`
`Rostom A, Dube C, Wells GA, Tugwell P, Welch V, Jolicoeur E, McGowan J, Lanas A
`
`This is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library
`2011, Issue 6
`
`http://www.thecochranelibrary.com
`
`Prevention of NSAID-induced gastroduodenal ulcers (Review)
`Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
`
`
`
`T A B L E O F C O N T E N T S
`
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`HEADER .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`ABSTRACT .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`PLAIN LANGUAGE SUMMARY
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`BACKGROUND .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`OBJECTIVES
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`METHODS .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`RESULTS .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`Figure 1.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`Figure 2.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`DISCUSSION .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`AUTHORS’ CONCLUSIONS
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`ACKNOWLEDGEMENTS
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`REFERENCES .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`CHARACTERISTICS OF STUDIES
`.
`.
`.
`.
`.
`.
`.
`.
`.
`DATA AND ANALYSES .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`FEEDBACK .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`WHAT’S NEW .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`HISTORY .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`CONTRIBUTIONS OF AUTHORS
`.
`.
`.
`.
`.
`.
`.
`.
`.
`DECLARATIONS OF INTEREST .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`SOURCES OF SUPPORT .
`.
`.
`.
`DIFFERENCES BETWEEN PROTOCOL AND REVIEW .
`NOTES .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`INDEX TERMS
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`1
`1
`2
`2
`3
`3
`5
`7
`8
`11
`13
`13
`13
`19
`49
`58
`59
`59
`60
`61
`61
`61
`61
`61
`
`Prevention of NSAID-induced gastroduodenal ulcers (Review)
`Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
`
`i
`
`
`
`[Intervention Review]
`
`Prevention of NSAID-induced gastroduodenal ulcers
`
`Alaa Rostom1, Catherine Dube1, George A Wells2, Peter Tugwell3, Vivian Welch4, Emilie Jolicoeur5, Jessie McGowan6, Angel Lanas7
`
`1University of Calgary Department of Medicine / Gastroenterology, Foothills Medical Centre, Calgary, Canada. 2Department of
`Epidemiology and Community Medicine, University of Ottawa, Ottawa, Canada. 3Centre for Global Health, Institute of Population
`Health, Faculty of Medicine, University of Ottawa, Ottawa, Canada. 4Centre for Global Health, Institute of Population Health,
`University of Ottawa, Ottawa, Canada. 5Division of Gastroenterology, A-1 Endoscopy Unit, Ottawa, Canada. 6Institute of Population
`Health/Ottawa Health Research Institute, University of Ottawa, Ottawa, Canada. 7University of Zaragoza, Zaragoza, Spain
`
`Contact address: Alaa Rostom, University of Calgary Department of Medicine / Gastroenterology, Foothills Medical Centre, 1403 -
`29 St. N.W, Calgary, Alberta, T2N 2T9, Canada. arostom@ucalgary.ca.
`
`Editorial group: Cochrane Upper Gastrointestinal and Pancreatic Diseases Group.
`Publication status and date: Edited (no change to conclusions), published in Issue 6, 2011.
`Review content assessed as up-to-date: 11 May 2009.
`
`Citation: Rostom A, Dube C, Wells GA, Tugwell P, Welch V, Jolicoeur E, McGowan J, Lanas A. Prevention of NSAID-induced gastro-
`duodenal ulcers. Cochrane Database of Systematic Reviews 2002, Issue 4. Art. No.: CD002296. DOI: 10.1002/14651858.CD002296.
`
`Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
`
`Background
`
`A B S T R A C T
`
`Non-steroidal anti-inflammatory drugs (NSAIDs) are important agents in the management of arthritic and inflammatory conditions,
`and are among the most frequently prescribed medications in North America and Europe. However, there is overwhelming evidence
`linking these agents to a variety of gastrointestinal (GI) toxicities.
`
`Objectives
`
`To review the effectiveness of common interventions for the prevention of NSAID induced upper GI toxicity.
