Risk of pneumonia associated with use of angiotensin converting enzyme inhibitors and angiotensin receptor blockers : systematic review and meta-analysis

Risk o pneumonia associa ed wi h use o angio ensin con e ing enzyme inhibi o s and angio ensin ecep o blocke s: sys ema ic e iew and me a-analysis OPEN ACCESS Daniel Caldei a ca diologis esiden , assis an o clinical pha macology 1, Joana Ala cão scien i ic consul an , assis an o clinical pha macology2, An ónio Vaz-Ca nei o clinical p o esso o medicine, di ec o o he Cen e o E idence-Based Medicine2 3, João Cos a p o esso o clinical pha macology, coo dina o o he Po uguese Coch ane Cen e 123 1Labo a o y o Clinical Pha macology and The apeu ics, Facul y o Medicine, Uni e si y o Lisbon; 2Cen e o E idence-Based Medicine, Facul y o Medicine, Uni e si y o Lisbon, A enue P o esso Egas Moniz, 1649-028, Lisbon, Po ugal; 3Coch ane Coo dina ing Cen e Po ugal, Facul y o Medicine, Uni e si y o Lisbon Abs ac Objec i e To sys ema ically e iew longi udinal s udies e alua ing use o angio ensin con e ing enzyme (ACE) inhibi o s o angio ensin ecep o blocke s (ARBs) and isk o pneumonia. Design Sys ema ic e iew and me a-analysis. Da a sou ces Medline h ough PubMed, Web o Science wi h con e ence p oceedings (incep ion o June 2011), and US Food and D ug Adminis a ion websi e (June 2011). Sys ema ic e iews and e e ences o e ie ed a icles we e also sea ched. S udy selec ion Two e iewe s independen ly selec ed andomised con olled ials and coho and case-con ol s udies e alua ing he use o ACE inhibi o s o ARBs and isk o pneumonia and e ie ed cha ac e is ics o he s udies and da a es ima es. Da a syn hesis The p ima y ou come was incidence o pneumonia and he seconda y ou come was pneumonia ela ed mo ali y. Subg oup analyses we e ca ied acco ding o baseline mo bidi ies (s oke, hea ailu e, and ch onic kidney disease) and pa ien s’ cha ac e is ics (Asian and non-Asian). Pooled es ima es o odds a ios and 95% con idence in e als we e de i ed by andom e ec s me a-analysis. Adjus ed equen is indi ec compa isons be ween ACE inhibi o s and ARBs we e es ima ed and combined wi h di ec e idence whene e a ailable. He e ogenei y was assessed using he I2 es . Resul s 37 eligible s udies we e included. ACE inhibi o s we e associa ed wi h a signi ican ly educed isk o pneumonia compa ed wi h con ol ea men (19 s udies: odds a io 0.66, 95% con idence in e al 0.55 o 0.80; I2=79%) and ARBs (combined di ec and indi ec odds a io es ima e 0.69, 0.56 o 0.85). In pa ien s wi h s oke, he isk o pneumonia was also lowe in hose ea ed wi h ACE inhibi o s compa ed wi h con ol ea men (odds a io 0.46, 0.34 o 0.62) and ARBs (0.42, 0.22 o 0.80). ACE inhibi o s we e associa ed wi h a signi ican ly educed isk o pneumonia among Asian pa ien s (0.43, 0.34 o 0.54) compa ed wi h non-Asian pa ien s (0.82, 0.67 o 1.00; P<0.001). Compa ed wi h con ol ea men s, bo h ACE inhibi o s (se en s udies: odds a io 0.73, 0.58 o 0.92; I2=51%) and ARBs (one andomised con olled ial: 0.63, 0.40 o 1.00) we e associa ed wi h a dec ease in pneumonia ela ed mo ali y, wi hou di e ences be ween in e en ions. Conclusions The bes e idence a ailable poin s owa ds a pu a i e p o ec i e ole o ACE inhibi o s bu no ARBs in isk o pneumonia. Pa ien popula ions ha may bene i mos a e hose wi h p e ious s oke and Asian pa ien s. ACE inhibi o s we e also associa ed wi h a dec ease in pneumonia ela ed mo ali y, bu he da a lacked s eng h. In oduc ion Pneumonia ep esen s an impo an clinical condi ion because o i s ela i ely high incidence (0.5% o 1.1% annually in he Uni ed Kingdom) and associa ed mo bidi y and mo ali y.1 2 Suscep ibili y is highe among elde ly people (≥65 yea s), hose wi h alcohol dependency, smoke s, and pa ien s wi h hea ailu e, p e ious s oke, diabe es, ch onic kidney disease, and ch onic lung disease.3-6 Pneumonia is a common eason o hospi al admission and a isk ac o o p olonged hospi al s ay, ca ying a conside able inancial bu den on heal hca e esou ces.7 8 Co espondence o: J Cos a [email p o ec ed] Ex a ma e ial supplied by he au ho (see h p://www.bmj.com/con en /345/bmj.e4260? ab= ela ed#webex a) Sea ch s a egy Supplemen a y igu es 1-13 No comme cial euse: See igh s and ep in s h p://www.bmj.com/pe missions Subsc ibe: h p://www.bmj.com/subsc ibe BMJ 2012;345:e4260 doi: 10.1136/bmj.e4260 (Published 11 July 2012) Page 1 o 20 Resea ch RESEARCH Usage o some d ugs has been shown o modula e he isk o pneumonia. Acid supp essan s can inc ease pa ien s’ suscep ibili y o pneumonia, whe eas s a ins may ha e a p o ec i e ole.9 10 Angio ensin con e ing enzyme (ACE) inhibi o s and angio ensin ecep o blocke s (ARBs) a e o en used in pa ien s wi h ca dio ascula disease. ACE inhibi o s a e known o ha e ad e se e ec s on he espi a o y sys em, in pa icula an inc eased incidence o cough. Basic in es iga ion has shown ha b adykinin and subs ance P sensi ise he senso y ne es o he ai ways and enhance he cough e lex,11-13 which may ha e a p o ec i e ole on he acheob onchial ee.14 15 These mechanisms also imp o e swallowing by a oiding he exposu e o he espi a o y ee o o opha ynx sec e ions.11 14 16 Taken oge he , he pleio opic e ec s o ACE inhibi o s we e sugges ed o educe he incidence o pneumonia, bu a ailable clinical e idence lacks s eng h17-19 and published esul s ha e been con adic o y.20-22 We sys ema ically e iewed and me a-analysed all s udies (expe imen al and obse a ional) e alua ing he use o ACE inhibi o s and incidence o pneumonia. Because he clinical cha ac e is ics and isk ac o s o popula ions using ARBs a e simila o hose o pa ien s using ACE inhibi o s, and he e o e s udies e alua ing hese in e en ions sha e iden ical po en ial clinical con ounde s, we also es ima ed he incidence o pneumonia in s udies e alua ing ARBs. Mo eo e , pa ien s ea ed wi h ARBs a e less likely o expe ience espi a o y ad e se e en s,23 24 and he e o e ARBs may ha e a p o ec i e ole. Me hods The sys ema ic e iew was ca ied ou in acco dance wi h he me a-analysis o obse a ional s udies in epidemiology and p e e ed epo ing i ems o sys ema ic e iews and me a-analyses s a emen s.25 26 Ou p ima y ou come was he incidence o pneumonia. We conside ed cases o pneumonia, lowe espi a o y ac in ec ions, and admissions o hospi al due o lowe espi a o y ac in ec ions. Da a we e ex ac ed i espec i e o whe he hey had been epo ed as p ede ined ou comes o as ad e se e ec s. I s udies epo ed da a o dea h om pneumonia only, o a oid duplica ion we did no conside hese cases o he p ima y ou come. The seconda y ou come was pneumonia ela ed mo ali y, de ined as dea h di ec ly ela ed o his condi ion o in-hospi al dea h o mo ali y wi hin 30 days a e onse o pneumonia.27 Fo bo h ou comes, we did no conside unde ined da a o da a on uppe espi a o y ac in ec ions. We conside ed andomised con olled pa allel ials, coho s udies, and case-con ol s udies wi h ACE inhibi o s o ARBs as in e en ions and wi h p ede ined ou comes. T ea men a ms could compa e ACE inhibi o s and ARBs wi h each o he o wi h placebo o any o he ac i e d ug. Coho s udies could be based on popula ions in he communi y o hose in ins i u ions o hospi al and had o ollow pa ien s o de e mine pneumonia ou comes. In case-con ol s udies, cases had o be de ined as pa ien s wi h new onse pneumonia iden i ied h ough clinical examina ion, adiological me hods, o da abase codes. Con ols had o be ma ched o cases, bu wi hou new onse pneumonia. Fo pneumonia ela ed mo ali y, we allowed case-con ol s udies wi h bo h cases and con ols ha ing pneumonia. We allowed all pa icipan s, i espec i e o baseline diseases and isk ac o s. In