Intended for healthcare professionals

Clinical Review Extracts from “Clinical Evidence”

Non-steroidal anti-inflammatory drugs

BMJ 2000; 320 doi: (Published 15 April 2000) Cite this as: BMJ 2000;320:1058
  1. Peter C Gøtzsche, director (p.c.gotzsche{at}
  1. Nordic Cochrane Centre, Rigshospitalet, 9 Blegdamsvej, DK 2100 Copenhagen Ø, Denmark


    Definition Non-steroidal anti-inflammatory drugs (NSAIDs) have anti-inflammatory, analgesic, and antipyretic effects and inhibit thrombocyte aggregation. The drugs have no documented effect on the disease process itself.

    Incidence/prevalence NSAIDs are widely used. Almost 10% of people in the Netherlands had used a non-aspirin NSAID in 1987; the overall use was 11 defined daily doses (see box) per 1000 persons a day.1 In Australia in 1994, overall use was 35 defined daily doses per 1000 persons a day, with 36% of the people receiving NSAIDs for osteoarthritis, 42% for sprain and strain or low back pain, and 4% for rheumatoid arthritis; 35% were aged over 60 years.2

    Aims To reduce symptoms in rheumatic disorders; to avoid severe gastrointestinal adverse effects.

    Outcomes Primary outcomes: pain intensity, person's preference for one drug over another, global efficacy, and clinically significant gastrointestinal complications. Secondary outcomes: number of tender joints, perforation, gastrointestinal haemorrhage, dyspepsia, and ulcer detected by routine endoscopy.

    Defined daily dose: The assumed average daily dose for the main indication of a specified drug. The defined daily dose per 1000 population per day is an estimate of the proportion of that population receiving treatment with that drug.


    We searched Medline and the Cochrane Library in July 1999 for systematic reviews and randomised controlled trials (RCTs) that included at least 100 people. More than 100 meta-analyses and thousands of RCTs have compared various NSAIDs. Many trials are unpublished or published in sources that are not indexed in publicly available databases. The quality of the trials is variable and bias is common, both in the design and analysis of the trials, to such an extent that a systematic review identified false significant findings favouring new drugs over control drugs in 6% of trials.3



    NSAIDs in rheumatoid arthritis

    Misoprostol in high risk patients who cannot avoid NSAIDs

    Topical NSAIDs in acute and chronic pain conditions

    Likely to be beneficial:

    Omeprazole and, to a lesser extent, H2 blockers in high risk patients who cannot avoid NSAIDs

    Trade-off between benefits and harms:

    NSAIDs in osteoarthritis

    Unknown effectiveness:

    NSAIDs versus simple analgesics in acute musculoskeletal syndromes

    Switch between different NSAIDs

    Topical NSAIDs versus systemic NSAIDs or alternative analgesics

    Unlikely to be beneficial:

    NSAIDs in increased doses

    Question: Are there any important differences between available NSAIDs?

    Systematic reviews have found no important differences in effect between different NSAIDs or doses but have found differences in toxicity related to increased doses and possibly to the nature of the NSAID itself. In acute musculoskeletal syndromes, no large double blind trials have compared an NSAID with paracetamol. We found no evidence that NSAIDs are more effective than simple analgesics.


