This review concluded that misoprostol, cyclooxygenase-2 (COX-2) and selective non-steroidal anti-inflammatory drugs (NSAIDs), and possibly proton-pump inhibitors, reduce the risk of symptomatic ulcers; misoprostol is less well tolerated than proton-pump inhibitors or COX-2 non-steroidal anti-inflammatory drugs; and proton-pump inhibitors reduce the incidence of gastrointestinal symptoms, but not serious gastrointestinal complications, compared with COX-2 NSAIDs. The authors' conclusions seem reliable.

To assess the effectivenes of five specified gastroprotective strategies for people taking non-steroidal anti-inflammatory drugs (NSAIDs).

The Cochrane Library (Issue 2, 2002), MEDLINE (1966 to 2002), EMBASE (1980 to 2002), Current Controlled Trials (2002) and SIGLE (2002) were searched. The search terms were given on the NHS Health Technology Assessment Programme website (accessed 25/06/2005). See Web Address at end of abstract. In addition, the reference lists of included studies and identified systematic reviews were checked and authors were contacted for additional studies. Non-English reports were eligible.

Study designs of evaluations included in the review

Randomised controlled trials (RCTs) of at least 21 days in duration were eligible for inclusion.

Specific interventions included in the review

Studies that compared one of the following gastroprotective strategies with non-selected NSAIDs alone were eligible for inclusion: non-selective NSAID plus histamine-2 receptor antagonists (H2RAs); non-selective NSAID plus proton-pump inhibitors (PPIs); non-selective NSAID plus misoprostol; cyclooxygenase-2 (COX-2) selective NSAID only; or COX-2 specific NSAID only. The authors stated that some of the included studies used NSAIDs and gastroprotective agents at higher than the recommended doses.

Participants included in the review

Studies that evaluated exclusively children or healthy volunteers were excluded. The baseline gastrointestinal (GI) status (based on a score of 1 to 6, and defined as the percentage of patients with history of ulcers or bleeds) varied across the included studies. Further details were reported.

Outcomes assessed in the review

The primary outcomes of interest were serious GI complications, symptomatic ulcers, serious cardiovascular or renal illness, health-related quality of life (not measures of arthritis pain or disability) and mortality. GI complications included haemorrhage, haemorrhagic erosions, recurrent upper GI bleeds, perforation, pyloric obstruction and melaena. The secondary outcomes of interest included total GI symptoms, endoscopic ulcers, anaemia, occult bleeding, the total number of drop-outs and drop-outs due to GI symptoms.

How were decisions on the relevance of primary studies made?

Two reviewers independently selected the studies and resolved any disagreements through discussion.

The studies were assessed for methods of randomisation, allocation concealment, baseline comparability, blinding of the participants, care providers and outcome assessors, and losses to follow-up. Studies were classified for risk of bias (low, medium or high) based on allocation concealment and baseline comparability. Two reviewers independently assessed validity and resolved any disagreements through discussion.

Two reviewers independently extracted the data and resolved any disagreements through discussion. For each comparison, data were extracted on the number of people with each outcome for the longest follow-up point and used to derive a relative risk (RR).

How were the studies combined?

The results from the individual studies were combined using a random-effects meta-analysis, where appropriate. A pooled RR with 95% confidence intervals (CIs) was calculated separately for each gastroprotective strategy for each outcome of interest. The number-needed-to-treat to prevent one symptomatic ulcer was calculated where data permitted. Publication bias was assessed using funnel plots and Egger's and Begg's tests.

How were differences between studies investigated?

Statistical heterogeneity was assessed using the Cochran test (with a significance level of p<0.1) and by visual assessment of forest plots. A random-effects meta-regression was used to examine the influence on symptomatic ulcers of length of follow-up, mean age of the participants, baseline GI status and a number of initial risk factors for GI harms. A sensitivity analysis was performed by omitting studies with a 'high' risk of bias, treatment arms with higher than recommended doses of NSAID or gastroprotective agent, and naproxen treatment arms (for deaths and serious cardiovascular and renal events).

