- Email: [email protected]
Improving blood sugar control during critical illness: A cohort study
Journal of Critical Care (2010) 25, 78–83
Glucose Control
Improving blood sugar control during critical illness: A cohort study Enda O'Connor MB, BCh, BaO, MRCPI, FJFICMa,⁎, David Tragen MBChB, MRCAa , Paul Fahey BSc, MMedStatb , Michael Robinson MBChBc , Theresa Cremasco RNd a
University of Queensland, Toowoomba Health Service District, Toowoomba, Qld 4350, Australia Centre for Rural and Remote Area Health, University of Southern Queensland, Toowoomba, Qld 4350, Australia c Royal Hallamshire Hospital, Sheffield, South Yorkshire, United Kingdom d Intensive Care Unit, Toowoomba Health Service District, Toowoomba, Qld 4350, Australia b
Keywords: Critical illness; Glucose; Insulin; Nurse; Hyperglycemia
Abstract Purpose: The aim of this study is to compare blood sugar control and safety profile of nurse-titrated and medically ordered glucose-insulin regimens. Materials and Methods: We conducted a retrospective cohort study in a 9-bedded regional intensive care unit (ICU) in Queensland, Australia. Seventy critically ill patients requiring one-on-one nursing and intravenous insulin were included. In the nursing group, the ICU nurse decided initial and ongoing insulin infusion rates and glucose measurement frequency. The medical group had a traditional insulin sliding scale prescription. Results: Thirty-seven patients in the nursing group had 1949 glucose measurements. Thirty-three patients in the medical group had 2118 measurements. Mean blood sugar levels (±SD) were 8.33 ± 2.34 and 8.78 ± 2.74 in nursing and medical groups (P b .001). Eighteen percent of glucose readings were greater than 10 mmol/L in the nursing group compared with 27% in the medical group (P = .038). The incidence of hypoglycemia (b2.2 mmol/L) was similar in the 2 groups. Conclusions: In a regional ICU, nurse-titrated glycemic control is safe, effective, and results in high compliance with a glucose target range. Crown Copyright © 2010 Published by Elsevier Inc. All rights reserved.
1. Introduction Hyperglycemia frequently complicates the course of critical illness. The etiologies are multifactorial but include This study was presented as a poster presentation at the Annual Scientific Meeting of the Joint Faculty of Intensive Care Medicine, Melbourne, Australia, May 31, 2008. ⁎ Corresponding author. Tel.: +61 7 31394000; fax: +61 7 38576817. E-mail address: [email protected] (E. O'Connor).
an imbalance between catabolism and anabolism along with a concomitant resistance to circulating insulin [1]. It can be precipitated by iatrogenic factors such as total parenteral nutrition and a number of drugs including glucocoticosteroids and exogenous catecholamines [2]. It is therefore not surprising that hyperglycemia is more the rule rather than the exception in the critically ill. Although hyperglycemia in acute illness is in part a physiological response, multiple studies in diverse populations have demonstrated adverse outcomes with poor
0883-9441/$ – see front matter. Crown Copyright © 2010 Published by Elsevier Inc. All rights reserved. doi:10.1016/j.jcrc.2008.10.008
Improving blood sugar control during critical illness glycemic control. These groups include postoperative patients [3] and patients with multitrauma [4], myocardial infarction [5], subarachnoid hemorrhage [6], burns [7], and general medical and surgical critical illness [8-11]. Predictably, efforts have been made to show an outcome benefit with tight control of blood sugar within lower target ranges [10-12]. These efforts have been met with 2 problems. Firstly, tight glycemic control comes at a cost of increasing incidence of life-threatening hypoglycemia [11,12], a contributing factor in the early termination of a recent randomized controlled trial [12]. Secondly, maintaining a blood glucose level (BGL) within a target range is challenging [2]. In experimental studies, up to 50% of blood sugars fall outside target ranges [10,11,13]. Therefore, it is conceivable that outcome benefits may be realized by better adherence to target ranges as much as by changing BGL targets [2]. There are multiple competing factors contributing to efficacy of glycemic control during acute illness. Although the influence of severity of illness, drug therapy, and patient's comorbidities cannot be overestimated [14], the staff involved in the prescription and administration of insulin and the level of experience thereof may affect the performance of an insulin-glucose titration scale and thus the quality of glucose control [15]. Although the most common glucose-insulin protocol in intensive care unit (ICU) comprises nurse-administered insulin according to a medical prescription or unit insulin protocol, there are grounds for considering an alternative model. In ICUs where qualified nurses semi-autonomously titrate infusion rates of intravenous sedatives and vasoactive drugs, a strong case could be made for glucose-insulin titration by the nurse at the bedside, within a strict predetermined protocol. Therefore, the bases for this study are (a) the importance of adherence to a target BGL range and avoidance of lifethreatening deviations from this range, and (b) to our knowledge, no previous study has compared the performance of medical and independent nurse-titrated insulin scales. We performed a retrospective cohort study of 2 insulin regimens (medically ordered and nurse titrated) and evaluated their performance in terms of overall glucose control and relative safety.
2. Methods 2.1. Study design and participants The study was carried out in a 9-bed regional Australian ICU. After obtaining approval from the local ethics committee, a retrospective chart review was conducted on patients admitted to the ICU during 2 predefined 9-month periods. For the nurse-titrated and medically ordered groups, consecutive eligible patients were enrolled from April to December 2004 and from July 2005 to March
79 2006, respectively. To minimize selection and recording bias, the patient charts were reviewed in a random order. Patients were included if (a) they required a nurse to patient ratio of one-to-one (ie, for invasive or noninvasive positive pressure ventilation, renal replacement therapy, or the use of vasopressors and/or inotropes), (b) they required intravenous insulin to control blood glucose, and (c) they had a predicted ICU length of stay 24 hours or longer. Patients were excluded if they had a condition that necessitated a medically prescribed insulin titration scale (diabetic ketoacidosis, hyperosmolar coma) or one that was managed with a different target range than usual ICU patients (necrotizing fasciitis).
