Three Part Question
In [critically ill patients] does the use of [intensive glucose control] improve [survival]?
You start at a new hospital and find that there is a protocol in place to guide the treatment of patients with sepsis. It states that if the patient's blood glucose is found to be greater than 7.8 mmol/l then insulin should to be started to keep it tightly controlled . You wonder if some degree of hyperglycaemia is not such a bad thing, given that it is part of a natural physiological response to stress and glucose is required for brain metabolism. A quick review of the literature throws up contrasting results and so you decide that you should really examine the evidence in a more systematic manner.
Medline through OVID interface June 2010.
EMBASE through OVID interface June 2010.
The following search was applied to Medline and EMBASE:
[(tight adj5 glucose).mp. OR (intensive adj5 glucose).mp. OR(tight adj5 insulin).mp. OR (intensive adj5 insulin).mp. OR (strict adj5 glucose).mp. OR (strict adj5 insulin).mp. OR (hyperglyc$ adj5 stroke).mp. OR (critical adj5 glucose).mp. OR (critical adj5 insulin).mp. OR (critical adj5 glyc?emic).mp. OR (intensive adj5 glyc?emic).mp. OR (tight adj5 glyc?emic).mp. OR (strict adj5 glyc?emic).mp.] AND [exp Critical Care/ OR exp Intensive Care Units/ OR exp Intensive Care/ OR itu.mp. OR icu.mp. OR critical$ ill$.mp. OR sepsis.mp. OR exp Sepsis/ OR exp Shock, Traumatic/ OR exp Burns/OR severe trauma$.mp. OR exp Multiple Trauma/ OR exp Shock, Hemorrhagic/ OR stroke.mp. OR exp Stroke/ OR myocardial infarct$.mp. OR exp Myocardial Infarction/ OR cva.mp. OR cerebrovascular accident.mp.] LIMIT to English language AND humans AND "all adult (19 plus years)".
The Cochrane website was searched using the term, 'insulin'.
The NHS Evidence Health Information Resource website was also searched using the term 'insulin'.
Google Scholar was searched using 'intensive glucose control' and the first 50 results were checked.
The Medline search obtained 475 papers. One systematic review and meta-analysis was found incorporating 34 randomised controlled trials. 5 additional relevant papers were found, 4 published following the meta-analysis and 1 that was not included in the meta-analysis.
The EMBASE search obtained 386 papers including the same meta-analysis. 4 of the papers were relevant and not included in the meta-analyses. 2 of the papers were duplicates of those found with the Medline search.
The search of the Cochrane library found two protocols for reviews regarding strict glucose control for critically ill patients. These were both dated January 2009, there were no abstracts available and they had not been completed at this time.
The Database of Abstracts and Reviews of Effects (DARE) that was searched as part of the NHS Evidence site revealed another meta-analysis.
The first 50 hits from Google Scholar were examined but did not reveal any original papers.
