CHEST Originl Reserch Inhled Corticosteroids nd the Risk of Pneumoni in People With Asthm A Cse-Control Study Trici McKeever, PhD ; Timothy W. Hrrison, MD ; Richrd Hubbrd, MD ; nd Dominick Shw, MD ASTHMA Bckground: In clinicl trils, the use of inhled corticosteroids is ssocited with n incresed risk of pneumoni in people with COPD, but whether the sme is true for people with sthm is not known. Methods: With the use of primry cre dt from The Helth Improvement Network, we identified people with sthm, nd from this cohort, we identified ptients with pneumoni or lower respirtory trct infection nd ge- nd sex-mtched control subjects. Conditionl logistic regression ws used to determine the ssocition between the dose nd type of inhled corticosteroid nd the risk of pneumoni or lower respirtory trct infection. Results: A dose-response reltionship ws found between the strength of inhled corticosteroid dose nd risk of pneumoni or lower respirtory trct infection ( P,.001 for trend) such tht fter djusting for confounders, people receiving the highest strength of inhled corticosteroid ( 1,000 m g) hd 2.04 (95% CI, 1.59-2.64) incresed risk of pneumoni or lower respirtory trct infection compred with those with sthm who did not hve prescription for inhled corticosteroids within the previous 90 dys. Conclusions: People with sthm receiving inhled corticosteroids re t n incresed risk of pneumoni or lower respirtory infection, with those receiving higher doses being t greter risk. Pneumoni should be considered s possible side effect of inhled corticosteroids, nd the lowest possible dose of inhled corticosteroids should be used in the mngement of sthm. CHEST 2013; 144(6):1788 1794 Abbrevitions: ICS 5 inhled corticosteroid ; LRTI 5 lower respirtory trct infection Inhled corticosteroids (ICSs) re prescribed widely to people with sthm 1 nd COPD 2 to improve symptoms, mximize lung function, nd reduce excerbtion risk. Evidence suggests tht ICS use my be Mnuscript received April 10, 2013; revision ccepted July 31, 2013. Affilitions: From the Division of Epidemiology nd Public Helth (Drs McKeever nd Hubbrd) nd Respirtory Reserch Unit (Drs Hrrison nd Shw), University of Nottinghm, Nottinghm, Englnd. Funding/Support: The uthors hve reported to CHEST tht no funding ws received for this study. Correspondence to: Dominick Shw, MD, Respirtory Reserch Unit, University of Nottinghm, Edwrds Ln, Nottinghm, NG5 1PB, Englnd; e-mil: dominic.shw@nottinghm.c.uk 2013 Americn College of Chest Physicins. Reproduction of this rticle is prohibited without written permission from the Americn College of Chest Physicins. See online for more detils. DOI: 10.1378/chest.13-0871 ssocited with n incresed risk of pneumoni in people with COPD. 3-5 Although sthm is n independent risk fctor for invsive pneumococcl disese, 6 it is not known whether ICSs re independently ssocited with n incresed risk of pneumoni in people with sthm. We set out to exmine whether ICSs re ssocited with n incresed risk of pneumoni or lower respirtory trct infection (LRTI) in people with sthm by using computerized primry cre dt nd mtched cse-control design. Mterils nd Methods We extrcted dt from The Helth Improvement Network dtbse (www.thin-uk.com). This dtbse contins electronic medicl records of 9.1 million ptients collected from. 479 generl prctices in the United Kingdom. We identified cohort of 1788 Originl Reserch