2.2. Study protocol All patients received short-acting insulin (Actrapid, Novo Nordisk, Denmark) prepared as 50 units in 50 mL 0.9% saline. Capillary blood was drawn by the fingerprick method. Glucose levels were measured using a point-of-care device (Medisense Optium, Massachusetts, USA) and recorded on a standard ICU observation flowsheet that included both the glucose reading and the insulin infusion rate. During both study periods, the ICU had a policy of commencing intravenous insulin therapy when the measured blood glucose exceeded 10 mmol/L. Insulin was adjusted to maintain BGL between 5 and 10 mmol/L. In the nursetitrated group, patients were cared for by a senior ICU nurse or by a junior registered nurse with senior supervision. In the medical group, insulin was prescribed by an intensive care consultant or a junior medical officer under consultant supervision. The treatment protocols for each group are summarized in Table 1. Sample prescriptions for each group are shown in Table 2. Patients were commenced on enteral nutrition within 24 hours of admission with the aim of achieving goal caloric intake within 72 hours. Parenteral nutrition was commenced on patients not reaching enteral goals or in whom enteral feeding was contraindicated. The use of systemic steroid therapy in the management of catecholamine-resistant septic shock was commenced in August 2004 (midway through the first study period) and was continued throughout the study.
2.3. Data collection For each patient, we recorded age, sex, admission type (surgical, medical), Acute Physiology and Chronic Health Evaluation III (APACHE III) score, administration of exogenous steroids during admission, administration of exogenous inotropes/vasopressors during admission, and previous diagnosis of diabetes mellitus. To eliminate the variability of admission glucose levels, we commenced glucose recording 6 hours after ICU admission. Recording was discontinued when any of the following events occurred: the patient was discharged from ICU, the patient did not
80 Table 1
E. O'Connor et al. Study protocol for medical and nurse-titrated glucose-insulin scales
Nurse-titrated scale
Medically ordered scale
Insulin commenced when BGL N10 mmol/L Insulin preparation prescribed by MO but no dose prescription written Nurse-initiated decisions regarding: Initial rate of insulin infusion Frequency of BGL measurement Change in rate of insulin infusion according to BGL Medical review of BGL control could be requested by nurse
Insulin commenced when BGL N10 mmol/L Insulin preparation and dose titration scale prescribed by MO MO prescription of: Initial rate of insulin infusion Frequency of BGL measurement Change in rate of insulin infusion according to BGL Medical review of BGL control at least twice each 24 h by junior MO and consultant Medical review could be requested at other times by nurse Nurse not permitted to titrate insulin dose other than that prescribed by MO
MO indicates medical officer.
require intravenous insulin for at least 24 hours, or blood sugar measurements were more than 6 hours apart. For each datum entry, the time of measurement, the blood sugar level, and the insulin infusion rate were recorded. The primary study end point was quality of glycemic control. This was described as the mean blood sugar and the percentage of readings above the upper limit of the target range (ie, N10 mmol/L) for each group. Secondary end points were the number of hypoglycemic episodes (b2.2 mmol/L), average amount of insulin administered per hour, the frequency of BGL measurement, and survival to ICU discharge. We performed a power analysis which indicated that to detect a difference in mean glucose of 0.80 standard deviations, each study group required 50 patients with 1000 to 1500 glucose measurement readings, acknowledging an α value of .05 and a β value of .10. Analyses of patient data consisted of Pearson χ2 tests for categorical variables, independent samples t tests for variables with symmetric distributions, Mann-Whitney U tests for variables with skewed distributions, and multiple regression analysis (with logarithm transformations on skewed variables) to model the relative impacts of multiple predictor variables (including correction for possible baseline differences between groups). Where there were multiple observations per patient, 2-level hierarchical models were used to ensure appropriate accounting of within-patient and between-patient variation. All analyses were conducted on SPSS and MLwiN software [16].
3. Results A total of 70 (nurse titrated 37, medically ordered 33) patients had 4067 blood sugar readings recorded. The baseline characteristics are shown in Table 3. There were more females in the medical group and a trend toward more total parenteral nutrition (TPN) usage in the nurse group. There was no significant difference between groups in baseline characteristics including primary diagnosis, age,
inotrope use, incidence of diabetes, APACHE score, or use of steroids. Clinical end points are summarized in Table 4. Thirtyseven patients in the nurse-titrated group had 1949 blood sugar measurements (mean ± SD, 8.33 ± 2.34), and 33 patients in the medical group had 2118 measurements (mean ± SD, 8.78 ± 2.74). The difference in mean glucose readings was statistically significant (P b .001). Glucose readings above the target range of 10 mmol/L occurred in 27.4% and 18.3% of patients managed by doctors and nurses, respectively (P = .038; odds ratio, 1.65). There was a nonsignificant trend toward increased hourly insulin administration in the nurse-titrated group, and there was no
Table 2 Sample medical prescriptions for medical and nursetitrated glucose-insulin scales Nurse-titrated insulin Actrapid 50 units in 50 mL 0.9% saline continuous intravenous infusion
Medically ordered insulin
Actrapid 50 units in 50 mL 0.9% saline continuous intravenous infusion Commence insulin when Commence insulin when BGL N10 mmol/L BGL N10 mmol/L Insulin rate and BGL frequency BGL measurement every 2 h as per nursing order until BGL between 5 and 10 mmol/L then every 4 h Insulin infusion rate as per sliding scale below BGL(mmol/L) Insulin (unit/h) 0-5 0 5.1-7 1 7.1-9 2 9.1-11 3 11.1-14 4 14.1-17 5 17.1-20 6 N20 Notify MO MO indicates medical officer.
