial e ec modi ie , such as age. Do plo s e ealed a end owa ds lowe median baseline se um 25-hyd oxy i amin D con- cen a ion and highe median age o s udies employing bolus compa ed wi h daily o weekly dosing (see supple- men a y igu es S2 and S3). To es ablish which o hese po en ial e ec modi ie s was ac ing independen ly, we epea ed he analysis o include ea men -co a ia e in e ac ion e ms o baseline i amin D s a us, dosing equency, and age. In his model, in e ac ion e ms o baseline i amin D s a us and dosing equency we e s a is ically signi ican (P=0.01 and P=0.004, espec- i ely), bu he in e ac ion e m o age was no (P=0.20), consis en wi h he hypo hesis ha baseline i amin D s a us and dosing equency, bu no age, independen ly modi ied he e ec o i amin D supplemen a ion on isk o acu e espi a o y ac in ec ion. We hen p oceeded o s a i y he subg oup analysis p esen ed in able 2 acco ding o dosing equency, o p o ide a “cleane ” look a he esul s o subg oup analyses unde he assump ion ha use o bolus doses was ine ec i e. Table 3 p esen s he esul s: hese e eal ha daily o weekly i amin D ea men was associa ed wi h an e en g ea e deg ee o p o ec ion agains acu e espi a o y ac in ec ion among pa ic- ipan s wi h baseline ci cula ing 25-hyd oxy i amin D Li-Ng 2009 U ashima 2010 Manaseki-Holland 2010 Laaksi 2010 Majak 2011 T ilok-Kuma 2011 Lehouck 2012 Manaseki-Holland 2012 Cama go 2012 Mu doch 2012 Be gman 2012 Ma chisio 2013 Rees 2013 T an 2014 Goodall 2014 U ashima 2014 G an 2014 Ma ineau 2015 (ViDiCO) Ma ineau 2015 (ViDiAs) Ma ineau 2015 (ViDiFlu) Dubno -Raz 2015 Denlinge 2016 Tachimo o 2016 Ginde 2016 Simpson 2015 O e all: I 2=53.3%, P=0.001 No e: Weigh s a e om andom e ec s analysis 0.85 (0.44 o 1.64) 0.90 (0.58 o 1.41) 0.60 (0.41 o 0.88) 0.51 (0.27 o 0.96) 0.20 (0.05 o 0.82) 0.92 (0.77 o 1.11) 1.00 (0.53 o 1.90) 1.08 (0.89 o 1.30) 0.38 (0.22 o 0.65) 0.97 (0.30 o 3.15) 0.42 (0.20 o 0.89) 0.44 (0.21 o 0.95) 1.03 (0.72 o 1.49) 0.92 (0.65 o 1.30) 0.66 (0.45 o 0.98) 1.43 (0.73 o 2.78) 0.77 (0.43 o 1.36) 0.87 (0.48 o 1.57) 0.71 (0.38 o 1.31) 1.13 (0.66 o 1.95) 0.23 (0.01 o 3.82) 1.52 (1.02 o 2.28) 0.45 (0.11 o 1.89) 0.44 (0.19 o 1.02) Excluded 0.80 (0.69 o 0.93) 3.48 5.36 6.12 3.58 1.00 8.69 3.57 8.58 4.36 1.43 2.89 2.84 6.35 6.60 5.94 3.41 4.12 3.98 3.74 4.38 0.28 5.86 1.01 2.44 0.00 100.00 0.125 0.25 0.5 124 S udy Adjus ed odds a io (95% CI) Adjus ed odds a io (95% CI) Weigh (%) 33/76 (43.4) 69/167 (41.3) 126/229 (55.0) 54/84 (64.3) 11/24 (45.8) 458/1030 (44.5) 29/89 (32.6) 245/1505 (16.3) 53/103 (51.5) 155/161 (96.3) 39/62 (62.9) 38/58 (65.5) 276/360 (76.7) 96/197 (48.7) 80/234 (34.2) 17/99 (17.2) 53/80 (66.3) 75/118 (63.6) 93/125 (74.4) 58/103 (56.3) 10/11 (90.9) 93/207 (44.9) 5/35 (14.3) 24/52 (46.2) 14/16 (87.5) Con ol 32/81 (39.5) 68/167 (40.7) 97/224 (43.3) 39/80 (48.8) 4/24 (16.7) 438/1034 (42.4) 30/86 (34.9) 260/1506 (17.3) 44/141 (31.2) 154/161 (95.7) 26/62 (41.9) 26/58 (44.8) 303/399 (75.9) 185/397 (46.6) 70/258 (27.1) 32/148 (21.6) 94/156 (60.3) 76/122 (62.3) 85/125 (68.0) 83/137 (60.6) 10/14 (71.4) 110/201 (54.7) 4/54 (7.4) 17/55 (30.9) 16/18 (88.9) In e en ion P opo ion wi h ≥1 ARTI (%) ig 2 | wo s ep indi idual pa icipan da a me a-analysis: p opo ion o pa icipan s expe iencing a leas one acu e espi a o y ac in ec ion (a i). Da a om ial by simpson e al we e no included