`
`Search methods
`
`We searched MEDLINE from 1966 to May 2009, Current Contents for six months prior to May 2009, EMBASE to May 2009, and
`the Cochrane Controlled Trials Register from 1973 to May 2009. Recent conference proceedings were reviewed and content experts
`and companies were contacted.
`
`Selection criteria
`
`Randomized controlled clinical trials (RCTs) of prostaglandin analogues (PA), H2-receptor antagonists (H2RA) or proton pump
`inhibitors (PPI) for the prevention of chronic NSAID induced upper GI toxicity were included.
`
`Data collection and analysis
`
`Two independent authors extracted data regarding population characteristics, study design, methodological quality and number of
`participants with endoscopic ulcers, ulcer complications, symptoms, overall drop-outs, drop outs due to symptoms. Dichotomous data
`were pooled using RevMan 5.0. Heterogeneity was evaluated using a chi square test, and the I square statistic.
`
`Prevention of NSAID-induced gastroduodenal ulcers (Review)
`Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
`
`1
`
`
`
`Main results
`
`Forty-one RCTs met the inclusion criteria. All doses of misoprostol significantly reduced the risk of endoscopic ulcers. Misoprostol
`800 ug/day was superior to 400 ug/day for the prevention of endoscopic gastric ulcers (RR 0.17, and RR 0.39 respectively, P=0.0055).
`A dose response relationship was not seen with duodenal ulcers. Misoprostol caused diarrhoea at all doses, although significantly more
`at 800 ug/day than 400 ug/day (P=0.0012). Misoprostol also reduced the risk of clinical ulcer complications.
`
`Standard doses of H2RAs were effective at reducing the risk of endoscopic duodenal (RR 0.36; 95% CI 0.18 to 0.74) but not gastric
`ulcers (RR 0.73; 95% CI 0.50 to 1.08). Both double dose H2RAs and PPIs were effective at reducing the risk of endoscopic duodenal
`and gastric ulcers (RR 0.44; 95% CI 0.26 to 0.74) and RR=0.40;95% CI; 0.32-0.51 respectively for gastric ulcer), and were better
`tolerated than misoprostol.
`
`Authors’ conclusions
`
`Misoprostol, PPIs, and double dose H2RAs are effective at preventing chronic NSAID related endoscopic gastric and duodenal ulcers.
`Lower doses of misoprostol are less effective and are still associated with diarrhoea. In patients with previous NSAID bleeds, a COX-2
`inhibitor alone is equivalent to a tNSAID+PPI, though the re-bleeding rates with both strategies are still relatively high. A strategy of
`a COX-2 inhibitor+PPI appears to offer the greatest GI safety in high risk patients.
`
`P L A I N L A N G U A G E S U M M A R Y
`
`Medications to prevent NSAID-induced gastroduodenal ulcers
`
`The results of this meta-analysis demonstrate that misoprostol, proton pump inhibitors, and double doses of H2-receptor antagonists
`are effective at reducing the risk of both gastric and duodenal non steroidal anti-inflammatory (NSAID) medications induced ulcers.
`In high risk patients, the use of a traditional NSAID + PPI appears equivalent to a COX-2 inhibitor alone. The most effective strategy
`in high risk GI patients appears to be the combination of a COX-2 inhibitor + PPI.
`
`B A C K G R O U N D
`
`Description of the condition
`
`Non-steroidal anti-inflammatory drugs (NSAIDs) are important
`agents in the management of arthritic and inflammatory condi-
`tions, and are among the most frequently prescribed medications
`in North America and Europe (Fries 1990; Wallace 1996). How-
`ever, there is overwhelming evidence linking these agents to a va-
`riety of gastrointestinal (GI) toxicities (Fries 1990; Stalnikowicz
`1993; Smalley 1996; Fries 1991; Griffin 1988; Bollini 1992;
`McMahon 1997; Gabriel 1991; Langman 1994; MacDonald
`1997; Armstrong 1987; Silverstein 1995). Common side effects
`such as nausea and dyspepsia correlate poorly with serious adverse
`GI events (Silverstein 1995; Larkai 1987). While endoscopic ul-
`cers can be documented in up to 40% of chronic NSAID users
`(Stalnikowicz 1993), it is estimated that as many as 85% of these
`never become clinically apparent (Silverstein 1995; Maetzel 1998).
