556318 research-article2015
PENXXX10.1177/0148607114556318Journal of Parenteral and Enteral NutritionKoretz
Journal Club Journal of Parenteral and Enteral Nutrition Volume 39 Number 3 March 2015 369–371 © 2015 American Society for Parenteral and Enteral Nutrition DOI: 10.1177/0148607114556318 jpen.sagepub.com hosted at online.sagepub.com
JPEN Journal Club 11. Early Stopping Ronald L. Koretz, MD1,2
The article to be considered in this installment of the JPEN Journal Club is the classic trial promoting intensive insulin therapy in critically ill patients that was published in 2001.1 The Belgian investigators noted an association between hyperglycemia and a poorer clinical outcome and tested the hypothesis that this association was causative. They randomized patients who were admitted to their intensive care unit (ICU), one that largely cared for surgical patients, into 1 of 2 arms. The experimental arm was provided with intensive insulin infusion therapy designed to keep the blood glucose between 80 mg% and 110 mg%; the control subjects received insulin infusions only if the blood glucose exceeded 215 mg%, and the goal of therapy was to keep the blood glucose between 180 mg% and 210 mg%. The primary outcome was ICU mortality. There were 13 secondary outcomes including hospital mortality, length of stay in the ICU, need for a variety of medical support systems, septicemia, organ dysfunctions, transfusion requirements, fever, and laboratory tests. A sample size calculation that was based on seeing a difference in mortality in the subgroup that remained in the ICU for >5 days of 5% and an overall difference in mortality of 2% indicated that 2,500 patients would need to be randomized. Interim analyses were planned every 3 months with a plan to stop the trial if the P value for the difference in survival was 5 days and at the time of discharge from the hospital. These patients had fewer episodes of septicemia, a shorter ICU length of stay, less need for ventilator support or renal replacement therapy, less hyperbilirubinemia, and less abnormal markers of inflammation. The investigators concluded that intensive insulin therapy improved mortality and morbidity in critically ill patients housed in a surgical ICU. The precise method by which the patients were assigned to groups was not explained, but a “balanced use of permuted blocks” suggests that some type of formal randomization mechanism was employed. While the assignments were placed in sealed envelopes, we cannot be sure that the investigators
were completely unaware of the treatment arm into which the patient was going to be assigned (concealment of allocation) because they could have opened the next envelope (or at least held it up to the light); furthermore, since the trial was not blinded, the allocation of at least every 10th patient would be known because of the balanced permutation blocks.2 The trial was not blinded to the assessors of the primary and a number of secondary outcomes, although mortality is an objective outcome. The randomization scheme appeared to have been successful, as there were no baseline differences between the 2 groups. However, there were some other methodologic issues that we should consider. It was curious to me that adverse events, hypoglycemia and hyperglycemia in particular, were not listed as secondary outcomes. After all, the interventions being assessed were different regimens of insulin. The investigators did report, early in the Results section of the article, that 39 (5%) and 6 (1%) patients in the intensive and standard insulin groups, respectively, developed hypoglycemia. While no statistical analysis was reported, when I performed a χ2 calculation, the P value was
PENXXX10.1177/0148607114556318Journal of Parenteral and Enteral NutritionKoretz
Journal Club Journal of Parenteral and Enteral Nutrition Volume 39 Number 3 March 2015 369–371 © 2015 American Society for Parenteral and Enteral Nutrition DOI: 10.1177/0148607114556318 jpen.sagepub.com hosted at online.sagepub.com
JPEN Journal Club 11. Early Stopping Ronald L. Koretz, MD1,2
The article to be considered in this installment of the JPEN Journal Club is the classic trial promoting intensive insulin therapy in critically ill patients that was published in 2001.1 The Belgian investigators noted an association between hyperglycemia and a poorer clinical outcome and tested the hypothesis that this association was causative. They randomized patients who were admitted to their intensive care unit (ICU), one that largely cared for surgical patients, into 1 of 2 arms. The experimental arm was provided with intensive insulin infusion therapy designed to keep the blood glucose between 80 mg% and 110 mg%; the control subjects received insulin infusions only if the blood glucose exceeded 215 mg%, and the goal of therapy was to keep the blood glucose between 180 mg% and 210 mg%. The primary outcome was ICU mortality. There were 13 secondary outcomes including hospital mortality, length of stay in the ICU, need for a variety of medical support systems, septicemia, organ dysfunctions, transfusion requirements, fever, and laboratory tests. A sample size calculation that was based on seeing a difference in mortality in the subgroup that remained in the ICU for >5 days of 5% and an overall difference in mortality of 2% indicated that 2,500 patients would need to be randomized. Interim analyses were planned every 3 months with a plan to stop the trial if the P value for the difference in survival was 5 days and at the time of discharge from the hospital. These patients had fewer episodes of septicemia, a shorter ICU length of stay, less need for ventilator support or renal replacement therapy, less hyperbilirubinemia, and less abnormal markers of inflammation. The investigators concluded that intensive insulin therapy improved mortality and morbidity in critically ill patients housed in a surgical ICU. The precise method by which the patients were assigned to groups was not explained, but a “balanced use of permuted blocks” suggests that some type of formal randomization mechanism was employed. While the assignments were placed in sealed envelopes, we cannot be sure that the investigators
were completely unaware of the treatment arm into which the patient was going to be assigned (concealment of allocation) because they could have opened the next envelope (or at least held it up to the light); furthermore, since the trial was not blinded, the allocation of at least every 10th patient would be known because of the balanced permutation blocks.2 The trial was not blinded to the assessors of the primary and a number of secondary outcomes, although mortality is an objective outcome. The randomization scheme appeared to have been successful, as there were no baseline differences between the 2 groups. However, there were some other methodologic issues that we should consider. It was curious to me that adverse events, hypoglycemia and hyperglycemia in particular, were not listed as secondary outcomes. After all, the interventions being assessed were different regimens of insulin. The investigators did report, early in the Results section of the article, that 39 (5%) and 6 (1%) patients in the intensive and standard insulin groups, respectively, developed hypoglycemia. While no statistical analysis was reported, when I performed a χ2 calculation, the P value was
