In the second paragraph of this article, it was noted that patients with the most to least adequate correction of elevated Th1/Th2 ratios had a statistically significant improvement in birth rate. The authors reported that only the implantation rate was significantly improved (P = 0.04), and the live birth rate per embryo (LBR/e) showed a trend to improvement (P = 0.09). However, on checking the P values, it became apparent that the authors had used a two-tail Fisher’s Exact test, but as one knew a priori in which direction a treatment difference was expected; the appropriate P values reported should have been P = 0.02 for improved implantation rate and P = 0.045 for improved live birth rate per embryo.
Further, a priori, one would have expected the best outcome for Group II (moderately high Th1/Th2) with adequate correction, then Group I (very high Th1/Th2) with adequate correction, then Group II with inadequate correction, then Group I with inadequate correction, and finally, the untreated patient controls. For LBR/e, the respective success rates were 51%, 41%, 36%, 33% and 13%. Even omitting the untreated control group, by a non-parametric Spearman Rank correlation coefficient, the P value is impressive, and a similar result obtains for implantation rates. The ‘dose response’ relationship discussed in my commentary supports the idea that the high success rate in the most successful group compared to the least successful group was not a fluke. The author regrets this was not made sufficiently clear; there are very short time lines to write a commentary on an accepted paper if it is to be reviewed and revised and appear in the same issue as the accepted paper.