|Author, date and country
||Study type (level of evidence)
|Cheung, N; Wong, V; McLean, M|
|240 patients presenting with AMI and a blood glucose level of =/> 7.8mmol/l.||PRCT. Patients were randomised to receive a dextrose/insulin infusion to maintain glucose <10mmol/l or conventional treatment which consisted of normal diabetic control if known diabetic or s/c insulin in patients with a blood glucose =/> 16.0 mmol/l.||Inpatient mortality||4.8% in the intensive group vs. 3.5% in the conventional group (p=0.75)||The authors had estimated that a sample size of 850 patients was necessary to obtain a significant result but stopped the study early due to lower than expected recruitment numbers and a failure to demonstrate a benefit on the interim analysis. The target levels of glucose are much higher than in the other papers included and this is probably why it was not included in either of the meta-analyses.|
|6 month mortality||7.9% in the intensive group vs. 6.1% in the conventional group (p=0.62)|
|Episodes of hypoglycaemia||13 episodes in the intensive group vs. 2 episodes in the conventional group (p=0.02)|
|Wiener, R; Wiener, D; Larson, R|
|29 randomized controlled trials with a total of 8432 critically ill patients. The inclusion criteria for the trials were: adult ITU setting; intervention group received tight glucose control; comparison group received usual care and one of the outcomes included hospital or short-term mortality.||Meta-analysis of randomized controlled trials comparing 'tight' versus conventional glucose control in critically ill, adult patients. ||Overall mortality||21.6% in the intensive control group vs. 23.3% in the conventional treatment group. (RR 0.93 95% CI 0.85-1.03)|
|Risk of hypoglycaemia||13.7% in the intensive group vs. 2.5% in the standard care group (RR 5.13, 95% CI 4.09-6.43).|
|Griesdale, D; de Souza, R; van Dam, R; Heyland, D; Cook, D; Malhotra, A; Dhaliwal, R; et al|
|26 papers which involved a total of 13, 567 patients admitted to critical care settings.||Literature search carried out seeking randomised controlled trials comparing standard versus intensive glucose control for patients in intensive care settings. The intensive arm of the trials had to involve a target glucose of 8.3mmol/l or less and the study had to document mortality.||Mortality||1681/6812 (24.7%) patients in the intensive group vs. 1681/6760 (24.9%) in the standard treatment group. (RR 0.93 95% CI 0.83-1.04)|
|Hypoglycaemia (reported in 14 trials)||RR 6.0 (95% CI 4.5-8.0) for patients in the intensive group vs. conventional group.|
|Bilotta, F; Caramia, R; Paolini, F; Delfini, R; Rosa, G|
|483 patients admitted to an ITU setting following either emergency or elective brain surgery.||Patients prospectively randomised to receive insulin titration to maintain blood glucose of 4.44-6.11mmol/l or conventional treatment where insulin was used if the patients blood glucose was greater than 11.94mmol/l.||Mortality||63/241 (26%) of patients in the intensive group died vs. 68/242 (28%) in the conventional group (p=0.8).||The patients were a mixture of elective and emergency patients. The effects of randomisation meant that 138/242 patients in the conventional group were emergency vs. 120/241 in the intensive treatment group. Data regarding patients from the same centre over the same time period has already been published as two separate papers with 78 and 97 patients in each. These two papers have been incorporated in both of the meta-analyses.|
|Hypoglycaemia||226/241 (94%) in the intensive group vs. 152/242 (63%) in the conventional group. (p<0.0001).|
|Meng, Y; Qingje, G; Xiangtong, Z; Shugang, S; Yaohua, W; et al|
|240 patients with severe traumatic brain injury admitted to one neurological intensive care unit.||Patients were randomised to receive intensive treatment with a glucose kept between 4.4-6.1mmol/l or conventional treatment where insulin was given to patients with a blood glucose greater than 11.1mmol/l.||6 month mortality.||61/117 (52.1%) in the intensive group vs. 62/116 in the conventional treatment group (p=0.8).||Small numbers. Select group of patients so not necessarily possible to generalise results.|
|Taslimi, R; Azizkhani, R; Talebian, M; Abtahi, H; Jalili, M; Nejati, A |
|Patients admitted onto the general ICU in one hospital with a blood glucose >200mg/dl and an APACHE score or ≥10.||Patients were randomised to intensive glucose control (80-110mg/dl) (4.4-6.1mmol/l) or conventional treatment where insulin was commenced if blood glucose levels were >125mg/dl(6.9mmol/l).||Mortality on ICU||19/62 (30.6) of the conventional treatment group vs. 26/67 (38.8%) of the intensive group (NS).||Small numbers. 51/62 patients on conventional treatment admitted for surgical rather than medical reasons, 46/67 patients on the intensive treatment were for surgical reasons.|
|Preiser, JC; Devos, P; Ruiz-Santana, S; Melot, C; Annane, D; Groeneveld, G; et al|