Improving blood sugar control during critical illness Table 3
Baseline characteristics of study groups Nurse titrated Medically ordered P (n = 37) (n = 33)
Male, n (%) Med dx, n (%) Inotrope use, n (%) Age (y), mean ± SD Diabetes, n (%) Steroids, n (%) TPN use, n (%) APACHE III score
23 (62.2) 23 (62.2) 23 (62.2) 64.8 ± 16.4 12 (32.4) 15 (40.5) 5 (13.5) 78.95
11 (33.3) 24 (72.7) 16 (48.5) 62.6 ± 14.4 16 (48.5) 13 (39.4) 0 (0) 75.58
.03 ⁎ .31 .34 .55 .22 1.0 .055 .65
Med dx indicates medical diagnosis. ⁎ Statistical significance with P b .05.
difference in frequency of blood glucose measurement between groups. Only 3 blood sugar readings were less than 2.2 mmol/L. Twenty-nine (87.9%) patients in medical group and 31 (83.8%) in nurse group survived to ICU discharge (P = .74). When results were corrected for differences in baseline characteristics, the study end points were unchanged.
4. Discussion In this retrospective study, nurse-titrated glycemic control was shown to be feasible and safe and resulted in better compliance with a target glucose level of 5 to 10 mmol/L when compared with a medically ordered titration regimen. To our knowledge, this is the first study directly comparing 2 such insulin regimens. Several models of insulin-glucose titration have been shown to improve glycemic control in the ICU. These include a Web-based titration calculator [17], the presence of a nurse quality supervisor [18], use of a standardized insulin adjustment protocol [10-13,19-21], and use of a nomogram for insulin administration in place of the traditional insulin sliding scale [15]. Despite clear benefits associated with their use, maintaining glucose levels consistently within a target range remains difficult and target compliance of greater than 70% is infrequently achieved [22,23]. Two recent randomized controlled trials Table 4
81 of tight blood glucose control demonstrated that 30% of patients in the intervention group had glucose levels above the target range [10,11]. Similarly, a large observational study in Australasian ICUs reported 49% compliance with a target glucose range when a new insulin regimen was used [13]. This highlights the difficulty in applying one insulin protocol to diverse groups of critically ill patients [14]. In contrast, the intervention group in our study had nurse-initiated bedside adjustments made to insulin infusion without the input from a unit protocol or attending physician and achieved 76.9% of glucose readings within target range, significantly greater than the medical group. Why might this have occurred? Although there was no difference in the frequency of glucose measurement between groups, there was a trend toward more insulin delivery by nurses. Furthermore, as detection of poor glucose control in the medically ordered group depended on either regular medical review or on notification to the doctor, the availability of medical staff may have influenced quality of glycemic control in these patients. This factor was not explored in our study. In the nursing group, autonomous decision making may have allowed the nurse to preempt insulin requirements when, for example, changing caloric intake or commencing steroid therapy. It may have permitted them to change measurement frequency according to glucose stability or, crucially, to evaluate the insulin needs of each individual patient and to respond to these needs without recourse to a medical officer, who may not have been available at the time. Thus, it is likely that the decision cycle-time [24] (the time taken from discovery of an abnormal BGL to a change made in insulin dose) was reduced in the nurse-titrated group because for the most part, the nurse was present at the bedside to make a decision when one was required. This phenomenon of long decision cycle-time in the clinical environment has been well described [25]. Lastly, although our relatively wide target range might have improved compliance, recent studies would support the use of such a conservative blood glucose target [10,12]. Different end points have been used as measures of glucose control in critical illness; mean glucose [17], mean morning glucose [10,11,13], peak glucose [26], percentage of total time within prespecified range [27], actual time over 24 hours within target range [19], area under a curve of blood
Primary and secondary end points
Glucose, mmol/L (mean ± SD) Glucose N10 mmol/L (%) Glucose b2.2 mmol/L(n) Glucose readings per 24 h, n, median (IQR) Insulin per hour, u, median (IQR) ICU survival, n (%) ⁎ Statistical significance with P b .05.
Nurse titrated (total 1949 BGL readings)
Medically ordered (total 2118 BGL readings)
P
8.33 ± 2.34 18.3 2 10.67 (9.79/12.72) 2.09 (1.21/3.18) 31 (83.8)
8.78 ± 2.74 27.4 1 11.89 (10.75/13.36) 1.52 (1.10/2.67) 29 (87.9)
b.001 ⁎ .038 ⁎ n/a .53 .151 .74
82 glucose concentration vs time [15,28], median glucose [9], and admission blood glucose [13]. Measurements such as peak glucose and mean morning glucose emphasize single or subset values rather than overall glucose control. Mean glucose can be altered by outlying values but may also conceal multiple glucose readings above and below target ranges [28]. Area under the curve implies a linear change between consecutive measurements, and percentage values vary with measurement frequency. In appreciation of these limitations, we selected 2 primary end points (mean glucose and percentage readings above upper target level) to summarize overall glucose control as well as spread of levels above the target range. Independent analysis of measurement frequency allowed us to evaluate its effect on the study results. Our finding of improved compliance with a target glucose range is a clinically relevant end point. Although there was no group difference in the incidence of hypoglycemia in our analysis, adherence to target levels does reduce the incidence over time of both hypo- and hyperglycemic events [2]. Although our study was underpowered to evaluate the effect of such events, poor glucose control has repeatedly been shown to adversely affect morbidity and mortality [6,20]. Our study has limitations. The retrospective nature precluded randomization, there was no standardization of nonexperimental therapies, and treatments were not blinded. The differences noted in baseline characteristics may have contributed to study results. Conversely, the study represents the day-to-day challenges associated with glucose control in critical illness in a nonexperimental setting and eliminates the confounding Hawthorne effect seen in prospective studies [29]. There was no standardization of ICU staff between the 2 study periods, there was minimal change in the nursing staff, but all junior medical staff had changed and one additional intensive care consultant had been employed. There was no formal tuition for nurses regarding insulin titration in critical illness, but most patients were cared for either by a senior registered nurse or a junior nurse with support from a senior colleague. In Queensland, Australia, standards for registered nurses have been published and include the ability to care for specific patient groups in complex clinical situations, advanced clinical skills, and ability to supervise junior colleagues [30]. In conclusion, a glucose-insulin protocol controlled by intensive care nurses in a regional ICU is a feasible proposal and may result in better adherence to a target blood sugar range without increasing frequency of blood sampling or risk of life-threatening hypoglycemia.