in his wo s ep me a-analysis, as an es ima e o he e ec o he in e en ion in he s udy could no be ob ained in he eg ession model owing o small sample size he bmj | BMJ 2017;356:i6583 | doi: 10.1136/bmj.i6583 RESEARCH 9 able 3 | One s ep indi idual pa icipan da a me a-analysis, p opo ion o pa icipan s expe iencing a leas one acu e espi a o y ac in ec ion (a i): o e all and by subg oup, s a i ied by dosing equency a iables bolus dosing Daily o weekly dosing no o ials* P opo ion wi h ≥1 a i, con ol g oup (%) P opo ion wi h ≥1 a i, in e en ion g oup (%) adjus ed odds a io (95% ci)† P alue P alue o in e ac ion no o ials* P opo ion wi h ≥1 a i, con ol g oup (%) P opo ion wi h ≥1 a i, in e en ion g oup (%) adjus ed odds a io (95% ci)† P alue P alue o in e ac ion O e all 10 994/2786 (35.7) 1097/3014 (36.4) 0.97 (0.86 o 1.10) 0.67 -- 15 1210/2439 (49.6) 1206/2694 (44.8) 0.81 (0.72 o 0.91) 0.001 -- Baseline 25(OH)D (nmol/L): <25 873/142 (51.4) 77/162 (47.5) 0.82 (0.51 o 1.33) 0.43 0.42 664/107 (59.8) 40/127 (31.5) 0.30 (0.17 o 0.53) <0.001 0.006 ≥25 8550/910 (60.4) 663/1121 (59.1) 1.02 (0.83 o 1.24) 0.87 11 477/729 (65.4) 516/874 (59.0) 0.75 (0.60 o 0.95) 0.02 Daily dose equi alen (µg): <20 . . . . . 0.56 5629/1321 (47.6) 619/1435 (43.1) 0.80 (0.68 o 0.94) 0.006 0.82 20-50 3467/1931 (24.2) 542/2127 (25.5) 0.95 (0.81 o 1.10) 0.50 6478/865 (55.3) 481/950 (50.6) 0.81 (0.66 o 1.01) 0.06 ≥50 7527/855 (61.6) 555/887 (62.6) 1.03 (0.83 o 1.28) 0.81 4103/253 (40.7) 106/309 (34.3) 0.85 (0.58 o 1.24) 0.39 Age (yea s): ≤1 2 321/1634 (19.6) 322/1637 (19.7) 0.99 (0.83 o 1.19) 0.93 0.72 2511/1110 (46.0) 532/1190 (44.7) 0.91 (0.77 o 1.08) 0.30 0.37 1.1-15.9 150/100 (50.0) 35/93 (37.6) 0.62 (0.35 o 1.11) 0.11 7191/413 (46.2) 159/473 (33.6) 0.59 (0.45 o 0.79) <0.001 16-65 8432/678 (63.7) 466/716 (65.1) 1.15 (0.90 o 1.48) 0.27 9422/781 (54.0) 419/876 (47.8) 0.79 (0.63 o 0.99) 0.04 >65 8191/374 (51.1) 274/568 (48.2) 0.85 (0.65 o 1.12) 0.25 386/135 (63.7) 96/155 (61.9) 0.88 (0.52 o 1.52) 0.66 Body mass index (kg/m2): <25 8215/372 (57.8) 231/417 (55.4) 1.01 (0.72 o 1.40) 0.97 0.70 11 757/1571 (48.2) 725/1657 (43.8) 0.82 (0.71 o 0.95) 0.009 >0.99 ≥25 8406/677 (60.0) 509/867 (58.7) 1.00 (0.80 o 1.25) 0.98 9253/358 (70.7) 245/367 (66.8) 0.83 (0.59 o 1.17) 0.30 As hma: No 5303/484 (62.6) 323/523 (61.8) 0.95 (0.71 o 1.28) 0.75 0.40 6215/524 (41.0) 197/578 (34.1) 0.74 (0.58 o 0.95) 0.02 0.40 Yes 4224/371 (60.4) 232/364 (63.7) 1.18 (0.85 o 1.65) 0.32 772/163 (44.2) 53/178 (29.8) 0.60 (0.37 o 0.98) 0.04 COPD: No 5410/632 (64.9) 436/656 (66.5) --‡ --‡ --‡ 267/131 (51.1) 57/135 (42.2) --‡ --‡ --‡ Yes 4117/223 (52.5) 119/231 (51.5) --‡ --‡ --‡ 25/7 (71.4) 1/7 (14.3) --‡ --‡ --‡ In luenza accina ion No 5119/163 (73.0) 121/178 (68.0) --‡ --‡ --‡ 5136/210 (64.8) 132/229 (57.6) --‡ --‡ --‡ Yes 5286/396 (72.2) 294/421 (69.8) . . . 5 278/383 (72.6) 283/405 (69.9) 25(OH)D=25-hyd oxy i amin D; COPD=ch onic obs uc i e pulmona y disease; 1 µg i amin D3=40 in e na ional uni s (IU). *Some ials did no con ibu e da a o a gi en subg oup, ei he because indi iduals wi hin ha subg oup we e no ep esen ed o because da a ela ing o he po en ial e ec modi ie we e no eco ded; acco dingly he numbe o ials ep esen ed a ies be ween subg oups. †Adjus ed o age, sex, and s udy du a ion. ‡Values could no be es ima ed as models did no con e ge.