art%3A10.1007%2FBF03261999
-
Upload
jesus-carlos -
Category
Documents
-
view
217 -
download
0
Transcript of art%3A10.1007%2FBF03261999
-
7/21/2019 art%3A10.1007%2FBF03261999
1/20
Individual NSAIDs and UpperGastrointestinal ComplicationsA Systematic Review and Meta-Analysis of ObservationalStudies (the SOS Project)
Jordi Castellsague,1 Nuria Riera-Guardia,1 Brian Calingaert,2 Cristina Varas-Lorenzo,1
Annie Fourrier-Reglat,3Federica Nicotra,4Miriam Sturkenboom5 andSusana Perez-Gutthann1,
on behalf of the investigators of the Safety of Non-Steroidal Anti-InflammatoryDrugs (SOS) Project
1 RTI Health Solutions, Barcelona, Spain
2 RTI Health Solutions, Research Triangle Park, NC, USA
3 Universite V. Segalen, Bordeaux, France
4 University Milan-Bicocca, Milan, Italy
5 Erasmus University Medical Center, Rotterdam, the Netherlands
Abstract Background:The risk of upper gastrointestinal (GI) complications associatedwith the use of NSAIDs is a serious public health concern. The risk variesbetween individual NSAIDs; however, there is little information on the risk
associated with some NSAIDs and on the impact of risk factors. These data
are necessary to evaluate the benefit-risk of individual NSAIDs for clinical
and health policy decision making. Within the European Communitys Seventh
Framework Programme, the Safety Of non-Steroidal anti-inflammatory drugs
(NSAIDs) [SOS] project aims to develop decision models for regulatory and
clinical use of individual NSAIDs according to their GI and cardiovascular
safety.
Objective:The aim of this study was to conduct a systematic review and meta-analysis of observational studies to provide summary relative risks (RR) of
upper GI complications (UGIC) associated with the use of individual
NSAIDs, including selective cyclooxygenase-2 inhibitors.
Methods:We used the MEDLINE database to identify cohort and case-control
studies published between 1 January 1980 and 31 May 2011, providing adjusted
effect estimates for UGIC comparing individual NSAIDs with non-use of
NSAIDs. We estimated pooled RR and 95% CIs of UGIC for individual
NSAIDs overall and by dose using fixed- and random-effects methods. Sub-
group analyses were conducted to evaluate methodological and clinical
heterogeneity between studies.Results:A total of 2984 articles were identified and 59 were selected for data
abstraction. After review of the abstracted information, 28 studies met the
SYSTEMATIC REVIEW Drug Saf 2012; 35 (12): 1127-1146
0114-5916/12/0012-1127
Adis 2012 Castellsague et al., publisher and licensee Springer International Publishing AG.This is an open access article published under the terms of the Creative Commons License
Attribution-NonCommercial-NoDerivative 3.0 (http://creativecommons.org/licenses/by-nc-nd/3.0/)which permits non-commercial use, distribution, and reproduction,
provided the original work is properly cited and not altered.
-
7/21/2019 art%3A10.1007%2FBF03261999
2/20
meta-analysis inclusion criteria. Pooled RR ranged from 1.43 (95%CI 0.65,
3.15) for aceclofenac to 18.45 (95%
CI 10.99, 30.97) for azapropazone. RRwas less than 2 for aceclofenac, celecoxib (RR 1.45; 95% CI 1.17, 1.81) and
ibuprofen (RR 1.84; 95% CI 1.54, 2.20); 2 to less than 4 for rofecoxib (RR
2.32; 95% CI 1.89, 2.86), sulindac (RR 2.89; 95% CI 1.90, 4.42), diclofenac
(RR 3.34; 95% CI 2.79, 3.99), meloxicam (RR 3.47; 95% CI 2.19, 5.50), ni-
mesulide (RR 3.83; 95% CI 3.20, 4.60) and ketoprofen (RR 3.92; 95% CI 2.70,
5.69); 45 for tenoxicam (RR 4.10; 95%CI 2.16, 7.79), naproxen (RR 4.10;
95%CI 3.22, 5.23), indometacin (RR 4.14; 95%CI 2.91, 5.90) and diflunisal
(RR 4.37; 95% CI 1.07, 17.81); and greater than 5 for piroxicam (RR 7.43;
95%CI 5.19, 10.63), ketorolac (RR 11.50; 95%CI 5.56, 23.78) and azapro-
pazone. RRs for the use of high daily doses of NSAIDs versus non-use were23 times higher than those associated with low daily doses.