people with recorded dignosis of sthm fter Jnury 4, 2004, ged 18 to 80 yers. We performed nested cse-control study in this cohort, with cses defined by the first-recorded dignosis of pneumoni or LRTI by previously defined Red codes for pneumoni. 7 The dte of the dignosis of pneumoni or LRTI ws considered the index dte. From the remining popultion of people with sthm, six control subjects per ptient were mtched on the bsis of sex nd ge t index dte (within 3 yers). Ptients nd control subjects with COPD were excluded from the dtset. Ethics pprovl to use dt from The Helth Improvement Network ws given by the NHS South-Est Multi-center Reserch Ethics Committee for studies tht use precollected, nonymized dt (reference number 07/H1102/103). The min hypothesis ws tht the risk of pneumoni or LRTI would be ssocited with the current use of ICS, nd for this reson, we initilly identified ll prescriptions for ICS up to 90 dys before the index dte. In the United Kingdom, the verge durtion of prescription is up to 90 dys. We grouped ICSs ccording to type s follows: beclomethsone dipropionte; fluticsone propionte, budesonide; nd combined group of ciclesonide nd mometsone furote tht ws not presented in ll nlyses becuse of smll numbers. Where the type of ICS ws chnged in the previous 90 dys, we defined exposure ccording to the lst prescription. We excluded individuls who were prescribed more thn one type of ICS inhler on the sme dy (n 5 78). To define the dose of ICS, we used the dose of drug delivered with ech inhltion becuse informtion on puffs prescribed per dy is recorded inconsistently in primry cre dtsets. To llow for different dose equivlence between drugs, budesonide ws considered equivlent to beclomethsone, nd dose multiplying fctor of 2 for fluticsone propionte ws used. For ech drug type, we defined high nd low dose bsed on cut point of 200 mg, except for fluticsone where we used cut point of 250 mg. These cut points represent the step t which long-cting b 2 -gonist is introduced ccording to UK sthm guidelines. 1 The priori confounders were smoking sttus (most recent recording) nd comorbidity s defined by the Chrlson Comorbidity Index score. 8 Other potentil confounders considered were influenz vccintion in the previous yer, number of courses of reliever inhlers in the previous yer (both totl nd seprted by short-cting b 2 -gonists nd long-cting b 2 -gonists), effect of orl corticosteroids (number of courses in the yer before the index dte), nd Townsend socioeconomic sttus score. We estimted the ssocition between ICS use nd pneumoni or LRTI by conditionl logistic regression nd vrious different exposures, including, in stepwise fshion, type of ICS, dose of ICS, nd combintion of type nd dose of ICS. We then djusted the models for priori confounders nd included other confounders if they ltered the ORs for the exposure nd outcome ssocition by. 10%. We performed number of sensitivity nlyses to determine whether the effect ws similr in different ptient popultions nd exmined whether the effect ws similr for the dignosis of pneumoni vs LRTI. We exmined the dt fter excluding ptients with bronchiectsis. We lso exmined the dt only in those who hd not chnged the type of inhler in the previous 90 dys nd included the number of short-cting nd long-cting inhlers in the pst yer seprtely in the model. All nlyses were completed with Stt 11 softwre (SttCorp LP). Results We identified 6,857 ptients with sthm nd pneumoni or LRTI nd 36,312 ge- nd sex-mtched control subjects from cohort of 359,172 people with sthm ( Tble 1 ). The men ge of the popultion ws 54 yers (rnge, 18-80 