Nonsteroidal antiinflammatory drugs, ulcers and risk:A collaborative meta-analysis

Nonsteroidal antiinflammatory drugs, ulcers and risk:A collaborative meta-analysis

Nonsteroidal Antiinflammatory Drugs, Ulcers and Risk: A Collaborative Meta-Analysis Luis Alberto Garcia Rodriguez Nonsteroidal antiinflammatory drugs ...

439KB Sizes 3 Downloads 113 Views

Nonsteroidal Antiinflammatory Drugs, Ulcers and Risk: A Collaborative Meta-Analysis Luis Alberto Garcia Rodriguez Nonsteroidal antiinflammatory drugs (NSAIDs) are associated with a high prevalence of gastrointestinal toxicity. Comparisons of the risks associated with individual NSAIDs are needed, but most clinical studies in this area are not ideally suited for use in meta-analyses. These difficulties can be overcome by using strict criteria for the inclusion of studies, and by reanalyzing previous data, updated by authors where possible. In doing so, this meta-analysis compared the relative risk of serious toxicity associated with 14 NSAIDs from 12 studies with the risks of ibuprofen. These results were supported by a novel "summary ranking" analysis, which was weighted to limit the influence of smaller studies. Wide variations in relative risk among NSAIDs were observed with piroxicam and azapropazone being the most toxic. Ibuprofen was associated with the least risk, probably because of its widespread use as a low-dose analgesic. Five studies provided comparative data on NSAIDs at "high" and " l o w " doses (as defined in the original reports), showing that the risk of toxicity was dose related. Furthermore, at full antiinflammatory doses, the risk associated with ibuprofen was similar to that of naproxen and diclofenac. These analyses show that NSAIDs vary more in toxicity than in efficacy. First-line therapy should be started with the lowest effective dose of a less toxic NSAID, moving to higher doses or a more toxic NSAID only if the clinical situation demands it. Newer NSAIDs, such as selective cyclooxygenase-2 inhibitors, may provide safer antiinflammatory therapy. Semin Arthritis Rheum 26:16-20, Copyright © 1997 by W,B. Saunders Company

INDEX WORDS: NSAIDs; gastrointestinal toxicity; meta-analysis. ' O N S T E R O I D A L ANTIINFLAMMATORY drugs (NSAIDs) are now established beyond question as highly effective in the treatment of rheumatic disease, but it is also clear that they carry a high risk of adverse events, particularly in the gastrointestinal (GI) tract. Studies have indicated that the prevalence of GI symptoms among patients taking these agents is in the region of 30% to 40% (1,2) with symptoms ranging in severity from occasional heartburn or indigestion to daily dyspepsia that cannot be relieved by antacids or H2-receptor blockers. In up to 10% of patients, the symptoms have been so severe that it was necessary to


From the Centro Espafiol de lnvestigaci6n Farmacoepidemio16gica, Universidad CompIutense de Madrid, Madrid, Spain. Address reprint requests to Dr LA Garcfa Roddguez, MD, Centro Espafiol de Investigaci6n FarmacoepidemiolOgica, Almirante 28 (2°), Madrid 28004, Spain. Copyright © 1997by W.B. Saunders Company 0049-0172/97/2606-100455.00/0


withdraw from treatment (2). Even more serious, from the point of view of the physician, is the 1% of patients who suffer life-threatening events such as severe GI bleeding or perforation as a result of NSAID treatment (2). Attempts to limit the morbidity and mortality of these agents include educational programs for physicians aimed at reducing prescribing, simultaneous prescription of a protective agent for the gastric mucosa (ie, misoprostol), and the use of paracetamol as an analgesic when possible (3,4). Another approach is to adopt a policy of prescribing a drug with comparatively low GI toxicity initially and to remove drugs with greater toxic potential only if and when clinically indicated (5). However, published studies vary in their coverage of individual drugs and as a result, estimating the relative risk of different agents has not been easy. This can create difficulties in meta-analytical studies that pool data from different studies (6). Further confusion can arise from methodological differ-

Seminars in Arthritis and Rheumatism, Vo126, No 6 (SUPPL 1) (June), 1997: pp 16-20



ences such as the dosages compared and differences in the patient population that might make them more or less susceptible to the development of GI complications. This article describes a study that used metaanalytical techniques to reexamine data from previous studies, applying quality control to ensure comparability of data and updating that information where appropriate (7). RISK FACTORS FOR PEPTIC ULCER COMPLICATIONS

