REVIEW ARTICLE

Long-term Use of Inhaled Corticosteroids

and the Risk of Pneumonia in Chronic
Obstructive Pulmonary Disease
A Meta-analysis
Sonal Singh, MD, MPH; Aman V. Amin, MD; Yoon K. Loke, MD

Background: Recent studies have suggested a possible
association between pneumonia and the use of inhaled
corticosteroids. We aimed to ascertain the risk of pneu-
monia with long-term inhaled corticosteroid use among
patients with chronic obstructive pulmonary disease
(COPD).
Methods: We performed systematic searches with no
date restrictions through June 30, 2008, of MEDLINE,
EMBASE, the Cochrane Database of Systematic Re-
views, regulatory documents, and trial registries. We in-
cluded randomized controlled trials of any inhaled cor-
ticosteroid vs a control treatment for COPD, with at least
24 weeks of follow-up and reporting of pneumonia as an
adverse event. Outcomes evaluated included any pneu-
monia, serious pneumonia, pneumonia-related mortal-
ity, and overall mortality.
Results: Eighteen randomized controlled trials (n=16 996)
with 24 to 156 weeks of follow-up were included after a
detailed screening of 97 articles. Inhaled corticosteroids were
associated with a significantly increased risk of any pneu-
monia (relative risk [RR], 1.60; 95% confidence interval
[CI], 1.33-1.92 [P.001]; I2 =16%) and serious pneumo-
nia (1.71; 1.46-1.99 [P.001]; I2 =0%) but without a sig-
nificantly increased risk of pneumonia-related mortality
(1.27; 0.80-2.03 [P=.31]; I2 =0%) or overall mortality (0.96;
0.86-1.08 [P=.51]; I2 =0%). Inhaled corticosteroids were
associated with a significantly increased risk of serious pneu-
monia when compared with placebo (RR, 1.81; 95% CI,
1.44-2.29 [P.001]) or when the combination of inhaled
corticosteroids and long-acting -agonists was compared
with long-acting -agonists (1.68; 1.20-2.34 [P=.002]).
Conclusion: Among patients with COPD, inhaled cor-
ticosteroid use for at least 24 weeks is associated with a
significantly increased risk of serious pneumonia, with-
out a significantly increased risk of death.
Arch Intern Med. 2009;169(3):219-229

Author Affiliations:
Department of Medicine, Wake
Forest University School of
Medicine, Winston-Salem,
North Carolina (Drs Singh and
Amin); Clinical Pharmacology,
School of Medicine, Health
Policy and Practice, University
of East Anglia, Norwich,
England (Dr Loke).
I
NHALED CORTICOSTEROIDS SUCH AS
fluticasone propionate, budes-
onide, and beclomethasone di-
propionate are widely used for the
treatment of chronic obstruc-
tive pulmonary disease (COPD).1,2 They
are recommended in combination with
long-acting bronchodilators to reduce the
frequency of exacerbations in sympto-
matic patients with a forced expiratory vol-
ume in the first second of expiration of less
than 50% predicted (stage III, severe
COPD; stage IV, very severe COPD) and
repeated exacerbations.1
The known adverse effects associated
with inhaled corticosteroid use include the
development of oropharyngeal candidia-
sis,2 cataracts,3 and fractures.4 In a recent
trial, patients with COPD who received in-
halers containing fluticasone had a higher
probability of pneumonia.5 Another re-
cent observational study raised the possi-
bility of an association between the use of
inhaled corticosteroids and the risk of hos-
pitalization for pneumonia and mortality in
elderly patients with COPD.6 It is unclear
whether the association with pneumonia is
specific to the inhaled corticosteroid com-
ponent or occurs because of an interaction
with the long-acting -agonist component
of the combination inhaler. The magni-
tude and the strength of this potential as-
sociation, if any, are also unknown. Lim-
ited information is currently available to
clinicians on the long-term safety of in-
haled corticosteroids in COPD.1
Our primary objectives were to system-
atically review the current evidence of the
risks of pneumonia with long-term use of
inhaled corticosteroids in patients with
COPD. We also aimed to ascertain the risk
of pneumonia-related mortality and over-
all mortality in these trials as a secondary
objective.

(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 3), FEB 9, 2009 WWW.ARCHINTERNMED.COM
219
Downloaded from www.archinternmed.com on June 14, 2010
2009 American Medical Association. All rights reserved.

651 Potentially relevant citations identified by
initial search and titles and abstracts
screened against inclusion criteria
554 Excluded on basis of title and abstract for not
fulfilling inclusion criteria on basis of design,
intervention, population, or duration of study
404 Not RCTs of inhaled corticosteroid use
for >24 wk in COPD
113 Duplicates or multiple reports
37 Review articles
97 Full-text RCTs screened in detail for inclusion
67 Excluded for not meeting inclusion criteria
20 Crossover trials
47 Wrong comparators
30 Long-term (>24 wk) RCTs of inhaled
corticosteroid use in COPD selected for
detailed evaluation of pneumonia
12 RCTs excluded because of no report on
pneumonia adverse events
18 Long-term RCTs of inhaled corticosteroid use
in COPD included in the analyses of pneumonia
Figure 1. Flowchart describing study selection and excluded studies. COPD indicates chronic obstructive
pulmonary disease; RCT, randomized controlled trial.
METHODS PubMed and EMBASE with the clinical
trial filters using the search terms flutica-
sone or budesonide or beclomethasone or be-
ELIGIBILITY CRITERIA clomethasone and chronic and obstructive
with no language or date restrictions
Our specific inclusion criteria were through June 30, 2008. Published or un-
(1) study design consisting of a ran- published trials were retrieved from the
domized controlled trial (RCT) for Cochrane Database of Systematic Re-
any inhaled corticosteroid (flutica- views, Web sites of the US Food and Drug
sone, beclomethasone, or budesonide) Administration and European regula-
with at least 24 weeks of follow-up; tory authorities, the manufacturers prod-
(2) study participants with COPD; (3) uct information sheets, and the manufac-
an inhaled corticosteroid as the inter- turers clinical trials register of fluticasone
vention drug vs a control treatment, and beclomethasone (GlaxoSmithKline)7
in which the comparison groups con- and budesonide (AstraZeneca).8 We
sisted of inhaled corticosteroids vs checked the included and excluded stud-
placebo or inhaled corticosteroid in ies lists from systematic reviews and meta-
combination with a long-acting analyses of inhaled corticosteroids in
-agonist vs a long-acting -agonist; COPD9-14 and the bibliographies of in-
and (4) data on the incidence of pneu- cluded studies and used the Web of Sci-
monia (including 0 events) as an ence citation index to identify relevant
adverse event. articles.
The analysis was restricted to RCTs
of more than 24 weeks duration to
evaluate the risk of pneumonia associ- STUDY SELECTION
ated with long-term use of inhaled cor-
ticosteroids. Randomized controlled Two reviewers (A.V.A. and Y.K.L.) in-
trials of inhaled corticosteroids in pa- dependently and in duplicate scanned all
tients with asthma were ineligible for in- titles and abstracts that indicated the
clusion. Observational studies suscep- study was an RCT evaluating the use of
tible to confounding and channeling bias inhaled corticosteroids by patients with
were also excluded. COPD. After obtaining full reports of po-
tentially relevant trials, the same review-
SEARCH STRATEGY ers independently assessed eligibility
from full-text articles. Disagreements re-
Two reviewers (A.V.A. and Y.K.L.) inde- garding eligibility were resolved with a
pendently and in duplicate searched third reviewer through consensus.
(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 3), FEB 9, 2009
220
Downloaded from www.archinternmed.com on June 14, 2010
2009 American Medical Association. All rights reserved.
STUDY CHARACTERISTICS
A standard protocol was used to record
the location and duration of the trial, the
criteria used to diagnose COPD in par-
ticipants, the primary outcome mea-
sure, the dose and frequency of inhaled
corticosteroid and control interven-
tions, the mean age and sex of partici-
pants, the severity of COPD in the par-
ticipants as mean predicted forced
expiratory volume in the first second of
expiration, previous inhaled corticoste-
roid use, and the percentage of current
smokers enrolled.
QUALITY ASSESSMENT
The Cochrane Toolkit was used for the
assessment of bias in evaluating each trial
for the reporting of sequence genera-
tion, allocation concealment, the use of
blinding of participants and personnel,
and information on loss to follow-up.15
Information was extracted on addi-
tional potential sources of bias such as
withdrawal rates. The frequency and type
of adverse event monitoring during the
follow-up period were evaluated as rec-
ommended in the Cochrane Handbook
for Systematic Reviews of Interven-
tions15 to determine the strength of ad-
verse event monitoring.
OUTCOME MEASURES
The end points of any pneumonia
(pneumonia as an adverse event
irrespective of severity) and serious
pneumonia (pneumonia as a serious
adverse event) were prespecified as
the primary outcome measures.
Serious adverse events are life-
threatening, require hospitalization,
or lead to significant disability
or death.16 The end points of
pneumonia-related mortality (any
death in which pneumonia was the
cause of death) and overall mortality
were prespecified as secondary out-
come measures. The primary outcome
of any pneumonia was inclusive of
pneumonia reported as a serious
adverse event, and the primary out-
come of serious pneumonia was inclu-
sive of pneumonia-related mortality.
DATA EXTRACTION
Two reviewers (A.V.A. and Y.K.L.) in-
dependently and separately extracted
data (including 0 events) on any pneu-
monia among listings of adverse events,
serious pneumonia among listings of
pneumonia terms reported as serious ad-
verse events, and pneumonia-related
mortality and overall mortality from
WWW.ARCHINTERNMED.COM
 Table 1. Characteristics of RCTs of Inhaled Corticosteroid Use Included in the Analysis of Pneumonia Adverse Events

