Controlled trial of anti-tuberculous chemotherapy for two years in Crohn's disease.

One hundred and thirty patients with active symptoms of Crohn's disease were treated in a double blind randomised controlled trial with rifampicin, isoniazid, and ethambutol, or identical placebos for up to two years. All other treatment considered necessary was continued. Analyses were based on 126 patients, 63 in each treatment group. Thirty seven in the active and 30 in the placebo group had previous surgical procedures. There was no difference in concomitant treatment between the two groups. Thirty in the active and 46 in the placebo groups were taking corticosteroids at entry to the trial. Forty eight of 63 patients in the active and 49 of 63 in the placebo group, completed at least 12 months' therapy. Reasons for early withdrawal included pregnancy, adverse reaction, and failure to comply. There was no significant difference in the mean number of months completed between the two groups. Nineteen adverse reactions were recorded for 17 patients in the active group compared with three reactions in patients on placebo. All of the nine patients withdrawn early because of adverse reactions were in the active group. Fifteen patients on active treatment and 14 on placebo had surgery during the trial with no difference in the type of surgery required between the groups. Radiological assessments based on 98 patients at the end of the trial showed no significant differences between groups in changes of extent of disease. More patients developed strictures on placebo compared with active treatment but without a statistically significant difference. No differences were found between groups for the total prednisolone dose or the number of days on which prednisolone dose was 10 mg or above. Serial measurements of body weight and Crohn's disease activity index (CDAI) together with blood values for albumin, haemoglobin, white cell count, and platelets showed no consistent different differences between groups. There were occasional significant differences for some of these values between groups, which were not sustained. The trail provides little evidence of tangible benefit from the trail treatment.