References [1] Marik PE, Raghavan M. Stress-hyperglycaemia, insulin and immunomodulation in sepsis. Intensive Care Med 2004;30:748-56.
E. O'Connor et al. [2] Vanhorebeek I, Langouche L, Van den Berghe G. Tight blood glucose control with insulin in ICU. Facts and controversies. Chest 2007;132: 268-78. [3] Grey NJ, Perdrizet GA. Reduction of nosocomial infections in the surgical intensive care unit by strict glycaemic control. Endocr Pract 2004;10(Suppl 2):46-52. [4] Gale SC, Sicoutris C, Reilly PM, et al. Poor glycaemic control is associated with increased mortality in critically ill trauma patients. Am Surg 2004;73:454-60. [5] Bhadriraju S, Ray KK, DeFranco AC, et al. Association between blood glucose and long term mortality in patients with acute coronary syndromes in the OPUS-TIMI 16 trial. Am J Cardiol 2006;97:1573-7. [6] Lanzino G, Kassell NF, Germanson T, et al. Plasma glucose levels and outcome after aneurysmal subarachnoid haemorrhage. J Neurosurg 1993;79:885-91. [7] Gore DC, Chinkes D, Heggers J, et al. Association of hyperglycaemia with increased mortality after severe burn injury. J Trauma 2001;51: 540-4. [8] Van den Berghe G, Schoonheydt K, Becx P, et al. Insulin therapy protects the central and peripheral nervous system in intensive care patients. Neurology 2005;64:1348-53. [9] Finney SJ, Zekveld C, Elia A, et al. Glucose control and mortality in critically ill patients. JAMA 2003;290:2041-7. [10] Van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in critically ill patients. NEJM 2001;345:1359-67. [11] Van den Berghe G, Wilmer A, Hermans G, et al. Intensive insulin therapy in the medical ICU. NEJM 2006;354:449-61. [12] Brunkhorst FM, Engel C, Bloos F, et al. Intensive insulin therapy and pentastarch resuscitation in severe sepsis. NEJM 2008;358:125-39. [13] Orford N, Stow P, Green D, et al. Safety and feasibility of an insulin adjustment protocol to maintain blood glucose concentrations within a narrow range in critically ill patients in an Australian level III adult intensive care unit. Crit Care Resusc 2004;6:92-8. [14] Rady MY, Johnson DJ, Patel BM, et al. Influence of individual characteristics on outcome of glycaemic control in intensive care patients with or without diabetes mellitus. Mayo Clin Proc 2005;80: 1558-67. [15] Brown G, Dodek P. Intravenous insulin nomogram improves blood glucose control in the critically ill. Crit Care Med 2001;29:1714-9. [16] Rasbash J, Steele F, Browne WJ, et al. A user's guide to MLwiN Version 2. Centre for Multilevel Modelling, University of Bristol. Available at http://www.cmm.bristol.ac.uk/MLwiN/index.shtml. Accessed November 1 2008. [17] Thomas AN, Marchant AE, Ogden MC, et al. Implementation of a tight glycaemic control protocol using a Web-based insulin dose calculator. Anaesthesia 2005;60:1093-100. [18] Reed CC, Stewart RM, Sherman M, et al. Intensive insulin protocol improves glucose control and is associated with a reduction in intensive care mortality. J Am Coll Surg 2007;204:1048-55. [19] Kanji S, Singh A, Tierney M, et al. Standardisation of intravenous insulin therapy improves the efficiency and safety of blood glucose control in critically ill adults. Intensive Care Med 2004;30:804-10. [20] Krinsley JS. Effect of an intensive glucose management protocol on the mortality of critically ill adult patients. Mayo Clin Proc 2004;79: 992-1000. [21] Pachler C, Plank J, Weinhandl H, et al. Tight glycaemic control by an automated algorithm with time-variant sampling in medical ICU patients. Intensive Care Med 2008;34:1224-30. [22] Bland DK, Fankhanel Y, Langford E, et al. Intensive versus modified conventional control of blood glucose level in medical intensive care patients: a pilot study. Am J Crit Care 2005;14:370-6. [23] Vogelzang M, Loef BG, Regtien JG, et al. Computer assisted glucose control in critically ill patients. Intensive Care Med 2008;34:1421-7. [24] Wetherbe JC. Principles of cycle time reduction: you can have your cake and eat it too. Cycle Time Res 1995;1:1-24. [25] Lefrant JY, Muller L, Bruelle P, et al. Insertion time of the pulmonary artery catheter in critically ill patients. Crit Care Med 2000;28:355-9.
Improving blood sugar control during critical illness [26] Mitchell I, Finfer S, Bellomo R, et al. Management of blood glucose in the critically ill in Australia and New Zealand: a practice survey and inception cohort study. Intensive Care Med 2006;32: 867-74. [27] Scheuren L, Baetz B, Cawley MJ, et al. Pharmacist designed and nursing driven insulin infusion protocol to achieve and maintain glycaemic control in critical care patients. J Trauma Nurs 2006;13: 140-5.
83 [28] Vogelzang M, van der Horst ICC, Nijsten MWN. Hyperglycaemic index as a tool to assess glucose control: a retrospective study. Crit Care 2004;8:R122-7. [29] Landsberger HA. (1958) Hawthorne revisited. New York (NY): Ithaca; 1956. [30] Nursing and Midwifery Classification Structure. Queensland Government. Available via http://www.health.qld.gov.au/hrpolicies/resourcing/b_7.pdf. Accessed 2 July 2008.