% of Subjects
Treatment Age, Mean (SD)
Duration, COPD Primary Male, Mean Predicted Prior Current
Source Location wk Criteria, % Outcome Drug % (SD), y FEV1, % ICS Use Smokers
Aaron et al,20 27 Centers in 52 FEV1:FVC COPD SFC, 50/500 g BID 57.9 67.5 (8.9) 39.4 (11.9) 70.8 32.4
2007 a Canada ratio, 70 exacerbation Sal, 50 g 57.4 67.6 (8.2) 38.0 (13.1) 78.8 24.3
Burge et al,21 18 UK hospitals 156 FEV1:FVC Decline in FEV1 Flu, 500 g BID 75.0 63.7 (7.1) 50.3 (14.9) 51.1 36.4
2000 ratio, 70 Placebo 74.2 63.8 (7.1) 15.0 (14.9) 57.0 39.2
Calverley 196 Centers in 52 ERS FEV1 SFC, 50/500 g BID 75 62.7 (8.7) 44.8 (14.7) 50 52
et al,22 25 countries Sal, 50 g BID 70 63.2 (8.6) 44.3 (13.8) 49 51
2003 Flu, 500 g BID 70 63.5 (8.5) 45.0 (13.6) 54 53
Placebo 75 63.4 (8.6) 44.2 (13.7) 52 47
Calverley 109 Centers in 52 GOLD FEV1 and For, 9 gBud, 400 g 78 64 (NA) 36 (10) 47 33
et al,23 15 countries HRQOL BID
2003 For, 9 g BID 75 63 (NA) 36 (10) 48 36
Bud, 400 g BID 74 64 (NA) 36 (10) 51 39
Placebo 75 65 (NA) 36 (10) 46 30
Calverley 44 Centers in 156 ERS Mortality SFC, 50/500 g BID 75 65 (8.3) 44.3 (12.3) 47 43
et al,5 42 countries Sal, 50 g 76 65.1 (8.2) 43.6 (12.6) 45 43
2007 Flu, 500 g BID 75 65 (8.4) 44.1 (12.3) 47 43
Placebo 76 65 (8.2) 44.1 (12.3) 51 43
Ferguson 94 Centers in 52 ATS Rate of SFC, 50/250 g BID 58.3 64.9 (9.0) 39.8 (13.9) 15 40
et al,24 North America exacerbations Sal, 50 g BID 52.0 65.0 (9.1) 50.6 (15.4) 18 38
2008
FLTA3025,25 55 US centers 24 ATS FEV1 Flu, 500 g BID 66 63.3 (10) 1301 (500) b NA NA
2005 Flu, 250 g BID 72 65.2 (8.7) 1240 (486) b NA NA
Placebo 70 64.8 (9.5) 1221 (502) b NA NA
Hanania 76 US centers 24 ATS FEV1 SFC, 50/250 g BID 61 63 (NA) 41 (11) 23 43
et al,26 Sal, 50 g BID 58 64 (NA) 42 (12) 20 51
2003 Flu, 250 g BID 66 63 (NA) 42 (11) 28 48
Placebo 58 65 (NA) 42 (12) 30 47
Kardos 95 Centers in 52 GOLD COPD SFC, 50/500 g BID 74.0 63.8 (8.3) 40.4 (8.9) 40.6 49.7
et al,27 Germany exacerbations Sal, 50 g BID 77.6 64 (8.2) 40.3 (8.5) 44.4 49.9
2007
Mahler Multicenter US 24 ATS FEV1 and TDI SFC, 50/500 g BID 62 61.9 (NA) 41 (NA) 28 46
et al,28 trial Sal, 50 g BID 64 63.5 (NA) 40 (NA) 31 46
2002 Flu, 500 g BID 61 64.4 (NA) 41 (NA) 25 46
Placebo 75 64 (NA) 41 (NA) 18 54
Paggiaro 13 European 24 ERS Exacerbations Flu, 500 g BID 99 62 (NA) 59 (18) NA 49
et al,29 centers Placebo 78 64 (NA) 55 (17) NA 49
1998
SCO100250,30 98 US and 52 FEV1:FVC Rate of SFC, 50/250 g BID 51 65.4 (NA) 50 c NA NA
2008 Canadian ratio, 70 exacerbations Sal, 50 g BID 57 65.3 (NA) 50 c NA NA
centers
SCO100470,31 135 Centers in 24 GOLD FEV1 and TDI SFC, 50/250 g BID 78.3 63.5 (9.3) 1654 (459) b NA 42
2006 Europe and Sal, 50 g BID 77.2 63.7 (9.0) 1681 (465) b NA 44
Asia-Pacific
SCO40041,32 31 US centers 156 GOLD Bone mineral SFC, 50/250 g BID 59.7 65.4 (8.36) 70 NA NA
2008 density Sal, 50 g BID 62.7 65.9 (9.52) 70
SFCT01/ 49 Centers in 52 FEV1:FVC Time to Flu, 500 g BID 83.9 64.6 (8.7) NA NA NA
SCO30002,33 Italy and ratio, 88 exacerbations Placebo 80.0 65.7 (9.0) NA NA NA
2005 Poland
van der Valk Pulmonary 26 ATS Exacerbations Flu, 500 g BID 85.4 64.1 (6.8) 57.5 (14.1) 86.2 22.0
et al,34 clinics and HRQOL Placebo 83.5 64 (6.7) 56.1 (14.8) 80.2 33
2002
Vestbo et al,35 Community in 156 FEV1:FVC Rate of FEV1 Bud, 400 g BID 58.6 59.0 (8.3) 86.2 (20.6) NA 75.9
1999 Denmark ratio, 70 decline Placebo 62.1 59.1 (9.7) 86.9 (21.1) NA 77.2
Wouters 39 Centers in the 52 FEV1:FVC Rate of FEV1 SFC, 50/500 g BID 73 63 (7.9) 47.4 (13.9) 85 39
et al,36 Netherlands ratio, 88 decline Sal, 50 g BID 75 64 (7.7) 48.2 (12.9) 87 35
2005

Abbreviations: ATS, American Thoracic Society; BID, twice daily; Bud, budesonide; COPD, chronic obstructive pulmonary disease; ERS, European Respiratory
Society; FEV1, forced expiratory volume in the first second of expiration; Flu, fluticasone propionate; For, formoterol; FVC, forced vital capacity; GOLD, Global Initiative
for Chronic Obstructive Lung Disease; HRQOL, health-related quality of life; NA, not available; RCT, randomized controlled trial; Sal, salmeterol xinafoate; SFC,
combination of salmeterol and fluticasone; TDI, transitional dyspnea index.
a Indicates treatment with tiotropium bromide, 18 g/d, in both arms and an additional tiotropium arm. Influenza vaccination was only reported for Aaron et al20 at
75.5% and unavailable for the other studies.
b Reported mean FEV in milliliters as percentage of predicted unavailable.
1
c Indicates FEV less than 50% predicted (exact mean and SD data were unavailable).
1

safety data collected as part of serious
adverse event reporting for each study.
Data in the clinical trials register and
the regulatory documents were recon-
ciled with that of the published journal
article when possible. If there were mul-
tiple reports for a particular study, data
from the most recent version were ex-
tracted. When specific aspects of the data
required clarification, the authors were
contacted.
Two reviewers (A.V.A. and Y.K.L.)
were independently involved in all stages
of study selection, data extraction, and
quality assessment. All discrepancies
were resolved with 100% agreement af-
ter rechecking the source papers, fur-
ther discussion among the reviewers, and
consultation with a third reviewer (S.S.),
with full consensus obtained before
drafting the article.