o ma ion sou ces and sea ch me hod We iden i ied po en ially eligible s udies h ough an elec onic sea ch o bibliog aphic da abases om incep ion o June 2011 (Medline h ough PubMed and Web o Science wi h con e ence p oceedings). See he supplemen a y ile o de ails o he sea ch s a egy. No language es ic ions we e applied. We sc eened and c oss checked iden i ied sys ema ic e iews and me a-analyses e alua ing ACE inhibi o s o ARBs, as well as e e ence lis s o pape s o po en ial addi ional s udies. We also sea ched he Food and D ug Adminis a ion websi e (10 June 2011) o egula o y documen s wi h unpublished da a om clinical ials. S udy selec ion and da a collec ion p ocess The i les and abs ac s o ob ained eco ds we e sc eened. Doub s and disag eemen s we e esol ed by consensus. We assessed he selec ed s udies in ull ex o de e mine app op ia eness o inclusion. Two au ho s independen ly ex ac ed da a on s udy design, loca ion, pe iod o s udy, pa ien s’ cha ac e is ics, d ug use and how i was assessed, p ima y ou comes, da a o equi ed ou comes, and adjus men s o es ima es. When s udies p esen ed di e en es ima es on p ima y ou comes acco ding o he se e i y o pneumonia, we ex ac ed o analysis only hose epo ing he mos se e e cases. Fo he p ima y ou come we conside ed d ug wi hd awals due o pneumonia only when no o he es ima es we e a ailable. When mo e han one isk es ima e was a ailable om se e al sou ces, we used only he mos p ecise o adjus ed measu es o associa ion om each epo . O he wise we used he c ude odds a io o de i ed i om he aw da a. Two au ho s (DC and JA) independen ly analysed he quali y o epo ing by using a quali a i e classi ica ion acco ding o isk o bias (high, unclea , o low). Fo obse a ional s udies we used a six i em classi ica ion based on he me a-analysis o obse a ional s udies in epidemiology,25 he quali y assessmen ool o sys ema ic e iews o obse a ional s udies,28 and s eng hening he epo ing o obse a ional s udies in epidemiology.29 This sys em was adap ed om a p e iously published sys ema ic e iew30 31 and ook in o conside a ion he pa icipan s (i any jus i ica ion was gi en o he coho and he s udy epo ed app op ia e inclusion and exclusion c i e ia), in e en ion (i d ug use was adequa ely assessed and no based on sel epo ), ou come (i pneumonia was assessed by clinical examina ion, adiological me hods, o da abase codes and no based on sel epo ), and ou come adjus men s ( o bo h age and a leas one o he ollowing: smoke o pulmona y disease, ca dio ascula diseases o d ug use o ch onic kidney disease; o he adjus men s). Fo andomised con olled ials we adap ed he Coch ane Collabo a ion’s ool o assessing isk o bias o e alua e he quali y o epo ing: andomisa ion me hod, alloca ion concealmen , blinding o pa icipan s and s a , blinding o ou come assessmen , selec i e epo ing (i pneumonia was a p especi ied ou come), and desc ip ion o wi hd awals.32 F om hese ools we de i ed isk o bias g aphs. S a is ical analysis We used Re Man 5.1.4 so wa e o s a is ical analysis (No dic Coch ane Cen e, Coch ane Collabo a ion, 2011) and o de i e o es plo s showing he esul s o indi idual s udies and pooled analysis. We ca ied ou h ee analyses. Fi s ly, we compa ed ACE inhibi o s and ARBs wi h each con ol g oup using andom e ec s me a-analysis weigh ed by he in e se a iance me hod No comme cial euse: See igh s and ep in s h p://www.bmj.com/pe missions Subsc ibe: h p://www.bmj.com/subsc ibe BMJ 2012;345:e4260 doi: 10.1136/bmj.e4260 (Published 11 July 2012) Page 2 o 20 RESEARCH o es ima e pooled odds a ios and 95% con idence in e als. He e ogenei y was assessed wi h he I2 es , which measu es he pe cen age o o al a ia ion be ween s udies due o he e ogenei y.33 We used he andom e ec s model independen ly o he exis ence (I2≥50%) o subs an ial he e ogenei y be ween he esul s o ials, as we pooled he esul s o s udies wi h di e en designs and pa ien s’ cha ac e is ics. We chose he odds a io as he measu emen es ima e o e ec because ela i e es ima es a e mo e simila han absolu e e ec s ac oss s udies wi h di e en designs, popula ions, and leng hs o ollow-up.34 Raw da a we e i s con e ed o odds a ios h ough classic me hods, o h ough Pe o’s me hod i one a m had a ze o coun cell. When aw da a o odds a ios we e no a ailable we ook he haza d a io o isk a io o analysis. To explo e di e ences in es ima es o ou comes we p esen ed he esul s s a i ied acco ding o s udy design. Fo he pu pose o he analysis we ea ed nes ed case-con ol s udies as coho s udies. We ca ied ou subg oup analyses o pa ien s wi h p e ious s oke, hea ailu e, and ch onic kidney disease because such pa ien s a e known o be pa icula ly suscep ible o pneumonia and all indica ions a e clinically app o ed o ea men wi h ACE inhibi o s and ARBs.5 6 35 36 In iew o he sugges ion ha ACE inhibi o s may be mo e e icien in educing he isk o pneumonia in Asian pa ien s we also calcula ed es ima es o Asian and non-Asian popula ions.37 We e alua ed di e ences be ween subg oups wi h he me hod desc ibed by Deeks e al, based on he in e se a iance me hod.33 In he second analysis we ca ied ou adjus ed indi ec compa isons be ween he pooled es ima e o ACE inhibi o s ( e sus con ol) and ARBs ( e sus con ol) using he Buche equen is me hod, which compa es di e en ea men s adjus ed o he esul s o hei di ec compa ison wi h a common con ol.38 This me hod pa ially o e comes he p oblem o di e en p ognos ic cha ac e is ics be ween pa icipan s among s udies, and i is belie ed o be alid assuming ha he ela i e e ec o in e en ions is consis en ac oss di e en s udies, as e i ied in ou case.39 By de aul we used he andom e ec s model because adjus ed indi ec compa isons ha used he ixed e ec s model end o unde es ima e he s anda d e o s o pooled es ima es.39 Thi dly, we combined e idence gene a ed by indi ec compa isons wi h e idence om head o head s udies compa ing ACE inhibi o s wi h ARBs using he andom e ec s model o quan i a i e pooling,40 41 and we de e mined he disc epancy and he e ogenei y be ween di ec and indi ec es ima es.42 We also calcula ed he numbe needed o ea (NNT) and 95% con idence in e als, aking in o accoun he baseline isk (weigh ed p opo ion o e en a e in con ol g oup) because o he di e ences in he p edic ed absolu e bene i o ea men acco ding o a ia ion in baseline isk be ween g oups.34 43 In ou case, he weigh ed isk o pneumonia in he con ol g oups was 4.6% (95% con idence in e al 3.1% o 6.7%). Publica ion bias was assessed h ough isual inspec ion o he asymme y in unnel plo s. Resul s The sea ch o he elec onic da abases yielded 807 published s udies. A e applying he inclusion and exclusion c i e ia 29 s udies we e included o analysis ( ig 1⇓). The esul s om eigh addi ional s udies we e iden i ied in he FDA egula o y documen s. O e all, da a we e ob ained om 37 s udies.20-22 44-79 Desc ip ion o s udies The 37 s udies included 18 andomised con olled ials,20 44-61 11 coho s udies,62-73 wo nes ed case-con ol s udies,21 74 and six case-con ol s udies.22 75-79 Among andomised con olled ials, eigh we e done wo ldwide,47 48 51 52 55 57 58 61 six in Eu ope,44-46 53 56 60 h ee in Asia,49 50 59, and one in Eu ope and he Uni ed S a es.54 Mos o he andomised con olled ials we e mul icen e (n=16). Se en andomised con olled ials compa ed ACE inhibi o s wi h con ols,44-50 nine compa ed ARBs wi h con ols,51-59 and wo compa ed ACE inhibi o s wi h ARBs.60 61 Se en ials epo ed speci ic da a o se ious pneumonia, i e o a al pneumonia, and eigh epo ed pneumonia wi hou speci ying he se e i y o disease. In only wo ials was pneumonia a p especi ied ou come.49 59 Among obse a ional s