Casual observation suggests that many physicians and patients have concluded that the adverse outcomes are less important than the clinical benefits. Despite a flurry of media "revelations," NSAIDs continue to be one of the most commonly prescribed classes of drugs in the United States [3], and the Food and Drug Administration (FDA) has decided that ibuprofen, naproxen, and aspirin can be safely sold over the counter to be used at the patient's discretion.

Because cumulative knowledge showed that the gastropathy was associated with depletion of tissue prostaglandins and that the damage could be prevented by pretreatment with prostaglandins, the oral prostaglandin congener misoprostol (Cytotec; G.D. Searle and Co., Chicago, Illinois) was submitted for FDA approval. Despite initial concerns about misoprostol's abortifacient property, the drug was approved as "indicated for the prevention of NSAID-induced gastric ulcers in patients at high risk of complications of gastric ulcer, e.g., the elderly and patients with concomitant debilitating disease, as well as patients at high risk of developing gastric ulceration, such as patients with a history of ulcer" [4]. Despite the approved indication, debate has continued over the appropriate role of misoprostol [5-7], fostered by uncertainty over the drug's efficacy in preventing the severe complications of chronic ulcer disease, its gastrointestinal side effects, and its cost. In this issue, the results of a large randomized, double-blind, placebo-controlled study are reported [8], further fueling the fires of the debate.

Silverstein and colleagues [8] provide the results of a long-awaited trial that prospectively assessed clinically relevant severe gastrointestinal complications (outcomes) in a large group (n = 8843) of patients older than 52 years who had rheumatoid arthritis and were receiving long-term NSAID therapy. Half of these patients received 200 µg of misoprostol four times a day and were followed for 6 months. The misoprostol intention-to-treat group developed significantly fewer perforations and episodes of overt upper gastrointestinal bleeding during the observation period. However, the absolute risk for these complications, as predicted in previous epidemiologic studies [9, 10], was infrequent (absolute risk for placebo recipients, 42; risk for misoprostol recipients, 25). One placebo recipient died. It is likely that this information also applies to patients without rheumatoid arthritis who are receiving long-term NSAID therapy, but this has not been tested. In addition, overall functional outcomes (how the patients felt during the 6 months) were not measured. The dropout rate from gastrointestinal symptoms was large in both groups but was significantly greater in patients receiving misoprostol (27.5%) than in those receiving placebo (20.2%). Thus, for the first time we have strong evidence that misoprostol not only prevents gastropathy of uncertain importance but is effective in reducing the risk for the infrequent serious gastrointestinal outcomes (overt bleeding and perforation) that are only caused by large chronic ulcers. Now we must return to an essential question: Is this treatment cost-effective?

Unfortunately, the study by Silverstein and colleagues [8] was not designed to measure either the direct or the indirect costs of the medical interventions. Thus, we do not know whether the gastrointestinal side effects in 27% of patients in the misoprostol group led to further interventions (endoscopy, stool cultures, revisits to providers, treatment with antidiarrheal agents), how many units of blood were transfused in the patients with severe gastrointestinal complications, how many colonoscopies were done for occult blood in the stool, or how many work days were lost in either group. Despite these limitations, I will provide rough cost estimates to allow some context for individual and institutional decision making.

In Denver pharmacies (average of four national chains), the current cost to patients for 1 month of misoprostol at 200 µg four times a day is $90. The cost of treating 4404 patients for 6 months is $3 963 600, or $233 000 for each of the 17 serious gastrointestinal complications prevented in the misoprostol group. We can reduce these numbers to $2 021 490 and $119 000, respectively, by recognizing that 1230 of the 4404 patients could not tolerate 800 µg per day and therefore received only 400 µg per day. The average 1993 charge (not cost) for treating 17 similar complications (8 gastrointestinal bleeding episodes and 9 perforations or obstructions requiring surgery) in hospitals in the Denver metropolitan area was $318 000 ($18 700 per case). Thus, in a nonselected population of patients with rheumatoid arthritis who are receiving NSAIDs, the cost of preventive therapy with misoprostol is 6.5 times that of no treatment, a number unlikely to catch the eye of most managed care medical directors.

However, by using logistic regression, Silverstein and colleagues identified factors (previous gastrointestinal bleeding, previous peptic ulcer, age greater than 75 years, and a history of cardiac disease) that identified patients at increased risk. Whereas patients with none of these factors had a 0.4% risk for severe gastrointestinal complications, the presence of all four factors conferred a 9% risk. Expressed monetarily, the cost of preventing a single complication is $276 916 in the former group and $12 486 in the latter group.

Practitioners should recognize 1) that even this rudimentary cost analysis would need to be evaluated in the context of considering whose costs we are talking about [for example, the patient, the Health Care Financing Administration, a staff-model health maintenance organization, or a physician receiving capitation payments]; 2) that newer NSAIDs that produce nitric oxide may make this issue moot by completely separating the therapeutic effects of these drugs from their gastrointestinal toxic effects [11]; and 3) that potent, and expensive, acid suppression may be shown to be equally efficacious with fewer medication side effects [12]. Given these considerations, we must decide whether to recommend misoprostol for the next patient with rheumatoid arthritis receiving long-term NSAID therapy who comes into the office.

Currently, I would first consider whether therapy with the NSAID can be stopped and replaced by a joint-sparing medication [13]. If this is not possible, I would prescribe low doses of NSAIDs such as ibuprofen and nabumetone, which induce less gastropathy [14]. I would not use misoprostol in most patients requiring NSAIDs. However, for the small subgroup of patients with a history of peptic ulcer or gastrointestinal bleeding, especially elderly patients with concomitant cardiac disease, I would prescribe a dose of misoprostol that the patient could tolerate, such as 100 to 200 µg four times a day.