`Serious NSAID induced GI complications such as haemorrhage,
`perforation or death are much less common, occurring collectively
`
`with an incidence of about 1.5% per year (Silverstein 1995). How-
`ever, the number of individuals prescribed NSAIDs and the po-
`tential for life-threatening adverse events make NSAID toxicity an
`important clinical and economic problem.
`
`Description of the intervention
`
`In the late 1990s, evidence from non-clinical and early clinical
`trials suggested that the gastrointestinal (GI) safety of the newer
`cyclooxygenase-2 (COX-2) selective NSAIDs may be such that a
`fundamental change in clinicians’ choice from the use of standard
`NSAIDs with a gastro-protective agent to monotherapy with a
`COX-2 selective NSAID (COX-2 inhibitors) was on the horizon.
`However, much has changed since then in the NSAID field.
`The release of the cyclo-oxygenase-2 selective inhibitors (COX-
`2s) brought about significant changes in the NSAID market place.
`Traditional nonselective NSAIDs (tNSAIDs) prescription num-
`bers fell rapidly, to be replaced by COX-2 prescriptions. Addition-
`ally, overall NSAID prescriptions rose in number suggesting that
`
`Prevention of NSAID-induced gastroduodenal ulcers (Review)
`Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
`
`2
`
`
`
`clinicians were starting COX-2s on patients who were not con-
`sidered candidates for tNSAIDs. The rise of COX-2s continued
`until 2004 when greater data regarding their cardiovascular and
`other toxicities became available, leading to the withdrawal of most
`of these agents from the market over the following years. Non-
`naproxen tNSAIDs were also suggested to have important car-
`diovascular toxicity, leading to considerable uncertainty amongst
`clinicians treating patients with arthritis and other pain disorders
`as to the choice of agent to use (Rostom 2009; Rostom 2009b).
`
`How the intervention might work
`
`NSAIDs are believed to cause gastroduodenal mucosal injury
`through their inhibition of mucosal prostaglandin production.
`Prostaglandins promote mucosal integrity through several mecha-
`nisms including: maintenance of mucosal blood flow; promoting
`mucosal bicarbonate formation; promoting mucosal mucus for-
`mation; and reducing mucosal acid secretion. Three intervention
`classes were assessed in this review: misoprostol; H2RAs; and PPIs.
`H2RAs and PPIs are believed to exert their gastro-protective ef-
`fects from NSAID gastroduodenal injury through the reduction of
`gastric acid secretion. Prostaglandin analogues such as misoprostol
`are believed to exert their gastro-protection by restoring mucosal
`prostaglandin effects (Rostom 2004).
`
`Why it is important to do this review
`
`NSAIDs are amongst the most commonly prescribed medica-
`tions worldwide, and are associated with important gastrointesti-
`nal harms. The introduction of COX-2’s with their greater GI
`safety resulted in important declines in tNSAID prescriptions.
`However, with the discovery of cardiovascular (CVS) and other
`adverse effects associated with COX-2s, practitioners are returning
`to prescribing tNSAIDs with a gastro-protective agent in an effort
`to overcome some of the adverse GI effects of tNSAIDs.
`
`O B J E C T I V E S
`
`The primary objective of this study was to systematically review
`the available literature on the effectiveness of the prostaglandin
`analogue (PA) misoprostol, H2-receptor antagonists (H2RA), and
`proton pump inhibitors (PPI) for the prevention of NSAID in-
`duced upper GI toxicity, among patients requiring chronic NSAID
`use. The secondary objectives were to review the effect of these
`agents on NSAID induced GI symptoms, and to assess the rela-
`tionship between the effectiveness of PAs at various doses and their
`associated drug induced adverse events.
`
`M E T H O D S
`
`Criteria for considering studies for this review
`
`Types of studies
`
`Randomized controlled trials were eligible for this meta-analysis.
