Intensive insulin therapy and starch (HES 200/0.5) had some risk and no clear benefit in severe sepsis.

Abstract

Questions In critically ill patients with severe sepsis or septic shock, how does intensive insulin therapy (IIT) compare with conventional therapy for mortality? How does fluid resuscitation with hydroxyethyl starch (HES) compare with Ringer's lactate (RL)? Methods Design Randomized, 2 2 factorial design, controlled trial (Efficacy of Volume Substitution and Insulin Therapy in Severe Sepsis [VISEP] study). Allocation {Concealed}.* Blinding {Unblinded}.* Follow-up period 28 days. Setting 18 intensive care units (ICUs) in Germany. Patients 537 patients (mean age 65 y, 60% men) with severe sepsis or septic shock developing < 24 hours before ICU admission or < 12 hours after admission if condition developed in the ICU. Exclusion criteria included receipt of HES >1000 mL in the 24 hours before randomization. Intervention IIT (n =247) or conventional therapy (n =290) and fluid resuscitation with HES (n =262) or RL (n =275). IIT consisted of insulin infusion when plasma glucose levels were >110 mg/dL, then adjusted to maintain euglycemia (80 to 110 mg/dL). Conventional therapy consisted of continuous insulin infusion when glucose levels were >200 mg/dL, then adjusted to maintain levels at 180 to 200 mg/dL. Outcomes All-cause mortality and Sequential Organ Failure Assessment (SOFA) score. Secondary outcomes included length of ICU stay, acute renal failure, renal replacement therapy (RRT), hypoglycemia (40 mg/dL), and development of 1 adverse event and 1 serious adverse event. Patient follow-up 89% (intention-to-treat analysis). Main results The trial was stopped early by the safety committee. More patients in the IIT group had hypoglycemia, 1 adverse event, and 1 serious adverse event than the conventional therapy group (Table), but groups did not differ for all-cause mortality (Table), SOFA scores, length of ICU stay, acute renal failure, or RRT. Cumulative median dose of HES was 70.4 mL/kg; 38% of patients in the HES group exceeded the maximum dose of 20 mL/kg/d. HES led to more acute renal failure and RRT than RL (Table); groups did not differ for mortality (Table), SOFA scores, length of ICU stay, hypoglycemia, or adverse events. Conclusions Intensive insulin therapy increased hypoglycemia and adverse events compared with conventional therapy. Fluid resuscitation with hydroxyethyl starch increased acute renal failure and renal replacement therapy compared with Ringer's lactate. Intensive insulin therapy (IIT) vs conventional insulin therapy (CIT) and fluid resuscitation with hydroxyethyl starch (HES) vs Ringer's lactate (RL) in severe sepsis or septic shock Outcomes at 28 d IIT CIT RRR (95% CI) NNT All-cause mortality 25% 26% 4.8% (27 to 29) Not significant RRI (CI) NNH (CI) Hypoglycemia (40 mg/dL) 17% 4.1% 311% (124 to 658) 8 (6 to 13) 1 adverse event 22% 8.6% 158% (67 to 301) 8 (6 to 14) 1 serious adverse event 11% 5.2% 111% (16 to 285) 18 (10 to 83) HES RL All-cause mortality 27% 24% 11% (17 to 48) Not significant Acute renal failure 35% 23% 53% (17 to 102) 9 (6 to 23) Renal replacement therapy 31% 19% 66% (22 to 125) 9 (6 to 21) Abbreviations defined in Glossary. RRR, RRI, NNT, NNH, and CI calculated from data in article. Commentary The multicenter, randomized, factorial trial by Brunkhorst and colleagues evaluated 2 interventions for severe sepsis that did not influence mortality. Resuscitation with HES compared with RL increased risk for acute renal failure (number needed to harm [NNH] 9) and use of RRT (NNH 9), consistent with a meta-analysis of 22 trials (n =1866) showing that synthetic starch solutions in critical illness increase use of RRT (odds ratio 1.9, 95% CI 1.2 to 3.0) compared with other fluids (1). Although RRT thresholds vary considerably and may be subject to bias in unblinded trials, this raises concerns about frequent or prolonged use of HES, particularly in patients with antecedent renal insufficiency. The trial also confirmed that IIT targeting euglycemia causes more hypoglycemia than more liberal glucose management (NNH 8). The laboratory outcome was the impetus to stop the trial in the absence of a mortality difference. Whether the downside of transient hypoglycemia (which led other IIT trials to stop early) is offset by the elusive survival benefit observed in the initial (2) but not subsequent trials remains uncertain. The ongoing Normoglycemia in Intensive Care Evaluation and Survival Using Glucose Algorithm Regulation trial with a large a priori severe sepsis subgroup (projected n =1200) will inform this issue (3).