An earlier population-based, case-control study, performed retrospectively in the United Kingdom identified a number of independent risk factors for upper GI bleeding associated with NSAID use (8). Factors associated with a higher risk of GI bleeding are shown in Figure 1. The overall risk of upper GI bleeding was 4.7 in this study, although it was significantly higher for patients taking multiple NSAIDs than those taking only one. Men carried a two-fold increased risk of upper GI bleeding compared with women, and smoking increased the risk by 40%. Risk of developing NSAID complications increased with age; patients 60 to 69 years old had a relative risk of 3.1 compared with patients under 50 years of age. The relative risk for patients between 70 and 80 years of age was 5.6. A past history of GI bleeding or perforation was the single most important predictor of upper GI bleeding in this study (relative risk 13.5), although Increasing age: >60 years


Gender: men versus women

~ 2 - 3




Prior peptic ulcer disease


Alcohol consumption





1.5-2 1.3 - 1.5

Current use of: NSAIDs Anticoagulants Corticosteroids i

0 5 10 Range of estimates of relative risk

Fig 1. Risk factors for peptic ulcer complications. Shown are relative risks of upper gastrointestinal (GI) bleeding and perforation associated with several risk factors (95% confidence intervals). (Data from Garcia Rodriguez et al [8])

patients with no history carried a greater relative risk of developing these symptoms while on NSAID treatment than those with a positive history. Compared with people not currently taking NSAIDs and with no history of peptic ulceration, the relative risks were 5.4 for those exposed to NSAIDs with no upper GI bleeding history, 8.7 for those with a positive history but not currently taking NSAIDs, and 17.2 for those with both risk factors. Because people with a past history of peptic ulceration have almost nine times the risk of upper GI bleeding than people without, even in the absence of exposure to NSAIDs, these drugs should be used with great caution in such patients. When the severity of the previous episode of peptic ulcer is examined, another study has shown that the risk of developing GI complications with NSAID use increases from an odds ratio of 3.7 for those who have less serious conditions such as dyspepsia or use of antiulcer drugs to 5.3 for those with peptic ulcer-related hospitalization and up to 19.7 for those with a hospitalization caused by ulcer complications (9). In another study (9) that also examined the relative risks of different NSAIDs, the authors found that, in accordance with other studies, ibuprofen in low doses carried the lowest risk of any of the agents examined. In this study, piroxicam and ketorolac carried the greatest relative risk (10 and 25, respectively) of causing upper GI bleeding. A COLLABORATIVE META-ANALYSIS

Study Methods A literature search covering the period of 1985 to 1994 was undertaken on MedLine and the bibliographies of previously published meta-analyses reviewed. Studies identified in this manner then underwent a quality control procedure, and those that did not meet prespecified criteria were excluded from further consideration. The criteria for inclusion in the meta-analysis are shown in Table 1. The authors of the suitable studies were contacted and asked to submit updated data and asked for details of any unpublished or unlisted work of which they might be aware. A total of 12 studies were included in the final analysis (6,8,10-19). This study took a different approach than previous studies of this type by calculating the relative risk of all other NSAIDs relative to ibuprofen users rather than relative to nonusers. This required some reanalysis of data obtained at a specially convened



Table 1, Criteria for Inclusion of Studies into the Meta-Analysis

Mean rank

Results A total of 14 NSAIDs were investigated in the 12 studies included in the meta-analysis, 11 of which compared ibuprofen with other NSAIDs. The overall relative risk of complications with NSAIDs in these studies are mainly in the range 3.0 to 5.0, which is consistent with the results obtained from other meta-analyses (20,21). Calculations of the relative risks from individual studies showed a wide variation, but a trend of higher values with some NSAIDs over others suggests a genuine difference between the drugs. This was particularly marked for azapropazone and was confirmed when the individual estimates of relative risk were pooled together. Drugs that appeared in two or more studies were included in the summary ranking analysis. Ibuprofen achieved the lowest ranking, followed by diclofenac, whereas ketoprofen, piroxicam, and azapropazone had the greatest risk of GI toxicity.


Controlled epidemiological studies Relate NSAID use to serious peptic ulcer Quality control: Measurement/validation of outcomes Selection/comparability of control Control of confounding factors

authors' workshop. Point estimates from individual studies relative to ibuprofen were computed and pooled across all studies for each drug. In addition, separate estimates of risk associated with low/ medium and high drug dosage were pooled for three drugs. Only five studies provided dosespecific data for individual NSAIDs. Calculation of the comparative toxicity of each agent was undertaken using a novel "summary ranking" method. This entailed assigning a score to the rank order of toxicity in individual studies, weighted so that smaller studies exerted less influence on the overall result than larger studies. Scores were then added together to provide a rank order of toxicity which best fitted the results obtained from individual studies. This analysis was complementary to the main analysis, and offered the advantage of comparing each NSAID with all of the others.