`
`Types of participants
`
`Participants were eligible if they had taken NSAIDs for greater
`than 3 weeks and were enrolled for the prophylaxis of NSAID-
`induced ulcers.
`
`Types of interventions
`
`Interventions that were examined include: H2-antagonists, proton
`pump inhibitors and misoprostol each used for the prophylaxis of
`NSAID induced gastroduodenal ulcers.
`
`Types of outcome measures
`
`Primary outcomes
`
`For each study, the number of participants with: endoscopic ul-
`cers; ulcer complications (haemorrhage, perforation, pyloric ob-
`struction or death); symptoms (nausea, vomiting, dyspepsia, ab-
`dominal pain or diarrhoea); overall drop-outs; and drop outs due
`to symptoms were identified. Included studies were also classified
`into primary or secondary prophylaxis trials and by the time peri-
`ods of outcome measures.
`The primary outcomes were:
`• endoscopic ulcers (gastric, duodenal and gastroduodenal);
`• clinical ulcer complications.
`
`Secondary outcomes
`
`The secondary outcomes were:
`• symptoms;
`• drop-outs and drop outs due to symptoms
`
`Search methods for identification of studies
`
`Electronic searches
`
`Randomized controlled clinical trials (RCTs) of PA, H2RA or
`PPIs for the prevention of NSAID induced upper GI toxicity
`were identified in accordance with published recommendations
`(Haynes 1994; Hunt 1997). This included identification of articles
`
`Prevention of NSAID-induced gastroduodenal ulcers (Review)
`Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
`
`3
`
`
`
`through electronic databases including originally a MEDLINE
`search from 1966 to June 2002, Current Contents for 6 months
`prior to June 2002, EMBASE to May 2002, and a search of the
`Cochrane Controlled Trials Register from 1973 to 2002. These
`searches were updated from 2002 to Dec 2004 and then from
`2004 to May 2009.
`
`Searching other resources
`
`In addition to update searches by the Cochrane UGPD Group, the
`search strategy for this review was supplemented on two occasions:
`1. for the Canadian Coordinating Office for Health
`Technology Assessment (CCOHTA) review in 2003;
`2. for the Canadian Consensus conference on the use of ASA.
`NSAIDs, and COX-2 Inhibitors in 2007-2008.
`The CCOHTA search Description: A Dialog® OneSearch® on
`MEDLINE®, ToxFile, EMBASE®, BIOSIS Previews®, Phar-
`maceutical News Index (PNI)® and Current Contents Search®
`for published and scientific meeting literature was performed. The
`Cochrane Library was searched separately. The web sites for Inter-
`national Agencies for Health Technology Assessment (INAHTA)
`web sites, specialized databases (e.g. University of York National
`Health Service (NHS) Center for Reviews and Dissemination
`(CRD)) and Conference Papers Index, as well as internet searching
`(i.e. Google searching), were searched in order to identify health
`technology assessment reports, meeting abstracts, and other grey
`literature. Trial registries were searched for ongoing trials. Recent
`conference proceedings were consulted and content experts and
`companies were contacted. The reference lists of all potentially
`relevant articles including reviews were reviewed for the identi-
`fication of other potential studies. The search strategies used are
`shown in Appendix 1; Appendix 2; Appendix 3; Appendix 4.
`The search strategy used to supplement the Canadian NSAID
`consensus conference is listed in Appendix 5.
`
`Data collection and analysis
`
`Selection of studies
`
`Study selection was performed in duplicate by two independent
`authors (AR, CD, VW or EJ). A first level screen to obtain a list
`of potentially relevant articles was performed by reviewing the
`titles and abstracts of the search results (Rostom 2000). RCTs of
`PA, H2RAs or PPIs were considered eligible for inclusion if: these
`drugs were used for the prevention of NSAID induced upper GI
`toxicity in adults; the duration of NSAID exposure was 4 weeks
`or greater (equivalent to > 3 weeks); and endoscopic ulcers were
`defined as at least 3mm in diameter and/or could be distinguished
`from erosions based on the authors’ description. Studies in healthy
`volunteers were excluded. Double doses of H2RAs were defined
`
`as a dose equivalent to or greater than 300mg of ranitidine twice
`daily.
