To the Editor The recent review by Van Der Linden et al.1 concludes “that data in the peer reviewed literature do not suggest an adverse safety signal” when tetrastarch is used perioperatively. The review as acknowledged by the authors was commissioned by a tetrastarch manufacturer. However, the unfortunate timing of publication immediately after publication of the Crystalloid Versus Hydroxyethyl Starch Trial2 may lead readers to underestimate safety concerns related to use of hydroxyethyl starches (HESs) in surgical patients. Major trials purely in the surgical setting are still lacking.3–5 The Crystalloid Versus Hydroxyethyl Starch Trial included 7000 intensive care unit patients,2 2829 of them after surgery. It showed that tetrastarch—like pentastarch—is not beneficial but harmful. Patients receiving tetrastarch received significantly more renal replacement therapy (RRT) (7.0% vs 5.8%), blood products (78 ± 250 mL vs 60 ± 190 mL) and had more pruritus (4.0% vs 2.2%). Earlier this year, another high-quality randomized control trial showed that tetrastarch significantly increased the need for RRT and 90-day mortality in sepsis.6 Indeed, all prospective studies assessing 90-day survival status after tetra- or pentastarch found significantly or slightly increased mortality rates.2,6–10 Studies in the surgical setting were not designed to assess patient-relevant safety outcomes. Most studies use other starches or synthetic colloids with similar risks11 as comparator and observation periods of <24 hours5 are too short to evaluate safety. Given these limitations, cumulative data analysis of survival, renal outcomes, or blood loss is meaningless. The Van Der Linden et al.1 review lacks important elements of systematic reviews, e.g., the assessment of risk of bias in identified studies, additional information requested from study authors, a forest plot with results from individual studies or a funnel plot to assess risk of bias across studies, which would have helped authors and readers critically appraise the cumulative data.12 The authors perform some meta-analyses which find that tetrastarch is beneficial in surgical patients. The odds ratio of mortality at any time is 0.51 (0.24-1.05) with a favorable trend for HES from 21 studies with 1918 patients and 11 deaths after tetrastarch.1 Furthermore, we find the use of a fixed effects model problematic in this context since it can result in inaccurate confidence intervals. More importantly, however, the authors omitted 18 deaths (12/56 patients in the HES 130/0.4 group vs 6/53 in the saline group) from the Fluids in Resuscitation of Severe Trauma study with trauma patients.9 These figures were only published in reply to a letter,13 but should be known to the authors since the first author of the trauma study is a coauthor of the present paper. Recently, the first study to assess 90-day mortality in the surgical setting found 5 of 24 patients died after tetrastarch compared with 0 of 24 patients in the crystalloid group (P = 0.051).10 The authors report a reduced risk for allogeneic blood transfusion after tetrastarch compared with comparator fluids (odds ratio, 0.73 [0.61–0.87]; P = 0.0004).1 Again, a forest plot with study references should have been provided. More importantly, most studies use unsuitable comparators, and the 2 largest studies used a hemoglobin-based oxygen carrier, which is not approved for clinical use on account of increased risk of death and myocardial infarction.14 After hemoglobin-based oxygen carrier, 10.5% more patients required allogeneic blood transfusions which may have shifted the signal considerably in favor of tetrastarch.15 The findings of the Mittermayr et al. study16 should be considered: more patients receiving tetrastarch than patients receiving Ringer’s lactate during orthopedic surgery required allogeneic blood products (3/19 vs 1/20), fibrinogen (6/19 vs 0/20), and more patients received red blood cell units (9 vs 2). Schramko et al.17 reported similar impairment in thromboelastometry after HES 130 and HES 200 compared with human albumin. A systematic review found that HES 130/0.4 administration results in a weaker and smaller clot in comparison with crystalloid or albumin control fluids.18 Finally, the authors found no signal for an increased risk for RRT based on 7 studies with 790 patients and various control fluids (mostly colloids). The number of patients appears small; e.g., to detect an increase in RRT in cardiac surgical patients, approximately 1000 patients need to be randomized into 2 groups with relevant control fluids.5 For the sake of further transparency, the affiliation of Dr. Edward Burdett (responsible for the literature search) and the role of the company representatives acknowledged for noninterference should be provided. Low-quality reviews overwhelm the results of well-performed meta-analyses on HESs;19 and the limitations described above do little to correct this imbalance in the literature. Jukka Takala, MD, PhD Department of Intensive Care Medicine University Hospital Bern (Inselspital) Bern, Switzerland Christiane Hartog, MD Konrad Reinhart, MD Department of Anesthesiology and Intensive Care Jena University Hospital Jena, Germany [email protected]