(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 3), FEB 9, 2009 WWW.ARCHINTERNMED.COM
221
Downloaded from www.archinternmed.com on June 14, 2010
2009 American Medical Association. All rights reserved.
 Table 2. Quality Assessment of Included RCTs of Inhaled Corticosteroids in COPD a

No. (%)

Sequence Allocation Drug (No. of Withdrawal Lost to

Source Generation Concealment Monitoring of AEs Subjects) Rates Follow-up
Aaron et al,20 2007 Adequate, central Adequate Captured through monthly telephone interviews SFC (145) 15 (10.3) 2 (1.4)
allocation and checklist; pneumonia recorded only as Sal (148) 20 (13.5) 2 (1.4)
SAE leading to hospitalization or death
Burge et al,21 2000 Adequate, computer Adequate AEs and SAEs recorded throughout study Flu (372) 160 (43.0) 16 (4.3)
generated, Placebo (370) 195 (52.7) 18 (4.9)
stratified by center
Calverley et al,22 2003 Adequate computer Adequate AE or SAE occurring during therapy SFC (358) 89 (24.9) 8 (2.2)
generated Sal (372) 119 (32.0) 8 (2.2)
Flu (374) 108 (29.0) 8 (2.1)
Placebo (361) 140 (38.8) 6 (1.7)
Calverley et al,23 2003 Unclear Unclear AEs recorded at 1, 2, 3, 6, 9, and 12 mo of For/Bud (254) 74 (29.1) 4 (1.6)
treatment For (255) 111 (43.5) 4 (1.6)
Bud (257) 102 (40.0) 4 (1.6)
Placebo (256) 106 (41.4) 6 (2.3)
Calverley et al,5 2007 b Adequate, schedule Adequate AEs reviewed at each visit; no prospective SFC (1533) 522 (34.1) 29 (1.9)
generated by confirmation by radiographs; pneumonia Sal (1521) 561 (36.9) 15 (1.0)
system for central recorded as subset of exacerbations Flu (1534) 587 (38.3) 24 (1.6)
allocation Placebo (1524) 673 (44.2) 21 (1.4)
Ferguson et al,24 2008 Unclear Unclear AEs collected at study start and end FSC (394) 117 (29.7) 10 (2.5)
Sal (388) 149 (38.4) 10 (2.6)
FLTA3025,25 2005 Unclear Unclear AEs and SAEs recorded at each visit Flu (434) 147 (33.9) NA
Placebo (206) 79 (38.3) NA
Hanania et al,26 2003 Unclear Unclear AE reporting at each visit SFC (178) 53 (30.0) NA
Sal (177) 57 (32.2) NA
Flu (183) 49 (26.8) NA
Placebo (185) 59 (31.9) NA
Kardos et al,27 2007 Adequate, centrally Adequate AEs and SAEs recorded during trial and SFC (507) 99 (19.5) 4 (0.8)
generated block follow-up Sal (487) 103 (21.1) 3 (0.6)
Mahler et al,28 2002 Unclear Unclear AEs and SAEs documented SFC (165) 52 (31.5) NA
Sal (160) 45 (28.2) NA
Flu (168) 68 (40.5) NA
Placebo (181) 69 (38.1) NA
Paggiaro et al,29 1998 Adequate, computer Adequate AE defined as untoward medical occurrence Flu (142) 19 (13.3) 0
generated during treatment Placebo (139) 27 (19.4) 2 (1.4)
SCO100250,30 2008 Unclear Unclear AEs and SAEs recorded after study medication SFC (394) 125 (31.7) NA
administration but no later than last date after Sal (403) 156 (38.7) NA
study medication administration
SCO100470,31 2006 Unclear Unclear AEs and SAEs recorded at each study visit SFC (518) 59 (11.4) NA
Sal (532) 74 (13.9) NA
SCO40041,32 2008 Unclear Unclear AEs and SAEs monitored during therapy SFC (92) 36 (39.1) NA
Sal (94) 39 (41.5) NA
SFCT01/ Unclear Unclear All AEs occurring after subject consented to Flu (131) 34 (26.0) NA
SCO30002,33 2005 participate until end of follow-up Placebo (125) 40 (32.0) NA
van der Valk et al,34 2002 Adequate, permuted Adequate 3- and 6-mo follow-up Flu (123) 1 (0.8) 0
blocks, stratified Placebo (121) 1 (0.8) 0
Vestbo et al,35 1999 Adequate, computer Adequate Participants seen every 3 mo Bud (145) 36 (24.8) 0
generated Placebo (145) 51 (35.2) 0
Wouters et al,36 2005 Adequate Adequate AE collected at start and end of treatment SFC (189) 34 (18.0) 0
Sal (184) 46 (25.0) 0

Abbreviations: AE, adverse event; Bud, budesonide; COPD, chronic obstructive pulmonary disease; Flu, fluticasone propionate; For, formoterol;
For/Bud, combination of formoterol and budesonide; NA, not available; RCT, randomized controlled trial; SAE, serious AE; Sal, salmeterol xinafoate;
SFC, combination of salmeterol and fluticasone.
a All RCTs were double blind.
b Data on pneumonia were extracted from the US Food and Drug Administration presentation37,38 because the published version provided information on the
probability of pneumonia and not the actual number of events.

QUANTITATIVE DATA
SYNTHESIS AND SENSITIVITY
ANALYSIS
We used Review Manager software, ver-
sion 5.0.15 (Nordic Cochrane Center,
Copenhagen, Denmark), to calculate
pooled relative risk (RR) and 95% con-
fidence intervals (CIs) using a random-
effects model. Outcome data and data on
trial participants were analyzed using a
22 format according to the intention-
to-treat principle. All reported P values
are 2 sided, with significance set at less
than .05. Statistical heterogeneity was as-
sessed using the I2 statistic; values of 50%
or more indicated a substantial level of
heterogeneity.17 If substantial statisti-
cal heterogeneity was present (I2 50%),
we planned to explore individual study
characteristics and those of subgroups
of the main body of evidence.
We performed a sensitivity analysis
to explore the influence on effect size for

(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 3), FEB 9, 2009 WWW.ARCHINTERNMED.COM
222
Downloaded from www.archinternmed.com on June 14, 2010
2009 American Medical Association. All rights reserved.
 ICS No ICS

No. of Total No. No. of Total No. Weight, Risk Ratio Favors Favors
Source or Subgroup Events of Patients Events of Patients % (95% CI) treatment control
ICS-LABA vs LABA
Aaron et al,20 2007 1 145 1 148 0.4 1.02 (0.06-16.16)
Calverley et al,22 2003 7 358 9 372 3.3 0.81 (0.30-2.15)
Calverley et al,23 2003 8 254 7 255 3.1 1.15 (0.42-3.12)
Calverley et al,5 2007 248 1546 162 1542 27.4 1.53 (1.27-1.84)
Ferguson et al,24 2008 24 394 9 388 5.2 2.63 (1.24-5.58)
Hanania et al,26 2003 0 178 1 177 0.3 0.33 (0.01-8.08)
Kardos et al,27 2007 23 507 7 487 4.3 3.16 (1.37-7.29)
Mahler et al,28 2002 2 165 0 160 0.4 4.85 (0.23-100.23)
SCO100250,30 2008 23 394 9 403 5.1 2.61 (1.22-5.58)
SCO100470,31 2008 2 532 4 518 1.2 0.49 (0.09-2.65)
SCO40041,32 2008 8 92 6 94 3.0 1.36 (0.49-3.77)
Wouters et al,36 2005 10 189 2 184 1.4 4.87 (1.08-21.92)

Subtotal 4754 4728 55.2 1.72 (1.28-2.30)

Total No. of Events 356 217
Heterogeneity: 2 = 0.05; 112 = 14.04 (P = .23); I2 = 22%
Test for overall effect: z = 3.62 (P <.001)

