Budesonide: Is No Evidence of a Difference the Same as Evidence of No Difference?

Corticosteroids have consistently been shown to be effective in the management of both ulcerative colitis and Crohn's disease (CD). In fact it could be argued that the establishment of corticosteroids by Truelove et al. (1) as effective therapy for the induction of remission converted a disease of high mortality into a disease with normal life expectancy. Their efficacy was so obvious to the medical community that relatively few studies were performed making life difficult for the eager metaanalyst. Munkholm and associates have skillfully documented the generally positive outcome that one can expect when corticosteroids are prescribed for patients with active CD (2). Eighty percent of patients with CD in this population-based study either entered remission or improved after 30 days of oral prednisolone. The other 20% either worsened or did not respond while on steroid therapy. For those who responded, 55% had a prolonged response. Many of the adverse events related to steroids are cumulative and can be reduced by trying to minimize the duration of therapy. Steroids are generally regarded as important only in the induction of remission. There is little evidence that conventional corticosteroids have a role to play in the maintenance of remission (3,4). The search for the “magic bullet” for CD continues, and it has been suggested that one candidate for the honor is budesonide. This second generation corticosteroid is currently available in Canada and Europe in two formulations. The controlled-release formulation consists of microgranules of budesonide coated with a pH-sensitive resin. The other product consists of a pellet coated with a pH-sensitive resin. For the purposes of this article, I will not distinguish between the two. Both preparations are promoted as effective as conventional corticosteroids with a much-improved adverse event profile. Is this the miracle steroid of the 21st century or just an old friend wrapped in a jazzy new set of clothes? Is the 20-fold cost increase for 1 month's budesonide (9 mg q. a.m.) compared with a 3-month course of prednisone (40 mg per day for 2 weeks followed by a taper) justifiable? Four randomized, double-dummy trials have compared budesonide with conventional corticosteroid therapy for the induction of remission in patients with active CD (5,–8). All four trials had the same general conclusions. Budesonide was as effective as conventional steroids without the adverse events. Roughly 600 patients were treated (300 in each study arm). Is this a logical conclusion to draw? How many patients would it require to answer this question? What does the p = NS really mean? Most clinical trials (particularly placebo-controlled trials) are designed to detect a difference between treatment arms. The p value calculated is an estimate of how confidently we can reject the null hypothesis (treatments are the same) and accept the alternative hypothesis (treatments differ). The sample size calculation for such a trial requires values to be assigned by the investigator for the α, β, and the minimum δ between the two interventions to be detected. The minimum difference is often large in placebo-controlled trials, and therefore the sample size required is relatively small. In this situation if p = NS, this means that the null hypothesis of no difference cannot be rejected (“no evidence of a difference”). In an equivalence or positive control study the issues are different. In theory the δ for an equivalence study is 0 and the sample size required is infinite. The suggested strategy for dealing with the sample size required is to arrive at a value for δ that represents the minimum difference below which both treatments might be considered to be equivalent. What should that be in the case of budesonide and conventional steroids: 5%, 10%, 20%? Figure 1 represents the sample sizes required to provide “evidence of no difference.” In this case the null hypothesis is that the two interventions differ by less than δ. The sample sizes required are based on the 80% complete or partial response by 30 days reported by Munkholm. As can be seen, the numbers required for equivalence studies for less than 10% difference are more than are found in most metaanalyses of inflammatory bowel disease therapy. The current dataset of budesonide and conventional corticosteroids will not allow such precision. Relationship between number of patients required per treatment arm to determine if there is sufficient evidence to conclude there is no difference between the two treatments. The model assumes an α of 0.05, a β of 0.10, and δ ranging from 1–20, assuming an 80% response to the control (conventional corticosteroids). In the absence of clear evidence that budesonide is superior, how else might decisions be made? Friedman and colleagues suggest several criteria for comparison of similar treatments: frequency and severity of adverse events, changes in quality of life, ease of applying the intervention, and cost (9). While one can admit that there are differences in short-term adverse events, it is not clear yet whether these differences can be extrapolated to long-term toxicity, for example, metabolic bone disease. Budesonide is a powerful steroid, and the 10% of the drug that survives first-pass hepatic metabolism may have long-term impact on adverse events. There are less data that compare quality of life offered by budesonide with conventional steroids. Ease of use is not an issue. In one trial, no significant statistical differences in quality of life (Inflammatory Bowel Diseases Questionnaire score) were reported. The cost (at least in Canada) is approximately 20-fold higher for budesonide compared with prednisone. Based on the current literature it is difficult to recommend that budesonide should replace conventional corticosteroids in the management of inflammatory bowel disease. The search for the “magic bullet” continues.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.