Conclusions:We confirmed variability in the risk of UGIC among individual
NSAIDs as used in clinical practice. Factors influencing findings across
studies (e.g. definition and validation of UGIC, exposure assessment, anal-
ysis of new vs prevalent users) and the scarce data on the effect of dose and
duration of use of NSAIDs and on concurrent use of other medications need
to be addressed in future studies, including SOS.
1. Background
NSAIDs are widely used for the symptomatictreatment of acute pain and chronic inflammatoryand degenerative joint diseases. However, their useis restricted by the occurrence of upper gastro-intestinal (GI) complications (UGIC) such as pep-tic ulcer perforations, obstructions and bleeding.The use of NSAIDs has been associated with a 3- to5-fold increase in the risk of UGIC.[1,2] Clinicaltrials and observational studies have shown that
the use of selective cyclooxygenase (COX)-2 inhibi-tors is associated with a lower risk of UGIC;[3-5]
however, they have been also associated withan increased risk of serious cardiovascular (CV)events.[6] Further data are necessary to quantify therisk of UGIC associated with many individualNSAIDs, including selective COX-2 inhibitors, andto evaluate the benefit-risk balance of the NSAIDsmost often used in regular clinical practice, takinginto account dose, duration and effect of other riskfactors. These data can help clinicians select
treatments for individual patients and help healthpolicy regulators assess the public health impactof therapy.
Within the European Communitys SeventhFramework Programme, the Safety Of non-Steroidal anti-inflammatory drugs (SOS) colla-borative project started in 2008 with the goal ofdeveloping statistical and decision models to fa-cilitate regulatory and treatment decisions basedon the GI and CV safety of individual NSAIDs.One of the initial tasks of the SOS project was tosummarize the data available on the risk of GIand CV events from observational studies. In thiscontext, we conducted a systematic review and
meta-analysis of published observational studiesto provide pooled relative risks (RR) for UGICassociated with the use of individual NSAIDsversus non-use of NSAIDs. We followed theMOOSE guidelines for reporting meta-analysesof observational studies (http://www.equator-network.org/resource-centre/).
2. Materials and Methods
We performed a literature search in PubMedusing medical subject headings (MeSH) and free-text terms for individual NSAIDs and selective
1128 Castellsague et al.
Adis 2012 Castellsague et al., publ ish er and l icensee Springer In tern ationa l Publ ish ing AG. Drug Saf 2012; 35 (12)
-
7/21/2019 art%3A10.1007%2FBF03261999
3/20
COX-2 inhibitors, GI disease, case-control stud-ies and cohort studies. The search was restrictedto observational studies published in the Englishlanguage between 1 January 1980 and 31 May 2011.Details of the search strategy are available inthe supplemental digital content (SDC; http://links.adisonline.com/DSZ/A78). Studies had to be(i) cohort, case-control or nested case-controlstudies; (ii) provide odds ratios or RRs of UGICcomparing individual NSAIDs with non-use ofNSAIDs; and (iii) provide effect estimates ad-
justed at least for age and sex. All titles and/or
abstracts of the articles identified were reviewedto select those potentially meeting the inclusioncriteria. Data from these articles were abstractedin a standardized database that included informa-tion on source population, inclusion and exclusioncriteria, study design, case definition and valida-tion, selection of controls, exposure definition,confounding factors and statistical analysis. Theaccuracy of the abstracted data was reviewedindependently by two of the authors (NR-G,JC). References from relevant studies and prior
meta-analyses were also reviewed. Study authorswere contacted when additional information wasneeded.[7]
The methodological quality of each study wasevaluated using the Newcastle-Ottawa Scale(NOS).[8] The NOS involves a score system inwhich the study design is evaluated on threebroad categories: (i) selection of the study groups;(ii) comparability between the study groups; and(iii) exposure/outcome ascertainment. For eachstudy, the NOS was evaluated independently by
two of the authors (NR-G and JC), and any dif-ferences were resolved by consensus.