ICS vs placebo
Burge et al,21 2000 18 372 8 370 4.5 2.24 (0.99-5.08)
Calverley et al,22 2003 9 374 3 361 1.9 2.90 (0.79-10.61)
Calverley et al,23 2003 5 257 2 256 1.2 2.49 (0.49-12.72)
Calverley et al,5 2007 224 1552 139 1544 26.3 1.60 (1.31-1.96)
FLTA3025,25 2005 4 434 1 206 0.7 1.90 (0.21-16.88)
Hanania et al,26 2003 1 183 0 185 0.3 3.03 (0.12-73.96)
Mahler et al,28 2002 2 168 0 181 0.4 5.38 (0.26-111.35)
Paggiaro et al,29 1998 2 142 2 139 0.9 0.98 (0.14-6.85)
SFCT01/SCO30002,33 2005 1 131 1 125 0.6 0.95 (0.06-15.09)
van der Valk et al,34 2002 3 123 0 121 0.4 6.89 (0.36-131.93)
Vestbo et al,35 1999 16 145 24 145 7.8 0.67 (0.37-1.20)

Subtotal 3881 3633 44.8 1.51 (1.08-2.10)

Total No. of Events 285 180
Heterogeneity: 2 = 0.05; 102 = 12.15 (P = .27); I2 = 18%
Test for overall effect: z = 2.40 (P = .02)

Total 8635 8361 100.0 1.60 (1.33-1.92)

Total No. of Events 641 397
Heterogeneity: 2 = 0.02; 222 = 26.32 (P = .24); I2 = 16%
Test for overall effect: z = 4.99 (P <.001)
0.01 0.1 1.0 10 100
Risk Ratio (95% CI)

Figure 2. Meta-analysis of randomized controlled trials of inhaled corticosteroid (ICS) use vs control treatment for any pneumonia. CI indicates confidence
interval; LABA, long-acting -agonist.

statistical models (fixed vs random ef- cebo or active comparators, for 1 addi- haled corticosteroids and long-
fects), the influence of individual trials, tional patient to be harmed by an adverse acting -agonist combinations with
and the risk of bias by restricting the event of pneumonia. long-acting -agonists. Five
analysis to RCTs considered to be at low
RCTs5,23,26,28,31 had 4 arms that com-
risk of bias (adequate sequence genera-
tion, allocation concealment, and double RESULTS pared inhaled corticosteroids with
blinding, with clear reporting of loss to placebo and the combination of in-
Of the 651 potentially relevant ci-
follow-up). The fail-safe numbers using haled corticosteroids and long-
tations identified, and after a de-
the Rosenberg method18 were calcu- acting -agonist with long-acting
tailed review of 97 of those studies,
lated to assess for the influence of pub- -agonists. Ten RCTs5,20-22,27-29,33,34,36
lication bias on the meta-analysis. The 18 trials5,20-36 fulfilled our inclusion
criteria. The flow of the trial is shown evaluated inhaled fluticasone pro-
fail-safe number represents the num-
in Figure 1. Trial characteristics are pionate, alone or in combination
ber of nonsignificant studies that would
need to be added to a meta-analysis to given in Table 1. with salmeterol xinafoate, at a dos-
reverse an overall statistically signifi- The trials included 16 996 pa- age of 500 g twice daily, and 6
cant result to nonsignificance. tients, of whom 8635 received in- trials24-26,30-32 evaluated inhaled flu-
The number needed to harm with in- haled corticosteroids and 8361 re- ticasone propionate at a dosage of
haled corticosteroids was calculated by ceived control therapy. The trials 250 g twice daily. Two RCTs23,35
applying the pooled RR estimates from evaluated inhaled budesonide at a
ranged in duration from 24 weeks
the meta-analysis to the control event rate
to 3 years. The sample size ranged dosage of 400 g twice daily.
in a large trial population using Visual Rx,
version 2.0.19 The number needed to from 186 to 6184 in these trials. Six Trial quality was variable
harm is the number of patients with RCTs21,25,29,33-35 compared inhaled (Table 2). Nine RCTs5,20-22,27,29,34-36
COPD who needed to be treated with in- corticosteroids with placebo. Seven were judged to be at low risk of bias
haled corticosteroid, rather than with pla- RCTs 20,24,27,30-32,36 compared in- (adequate sequence generation, allo-

(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 3), FEB 9, 2009 WWW.ARCHINTERNMED.COM
223
Downloaded from www.archinternmed.com on June 14, 2010
2009 American Medical Association. All rights reserved.
 ICS No ICS

No. of Total No. No. of Total No. Weight, Risk Ratio Favors Favors
Source or Subgroup Events of Patients Events of Patients % (95% CI) treatment control
ICS-LABA vs LABA
Aaron et al,20 2007 1 145 1 148 0.3 1.02 (0.06-16.16)
Calverley et al,22 2003 7 358 9 372 2.5 0.81 (0.30-2.15)
Calverley et al,5 2007 157 1546 99 1542 41.6 1.58 (1.24-2.01)
Ferguson et al,24 2008 18 394 5 388 2.5 3.55 (1.33-9.45)
Hanania et al,26 2003 0 178 1 177 0.2 0.33 (0.01-8.08)
Kardos et al,27 2007 14 507 4 487 2.0 3.36 (1.11-10.14)
Mahler et al,28 2002 2 165 0 160 0.3 4.85 (0.23-100.23)
SCO100250,30 2008 11 394 7 403 2.8 1.61 (0.63-4.10)
SCO100470,31 2006 2 532 4 518 0.8 0.49 (0.09-2.65)
SCO40041,32 2008 5 92 4 94 1.5 1.28 (0.35-4.61)
Wouters et al,36 2005 10 189 2 184 1.1 4.87 (1.08-21.92)

Subtotal 4500 4473 55.6 1.68 (1.20-2.34)

Total No. of Events 227 136
Heterogeneity: 2 = 0.05; 102 = 11.83 (P = .30); I2 = 15%
Test for overall effect: z = 3.06 (P = .002)

ICS vs placebo
Burge et al,21 2000 18 372 8 370 3.6 2.24 (0.99-5.08)
Calverley et al,22 2003 9 374 3 361 1.4 2.90 (0.79-10.61)
Calverley et al,5 2007 150 1552 86 1544 37.1 1.74 (1.34-2.24)
FLTA3025,25 2005 4 434 1 206 0.5 1.90 (0.21-16.88)
Hanania et al,26 2003 1 183 0 185 0.2 3.03 (0.12-73.96)
Mahler et al,28 2002 2 168 0 181 0.3 5.38 (0.26-111.35)
Paggiaro et al,29 1998 2 142 2 139 0.6 0.98 (0.14-6.85)
SFCT01/SCO30002,33 2005 1 131 1 125 0.3 0.95 (0.06-15.09)
van der Valk et al,34 2002 3 123 0 121 0.3 6.89 (0.36-131.93)

Subtotal 3479 3232 44.4 1.81 (1.44-2.29)

Total No. of Events 190 101
Heterogeneity: 2 = 0.00; 28 = 2.85 (P = .94); I2 = 0%
Test for overall effect: z = 4.99 (P <.001)

Total 7979 7705 100.0 1.71 (1.46-1.99)

Total No. of Events 417 237
Heterogeneity: 2 = 0.00; 192 = 15.14 (P = .71); I2 = 0%
Test for overall effect: z = 6.73 (P <.001)
0.01 0.1 1.0 10 100
Risk Ratio (95% CI)

Figure 3. Meta-analysis of randomized controlled trials of inhaled corticosteroid (ICS) use vs control treatment for serious pneumonia. CI indicates confidence
interval; LABA, long-acting -agonist.