yers). Ptients with sthm nd pneumoni or LRTI were more likely to smoke, hd higher Chrlson Comorbidity Index score (ie, more comorbid illness), nd were from lower socil clss compred with control subjects ( Tble 1 ). Ptients were more likely to hve received flu vccintion in the previous yer ( Tble 1 ) nd to use more reliever inhlers nd hd more prescriptions for orl steroids in the previous yer thn control subjects ( Tble 2 ). Ptients with sthm nd pneumoni or LRTI were more likely to hve prescription for ICS in the pst 90 dys thn control subjects ( Tble 2 ). After djusting for confounders, the OR for this ssocition ws 1.24 (95% CI, 1.15-1.33) ( Tble 3 ). Steroid Type We exmined the risk of pneumoni or LRTI by ICS type. After djusting for confounders, budesonide use demonstrted smll incresed risk for pneumoni or LRTI (OR, 1.20; 95% CI, 1.06-1.35; P 5.003). There ws higher risk of pneumoni or LRTI in ptients receiving fluticsone propionte (OR, 1.64; 95% CI, 1.50-1.79; P,.001). None of the remining steroid inhlers ws ssocited with evidence of n incresed risk ( Tble 3 ). For fluticsone propionte the results were consistent if the 267 individuls who chnged their steroid inhler in the previous 90 dys were excluded. The results were lso similr when the cses were seprted by dignosis of LRTI or pneumoni; the OR for fluticsone propionte use nd risk of LRTI (n 53,410) ws 1.67 (95% CI, 1.46-1.91) nd for the risk of pneumoni (n 53,447), 1.56 (95% CI, 1.38-1.77). The results were consistent when ptients with bronchiectsis were excluded nd when the number of short- nd long-cting reliever inhlers were included seprtely in the model. Steroid Dose Of the ptients with sthm nd pneumoni or LRTI, 1.7% were prescribed the highest doses of ICSs ( 1,000 m g) compred with only 0.6% of the control subjects ( P,.001). There ws dose-response reltionship between strength of ICS nd infection risk ( P,.001 for trend) such tht fter djusting for confounders, ptients receiving the highest doses of ICSs were 2.04 (95% CI, 1.59-2.64) times more likely to hve pneumoni or LRTI ( Tble 3 ). These results remined consistent in the vrious sensitivity nlyses nd fter excluding the 267 ptients with chnge of inhler in the previous 90 dys (OR for steroid dose 1,000 m g, 1.99; 95% CI, 1.54-2.57; P,.001 for trend). When seprting the ptient cses journl.publictions.chestnet.org CHEST / 144 / 6 / DECEMBER 2013 1789
Tble 1 Demogrphic Dt for Ptients nd Control Subjects Chrcteristic Ptients (n 5 6,857) Control Subjects (n 5 36,312) Univrite OR (95% CI) Sex Mle 2,683 (39.1) 13,682 (37.7) Femle 4,174 (60.9) 22,630 (62.3) Age (men SD), y 55.5 17.8 53.7 17.9 Smoking sttus Never 2,824 (41.2) 17,784 (49.0) 1.00 Ex 2,071 (30.2) 8,967 (24.7) 1.39 (1.31-1.48) Current 1,621 (23.6) 6,890 (19.0) 1.55 (1.45-1.66) Unknown 341 (5.0) 2,671 (7.4) 0.81 (0.72-0.91) Chrlson Comorbidity Index score 0 3,132 (45.7) 21,822 (60.1) 1.00 1 2,598 (37.9) 11,375 (31.3) 1.67 (1.57-1.77) 2 717 (10.5) 2,235 (6.2) 2.46 (2.22-2.72) 3 277 (4.0) 662 (1.8) 3.19 (2.73-3.73) 4 103 (1.5) 169 (0.5) 4.81 (3.73-6.22) 5 30 (0.4) 49 (0.1) 5.13 (3.23-8.15) Townsend score 1 high 1,452 (21.2) 8,651 (23.8) 1.00 2 1,390 (20.3) 7,576 (20.9) 1.09 (1.01-1.18) 3 1,312 (19.1) 7,024 (19.3) 1.12 (1.04-1.22) 4 1,330 (19.4) 6,487 (17.9) 1.24 (1.14-1.34) 5 low 1,043 (15.2) 4,409 (12.1) 1.43 (1.32-1.57) Unknown 330 (4.8) 2,165 (6.0) 0.92 (0.81-1.05) Influenz vccintion No 3,418 (49.8) 20,939 (57.7) 1.00 Yes 3,439 (50.2) 15,373 (42.3) 1.35 (1.27-1.44) Dt re presented