udies, 10 we e ca ied ou in Asia, i e in he Uni ed S a es, and ou in Eu ope. Ele en s udies we e e ospec i e and eigh we e p ospec i e. Se en een e alua ed ACE inhibi o s, wo ARBs, and wo compa ed ACE inhibi o s wi h ARBs. Tables 1 o 3⇓⇓⇓ summa ise he main cha ac e is ics o he included s udies. The o e all quali y o he s udies was conside ed o be good. All he andomised con olled ials, excep one,51 me he c i e ia o andom sequence gene a ion and abou hal speci ically epo ed adequa e alloca ion concealmen .46 48 50 54 55 57 58 61 Only wo andomised con olled ials56 59 we e conside ed o be a high isk o pe o mance bias. Adequa e blinding o ou come assessmen 45-48 50 52-59 61 and ull desc ip ion o s udy wi hd awals44-55 57 58 60 61 we e epo ed in 78% and 89% o andomised con olled ials, espec i ely. The highes isk o bias was ound o po en ial epo ing bias because only wo andomised con olled ials p esen ed esul s o pneumonia as a p especi ied ou come.49 59 Supplemen a y igu es 1 and 2 show he esul s o he quali y app aisal o he andomised con olled ials. All obse a ional s udies we e conside ed o ha e adequa e inclusion and exclusion c i e ia and p o ided jus i ica ion o he coho . Fi e s udies (26%)62-64 66 69 did no clea ly s a ed how he d ug use was assessed, and ou s udies (21%)63 64 66 73 did no p o ide de ails abou ou come assessmen . Ele en s udies (58%)21 22 68 70-72 74 75 77 79 p o ided esul s a e adjus men o a leas one po en ial a iable con ounde . In one s udy76 i was unclea o which a iables he esul s we e adjus ed. Few s udies (26%) epo ed esul s adjus ed o mul iple con ounde s, and in se en s udies (37%) no ype o adjus men was men ioned. Supplemen a y igu es 3 and 4 show he esul s o he quali y o he obse a ional s udies. P ima y ou come: incidence o pneumonia P ima y ou come da a we e a ailable om 19 s udies compa ing ACE inhibi o s wi h con ols ( i e andomised con olled ials, eigh coho o nes ed case-con ol s udies, and six case-con ol s udies), 11 s udies compa ing ARBs wi h con ols (nine andomised con olled ials and wo coho o nes ed case-con ol s udies), and wo s udies compa ing ACE inhibi o s wi h ARBs (one andomised con olled ial and one coho s udy). Use o ACE inhibi o s was associa ed wi h a signi ican 34% educ ion in isk o pneumonia compa ed wi h con ols (odds a io 0.66, 95% con idence in e al 0.55 o 0.80; I2=79%). The NNT o 2.0 yea s was 65 (48 o 112). The magni ude o he isk educ ion was simila ac oss all s udy designs (P=0.78 o No comme cial euse: See igh s and ep in s h p://www.bmj.com/pe missions Subsc ibe: h p://www.bmj.com/subsc ibe BMJ 2012;345:e4260 doi: 10.1136/bmj.e4260 (Published 11 July 2012) Page 3 o 20 RESEARCH subg oup di e ences). The odds a ios o andomised con olled ials, coho o nes ed case-con ol s udies, and case-con ol s udies we e 0.69 (0.56 o 0.85; I2=0%), 0.58 (0.38 o 0.88; I2=79%), and 0.67 (0.49 o 0.93; I2=73%), espec i ely ( ig 2⇓). The isk o pneumonia was no , howe e , di e en be ween pa ien s who did o did no use ARBs (0.95, 0.87 o 1.04; I2=14%). Odds a io es ima es o andomised con olled ials (0.90, 0.79 o 1.01; I2=7%) and coho o nes ed case-con ol s udies (1.01, 0.94 o 1.09; I2=0%) did no di e signi ican ly (P=0.10; ig 3⇓). Pooled esul s om he wo head o head s udies showed a non-signi ican 37% educ ion in isk o pneumonia associa ed wi h use o ACE inhibi o s (0.63, 0.28 o 1.44; I2=78%). In his case, es ima es om he andomised con olled ial and coho s udy di e ed signi ican ly (P=0.03) (see supplemen a y igu e 5). Indi ec compa ison o ACE inhibi o s wi h ARBs showed a signi ican 30% educ ion in isk o pneumonia associa ed wi h use o ACE inhibi o s (0.70, 0.56 o 0.86). Simila esul s we e ob ained om pooled di ec and indi ec es ima es (0.69, 0.56 o 0.85) wi hou disc epancy (P=0.82) o he e ogenei y (I2=0%) be ween bo h es ima es ( ig 4⇓). The NNT o 2.2 yea s based on his es ima e was 72 (51 o 147). Subg oup analyses o p ima y ou come Pa ien s wi h p e ious s oke In pa ien s wi h p e ious s oke, use o ACE inhibi o s was associa ed wi h a 54% educ ion in isk o pneumonia compa ed wi h con ols (0.46, 0.34 o 0.62, I2=0%; se en s udies pooled) (see supplemen a y igu e 6). In he same popula ion, howe e , use o ARBs was no associa ed wi h a signi ican educ ion in isk (0.86, 0.67 o 1.09; I2=0%; wo s udies pooled) (see supplemen a y igu e 7). The pooled es ima e om indi ec (odds a io 0.53, 95% con idence in e al 0.16 o 1.79) and di ec (0.38, 0.17 o 0.81) e idence o ACE inhibi o s compa ed wi h ARBs showed a signi ican 58% educ ion in isk o pneumonia (0.42, 0.22 o 0.80; ig 4), wi hou disc epancy (P=0.44) o he e ogenei y (I2=0%) be ween indi ec and di ec es ima es. Pa ien s wi h hea ailu e In pa ien s wi h hea ailu e, wo s udies e alua ed he isk o pneumonia in hose ea ed wi h ACE inhibi o s44 45 and wo o he s udies epo ed da a o hose ea ed wi h ARBs.51 52 ACE inhibi o s we e associa ed wi h a signi ican 37% educ ion in isk o pneumonia (0.63, 0.47 o 0.84; I2=0%), whe eas ARBs showed no signi ican e ec (0.85, 0.49 o 1.47; I2=15%) (see supplemen a y igu e 8). Pa ien s wi h ch onic kidney disease In pa ien s wi h ch onic kidney disease, he esul s om one andomised con olled ial o ACE inhibi o s46 (odds a io 0.15, 95% con idence in e al 0.00 o 7.70) and wo andomised con olled ials o ARBs52 53 (1.21, 0.32 o 4.52; I2=77%) did no di e signi ican ly when compa ed wi h con ols (see supplemen a y igu e 9). Asian and non-Asian pa ien s Ele en s udies we e ca ied ou in Asian coun ies and 11 we e done ou side o Asia. The PROGRESS20 s udy was he only mul icen e s udy ca ied ou wo ldwide ha supplied sepa a e da a o Asian and non-Asian pa ien s. To lowe analysis bias, he o he s udies ca ied ou wo ldwide ha did no p o ide sepa a e da a o bo h g oups we e excluded. The educ ion in isk o pneumonia associa ed wi h ACE inhibi o s was signi ican ly highe among Asian pa ien s (0.43, 0.34 o 0.54; I2=0%) compa ed wi h non-Asian pa ien s (0.82, 0.67 o 1.00, I2=80%, P<0.001 o subg oup di e ences) (see supplemen a y igu e 10). ARBs, howe e , we e no associa ed wi h a educ ion in isk o pneumonia in Asian pa ien s (1.04, 0.59 o 1.84; one andomised con olled ial HIJ-CREATE59) o non-Asian pa ien s (0.97, 0.84 o 1.12; I2=27%; i e s udies pooled; ig 4 and supplemen a y igu e 11). Seconda y ou come: pneumonia ela ed mo ali y Da a o seconda y ou comes we e ex ac ed om se en s udies compa ing ACE inhibi o s wi h con ols ( h ee andomised con olled ials and ou coho s udies),20 49 50 68 70-72 one andomised con olled ial compa ing ARBs wi h con ol,55 and one head o head andomised con olled ial.60 Fi e s udies compa ing ACE inhibi o s wi h con ols we e ca ied ou on an en iched popula ion— ha is, en olled pa ien s wi h pneumonia.49 68 70-72 T ea men wi h ACE inhibi o s was associa ed wi h a signi ican 27% educ ion in isk o pneumonia ela ed mo ali y compa ed wi h con ols (0.73, 0.58 o 0.92; I2=51%), wi hou signi ican di e ences be ween es ima es om andomised con olled ials and obse a ional s udies (P=0.76). The pooled esul om andomised con olled ials, howe e , ailed o each s a is ical signi icance (0.61, 0.20 o 1.90; I2=61%) ( ig 5⇓). Only one andomised con olled ial55 epo ed he e ec o ea men wi h ARBs on pneumonia ela ed mo ali y (odds a io 0.63, 95% con idence in e al 0.40 o 1.00) ( ig 5). The isk o pneumonia ela ed mo ali y in indi ec (1.16, 0.69 o 1.94), di ec (HEAVEN andomised con olled ial60 7.29, 0.14 o 367.24), and pooled compa isons (1.19, 0.71 o 1.98) did no di e be ween ACE inhibi o s and ARBs ( ig 4). The e was no disc epancy (P=0.36) o he e ogenei y (I2=0%) be ween indi ec and di ec es