Is there enough evidence to support use of N-acetylcysteine in contrast-induced nephropathy?

TO THE EDITOR: I would like to applaud Kelly and colleagues for their exhaustive meta-analysis (1), and I have some comments about their analysis. First, exactly how many studies involving N-acetylcysteine were included in the meta-analysis? The number of N-acetylcysteine studies in Figure 1 differs from the numbers listed in the Table and Figure 2. Why were 4 studies excluded after they were deemed to have satisfied inclusion criteria? Second, the paper cited as a study involving furosemide and mannitol (Table) is a review relating to selection of contrast media (2). The actual mannitol and furosemide study was published elsewhere (3). Such errors in referencing tend to perpetuate. Second, the statistically significant heterogeneity with respect to N-acetylcysteine effect merits comment. Some statistical authorities suggest that if there is substantial heterogeneity, then a summary effect should not be derived (4). This is a point that warrants discussion. The random-effects model does not necessarily eliminate the problem of heterogeneity (5). Particularly, if a substantial number of studies differ in treatment effect and direction (Figure 2), then combining them could lead to misleading conclusions (5). Thus, consistent with past meta-analyses, the effect of N-acetylcysteine remains uncertain. To conclude otherwise is incorrect. Furthermore, I want to draw attention to the Editors’ Notes. The paper presents no evidence of the benefit of theophylline. The analysis itself, which includes 1 aminophylline trial (6), did not reach statistical significance. On closer examination, 2 included studies were published from the same institute and had similar methods (7, 8). There are differences: For example, the first study was not restricted to individuals undergoing coronary angiography but had 54 such participants. The second study included only participants undergoing coronary angiography. It is possible that the second study included the same 54 coronary angiography patients from the first study. If so, because both studies were positive, this could bias results in favor of theophylline. Finally, the editors’ comment regarding the benefit of mannitol merits correction. The authors presented only 1 mannitol study, which had shown that mannitol was harmful compared with one half the normal dose of saline (3). I applaud the authors and the editors for making the point that change in creatinine level is a surrogate laboratory outcome. It is of greater importance to show the benefit of preventive interventions on clinical outcomes, such as death, on which data are lacking.