Glucose Control
Improving blood sugar control during critical illness: A cohort study Enda O'Connor MB, BCh, BaO, MRCPI, FJFICMa,⁎, David Tragen MBChB, MRCAa , Paul Fahey BSc, MMedStatb , Michael Robinson MBChBc , Theresa Cremasco RNd a
University of Queensland, Toowoomba Health Service District, Toowoomba, Qld 4350, Australia Centre for Rural and Remote Area Health, University of Southern Queensland, Toowoomba, Qld 4350, Australia c Royal Hallamshire Hospital, Sheffield, South Yorkshire, United Kingdom d Intensive Care Unit, Toowoomba Health Service District, Toowoomba, Qld 4350, Australia b
Keywords: Critical illness; Glucose; Insulin; Nurse; Hyperglycemia
Abstract Purpose: The aim of this study is to compare blood sugar control and safety profile of nurse-titrated and medically ordered glucose-insulin regimens. Materials and Methods: We conducted a retrospective cohort study in a 9-bedded regional intensive care unit (ICU) in Queensland, Australia. Seventy critically ill patients requiring one-on-one nursing and intravenous insulin were included. In the nursing group, the ICU nurse decided initial and ongoing insulin infusion rates and glucose measurement frequency. The medical group had a traditional insulin sliding scale prescription. Results: Thirty-seven patients in the nursing group had 1949 glucose measurements. Thirty-three patients in the medical group had 2118 measurements. Mean blood sugar levels (±SD) were 8.33 ± 2.34 and 8.78 ± 2.74 in nursing and medical groups (P b .001). Eighteen percent of glucose readings were greater than 10 mmol/L in the nursing group compared with 27% in the medical group (P = .038). The incidence of hypoglycemia (b2.2 mmol/L) was similar in the 2 groups. Conclusions: In a regional ICU, nurse-titrated glycemic control is safe, effective, and results in high compliance with a glucose target range. Crown Copyright © 2010 Published by Elsevier Inc. All rights reserved.
1. Introduction Hyperglycemia frequently complicates the course of critical illness. The etiologies are multifactorial but include This study was presented as a poster presentation at the Annual Scientific Meeting of the Joint Faculty of Intensive Care Medicine, Melbourne, Australia, May 31, 2008. ⁎ Corresponding author. Tel.: +61 7 31394000; fax: +61 7 38576817. E-mail address: [email protected] (E. O'Connor).
an imbalance between catabolism and anabolism along with a concomitant resistance to circulating insulin [1]. It can be precipitated by iatrogenic factors such as total parenteral nutrition and a number of drugs including glucocoticosteroids and exogenous catecholamines [2]. It is therefore not surprising that hyperglycemia is more the rule rather than the exception in the critically ill. Although hyperglycemia in acute illness is in part a physiological response, multiple studies in diverse populations have demonstrated adverse outcomes with poor
0883-9441/$ – see front matter. Crown Copyright © 2010 Published by Elsevier Inc. All rights reserved. doi:10.1016/j.jcrc.2008.10.008
Improving blood sugar control during critical illness glycemic control. These groups include postoperative patients [3] and patients with multitrauma [4], myocardial infarction [5], subarachnoid hemorrhage [6], burns [7], and general medical and surgical critical illness [8-11]. Predictably, efforts have been made to show an outcome benefit with tight control of blood sugar within lower target ranges [10-12]. These efforts have been met with 2 problems. Firstly, tight glycemic control comes at a cost of increasing incidence of life-threatening hypoglycemia [11,12], a contributing factor in the early termination of a recent randomized controlled trial [12]. Secondly, maintaining a blood glucose level (BGL) within a target range is challenging [2]. In experimental studies, up to 50% of blood sugars fall outside target ranges [10,11,13]. Therefore, it is conceivable that outcome benefits may be realized by better adherence to target ranges as much as by changing BGL targets [2]. There are multiple competing factors contributing to efficacy of glycemic control during acute illness. Although the influence of severity of illness, drug therapy, and patient's comorbidities cannot be overestimated [14], the staff involved in the prescription and administration of insulin and the level of experience thereof may affect the performance of an insulin-glucose titration scale and thus the quality of glucose control [15]. Although the most common glucose-insulin protocol in intensive care unit (ICU) comprises nurse-administered insulin according to a medical prescription or unit insulin protocol, there are grounds for considering an alternative model. In ICUs where qualified nurses semi-autonomously titrate infusion rates of intravenous sedatives and vasoactive drugs, a strong case could be made for glucose-insulin titration by the nurse at the bedside, within a strict predetermined protocol. Therefore, the bases for this study are (a) the importance of adherence to a target BGL range and avoidance of lifethreatening deviations from this range, and (b) to our knowledge, no previous study has compared the performance of medical and independent nurse-titrated insulin scales. We performed a retrospective cohort study of 2 insulin regimens (medically ordered and nurse titrated) and evaluated their performance in terms of overall glucose control and relative safety.
2. Methods 2.1. Study design and participants The study was carried out in a 9-bed regional Australian ICU. After obtaining approval from the local ethics committee, a retrospective chart review was conducted on patients admitted to the ICU during 2 predefined 9-month periods. For the nurse-titrated and medically ordered groups, consecutive eligible patients were enrolled from April to December 2004 and from July 2005 to March
79 2006, respectively. To minimize selection and recording bias, the patient charts were reviewed in a random order. Patients were included if (a) they required a nurse to patient ratio of one-to-one (ie, for invasive or noninvasive positive pressure ventilation, renal replacement therapy, or the use of vasopressors and/or inotropes), (b) they required intravenous insulin to control blood glucose, and (c) they had a predicted ICU length of stay 24 hours or longer. Patients were excluded if they had a condition that necessitated a medically prescribed insulin titration scale (diabetic ketoacidosis, hyperosmolar coma) or one that was managed with a different target range than usual ICU patients (necrotizing fasciitis).