ima es. Publica ion bias Visual inspec ion o unnel plo s did no e eal any ob ious asymme ical ail (see supplemen a y igu e 12). Publica ion bias was no sugges ed by sensi i i y analysis aking in o accoun published and unpublished ials (see supplemen a y igu e 13). Discussion In his sys ema ic e iew we ound ha ea men wi h angio ensin con e ing enzyme (ACE) inhibi o s was associa ed wi h a signi ican educ ion in isk o pneumonia compa ed wi h con ol ea men and angio ensin ecep o blocke s (ARBs); he magni ude o his educ ion (abou one hi d) was simila ac oss s udies wi h di e en designs ( andomised con olled ials, coho , and case-con ol s udies). The isk o pneumonia was also educed in pa ien s ea ed wi h ACE inhibi o s who we e a highe isk o pneumonia, in pa icula hose wi h s oke and hea ailu e. Mos o he po en ial p o ec i e bene i om ACE inhibi o s seemed o be in Asian pa ien s; i is unclea whe he he me hodology o he s udies o he clinical and gene ic cha ac e is ics o he pa ien s we e esponsible o his inding. Use o ACE inhibi o s was also associa ed wi h a educ ion in pneumonia ela ed mo ali y, al hough he esul s we e less obus han o o e all isk o pneumonia; i is No comme cial euse: See igh s and ep in s h p://www.bmj.com/pe missions Subsc ibe: h p://www.bmj.com/subsc ibe BMJ 2012;345:e4260 doi: 10.1136/bmj.e4260 (Published 11 July 2012) Page 4 o 20 RESEARCH unce ain i di e ences exis be ween ACE inhibi o s and ARBs o his ou come. The p esen e iew was designed o de e mine he e ec o ea men wi h ACE inhibi o s and ARBs on isk o pneumonia. We combined da a om bo h expe imen al and obse a ional s udies o ob ain mo e obus esul s, mainly because no andomised con olled ial was p ima ily designed wi h his objec i e. Pneumonia is no a a e ou come (pa icula ly in popula ions ea ed wi h ACE inhibi o s o ARBs) o an ou come ha only occu s mon hs o yea s a e use o ACE inhibi o s o ARBs. The e o e andomised con olled ials would ha e been an app op ia e s udy design o deal wi h his p oblem. We ound signi ican s a is ical he e ogenei y o ACE inhibi o s bu no o ARB esul s. This was due o he esul s o obse a ional s udies (no he e ogenei y was ound among andomised con olled ials). Ne e heless, he obse ed s a is ical he e ogenei y was mo e quan i a i e han quali a i e because all es ima es o s udy designs sha e he same di ec ion. This consis ency, as well as he obus ness o educ ion in he isk o pneumonia ac oss all s udy designs, sugges s ha use o ACE inhibi o s dese es a en ion. Fu he mo e, ha ACE inhibi o s educed he isk o pneumonia compa ed no only wi h he con ol g oup bu also wi h ARB ea men , is eassu ing because pa ien s’ cha ac e is ics and isk ac o s, as well as o he po en ial clinical and me hodological con ounde s a e p obably simila be ween s udies on ARBs and hose on ACE inhibi o s. We we e also conse a i e in ou analysis because we did no conside unde ined da a o da a on uppe espi a o y ac in ec ions, and when s udies p esen ed di e en es ima es acco ding o he se e i y o pneumonia we ex ac ed hose epo ing only he mos se e e cases and he mos p ecise o adjus ed measu e. Ou indings ha e po en ial clinical implica ions. ACE inhibi o s a e widely p esc ibed and p esc ip ions may be in luenced by conce ns abou po en ial ad e se e ec s, in pa icula cough, which may be p o ec i e. The incidence o ACE inhibi o induced cough has been epo ed o be in he ange o 5% o 35%.80 Ou esul s sugges ha pa ien s aking ACE inhibi o s who de elop cough should, p o iding ha cough is ole able, pe sis wi h ea men . Compliance and pe sis ence wi h ea men is impo an . Fu he mo e, om an e idence based pe spec i e, he e is li le o choose be ween ACE inhibi o s and he mo e expensi e ARBs. Howe e , in he case o a pa icula pa ien , in whom ACE inhibi o s and ARBs a e p esumed o ha e simila clinical bene i , ou esul s may also in luence he choice o p esc ip ion in hose a high isk o pneumonia. The e o e pa ien s wi h isk ac o s o pneumonia and mo bidi ies ha equi e ea men wi h ACE inhibi o s may ha e an addi ional eason o con inue ea men . A u he impo an aspec o ou esul s was he educ ion in isk o pneumonia ac oss high isk pa ien s, which p o ided consis ency o he o e all esul s. Pa ien s wi h p e ious s oke ha e inc eased suscep ibili y o pneumonia owing o isk o aspi a ion associa ed wi h dec eased p o ec i e e lexes o he espi a o y sys em media ed by subs ance P and pos -s oke dysphagia.14 81 Abou 20% o hese pa ien s will de elop pneumonia,82 which is a p edic o o poo unc ional ou come83 84 and a ele an cause o dea h.84 85 The pu a i e p o ec i e e ec o ACE inhibi o s in his popula ion was p edic able gi en he impo ance o dysphagia and subs ance P in hese pa ien s. Acco ding o one s udy, ARBs do no inc ease he le els o subs ance P o imp o e asymp oma ic dysphagia.86 This highligh s he impo ance o using ACE inhibi o s in pa ien s wi h p e ious s oke who ha e como bidi ies o which ACE inhibi o s a e ecommended. Only a ew s udies e alua ed o he popula ions wi h inc eased isk, such as pa ien s wi h hea ailu e o ch onic kidney disease. Fo pa ien s wi h hea ailu e, he dec eased isk o pneumonia was also ound in pa ien s ea ed wi h ACE inhibi o s. The sugges ed e ec was signi ican bu his e alua ion lacked obus da a. ARBs did no show any p o ec i e e ec . The pu a i e p e en i e e ec o ACE inhibi o s on pneumonia in Asian pa ien s has been sugges ed.37 We explo ed his subg oup and compa ed he e ec wi h non-Asian pa ien s. Fu he mo e, we ob ained a conside able weigh o e idence om s udies ha e alua ed Asian pa ien s. ACE inhibi o s signi ican ly educed he isk o pneumonia in bo h Asian and non-Asian pa ien s, al hough he odds educ ion was signi ican ly highe in Asian pa ien s (57% 12%; P<0.001). ARBs did no educe he isk o pneumonia in ei he popula ion. Gene ic di e ences in ACE polymo phisms be ween Asian and non-Asian pa ien s ha e been sugges ed o explain he di e ence in p o ec i e e ec s. Polymo phisms I/I and I/D, which a e mo e p e alen in Asian popula ion, showed a p o ec i e end in he pos -hoc analysis in PROGRESS, whe eas he D/D polymo phism was less p o ec i e.20 77 87 This las polymo phism is associa ed wi h acu e espi a o y dis ess synd ome, pa icula ly in whi e popula ions.88 This po en ial loss o p o ec i e e ec may be explained by inc eased le els o se um ACE inhibi o s and ca abolism o kinins in pa ien s wi h he D/D polymo phism.89 Howe e , gene ic e idence is equi ocal. One s udy did no ind an associa ion be ween any speci ic geno ype and pneumonia.90 O he ac o s should be explo ed o explain hese di e ences in e hnic g oups o by geog aphical loca ion o be e de ine hose who can bene i mo e. Ou conclusions a e weake o pneumonia ela ed mo ali y because ewe s udies p o ided da a o his ou come and signi ican he e ogenei y exis ed o he esul s o ACE inhibi o s. This unce ain y was e lec ed by he wide con idence in e als. T ea men wi h ACE inhibi o s ( h ee andomised con olled ials and ou coho s udies) and ARBs (one andomised con olled ial) we e bo h associa ed wi h a dec eased isk o pneumonia ela ed mo ali y. Explana ions o such indings may ely on modula ion o ca dio ascula isk by ACE inhibi o s and ARBs because dea hs due o ca dio ascula disease a e no uncommon among pa ien s wi h pneumonia.27 91 Dec eased mo ali y may also be explained by he ole o ACE inhibi o s in pulmona y inju y and p oduc ion o cy okines, which may be ela ed o se e i y o pneumonia.92-94 ACE inhibi o s may in luence he pa e n o elease o cy okines exe ing an i-in lamma o y e ec s ha could educe he se e i y o and mo ali y om pneumonia.95 The in luence