One hundred and eighteen RCTs (n=76,322) were included.

Five studies were considered to be at low risk for bias. Methodological flaws of the studies included lack of systematic reporting of important outcomes, overlapping of some outcomes, and lack of reporting allocation concealment and blinding of the outcomes assessment. The authors stated that few authors provided, or were able to provide, additional data when contacted and that few studies did not receive funding from pharmaceutical companies.

Non-selective NSAID plus H2RA versus placebo plus non-selective NSAID: 15 RCTs (2,621 patients) were identified, none of which were judged to be at low risk of bias. There were insufficient data to adequately compare H2RAs with placebo on any of the primary outcomes. H2RAs significantly reduced endoscopic ulcers compared with placebo (RR 0.55, 95% CI: 0.4, 0.7).

Non-selective NSAID plus PPIs versus placebo plus non-selective NSAID: 6 RCTs (1,358 patients) were identified, one of which was judged to be at low risk of bias. There were insufficient data to compare PPIs with placebo in terms of serious GI complications, serious cardiovascular or renal conditions, quality of life, or death. The significant reduction in symptomatic ulcers with PPI versus placebo (RR 0.09, 95% CI: 0.0, 0.5) was lost on sensitivity analysis. The significant reduction in endoscopic ulcers with PPI versus placebo (RR 0.37, 95% CI: 0.3, 0.5) remained on sensitivity analysis.

Non-selective NSAID plus misoprostol versus placebo plus non-selective NSAID: 23 RCTs (16,945 patients) were identified, one of which was judged to be at low risk of bias. Misoprostol significantly reduced serious GI complications (RR 0.57, 95% CI: 0.4, 0.9), symptomatic ulcers (RR 0.36, 95% CI: 0.2, 0.7) and endoscopic ulcers (RR 0.33, 95% CI: 0.3, 0.4). No statistically significant heterogeneity was detected and the results were stable to sensitivity analysis.

COX-2 selective NSAID only versus non-selective NSAIDs: 51 RCTs (28,178 patients) were identified, none of which were judged to be at low risk of bias. COX-2 selective NSAIDs significantly reduced symptomatic ulcers compared with placebo (RR 0.41, 95% CI: 0.3, 0.7). There were few events reported for the other primary outcomes.

COX-2 specific NSAID only versus non-selective NSAIDs: 17 RCTs (25,564 patients) were identified, three of which were judged to be at low risk of bias. COX-2 specific NSAIDs appeared to significantly reduce serious GI complications (RR 0.55, 95% CI: 0.4, 0.8) and symptomatic ulcers (RR 0.49, 95% CI: 0.4, 0.6). No statistically significant heterogeneity was detected for either meta-analysis, but the results for serious GI complications were not robust to sensitivity analysis. No significant difference was found in the occurrence of serious cardiovascular or renal illness, or total deaths. COX-2 specific NSAIDs were associated with significantly fewer endoscopic ulcers.

The meta-regression found no significant association between the RR of symptomatic ulcers (for COX-2 selectives and specifics) or endoscopic ulcers (for H2RAs, PPIs and misoprostol) and length of follow-up, mean age, or baseline GI status.

The number-needed-to-treat was infinite for H2RAs and COX-2 specific NSAIDs and 14 (8 to 100) for PPIs.

Following the initial writing of this abstract based on a paper (see Other Publications of Related Interest no.1), the full report has become available in which 10 further comparisons were reported.

H2RA plus NSAID versus misoprostol plus NSAID (3 RCTs; at least 599 patients): significant results were not seen for any primary outcome, and the CIs were wide. The only secondary outcome to show a significant difference when H2RA and misoprostol were compared directly was endoscopic ulcers (RR 4.35, 95% CI: 1.51, 12.55; 3 RCTs).