cation concealment and double blind- (TORCH) Study5 were extracted creased risk of serious pneumonia
ing, and clear reporting of loss to fol- from regulatory documents37,38 be- (417 of 7979 [5.2%] vs 237 of 7705
low-up), whereas 9 RCTs23-26,28,30-33 cause the published version re- [3.1%]; RR, 1.71; 95% CI, 1.46-1.99
were at unclear risk of bias. None of ported the probability of pneumo- [P.001]) in a meta-analysis of 16
theincludedtrialsusedobjectivepneu- nia rather than actual rates. trials5,20-22,24-34,36 involving 15 684 pa-
monia definitions, required radio- tients. There was no evidence of sta-
graphic confirmation of pneumonia, PRIMARY OUTCOME tistical heterogeneity among the trials
or specifically monitored pneumonia (I2 =0%) (Figure 3).
as an outcome of interest. Any Pneumonia The significantly increased risk of
Only 2 trials23,35 that evaluated serious pneumonia associated with
Use of inhaled corticosteroids was
inhaled budesonide reported on inhaled corticosteroid use was ro-
associated with a significantly in-
pneumonia as an adverse event bust to the choice of comparators.
creased risk of any pneumonia when
irrespective of severity. Sixteen The combination of inhaled corti-
compared with controls (641 of 8635
trials5,20-22,24-34,36 evaluated inhaled flu- costeroids and long-acting -
[7.4%] vs 397 of 8361 [4.7%]; RR,
ticasone vs a control and reported on agonists was associated with a sig-
1.60; 95% CI, 1.33-1.92 [P.001])
pneumonia as a serious adverse event, nificantly increased risk of serious
in a meta-analysis of 18 trials in-
with 5 of these trials5,24,27,30,32 addi- pneumonia when compared with
volving 16 996 patients.5,20-36 There
tionally reporting on pneumonia as long-acting -agonists (227 of 4500
was evidence of low statistical
an adverse event, irrespective of se- [5.0%] vs 136 of 4473 [3.0%]; RR,
heterogeneity among the included
verity. Only 15 trials5,20,22,24-34,36 re- 1.68; 95% CI, 1.20-2.34 [P=.002])
trials (I2 =16%) (Figure 2).
ported on pneumonia-related mor- in a meta-analysis of 11 trials* in-
tality, whereas all 18 trials 5,20-36 Serious Pneumonia volving 8973 patients. There was evi-
provided data on overall mortality. dence of low statistical heteroge-
Data on pneumonia for the To- Use of inhaled corticosteroids was as-
ward a Revolution in COPD Health sociated with a significantly in- *References 5, 20, 22-24, 26-28, 30-32, 36.

(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 3), FEB 9, 2009 WWW.ARCHINTERNMED.COM
224
Downloaded from www.archinternmed.com on June 14, 2010
2009 American Medical Association. All rights reserved.
 ICS No ICS

No. of Total No. No. of Total No. Weight, Risk Ratio Favors Favors
Source or Subgroup Events of Patients Events of Patients % (95% CI) treatment control
ICS-LABA vs LABA
Aaron et al,20 2007 1 145 1 148 2.9 1.02 (0.06-16.16)
Calverley et al,22 2003 0 358 0 372 Not estimable
Calverley et al,5 2007 15 1546 15 1542 43.1 1.00 (0.49-2.03)
Ferguson et al,24 2008 1 394 0 388 2.1 2.95 (0.12-72.30)
Hanania et al,26 2003 0 178 0 177 Not estimable
Kardos et al,27 2007 1 507 1 487 2.9 0.96 (0.06-15.31)
Mahler et al,28 2002 0 165 0 160 Not estimable
SCO100250,30 2008 0 394 0 403 Not estimable
SCO100470,31 2008 0 532 0 518 Not estimable
SCO40041,32 2008 1 92 1 94 2.9 1.02 (0.06-16.09)
Wouters et al,36 2005 0 189 0 184 Not estimable

Subtotal 4500 4473 53.8 1.04 (0.55-1.97)

Total No. of Events 19 18
Heterogeneity: 2 = 0.00; 42 = 0.43 (P = .98); I2 = 0%
Test for overall effect: z = 0.13 (P = .90)

ICS vs placebo
Calverley et al,22 2003 0 374 0 361 Not estimable
Calverley et al,5 2007 21 1552 13 1544 46.2 1.61 (0.81-3.20)
FLTA3025,25 2005 0 434 0 206 Not estimable
Hanania et al,26 2003 0 183 0 185 Not estimable
Mahler et al,28 2002 0 168 0 181 Not estimable
Paggiaro et al,29 1998 0 142 0 139 Not estimable
SFCT01/SC30002,33 2005 0 131 0 125 Not estimable
van der Valk et al,34 2002 0 123 0 121 Not estimable

Subtotal 3107 2862 46.2 1.61 (0.81-3.20)

Total No. of Events 21 13
Heterogeneity: not applicable
Test for overall effect: z = 1.35 (P = .18)

Total 7607 7335 100.0 1.27 (0.80-2.03)

Total No. of Events 40 31
Heterogeneity: 2 = 0.00; 52 = 1.25 (P = .94); I2 = 0%
Test for overall effect: z = 1.01 (P = .31)
0.01 0.1 1.0 10 100
Risk Ratio (95% CI)

Figure 4. Meta-analysis of randomized controlled trials of inhaled corticosteroid (ICS) vs control for pneumonia-related mortality. CI indicates confidence interval;
LABA, long-acting -agonist.

neity among the included trials 15 trials5,20,22,24-34,36 involving 14 942 direction to those obtained from ran-
(I2 = 15%). Use of inhaled cortico- patients. There was no evidence of dom-effects analysis.
steroids was associated with a sig- statistical heterogeneity among the The sensitivity analysis of pneu-
nificantly increased risk of serious included trials (I2 =0%) (Figure 4). monia limited to the 9 RCTs at low
pneumonia when compared with risk of bias5,20-22,27,29,34-36 found that
placebo (190 of 3479 [5.5%] vs 101 Overall Mortality the risk of pneumonia (RR, 1.55;
of 3232 [3.1%]; RR, 1.81; 95% CI. 95% CI, 1.19-2.00 [P=.001]) asso-
1.44-2.29 [P .001]) in a meta- Use of inhaled corticosteroids was ciated with inhaled corticosteroid
analysis of 9 trials involving 6711 not associated with a significantly use was similar in magnitude and di-
patients. There was no evidence of increased risk of overall mortality rection to those from 9 trials in
statistical heterogeneity among the when compared with controls (525 which the risk of bias was unclear
included trials (I2 = 0%) (Figure 3). of 8635 [6.1%] vs 549 of 8361 (1.88; 1.29-2.75 [P=.001]).23-26,28,30-33
[6.6%]; RR, 0.96; 95% CI, 0.86- After excluding the trial with the
SECONDARY OUTCOMES 1.08 [P=.51]) in a meta-analysis of largest number of participants and
18 trials 5,20-36 involving 16 996 the longest duration of follow-up,5
Pneumonia-Related Mortality patients. There was no evidence the pooled analysis of pneumonia
of statistical heterogeneity among (RR, 1.69; 95% CI, 1.22-2.33
Use of inhaled corticosteroids was the included trials (I2 = 0%) [P = .001]) associated with inhaled
not associated with a significantly in- (Figure 5). corticosteroid use from 17 RCTs was
creased risk of pneumonia-related similar in magnitude and direction
mortality when compared with con- SENSITIVITY ANALYSIS to those obtained from 18 RCTs.
trols (40 of 7607 [0.5%] vs 31 of
7335 [0.4%]; RR, 1.27; 95% CI, 0.80- The fixed-effects analysis of pneu- FAIL-SAFE NUMBER
2.03 [P=.31]) in a meta-analysis of monia from 18 trials yielded effect
sizes (RR 1.60; 95% CI, 1.42-1.80 According to the Rosenberg
References 5, 21, 22, 25, 26, 28, 29, 33, 34. [P.001]) similar in magnitude and method,18 16 nonsignificant trials of

(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 3), FEB 9, 2009 WWW.ARCHINTERNMED.COM
225
Downloaded from www.archinternmed.com on June 14, 2010
2009 American Medical Association. All rights reserved.
 ICS No ICS

No. of Total No. No. of Total No. Weight, Risk Ratio Favors Favors
Source or Subgroup Events of Patients Events of Patients % (95% CI) treatment control
ICS-LABA vs LABA
Aaron et al,20 2007 6 145 6 148 1.0 1.02 (0.34-3.09)
Calverley et al,22 2003 2 358 3 372 0.4 0.69 (0.12-4.12)
Calverley et al,23 2003 5 254 13 255 1.2 0.39 (0.14-1.07)
Calverley et al,5 2007 193 1546 205 1542 37.5 0.94 (0.78-1.13)
Ferguson et al,24 2008 6 394 3 388 0.7 1.97 (0.50-7.82)
Hanania et al,26 2003 0 178 0 177 Not estimable
Kardos et al,27 2007 7 507 9 487 1.3 0.75 (0.28-1.99)
Mahler et al,28 2002 0 165 0 160 Not estimable
SCO100250,30 2008 4 394 6 403 0.8 0.68 (0.19-2.40)
SCO100470,31 2008 3 532 3 518 0.5 0.97 (0.20-4.80)
SCO40041,32 2008 5 92 7 94 1.0 0.73 (0.24-2.22)
Wouters et al,36 2005 2 189 4 184 0.4 1.49 (0.09-2.63)

Subtotal 4754 4728 44.8 0.90 (0.76-1.07)

Total No. of Events 233 259
Heterogeneity: 2 = 0.00; 92 = 5.23 (P = .81); I2 = 0%
Test for overall effect: z = 1.18 (P = .24)

