REVIEW Beyond diabetes: saving lives with insulin in the ICU

Size: px

Start display at page:

Download "REVIEW Beyond diabetes: saving lives with insulin in the ICU"

Rafe Farmer
5 years ago
Views:

1 (2002) 26, Suppl 3, S3 S8 ß 2002 Nature Publishing Group All rights reserved /02 $ REVIEW Beyond diabetes: saving lives with insulin in the ICU 1 * 1 Department of Intensive Care Medicine, University Hospital Gasthuisberg, University of Leuven, Leuven, Belgium The risk of mortality or significant morbidity is high among long-stay intensive care unit (ICU) patients. Sepsis, polyneuropathy and multiple organ failure are prominent causes of mortality and morbidity in the ICU. Many ICU patients are hyperglycaemic, presumably reflecting an adaptive development of insulin resistance. We hypothesized that this hyperglycaemia predisposes patients to many of the typical ICU complications, prolonged intensive care dependence and excess mortality. Insulin therapy directed at establishing normoglycaemia was investigated in a series of 1548 ICU patients. An intensive treatment group received insulin infusion tailored to control blood glucose levels in the range mmol=l (80 110mg=dl), whereas the conventional treatment group only received insulin when glucose levels exceeded 11.1 mmol=l (200 mg=dl) and in that event were maintained in a target range of mmol=l ( mg=dl). Intensive management of blood glucose levels was reflected in a 43% reduction in intensive care mortality risk (P ¼ after correction for interim analyses) and a 34% reduction in hospital mortality (P ¼ 0.01). A reduced risk of infection was reflected in a 46% reduction in the risk of septicaemia (P ¼ 0.003) and a 35% reduction in the need for prolonged ( > 10 d) antibiotic therapy (P < 0.001). Regression analysis suggests that control of glucose levels, rather than insulin administration itself, was responsible for the clinical benefits observed. Use of insulin infusion to control glucose levels in ICU patients, at least in populations similar to those in our study, can be expected to achieve clinically welcome improvements in outcome. An algorithm is proposed for implementing this. Further data are needed to establish the applicability of this strategy to other patient groups in the ICU and in general hospital care. (2002) 26, Suppl 3, S3 S8. doi: =sj.ijo Keywords: intensive care unit; clinical outcome; hyperglycaemia; intensive treatment; mortality risk; algorithm; sepsis Introduction There is a pressing need to improve outcomes for long-stay patients in intensive care units (ICUs). Approximately 30% of patients in surgical ICUs require more than 5 days intensive care and, despite our best efforts, the risk of mortality or significant morbidity among these long-stay patients is at least 20%. 1 Outcomes would be expected to be even worse among ICU patients with diabetes. Of course, many ICU patients have grave underlying conditions or have undergone severe trauma or major surgery, evidenced by their very presence in the unit. Nevertheless, the nature of the complications frequently occurring in these patients suggests that there is scope for improvement. There is, for example, much interest in therapies aimed at reducing the risk of sepsis and, in particular, septic shock in the ICU patient. 2 Septic shock is the most common cause of death in the ICU, and accounts for more than deaths per year in the USA alone. 3 Sepsis is closely associated with a series of inflammatory and metabolic responses to infection that may contribute to the multiple organ failure also common in long-stay ICU patients. 2 Polyneuropathy and skeletal muscle wasting are common, and prolong the need for mechanical ventilation. 4 6 The causes of critical-illness polyneuropathy are unclear but its consequences are severe. As in many areas of medicine, management of critical illness has had its share of rising hopes and disappointments as magic bullets fail to reach their targets. 1,7 Those therapies that have shown promise can be costly and raise fresh ethical questions concerning fair distribution of resources. 8 *Correspondence:, Department of Intensive Care Medicine, University Hospital Gasthuisberg, University of Leuven, B-3000 Leuven, Belgium. greta.vandenberghe@med.kuleuven.ac.be Glucose and insulin as targets for intervention in the ICU It is well established that hyperglycaemia is common in critically-ill patients, even if there is no history of diabetes. 9 This reflects the development of insulin resistance, at receptor and postreceptor level, in liver and muscle. This may be