2.2. Study protocol All patients received short-acting insulin (Actrapid, Novo Nordisk, Denmark) prepared as 50 units in 50 mL 0.9% saline. Capillary blood was drawn by the fingerprick method. Glucose levels were measured using a point-of-care device (Medisense Optium, Massachusetts, USA) and recorded on a standard ICU observation flowsheet that included both the glucose reading and the insulin infusion rate. During both study periods, the ICU had a policy of commencing intravenous insulin therapy when the measured blood glucose exceeded 10 mmol/L. Insulin was adjusted to maintain BGL between 5 and 10 mmol/L. In the nursetitrated group, patients were cared for by a senior ICU nurse or by a junior registered nurse with senior supervision. In the medical group, insulin was prescribed by an intensive care consultant or a junior medical officer under consultant supervision. The treatment protocols for each group are summarized in Table 1. Sample prescriptions for each group are shown in Table 2. Patients were commenced on enteral nutrition within 24 hours of admission with the aim of achieving goal caloric intake within 72 hours. Parenteral nutrition was commenced on patients not reaching enteral goals or in whom enteral feeding was contraindicated. The use of systemic steroid therapy in the management of catecholamine-resistant septic shock was commenced in August 2004 (midway through the first study period) and was continued throughout the study.
2.3. Data collection For each patient, we recorded age, sex, admission type (surgical, medical), Acute Physiology and Chronic Health Evaluation III (APACHE III) score, administration of exogenous steroids during admission, administration of exogenous inotropes/vasopressors during admission, and previous diagnosis of diabetes mellitus. To eliminate the variability of admission glucose levels, we commenced glucose recording 6 hours after ICU admission. Recording was discontinued when any of the following events occurred: the patient was discharged from ICU, the patient did not
80 Table 1
E. O'Connor et al. Study protocol for medical and nurse-titrated glucose-insulin scales
Nurse-titrated scale
Medically ordered scale
Insulin commenced when BGL N10 mmol/L Insulin preparation prescribed by MO but no dose prescription written Nurse-initiated decisions regarding: Initial rate of insulin infusion Frequency of BGL measurement Change in rate of insulin infusion according to BGL Medical review of BGL control could be requested by nurse
Insulin commenced when BGL N10 mmol/L Insulin preparation and dose titration scale prescribed by MO MO prescription of: Initial rate of insulin infusion Frequency of BGL measurement Change in rate of insulin infusion according to BGL Medical review of BGL control at least twice each 24 h by junior MO and consultant Medical review could be requested at other times by nurse Nurse not permitted to titrate insulin dose other than that prescribed by MO
MO indicates medical officer.
require intravenous insulin for at least 24 hours, or blood sugar measurements were more than 6 hours apart. For each datum entry, the time of measurement, the blood sugar level, and the insulin infusion rate were recorded. The primary study end point was quality of glycemic control. This was described as the mean blood sugar and the percentage of readings above the upper limit of the target range (ie, N10 mmol/L) for each group. Secondary end points were the number of hypoglycemic episodes (b2.2 mmol/L), average amount of insulin administered per hour, the frequency of BGL measurement, and survival to ICU discharge. We performed a power analysis which indicated that to detect a difference in mean glucose of 0.80 standard deviations, each study group required 50 patients with 1000 to 1500 glucose measurement readings, acknowledging an α value of .05 and a β value of .10. Analyses of patient data consisted of Pearson χ2 tests for categorical variables, independent samples t tests for variables with symmetric distributions, Mann-Whitney U tests for variables with skewed distributions, and multiple regression analysis (with logarithm transformations on skewed variables) to model the relative impacts of multiple predictor variables (including correction for possible baseline differences between groups). Where there were multiple observations per patient, 2-level hierarchical models were used to ensure appropriate accounting of within-patient and between-patient variation. All analyses were conducted on SPSS and MLwiN software [16].
3. Results A total of 70 (nurse titrated 37, medically ordered 33) patients had 4067 blood sugar readings recorded. The baseline characteristics are shown in Table 3. There were more females in the medical group and a trend toward more total parenteral nutrition (TPN) usage in the nurse group. There was no significant difference between groups in baseline characteristics including primary diagnosis, age,
inotrope use, incidence of diabetes, APACHE score, or use of steroids. Clinical end points are summarized in Table 4. Thirtyseven patients in the nurse-titrated group had 1949 blood sugar measurements (mean ± SD, 8.33 ± 2.34), and 33 patients in the medical group had 2118 measurements (mean ± SD, 8.78 ± 2.74). The difference in mean glucose readings was statistically significant (P b .001). Glucose readings above the target range of 10 mmol/L occurred in 27.4% and 18.3% of patients managed by doctors and nurses, respectively (P = .038; odds ratio, 1.65). There was a nonsignificant trend toward increased hourly insulin administration in the nurse-titrated group, and there was no
Table 2 Sample medical prescriptions for medical and nursetitrated glucose-insulin scales Nurse-titrated insulin Actrapid 50 units in 50 mL 0.9% saline continuous intravenous infusion
Medically ordered insulin
Actrapid 50 units in 50 mL 0.9% saline continuous intravenous infusion Commence insulin when Commence insulin when BGL N10 mmol/L BGL N10 mmol/L Insulin rate and BGL frequency BGL measurement every 2 h as per nursing order until BGL between 5 and 10 mmol/L then every 4 h Insulin infusion rate as per sliding scale below BGL(mmol/L) Insulin (unit/h) 0-5 0 5.1-7 1 7.1-9 2 9.1-11 3 11.1-14 4 14.1-17 5 17.1-20 6 N20 Notify MO MO indicates medical officer.
