Use of hydroxyethyl starch in sepsis research: A systematic review with meta-analysis

Background: After being used for more than a decade the use of hydroxyethyl starch (HES) 130/0.38-0.45 in clinical practice was discouraged because of serious adverse events including bleeding, acute kidney injury and death. But could these adverse effects have been discovered sooner? By comparing the summary effect estimates of the adverse effects of HES 130/0.38-0.45 in different study designs we aimed to disclose any signal of this. Methods: We systematically searched MEDLINE, EMBASE and Cochrane Library and hand-searched the reference lists of relevant studies to identify studies for inclusion. Eligible trials were randomised clinical trials (RCTs) and observational studies in patients with sepsis and randomised trials in animals with induced sepsis comparing HES 130/0.38-0.45 to any type of crystalloid. Relevant outcomes were all-cause mortality at longest follow-up, renal replacement therapy (RRT), acute kidney injury (AKI) and bleeding. We extracted data, conducted conventional meta-analyses and assessed the risk of bias and the quality of evidence. Results: We included 8 RCTs including 3,273 patients, 1 observational study including 379 patients and 5 randomised animal trials including 94 test animals. There was no suggestion of interaction in subgroup analyses comparing the different study designs for any outcomes (all-cause mortality at longest follow-up p = .33; RRT p = .70; AKI p = .63; bleeding p = .20). Conclusions: We observed no interaction between the summary effect estimates of RCTs, observational studies in patients and randomised animal trials for any of the outcomes. Accordingly, we found no evidence indicating that the adverse effects of HES 130/0.38-0.45 could have been discovered sooner. Systematic review registration: PROSPERO ID: CRD42018103422. Protocol Johansen JR, Perner A, Brodtkorb JH, Moller MH. Use of hydroxyethyl starch in sepsis research: Protocol for a systematic review. Acta Anaesthesiol Scand. 2019;63:814-818.