o ACE inhibi o s on su i al in hese pa ien s should be in e p e ed ca e ully because obse a ional s udies wi h en iched popula ions accoun ed o mos o he weigh o he pooled analysis, whe eas me a-analysis o h ee andomised con olled ials (one wi h an en iched popula ion) did no show di e ences be ween ACE inhibi o s and con ols. Howe e , he e was no signi ican di e ence in e ec s be ween o e all andomised con olled ials and obse a ional s udies. Al hough he da a we e no obus , hey did sugges ha he e ec s o ea men wi h ACE inhibi o s on mo ali y we e mos ly no iceable in pa ien s wi h pneumonia. Limi a ions o he e iew The esul s and conclusion o his e iew a e weakened by limi a ions inhe en o me a-analysis and indi idual s udies. The o e all quali y o included s udies was good. Howe e , epo ing quali y o a ew s udies, pa icula ly obse a ional ones, was No comme cial euse: See igh s and ep in s h p://www.bmj.com/pe missions Subsc ibe: h p://www.bmj.com/subsc ibe BMJ 2012;345:e4260 doi: 10.1136/bmj.e4260 (Published 11 July 2012) Page 5 o 20 RESEARCH low as some o hese we e abs ac ed om cha ac e limi ed sec ions such as le e s o commen s. The highe isk o bias was ound o po en ial selec i e epo ing in andomised con olled ials and p esen a ion o unadjus ed isk es ima es in obse a ional s udies. Bo h limi he s eng h o ou conclusions. A key limi a ion is ha no one andomised con olled ial was p ima ily designed o assess he e ec s o ACE inhibi o s o ARBs on pneumonia. Al hough we sea ched a la ge numbe o s udies, only a ew epo ed his ou come. Among hese, only wo andomised con olled ials (<25%) had pneumonia o pneumonia ela ed mo ali y as a p ede ined ou come.49 59 As a consequence we we e able o ex ac da a only om s udies whe e au ho s conside ed pneumonia o be an impo an ou come, because o ei he scien i ic in e es o s a is ical signi icance. Obse a ional s udies had an impo an weigh in he esul s o he p ima y ou come and his should be aken in o accoun when in e p e ing he clinical implica ions o ou indings. Use o ca dio ascula d ugs in obse a ional s udies could bias esul s, because pa ien s using d ugs could be mo e conce ned o hei heal h and mo e willing o ollow medical ad ice han con ols, he so-called heal hy use e ec bias.96 Howe e , pa ien s wi h pneumonia a e likely o ha e a highe isk o ca dio ascula disease91 97 and a e mo e likely o be ea ed wi h ACE inhibi o s, coun e balancing he bias om a heal hy use e ec . Addi ionally, he magni ude o he odds isk educ ion was simila o andomised con olled ials, coho s udies, and case-con ol s udies. Pooling da a om s udies wi h di e en designs (con ounding bias in obse a ional s udies) ha e alua ed pa ien s in di e en se ings (communi y based and hospi al based s udies; e e al bias), as well as wi h di e en baseline mo bidi ies and he e ogeneous isk (membe ship bias) o pneumonia, should also be aken in o accoun as limi a ions o ou conclusions. The deg ee o s a is ical he e ogenei y was in ac high in some compa isons. Ne e heless, he pooled es ima es om expe imen al and obse a ional s udies we e simila . In his case, pooling expe imen al and obse a ional da a inc eased he powe and ex e nal alidi y o he indings. Included s udies compa ed di e en ACE inhibi o s and ARBs wi h di e en con ols, such as placebo, calcium channel blocke s, and β blocke s. In he p esen analysis we did no ca y ou se ial subg oup analysis o explo e i he e ec was di e en o a pa icula d ug because o he sca ci y o he da a and he isk o ob aining a esul by chance. Finally, we used adjus ed indi ec compa isons o es ima e he e ec o ACE inhibi o s compa ed wi h ARBs. Al hough combined indi ec and di ec e idence showed no disc epancies o he e ogenei y, he esul s should no be hough as de ini i e conclusions because o he possibili y o imbalanced da a om s udies wi h di e en designs, baseline isk o pa ien s, and leng h o ollow-up. Conclusions Ou esul s sugges an impo an ole o ACE inhibi o s, bu no ARBs, in educing he isk o pneumonia. These da a may discou age he wi hd awal o ACE inhibi o s in some pa ien s wi h ole able ad e se e en s (namely, cough) who a e a pa icula ly high isk o pneumonia. Speci ic designed andomised con olled ials a e equi ed o es ablish de ini e conclusions and o es ima e be e he ue magni ude o his pu a i e p o ec i e e ec . Pa ien s wi h p e ious s oke and Asian pa ien s a e pa ien popula ions ha could bene i mo e om ea men wi h ACE inhibi o s. ACE inhibi o s also lowe ed he isk o pneumonia ela ed mo ali y, mainly in pa ien s wi h es ablished disease, bu he obus ness o he e idence was weake . We hank he Coch ane Coo dina ing Cen e in Po ugal. Con ibu o s: DC and JA con ibu ed o he concep and design, da a acquisi ion, da a analysis, and in e p e a ion o he da a; w o e he i s d a o he manusc ip ; c i ically e ised he manusc ip ; and ga e inal app o al o he submi ed manusc ip . AVC con ibu ed o he in e p e a ion o da a, c i ically e ised he manusc ip , and ga e inal app o al o he submi ed manusc ip . JC con ibu ed o he concep and design, da a analysis, and in e p e a ion o he da a; w o e he i s d a o he manusc ip ; c i ically e ised he manusc ip ; and ga e inal app o al o he submi ed manusc ip . JC is he gua an o . Funding: This was an academic p ojec no unded by go e nmen o non-go e nmen g an s. Compe ing in e es s: All au ho s ha e comple ed he ICMJE uni o m disclosu e o m a www.icmje.o g/coi_disclosu e.pd (a ailable on eques om he co esponding au ho ) and decla e: no suppo om any o ganisa ion o he submi ed wo k; no inancial ela ionships wi h any o ganisa ions ha migh ha e an in e es in he submi ed wo k in he p e ious h ee yea s; and no o he ela ionships o ac i i ies ha could appea o ha e in luenced he submi ed wo k. E hical app o al: No equi ed. Da a sha ing: No addi ional da a a ailable. 1 File TM. Communi y-acqui ed pneumonia. Lance 2003;362:1991-2001. 2 Lim WS, Baudouin SV, Geo ge RC, Hill AT, Jamieson C, Le Jeune I, e al. BTS guidelines o he managemen o communi y acqui ed pneumonia in adul s: upda e 2009. Tho ax 2009;64(suppl 3):iii,1-55. 3Samokh alo AV, I ing HM, Rehm J. Alcohol consump ion as a isk ac o o pneumonia: a sys ema ic e iew and me a-analysis. Epidemiol In ec 2010;138:1789-95. 4 Almi all J, Bolíba I, Balanzó X, González CA. Risk ac o s o communi y-acqui ed pneumonia in adul s: a popula ion-based case-con ol s udy. Eu Respi J 1999;13:349-55. 5 Vinog ado a Y, Hippisley-Cox J, Coupland C. Iden i ica ion o new isk ac o s o pneumonia: popula ion-based case-con ol s udy. B J Gen P ac 2009;59:e329-38. 6 Fung HB, Mon eagudo-Chu MO. Communi y-acqui ed pneumonia in he elde ly. Am J Ge ia Pha maco he 2010;8:47-62. 7 Lee HC, Chang KC, Lan CF, Hong CT, Huang YC, Chang ML. Fac o s associa ed wi h p olonged hospi al s ay o acu e s oke in Taiwan. Ac a Neu ol Taiwan 2008;17:17-25. 8 Me e sky ML, Ta e JP, Fine MJ, Pe illo MK, Meehan TP. Tempo al ends in ou comes o olde pa ien s wi h pneumonia. A ch In e n Med 2000;160:3385-91. 9 Eom CS, Jeon CY, Lim JW, Cho EG, Pa k SM, Lee KS. Use o acid-supp essi e d ugs and isk o pneumonia: a sys ema ic e iew and me a-analysis. CMAJ 2011;183:310-9. 10 Tleyjeh IM, Kashou T, Hakim FA, Zimme man VA, E win PJ, Su on AJ, e al. S a ins o he p e en ion and ea men o in ec ions: a sys ema ic e iew and me a-analysis. A ch In e n Med 2009;169:1658-67. 11 Mo ice AH, Low y R, B own MJ, Higenbo am T. Angio ensin-con e ing enzyme and he cough e lex. Lance 198;2:1116-8. 12 Fox AJ, Lalloo UG, Bel isi MG, Be na eggi M, Chung KF, Ba nes PJ. B adykinin-e oked sensi iza ion o ai way senso y ne es: a mechanism o ACE-inhibi o cough. Na Med 1996;2:814-7. 