I would like to congratulate Li and co-authors for this meticulous analysis. Specifically, the subgroup analyses do help in a better understanding of this topic. A few comments:

1. The difference in effect size between the subgroups with and without renal impairment (figure 6), with a protective effect amongst the studies done in patients without renal impairment, is indeed quite perplexing. Indeed, if one performs a meta-regression with the baseline creatinine of the control group as a moderator, the slope is significant (p = 0.034), suggesting a greater beneficial effect in patients with lower creatinine, which would be somewhat paradoxical. This actually is similar to the protective effect of n-acetyl cysteine in low risk patients (as demonstrated elegantly by Gonzales et al in "BMC Medicine, 5:32, 2007").

2. The myth of the protective effects of n-acetyl-cysteine should be allowed to retire. The above alluded paper by Gonzales went a long way to explaining the incongruent and heterogeneous results seen with NAC. The more recent publication of the ACT study by Berwanger et al ("Circulation. 2011 Sep 13;124(11):1250-9") was the largest RCT of NAC with clinically relevant endpoints and showed no difference at all between high dose NAC and placebo.

Thanks again for a well done and written paper which does add significantly to the contrast nephropathy literature.

Swapnil Hiremath