Misoprostol plus NSAID versus COX-2 preferential NSAID (3 RCTs; 1,035 patients): there was a significant increase in serious GI events with misoprostol (RR 3.05, 95% CI: 1.03, 9.06; 2 RCTs). There was no significant difference in serious cardiovascular or renal events or deaths (1 RCT).

PPI plus NSAID versus misoprostol plus NSAID (2 RCTs; 973 patients): significant results were not seen for any primary outcome.

Only single studies evaluating each of the following comparisons were identified: H2RA plus NSAID versus PPI plus NSAID; PPI plus NSAID versus COX-2 coxib NSAID; misoprostol plus NSAID versus COX-2 coxib NSAID; and COX-2 coxib NSAID versus COX-2 preferential NSAID.

No studies were identified evaluating the following comparisons: H2RA plus NSAID versus COX-2 coxib NSAID or COX-2 preferential NSAID; and PPI plus NSAID versus COX-2 preferential NSAID.

The economic analysis suggested that COX-1 NSAIDs plus H2RAs may reduce the risk of endoscopic ulcer at a lower overall cost to the health care provider than NSAIDs alone, and COX-1 NSAIDs plus PPIs or misoprostol and COX-2 coxibs at an overall increased cost.

Misoprostol, COX-2 specific and selective NSAIDs, and possibly PPIs, significantly reduced the risk of symptomatic ulcers. Misoprostol and probably COX-2 specifics significantly reduced the risk of serious GI complications, but the quality of the data was poor. Misoprostol is not so well tolerated as either PPIs or a COX-2 NSAID. PPIs are probably more effective at reducing the incidence of GI symptoms, but not necessarily serious GI complications, compared with COX-2 NSAIDs.

The review question was clear in terms of the study design, participants, intervention and outcomes. Several relevant sources were searched and the search terms were stated. Attempts were made to minimise language and publication bias. Two reviewers independently selected the studies, assessed validity and extracted the data, thus reducing the potential for bias and errors. Validity was assessed using established criteria. Adequate details on the results of the included studies were given, although information on the participants and interventions was limited. The methods used to combine the studies appeared appropriate. Differences between the studies were assessed and associations between treatment effect and several study characteristics were explored. The synthesis of the evidence took account of study quality. The authors' conclusions appear reliable.

Practice: The authors stated that patients in high-risk groups should either be co-prescribed a gastroprotective agent, or a COX-2 NSAID, and that there is no justification for co-prescribing a H2RA over the other strategies.

Research: The authors stated that trials should report rare but important events (such as deaths, cardiovascular events or serious GI bleeds) in papers so that data are available for independent meta-analysis or, at the very least, are available to named contact authors. They further stated that there is a need for a large multicentre, independently funded trial lasting at least 1 year into gastroprotective agents such as H2RAs, PPIs, misoprostol, COX-2 selectives, COX-2 specifics and placebo.

NHS R&D Health Technology Assessment (HTA) Programme, project number 01/40/02.

1. Hooper L, Brown TJ, Elliott RA, Payne K, Roberts C, Rostom A, et al. The effectiveness of five strategies for the prevention of gastrointestinal toxicity induced by non-steroidal anti- inflammatory drugs: systematic review. BMJ 2004;329:948-52. 2. Deeks JJ, Smith LA, Bradley MD. Efficacy, tolerability, and upper gastrointestinal safety of celecoxib for treatment of osteoarthritis and rheumatoid arthritis: systematic review of randomised controlled trials. BMJ 2002;325:619-23. 3. National Institute for Clinical Excellence. Guidance on the use of cyclo-oxygenase (COX) II selective inhibitors, celecoxib, refecoxib, meloxicam and etodolac for osteoarthritis and rheumatoid arthritis. London: The Institute; 2001. Report No.: Technology Appraisal Guidance 27. This additional published commentary may also be of interest. Wong VW. Review: misoprostol or COX-2-specific or selective NSAIDs reduce gastrointestinal complications and symptomatic ulcers. ACP J Club 2005;142:75.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.