|Risk of serious NSAID-related gastrointestinal events during long-term exposure: a systematic review
|Schaffer D, Florin T, Eagle C, Marschner I, Singh G, Grobler M, Fenn C, Schou M, Curnow K M
This review investigated whether the occurrence of serious gastrointestinal events decrease over time in studies of long-term use of non-steroidal anti-inflammatory drugs (NSAIDs). The authors found that the probability of having a serious NSAID-related ulcer bleed, perforation or hospitalisation did not meaningfully decline over time. The use of meta-regression with few studies implies cautious interpretation of these results.
To assess whether the occurrence of serious gastrointestinal (GI) events reduces over time in studies of long-term use of non-steroidal anti-inflammatory drugs (NSAIDs).
Current Contents, EMBASE, MEDLINE and Derwent Drug File were searched from 1999 to 2002. Further studies were sought through manual searches of retrieved systematic reviews. The Cochrane Controlled Trials Register (Issue 3, 2002) was also searched from its start date. Studies published in English, French or German were included.
Patients exposed to prescription non-selective NSAIDs were eligible for inclusion. Adult patients with osteoarthritis and rheumatoid arthritis, subgroups aged over 65 years and over 40 years, and low-income elderly were included. The included studies investigated a wide range of NSAIDs taken in usual clinical doses. NSAID use was for pain treatment (or assumed to be if not specified). Most of the patients had taken NSAIDs before the study started.
Studies in which a 'serious GI event' incidence or risk was reported by duration of NSAID exposure, and in which events occurred during current use of medication, were eligible for inclusion. The included studies reported the occurrence of upper GI bleeds and perforations and several studies reported hospitalised symptomatic ulcers.
Randomised controlled trials (RCTs) and observational studies of NSAID exposure for more than 6 months were eligible for inclusion. The included studies were RCTs and cohort studies.
The authors did not state how the papers were selected for the review, or how many reviewers performed the selection.
Assessment of study quality
Study validity was assessed according to standard epidemiological principles and Australian Government Guidelines. The authors did not state how the validity assessment was performed.
Data for serious GI events at specific times (time to event data) were extracted for patients taking non-selective NSAIDs. If multiple definitions of serious GI event were reported, the strictest was extracted. Where data were not available, authors were contacted to obtain unpublished data.
The authors did not state how the data were extracted for the review, or how many reviewers performed the data extraction.
Methods of synthesis
RCT and cohort studies were analysed separately using a generalised estimating equations linear regression model across studies and time points. The annualised probability of having an event was estimated from the conditional probability of having a serious ulcer complication in a specified time interval, given that no event was experienced before the start of that interval, and derived from published cumulative probability estimates for each study. This was calculated for each time point in each study.
The outcome used for the meta-analysis was the estimated probablity of havng a GI event for each year in each study. This was derived from published cumulative probability estimates of having a serious event in a given time interval. The studies were combined using a generalised estimating equation model (form of meta-regression) to account for different time points and the model adjusted from time in months, study and treatment group. Analyses were weighted using the inverse variance method. Separate analyses were performed for RCTs and cohort studies.
The change in risk of a GI event over time was assessed using the statistical significance of the slope from the the meta-regression model. Formal tests of heterogeneity were not performed since there were only a few studies, but the authors did conduct a pre-assessment of heterogeneity (not reported but available on request).
Results of the review
Seven studies (n=19,841) were included in the review: 4 RCTs (n=11,723) and 3 cohort studies (n=8,118).
RCTs showed a small significant reduction in the annualised probability of having an upper GI bleed, perforation or obstruction, or being hospitalised for a symptomatic ulcer over time, of 0.005% per month (five groups from 4 RCTs; 95% confidence interval, CI: -0.008, -0.001). Sensitivity analyses showed that the results were affected by one trial which showed a greater decline in the risk of a GI event of -0.16% per exposure month.
There was no evidence of a significant decline in serious GI events over time in cohort studies (-0.003%, 95% CI: -0.008, 0.003). Sensitivity analyses showed similar results.
The probability of having a serious NSAID-related ulcer bleed, perforation or hospitalisation does not meaningfully decline over time.
The research question was well-defined and the inclusion criteria were clear with regard to the intervention, outcome and study design. The authors searched several appropriate electronic databases but did not report any attempt to identify unpublished literature, which may have led to publication bias. Only studies in English, French and German were included, which increases the possibility of language bias. The authors stated they assessed study validity, but few details were reported and there was no attempt to link study quality to the results of the analysis. The authors did not state how the validity assessment was performed or how the data were extracted, so it is not known whether any steps were taken to minimise bias and error in the review process. There was also no information reported about the study participants, or actual time to event results for each study. The meta-regression performed may have been inappropriate given the small number of studies included in the analysis. The authors used analysis methods which accounted for event rates at multiple times within each study and performed separate analyses for the RCTs and cohort studies, which was appropriate. However, as they pointed out, meta-regression results are only exploratory and limited when there are only a small number of studies. This, and the limited amount of information presented about the studies, means that the conclusions should be treated with some caution.
The authors stated that the project was conducted to support the Pharmaceutical Benefits Advisory Committee submission for celecoxib by Pfizer.
Implications of the review for practice and research
Practice: The authors stated that the duration of treatment should be minimised and, if long-term treatment is required, precautionary measures should be considered to reduce the risk of serious GI events.
Research: The authors stated that further large trials are needed. In addition, future studies should minimise the effect of rare event bias and depletion of susceptibilities, maximise the number of participants remaining in the study long-term, and capture potentially important confounding factors.
There was no specific monetary funding.
Schaffer D, Florin T, Eagle C, Marschner I, Singh G, Grobler M, Fenn C, Schou M, Curnow K M. Risk of serious NSAID-related gastrointestinal events during long-term exposure: a systematic review. Medical Journal of Australia 2006; 185(9): 501-506
Subject indexing assigned by NLM
Anti-Inflammatory Agents, Non-Steroidal /administration & Drug Administration Schedule; Gastrointestinal Diseases /chemically induced /epidemiology; Humans; Incidence; Risk Assessment; Time Factors; dosage /adverse effects
Date bibliographic record published
Date abstract record published
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.