Skip navigation

GI complications and NSAIDs

Gastrointestinal complications, up to and including serious bleeding episodes, are a recognised complication of oral NSAID use. Much of the information about this comes from case-control or cohort studies, which have shown an incidence of such events of about 2% - that is occurring in 2 people out of every 100 treated in any one year.

Imagine an RCT...

Randomised controlled studies are usually about efficacy and have relatively small group sizes. While RCTs provide useful information about adverse events which are common, they are usually poor indicators of adverse events which are rare.

Imagine doing a study to examine the effects of co-administration of misoprostol with NSAIDs to protect against serious upper gastrointestinal complications in older patients with chronic rheumatoid arthritis who are taking NSAIDs. Imagine doing a randomised controlled study - what sort of numbers would you need?

If we assume that misoprostol reduces adverse events by 50% or so, and that most patients finished the study, then for a study over six months we would need 7,500 patients.

The actual RCT

Such a study was done in the USA sponsored by GD Searle [1]. Six hundred and sixty four practices in the USA and Canada recruited 8,843 men and women, with a mean age of 68 years, and with most over age 60. These patients took their regular NSAID, but were randomised to receive 400-800 µg misoprostol or matched placebo daily for six months.


All suspicious events were reported. These were then examined by an external review committee of a gastroenterologist, a rheumatologist and an epidemiologist. Without unblinding the treatments the committee determined whether the patient had upper gastrointestinal bleeding. The committee developed a range of 11 different definitions - eight of definite upper GI complications (including bleeding and perforation) and three of "probable" complications.

The paper reports each of these separately, but for simplicity, and because the definitions were not made before the trial started, Bandolier here addresses information about all 11 definitions.

Complications in placebo group

61 of 4439 patients receiving placebo had complications - about 1.4% over six months, or 2.8% over one year.

A multiple linear regression model which looked at 18 potential risk factors indicated that risk factors for serious complications with oral NSAIDs were age 75 years or more, history of peptic ulcer, history of gastrointestinal bleeding and history of heart disease.

The model predicted that for patients with none of the risk factors, the one-year risk of a complication was 0.8%, for patients with any single risk factor it was 2%, and for patients with all four factors it was 18%. With combinations of three of the factors, the one year risk was 8 - 10%.

Effects of misoprostol

34 of 4404 patients receiving misoprostol had bleeding episodes - 0.8% over six months, or 1.6% over one year. This reduction in total complications was significant (odds ratio 0.56; 95%CI 0.37 - 0.85).

Adverse events

Withdrawal because of diarrhoea, abdominal pain or flatulence in the first month of treatment occurred in 20% of patients receiving misoprostol significantly more than 15% of patients receiving placebo. There were some deaths during the study, but these were similar in number in each treatment group.


For misoprostol to prevent one bleeding event compared with placebo in one year, the number-needed-to-treat was 83 (95%CI 55 - 160). This seems to be high, but NNT is dependent upon the risk of an event happening; the more infrequent the event is, the more patients needed to be treated to prevent one event.

If misoprostol was perfect, and prevented all bleeding events, the NNT would still be 36. NNT values for preventative measures tend to be relatively high, like the NNT for 40 for aspirin against nothing to prevent one death at five weeks after infarction, or an NNT of 15 to prevent one coronary event for simvastatin compared with placebo over five years ( Bandolier 17 ).

NNT and underlying risk

Just how the NNT is affected by the risk of an event happening is shown in the box and figure. A similar degree of risk reduction gives progressively larger NNTs (and wider confidence intervals) as the underlying risk of the event happening diminishes.


This was an interesting, and possibly heroic study. Because of the large numbers of patients studied, it provides useful RCT data about upper gastrointestinal complications in elderly people taking NSAIDS. It confirms information on the incidence of these complications obtained from other sources.

Does it tell us anything about misoprostol? Well, it does demonstrate that misoprostol can reduce the incidence of upper gastrointestinal complications. The number of patients needed to be treated to prevent one event was large because of the comparative rarity of the event.

It may well be that in patients with the risk factors identified in this study treatment with misoprostol would be appropriate. As an example, with the same treatment effect of misoprostol as for the whole study (about a 40% reduction in serious complications), but with an 18% incidence of complications, the NNT would be 14.

It may well be the case that NNT calculations done from the trial would show even lower NNTs than this in patients with higher risk. It is likely that clear and strong guidelines for prophylactic strategy could be derived from the information on patients at higher risk if the information from this study could be made available.

Bandolier 16 examined some of the issues about prophylaxis, and produced a diagram as an aide-memoir for the issues, which is reproduced here as well.

It must be remembered that there may be other treatment options available which might have lower risks of gastrointestinal complications. One such would be the use of simple analgesics like paracetamol, rather than the NSAID [2], and another might be topical application of NSAID, which do not cause gastrointestinal complications [3].


  1. FE Silverstein, DY Graham, JR Senior et al. Misoprostol reduces serious gastrointestinal complications in patients with rheumatoid arthritis receiving nonsteroidal anti-inflammatory drugs. Annals of Internal Medicine 1995 123:241-9.
  2. L March, L Irwig, J Schwartz et al. n Of 1 trials comparing a non-steroidal anti-inflammatory drug with paracetamol in osteoarthritis. British Medical Journal 1994 309:1041-6.
  3. Evans JMM, McMahon A, McGilchrist M, et al. Topical non-steroidal anti-inflammatory drugs and admission to hospital for upper gastrointestinal bleeding and perforation: a record linkage case-control study. British Medical Journal 1995 311:22-6.

next story in this issue