Improving blood sugar control during critical illness Table 3
Baseline characteristics of study groups Nurse titrated Medically ordered P (n = 37) (n = 33)
Male, n (%) Med dx, n (%) Inotrope use, n (%) Age (y), mean ± SD Diabetes, n (%) Steroids, n (%) TPN use, n (%) APACHE III score
23 (62.2) 23 (62.2) 23 (62.2) 64.8 ± 16.4 12 (32.4) 15 (40.5) 5 (13.5) 78.95
11 (33.3) 24 (72.7) 16 (48.5) 62.6 ± 14.4 16 (48.5) 13 (39.4) 0 (0) 75.58
.03 ⁎ .31 .34 .55 .22 1.0 .055 .65
Med dx indicates medical diagnosis. ⁎ Statistical significance with P b .05.
difference in frequency of blood glucose measurement between groups. Only 3 blood sugar readings were less than 2.2 mmol/L. Twenty-nine (87.9%) patients in medical group and 31 (83.8%) in nurse group survived to ICU discharge (P = .74). When results were corrected for differences in baseline characteristics, the study end points were unchanged.
4. Discussion In this retrospective study, nurse-titrated glycemic control was shown to be feasible and safe and resulted in better compliance with a target glucose level of 5 to 10 mmol/L when compared with a medically ordered titration regimen. To our knowledge, this is the first study directly comparing 2 such insulin regimens. Several models of insulin-glucose titration have been shown to improve glycemic control in the ICU. These include a Web-based titration calculator [17], the presence of a nurse quality supervisor [18], use of a standardized insulin adjustment protocol [10-13,19-21], and use of a nomogram for insulin administration in place of the traditional insulin sliding scale [15]. Despite clear benefits associated with their use, maintaining glucose levels consistently within a target range remains difficult and target compliance of greater than 70% is infrequently achieved [22,23]. Two recent randomized controlled trials Table 4
81 of tight blood glucose control demonstrated that 30% of patients in the intervention group had glucose levels above the target range [10,11]. Similarly, a large observational study in Australasian ICUs reported 49% compliance with a target glucose range when a new insulin regimen was used [13]. This highlights the difficulty in applying one insulin protocol to diverse groups of critically ill patients [14]. In contrast, the intervention group in our study had nurse-initiated bedside adjustments made to insulin infusion without the input from a unit protocol or attending physician and achieved 76.9% of glucose readings within target range, significantly greater than the medical group. Why might this have occurred? Although there was no difference in the frequency of glucose measurement between groups, there was a trend toward more insulin delivery by nurses. Furthermore, as detection of poor glucose control in the medically ordered group depended on either regular medical review or on notification to the doctor, the availability of medical staff may have influenced quality of glycemic control in these patients. This factor was not explored in our study. In the nursing group, autonomous decision making may have allowed the nurse to preempt insulin requirements when, for example, changing caloric intake or commencing steroid therapy. It may have permitted them to change measurement frequency according to glucose stability or, crucially, to evaluate the insulin needs of each individual patient and to respond to these needs without recourse to a medical officer, who may not have been available at the time. Thus, it is likely that the decision cycle-time [24] (the time taken from discovery of an abnormal BGL to a change made in insulin dose) was reduced in the nurse-titrated group because for the most part, the nurse was present at the bedside to make a decision when one was required. This phenomenon of long decision cycle-time in the clinical environment has been well described [25]. Lastly, although our relatively wide target range might have improved compliance, recent studies would support the use of such a conservative blood glucose target [10,12]. Different end points have been used as measures of glucose control in critical illness; mean glucose [17], mean morning glucose [10,11,13], peak glucose [26], percentage of total time within prespecified range [27], actual time over 24 hours within target range [19], area under a curve of blood
Primary and secondary end points
Glucose, mmol/L (mean ± SD) Glucose N10 mmol/L (%) Glucose b2.2 mmol/L(n) Glucose readings per 24 h, n, median (IQR) Insulin per hour, u, median (IQR) ICU survival, n (%) ⁎ Statistical significance with P b .05.
Nurse titrated (total 1949 BGL readings)
Medically ordered (total 2118 BGL readings)
P
8.33 ± 2.34 18.3 2 10.67 (9.79/12.72) 2.09 (1.21/3.18) 31 (83.8)
8.78 ± 2.74 27.4 1 11.89 (10.75/13.36) 1.52 (1.10/2.67) 29 (87.9)
b.001 ⁎ .038 ⁎ n/a .53 .151 .74
82 glucose concentration vs time [15,28], median glucose [9], and admission blood glucose [13]. Measurements such as peak glucose and mean morning glucose emphasize single or subset values rather than overall glucose control. Mean glucose can be altered by outlying values but may also conceal multiple glucose readings above and below target ranges [28]. Area under the curve implies a linear change between consecutive measurements, and percentage values vary with measurement frequency. In appreciation of these limitations, we selected 2 primary end points (mean glucose and percentage readings above upper target level) to summarize overall glucose control as well as spread of levels above the target range. Independent analysis of measurement frequency allowed us to evaluate its effect on the study results. Our finding of improved compliance with a target glucose range is a clinically relevant end point. Although there was no group difference in the incidence of hypoglycemia in our analysis, adherence to target levels does reduce the incidence over time of both hypo- and hyperglycemic events [2]. Although our study was underpowered to evaluate the effect of such events, poor glucose control has repeatedly been shown to adversely affect morbidity and mortality [6,20]. Our study has limitations. The retrospective nature precluded randomization, there was no standardization of nonexperimental therapies, and treatments were not blinded. The differences noted in baseline characteristics may have contributed to study results. Conversely, the study represents the day-to-day challenges associated with glucose control in critical illness in a nonexperimental setting and eliminates the confounding Hawthorne effect seen in prospective studies [29]. There was no standardization of ICU staff between the 2 study periods, there was minimal change in the nursing staff, but all junior medical staff had changed and one additional intensive care consultant had been employed. There was no formal tuition for nurses regarding insulin titration in critical illness, but most patients were cared for either by a senior registered nurse or a junior nurse with support from a senior colleague. In Queensland, Australia, standards for registered nurses have been published and include the ability to care for specific patient groups in complex clinical situations, advanced clinical skills, and ability to supervise junior colleagues [30]. In conclusion, a glucose-insulin protocol controlled by intensive care nurses in a regional ICU is a feasible proposal and may result in better adherence to a target blood sugar range without increasing frequency of blood sampling or risk of life-threatening hypoglycemia.