13 Tomaki M, Ichinose M, Miu a M, Hi ayama Y, Kageyama N, Yamauchi H, e al. Angio ensin con e ing enzyme (ACE) inhibi o -induced cough and subs ance P. Tho ax 1996;51:199-201. 14 Sekizawa K, Ujiie Y, I abashi S, Sasaki H, Takishima T. Lack o cough e lex in aspi a ion pneumonia. Lance 1990;335:1228-9. 15 Pon oppidan H, Beeche HK. P og essi e loss o p o ec i e e lexes in he ai way wi h he ad ance o age. JAMA 1960;174:2209-13. 16 Nakayama K, Sekizawa K, Sasaki H. ACE inhibi o and swallowing e lex. Ches 1998;113:1425. 17 El Solh AA, Saliba R. Pha macologic p e en ion o aspi a ion pneumonia: a sys ema ic e iew. Am J Ge ia Pha maco he 2007;5:352-62. 18 Ra ailidis PI, Ma haiou DK, Va bobi is I, Falagas ME. Use o ACE inhibi o s and isk o communi y-acqui ed pneumonia: a e iew. Eu J Clin Pha macol 2008;64:565-73. 19 Siempos II, Va dakas KZ, Kop e ides P, Falagas ME. Adjunc i e he apies o communi y-acqui ed pneumonia: a sys ema ic e iew. J An imic ob Chemo he 2008;62:661-8. 20 Ohkubo T, Chapman N, Neal B, Woodwa d M, Omae T, Chalme s J, e al. E ec s o an angio ensin-con e ing enzyme inhibi o -based egimen on pneumonia isk. Am J Respi C i Ca e Med 2004;169:1041-5. 21 E minan M, Zhang B, Fi zge ald M, B ophy JM. Do angio ensin-con e ing enzyme inhibi o s o angio ensin II ecep o blocke s dec ease he isk o hospi aliza ion seconda y o communi y-acqui ed pneumonia? A nes ed case-con ol s udy. Pha maco he apy 2006;26:479-82. 22 Van de Ga de EM, Sou e ein PC, an den Bosch JM, Denee V H, Leu kens HG. Angio ensin-con e ing enzyme inhibi o use and pneumonia isk in a gene al popula ion. Eu Respi J 2006;27:1217-22. 23 Dicks ein K, Kjekshus J; OPTIMAAL S ee ing Commi ee o he OPTIMAAL S udy G oup. E ec s o losa an and cap op il on mo ali y and mo bidi y in high- isk pa ien s a e acu e myoca dial in a c ion: he OPTIMAAL andomised ial. Op imal T ial in Myoca dial In a c ion wi h Angio ensin II An agonis Losa an. Lance 2002;360:752-60. No comme cial euse: See igh s and ep in s h p://www.bmj.com/pe missions Subsc ibe: h p://www.bmj.com/subsc ibe BMJ 2012;345:e4260 doi: 10.1136/bmj.e4260 (Published 11 July 2012) Page 6 o 20 RESEARCH Wha is al eady known on his opic Angio ensin con e ing enzyme (ACE) inhibi o s educe mo bidi y and mo ali y in pa ien s wi h ca dio ascula disease These d ugs also ha e seconda y e ec s on he espi a o y sys em, sugges ed o p o ec agains pneumonia Mos o he da a on his issue a e p o ided by he e ogeneous obse a ional s udies wi h inconclusi e esul s Wha his s udy adds In pooled esul s om bo h in e en ional and obse a ional s udies, ACE inhibi o s, bu no angio ensin ecep o blocke s (ARBs), showed a s a is ical and pu a i e clinically signi ican p o ec i e ole agains pneumonia This esul may discou age he wi hd awal o ACE inhibi o s in pa ien s wi h ole able ad e se e en s—namely, cough This p o ec i e e ec was highe among Asian pa ien s and in hose wi h p e ious s oke; pa ien popula ions ha may bene i mos om ACE inhibi o s 24 P e e MA, McMu ay JJ, Velazquez EJ, Rouleau JL, Købe L, Maggioni AP, e al. Valsa an, cap op il, o bo h in myoca dial in a c ion complica ed by hea ailu e, le en icula dys unc ion, o bo h. N Engl J Med 2003;349:1893-906. 25 S oup DF, Be lin JA, Mo on SC, Olkin I, Williamson GD, Rennie D, e al. Me a-analysis o obse a ional s udies in epidemiology: a p oposal o epo ing. Me a-analysis O Obse a ional S udies in Epidemiology (MOOSE) g oup. JAMA 2000;283:2008-12. 26 Libe a i A, Al man DG, Te zla J, Mul ow C, Gø zsche PC, Ioannidis JP, e al. The PRISMA s a emen o epo ing sys ema ic e iews and me a-analyses o s udies ha e alua e heal hca e in e en ions: explana ion and elabo a ion. BMJ 2009;339:b2700. 27 Mo ensen EM, Coley CM, Singe DE, Ma ie TJ, Ob osky DS, Kapoo WN, e al. Causes o dea h o pa ien s wi h communi y-acqui ed pneumonia: esul s om he Pneumonia Pa ien Ou comes Resea ch Team coho s udy. A ch In e n Med 2002;162:1059-64. 28 Wong WC, Cheung CS, Ha GJ. De elopmen o a quali y assessmen ool o sys ema ic e iews o obse a ional s udies (QATSO) o HIV p e alence in men ha ing sex wi h men and associa ed isk beha iou s. Eme g Themes Epidemiol 2008;5:23. 29 Von Elm E, Al man DG, Egge M, Pocock SJ, Gø zsche PC, Vandenb oucke JP, e al. The S eng hening he Repo ing o Obse a ional S udies in Epidemiology (STROBE) s a emen : guidelines o epo ing obse a ional s udies. Epidemiology 2007;18:800-4. 30 Ca e P, G ay LJ, T ough on J, Khun i K, Da ies MJ. F ui and ege able in ake and incidence o ype 2 diabe es melli us: sys ema ic e iew and me a-analysis. BMJ 2010;341:c4229. 31 Bui ago-Lopez A, Sande son J, Johnson L, Wa nakula S, Wood A, Di Angelan onio E, e al. Chocola e consump ion and ca diome abolic diso de s: sys ema ic e iew and me a-analysis. BMJ 2011;343:d4488. 32 Higgins JPT, Al man DG, S e ne JAC. Assessing isk o bias in included s udies. In: Higgins JPT, G een S, eds. Coch ane handbook o sys ema ic e iews o in e en ions e sion 5.1.0 . Coch ane Collabo a ion, 2011. 33 Deeks JJ, Al man DG, B adbu n MJ. S a is ical me hods o examining he e ogenei y and combining esul s om se e al s udies in me aanalysis. In Egge M, Da ey Smi h G, Al man DG, eds. Sys ema ic e iews in heal h ca e: me a-analysis in con ex . 2nd ed. BMJ Publishing G oup, 2001:313-35. 34 Deeks JJ. Issues in he selec ion o a summa y s a is ic o me a-analysis o clinical ials wi h bina y ou comes. S a Med 2002;21:1575-600. 35 Nakagawa T, Sekizawa K, A ai H, Kikuchi R, Manabe K, Sasaki H. High incidence o pneumonia in elde ly pa ien s wi h basal ganglia in a c ion. A ch In e n Med 1997;157:321-4. 36 Nakagawa T, Sekizawa K, Nakajoh K, Tanji H, A ai H, Sasaki H. Silen ce eb al in a c ion: a po en ial isk o pneumonia in he elde ly. J In e n Med 2000;247:255-9. 37 Te amo o S, Yamamo o H, Yamaguchi Y, Hanaoka Y, Ishii M, Hibi S, e al. ACE inhibi o s p e en aspi a ion pneumonia in Asian, bu no Caucasian, elde ly pa ien s wi h s oke. Eu Respi J 2007;29:218-9 38 Buche HC, Guya GH, G i i h LE, Wal e SD. The esul s o di ec and indi ec ea men compa isons in me a-analysis o andomized con olled ials. J Clin Epidemiol 1997;50:683-91. 39 Song F, Al man DG, Glenny AM, Deeks JJ. Validi y o indi ec compa ison o es ima ing e icacy o compe ing in e en ions: empi ical e idence om published me a-analyses. BMJ 2003;326:472. 40 De Simonian R, Lai d N. Me a-analysis in clinical ials. Con ol Clin T ials 1986;7:177-88. 41 Glenny AM, Al man DG, Song F, Saka o i ch C, Deeks JJ, D’Amico R, e al. Indi ec compa isons o compe ing in e en ions. Heal h Technol Assess 2005;9:1-134,iii-i . 42 Song F, Ha ey I, Lil o d R. Adjus ed indi ec compa ison may be less biased han di ec compa ison o e alua ing new pha maceu ical in e en ions. J Clin Epidemiol 2008;61:455-63. 43 Smee h L, Haines A, Eb ahim S. Numbe s needed o ea de i ed om me a-analyses—some imes in o ma i e, usually misleading. BMJ 1999;318:1548-51. 44 Widimský J, K eme HJ, Je ie P, Uhlí O. Czech and Slo ak spi ap il in e en ion s udy (CASSIS). A andomized, placebo and ac i e-con olled, double-blind mul icen e ial in pa ien s wi h conges i e hea ailu e. Eu J Clin Pha macol 1995;49:95-102. 45 Købe L, To p-Pede sen C, Ca lsen JE, Bagge H, Eliasen P, Lyngbo g K, e al. A clinical ial o he angio ensin-con e ing-enzyme inhibi o andolap il in pa ien s wi h le en icula dys unc ion a e myoca dial in a c ion. T andolap il Ca diac E alua ion (TRACE) S udy G oup. N Engl J Med 1995;333:1670-6. 46 Randomised placebo-con olled ial o e ec o amip il on decline in glome ula il a ion a e and isk o e minal enal ailu e in p o einu ic, non-diabe ic neph opa hy. The GISEN G oup (G uppo I aliano di S udi Epidemiologici in Ne ologia). Lance 1997;349:1857-63. 