ICS vs placebo
Burge et al,21 2000 32 372 36 370 6.1 0.88 (0.56-1.39)
Calverley et al,22 2003 3 374 7 361 0.7 0.41 (0.11-1.59)
Calverley et al,23 2003 6 257 5 256 0.9 1.20 (0.37-3.87)
Calverley et al,5 2007 246 1552 231 1544 46.2 1.06 (0.90-1.25)
FLTA3025,25 2005 0 434 0 206 Not estimable
Hanania et al,26 2003 0 183 0 185 Not estimable
Mahler et al,28 2002 0 168 3 181 0.1 0.15 (0.01-2.96)
Paggiaro et al,29 1998 0 142 2 139 0.1 0.20 (0.01-4.04)
SFCT01/SCO30002,33 2005 0 131 0 125 Not estimable
van der Valk et al,34 2002 1 123 1 121 0.2 0.98 (0.06-15.55)
Vestbo et al,35 1999 4 145 5 145 0.8 0.80 (0.22-2.92)

Subtotal 3881 3633 55.2 1.01 (0.87-1.18)

Total No. of Events 292 290
Heterogeneity: 2 = 0.00; 27 = 5.26 (P = .63); I2 = 0%
Test for overall effect: z = 0.19 (P = .85)

Total 8635 8361 100.0 0.96 (0.86-1.08)

Total No. of Events 525 549
Heterogeneity: 2 = 0.00; 172 = 11.49 (P = .83); I2 = 0%
Test for overall effect: z = 0.65 (P = .51)
0.01 0.1 1.0 10 100
Risk Ratio (95% CI)

Figure 5. Meta-analysis of randomized controlled trials of inhaled corticosteroids (ICS) vs control for overall mortality. In the Forest plot, the white horizontal line
within the black box for Calverley et al5 indicates the 95% confidence interval (CI). LABA, indicates long-acting -agonist.

inhaled corticosteroids with an av- COMMENT 1.30-1.80) in a case-control study

erage sample size of 945 partici- among elderly patients with COPD
pants each would be required to re- in Canada.6 Another unpublished
verse the significantly increased risk The use of inhaled corticosteroids for population-based cohort study in the
of pneumonia in the random ef- more than 24 weeks in patients with General Practice Research Data-
fects meta-analysis. COPD is associated with a signifi- base in the United Kingdom re-
cantly increased risk of any pneu- ported a doubling in the risk of
ESTIMATED NUMBER monia (RRs, an approximately 60% pneumonia 4 years after inhaled flu-
NEEDED TO HARM increased risk) and serious pneu- ticasone exposure (odds ratio [OR],
monia (RRs, an approximately 70% 2.08; 95% CI, 1.01-7.77). 39 An-
Assuming a baseline event rate of increased risk) without a signifi- other unpublished case-control
30 per 1000 person-years for seri- cant increase in the risk of pneumo- study based on the General Prac-
ous pneumonia in adult patients nia-related death or overall death. tice Research Database reported no
with COPD similar to the control Our findings need to be inter- significant association with pneu-
event rate in the trial population preted in the context of evidence monia with inhaled corticosteroid
of the TORCH Study,38 the annu- from recent database studies. Use of exposure of up to 24 months, with
alized number needed to harm for inhaled corticosteroids was associ- or without long-acting -agonists,
serious pneumonia associated ated with a dose-dependent in- when compared with short-acting -
with inhaled corticosteroid use creased risk of hospitalization for agonists.40 However, the highest risk
added to long-acting -agonist pneumonia (RR, 1.70; 95% CI, 1.63- for pneumonia associated with in-
therapy is estimated to be 47 (95% 1.77) and an increase in pneumonia- haled corticosteroid use in combi-
CI, 34-73). related mortality in 30 days (1.53; nation with a long-acting -agonist