`
`Data extraction and management
`
`Data extraction was performed independently by two authors
`(AR,CD, VW, or EJ), with a standardized data extraction sheet.
`Differences were resolved by consensus.
`
`Assessment of risk of bias in included studies
`
`Studies included in the review were independently assessed for
`quality by two authors (AR, CD, VW) using a validated qual-
`ity instrument (Jadad 1996). This instrument rates studies on 3
`domains: randomisation; blinding; and completeness of follow-
`up accounting. Additionally the studies were assessed for the ad-
`equacy of allocation concealment. Differences in ratings were re-
`solved by consensus.
`
`Measures of treatment effect
`
`The dichotomous outcomes were analysed with Metaview 5, using
`the Mantel-Haenszel relative risk (Greenland 1985) using a fixed
`effects model. The risk difference is also presented. A global chi
`Square test (1 degree of freedom) was used to assess the difference
`between the estimated adjusted relative risk (RR) for high and low
`dose misoprostol.
`
`Unit of analysis issues
`
`The primary analysis unit was the proportion of participants with
`an outcome over the total number of participants in the group.
`Mantel-Haenzel relative risks were calculated for treatment groups
`compared to control. No serious analysis issues arose.
`
`Dealing with missing data
`
`Reporting of a primary outcome was an inclusion criteria of this
`review. However, some studies did not report on all primary end-
`points or all secondary outcomes. In those cases, the studies were
`used in the analyses of the endpoints they reported. For each end-
`point, the studies contributing to the analysis and the total num-
`ber of participants are indicated in the text. In some cases, missing
`data could be estimated from survival graphs, or could be calcu-
`lated by straight forward estimates (e.g. total ulcers from individ-
`ual reporting of gastric and duodenal ulcer data).
`
`Assessment of heterogeneity
`
`Heterogeneity was tested using a Chi-square test with (N-1) de-
`grees of freedom, where N equals the number of trials contribut-
`ing data. A P value of less than 0.10 was considered evidence of
`statistical heterogeneity. Additionally, heterogeneity was deemed
`
`Prevention of NSAID-induced gastroduodenal ulcers (Review)
`Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
`
`4
`
`
`
`to be present if the I square statistic was greater than 50%. Efforts
`were made a priori to reduce clinical and statistical heterogeneity
`by subdividing analyses by intervention class; dose of intervention
`used when available; and study durations. Analyses demonstrating
`heterogeneity were presented using a random effects model only
`if clinically and statistically appropriate.
`
`Assessment of reporting biases
`
`The presence of publication bias was explored through the use of
`an inverted funnel plot (studies’ effect size vs sample size). A review
`can be biased if small negative studies are not considered because
`these studies are often not published. Heterogeneity was tested
`using a chi square test at an alpha of 0.10, and represented graphi-
`cally with a L’Abbe plot (L’Abbe 1987). Estimates of heterogeneous
`data were obtained using a random effects model (DerSimonian
`1986) only if clinically and statistically appropriate.
`
`Data synthesis
`
`Data were analysed using Review Manager (RevMan) version 5.0.
`Endoscopic, clinical and symptom-based outcomes were analysed
`separately. The primary analyses were expressed as relative risks
`using a fixed effects model. A random-effects model was used to
`combine ’heterogeneous trials’ only if it was clinically and statis-
`tically appropriate. The absolute risk reduction (ARR) and the
`number needed to treat (NNT) were calculated for appropriate
`clinical endpoints.
`
`Subgroup analysis and investigation of heterogeneity
`
`In addition to dividing the data by the intervention and the specific
`outcome under study, subgroup analyses were performed by the
`dosages of the intervention used, and the length of follow-up.
`
`Sensitivity analysis
`
`Sensitivity analyses were performed by: the study quality, with the
`median quality score used as the cut off to define lower and higher
`quality studies; and by primary vs secondary prophylaxis trials.