    Rheumatoid arthritis: A systematic review of 37 crossover trials compared indomethacin with 10 newer NSAIDs; it included 1416 patients treated for a median of two weeks with each drug. Only 5% more people (95% confidence interval 0% to 10%) preferred the newer NSAID to indomethacin.4 Four of the trials included a placebo period and one trial compared four drugs. Another systematic review of 88 trials comparing two NSAIDs, and 27 trials comparing an NSAID with placebo, in a total of 6440 patients, found no differences in the number of tender joints in 17 different NSAIDs.5 Studies claiming that individuals may benefit from different NSAIDs are not convincing; people's preferences for particular drugs have not been replicated and could therefore be due to chance or to the considerable natural fluctuations in disease activity.6 7 Osteoarthritis: Two reviews found no clear differences between various NSAIDs used to treat osteoarthritis of the hip (39 trials)8 or the knee (16 trials).9 A systematic review comparing tenoxicam with three other NSAIDs for osteoarthritis10 found superiority of tenoxicam over piroxicam both for global efficacy (10 trials, 834 people; odds ratio 1.46, 1.08 to 2.03) and for global tolerability (seven trials, 974 people; 1.46, 1.01 to 2.15). This result is at variance with a large RCT of 1328 people with osteoarthritis or rheumatoid arthritis, which found no significant differences in global efficacy or tolerability between the two drugs; improvement was noted for 55% of patients receiving tenoxicam compared with 53% of patients on piroxicam (difference 2%, −5% to 9%)11 12 Acute musculoskeletal syndromes: The RCTs were generally of poor quality, 13 and we found few systematic reviews. One review of 17 NSAID trials for shoulder pain was inconclusive.14 15 Another systematic review of 84 studies included 32 025 people with soft tissue injuries of the ankle; it was unable to pool data to perform a statistical analysis of the different forms of treatment.13 We found no good evidence that NSAIDs are more effective than paracetamol in acute musculoskeletal syndromes. Dose-response relation: A systematic review of 19 trials in which participants were randomised to more than one dose of nine different NSAIDs found a dose-response relation that saturated at high doses, 16 as for most other drugs. This and another systematic review found that the recommended dosages were close to providing a ceiling effect.16 17 The review of the 115 trials mentioned above found no differences between various doses of drugs5; 10 of 21 studies of ibuprofen had used a daily dosage of 1200 mg or less.


    Versus placebo: A systematic review of 100 trials (12 853 people) found an increased risk of gross haemorrhage (ARI (absolute risk increase) for NSAID v placebo 0.7%, −0.1% to 1.5%) and for proved ulcer (ARI for NSAID v placebo 0.05%, −0.01% to 0.11%); the effect was not significant but mean treatment duration was only two months.18 For 40 aspirin trials (22 234 people, mean treatment duration one year) the review found an increased risk of gross haemorrhage (ARI 0.6%, 0.2% to 1.0%) and for proved ulcer (ARI 0.6%, 0% to 1.2%).18 The number needed to harm was in the range 100-1000. A systematic review of 38 placebo controlled trials found that NSAIDs raised mean blood pressure by 5.0 mm Hg (1.2 mm Hg to 8.7 mm Hg).19 Versus other NSAIDs: A meta-analysis of 11 case-control studies and one cohort study found that ibuprofen was significantly less toxic than other NSAIDs; the 11 comparator drugs were associated with a 1.6-9.2 fold increase in risk of serious upper gastrointestinal complications.20 An overall assessment of newer cox-2 inhibitors is not yet possible as trials assessing the frequency of clinically important ulceration have been published only as abstracts, and there are concerns that these drugs may retard ulcer healing.21 Dose response relation: Three systematic reviews found no ceiling effect for adverse effects; the incidence of adverse effects increased in an approximately linear fashion with dose.17 20 22


    Important differences in adverse effects seem to exist between different NSAIDs. In contrast, their beneficial effects seem to be similar. The only meta-analysis that found one drug to be more effective than another was funded by the manufacturer; the studies it included were small and the result was not replicated in a large RCT. If the effect of one NSAID is not satisfactory, a switch to another NSAID cannot be expected to solve the problem. Likewise, doubling the dose of an NSAID leads to only a small increase in effect, which may not be clinically relevant. In acute musculoskeletal problems it is doubtful whether NSAIDs have any clinically relevant anti-inflammatory effect; it is therefore unfortunate that not one, large double blind RCT has compared an NSAID with paracetamol. Paracetamol has been studied in osteoarthritis, where it had much the same effect as ibuprofen or naproxen.12

    Question: What are the effects of cotreatments to reduce the risk of gastrointestinal adverse effects of NSAIDs?

    One large RCT found that misoprostol slightly reduced the incidence of clinically important gastrointestinal complications of oral NSAIDs. A large RCT found that omeprazole 20 mg and 40 mg daily and misoprostol 800 μg daily had similar effects in reducing endoscopically diagnosed adverse effects. However, misoprostol caused more adverse effects (mostly diarrhoea and abdominal pain). RCTs found that H2 blockers were inferior to omeprazole and misoprostol. The absolute risk of serious upper gastrointestinal complications is approximately doubled for patients aged over 75 and for those with a history of peptic ulcer, bleeding, or cardiovascular disease.


