Re: Characterization of chronic and acut e ESA hyporesponse(Sibbel et al 2015)P-values wrong?
Hans-Joachim Kremer, Medical Writing Service
26 February 2016
Sirs,
I have read your article and found the p-values in Table 2 implausible.
Calcium has identical means and medians throughout, but the p-value is given as "<0.0001". Instead, CRP has large numerical difference and my quick'n'dirty check indicated dozens of zeros before the first digit, but the table tells us 0.0157. Please check all p-values. It appears that CRP is a superb predictor for hyporesponse, and that this, althought not so new, might be the major message of the trial.
If, as I expect, CRP is the superb predictor, then you should have run some additional analysis based on this factor. Maybe transfusions or deaths are better predicted by CRP than by the ESA dose.
Finally, I wonder why you did not define your dose cohorts based on a weight-normalised dose. The SPCs recommend dosing by weight, and in fact, it appears that the cohort of "chronic hyporesponders" is a bit more overweight than the "ESA responders", maybe because their weight-normalised dose was too low. The slight (relative) overrepresentation of females among the "ESA responnders" appears to have the same cause, namely the usually lower weight.
As, however, CRP is also correlated with overweight, it could even turn out that only CRP is of importance.
Re: Characterization of chronic and acut e ESA hyporesponse(Sibbel et al 2015)P-values wrong?
26 February 2016
Sirs,
I have read your article and found the p-values in Table 2 implausible.
Calcium has identical means and medians throughout, but the p-value is given as "<0.0001". Instead, CRP has large numerical difference and my quick'n'dirty check indicated dozens of zeros before the first digit, but the table tells us 0.0157. Please check all p-values. It appears that CRP is a superb predictor for hyporesponse, and that this, althought not so new, might be the major message of the trial.
If, as I expect, CRP is the superb predictor, then you should have run some additional analysis based on this factor. Maybe transfusions or deaths are better predicted by CRP than by the ESA dose.
Finally, I wonder why you did not define your dose cohorts based on a weight-normalised dose. The SPCs recommend dosing by weight, and in fact, it appears that the cohort of "chronic hyporesponders" is a bit more overweight than the "ESA responders", maybe because their weight-normalised dose was too low. The slight (relative) overrepresentation of females among the "ESA responnders" appears to have the same cause, namely the usually lower weight.
As, however, CRP is also correlated with overweight, it could even turn out that only CRP is of importance.
Competing interests
None.