s No. (%) unless otherwise indicted. into pneumoni or LRTI, the effect ws stronger for pneumoni; for the highest doses of ICSs ( 1,000 mg), the OR for pneumoni (n 5 4,393) ws 2.37 (95% CI, 1.67-3.37; P,.001 for trend) compred with the OR for LRTI (n 5 4,666) (1.75; 95% CI, 1.21-2.53; P,.001 for trend). Steroid Dose nd Type Combined There ws significntly incresed risk of pneu mo ni or LRTI in ptients with sthm receiving beclom ethsone, budesonide, nd fluticsone, but only fluticsone when the nlysis ws restricted to those ged, 40 yers without bronchiectsis ( Tble 4 ). To control for sthm severity s confounder, we restricted the nlyses to ptients only receiving ICS, with those receiving low-dose beclomethsone s the bseline group. With this restricted dtset in the djusted nlyses, we found tht ptients prescribed high-dose beclomethsone were 24% more likely to hve n episode of pneumoni or LRTI (OR, 1.24; 95% CI, 0.99-1.54). Prescriptions of low- nd highdose budesonide nd low- nd high-dose fluticsone incresed the risk of infection; the lrgest effect ws found in ptients prescribed high-dose fluticsone, who hd n 87% (OR, 1.87; 95% CI, 1.63-2.15) incresed risk of hving pneumoni or LRTI compred with those prescribed low-dose beclomethsone ( Tble 5 ). Discussion To our knowledge, this study is the first to demonstrte reltionship between ICS use by type nd dose nd n incresed risk of pneumoni or LRTI in sthm. Ptients with sthm nd pneumoni or LRTI were more likely to be receiving high-dose ICS. These results re consistent for dignosis of both LRTI nd pneumoni nd re similr fter number of different sensitivity nlyses, including controlling for sthm severity. People with coexistent sthm nd COPD were excluded from the nlysis, nd the results remined consistent when restricted to people ged, 40 yers, negting ny residul confounding effect of COPD. There ws cler dose-response reltionship, with higher prescribed doses of ICS being ssocited with higher risk of infection. The only ICS ssocited with n incresed risk of LRTI or pneumoni cross ll nlyses ws fluticsone propionte. A mjor strength of this study is the popultion size. To our knowledge, it is the lrgest to dte, with 6,857 cses of pneumoni or LRTI identified. The ssocitions described remined consistent fter number of sensitivity nlyses, including n exmintion of effect in different ge popultions. The use of dt from primry cre records hs limittions. We cnnot confirm complince with ICS use or ssess device type. Furthermore, individuls my 1790 Originl Reserch
Tble 2 Study Popultion Mediction Use for Ptients nd Control Subjects Mediction Use Ptients (n 5 6,857) Control Subjects (n 5 36,312) Univrite OR (95% CI) Reliever use in the pst yer 0 2,002 (29.2) 16,547 (45.6) 1.00 1-2 1,248 (18.2) 7,387 (20.3) 1.43 (1.32-1.54) 3-6 1,338 (19.5) 5,938 (16.4) 1.91 (1.76-2.06) 7-12 1,194 (17.4) 3,888 (10.7) 2.59 (2.38-2.81) 13 1,075 (15.7) 2,552 (7.0) 3.53 (3.23-3.85) Orl steroid courses in the pst yer 0 5,061 (73.8) 32,324 (89.0) 1.00 1 554 (8.1) 1,499 (4.1) 2.38 (2.14-2.63) 2-3 494 (7.2) 1,181 (3.3) 2.68 (2.40-2.99) 4-8 388 (5.7) 737 (2.0) 3.35 (2.95-3.82) 9 360 (5.3) 571 (1.6) 3.92 (3.41-4.50) Most recent ICSs in the pst 90 d No steroids 3,432 (50.1) 24,143 (66.5) 1.00 Beclomethsone 1,214 (17.7) 5,743 (15.8) 1.46 (1.36-1.57) Budesonide 536 (7.8) 2,065 (5.7) 1.82 (1.64-2.01) Ciclesonide/mometsone 6 (0.1) 44 (0.1) 0.95 (0.41-2.25) Fluticsone propionte 1,669 (24.3) 4317 (11.9) 2.71 (2.53-2.91) ICS dose, m g 0 3,432 (50.1) 24,143 (66.5) 1.00, 200 1,227 (17.9) 5,592 (15.4) 1.53 (1.43-1.65) 200-249 591 (8.6) 2,400 (6.6) 1.71 (1.55-1.89) 250-399 569 (8.3) 1,896 (5.2) 2.07 (1.87-2.29) 400-499 132 (1.9) 441 (1.2) 2.07 (1.69-2.52) 500-999 778 (11.4) 1,626 (4.5) 3.36 (3.06-3.70) 1,000 117 (1.7) 200 (0.6) 4.19 (3.31-5.30) Dt re presented s No. (%) unless otherwise indicted. ICS 5 inhled corticosteroid. Ws not ble to determine strength in 25 individuls. be using drug tht ws prescribed before the 90-dy index period; however, this misclssifiction would bis the results towrd the null hypothesis. The present study lso shres similr limittion to studies of pneumoni in COPD, nmely tht pneumoni ws not confirmed rdiogrphiclly; however, the dignosis of pneumoni nd LRTI hs been previously shown to be resonbly ccurte. 7 Tble 3 Assocition Between Dose nd Type of ICS Use nd Risk of Pneumoni or LRTI (n 5 43,169) ICS Dose nd Type OR Adjusted OR 95% CI Adjusted OR b 95% CI Any ICS use in the pst 90 d No 1.00 1.00 Yes 1.96 1.24 1.15-1.33 1.24 1.07-1.44 ICS use in the pst 90 d No steroids 1.00 1.00 1.00 Beclomethsone 1.46 1.09 1.00-1.18 1.13 0.95-1.34 Budesonide 1.82 1.20 1.06-1.35 1.13 0.87-1.47 Ciclesonide/mometsone 0.95 0.71 0.30-1.69 Fluticsone 2.71 1.64 1.50-1.79 1.58 1.29-1.93 ICS dose, c m g 0 1.00 1.00 1.00, 200 1.53 1.11 1.02-1.21 1.07 0.98-1.38 200-249 1.71 1.16 1.04-1.30 1.02 0.80-1.30 250-399 2.07 1.32 1.18-1.49 1.43 1.09-1.89 400-499 2.07 1.16 0.93-1.43 1.18 0.62-1.94 500-999 3.36 1.83 1.63-2.05 2.00 1.52-2.64. 1,000 4.19 2.04 1.59-2.63 2.51 1.05-5.99 Individuls tking ciclesonide/mometsone were excluded from the nlyses becuse of smll numbers. LRTI 5 lower respirtory trct infection. See Tble 2 legend for expnsion of other bbrevition. Adjusted for number of relievers in the pst yer, Chrlson Comorbidity Index score, smoking, socil clss, nd use of orl steroids in the pst yer. b Restricted to individuls ged, 40 y who did not hve bronchiectsis nd did not chnge steroid in the previous 90 d nd djusted for number of relievers in the pst yer, Chrlson Comorbidity Index score, smoking, socil clss, nd use of orl steroids in the pst yer. c Ws not ble to determine strength in 25 individuls. journl.publictions.chestnet.org CHEST / 144 / 6 / DECEMBER 2013 1791
Tble 4 Anlysis Combining Type of ICS nd Dose With Risk of Pneumoni or LRTI (n 5 43,095) ICS Use in the Pst 90 d Cses, No. (%) Control Subjects No. (%) OR Adjusted OR 95% CI Adjusted OR b 95% CI No steroids 3,432 (50.1) 24,143 (66.5) 1.00 1.00 1.00 Beclomethsone low dose 200 m g 1,031 (15.0) 5,101 (14.1) 1.41 1.07 0.98-1.16 1.12 0.94-1.33 Beclomethsone high dose. 200 m g 183 (2.7) 642 (1.8) 1.86 1.24 1.03-1.49 1.33 0.76-2.32 Budesonide low dose 200 m g 417 (6.1) 1,671 (4.6) 1.75 1.18 1.04-1.35 1.14 0.86-1.49 Budesonide high dose. 200 m g 119 (1.7) 394 (1.1) 2.10 1.16 0.93-1.46 1.19 0.66-2.16 Fluticsone low dose 250 m g 774 (11.3) 2,491 (6.9) 2.17 1.39 1.25-1.55 1.33 1.04-1.68 Fluticsone high dose. 250 m g 895 (13.1) 1,826 (5.0) 3.45 1.89 1.69-2.11 2.06 1.57-2.69 Individuls tking ciclesonide/mometsone were excluded from the nlyses becuse of smll numbers. See Tble 2 nd 3 legends for expnsion of bbrevitions. Adjusted for number of relievers in the pst yer, Chrlson Comorbidity Index score, smoking, socil clss, nd use of orl steroids in the pst yer. b Restricted to individuls ged, 40 y who did not hve bronchiectsis nd did not chnge steroid in the previous 90 d nd djusted for number of relievers in the pst yer, Chrlson Comorbidity Index score, smoking, socil