We estimated pooled RRs for those individualNSAIDs that had effect estimates reported in atleast three different studies. Pooled RRs and 95%CIs were estimated using both the inverse-varianceLagrange fixed-effects method and the DerSimonianand Laird random-effects method.[9] We gener-ated forest plots from the random-effects models.Heterogeneity between studies was assessed bygraphical inspections of the forest plots and by
Cochrans Chi-squared (w2
) test of homogeneity,and subgroup analyses evaluating methodologicaland clinical heterogeneity between studies. Sub-
group analyses included stratification by studydesign, prior history of UGIC, bleeding complica-tions, study period and dose of NSAIDs. Pooledestimates for dose were calculated according tothe dose categorization used in each study. In thesubgroup analyses, pooled RRs were also calcu-lated for those NSAIDs with only two effect es-timates available. The Higgins inconsistency I2
statistic was used to describe the percentage of thevariability in effect estimates that is due toheterogeneity rather than chance.[10] The meta-analysis was conducted using Review Manager
(RevMan), Version 5.0.22 (The Nordic CochraneCentre, The Cochrane Collaboration, Copenhagen,Denmark, 2009).
3. Results
A total of 2984 articles on NSAIDs and GIcomplications were identified. Of these, 2974 ar-ticles were identified in the PubMed search andten additional articles were identified through thereferences of relevant studies (figure 1). The re-
view of titles and abstracts of these studies led toselect 59 articles for full data abstraction. Afterreview of the abstracted information, 28 stud-ies on the use of individual NSAIDs and the riskof UGIC met the inclusion criteria and were in-cluded in the meta-analysis.[7,11-35] The remaining31 articles were excluded for the following reasons:the reference group was other than non-use ofNSAIDs in nine studies;[36-44] the outcome wasoverall upper and lower GI complications in threestudies;[45-47] the outcome was uncomplicated upper
GI events in two studies;[48,49]
the study populationwas restricted to users of specific drugs or to patientswith specific diseases in three studies;[50-52] the studypopulation and the study period overlapped in fourstudies;[53-56] and the study design did not meet theinclusion criteria in ten studies (i.e. different typeof study or measures of association and exposureassessment).[57-66]
Selected characteristics of the 28 studies includedin the meta-analysis are summarized in table I; 3studies were cohort studies,[20,24,26] 10 were nested
case-control studies,[7,11,17-19,23,25,33,35,67] and 15were case-control studies.[12-16,21,22,27-32,34,68] Twelvestudies, all case-control studies, were field studies
Individual NSAIDs and Upper GI Complications (SOS Project) 1129
Adis 2012 Castellsague et al., publ ish er and l icensee Springer In tern ationa l Publ ish ing AG. Drug Saf 2012; 35 (12)
-
7/21/2019 art%3A10.1007%2FBF03261999
4/20
collecting individual information by standardized
questionnaires. The 16 remaining studies usedinformation recorded in healthcare databases.Cases were defined as hospitalization or re-
ferral to a specialist for upper GI bleeding in13 studies,[12,15,16,19-22,27-30,33,34] and for bleedingand/or perforation in 15 other studies; four studiesalso included cases of uncomplicated pepticulcer.[14,17,33,68] The site of complication was de-fined as gastric and/or duodenal in all studies,and two included oesophageal complica-tions.[12,32] Most studies, both field and database