o -~












Fig 2, Comparative GI toxicity (upper GI bleeding and perforation) of nonsteroidal antiinflammatory drugs (NSAIDs) obtained using the summary ranking method, (Data from Henry et al [7])

Aspirin, sulindac, naproxen, and indomethacin carried an intermediate risk (Fig 2). The very high relative risks of piroxicam and azapropazone are alarming, and have prompted calls for their withdrawal (22). This could be undesirable for those patients who do not respond to other agents in which the benefits of treatment with piroxicam and azapropazone will exceed the risks. However, there must be doubts about the suitability of these agents, particularly azapropazone, for routine use. Five of the 12 studies provided information about relative toxicity at different doses. Comparisons were made between ibuprofen, indomethacin, and naproxen (6,8,10-12). Definitions of "high" and "low" dose for these purposes reflected those used in the original studies and were subject to some variation between studies. Figure 3 shows that there is a clear dose-response effect, and that ibuprofen at high dosage carries a similar risk to Relative 14 risk of


12 10

proxen omethac n]

8 6 4 2 0 Low/medium dose

High dose

(Low/medium and high doses as were defined in original studies)

Fig 3, Relationship between dose and relative risk of GI toxicity (upper GI bleeding and perforation) with three commonly used NSAIDs (95% confidence intervals), (Data from Henry et al [7])


19 36.4

Relative 20 risk of 18 1 toxicity 16 14


ibuprofen are at much greater risk of upper GI bleeding.




10 8 6 4 2 0


Ibuprofen >1,500 mg

Naproxen >750 mg

Diclofenac Indomethacin >100 mg >75 mg

Fig 4. Relative risk (95% confidence intervals) of GI toxicity (upper GI bleeding and perforation) with four commonly used NSAIDs at high antiinflammatory doses (95% confidence intervals), (Data from Garcia Rodriguez et al [8])

other NSAIDs at high dosage. Likewise, at lower doses, all three agents carry a much lower risk of GI toxicity. These findings confirm the results of another study (8), which showed that the relative risk of GI toxicity between ibuprofen at full antiinflammatory dosage (>1,500 mg/day) is similar to that of corresponding doses of diclofenac (> 100 mg/day), naproxen (>750 mg/day), and indomethacin (>75 rag/day) (Fig 4). In this study, as in the meta-analysis, there was no significant difference between ibuprofen and the other NSAIDs at either dose level. Ibuprofen is often used in low (analgesic) dosage in clinical practice. However, these results suggest that when it is used at full antiinflammatory dosage (as required for patients with arthritis), the apparent safety advantages are lost and the risk of GI toxicity is broadly similar to those of indomethacin, didofenac, and naproxen at equivalent dosage. In clinical practice, the incidence of toxicity observed is largely determined by the dosage used, and the apparent safety advantages of ibuprofen may be explained by the fact that 80% of users took it in low dosage (8). The entire study population exposed to ibuprofen, therefore, carried a low risk, but individuals who take high doses of

Many meta-analytical studies have encountered great difficulty in extracting comparable data sets from different studies. These may be compounded by the limited dosage options available for some agents, with some patients receiving higher or lower doses than they actually need. Definitions of "high" and "low" dose are often arbitrary and cause further difficulties in investigating the relationship between dose and toxicity. Finally, it is important to be aware that epidemiological studies do not always compare like with like, and careful attention must be given to the methods and exposure definitions used. The meta-analysis described here overcame these problems as far as possible by setting strict criteria for the eligibility of studies into the analysis and augmenting the original results with additional data from the authors. This approach is validated by the similarity of results obtained by using either pooled relative risks or the summary ranking method. The summary ranking method has the advantage of comparing each drug with each other one, and offers a powerful technique for identifying real differences between them. The low risk of complications seen with ibuprofen in the main analysis relates to the relatively low doses taken by the majority of patients. When used at higher arltiinflammatory dosage, it carries a similar risk to other agents at equivalent dosage. NSAIDs show much greater variability in their toxicity than in their efficacy. Prescribers and patients need clear information about the advantages and disadvantages of individual agents to enable them to choose the optimum regimen. For first-line treatment, the lowest effective dose of an NSAID with a good safety record should be used, moving to higher doses or a more toxic agent only if the clinical situation demands it. Newer NSAIDs, such as selective cyclooxygenase-2 inhibitors, may provide safer antiinflammatory therapy.