clss, nd use of orl steroids in the pst yer. Lower lung function 9 is known to be ssocited with n incresed risk of pneumoni, with more severe sthm nd lower lung function crrying the highest risk; however, the incresed risk of pneumoni or LRTI remined despite correction for ge, smoking sttus, nd Chrlson Comorbidity Index score, reducing the effect of these limittions. We lso corrected for the number of orl steroids prescribed in the pst yer, which is ssocited with both disese severity nd decline in lung function, 10 nd for use of reliever inhlers, which is ssocited with n incresed risk of n dverse sthm outcome. 11 Becuse sthm severity is n independent risk fctor for pneumoni, 6,12 we ttempted to control for sthm severity s confounder by restricting the nly sis. We used low-dose beclomethsone s the bseline nd nlyzed only those ptients who received ICSs. The nlysis did not substntilly chnge the results, nd the dose-response trend remined unffected; however, we cknowledge tht it is impossible to fully remove severity s confounder. The possibility tht ICS my be ssocited with n incresed risk of pneumoni in obstructive lung disese greed with the results of the TORCH (Towrd Revolution in COPD Helth) study, 5 which exm- ined the potentil benefit of fluticsone propionte nd slmeterol on mortlity in COPD. The TORCH study reported 19% 3-yer rte of pneumoni in ptients receiving fluticsone t 1,000 m g/d, corresponding to significnt 1.6-fold increse over plcebo. There hve been few studies in sthm. O Byrne nd collegues 9 ssessed the risk of pneumoni in retrospective nlysis of budesonide use in sthm. The primry dtset ws ll double-blind, plcebocontrolled trils lsting t lest 3 months tht studied budesonide (26 trils, n 5 9,067 for budesonide nd n 5 5,926 for the comprtor); 62 cses of pneumoni were reported s either n dverse or serious dverse event for budesonide compred with 82 for the comprtor. In the primry dtset, the rte of pneumoni dverse events ws 0.5% (10 of 1,000 ptient-yers) for budesonide nd 1.2% (19.3 of 1,000 ptient-yers) for plcebo. The occurrence of pneumoni serious dverse events ws 0.15% for budesonide nd 0.13% for plcebo, resulting in hzrd rtio of 1.29 (95% CI, 0.53-3.12), which is similr to the present OR of 1.10 for budesonide nd risk of pneumoni or LRTI. Overll, there ws no incresed risk with higher budesonide doses or ny difference between budesonide Tble 5 Restricted Anlysis of Assocition Between Type nd Dose of ICS in Ptients With Prescription for Steroids in the Pst 90 d nd Risk of Pneumoni or LRTI, With Low-Dose Beclomethsone s Control (n 5 9,324) ICS Use in the Pst 90 d OR Adjusted OR 95% CI Adjusted OR b 95% CI Beclomethsone low dose 200 m g 1.00 1.00 1.00 Beclomethsone high dose. 200 m g 1.38 1.24 0.99-1.54 1.25 0.59-2.67 Budesonide low dose 200 m g 1.27 1.16 0.99-1.35 1.00 0.69-1.45 Budesonide high dose. 200 m g 1.64 1.26 0.96-1.65 1.45 0.65-3.23 Fluticsone low dose 250 m g 1.56 1.36 1.19-1.56 1.39 1.00-1.94 Fluticsone high dose. 250 m g 2.44 1.87 1.63-2.15 2.10 1.43-3.06 Individuls tking ciclesonide/mometsone were excluded from the nlyses becuse of smll numbers. See Tble 2 nd 3 legends for expnsion of bbrevitions. Adjusted for number of relievers in the lst yer, Chrlson Comorbidity Index score, smoking, socil clss, nd use of orl steroids in the pst yer. b Restricted to individuls ged, 40 y who did not hve bronchiectsis nd did not chnge steroid in the previous 90 d nd djusted for number of relievers in the pst yer, Chrlson Comorbidity Index score, smoking, socil clss, nd use of orl steroids in the pst yer. 1792 Originl Reserch