`In the first version of the review, a sensitivity analysis varying the
`obtained point estimates from efficacy to intention to treat was
`performed and did not have a significant effect on the estimates.
`This analysis was not repeated in the updates to the review.
`
`R E S U L T S
`
`Description of studies
`
`See: Characteristics of included studies; Characteristics of excluded
`studies.
`Thirty-nine RCTs met the inclusion criteria: 23 misoprostol trials;
`12 H2RA; and 9 PPI trials. Some studies considered more than
`one active intervention and seven studies were used for head to
`head comparisons. See Characteristics of included studies for a
`summary of included studies.
`A description of the search result updates over time from the orig-
`inal review is detailed below.
`
`Results of the search
`
`Original Cochrane Search Strategy results
`
`Of a total of 970 references identified in the first review, 33 RCTs
`met the inclusion criteria: 18 misoprostol trials; nine standard dose
`H2RA; three double dose H2RA trials; and four PPI trials. (See
`Rostom 2000).
`At the July 2001 update, four potentially relevant articles were
`found by repeating the electronic searches and reviewing con-
`ference proceedings. Of these, two fulfilled the inclusion criteria
`(misoprostol Bocanegra 1998; 1 PPI-misoprostol abstract Jensen
`2000).
`
`CCOHTA Search Strategy Results (2002)
`
`The updated search identified for tNSAIDs and COX-2s identi-
`fied 898 references. Of 241 potentially relevant papers, three new
`studies met the inclusion criteria for this review: two PPI papers
`(Graham 2002; Bianchi Porro 2000); and two misoprostol papers
`(Graham 2002; Chan 2001a). The Graham study considered two
`interventions.
`
`Updated Cochrane Searches
`
`An updated search in August 2003 and August 2004 did not reveal
`any further new studies.
`The Cochrane search was updated from 2004 to May 2009. The
`MEDLINE search identified 235 articles with ten potentially rel-
`evant studies (Desai 2008; Goldstein 2007; Chan 2007; Lanas
`2007; Regula 2006; Scheiman 2006; Goldstein 2005; Lai 2005;
`Miyake 2005; Hawkey 2005; Chan 2004b). While some of these
`papers provided supporting evidence, none met the inclusion cri-
`teria. The EMBASE search identified 549 articles and only iden-
`tified eight of the eleven potentially relevant studies identified by
`MEDLINE. Similary, out of 41 articles identified in the EBMR
`search, only four of the potentially relevant studies were identified.
`The non-MEDLINE searches did not identify studies that were
`not identified by MEDLINE. While the Chan 2004 paper did not
`meet inclusion criteria, the results are presented in the PPI section
`as indirect evidence.
`
`Prevention of NSAID-induced gastroduodenal ulcers (Review)
`Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
`
`5
`
`
`
`Canadian Consensus Conference Search Strategy identified two
`additional studies not identified in the previous searches (Stupnicki
`2003; Lai 2003).
`
`quality with a Jadad score of three or greater (20 studies - Q3; ten
`studies - Q4; two studies Q5). Nine studies received a Jadad Score
`of two or less.
`
`Included studies
`
`Allocation
`
`See Characteristics of included studies. Thirty-nine studies were
`included in the review: 23 misoprostol studies (includes six head
`to head); twelve H2RA trials (nine standard dose; three double
`dose; one head to head); and nine PPI trials (six direct; five head to
`head). Some studies considered more than one intervention. All
`the included studies were RCTs in participants with arthritis who
`were taking traditional NSAIDs in an outpatient setting.
`
`Excluded studies
`
`See Characteristics of excluded studies. The most common rea-
`sons for exclusions were: short term studies <4 weeks; studies not
`reporting on desired outcomes; studies that were not RCTs.
`
`Risk of bias in included studies
`
`Methodological quality was assessed by two independent authors
`(AR, VW or EJ) using Jadad’s scale (Jadad 1996) with consid-
`eration of allocation concealment. Differences were resolved by
`consensus. A third reviewer was consulted to resolve any disagree-
`ments (CD).
