Hydroxyethyl starch in patients with trauma.

Editor—The Fluids in Resuscitation of Severe Trauma (FIRST) randomized trial conducted by James and colleagues compared hydroxyethyl starch (HES) 130/0.4 with 0.9% saline in the resuscitation of 115 patients with severe blunt or penetrating trauma at a single centre in South Africa. In blunt trauma, which accounts for 86–97% of all traumatic injuries in Europe, 3 no differences were found between HES 130/ 0.4 and 0.9% saline in either of the two primary study endpoints: fluid volume needed during the first 24 h and number of patients achieving normal gastrointestinal function by day 5. In penetrating trauma, HES 130/0.4 reduced fluid requirement significantly but less than generally expected and showed no effect on normalization of gastrointestinal function. In the published report of the trial, the FIRST investigators emphasized improvement in renal function (P1⁄40.018) and lactate clearance (P,0.05) among patients with penetrating trauma receiving HES 130/0.4. Conversely, they downplayed an observed increase in transfusion of blood products among HES130/0.4 recipients with blunt trauma. For that type of trauma, mean transfusion of packed red blood cells was twice as great in the HES 130/ 0.4 as the 0.9% saline group (P1⁄40.005), fresh-frozen plasma (FFP) three times as great (P1⁄40.005), and platelets five times as great (P1⁄40.005). Two primary endpoints and seven secondary endpoints were pre-specified (www. controlled-trials.com/ISRCTN42061860/42061860). However, there is considerable discrepancy between the predefined study endpoints and those published in the report. Three safety endpoints were defined in the publication, which were not included in the register entry. Among the non-pre-specified study and safety endpoints are acute kidney injury (AKI) and lactate clearance, which are highlighted as the main results in the title of the paper. However, the post hoc analyses of the data for those endpoints presented in the publication should be considered exploratory and not conclusive. Furthermore, the FIRST trial was not powered to assess differences in AKI. The baseline data that were presented make it clear that the study groups in the FIRST trial were not well matched. Among patients with blunt trauma, both Injury Severity Score and New Injury Severity Score (NISS) were significantly higher in the group allocated to HES 130/0.4 (P,0.01 for each comparison). The small P-values indicate that the imbalances were unlikely to have arisen by chance, suggesting the possibility of flawed randomization. Although not statistically significant, substantial imbalances were also present in other baseline characteristics. Thus, the penetrating trauma group allocated to HES 130/0.4 (P-HES), that is, the group with apparently improved renal function and lactate clearance, was .5 yr younger on average than any other group and weighed .5 kg less. Furthermore, the mean baseline plasma lactate level was 40% higher in the group with blunt trauma allocated to HES 130/0.4 than 0.9% saline, but 20% lower with penetrating trauma. The pervasive baseline imbalances in this trial raise doubt as to whether observed between-group differences were due to study fluid or confounding variables. Interpretation of the trial results is further complicated by the absence of key data such as site of injury, pre-study crystalloid volume, baseline SOFA score, and the time to resuscitation targets, that is, mean arterial pressure, heart rate, central venous pressure and oxygen saturation, serum creatinine, and the mortality by group. No analysis was reported on the influence, if any, of baseline imbalance in blunt injury severity on blood product transfusion. The finding of only 31% correlation between baseline NISS and blood product usage suggests that the higher transfusion requirement of HES 130/0.4 recipients with blunt trauma may have persisted after adjustment for injury severity. Appropriate statistical analysis should have been performed to resolve this important question. The claim that HES 130/0.4 resulted in more effective resuscitation by increased lactate clearance remains highly questionable for several reasons. First, as shown in Figure 3B of the publication, the mean plasma lactate level was higher at baseline in the P-SAL than in the P-HES group, and mean levels in the two groups closely paralleled each other over the first 3 h of resuscitation. Those observations are not consistent with a significant between-group difference in plasma lactate. However, such a difference is reported in Figure 3C based upon a mixed-effect linear regression analysis. No details are provided about that analysis, which assumes that the data are normally distributed. However, since the coefficients of variation shown in Figure 3B exceed 50%, the normality assumption is clearly violated. No assurance is provided in the FIRST publication that the mixed-effect linear regression analysis is valid. Secondly, plasma group difference with respect to epinephrine use, which can cause hyperlactaemia, might also account for the perceived differences in lactate clearance. The maximum SOFA score was significantly higher in the P-HES group than the corresponding control group (P-SAL), but this observation would only be meaningful if baseline scores were closely matched. With no baseline SOFA data, the fidelity of matching with respect to that score remains indeterminate. AKI was scored in accordance with the RIFLE criteria which are based on serum creatinine or urine output. Differences in urine output were not found in the FIRST trial, so the AKI results were evidently derived from serum creatinine measurements. However, serum creatinine data were not reported, and so the basis for a difference in AKI is difficult to assess. Thrombelastography (TEG) data were assessed, but data were not provided; British Journal of Anaesthesia 108 (2): 321–330 (2012)