nd fluticsone propionte; however, pneumoni ws not the primry end point in the trils. Although the present dt show possible doseresponse reltionship, with higher doses of ICS being ssocited with n incresed risk of pneumoni or LRTI, the biologic mechnism explining the ssocition between risk of infection in sthm nd ICS is not cler. There is contrdictory evidence on the bil ity of ICS to influence bcteril numbers. Mouse models 13,14 nd in vitro studies of humn bronchil epithelil cells 15,16 hve demonstrted tht ICSs reduce bcteril lod or invsion, wheres steroid tretment cn rectivte chronic infection by typicl bcteri in vitro nd in niml models. 17,18 People receiving ICS for chronic respirtory disese hve lso been found to hve n incresed risk of nontuberculous mycobcteril infection 19,20 nd ltered irwy microbiot compred with norml subjects. 21 Studying the lung microbiot to delinete whether sthm itself or tretment with ICS lters the lung microbiome my help to nswer questions of custion. Conclusions We show tht ICSs re ssocited with n incresed risk of pneumoni nd LRTI in people with sthm. The results suggest tht the dose of ICS prescribed should be kept to the minimum necessry to tret symptoms nd tht the dose should be stepped down if the ptient is well controlled. Furthermore, prescribers should consider the possibility tht infection rther thn underlying sthm my be driving symptoms before incresing the ICS dose. This my be importnt before prescribers increse the ICS dose to tret recurring symptoms. A review nd met-nlysis of sthm studies involving ICS nd n explortion of the effect of ICS on the lung microbiot re now required. Acknowledgments Author contributions: Drs McKeever nd Shw hd full ccess to ll the dt in the study nd tke responsibility for the integrity of the dt nd the ccurcy of the dt nlysis. Dr McKeever: contributed to the study design, nlysis, nd review of the mnuscript. Dr Hrrison: contributed to the study design nd review of the mnuscript. Dr Hubbrd: contributed to the study design, nlysis, nd review of the mnuscript. Dr Shw: contributed to the study design nd writing of the first drft of the mnuscript. Finncil/nonfinncil disclosures: The uthors hve reported to CHEST the following conflicts of interest: Dr Hrrison hs received finncil support from Boehringer Ingelheim GmbH nd GlxoSmithKline to ttend scientific meetings nd honorrium pyments for dvisory bords from GlxoSmithKline, Npp Phrmceuticls Limited, nd Boehringer Ingelheim GmbH. Dr Hubbrd hs two Medicl Reserch Council grnts to investigte the cuses nd nturl history of lung fibrosis nd Roy Cstle clinicl fellowship wrd to study cre pthwys for people with lung cncer. Dr Hubbrd is the current GlxoSmithKline/British Lung Foundtion Professor of Respirtory Epidemiology. GlxoSmithKline hs cofunded cohort study of lung biomrkers for people with idiopthic pulmonry fibrosis for which Dr Hubbrd is copplicnt. Drs McKeever nd Shw hve reported tht no potentil conflicts of interest exist with ny compnies/orgniztions whose products or services my be discussed in this rticle. Other contributions: The uthors thnk Christopher Smith, PhD, for help in the preprtion of this mnuscript. References 1. British Thorcic Society nd Scottish Intercollegite Guidelines Network. British Guideline on the Mngement of Asthm. A Ntionl Clinicl Guideline. Edinburgh, Scotlnd; 2012. 