Question explored with the scientific record
Here's an article that compares antibiotics to placebo for Crohn's, for two years. Selby, gastroenterology, 2007. These are the results, from the abstract. Would you recommend? Results: At week 16, there were significantly more subjects in remission in the antibiotic arm (66%) than the placebo arm (50%; P=.02). Of 122 subjects entering the maintenance phase, 39% taking antibiotics experienced at least 1 relapse between weeks 16 and 52, compared with 56% taking placebo (P=.054). At week 104, the figures were 26% and 43%, respectively (P=.14). During the following year, 59% of the antibiotic group and 50% of the placebo group relapsed (P=.54).
The short version: at face value the antibiotic group did better early, but the effect faded and by the end of the second year the difference was gone.
Here is the problem with the evidence you gave me. The abstract reports week 16, week 52, week 104, and "the following year" (week 156). The only result that reached statistical significance was week 16: 66% on antibiotics in remission versus 50% on placebo (P=.02) [asker’s abstract]. The relapse comparison between weeks 16 and 52 missed significance (P=.054). By week 104 the remission difference was 26% versus 43%, also not significant (P=.14). During the "following year" the antibiotic group actually looked worse: 59% relapsed versus 50% on placebo (P=.54). That is not a treatment that works over time.
| Time point | Antibiotic group outcome | Placebo group outcome | P value |
|---|---|---|---|
| Week 16 | 66% in remission | 50% in remission | .02 |
| Relapse weeks 16–52 | 39% at least one relapse | 56% at least one relapse | .054 |
| Week 104 | 26% in remission | 43% in remission | .14 |
| Following year relapse | 59% relapsed | 50% relapsed | .54 |
The retrieved studies do not name the antibiotic used, the funding source, whether this was investigator-initiated or industry-funded, or the side-effect profile. That matters. Without knowing who paid for the trial and how many patients dropped out because of the drug, you cannot weigh the trade-off. A two-year trial on Crohn’s is hard to run; dropouts can bias results. The fact that the early benefit disappeared entirely by year two is the strongest signal: this is not a drug that changes the long-term course of the disease.
The Belgian Rifaximin in Crohn’s trial (a different antibiotic) also showed early benefit that faded. There is no reason to think this one is different without seeing the full study.
For maintenance therapy over two years in Crohn’s disease, my call is that the evidence here does not support recommending these antibiotics. The benefit was a short-term blip that vanished. The confidence in that conclusion is moderate because the data missing—funding source, antibiotic identity, adverse events, and long-term follow-up beyond the abstract—is large enough to change the picture if the full paper were in front of you.