studies, required information from endoscopy orother diagnostic procedures to confirm UGIC.Six studies conducted in healthcare databases didnot conduct any validation of the cases identi-fied.[14,17-20,68] Sixteen studies reported aggregateresults for patients with and without a history ofUGIC,[7,12-14,16,20-23,28,31-34,67,68] and 12 provid-ed results for patients without a history ofUGIC;[11,15,17-19,24-27,29,30,35] the remaining studywas a case-crossover study.[14] Most studies ex-cluded subjects with a history of a known cause of
UGIC, including the use of gastrotoxic medicationsand life-threatening diseases (table II). Four stud-ies did not have any exclusion criteria.[20,22,28,34]
Among the 14 case-control studies, seven included
hospital controls;[12,13,16,21,29-31]
five included bothhospital and community controls;[22,27,28,32,34] oneincluded community controls;[15] and two werecase-crossover studies.[14,68] All case-control stud-ies were matched on age, sex (except one study[16]),hospital or geographic area, and index date. Threeof the case-control studies with hospital and com-munity controls estimated separate results for eachset of controls;[22,28,34] as results between the twosets were similar, we included in the meta-analysisresults reported using hospital controls. In cohort
studies, current use of NSAIDs was defined asthe time covered by each prescription,[20,24] andone study extended the coverage by 15 days. [26]
Most case-control studies defined current use ofNSAIDs as any use ending at the index date orwithin 7 days before the index date. A few case-control studies considered current use as thatending up to 30 days[11,13,17,19,28,35] or 90 days[18]
before the index date. Two cohort studies focusedon new users of NSAIDs,[20,26] and one nestedcase-control study provided results for both new
users and all users (incident and prevalent) ofNSAIDs.[17] In addition to age and sex, the mostfrequent confounders considered were a history
Articles identified with
PubMed search strategy
on GI events(n =2974)
Articles retrievedthrough
cross-referencing(n =10)
Articles screened
(n =2984)
Articles excluded after reviewing
titles and abstracts
(n =2925)
Articles excluded (n =31)
Other reference category (n =9) Upper and lower GI complications combined (n =3)
Uncomplicated upper GI events (n =2)
Specific populations (n =3) Same population, overlapping period (n =4) Different type of study (n =10)
Articles included in qualitative synthesis
(n =59)
Articles included in the meta-analysis(n =28)
Fig. 1. Flow diagram for identification of studies on upper gastrointestinal complications and individual NSAIDs. GI =gastrointestinal.
1130 Castellsague et al.
Adis 2012 Castellsague et al., publ ish er and l icensee Springer In tern ationa l Publ ish ing AG. Drug Saf 2012; 35 (12)
-
7/21/2019 art%3A10.1007%2FBF03261999
5/20
Table
I.
Descriptionofobservationalstudiesontheriskofuppergastrointestinalcomplica
tionsassociatedwiththeuseofindividua
lNSAIDs
Study,y
Source
population;
studyperiod
Study
popula
tion
(N);ag
e(y)
Study
design
Typeof
endpoint
Studyen
dpoint
Case
validation
Dose
assessment
Exposure
assessment
Definitionof
curren
tusea
Mamdanietal.,
2002[20]
Ontario,
Canada;
20001
14396
9;
>65
Cohort
All
Bleeding
Site:gas
tric,duodenal
Lesion:pepticulcer,upperGIb
No
No
Dispensed
prescriptions
Prescription
covera
ge
Menniti-Ippolitoetal.,
1998[24]
Umbria,Italy;
19934
20135
7;
3584
Cohort
Incident
Bleeding
,perforation
Site:gas
tric,duodenal
Lesion:pepticulcer,upperGIb
Review
medical
records
No
Dispensed
prescriptions
Estima
ted
prescription
covera
ge
McMahonetal.,
1997[26]
Tayside,
Scotland;
1989
15639
8;
All
Cohort
Incident
Bleeding
,perforation
Site:gas
tric,duodenal
Lesion:upperGIb
Review
medical
records
No
Dispensed