    We found one systematic review and four large subsequent RCTs. Misoprostol versus placebo: The systematic review of 24 trials included 5005 people who all received NSAIDs.23 In trials lasting more than four weeks, misoprostol compared with placebo reduced the development of gastric ulcer detected by routine endoscopy (ARR (absolute risk reduction) 8%, 1% to 18%). H2 blockers had no effect on gastric ulcer (ARR 1%, −2% to 3%) but reduced the risk of duodenal ulcer (ARR 2%, 0% to 5%). Misoprostol also reduced duodenal ulcer risk with long term use (ARR 3%, 0% to 6%) but not with short term use (ARR 2%, −2% to 6%). In a six month RCT published after the systematic review, 8843 people with rheumatoid arthritis (mean age 68 years, all treated with NSAIDs) were randomised to misoprostol 800 μg daily or placebo.24 Serious upper gastrointestinal complications (such as perforation, gastric outlet obstruction, or bleeding detected by clinical symptoms or investigation) were reduced by misoprostol compared with placebo (absolute risk 0.6% in misoprostol group, 1.0% in placebo group; ARR 0.4%, 0% to 0.7%; NNT (number needed to treat) 263). The risk of serious upper gastrointestinal complications was approximately doubled for people aged over 75 and for those with a history of peptic ulcer, bleeding, or cardiovascular disease. People with all four risk factors had an absolute risk of a major complication in six months of 9%, corresponding to a number needed to treat with misoprostol rather than placebo of 28 people. Misoprostol versus omeprazole: In an RCT of 935 patients treated with NSAIDs who had ulcers or more than 10 erosions at endoscopy, treatment success was defined as fewer than five erosions at each site, no ulcers, and not more than mild dyspepsia.25 At eight weeks, treatment was successful in 76% of those given omeprazole 20 mg daily, 75% given omeprazole 40 mg daily, and 71% of those given misoprostol 800 μg daily. Participants (732) in whom treatment was successful were re-randomised to maintenance treatment for six months. More people remained in remission with 20 mg of omeprazole (61%) than with 400 μg of misoprostol (48%), and with either drug than with placebo (27%) (ARR for omeprazole v placebo 34%, 25% to 43%; NNT 3; ARR for misoprostol v placebo 21%, 12% to 30%; NNT 5). H2blockers: In a similarly designed trial in 541 participants, treatment was successful in 80%, 79%, and 63% of participants given 20 mg or 40 mg of omeprazole or 300 mg of ranitidine daily, respectively.26 The estimated proportions in remission after six months were 72% with omeprazole 20 mg and 59% with ranitidine 300 mg (ARR for omeprazole v ranitidine 13%, 4% to 22%; NNT 8). In an eight week study, 538 people with NSAID related upper gastrointestinal pain without endoscopic evidence of ulcers were randomised to either misoprostol 800 μg or ranitidine 300 mg daily.27 One third were excluded from analysis because of problems with adherence and missing endoscopic examinations. Gastric ulcers at least 3 mm in diameter were found in 1% of people with misoprostol and in 6% of people with ranitidine (ARR for misoprostol v ranitidine 5%, 2% to 9%; NNT 20). Duodenal ulcer rate was 1% with both drugs.


    In one of the large studies more participants receiving misoprostol (27%) than placebo (20%) withdrew from the study because of adverse events, primarily diarrhoea and abdominal pain.24 There were 17 and 21 deaths respectively. One death was a direct result of gastrointestinal toxicity. In another study, adverse events were noted in 48%, 46%, and 59% of participants receiving omeprazole 20 mg and 40 mg, and misoprostol, respectively; treatment discontinuations (all causes) occurred in 10%, 11%, and 17%.25 Few adverse events were reported in the third study; treatment discontinuations (all causes) occurred in 10%, 10%, and 14%.26 In the fourth study, adverse events (mostly gastrointestinal) occurred in 77% of participants taking misoprostol and 66% taking ranitidine, with withdrawal rates of 13% and 7%, respectively (ARR for ranitidine v misoprostol 6%, 1% to 11%; NNT 17).27


    The clinical relevance of these findings is doubtful. The only trial that used clinically relevant outcomes found little difference between active drug and placebo, except for high risk patients. The rate of ulcers was more than 10 times higher in the studies where the investigators looked for them with regular endoscopy, than in earlier trials of NSAIDs.18 These ulcers were sometimes defined as endoscopic lesions with a size of only 3 mm, or as any lesion of an unequivocal depth, or no definition was provided at all.23

    Question: What are the effects of topical NSAIDs?

