In Reply

We thank Drs. Groeneveld et al. and Widermann for their interesting comments on our article.¹  Groeneveld et al. made a remark that in CHEST trial, the temporal effects of serum creatinine increase became apparent only between days 1 and 4. Yet the increase was from ±110 to ±116 µm, which is certainly not clinically relevant. Serum creatinine or creatinine clearance are not perfect biomarkers for renal injury, but no other marker is universally accepted in the field even though many have been studied. Although the CHEST trial²  might be a landmark study, it was conducted in patients in intensive care unit, and we purposely excluded this patient population from our meta-analysis. In the CHEST trial²  only 1,574 of 6,742 patients were randomized after elective surgery (the type of patients we evaluated). When such patients were admitted to intensive care units, it was probably because they suffered intraoperative complications, which may have put them at higher risk of delayed renal complications.

We do not think that the comparison of our work with the meta-analysis by Zarychanski et al.³  is adequate. This meta-analysis has included very heterogeneous groups of patients, including severely ill patients in intensive care units, many of them with different forms of shock including septic shock. Such a population is at high risk of organ dysfunction, including acute kidney failure, and cannot be compared with elective surgical patients. Also, we do not think that our meta-analysis is comparable with analysis by Gattas et al.⁴  In this meta-analysis, the studies showing a higher risk for the need of renal replacement therapy (fig 3, top panel⁴ ) were all conducted in patients in intensive care unit, with only one exception (Nagpal et al.). In the middle panel of figure 3 of our article,¹  trials conducted in surgical patients are presented. These trials were not associated with a higher risk of renal replacement therapy (risk ratio, 0.46; 95% CI, 0.10–2.05; P = 0.794). Therefore both meta-analyses either are inadequate to address the clinical question that we wanted to address or found comparable evidence.

Surgical patients were also evaluated by Van Der Linden et al.⁵ . In their meta-analysis, 2,139 patients treated with tetrastarches were compared with 2,390 patients treated with a comparator. From 39 trials, the authors concluded that tetrastarches used during surgery did not induce adverse renal effects as assessed by changes in serum creatinine or need for renal replacement therapy. The authors reported 21 studies documenting serum creatinine or creatinine clearance after administration of 130/0.4 starch or other tested fluids. One thousand five patients were given a tetrastarch and 1,051 patients were given a comparator. The period for which creatinine was reported covered up to 14 days after administration. All but three studies showed no difference in peak creatinine concentration. Two studies found a statistically better outcome for the tetrastarch. The authors concluded that they could not detect a hint for an adverse signal after the use of modern starch in surgical patients.⁵  The risk of excessive bleeding was out of the scope of our meta-analysis, but the results of the Van Der Linden meta-analysis are reassuring with this regard. Every meta-analysis can only be as reliable as the data available. In this way, it is in fact limited, and this point was emphasized at the end of our discussion. But, even though only two of the trials in the meta-analysis were primarily designed to evaluate the renal effect of hydroxyethyl starch 130/0.4, this side effect of colloids was well known since long and thus was an integral safety parameter in all of these trials.

We are confident that our conclusions are meaningful today and can hold in the light of upcoming evidence. Although Groeneveld et al. point out that a retrospective analysis has found an association between hydroxyethyl starch 130/0.4 and renal replacement therapy, we would like to draw the readers’ attention to a recently published prospective randomized study in patients undergoing abdominal surgery by Feldheiser et al.⁶  demonstrating that a stringent treatment algorithm and an adequate monitoring results in better hemodynamic stability and reduced need for fresh-frozen plasma. This study included a 3-month follow-up and measured the sensitive renal marker neutrophil gelatinase-associated lipocalin. If older studies might not provide the evidence, we would wish for today, this is also true for all studies that did not use rigorous protocols to identify patients who were in need of volume therapy.