47 Yusu S, Sleigh P, Pogue J, Bosch J, Da ies R, Dagenais G. E ec s o an angio ensin-con e ing-enzyme inhibi o , amip il, on ca dio ascula e en s in high- isk pa ien s. The Hea Ou comes P e en ion E alua ion S udy In es iga o s. N Engl J Med 2000;342:145-53. 48 PROGRESS Collabo a i e G oup. Randomised ial o a pe indop il-based blood-p essu e-lowe ing egimen among 6,105 indi iduals wi h p e ious s oke o ansien ischaemic a ack. Lance 2001;358:1033-41. 49 Kanda A, Ebiha a S, Yasuda H, Takashi O, Sasaki T, Sasaki H. A combina o ial he apy o pneumonia in elde ly people. J Am Ge ia Soc 2004;52:846-7. 50 Hou FF, Zhang X, Zhang GH, Xie D, Chen PY, Zhang WR, e al. E icacy and sa e y o benazep il o ad anced ch onic enal insu iciency. N Engl J Med 2006;354:131-40. 51 Webe M. Clinical sa e y and ole abili y o losa an. Clin The 1997;19:604-16. 52 Lewis EJ, Hunsicke LG, Cla ke WR, Be l T, Pohl MA, Lewis JB, e al. Renop o ec i e e ec o he angio ensin- ecep o an agonis i besa an in pa ien s wi h neph opa hy due o ype 2 diabe es. N Engl J Med 2001;345:851-60. 53 Pa ing HH, Lehne H, B öchne -Mo ensen J, Gomis R, Ande sen S, A ne P, e al. The e ec o i besa an on he de elopmen o diabe ic neph opa hy in pa ien s wi h ype 2 diabe es. N Engl J Med 2001;345:870-8. 54 Dahlö B, De e eux RB, Kjeldsen SE, Julius S, Bee e s G, de Fai e U, e al. Ca dio ascula mo bidi y and mo ali y in he Losa an In e en ion Fo Endpoin educ ion in hype ension s udy (LIFE): a andomised ial agains a enolol. Lance 2002;359:995-1003. 55 P e e MA, Swedbe g K, G ange CB, Held P, McMu ay JJ, Michelson EL, e al. E ec s o candesa an on mo ali y and mo bidi y in pa ien s wi h ch onic hea ailu e: he CHARM-O e all p og amme. Lance 2003;362:759-66. 56 Sch ade J, Lüde s S, Kulschewski A, Hamme sen F, Pla e K, Be ge J, e al. Mo bidi y and mo ali y a e s oke, ep osa an compa ed wi h ni endipine o seconda y p e en ion: p incipal esul s o a p ospec i e andomized con olled s udy (MOSES). S oke 2005;36:1218-26. 57 Telmisa an Randomised AssessmeN S udy in ACE iN ole an subjec s wi h ca dio ascula Disease (TRANSCEND) In es iga o s: Yusu S, Teo K, Ande son C, Pogue J, Dyal L, Copland I, e al. E ec s o he angio ensin- ecep o blocke elmisa an on ca dio ascula e en s in high- isk pa ien s in ole an o angio ensin-con e ing enzyme inhibi o s: a andomised con olled ial. Lance 2008;372:1174-83. 58 Yusu S, Diene HC, Sacco RL, Co on D, Ounpuu S, Law on WA, e al. Telmisa an o p e en ecu en s oke and ca dio ascula e en s. N Engl J Med 2008;359:1225-37. 59 Kasanuki H, Hagiwa a N, Hosoda S, Sumiyoshi T, Honda T, Haze K, e al. Angio ensin II ecep o blocke -based s non-angio ensin II ecep o blocke -based he apy in pa ien s wi h angiog aphically documen ed co ona y a e y disease and hype ension: he Hea Ins i u e o Japan Candesa an Randomized T ial o E alua ion in Co ona y A e y Disease (HIJ-CREATE). Eu Hea J 2009;30:1203-12. 60 Willenheime R, Helme s C, Pan e E, Rydbe g E, Lö dahl P, Go don A, e al. Sa e y and e icacy o alsa an e sus enalap il in hea ailu e pa ien s. In J Ca diol 2002;85:261-70. 61 ONTARGET In es iga o s: Yusu S, Teo KK, Pogue J, Dyal L, Copland I, Schumache H, e al. Telmisa an, amip il, o bo h in pa ien s a high isk o ascula e en s. N Engl J Med 2008;358:1547-59. 62 Sekizawa K, Ma sui T, Nakagawa T, Nakayama K, Sasaki H. ACE inhibi o s and pneumonia. Lance 1998;352:1069. 63 Te amo o S, Ouchi Y. ACE inhibi o s and p e en ion o aspi a ion pneumonia in elde ly hype ensi es. Lance 1999;353:843. 64 A ai T, Yasuda Y, Takaya T, Toshima S, Kashiki Y, Yoshimi N, e al. ACE inhibi o s and educ ion o he isk o pneumonia in elde ly people. Am J Hype ens 2000;13:1050-1. 65 A ai T, Yasuda Y, Toshima S, Yoshimi N, Kashiki Y. ACE inhibi o s and pneumonia in elde ly people. Lance 1998;352:1937-8. 66 A ai T, Yasuda Y, Takaya T, Toshima S, Kashiki Y, Shibayama M, e al. Angio ensin-con e ing enzyme inhibi o s, angio ensin-II ecep o an agonis s, and pneumonia in elde ly hype ensi e pa ien s wi h s oke. Ches 2001;119:660-1. 67 A ai T, Sekizawa K, Oh ui T, Fujiwa a H, Yoshimi N, Ma suoka H, e al. ACE inhibi o s and p o ec ion agains pneumonia in elde ly pa ien s wi h s oke. Neu ology 2005;64:573-4. 68 Mo ensen EM, Res epo MI, Anzue o A, Pugh J. The impac o p io ou pa ien ACE inhibi o use on 30-day mo ali y o pa ien s hospi alized wi h communi y-acqui ed pneumonia. BMC Pulm Med 2005;5:12. 69 Ha ada J, Sekizawa K. Angio ensin-con e ing enzyme inhibi o s and pneumonia in elde ly pa ien s wi h in ace eb al hemo hage. J Am Ge ia Soc 2006;54:175-6. 70 Mo ensen EM, Pugh MJ, Copeland LA, Res epo MI, Co nell JE, Anzue o A, e al. Impac o s a ins and angio ensin-con e ing enzyme inhibi o s on mo ali y o subjec s hospi alised wi h pneumonia. Eu Respi J 2008;31:611-7. 71 Chalme s JD, Singanayagam A, Mu ay MP, Hill AT. P io s a in use is associa ed wi h imp o ed ou comes in communi y-acqui ed pneumonia. Am J Med 2008;121:1002-7. 72 Myles PR, Hubba d RB, Gibson JE, Pogson Z, Smi h CJ, McKee e TM. The impac o s a ins, ACE inhibi o s and gas ic acid supp essan s on pneumonia mo ali y in a UK gene al p ac ice popula ion coho . Pha macoepidemiol D ug Sa 2009;18:697-703. 73 Pos e p esen a ions om he Wo ld Cong ess o Ca diology Scien i ic Sessions 2010. Ci cula ion 2010;xx:e348. 74 Mukamal KJ, Ghimi e S, Pandey R, O’Mea a ES, Gau am S. An ihype ensi e medica ions and isk o communi y-acqui ed pneumonia. J Hype ens 2010;28:401-5. 75 Okaishi K, Mo imo o S, Fukuo K, Niinobu T, Ha a S, Onishi T, e al. Reduc ion o isk o pneumonia associa ed wi h use o angio ensin I con e ing enzyme inhibi o s in elde ly inpa ien s. Am J Hype ens 1999;12:778-83. 76 El Solh AA, B ewe T, Okada M, Bashi O, Gough M. Indica o s o ecu en hospi aliza ion o pneumonia in he elde ly. J Am Ge ia Soc 2004;52:2010-5. 77 Takahashi T, Mo imo o S, Okaishi K, Kanda T, Nakahashi T, Oku o M, e al. Reduc ion o pneumonia isk by an angio ensin I-con e ing enzyme inhibi o in elde ly Japanese inpa ien s acco ding o inse ion/dele ion polymo phism o he angio ensin I-con e ing enzyme gene. Am J Hype ens 2005;18:1353-9. 78 Van de Ga de EM, Sou e ein PC, Hak E, Denee VH, an den Bosch JM, Leu kens HG. Angio ensin-con e ing enzyme inhibi o use and p o ec ion agains pneumonia in pa ien s wi h diabe es. J Hype ens 2007;25:235-9. No comme cial euse: See igh s and ep in s h p://www.bmj.com/pe missions Subsc ibe: h p://www.bmj.com/subsc ibe BMJ 2012;345:e4260 doi: 10.1136/bmj.e4260 (Published 11 July 2012) Page 7 o 20 RESEARCH 79 Ma ciniak C, Ko u z AW, Lin E, Ro h E, Wel y L, Lo ell L. Examina ion o selec ed clinical ac o s and medica ion use as isk ac o s o pneumonia du ing s oke ehabili a ion: a case-con ol s udy. Am J Phys Med Rehabil 2009;88:30-8. 80 Dicpinigai is PV. Angio ensin-con e ing enzyme inhibi o -induced cough: ACCP e idence-based clinical p ac ice guidelines. Ches 2006;129:169-73S. 81 Wal e U, Knoblich R, S einhagen V, Dona M, Benecke R, Klo h A. P edic o s o pneumonia in acu e s oke pa ien s admi ed o a neu ological in ensi e ca e uni . J Neu ol 2007;254:1323-9. 82 Ro h EJ, Lo ell L, Ha ey RL, Heinemann AW, Semik P, Diaz S. Incidence o and isk ac o s o medical complica ions du ing s oke ehabili a ion. S oke 2001;32:523-9. 83 Ve meij FH, Schol e op Reime WJ, de Man P, an Oos enb ugge RJ, F anke CL, de Jong G, e al. S oke-associa ed in ec ion is an independen isk ac o o poo ou come a e acu e ischemic s oke: da a om he Ne he lands S oke Su ey. Ce eb o asc Dis 2009;27:465-71. 84 Hilke R, Poe e C, Findeisen N, Sobesky J, Jacobs A, Ne eling M, e al. Nosocomial pneumonia a e acu e s oke: implica ions o neu ological in ensi e ca e medicine. S oke 2003;34:975-81. 