prescriptions
Prescription
covera
ge
plus15days
Helin-Salmivaara
etal.,2007[17]
Finland;
20004
50971
;
All
Nested
case-
control
Incident
Bleeding
,perforation,ulcer
Site:gas
tric,duodenal
Lesion:pepticulcer
No
No
Dispensed
prescriptions
030d
ays
Hippisley
-Coxetal.,
2005[18]
QResearch,
UK;
20004
98274
;
25+
Nested
case-
control
Incident
Bleeding
,perforation
Site:gas
tric,duodenal
Lesion:pepticulcer,upperGIb
No
No
Electronic
prescriptions
090d
ays
Nrgardetal.,
2004[19]
NorthJutland,
Denmark;
20002
3686;
1890
Nested
case-
control
Incident
Bleeding
Site:gas
tric,duodenal
Lesion:pepticulcer,gastritis,
upperGIb
No
No
Dispensed
prescriptions
030d
ays
Castellsa
gueetal.,
2009[67]
Saskatchewan,
Canada;
19992001
20728
;
2089
Nested
case-
control
All
Bleeding
,perforation
Site:gas
tric,duodenal
Lesion:pepticulcer,upperGIb
Review
medical
records
Yes
Electronic
prescriptions
07da
ys
GarcaRodrguez
andBarreales,
2007[7]
THIN,UK;
20005
11561
;
4085
Nested
case-
control
All
Bleeding
,perforation
Site:gas
tric,duodenal
Lesion:pepticulcer,erosions,
inflammation
Review
computerized
information
andfreetext,
andrandom
sample
medical
records
Yes
Electronic
prescriptions
07da
ys
GarcaRodrguez
andHern
andez-Daz,
2001[23]
GPRD,UK;
19938
13605
;
4079
Nested
case-
control
All
Bleeding
,perforation
Site:gas
tric,duodenal
Lesion:pepticulcer,upperGIb
Review
medical
records
No
Electronic
prescriptions
06da
ys
GarcaRodrguez
etal.,1998[25]
FriuliVenezia
Giulia,Italy;
19915
21505
;
2589
Nested
case-
control
Incident
Bleeding
,perforation
Site:gas
tric,duodenal
Lesion:pepticulcer,upperGIb
Review
medical
records
Yes
Dispensed
prescriptions
0days
(on
dateof
event)
Continuednext
page
Individual NSAIDs and Upper GI Complications (SOS Project) 1131
Adis 2012 Castellsague et al., publ ish er and l icensee Springer In tern ationa l Publ ish ing AG. Drug Saf 2012; 35 (12)
-
7/21/2019 art%3A10.1007%2FBF03261999
6/20
TableI.C
ontd
Study,y
Source
population;
studyperiod
Study
popula
tion
(N);ag
e(y)
Study
design
Typeof
endpoint
Studyen
dpoint
Case
validation
Dose
assessment
Exposure
assessment
Definitionof
curren
tusea
GarcaRodrguez
andJick,
1994[11]
GPRD,UK;
19903
9879;
2577
Nested
case-
control
Incident
Bleeding
,perforation
Site:gas
tric,duodenal
Lesion:pepticulcer,upperGIb
Review
random
sample
medical
records
Yes
Electronic
prescriptions
030d
ays
Griffinet
al.,1991[33]
Tennessee
Medicaid,US;
19846
8478;
65+
Nested
case-
control
All
Bleeding
,ulcer
Site:gas
tric,duodenal,
Lesion:pepticulcer,upperGIb
Review
medical
records
Yes
Dispensed
prescriptions
0days
(on
dateof
event)
Perez-Gu
tthann
etal.,1997[35]
Saskatchewan,
Canada;
19826
11377
;
All
Nested
case-
control
Incident
Bleeding
,perforation
Site:gas
tric,duodenal
Lesion:pepticulcer,upperGIb
Review
medical
records
Yes
Dispensed
prescriptions
030d
ays
Changet
al.,2011[68]
Taiwan;2006
23million;
20+
Case-
crossover
All
Bleeding
,perforation,ulcer
Site:gas
tric,duodenal
Lesion:pepticulcer,gastritis,
duodenitis,upperGIb
No
Yes
Electronic
prescriptions
Prescription
covera
ge
Sakamotoetal.,
2006[15]
Selected
hospitalsin
Japan;
20024
522;
40+
Case-
control
Incident
Bleeding
Site:gas
tric,duodenal
Lesion:pepticulcer,gastritis
Fieldstudy.
Endoscopy
required
No
Questionnaire
07da
ys
Lanaset
al.,2006[16]
Selected
hospitalsin
Spain;
20014
3353;
2085
Case-
control
All
Bleeding
Site:gas
tric,duodenal
Lesion:pepticulcer
Fieldstudy.
Endoscopy
required
Yes
Questionnaire
07da
ys
Laportee
tal.,
2004[21]
Selected
hospitalsin
SpainandItaly;
19982001
10006
;
>18
Case-
control
All
Bleeding
Site:gas
tric,duodenal
Lesion:pepticulcer,acute
lesionso
fthegastricmucosa,
erosiveduodenitis,mixed
lesions
Fieldstudy.