2 S4 Insulin use during intensive care an adaptive response to life-threatening illness, ensuring adequate supply of glucose to the brain, to erythrocytes and to injured tissues. However, the benefits of this response may be outweighed by detrimental effects in the longer term. There is evidence linking hyperglycaemia with worse outcome after stroke and brain injury and with larger infarct size after myocardial infarction (MI). 10,11 Prevention of excessive hyperglycaemia [maintaining a level below 12 mmol=l (215 mg=dl)] in the DIGAMI trial was associated with a significant improvement in survival of diabetes patients with acute MI. 12 As well as the presumably macrovascular effects of excessive hyperglycaemia, there is the possible effect at any wound site: even mildly elevated glucose levels have been found to be associated with increased morbidity and mortality after burns or surgery. Although there is no clear evidence for a causal relationship, effects on a range of systems including mitochondrial metabolic pathways, endothelial function and cardiac potassium channels may be involved A role of hyperglycaemia in promoting bacterial infections of wound tissues and the blood is also intuitive. There are suggestions, however, that some of the beneficial effects of interventions that achieve glycaemic control via insulin in critically ill patients may reflect independent actions of insulin, rather than normalization of hyperglycaemia per se. 13 Glucose-insulin-potassium (GIK) infusion has been shown to salvage myocardium, increase heart function and improve mortality in patients with MI, perhaps independently of circulating glucose concentrations. 16,17 The Leuven Study: intensive insulin therapy in critically-ill patients We investigated the value of insulin therapy directed at establishing strict normoglycaemia ( mmol=l or mg=dl) in a series of 1548 patients entering the ICU at Leuven University Hospital. 18 All patients admitted to the unit and receiving mechanical ventilation were eligible for entry: only 14 patients were excluded (moribund state, donot-resuscitate order in place, participation in other trial). The largest fraction (63%) of patients in the study had undergone cardiac surgery, with the remaining patients admitted for a range of noncardiac indications (Table 1). Patients were randomised to receive intensive (n ¼ 765) or conventional (n ¼ 783) therapy on entry to the ICU. Glucose levels were measured at 1 4 hourly intervals in undiluted arterial blood. Conventionally-treated patients received a continuous insulin infusion only if their blood glucose level exceeded 12 mmol=l (215 mg=dl). Patients in the conventional group receiving insulin infusion were to be maintained in a target range of mmol=l or mg=dl. In contrast, patients in the intensive-treatment group received insulin infusion if blood glucose exceeded 6.1 mmol=l (110 mg=dl), the infusion then being adjusted to Table 1 Patients entering the Leuven Study: key characteristics at baseline 18 Intensive Conventional treatment treatment (n ¼ 765) (n ¼ 783) Gender (M=F) 544= =226 Age (mean, y) Reason for ICU entry Cardiac surgery 477 (62%) 493 (63%) Neurologic disease, cerebral 33 (4%) 30 (4%) trauma, brain surgery Thoracic surgery and=or respiratory 66 (9%) 56 (7%) insufficiency Abdominal surgery or peritonitis 45 (6%) 58 (7%) Non-cardiac vascular surgery 30 (4%) 32 (4%) Multiple trauma or severe burns 33 (4%) 35 (4%) Transplantation 46 (6%) 44 (6%) Other 35 (5%) 35 (4%) History of diabetes 101 (13%) 103 (13%) Blood glucose level > 6.1 mmol=l (110 mg=dl) 557 (73%) 598 (76%) > 11.1 mmol=l (200 mg=dl) 81 (11%) 101 (13%) maintain normoglycaemia ( mmol=l or mg= dl) (Figure 1). All patients were fed continuously with 9 g intravenous glucose per hour on the admission day followed by parenteral or enteral feeding according to a standard schedule, progressively increasing the caloric intake up to an average of 25 Cal=kgBW=day after 7 days. On discharge from the ICU, conventional blood glucose control management was adopted with a target range of mmol=l ( mg=dl). Data from the 1548 patients in the study indicate that the algorithms for glucose control were appropriately applied: the conventional treatment group recording a mean morning blood glucose level of 8.5 mmol=l (153 mg=dl) and the intensive group 5.7 mmol=l or 103 mg=dl (P < ). Caloric intake was similar in the two groups, while insulin dose was as expected significantly higher in the intensive group. The primary outcome measure of the study was in-icu mortality, with in-hospital mortality as a secondary endpoint. Other secondary outcomes included morbidity measures such as bloodstream infections. The Leuven Study was planned to continue for 2 y but an interim analysis after 1 y revealed a significant reduction of in-icu mortality in the intensive treatment group, leading to the termination of the study on ethical grounds (Figure 2). During the study period, the mortality rate in the intensivetreatment group was 4.6%, vs 8.0% in the conventional group, representing a significant risk reduction of 43% (P ¼ after adjustment for the interim analyses). The mortality reduction was especially evident among long-stay ICU patients two thirds being non-cardiac surgery patients,