References [1] Marik PE, Raghavan M. Stress-hyperglycaemia, insulin and immunomodulation in sepsis. Intensive Care Med 2004;30:748-56.
E. O'Connor et al. [2] Vanhorebeek I, Langouche L, Van den Berghe G. Tight blood glucose control with insulin in ICU. Facts and controversies. Chest 2007;132: 268-78. [3] Grey NJ, Perdrizet GA. Reduction of nosocomial infections in the surgical intensive care unit by strict glycaemic control. Endocr Pract 2004;10(Suppl 2):46-52. [4] Gale SC, Sicoutris C, Reilly PM, et al. Poor glycaemic control is associated with increased mortality in critically ill trauma patients. Am Surg 2004;73:454-60. [5] Bhadriraju S, Ray KK, DeFranco AC, et al. Association between blood glucose and long term mortality in patients with acute coronary syndromes in the OPUS-TIMI 16 trial. Am J Cardiol 2006;97:1573-7. [6] Lanzino G, Kassell NF, Germanson T, et al. Plasma glucose levels and outcome after aneurysmal subarachnoid haemorrhage. J Neurosurg 1993;79:885-91. [7] Gore DC, Chinkes D, Heggers J, et al. Association of hyperglycaemia with increased mortality after severe burn injury. J Trauma 2001;51: 540-4. [8] Van den Berghe G, Schoonheydt K, Becx P, et al. Insulin therapy protects the central and peripheral nervous system in intensive care patients. Neurology 2005;64:1348-53. [9] Finney SJ, Zekveld C, Elia A, et al. Glucose control and mortality in critically ill patients. JAMA 2003;290:2041-7. [10] Van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in critically ill patients. NEJM 2001;345:1359-67. [11] Van den Berghe G, Wilmer A, Hermans G, et al. Intensive insulin therapy in the medical ICU. NEJM 2006;354:449-61. [12] Brunkhorst FM, Engel C, Bloos F, et al. Intensive insulin therapy and pentastarch resuscitation in severe sepsis. NEJM 2008;358:125-39. [13] Orford N, Stow P, Green D, et al. Safety and feasibility of an insulin adjustment protocol to maintain blood glucose concentrations within a narrow range in critically ill patients in an Australian level III adult intensive care unit. Crit Care Resusc 2004;6:92-8. [14] Rady MY, Johnson DJ, Patel BM, et al. Influence of individual characteristics on outcome of glycaemic control in intensive care patients with or without diabetes mellitus. Mayo Clin Proc 2005;80: 1558-67. [15] Brown G, Dodek P. Intravenous insulin nomogram improves blood glucose control in the critically ill. Crit Care Med 2001;29:1714-9. [16] Rasbash J, Steele F, Browne WJ, et al. A user's guide to MLwiN Version 2. Centre for Multilevel Modelling, University of Bristol. Available at http://www.cmm.bristol.ac.uk/MLwiN/index.shtml. Accessed November 1 2008. [17] Thomas AN, Marchant AE, Ogden MC, et al. Implementation of a tight glycaemic control protocol using a Web-based insulin dose calculator. Anaesthesia 2005;60:1093-100. [18] Reed CC, Stewart RM, Sherman M, et al. Intensive insulin protocol improves glucose control and is associated with a reduction in intensive care mortality. J Am Coll Surg 2007;204:1048-55. [19] Kanji S, Singh A, Tierney M, et al. Standardisation of intravenous insulin therapy improves the efficiency and safety of blood glucose control in critically ill adults. Intensive Care Med 2004;30:804-10. [20] Krinsley JS. Effect of an intensive glucose management protocol on the mortality of critically ill adult patients. Mayo Clin Proc 2004;79: 992-1000. [21] Pachler C, Plank J, Weinhandl H, et al. Tight glycaemic control by an automated algorithm with time-variant sampling in medical ICU patients. Intensive Care Med 2008;34:1224-30. [22] Bland DK, Fankhanel Y, Langford E, et al. Intensive versus modified conventional control of blood glucose level in medical intensive care patients: a pilot study. Am J Crit Care 2005;14:370-6. [23] Vogelzang M, Loef BG, Regtien JG, et al. Computer assisted glucose control in critically ill patients. Intensive Care Med 2008;34:1421-7. [24] Wetherbe JC. Principles of cycle time reduction: you can have your cake and eat it too. Cycle Time Res 1995;1:1-24. [25] Lefrant JY, Muller L, Bruelle P, et al. Insertion time of the pulmonary artery catheter in critically ill patients. Crit Care Med 2000;28:355-9.
Improving blood sugar control during critical illness [26] Mitchell I, Finfer S, Bellomo R, et al. Management of blood glucose in the critically ill in Australia and New Zealand: a practice survey and inception cohort study. Intensive Care Med 2006;32: 867-74. [27] Scheuren L, Baetz B, Cawley MJ, et al. Pharmacist designed and nursing driven insulin infusion protocol to achieve and maintain glycaemic control in critical care patients. J Trauma Nurs 2006;13: 140-5.
83 [28] Vogelzang M, van der Horst ICC, Nijsten MWN. Hyperglycaemic index as a tool to assess glucose control: a retrospective study. Crit Care 2004;8:R122-7. [29] Landsberger HA. (1958) Hawthorne revisited. New York (NY): Ithaca; 1956. [30] Nursing and Midwifery Classification Structure. Queensland Government. Available via http://www.health.qld.gov.au/hrpolicies/resourcing/b_7.pdf. Accessed 2 July 2008.