Endoscopy
required
Yes
Questionnaire
07da
ys
Biskupiaketal.,
2006[14]
Outpatient
database,US;
19962004
60212
;
All
Case-
crossover
All
Bleeding
,perforation,ulcer
Site:gas
tric,duodenal
Lesion:pepticulcer,upperGIb
No
Lowdose
Electronic
prescriptions
90day
s
following
prescription
date
Continuednext
page
1132 Castellsague et al.
Adis 2012 Castellsague et al., publ ish er and l icensee Springer In tern ationa l Publ ish ing AG. Drug Saf 2012; 35 (12)
-
7/21/2019 art%3A10.1007%2FBF03261999
7/20
TableI.C
ontd
Study,y
Source
population;
studyperiod
Study
popula
tion
(N);ag
e(y)
Study
design
Typeof
endpoint
Studyen
dpoint
Case
validation
Dose
assessment
Exposure
assessment
Definitionof
curren
tusea
Lanaset
al.,2003[22]
Selected
hospitalsin
Spain;
19958
8309;
All
Case-
control
All
Bleeding
Site:gas
tric,duodenal
Lesion:pepticulcer,erosions,
acutelesionsofthegastric
mucosa
Fieldstudy.
Endoscopy
required
No
Questionnaire
07da
ys
Matikaine
nand
Kangas,1996[12]
Twohospitalsin
Finland;
19923
204;
All
Case-
control
All
Bleeding
Site:oes
ophageal,gastric,
duodena
l
Lesion:oesophagitis,peptic
ulcer,erosions
Fieldstudy.
Endoscopy
required
No
Questionnaire
07da
ys
Morideand
Abenhaim
,1994[13]
Selected
hospitalsin
Montreal,
Canada;
198890
859;
68+
Case-
control
All
Bleeding
,perforation
Site:gas
tric,duodenal
Lesion:pepticulcer,gastritis,
duodenitis,upperGIb
Review
medical
records
No
Dispensed
prescriptions
030d
ays
Kaufman
etal.,
1993[27]
US,
Sweden,
Hungary;
198792
1733;
1879
Case-
control
Incident
Bleeding
Site:gas
tric,duodenal
Lesion:pepticulcer,gastritis
Fieldstudy.
Endoscopy
required
No
Questionnaire
07da
ys
Langman
etal.,
1994[28]
Selected
hospitalsinthe
UK;
198791
3259;
60+
Case-
control
All
Bleeding
Site:gas
tric,duodenal
Lesion:pepticulcer
Fieldstudy.
Endoscopy
required
No
Questionnaire
030d
ays
Nobilietal.,1992[29]
Hospitalin
NorthernItaly;
19878
1764;
All
Case-
control
Incident
Bleeding
Site:gas
tric,duodenal
Lesion:pepticulcer,upperGIb
Fieldstudy.
Endoscopy
required
No
Questionnaire
07da
ys
Laportee
tal.,
1991[30]
Selected
hospitalsin
Spain;
19878
3557;
All
Case-
control
Incident
Bleeding
Site:gas
tric,duodenal
Lesion:pepticulcer,acute
lesionso
fthegastricmucosa,
erosiveduodenitis
Fieldstudy.
Endoscopy
required
No
Questionnaire
07da
ys
Savagee
tal.,
1993[31]
Selected
hospitalsin
Christchurch,
NewZealand;
1986
1466
All
Case-
control
All
Bleeding
,perforation
Site:gas
tric,duodenal
Lesion:pepticulcer
Fieldstudy.