REFERENCES 1. Larkai EN, Lacey Smith J, Lidsky MD, Sessoms SL, Graham DY. Dyspepsia in NSAID users: the size of the problem. J Clin Gastroenterol 1989; 11:158-62. 2. G>ercksky K-E, Huseby G, Rugstad H-E. Epidemiology of

NSAID-related gastrointestinal side effects. Scand J Gastroenterol 1989; 24(Suppl 163):3-8. 3. Silverstein FE, Graham DY, Senior JR, eta!. Misoprosto[ reduces serious gastrointestinal complications in patients with


rheumatoid arthritis receiving non-steroidal anti-inflammatory drugs. Ann Intern Med 1995; 123:241-9. 4, Bradley JD, Brandt KD, Katz BP, Kalasinski LA, Ryan SI. Comparison of an anti-inflammatory dose of ibuprofen, an analgesic dose of ibuprofen, and acetaminophen in the treatment of patients with osteoa~hritis of the knee. New Engl J Med 1991; 325:87-91. 5. Bateman DN. NSAIDs: time to re-evaluate gut toxicity. Lancet 1994; 343:1051-2. 6. Langman MJS, Weil J, Wainwright R et al. Risks of bleeding peptic ulcer associated with individual non-steroidal anti-inflammatory drugs. Lancet 1994; 343:1075- 8. 7. Henry D, Lim LL-Y, Garc/a Rodrfguez LA, et al. Variability in risk of gastrointestinal complications with individual non-steroidal anti-inflammatory drugs: results of a collaborative meta-analysis. Br Med 1996; 312:1563-6. 8. Garcfa Rodrfguez LA, Jick H. Risk of upper gastrointestinal bleeding and perforation associated with individual nonsteroidal anti-inflammatory dings. Lancet 1994; 343:769-72. 9. Garcfa Rodrfguez LA, Cattaruzzi C, Troncon MG, Agostihis L. Risk of hospitalization for upper gastrointestinal bleeding associated with ketorolac, other NSAIDs, calcium antagonists and other antihypertensive dings. Arch Intern Med, (in press) 10. Griffin MR, Piper JM, Daugherty JR, Snowden M, Ray WA. Non-steroidal anti-inflammatory drug use and increased risk for peptic ulcer disease in elderly persons. Ann Intern Med 1991; 114:257-63. 11. Kaufman DW, Kelly JR Sbeehan JE, et al. Non-steroidal anti-inflammatory drug use in relation to major upper gastrointestinal bleeding. Clin Pharmacol Ther 1993; 53:485-94. 12. Perez Gutthann S, Garcia Rodrfguez LA, Raiford DS. Individual non-steroidal anti-inflammatory drugs and the risk of hospitalisation for upper gastrointestinal bleeding and perforation in Saskatchewan: a nested case-control study. Pharmacoepidemiology and Drug Safety 1994; 3(Suppl 1):$63. 13. Somerville K, Faulkner G, Langman M. Non-steroidal


anti-inflammatory dings and bleeding peptic ulcer. Lancet 1986; 1:462-4. 14. Carson JL, Strom BL, Morse ML, et al. The relative gastrointestinal toxicity of the nonsteroidal anti-inflammatory dings. Arch Intein Med 1987; 147:1054-9. 15. Laporte JR, Came X, Vidal X, Moreno V, Juan J. Upper gastrointestinal bleeding in relation to previous use of analgesics and non-steroidal anti-inflammatory dings. Catalan countries study on upper gastrointestinal bleeding. Lancet 1991; 337: 85-9. 16. Nobiti A, Mosconi R Franzosi MG, Tognoni G. Nonsteroidal anti-inflammatory drugs and upper gastrointestinal bleeding, a post-marketing surveillance case-control study. Pharmacoepidemiology and Drug Safety 1992; 1:65-72. 17. Savage RL, Moller PW, Ballantyne CL, Wells JE. Variation in the risk of peptic ulcer complications with nonsteroidal antiinflammatory drug therapy. Arthritis Rheum 1993; 36:84-90. 18. Henry D, Dobson A, Turner C. Variability in the risk of major gastrointestinal complications from nonaspirin nonsteroidal anti-inflammatory drugs. Gastroenterology 1993; 105:107888. 19. Abenhaim L, Moride Y. The effect of baseline susceptibility on the relative gastrotoxicity of individual NSAIDs in the elderly: a study with the Quebec database. Postmarketing Surveillance 1993; 7:176. 20. Gabriel SE, Jaak!dmainen L, Bombadier C. Risk for serious gastrointestinal complications related to use of nonsteroidal anti-inflammatory drugs. A meta-analysis. Ann Intern Med 199l; 115:787-96. 21. Bollini P, Garcia Rodrfguez LA, Perez Gutthann S, Walker AM. The impact of research quality and study design on epidemiological estimates of the effect of nonsteroidal antiinflammatory drugs on upper gastrointestinal tract disease. Arch Intern Med 1992; 152:1289-92. 22. HRG revisits Feldene. Scrip 13 December: 26-27, 1994