Insulin use during intensive care S5 Figure 1 Control of blood glucose levels during the study. Data are means s.e. 1 0 timepoint is the peak level during day 1 when insulin was started.

3 Insulin use during intensive care S5 Figure 1 Control of blood glucose levels during the study. Data are means s.e. 1 0 timepoint is the peak level during day 1 when insulin was started. Adapted from reference 19. Figure 2 Kaplan-Meier cumulative survival curves for patients in the intensive and conventional treatment groups. (a) Survival in the ICU; (b) Survival in the hospital. Differences were significant in each case (ICU mortality adjusted P ¼ 0.036; hospital mortality nominal P ¼ 0.01). Adapted from reference 18.

6%; conventional treatment, 20.2%; P ¼ 0.005); indeed, mortality during the first 5 days of ICU stay was not different between treatment groups.

4 S6 Insulin use during intensive care Figure 4 Risk of critical illness polyneuropathy vs blood glucose level. Based on data from reference 18. Figure 3 Relative risk reductions for key measures of ICU morbidity. Based on data from reference 18. *P < 0.01; { P < Error bars: 95% confidence intervals. (intensive treatment, 10.6%; conventional treatment, 20.2%; P ¼ 0.005); indeed, mortality during the first 5 days of ICU stay was not different between treatment groups. The principal cause of death among patients in both groups was multiple organ failure (intensive group, 22/35 deaths; conventional group, 51=63 deaths). However, there was an evident reduction in the intensive-treatment group in the number of multiple organ failure deaths with a proven septic focus (8 vs 33 deaths). Morbidity analysis further showed that intensive treatment had reduced the risk of septicaemia by 46% (P ¼ 0.003), reflected also in a 35% reduction in risk of > 10 days antibiotic treatment (P < 0.001). This was accompanied by significant reductions in several other measures of criticalillness morbidity (Figure 3). Insulin or glucose? Our study showed that maintaining normoglycaemia with an insulin infusion dramatically reduces the risk of mortality in the ICU. But is the avoidance of hyperglycaemia, or the provision of insulin, the critical factor? We performed a multivariate logistic regression analysis on the effect of insulin dose and glucose level on risk of ICU mortality. 19 This indicated that both daily insulin dose and mean blood glucose level were independent positive predictors of in-icu mortality. Thus, a high glucose level increased the risk of death; meanwhile, a high daily insulin dose also increased the risk of death. This strongly suggests that the glucose level plays the critical role. The association of insulin dose with mortality risk does not however inevitably imply that insulin is itself hazardous. It is likely to reflect the severity of the hyperglycaemia and of the insulin resistance in the individuals requiring higher insulin doses. Further analysis of glucose levels and mortality risk indicates that, among long-stay ICU patients, those with better levels of control have the better prognosis, and vice versa, with no evident threshold for risk reduction above the normoglycaemic range (Figure 4). 18 Similarly, the risk of critical-illness polyneuropathy was continuously and linearly related to blood glucose level. Although suggestions have been made that insulin has anti-inflammatory effects, including an ability to suppress an impressive range of proinflammatory molecules, 20 the most compelling conclusion from the Leuven Study is that maintaining glucose levels in the normoglycaemic range is the key to reducing ICU mortality. Implications for critical care and an algorithm for insulin administration We have shown that strict maintenance of normoglycaemia during critical illness improves morbidity and mortality. We have also shown that this requires insulin administration in almost all ICU patients not only those with known diabetes. It is possible to develop a protocol for achieving and maintaining normoglycaemia in the ICU (Figure 5). As with any protocol, this algorithm alone does not guarantee improvement in quality of care or clinical outcome. 21 However, in the context of the treating physician s experience and intuition, this approach could be expected to realise significant reductions in ICU mortality and morbidity. It must be stressed, however, that insulin requirements in individual patients vary widely, depending on, for example, their insulin production reserves, their insulin sensitivity before and during critical illness, their caloric intake in the ICU and the severity and nature of their illness. The presence of infections or complications further affects insulin demand, as does the administration of concomitant medications such as corticosteroids. As a result, any algorithm can only be a recommendation that will have to be adapted to the individual condition of each patient. In particular, insulin dosing should be conducted with a degree of common sense: for example, a patient with a high glucose level on