(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 3), FEB 9, 2009 WWW.ARCHINTERNMED.COM
226
Downloaded from www.archinternmed.com on June 14, 2010
2009 American Medical Association. All rights reserved.
was seen in the 18- to 24-month cat-
egories (OR, 1.82; 95% CI, 0.95-
3.47 [P=.07]).40
Our findings must be distin-
guished from a previous meta-
analysis of a limited number of
mainly published trials. A Coch-
rane review of 7 trials5,22,23,26-28,31 re-
ported a significantly increased risk
of pneumonia with use of inhaled
corticosteroids in combination with
long-acting -agonists (OR, 1.58;
95% CI, 1.32-1.88 [P.001]) com-
pared with use of long-acting -
agonists alone.10 Another pooled
analysis of 2 trials5,23 showed an in-
creased risk of pneumonia with use
of inhaled corticosteroids com-
pared with placebo (RR, 1.55; 95%
CI, 1.33-1.80).9 Our robust meta-
analysis of 18 trials involving 16 996
patients clarifies that the risk of
pneumonia reported as an adverse
event (and a serious adverse event)
can be specifically attributed to the
long-term use of the inhaled corti-
costeroid component because the
RRs for serious pneumonia associ-
ated with inhaled corticosteroid use
are similar when inhaled corticoste-
roids are compared with placebo
(RR, 1.81; 95% CI, 1.44-2.29
[P .001]) or when use of inhaled
corticosteroids in combination with
long-acting -agonists are com-
pared with use of long-acting -
agonists alone (RR, 1.68; 95% CI,
1.20-2.34 [P=.002]). Our findings
also indicate that this increased risk
of pneumonia associated with in-
haled corticosteroid use is not ac-
companied by a proportionate in-
crease in pneumonia-related
mortality or overall mortality.
Another recent trial41 showed a
significant increase in the rates of se-
rious pneumonia in the inhaled sal-
meterol-fluticasone combination
compared with inhaled tiotropium
bromide for 2 years (8% vs 4%; haz-
ard ratio [HR], 1.94; 95% CI, 1.19-
3.17 [P=.008]), a risk likely attrib-
utable to the fluticasone component.
These risks of inhaled corticoste-
roid use must be weighed against the
benefits. The TORCH trial found no
reduction of mortality with use of in-
haled fluticasone (HR, 1.06; 95% CI,
0.89-1.27 [P=.53]).5 Inhaled fluti-
casone use failed to significantly re-
duce the annual rate ratio of severe
COPD exacerbations requiring hos-
(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 3), FEB 9, 2009
Downloaded from www.archinternmed.com on June 14, 2010
2009 American Medical Association. All rights reserved.
pitalization when compared with
placebo (0.88; 95% CI, 0.74-1.03
[P = .10]) or when the use of in-
haled fluticasone in combination
with salmeterol was compared with
use of salmeterol alone (1.02; 0.87-
1.20 [P = .79]).5 Serious pneumo-
nias, which constituted a subset of
exacerbations, were relatively infre-
quent compared with the number of
exacerbations, resulting in an in-
crease in the risk of serious pneu-
monia with fluticasone use with-
out an increase in the rates of
hospitalization for exacerbations in
the TORCH trial.5 Similarly, the HR
of the first exacerbations with in-
haled corticosteroid use among those
who were not using inhaled corti-
costeroids before randomization was
nonsignificant (1.11; 95% CI, 0.69-
1.79 [P=.68]) in another trial.20,42 Al-
though fluticasone use provided sta-
tistically significant improvements
in quality-of-life measures, such as
the St George Respiratory Question-
naire (SGRQ) (adjusted mean
change in SGRQ score for flutica-
sone vs placebo, 2; 95% CI, 2.9 to
1 [P .001]), these improve-
ments failed to meet the threshold
for clinical significance (a decre-
ment of 4 points on this 100-point
questionnaire is considered be clini-
cally significant).5 Thus, the ben-
efits of use of inhaled fluticasone in
combination with salmeterol in pa-
tients with COPD can be attributed
to salmeterol, without any synergis-
tic effect from the use of the com-
bination.42
The precise mechanism under-
lying the increased risk of pneumo-
nia with inhaled corticosteroid use
in patients with COPD is uncer-
tain. Inhaled corticosteroids achieve
locally high concentrations in the
lung and may increase the risk of
pneumonia owing to their immu-
nosuppressive effects.2 Inhaled flu-
ticasone propionate at dosages of
1000 g/d exerts effects on serum
cortisol levels that are equivalent to
10 mg of oral prednisone, a dose that
may double the risk of pneumonia
in patients with rheumatoid arthri-
tis. 43 The limited efficacy of in-
haled corticosteroid use in COPD
may be due to the progressive re-
duction in the activity of histone
deacetylases, particularly histone
deacetylase 2, in the lungs of pa-
WWW.ARCHINTERNMED.COM
227
tients with increasing severity of
COPD.44-46 Histone deacetylase 2 is
required by corticosteroids to switch
off activated inflammatory genes in
COPD.44-46 The contrasting effects of
inhaled corticosteroids on pneumo-
nia and exacerbations may possi-
bly be related to a differential effect
on viral compared with bacterial in-
fections and needs further investi-
gation.47
Although the risk-benefit pro-
file of inhaled corticosteroid use in
asthma is clearly favorable, the risk
of pneumonia associated with the
use of inhaled corticosteroids in pa-
tients with severe chronic asthma
who have clinical features similar to
those of patients with COPD is un-
known.
Our meta-analysis has several
limitations, which mainly stem from
the quality of reported data. The
trials did not consistently use an ob-
jective definition of pneumonia or
require radiographic confirmation.
Most of the trials were inad-
equately powered to detect any sig-
nificant difference in overall mor-
tality or in pneumonia-related
mortality. Nine trials in the analy-
sis were at unclear risk of bias. In the
absence of patient-level data, we
could not ascertain whether pa-
tients with mild COPD were less sus-
ceptible to the risk of pneumonia as-
sociated with inhaled corticosteroid
use. However, the direction of effect
for the RRs for pneumonia with in-
haled corticosteroid use favors con-
trols in all trials, except for 2 small
trials that recruited patients with less
severe COPD.26,35 We could not de-
termine the onset of pneumonia,
dose-response relationships, or the
influence of age, body mass index,
concomitant systemic corticoste-
roid use, influenza, and pneumo-
coccal vaccination status on the risk
of pneumonia associated with in-
haled corticosteroid use. We were
unable to discern intraclass differ-
ences in the risk of pneumonia.
The manufacturers of inhaled
corticosteroids need to make source
data available to external academ-
ics for independent analysis. Pro-
spective trials of inhaled corticoste-
roids need to monitor pneumonia as
a prespecified outcome using objec-
tive pneumonia definitions. Addi-
tional cost-benefit analysis may pro-
vide further guidance on optimal use
of inhaled corticosteroids among pa-
tients with COPD.
Despite these limitations, our
findings have potential implica-
tions. Our findings suggest the
possibility of a risk of pneumonia
associated with the long-term use
of inhaled corticosteroids in
patients with COPD. The magni-
tude of this risk of serious pneu-
monia associated with inhaled
corticosteroid use in patients with
COPD is substantial (RRs,
approximately 70% increased
risk) and may pose a substantial
public health burden. Chronic
obstructive pulmonary disease is a
known risk factor for the occur-
rence of pneumonia 48,49 and for
hospitalization for pneumonia.50
Patients with COPD and pneumo-
nia also have a higher risk of mor-
tality.51 Clinicians should remain
vigilant for the development of
pneumonia with inhaled cortico-
steroid use because the signs and
symptoms of pneumonia may
closely mimic those of COPD
exacerbations. Clinicians should
reevaluate the benefit-harm profile
of long-term inhaled corticoste-
roid use among patients with
COPD.
Accepted for Publication: Septem-
ber 3, 2008.
Corresponding Author: Sonal Singh,
MD, MPH, Department of Medi-
cine, Wake Forest University School
of Medicine, One Medical Center
Boulevard, Winston-Salem, NC
27157 (sosingh@wfubmc.edu).
Author Contributions: Dr Singh had
full access to all of the data in the
study and takes responsibility for the
integrity of the data and the accu-
racy of the data analysis. Study con-
cept and design: Singh and Loke. Ac-
quisition of data: Singh, Amin, and
Loke. Analysis and interpretation of
data: Singh and Loke. Drafting of the
manuscript: Singh, Amin, and Loke.
Critical revision of the manuscript for
important intellectual content: Singh,
Amin, and Loke. Statistical analy-
sis: Singh, Amin, and Loke. Admin-
istrative, technical, and material sup-
port: Singh. Study supervision: Singh.
Financial Disclosure: None re-
ported.
(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 3), FEB 9, 2009
Downloaded from www.archinternmed.com on June 14, 2010
2009 American Medical Association. All rights reserved.
REFERENCES
1. Global Initiative for Chronic Obstructive Lung Dis-
ease (GOLD). Global strategy for the diagnosis,
management, and prevention of COPD. 2007. http:
//www.goldcopd.org/Guidelineitem.asp
?l1=2&l2=1&intId=989. Accessed August 26,
2008.
2. Suissa S, McGhan R, Niewoehner D, Make B.
Inhaled corticosteroids in chronic obstructive pul-
monary disease. Proc Am Thorac Soc. 2007;
4(7):535-542.
3. Ernst P, Baltzan M, Deschenes J, Suissa S. Low-
dose inhaled and nasal corticosteroid use and the
risk of cataracts. Eur Respir J. 2006;27(6):1168-
1174.
4. Suissa S, Baltzan M, Kremer R, Ernst P. Inhaled
and nasal corticosteroid use and the risk of
fracture. Am J Respir Crit Care Med. 2004;169
(1):83-88.
5. Calverley PMA, Anderson JA, Celli B; TORCH
Investigators. Salmeterol and fluticasone propio-
nate and survival in chronic obstructive pulmo-
nary disease. N Engl J Med. 2007;356(8):775-
789.