|Adult patients requiring admission to 21 participating ICU departments.||Patients were randomised to have a blood glucose target of 7.8-10.0mmol/l or 4.4-6.1 mmol/l.||Mortality.||83/542 (15.3%) in the first group vs. 92/536 (17.2%) in the second group (NS).||The intention was to recruit 3,500 patients but following interim analysis the study was stopped early due to protocol violation. The intended glucose targets were not being achieved despite the insulin infusions.|
|Rate of hypoglycaemia||2.7% in the first group vs. 8.7% in the second group (p<0.0001).|
|COIITSS Study Investigators|
|509 patients requiring intensive care treatment due to septic shock admitted to 11 participating centres.||Patients were randomised to receive hydrocortisone with intensive insulin therapy or hydrocortisone with conventional insulin therapy. Each patient was also randomised to receive fludricortisone or not forming four groups in total.||In-patient mortality||117/255 (45.9%) of patients in the intensive treatment group vs. 109/254 (42.9%) in the conventional group (RR 1.07, 95% CI 0.88-1.30; p=0.5).||Relatively small number of patients given the number of variables that may have been involved in the outcomes.|
|Severe hypoglycaemic episodes||42/255 in the intensive group vs. 20/254 in the conventional treatment group (p=0.03).|
|Coester A, Neumann C, Schmidt M|
|88 adult patients admitted to one ITU with a GCS ≤ 8 following blunt head injury. Patients were excluded from the study if they died within 48h or if had shock lasting ≥12h.||Patients were randomised so that in the event of becoming hyperglycaemic they would be commenced on Intensive Insulin Treatment ( [glucose] maintained between 4.4-6.1mmol/l) or Conventional Treatment ([glucose] < 10mmol/l).||Mortality||11/39(28.2%) patients in Intensive group vs. 11/40 (27.5%) in the conventional group. (p=1)||Figures used for power calculation not justified. Small numbers, especially as not all patients in each group required any insulin. Unclear why exclusion criteria were chosen.|
|Hypoglycaemia||32/39 (82.1%) of patients in the Intensive group vs. 7/40 (17.5%) in the Conventional Group. (p=0.0001).|
There has been recognition for some time that the presence of hyperglycaemia in critically ill patients is associated with a worse outcome. This may be because the degree of hyperglycaemia reflects the stress response of the patient to the physiological insult of their illness or injury, or it may be that it is due to selection of diabetic patients who are more likely to have a poor outcome due to their underlying illness and its associated comorbidities. Following this recognition, studies have been undertaken to attempt to control the level of hyperglycaemia to see if this can modify outcome. The theory of benefit goes beyond merely ‘treating numbers’, as treatment with insulin is known to suppress free fatty acid oxidation, reduce thromboxane A production and decrease plasma plasminogen activator inhibitor-I production. Two landmark studies provided evidence of a marked benefit in survival outcome for ill patients with tight glucose control. Malberg et al looked at patients with acute myocardial infarction in the DIGAMI trial. They found a significant reduction in the level of mortality (29% reduction, p=0.027). Van den Berghe et al looked at patients in a surgical intensive care unit (ICU) setting, and also demonstrated a remarkable improvement in outcome in terms of survival (34% mortality reduction, p<0.04). The authors also found reduced rates of septicaemia, renal impairment requiring dialysis, transfusion requirements and length of stay in the ICU. The results appeared so significant that there was some extrapolation of the results, and tight glucose control was recommended for different groups of critically ill patients largely on the basis of this single study. Unfortunately, attempts to replicate the survival benefit of the original study have been unsuccessful and even achieving the recommended degree of control of hyperglycaemia has proven difficult. All of the studies have shown that attempts at strict control of glucose are associated with a higher level of episodes of hypoglycaemia and some studies have suggested a worse outcome for the patients in the treatment arms. The NICE-SUGAR Study was an international, multi-centre study involving 6104 patients at 42 centres. Patients admitted to ICUs were randomised to tight control (4.5–6 mmol/l) or conventional treatment (<10 mmol/l). Mortality rates were 27.5% in the tight group vs 24.9% in the conventional group (p=0.02). Two meta-analyses have been published with slightly different inclusion criteria but a large overlap in included papers. The Griesdale meta-analysis included the NICE-SUGAR study, the Weiner meta-analysis was published before this. Neither of the meta-analyses demonstrate any mortality benefit from intensive insulin therapy in this group of patients. The Griesdale meta-analysis suggested that there may be some benefit to surgical ICU patients, although this was largely due to the Van den Berghe results. Subgroup analysis of other studies including the NICE-SUGAR trial did not demonstrate improved survival in surgical ICU patients receiving intensive insulin control.