`The table of included studies details the Jadad score for the in-
`cluded studies. The majority of the studies were of reasonable
`
`The quality of allocation concealment was unclear in the majority
`of the included studies.
`
`Blinding
`
`All the included studies were blinded.
`
`Incomplete outcome data
`
`All the included studies reported on one of the primary endpoints
`(endoscopic ulcer; or clinical ulcer complication).
`
`Selective reporting
`
`The inverted funnel plot reveals a lack of small studies of miso-
`prostol with small effect sizes, suggesting the potential of publica-
`tion bias (Rostom 2000) (please see Figure 1). However, using the
`method described by Rosenthal, there would have to be 60 one-
`month, and 300 three-month small studies averaging null find-
`ings to negate the statistically significant reduction in endoscopic
`gastric ulcers observed with misoprostol (Rosenthal 1979). The
`symmetrical distribution of studies of H2RAs suggests the absence
`of publication bias. Six PPI trials were identified showing similar
`distribution as seen with misoprostol (Figure 2).
`
`Prevention of NSAID-induced gastroduodenal ulcers (Review)
`Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
`
`6
`
`
`
`Figure 1. Funnel plot of comparison: 1 misoprostol vs placebo - Primary Efficacy, outcome: 1.6 Total
`endoscopic ulcers - 12 weeks or longer studies.
`
`Prevention of NSAID-induced gastroduodenal ulcers (Review)
`Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
`
`7
`
`
`
`Figure 2. Funnel plot of comparison: 11 PPI vs placebo - 8 weeks or longer studies, outcome: 11.3 Total
`endoscopic ulcers.
`
`Other potential sources of bias
`
`The observed heterogeneity, based on the estimates for gastric ul-
`cers, is represented graphically (L’Abbe 1987) (see Rostom 2000).
`Significant heterogeneity was seen only for the three month duo-
`denal ulcer pooled estimate with misoprostol. This disappeared
`when the studies were grouped by misoprostol dose.
`
`Effects of interventions
`
`Misoprostol
`
`We found twenty-three studies that assessed the long term ef-
`fect of misoprostol on the prevention of NSAID ulcers (Graham
`2002; Chan 2001a; Hawkey 1998; Bocanegra 1998; Raskin
`1996; Agrawal 1995; Raskin 1995; Valentini 1995; Delmas 1994;
`Elliot 1994; Graham 1993; Henriksson 1993; Melo Gomes
`1993; Roth 1993; Bolten 1992; Verdickt 1992; Agrawal 1991;
`Chandresekaran 1991; Saggioro 1991; Graham 1988; Jensen
`2000; Silverstein 1995; Stupnicki 2003). Some of these were head
`to head studies.
`
`Endoscopic ulcers
`
`Eleven studies with 3,641participants compared the incidence
`of endoscopic ulcers, after at least three months, of misoprostol
`to that of placebo (Graham 2002; Chan 2001a; Hawkey 1998;
`Agrawal 1995; Raskin 1995; Elliot 1994; Graham 1993; Roth
`1993; Verdickt 1992; Agrawal 1991; Graham 1988). The cumu-
`lative incidence of endoscopic gastric and duodenal ulcers with
`placebo were 15% and 6% respectively. Misoprostol significantly
`reduced the relative risk of gastric ulcer and duodenal ulcers by
`74% (RR 0.26; 95% CI 0.17 to 0.39, random effects; Analysis
`1.4), and 58% (RR 0.42; 95% CI 0.22 to 0.81, random effects;
`Analysis 1.5). These relative risks correspond to a 12.0%, and 3%
`absolute risk reductions for gastric and duodenal ulcers respec-
`tively. The observed heterogeneity in these estimates was due to
`inclusion of all misoprostol doses in the analyses. Analysis of the
`misoprostol studies stratified by dose eliminated this heterogene-
`ity.