6. Ernst P, Gonzalez AV, Brassard P, Suissa S.
Inhaled corticosteroid use in chronic obstructive
pulmonary disease and the risk of hospitaliza-
tion for pneumonia. Am J Respir Crit Care Med.
2007;176(2):162-166.
7. GlaxoSmithKline. GlaxoSmithKline clinical trials
register. http://ctr.gsk.co.uk/medicinelist.asp. Ac-
cessed August 26, 2008.
8. AstraZeneca. Pulmicort (budesonide) clinical trial
report summaries. http://www.astrazenecaclini-
caltrials.com/article/528365.aspx. Accessed Oc-
tober 20, 2008.
9. Wilt TJ, Niewoehner D, MacDonald R, Kane RL.
Management of stable chronic obstructive pul-
monary disease: a systematic review for a clini-
cal practice guideline. Ann Intern Med. 2007;
147(9):639-653.
10. Nannini LJ, Cates CJ, Lasserson TJ, Poole P.
Combined corticosteroid and long-acting beta-
agonist in one inhaler versus long-acting beta-
agonists for chronic obstructive pulmonary
disease. Cochrane Database Syst Rev. 2007;
(4):CD006829.
11. Sin DD, Wu L, Anderson JA, et al. Inhaled corti-
costeroids and mortality in chronic obstructive pul-
monary disease. Thorax. 2005;60(12):992-997.
12. Gartlehner G, Hansen RA, Carson SS, Lohr KN.
Efficacy and safety of inhaled corticosteroids in
patients with COPD: a systematic review and meta-
analysis of health outcomes. Ann Fam Med. 2006;
4(3):253-262.
13. Soriano JB, Sin DD, Zhang X, et al. A pooled analy-
sis of FEV1 decline in COPD patients randomized
to inhaled corticosteroids or placebo. Chest. 2007;
131(3):682-689.
14. Highland KB, Strange C, Heffner JE. Long-term
effects of inhaled corticosteroids on FEV1 in pa-
tients with chronic obstructive pulmonary dis-
ease: a meta-analysis. Ann Intern Med. 2003;
138(12):969-973.
15. Higgins JPT, Green S, eds. Cochrane handbook
for systematic reviews of interventions, version
5.0.0 [updated February 2008]. Cochrane Col-
laboration, 2008. http://www.cochrane-handbook
.org. Accessed August 26,2008.
16. International Council on Harmonization. Interna-
tional conference on harmonisation of technical
requirements for registration of pharmaceuticals
for human use: ICH harmonized tripartite guide-
WWW.ARCHINTERNMED.COM
228
line: clinical safety data management: definitions
and standards for expedited reporting E2A. http:
//www.ich.org/LOB/media/MEDIA436.pdf. Ac-
cessed August 26, 2008.
17. Higgins JP, Thompson SG, Deeks JJ, Altman DG.
Measuring inconsistency in meta-analyses. BMJ.
2003;327(7414):557-560.
18. Rosenberg MS. The file-drawer problem revis-
ited: a general weighted method for calculating fail-
safe numbers in meta-analysis. Evolution. 2005;
59(2):464-468.
19. Cates CJ. Dr Chris Cates EBM Web site. http:
//www.nntonline.net. Accessed August 26, 2008.
20. Aaron SD, Vandemheen KL, Fergusson D, et al;
Canadian Thoracic Society/Canadian Respira-
tory Clinical Research Consortium. Tiotropium in
combination with placebo, salmeterol, or flutica-
sone salmeterol for treatment of chronic obstruc-
tive pulmonary disease: a randomized trial. Ann
Intern Med. 2007;146(8):545-555.
21. Burge PS, Calverley PMA, Jones PW, Spencer S,
Anderson JA, Maslen TK. Randomised, double
blind, placebo controlled study of fluticasone pro-
pionate in patients with moderate to severe chronic
obstructive pulmonary disease: the ISOLDE Trial.
BMJ. 2000;320(7245):1297-1303.
22. Calverley P, Pauwels R, Vestbo J, et al; Trial of
Inhaled Steroids and Long-Acting 2 Agonists
Study Group. Combined salmeterol and flutica-
sone in the treatment of chronic obstructive pul-
monary disease: a randomised controlled trial
[published correction appears in Lancet.
2003;361(9369):1660]. Lancet. 2003;361(9356):
449-456.
23. Calverley PM, Boonsawat W, Cseke Z, Zhong N,
Peterson S, Olsson H. Maintenance therapy with
budesonide and formoterol in chronic obstruc-
tive pulmonary disease. Eur Respir J. 2003;
22(6):912-919.
24. Ferguson GT, Anzueto A, Fei R, Emmett A, Kno-
bil K, Kalberg C. Effect of fluticasone propionate/
salmeterol (250/50 mcg) or salmeterol (50 mcg)
on COPD exacerbations. Respir Med. 2008;102
(8):1099-1108.
25. GlaxoSmithKline. Clinical study register. Study No.
FLTA3025. http://www.gsk-clinicalstudyregister
.com/files/pdf/1231.pdf. Accessed October 20,
2008.
26. Hanania NA, Darken P, Horstman D, et al. The ef-
ficacy and safety of fluticasone propionate (250
g)/salmeterol (50 g) combined in the Diskus in-
haler for the treatment of COPD. Chest. 2003;
124(3):834-843.
27. Kardos P, Wencker M, Glaab T, Vogelmeier C.
Impact of salmeterol/fluticasone propionate ver-
sus salmeterol on exacerbations in severe chronic
obstructive pulmonary disease. Am J Respir Crit
Care Med. 2007;175(2):144-149.
28. Mahler DA, Wire P, Horstman D, et al. Effective-
ness of fluticasone propionate and salmeterol com-
bination delivered via the Diskus device in the treat-
ment of chronic obstructive pulmonary disease.
Am J Respir Crit Care Med. 2002;166(8):1084-
1091.
29. Paggiaro PL, Dahle R, Bakran I, Frith L, Holling-
worth K, Efthimiou J. Multicenter randomised pla-
cebo-controlled trial of inhaled fluticasone pro-
pionate in patients with chronic obstructive
pulmonary disease. Lancet. 1998;351(9105):
773-780.
30. GlaxoSmithKline. Clinical study register. Study No.
SCO100250. http://www.gsk-clinicalstudyregister
.com/files/pdf/3452.pdf. Accessed October 20,
2008.
31. GlaxoSmithKline. Clinical study register. Study No.
 SCO100470 http://www.gsk-clinicalstudyregister
.com/files/pdf/1079.pdf. Accessed October 20,
2008.
32. GlaxoSmithKline. Clinical study register. Study No.
SCO40041. http://www.gsk-clinicalstudyregister
.com/files/pdf/21086.pdf. Accessed October 20,
2008.
33. GlaxoSmithKline. Clinical study register. SFCT 01/
SCO30002. http://www.gsk-clinicalstudyregister
.com/files/pdf/23674.pdf. Accessed October 20,
2008.
34. van der Valk P, Monninkhof E, van der Palen J,
Zielhuis G, van Herwaarden C. Effect of discon-
tinuation of inhaled corticosteroids in patients with
chronic obstructive pulmonary disease: the COPE
Study. Am J Respir Crit Care Med. 2002;166
(10):1358-1363.
35. Vestbo J, Sorensen T, Lange P, Brix A, Torre P,
Viskum K. Long-term effect of inhaled budes-
onide in mild and moderate chronic obstructive
pulmonary disease: a randomized controlled trial.
Lancet. 1999;353(9167):1819-1823.
36. Wouters EF, Postma DS, Fokkens B, et al; COS-
MIC (COPD and Seretide; a Multicenter Interven-
tion and Characterization) Study Group.
Withdrawal of fluticasone propionate from com-
bined salmeterol/fluticasone treatment in pa-
tients with COPD causes immediate and sus-
tained disease deterioration: a randomised
controlled trial. Thorax. 2005;60(6):480-487.
37. Medicine and Health Care Products Regulatory
Agency (MHRA). Drug safety update: latest ad-
vice for medicine users. October 2007;1(3). http:
//www.mhra.gov.uk/Publications/Safetyguidance
(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 3), FEB 9, 2009
Downloaded from www.archinternmed.com on June 14, 2010
2009 American Medical Association. All rights reserved.
/DrugSafetyUpdate/CON2032518. Accessed
October 20, 2008.
38. Knobil K. Efficacy and safety data from the Ad-
vair Diskus 500/50 clinical program. http://www
.fda.gov/ohrms/dockets/ac/07/slides
/2007-4298s1-02-GSK-Knobil_files/frame.htm.
Accessed August 26, 2008.
39. GlaxoSmithKline. Assessment of incidence of
pneumonia among COPD patients with or with-
out exposure to inhaled corticosteroids in Gen-
eral Practice Research Database. http://www
.gsk-clinicalstudyregister.com/files/pdf/24060
.pdf. Accessed October 20, 2008.
40. GlaxoSmithKline. Assessment of incidence of
pneumonia among COPD patients on Seretide or
inhaled corticosteroids in UK primary care. http:
//www.gsk-clinicalstudyregister.com/files/pdf
/24056.pdf. Accessed October 20,2008.
41. Wedzicha JA, Calverley PMA, Seemungal TA, Hagan
G, Ansari Z, Stockley RA; INSPIRE Investigators.
The prevention of chronic obstructive pulmo-
nary disease exacerbations by salmeterol/
fluticasone propionate or tiotropium bromide. Am
J Respir Crit Care Med. 2008;177(1):19-26.
42. Suissa S, Ernst P, Vandemheen KL, Aaron SD.
Methodological issues in therapeutic trials of
chronic obstructive pulmonary disease. Eur Respir
J. 2008;31(5):927-933.
43. Wolfe F, Caplan L, Michaud K. Treatment for rheu-
matoid arthritis and the risk of hospitalization for
pneumonia: associations with prednisone, disease-
modifying antirheumatic drugs, and anti-tumor ne-
crosis factor therapy Arthritis Rheum. 2006;
54(2):628-634.
WWW.ARCHINTERNMED.COM
229
44. Ito K, Ito M, Elliott WM, et al. Decreased histone
deacetylase activity in chronic obstructive pul-
monary disease. N Engl J Med. 2005;352(19):
1967-1976.
45. Barnes PJ. Reduced histone deacetylase in COPD:
clinical implications. Chest. 2006;129(1):151-
155.
46. Barnes PJ. Inhaled corticosteroids are not ben-
eficial in chronic obstructive pulmonary disease.
Am J Respir Crit Care Med. 2000;161(2, pt 1):
342-344.
47. Rohde G, Wiethege A, Borg I, et al. Respiratory
viruses in exacerbations of chronic obstructive pul-
monary disease requiring hospitalization: a case-
control study. Thorax. 2003;58(1):37-42.
48. Farr BM, Bartlett CL, Wadsworth J, Miller DL; Brit-
ish Thoracic Society Pneumonia Study Group.
Risk factors for community-acquired pneumo-
nia diagnosed upon hospital admission. Respir
Med. 2000;94(10):954-963.
49. Almirall J, Bolibar I, Balanzo X, Gonzalez CA.
Risk factors for community-acquired pneumo-
nia in adults: a population-based casecontrol
study. Eur Respir J. 1999;13(2):349-355.
50. LaCroix AZ, Lipson S, Miles TP, White L. Pro-
spective study of pneumonia hospitalizations and
mortality of US older people: the role of chronic
conditions, health behaviors, and nutritional status.
Public Health Rep. 1989;104(4):350-360.
51. Restrepo MI, Mortensen EM, Pugh JA, Anzueto
A. COPD is associated with increased mortality in
patients with community-acquired pneumonia. Eur
Respir J. 2006;28(2):346-351.