    One systematic review has found that topical NSAIDs are effective compared with placebo in acute pain conditions (NNT 5 to obtain good pain relief) and chronic pain conditions (NNT 3). Topical and oral formulations of the same drug have not been compared in high quality trials. Topical NSAIDs have not been compared directly with paracetamol. It is therefore uncertain whether topical treatment has advantages over these alternatives.

    Key messages

    Systematic reviews of RCTs have found no important differences in effect between different NSAIDs or doses but have found differences in toxicity related to increased doses and possibly to the nature of the NSAID itself

    In acute musculoskeletal syndromes, we found no large double blind trial that compared an NSAID with paracetamol

    One large RCT has found that misoprostol slightly reduces the incidence of clinically important gastrointestinal complications of oral NSAIDs

    One large RCT has found that omeprazole 20 mg and 40 mg daily and misoprostol 800 μg daily produce similar reductions of endoscopically diagnosed ulceration. Misoprostol causes more adverse effects (mostly diarrhoea and abdominal pain)

    A systematic review of RCTs has found that topical NSAIDs are effective compared with placebo in acute pain conditions (NNT 5 to obtain good pain relief) and chronic pain conditions (NNT 3)

    Topical NSAIDs have not been compared in high quality trials against oral forms of the same drug or against paracetamol. Whether or not topical administration is advantageous remains uncertain


    Versus placebo: We found one systematic review of 86 trials in 10 160 people, 28 most comparing a topical NSAID with placebo. The review was partly sponsored by two manufacturers. Many trials of acute pain conditions (soft tissue trauma, strains, and sprains) were small; in 37 trials, the average number of actively treated patients was 32, and the effect declined significantly with sample size. In seven trials with more than 80 patients per group, the relative benefit was 1.6 (1.3 to 1.9) and number needed to treat for a good outcome was 5 (4 to 6). In 12 trials in chronic pain conditions (osteoarthritis, tendinitis) the relative benefit was 2.0 (1.5 to 2.7) and the number needed to treat was 3 (3 to 4). In an additional trial, which was not included in this review, copper-salicylate gel or placebo was applied to the forearm of 116 people with osteoarthritis of the hip or knee.29 There was no difference between groups (22% v 21% patients reported good effect). Versus oral NSAIDs: Five trials in the systematic review compared topical with oral NSAIDs but they all had inadequate design and power.28 Versus paracetamol: We found no RCTs.


    In the systematic review, local adverse effects occurred in 3% of participants in both groups and systemic adverse events in 1%.28 In the trial of copper-salicylate gel, more participants taking the drug reported adverse reactions than with placebo (83% v 52%), most commonly skin reactions, and more withdrew from the trial because of these reactions (17% v 2%).29


    Sample size bias hampers the interpretation of the available trials. Further, no high quality trials have compared topical with systemic administration of the same NSAID, and no trials have compared a topical NSAID with paracetamol. It is therefore not possible to judge the relative benefit of this route of administration.


    Clinical Evidence musculoskeletal disorders advisers, Troels Mørk Hansen, Herlev, Denmark, and John Stothard, Middlesbrough, UK.


    • Competing interests None declared.

    • Embedded Image
      This review is one of 87 chapters from the second issue of Clinical Evidence

      Clinical Evidence is published by BMJ Publishing Group and American College of Physicians-American Society of Internal Medicine. The second issue is available now, and Clinical Evidence will be updated and expanded every six months. Individual subscription rate, issues 2 and 3, £55/$115; institution rate £132/$211. For more information including how to subscribe, please visit the Clinical Evidence website at


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    13. 13.
    14. 14.
    15. 15.
    16. 16.
    17. 17.
    18. 18.
    19. 19.
    20. 20.
    21. 21.
    22. 22.
    23. 23.
    24. 24.
    25. 25.
    26. 26.
    27. 27.
    28. 28.
    29. 29.
    View Abstract