85 Ka zan IL, Cebul RD, Husak SH, Dawson NV, Bake DW. The e ec o pneumonia on mo ali y among pa ien s hospi alized o acu e s oke. Neu ology 2003;60:620-5. 86 A ai T, Yasuda Y, Takaya T, Toshima S, Kashiki Y, Yoshimii N, e al. Angio ensin-con e ing enzyme inhibi o s, angio ensin II ecep o an agonis s, and symp omless dysphagia. Ches 2000;117:1819-20. 87 Sagnella GA, Ro hwell MJ, Onipinla AK, Wicks PD, Cook DG, Cappuccio FP. A popula ion s udy o e hnic a ia ions in he angio ensin-con e ing enzyme I/D polymo phism: ela ionships wi h gende , hype ension and impai ed glucose me abolism. J Hype ens 1999;17:657-64. 88 Hu Z, Jin X, Kang Y, Liu C, Zhou Y, Wu X, e al. Angio ensin-con e ing enzyme inse ion/dele ion polymo phism associa ed wi h acu e espi a o y dis ess synd ome among caucasians. J In Med Res 2010;38:415-22. 89 B own NJ, Blais C J , Gandhi SK, Adam A. ACE inse ion/dele ion geno ype a ec s b adykinin me abolism. J Ca dio asc Pha macol 1998;32:373-7. 90 Van de Ga de EM, Endeman H, Denee VH, Biesma DH, Sayed-Taba abaei FA, Ru en HJ, e al. Angio ensin-con e ing enzyme inse ion/dele ion polymo phism and isk and ou come o pneumonia. Ches 2008;133:220-5. 91 Rami ez J, Alibe i S, Mi saeidi M, Pey ani P, Fila do G, Ami A, e al. Acu e myoca dial in a c ion in hospi alized pa ien s wi h communi y-acqui ed pneumonia. Clin In ec Dis 2008;47:182-7. 92 K anzhö e R, Schmid J, P ei e CA, Hagl S, Libby P, Küble W. Angio ensin induces in lamma o y ac i a ion o human ascula smoo h muscle cells. A e ioscle Th omb Vasc Biol 1999;19:1623-9. 93 Wös en- an Aspe en RM, Lu e R, Hai sma JJ, Me kus MP, an Woensel JB, an de Loos CM, e al. ACE media es en ila o -induced lung inju y in a s ia angio ensin II bu no b adykinin. Eu Respi J 2008;31:363-71. 94 An unes G, E ans SA, Lo dan JL, F ew AJ. Sys emic cy okine le els in communi y-acqui ed pneumonia and hei associa ion wi h disease se e i y. Eu Respi J 2002;20:990-5. 95 Gulles ad L, Auk us P, Ueland T, Espe ik T, Yee G, Vagelos R, e al. E ec o high- e sus low-dose angio ensin con e ing enzyme inhibi ion on cy okine le els in ch onic hea ailu e. J Am Coll Ca diol 1999;34:2061-7. 96 Majumda SR, McAlis e FA, Eu ich DT, Padwal RS, Ma ie TJ. S a ins and ou comes in pa ien s admi ed o hospi al wi h communi y acqui ed pneumonia: popula ion based p ospec i e coho s udy. BMJ 2006;333:999. 97 Co ales-Medina VF, Suh KN, Rose G, Chi inos JA, Douce e S, Came on DW, e al. Ca diac complica ions in pa ien s wi h communi y-acqui ed pneumonia: a sys ema ic e iew and me a-analysis o obse a ional s udies. PLoS Med 2011;8:e1001048. Accep ed: 10 May 2012 Ci e his as: BMJ 2012;345:e4260 This is an open-access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion Non-comme cial License, which pe mi s use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed, he use is non comme cial and is o he wise in compliance wi h he license. See: h p://c ea i ecommons.o g/licenses/by- nc/2.0/ and h p://c ea i ecommons.o g/licenses/by-nc/2.0/legalcode. No comme cial euse: See igh s and ep in s h p://www.bmj.com/pe missions Subsc ibe: h p://www.bmj.com/subsc ibe BMJ 2012;345:e4260 doi: 10.1136/bmj.e4260 (Published 11 July 2012) Page 8 o 20 RESEARCH Tables Table 1| Main cha ac e is ics o andomised con olled ials included in e iew Da a* Ou comes abs ac ed P ima y ou come Mean (SD) ageNo ( o al)Compa isonPa ien s Mean ollow-up (yea s)Loca ionS udy ACE inhibi o s con ol: PublishedSe ious pneumonia†Assessmen o changes in exe cise ole ance 57.5 (10) 200 48 (248) Spi ap il o enalap il placeboPa ien s wi h ch onic conges i e hea ailu e 0.2Mul icen e Czech Republic and Slo akia CASSIS 199544 PublishedPneumonia†Dea h om any cause 67.5876 873 (1749) T andolap il placeboPa ien s wi h le en icula ejec ion ac ion a e myoca dial in a c ion 4.0Mul icen e Denma k TRACE 199545 PublishedD ug wi hd awal due o b onchopneumonia† Ra e o decline in glome ula il a ion 49.3 (13.6) 78 88 (166) Ramip il placeboPa ien s wi h ch onic neph opa hy and pe sis en p o einu ia 1.3Mul icen e I aly GISEN 199746 UnpublishedSe ious pneumonia†Myoca dial in a c ion, s oke, o dea h due o ca dio ascula disease 66 (7)4645 4652 (9297) Ramip il placeboPa ien s a high isk o de eloping a majo ca dio ascula e en 4.0Mul icen e wo ldwide (no Asia) HOPE 200047 PublishedFa al and non- a al pneumonia Fa al o non- a al s oke 64 (10) 3051 3054 (6105) Pe indop il placeboPa ien s wi h p e ious s oke o ansien ischaemic a ack 3.9Mul icen e wo ldwide PROGRESS 200420 48 PublishedHospi al dea hIn-hospi al mo ali y, du a ion o an ibio ic use, and in ec ion wi h MRSA 78 (8)33 35 (68)Imidap il+aman adine+s anda d ca e s anda d ca e Pa ien s aged ≥65 wi h his o y o s oke and admi ed wi h communi y acqui ed pneumonia 4.0Single cen e Japan Kanda 200449 PublishedPneumonia as cause o mo ali y Composi e o doubling o se um c ea inine le el, end s age enal disease, and dea h 44.8 (14.6) 216 112 (328) Benazep il placeboPa ien s wi h non-diabe ic ch onic kidney disease 3.4Single cen e China Hou 200650 ARBs con ol: PublishedPneumonia†Ad e se e en s54125 29Losa an placeboPa ien s wi h essen ial hype ension and hea ailu e 0.3Wo ldwideWebe 199751 UnpublishedPulmona y in ec ion†Time o i s occu ence o doubling o baseline se um c ea inine le el, end s age enal 58.9 (7.8) 579 569 (placebo) (1148) I besa an placebo o amlodipine Pa ien s wi h hype ension and wi h ype 2 diabe es and o e p o einu ia 4.8Mul icen e wo ldwide IDNT 200152 disease, o dea h UnpublishedPulmona y in ec ion†Time o occu ence o clinical o e albuminu ia 58.0 (8.1) 389 201 (590) I besa an 100 mg i besa an 300 mg placebo Pa ien s wi h hype ension and wi h ype 2 diabe es, mic oalbuminu ia, and no mal enal unc ion 2.0Mul icen e Eu ope IRMA-2 200153 No comme cial euse: See igh s and ep in s h p://www.bmj.com/pe missions Subsc ibe: h p://www.bmj.com/subsc ibe BMJ 2012;345:e4260 doi: 10.1136/bmj.e4260 (Published 11 July 2012) Page 9 o 20 RESEARCH Figu es Fig 1 Flow o s udies h ough e iew. ACE=angio ensin con e ing enzyme; ARBs=angio ensin ecep o blocke s No comme cial euse: See igh s and ep in s h p://www.bmj.com/pe missions Subsc ibe: h p://www.bmj.com/subsc ibe BMJ 2012;345:e4260 doi: 10.1136/bmj.e4260 (Published 11 July 2012) Page 16 o 20 RESEARCH Fig 2 Risk o pneumonia wi h use o angio ensin con e ing enzyme (ACE) inhibi o s compa ed wi h con ol ea men No comme cial euse: See igh s and ep in s h p://www.bmj.com/pe missions Subsc ibe: h p://www.bmj.com/subsc ibe BMJ 2012;345:e4260 doi: 10.1136/bmj.e4260 (Published 11 July 2012) Page 17 o 20 RESEARCH Fig 3 Risk o pneumonia wi h use o angio ensin ecep o blocke s (ARBs) compa ed wi h con ol ea men No comme cial euse: See igh s and ep in s h p://www.bmj.com/pe missions Subsc ibe: h p://www.bmj.com/subsc ibe BMJ 2012;345:e4260 doi: 10.1136/bmj.e4260 (Published 11 July 2012) Page 18 o 20 RESEARCH Fig 4 Summa y o me a-analysis es ima es and subg oup analyses. ACE=angio ensin con e ing enzyme; ARBs=angio ensin ecep o blocke s No comme cial euse: See igh s and ep in s h p://www.bmj.com/pe missions Subsc ibe: h p://www.bmj.com/subsc ibe BMJ 2012;345:e4260 doi: 10.1136/bmj.e4260 (Published 11 July 2012) Page 19 o 20 RESEARCH Fig 5 Pneumonia ela ed mo ali y in s udies compa ing angio ensin con e ing enzyme (ACE) inhibi o s o angio ensin ecep o blocke s (ARBs) wi h con ol ea men No comme cial euse: See igh s and ep in s h p://www.bmj.com/pe missions Subsc ibe: h p://www.bmj.com/subsc ibe BMJ 2012;345:e4260 doi: 10.1136/bmj.e4260 (Published 11 July 2012) Page 20 o 20 RESEARCH