2. Ntionl Institute for Helth nd Clinicl Excellence. Chronic Obstructive Pulmonry Disese: Mngement of Chronic Obstructive Pulmonry Disese in Adults in Primry nd Secondry Cre. London, Englnd; 2008. 3. Anzueto A, Ferguson GT, Feldmn G, et l. Effect of fluticsone propionte/slmeterol (250/50) on COPD excerbtions nd impct on ptient outcomes. COPD. 2009 ;6(5):320-329. 4. Rodrigo GJ, Cstro-Rodriguez JA, Plz V. Sfety nd efficcy of combined long-cting bet-gonists nd inhled corticosteroids vs long-cting bet-gonists monotherpy for stble COPD: systemtic review. Chest. 2009 ;136(4):1029-1038. 5. Crim C, Clverley PM, Anderson JA, et l. Pneumoni risk in COPD ptients receiving inhled corticosteroids lone or in combintion: TORCH study results. Eur Respir J. 2009 ; 34 (3):641-647. 6. Tlbot TR, Hrtert TV, Mitchel E, et l. Asthm s risk fctor for invsive pneumococcl disese. N Engl J Med. 2005 ;352(20):2082-2090. 7. Myles PR, McKeever TM, Pogson Z, Smith CJ, Hubbrd RB. The incidence of pneumoni using dt from computerized generl prctice dtbse. Epidemiol Infect. 2009 ;137(5): 709-716. 8. Chrlson ME, Pompei P, Ales KL, McKenzie CR. A new method of clssifying prognostic comorbidity in longitudinl studies: development nd vlidtion. J Chronic Dis. 1987 ; 40 (5):373-383. 9. O Byrne PM, Pedersen S, Crlsson LG, et l. Risks of pneumoni in ptients with sthm tking inhled corticosteroids. Am J Respir Crit Cre Med. 2011 ;183(5):589-595. 10. Bi TR, Vonk JM, Postm DS, Boezen HM. Severe excerbtions predict excess lung function decline in sthm. Eur Respir J. 2007 ;30(3):452-456. 11. Suiss S, Ernst P, Boivin JF, et l. A cohort nlysis of excess mortlity in sthm nd the use of inhled bet-gonists. Am J Respir Crit Cre Med. 1994 ;149(3 pt 1):604-610. 12. Klemets P, Lyytikäinen O, Ruutu P, et l. Risk of invsive pneumococcl infections mong working ge dults with sthm. Thorx. 2010 ;65(8):698-702. 13. Mrtin RJ, Chu HW, Honour JM, Hrbeck RJ. Airwy inflmmtion nd bronchil hyperresponsiveness fter Mycoplsm pneumonie infection in murine model. Am J Respir Cell Mol Biol. 2001 ;24(5):577-582. 14. Hnsbro PM, Begley KW, Horvt JC, Gibson PG. Role of typicl bcteril infection of the lung in predisposition/protection of sthm. Phrmcol Ther. 2004 ;101(3):193-210. 15. Brbier M, Agustí A, Albertí S. Fluticsone propionte reduces bcteril irwy epithelil invsion. Eur Respir J. 2008 ;32(5): 1283-1288. 16. Dowling RB, Johnson M, Cole PJ, Wilson R. Effect of fluticsone propionte nd slmeterol on Pseudomons eruginos infection of the respirtory mucos in vitro. Eur Respir J. 1999 ;14(2):363-369. journl.publictions.chestnet.org CHEST / 144 / 6 / DECEMBER 2013 1793
17. Blott MH, DeKruyff RH, Umetsu DT. Corticosteroids inhibit IL-12 production in humn monocytes nd enhnce their cpcity to induce IL-4 synthesis in CD4 1 lymphocytes. J Immunol. 1997 ;158 (12 ):5589-5595. 18. Litinen K, Luril AL, Leinonen M, Sikku P. Rectivtion of Chlmydi pneumonie infection in mice by cortisone tretment. Infect Immun. 1996 ;64 (4 ):1488-1490. 19. Hojo M, Iikur M, Hirno S, Sugiym H, Kobyshi N, Kudo K. Incresed risk of nontuberculous mycobcteril infection in sthmtic ptients using long-term inhled corticosteroid therpy. Respirology. 2012 ;17 (1 ):185-190. 20. Andréjk C, Nielsen R, Thomsen VØ, Duhut P, Sørensen HT, Thomsen RW. Chronic respirtory disese, inhled corticosteroids nd risk of non-tuberculous mycobcteriosis. Thorx. 2013 ;68(3):256-262. 21. Hilty M, Burke C, Pedro H, et l. Disordered microbil communities in sthmtic irwys. PLoS ONE. 2010 ; 5 ( 1 ): e8578. 1794 Originl Reserch