Hydroxyethyl starch on kidney and haemostatic function in cardiac surgical patients: is a non‐inferiority study design appropriate for this setting? A reply

Abstract

Drs Giordano, Pugliese and Bilotta [1] question equipoise for our recent trial and whether a non-inferiority design was appropriate. Specifically, they criticise our statement that the results of previous hydroxyethyl starch (HES) trials are inconsistent. Major trials show that third-generation HES vs. crystalloid slightly increases renal injury in critically ill and septic patients who are given large volumes of starch over a period of days, with variable effects on early mortality and no difference on long-term mortality [2, 3]. However, another randomised trial in hypovolaemic critical care patients (n = 2857) reported lower 90-day mortality, more vasopressor-free days, and ventilator-free days in patients given colloids, including HES [4]. Trials in non-cardiac surgical patients differ in showing no short-term or longterm renal injury (n = 1057) [5]. There is also compelling evidence that colloids, including HES, actually prevented kidney injury in trauma patients [6], perhaps because of faster resuscitation. Curiously, the authors cite a metaanalysis without specifying that it was restricted to critically ill and septic patients [7], and fail to mention others showing no increase in mortality or kidney injury in a broader population [8]. It is thus apparent that there is considerable heterogeneity among trials comparing HES with crystalloid solutions and that the results are highly context-sensitive. Septic patients suffer from inflammation, endothelial injury and microcirculatory dysfunction. Surgical patients differ in experiencing acute blood loss, hypotension and insufficient organ perfusion. Surgical patients thus require rapid intravascular volume administration and benefit from swift resolution of hypotension and improved systemic perfusion. The administration of HES enhances haemodynamic stability and improves cardiac index more than comparable volumes of crystalloid [7]. Cardiac surgery represents an intermediate condition in which patients are relatively sick, but rarely septic. One meta-analysis reports that risks of mortality and renal injury attributable to colloids are limited to critically ill septic patients, and that cardiac surgical patients are spared [5]. However, none have evaluated kidney injury in the detail and duration we have, thus providing strong equipoise for our trial. In asking ‘what is the benefit?’ Giordano et al. miss an important feature of our trial. We did not compare HES with crystalloid, but with albumin. Our trial was restricted to patients in whom clinicians believed a colloid was necessary. Albumin is in short supply and far more expensive than starch solutions, so starch solutions would be a reasonable substitute if they were comparably effective and no more toxic. We did not evaluate efficacy, but it widely believed that albumin and starch solutions similarly replete vascular volume [7]. Our goal was to compare the renal safety of the two solutions. Specifically, we tested the hypothesis that HES does not cause more renal injury than albumin. Noninferiority is exactly the right statistical approach for this question. In fact, there was no evidence of HES-induced kidney injury in our trial, as quantified by urinary neutrophil gelatinase-associated lipocalin (uNGAL) and urinary interleukin-18 (IL-18) [9]. Even a full year after surgery, there was no evidence of harm. Peri-operative haemostasis was also comparable with each solution. Our results therefore completely support our equipoise assumption and noninferiority approach, that HES solutions cause nomore renal injury than albumin in cardiac surgical patients.