`
`Analysis by dose
`
`All the studied doses of misoprostol significantly reduced the risk
`of endoscopic ulcers, and a dose response relationship was demon-
`
`Prevention of NSAID-induced gastroduodenal ulcers (Review)
`Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
`
`8
`
`
`
`strated for endoscopic gastric ulcers. Six studies with 2,461 par-
`ticipants used misoprostol 400 µg (Chan 2001a; Hawkey 1998;
`Agrawal 1995; Raskin 1995; Verdickt 1992; Graham 1988), one
`study with 928 participants used 600 µg daily (Raskin 1995), and
`seven with 2,423 participants used 800 µg daily (Graham 2002;
`Raskin 1995; Elliot 1994; Graham 1993; Roth 1993; Agrawal
`1991; Graham 1988). Misoprostol 800 µg daily was associated
`with the lowest risk (RR 0.18; 95% CI 0.12 to 0.27; Analysis 4.4)
`of endoscopic gastric ulcers when compared to placebo, whereas
`misoprostol 400 µg daily was associated with a relative risk of
`(RR 0.43; 95% CI 0.28 to 0.67, random effects model for hetero-
`geneity; Analysis 4.4). The observed heterogeneity in the 400 µg
`dose group was the result of the addition of the Chan study (Chan
`2001a). This study compared the relatively more toxic naproxen
`with low dose misoprostol to nabumatone alone. In this study the
`risk of ulcers was inexplicably greater in the misoprostol group,
`but we feel this result is based on the differences between the safety
`of the comparator NSAIDS rather than the prophylactic agent. As
`a sensitivity analysis, removal of the Chan 2001a study eliminates
`the observed heterogeneity without significantly altering the re-
`sults, giving low dose misoprostol prophylaxis a RR of 0.39 (95%
`CI 0.3 to 0.51; Analysis 4.5). This difference between high and
`low dose misoprostol reached statistical significance (P=0.0055).
`The intermediate misoprostol dose (600 µg daily) was not sta-
`tistically different from either the low or high dose. The pooled
`relative risk reduction of 78% (4.7% absolute risk difference, (RR
`0.21; 95% CI 0.09 to 0.49; Analysis 4.6) for duodenal ulcers with
`misoprostol 800 µg daily was not statistically different from those
`of the lower daily misoprostol dosages.
`
`Studies including data with less than 3 months
`NSAID exposure
`
`Eight studies, with 2206 participants, assessed the rates of endo-
`scopic ulcers with misoprostol compared to placebo at 1 to 1.5
`months (Bocanegra 1998; Delmas 1994; Elliot 1994; Henriksson
`1993; Melo Gomes 1993; Bolten 1992; Chandresekaran 1991;
`Saggioro 1991). The pooling of these studies revealed an 81% rel-
`ative risk reduction of gastric ulcers with misoprostol (RR 0.17;
`95% CI 0.09 to 0.31; Analysis 1.1) and an 70% relative risk reduc-
`tion of duodenal ulcers (RR 0.30; 95% CI 0.14 to 0.62; Analysis
`1.2).
`One study compared misoprostol to a newer cytoprotective agent,
`Dosmafate, for NSAID prophylaxis and found no statistically sig-
`nificant difference in ulcer rates between the two agents (Cohen
`2000).
`
`Clinical Ulcers
`
`Only one RCT, the MUCOSA trial, evaluated the efficacy of miso-
`prostol prophylaxis against clinically important NSAID induced
`ulcer complications. In this study, of 8,843 participants studied
`
`over six months, the overall GI event incidence was about 1.5%
`per year (Silverstein 1995). Misoprostol 800 µg/day was associ-
`ated with a statistically significant 40% risk reduction (OR 0.598;
`95% CI 0.364 to 0.982) in combined GI events (P=0.049), repre-
`senting a risk difference of 0.38% (from 0.95% to 0.57%). Over-
`all, approximately 260 participants would have to be treated with
`misoprostol to prevent one clinically important GI event. This
`NNT would drop as higher risk participants are considered. Miso-
`prostol appeared to be ineffective at preventing endoscopically
`proven GI haemorrhage alone. However a type II error is likely
`since the study was not powered to detect a difference in this end-
`point (Silverstein 1995).
`
`Adverse Effects
`
`Misoprostol was associated with a small but statistically significant
`1.6 fold excess risk of drop out due to