AMI, acute myocardial infarction; APACHE, acute physiology and chronic health evaluation; COIITSS, corticosteroid treatment and intensive insulin therapy for septic shock in adults; ICU, intensive care unit; ITU, intensive therapy unit; PRCT, prospective randomised controlled trial.
Clinical Bottom Line
There is no evidence of benefit, in terms of mortality, in attempting tight glucose control in the critically ill patient.
Level of Evidence
Level 1 - Recent well-done systematic review was considered or a study of high quality is available.
- Cheung, N; Wong, V; McLean, M The Hyperglycemia: Intensive Insulin Infusion in Infarction (HI-5) Study Diabetes Care April 2006; 29:4 p765-70
- Wiener, R; Wiener, D; Larson, R Benefits and Risks of Tight Glucose Control in Critically Ill Adults: A Meta-analysis JAMA 2008; 300: 933-44
- Griesdale, D; de Souza, R; van Dam, R; Heyland, D; Cook, D; Malhotra, A; Dhaliwal, R; et al Intensive insulin therapy and mortality among critically ill patients: a meta-analysis including NICE-SUGAR study data CMAJ 2009; 180(8):821-7
- Bilotta, F; Caramia, R; Paolini, F; Delfini, R; Rosa, G Safety and Efficacy of Intensive Insulin Therapy in Critical Neurosurgical Patients Anesthesiology 2009; 110:611-9
- Meng, Y; Qingje, G; Xiangtong, Z; Shugang, S; Yaohua, W; et al Intensive insulin therapy on infection rate, days in NICU, in-hospital mortality and neurological outcome in severe traumatic brain injury patients: A randomized controlled trial International Journal of Nursing Studies 2009; 46:753-58
- Taslimi, R; Azizkhani, R; Talebian, M; Abtahi, H; Jalili, M; Nejati, A The efficacy of intensive glucose management on hospitalized critically ill patients associated mortality rate in intensive care DARU 2009; 17(3):157-162
- Preiser, JC; Devos, P; Ruiz-Santana, S; Melot, C; Annane, D; Groeneveld, G; et al A prospective randomised multi-centre controlled trial on tight glucose control by intensive insulin therapy in adult intensive care units: the Glucontrol study Intensive Care Med 2009; 35:1738-48
- COITTSS Study Investigators Corticosteroid Treatment and Intensive Insulin Therapy for Septic Shock in Adults: A Randomized Controlled Trial JAMA 2010;303(4):341-48
- Krinsley JS. Association between hyperglycemia and increased hospital mortality in a heterogeneous population of critically ill patients Mayo Clin Proc 2003;78(12):1471-8
- Malmberg K, Ryden L, Efendic S, et al Randomized Trial of Insulin-Glucose Infusion Followed by Subcutaneous Insulin Treatment in Diabetic Patients With Acute Myocardial Infaction (DIGAMI Study): Effects on Mortality at 1 Year JACC 1995; 26(1):57-65
- Van den Berghe G, Wouters P, Weekers F, et al Intensive Insulin Therapy in Critically Ill Patients N Eng J Med 2001; 345(19):1359-67
- Dellinger RP, Carlet JM, Masur H, Gerlach H, et al Surviving Sepsis Campaign guidelines for management of severe sepsis and septic shock. Intensive Care Med 2004;30(4):536-55
- NICE-SUGAR Study Investigators Intensive versus Conventional Glucose Control in Critically Ill Patients N Eng J Med 2009; 360(13):1283-97
- Coester A, Neumann C, Schmidt M Intensive Insulin Therapy in Severe Traumatic Brain Injury: A Randomized Trial The Journal of Trauma, Injury, Infection, and Critical Care 2010;68: 904-11