Endoscopy
required
No
Questionnaire
07da
ys
Continuednext
page
Individual NSAIDs and Upper GI Complications (SOS Project) 1133
Adis 2012 Castellsague et al., publ ish er and l icensee Springer In tern ationa l Publ ish ing AG. Drug Saf 2012; 35 (12)
-
7/21/2019 art%3A10.1007%2FBF03261999
8/20
of peptic ulcer (21 studies),[7,11,15-31,35,67] smoking(13 studies),[7,11,15,18,20,21,23,27-32] alcohol use (9 stud-ies),[7,15,19,21,27,28,30-32]use of proton-pump inhibitorsand anti-ulcer medications (11 studies),[16-23,30,67,68]
and concurrent use of medications increasing the riskof UGIC (16 studies).[7,11,15,16,19-21,23,27,29-31,33,35,67,68]
The quality of the studies measured with the NOSwas, in general, very good: for the selection com-ponent, 12 studies had the maximum score of4[7,11,15,22-25,32,33,35,67,68] and 13 studies had thenext highest score of 3;[13,14,16-19,21,26,28,29,30,32,34]
for comparability, 24 studies had the maximum
score of 2;[7,11,14-19,21-33,35,67,68]
and for the exposure/
outcome component, 14 studies had the maximumscore of 3[7,11,13,17-19,23-26,33,35,67,68] and 10 studieshad the next highest score of 2.[12,14,16,20,21,28-30,32,34]
The studies included in the meta-analysis al-lowed us to estimate pooled RRs for the current useof 16 different NSAIDs (table III and figure 2).Forest plots for each individual NSAID are avail-able in the SDC. Using random-effects models,pooled RRs ranged from 1.43 (95%CI 0.65, 3.15)for aceclofenac to 18.45 (95%CI 10.99, 30.97) for
azapropazone. Pooled RR was less than 2 for ace-clofenac, celecoxib and ibuprofen; between 2 andless than 4 for rofecoxib, sulindac, diclofenac, me-loxicam, nimesulide and ketoprofen; between 4 andless than 5 for tenoxicam, naproxen, indometacinand diflunisal; and greater than 5 for piroxicam,ketorolac and azapropazone. Pooled RRs fromstudies providing results for patients without ahistory of UGIC were similar to those from theoverall analysis except for naproxen (more than10% change), 3.10 (95% CI 2.45, 3.91) and diclo-
fenac, 3.76 (95% CI 2.71, 5.21). Data in patientswith a history of peptic ulcer disease were availableonly for celecoxib (two studies) and rofecoxib (onestudy).[67,68] The pooled RR for celecoxib in thispopulation was 1.50 (95%CI 1.16, 1.94).
Pooled RRs from case-control studies werehigher than those from cohort studies for allNSAIDs except ibuprofen, ketorolac and sulindac.In general, pooled RRs from fixed-effects modelswere slightly lower than those from random-effects models. In general, there was significant
heterogeneity between studies, which decreasedin the subsequent subgroup analysis exploringmethodological and clinical diversity.Ta
bleI.C
ontd
Study,y
Source
population;
studyperiod
Study
popula
tion
(N);ag
e(y)
Study
design
Typeof
endpoint
Studyen
dpoint
Case
validation
Dose
assessment
Exposure
assessment
Definitionof
curren
tusea
Henryet
al.,1993[32]
Selected
hospitalsin
Newcastle,
Australia;
19859
1912;
All
Case-
control
All
Bleeding
,perforation
Site:oes
ophageal,gastric,
duodena
l
Lesion:pepticulcer,erosions,
inflammation,upperGIb
Fieldstudy.
Endoscopy
required
No
Questionnaire
07da
ys
Somervilleetal.,
1986[34]
Nottingham
Universityand
CityHospitals,
UK;
19835
667;
60+
Case-
control
All
Bleeding
Site:gas
tric,duodenal
Lesion:pepticulcer
Fieldstudy.
Endoscopy
required
No
Questionnaire
0days
(on
dateof
event)
a
Numb
erofdaysbeforetheindexdateifnotde
finedotherwise.
b
Uppe
rGIreferstodiagnosisofmelena,haem
atemesisorhaemorrhageoftheintestinaltractunspecified.
GI=gastrointestinal;GPRD=GeneralPracticeRe
searchDatabase;THIN=TheHealthImprovementNetwork.
1134 Castellsague et al.
Adis 2012 Castellsague et al., publ ish er and l icensee Springer In tern ationa l Publ ish ing AG. Drug Saf 2012; 35 (12)
-
7/21/2019 art%3A10.1007%2FBF03261999
9/20
Table
II.
