Probiotics and the irritable bowel syndrome: authors’ reply

Abstract

Sirs, We appreciate that doctors Chassany and Matuchansky comment1 on our publication about probiotic supplementation in the irritable bowel syndrome (IBS).2 This study strengthens the findings from an earlier study showing that a multispecies probiotic significantly alleviates symptoms of IBS.3 We agree that complete validity for the health-related quality of life (HRQOL) questionnaire remains to be achieved. We do not, on the other hand, agree that the slight difference between the groups in baseline symptom severity creates a methodological concern, as baseline-correction (analysis of covariance with baseline as covariate) was used in the statistical analyses. Despite proper randomisation, it may, in clinical trials, unwittingly happen that groups differ for some variables, and it is a recommended and accepted procedure to correct for this by using appropriate statistics. Concerning end points, current recommendations on trial design recommend that trials should be based on one, or at most two, primary end points,4, 5 and that the primary efficacy end point should combine, in a global assessment, the key symptoms of IBS. A summary score, as the one used in our study, is one advocated outcome measure. A classification of patients into responders and non-responders based on the primary end point may additionally be done if the definition of a responder has been decided a priori. If this has not been done, it is not correct to analyse the proportion of responders. We are criticized for not citing the recent trial by Guyonnet et al.6 in the HRQL part of the Discussion: the results in that trial showed that the HRQL score improved similarly in both the probiotic group and the placebo group (P < 0.001), but no difference between the groups could be observed. As the placebo response in IBS is estimated to be up to 70%,7 within-group analysis cannot, in our opinion, be considered an appropriate method to evaluate the effect of a treatment in this syndrome. In the responder rate analysis, Guyonnet et al. saw a significant difference between the groups in the HRQOL abdominal discomfort score only half-way through the trial, but not at the end. The effect on HRQOL was consequently not particularly obvious, as the beneficial effect was noticed at 3 weeks, but it did not remain at the end of the study at 6 weeks. With reference to the magnitude of the improvement of the HRQOL bowel symptoms domain in our trial, the median change in the score was 0.62 in the probiotic group, and 0.37 in the placebo group (P for difference between groups = 0.045). The authors of the questionnaire8 suggest that a mean change of 0.5 on the 1–7 scale approximates the minimal important difference in questionnaire score, but we agree that the actual clinical significance of HRQOL improvements in general is a demanding question. Because of the lack of a biological marker, clinical trials in IBS remain challenging to plan, conduct and interpret. We thus see all debates that aim to improve methodological issues in IBS trials as welcome and valuable. Declaration of personal interests: K. Kajander and R. Korpela are employees of Valio Ltd, Finland. Declaration of funding interests: The study by Kajander et al. was funded by Valio Ltd and the Finnish Funding Agency for Technology and Innovation. The preparation of the manuscript of the study was funded in part by the Academy of Finland.