5 Insulin use during intensive care insulin delivery. For example, if a patient is to be transported for investigation or surgery, all intravenous and enteral feeding is usually stopped. Insulin administration should therefore also be stopped, with glucose levels being checked before transport. A patient on full enteral tube feeding may have regular daily or twice-daily interruptions of feeding, in which case insulin should be stopped, or reduced to a low (0.5 IU=h) maintenance dose, during that time. When a patient is extubated prior to starting oral food intake, insulin dose will usually need to be reduced to match nutrient intake. When the patient is ready for discharge to an ordinary ward, insulin infusion can be stopped if the individual is near-normoglycaemic (11.1 mmol=l or 200 mg=dl or less) on low insulin doses (2 IU=h or less). If significant insulin doses are needed to maintain glucose levels below 11.1 mmol=l (200 mg=dl) it is likely that the patient has pre-existing diabetes and follow-up by an endocrinologist should be planned. An insulin regimen should be arranged with the consulting physician to ensure appropriate follow-up in the ordinary ward. S7 Figure 5 Suggested algorithm for achieving and maintaining normoglycaemia in the ICU. In all cases it is the responsibility of the treating physician, relying on experience and knowledge of the patient, to determine dosages and best treatment. Blood glucose levels are to be determined on site in undiluted arterial blood. Insulin, typically 50 IU soluble human insulin in 50 ml 0.9% saline, is administered by continuous infusion through a central venous line. BG, blood glucose. Based on data from reference 18. entry to the ICU (say, 400 mg=dl) should be treated with a relatively high starting insulin dose (say, 4 IU=h), whereas an individual with less severe hyperglycaemia would require a lower starting insulin dose. Subsequent insulin dose adjustments should be appropriate to previously-observed glucose levels, with the dose increment becoming smaller as the patient approaches the normoglycaemic range. Even after normoglycaemia is achieved, the need for monitoring of glucose levels and adjustment of insulin dose remains. In part, this anticipates an improvement in insulin sensitivity with time necessitating a reduction in insulin dose but also acknowledges that worsening infection, with increase in body temperature, increases insulin requirements. Special measures are necessary in patients at risk of acute renal failure. Regular substitution of urinary fluid loss is routinely performed in such patients to avoid fluctuations in intravascular filling status. In such cases, if the patient is receiving an insulin infusion the substitution solutions should also contain insulin appropriate to their glucose content, additional to the insulin infusion. Similarly, reduction or interruption of caloric intake should be accompanied by reduction or interruption of Significant improvements in clinical outcome in surgical ICU patients can be expected by improving their glycaemic status. Unresolved questions include the degree to which these data can be applied to other patient groups (eg medical ICU patients, children in the ICU, surgical patients in ordinary wards). It also remains to be established whether hyperglycaemia per se or the increased availability of insulin is responsible for the major benefits seen in our study. Acknowledgements Our study was performed at the University Hospital, Leuven, Belgium, with support from the Belgian Fund for Scientific Research, the Research Council of the University of Leuven, the Belgian Foundation for Congenital Heart Disease and an unrestricted research grant from Novo Nordisk. References 1 Takala J, Ruokonen E, Webster NR, Nielsen MS, Zandstra DF, Vundelinckx G, Hinds CJ. Increased mortality associated with growth hormone treatment in critically ill adults. New Engl J Med 1999; 341: Vincent J-L. Microvascular endothelial dysfunction: a renewed appreciation of sepsis pathophysiology. Crit Care 2001; 5(Suppl 2): S1 S5. 3 Parrillo JE. Pathogenetic mechanisms of septic shock. New Engl J Med 1993; 328: Zochodne DE, Bolton CF, Wells GA, Gilbert JJ, Hahn AF, Brown JD, Sibbald WA. Critical illness polyneuropathy. A complication of sepsis and multiple organ failure. Brain 1987; 110: Leijten FSS, de Weerd AW. Critical illness polyneuropathy: a review of the literature, definition and pathophysiology. Clin Neurol Neurosurg 1994; 96:

6 S8 Insulin use during intensive care 6 Bolton CF. Sepsis and the systemic inflammatory response syndrome: neuromuscular manifestations. Crit Care Med 1996; 24: Venn R. Sepsis, insulin and noninvasive ventilation. Crit Care 2002; 6: Hawryluck L, Crippen D. Ethics and critical care in the new millennium. Crit Care 2002; 6: Mizock BA. Alterations in fuel metabolism in critical illness: hyperglycaemia. Best Pract Res Clin Endocrinol Metab 2001; 15: Kagansky N, Levy S, Knobler H. The role of hyperglycemia in acute stroke. Arch Neurol 2001; 58: Capes SE, Hunt D, Malmberg K, Gerstine HC. Stress hyperglycemia and increased risk of death after myocardial infarction in patients with and without diabetes: a systematic overview. Lancet 2000; 355: Malmberg K, Norhammar A, Wedel H, Ryden L. Glycometabolic state at admission: important risk marker of mortality in conventionally treated patients with diabetes mellitus and acute myocardial infarction: long-term results from the Diabetes and Insulin-Glucose Infusion in Acute Myocardial Infarction (DIGAMI) study. Circulation 1999; 99: Groeneveld ABJ, Beishuizen A, Visser FC. Insulin: a wonder drug in the critically ill? Crit Care 2002; 6: Gore DC, Chinkes D, Heggers J, Herndon DN, Wolf SE, Desai M. Association of hyperglycemia with increased mortality after severe burn injury. J Trauma 2001; 51: Ljungqvist O, Nygren J, Thorell A. Insulin resistance and elective surgery. Surgery 2000; 128: Diaz R, Paolasso EC, Piegas LS, Tajer CD, Moreno MG, Coravalan R, Isea JE, Romero G, on behalf of the ECLA (Estudios Cardiologicos Latinoamerica) Collaborative Group. Metabolic modulation of acute myocardial infarction: the ECLA glucose-insulin-potassium pilot trial. Circulation 1998; 98: Jonassen AK, Sack MN, Mjøs OD, Yellon DM. Myocardial protection by insulin at reperfusion requires early administration and is mediated via Akt and p70s6 kinase cell-survival signaling. Circ Res 2001; 89: Van den Berghe G, Wouters P, Weekers F, Verwaest C, Bruyninckx F, Schetz M, Vlasselaers D, Ferdinande P, Lauwers P, Bouillon R. Intensive insulin therapy in critically ill patients. New Engl J Med 2001; 345: Van den Berghe G, Wouters PJ, Bouillon R, Weekers F, Verwaest C, Schetz M, Vlasselaers D, Ferdinande P, Lauwers P. Outcome benefit of intensive insulin therapy in the critically ill: insulin dose versus glycemic control. Crit Care Med (in press). 20 Das UN. Insulin and the critically ill. Crit Care 2002; 6: Wall RJ, Dittus RS, Ely EW. Protocol-driven care in the intensive care unit: a tool for quality. Crit Care 2001; 5:

Normal glucose values are associated with a lower risk of mortality in hospitalized patients

Diabetes Care Publish Ahead of Print, published online August Hyperglycemia 20, 2008 in hospital Normal glucose values are associated with a lower risk of mortality in hospitalized patients Alberto Bruno