Exclusionsappliedinobservationalstud
iesontheriskofuppergastrointestinalc
omplicationsa
Study,y
Priorpeptic
ulcerdisease
Malignancy
Oesophageal
varices
Mallory-
Weiss
syndrome
Alcohol-
related
disorders
Liver
diseases
Coagulopathies
Useof
anticoagulants
Nursing
home
residence
O
ther
Mamdanietal.,2002[20]
Menniti-Ippolitoetal.,
1998[24]
Excl.
Exc
l.
Excl.
Excl.
Excl.
McMahonetal.,1997[26]
Excl.
E
xcl.b
Helin-Salmivaaraetal.,
2007[17]
Excl.
Exc
l.
Excl.
Excl.
Hippisley
-Coxetal.,
2005[18]
Excl.
Nrgardetal.,2004[19]
Excl.
Castellsa
gueetal.,2011[67]
Exc
l.
Excl.
Excl.
Excl.
Excl.
Excl.
E
xcl.c
GarcaRodrguezand
Barreales,2007[7]
Exc
l.
Excl.
Excl.
Excl.
Excl.
Excl.
GarcaRodrguezand
Hernandez-Daz,2001[23]
Exc
l.
Excl.
Excl.
Excl.
Excl.
Excl.
GarcaRodrguezetal.,
1998[25]
Excl.
Exc
l.
Excl.
Excl.
Excl.
Excl.
Excl.
E
xcl.d
GarcaRodrguezandJick,
1994[11]
Excl.
Exc
l.
Excl.
Excl.
Excl.
Excl.
Excl.
Griffinet
al.,1991[33]
Exc
l.e
Excl.e
Perez-Gu
tthannetal.,
1997[35]
Excl.
Exc
l.
Excl.
Excl.
Excl.
Excl.
Excl.
Changet
al.,2011[68]
Exc
l.
Excl.
Sakamotoetal.,2006[15]
Excl.
Exc
l.
Excl.
Excl.
Excl.
Lanaset
al.,2006[16]
Exc
l.
Excl.
Excl.
Laportee
tal.,2004[21]
Exc
l.
Excl.
Excl.
Excl.
Excl.
Biskupiaketal.,2006[14]
Exc
l.
Excl.
E
xcl.f
Lanaset
al.,2003[22]
Matikaine
nandKangas,
1996[12]
Exc
l.
Excl.
Excl.
E
xcl.g
MorideandAbenhaim,
1994[13]
Exc
l.
Excl.
Excl.
Kaufman
etal.,1993[27]
Excl.
Excl.
Excl.
Excl.
Excl.
Langman
etal.,1994[28]
Nobilietal.,1992[29]
Excl.
Exc
l.
Excl.
Excl.
Excl.
Excl.
Excl.
Continuednext
page
Individual NSAIDs and Upper GI Complications (SOS Project) 1135
Adis 2012 Castellsague et al., publ ish er and l icensee Springer In tern ationa l Publ ish ing AG. Drug Saf 2012; 35 (12)
-
7/21/2019 art%3A10.1007%2FBF03261999
10/20
Pooled RRs for the effect of daily dose wereestimated for eight different NSAIDs. Cut-offvalues used in each study to define the daily doseof each NSAID are presented in table IV. Varia-tions in cut-off values were, in general, small ex-cept for those used for ibuprofen (200 mg) andnaproxen (220 mg) in one study,[14] and for ibu-profen (
-
7/21/2019 art%3A10.1007%2FBF03261999
11/20
-
7/21/2019 art%3A10.1007%2FBF03261999
12/20
TableIII.
Contd
Study,y
NSAID
Ibuprofen
Naproxen
Diclofenac
Indometacin
Piroxicam
Ketoprofen
PooledR
R
Random
effects
1.60(1.09
,2.33)
4.73(2.92,7.68)
3.37(2.55,
4.46)
4.24(2.42,7.45)
8.32(4.32,16.03)
4.25(2.06,8.76)
Fixedeffects
1.74(1.60
,1.88)
2.87(2.54,3.25)
2.36(2.23,
2.50)
2.27(1.94,2.66)
3.62(3.21,4.07)
2.35(1.97,2.79)
Heteroge
neity(w2
p-value)