1. Hyperglycemia: Is This Still a Concern? Lauren E. Healy BA, PharmD, BCPS NYSCHP Downstate Critical Care Program October 2, 2015

2. Conflicts of Interest  None to disclose 2

3. Learning Objectives  Explain the pathophysiology of hyperglycemia in critically ill patients  Define the association between hyperglycemia and clinical outcomes  Evaluate the recent literature on glycemic control in critically ill patients  Specify glycemic targets for individual patient populations  Compare the options for glycemic control 3

4. Stress Hyperglycemia  Blood glucose (BG) > 200 - 220 mg/dL in the presence of an acute illness  Usually resolves with treatment of underlying illness but can have lasting sequela 4 Donahey. Pharm Pract News. November 2013.

5. A Little History  1997  Malmberg and Colleagues of the Diabetes Mellitus Insulin Glucose Infusion in Acute Myocardial Infarction (DIGAMI) study group  Considered long-term, all cause mortality in patients with Diabetes who were post-MI  Compared intensive insulin treatment (IIT) (126 – 180 mg/dL) vs. control  Mean (range) follow-up was 3.4 (1.6 – 5.6) years  102 (33%) deaths in the treatment arm and 138 (44%) in the control arm (p = 0.011) 5 Malmberg. BMJ. 1997;314:1512.

6. Causes of Hyperglycemia in Critically Ill Patients Internal  Metabolic stress  Hormones  Cortisol, catecholamines, glucagon, growth hormone  Insulin Resistance  Demonstrated in > 80% of critically ill patients External  Poor glucose control  Lack of pharmacologic management  Medications  Glucocorticoids  Nutrition  TPN  Fluids 6 Donahey. Pharm Pract News. November 2013.

7. Effects of Hyperglycemia  Has been associated with poor clinical outcomes  Acute kidney injury (AKI)  Sepsis  Critical illness polyneuropathy (CIP)  Respiratory failure  Decreased wound healing  Increased mortality rates  Increased length of stay  Hospital and ICU 7 Donahey. Pharm Pract News. November 2013.

8. Van Den Berghe et al.  2001  Prospective, randomized, controlled study  IIT (80 – 110 mg/dL) versus conventional treatment (180 – 200 mg/dL)  1548 Surgical patients enrolled  12 months 8 Van Den Berghe. N Engl J Med. 2001;345:1359 – 67.

9. Van Den Berghe et al. Conventional (180 – 200 mg/dL) IIT (80 – 110 mg/dL) P-value Mortality during ICU stay (%) 8.0 %4.6 %<0.04 Treatment with antibiotics for >10d (%)17.1 %11.2 %<0.001 Need for RRT (%)8.2 %4.8 %0.007 EMG evidence of CIP (on more than 2 occasions) (%) 18.9 %7.0 %0.001 9 Van Den Berghe. N Engl J Med. 2001;345:1359 – 67.

10. Effects of Moderate Intensity Glycemic Control After Cardiac Surgery  Patients with Diabetes Mellitus or Random BG >150 mg/dL post cardiac surgery 10 Targets (mg/dL) Glucose levels (mg/dL) Infection rate (%) Hypoglycemia rate (%) Control (n = 207) -166 + 27112.5 Intervention (n = 410) 110 - 150151 + 1953.0 P - value-0.00010.0181.0 Leibowitz. Ann Thorac Surg. 2010;90:1825-32.

11. Association Between Hyperglycemia and Increased Hospital Mortality  Heterogeneous ICU patient population 11 Krinsley. Mayo Clin Proc. 2003;78(12):1471-8. Hospital Mortality vs. Mean BG Mean BG (mg/dL) Hospital Mortality (%)

12. Hyperglycemia and Mortality Risk  Retrospective cohort  173 United States Veteran’s Health Administration ICU’s  N = 259,040 admissions from 10/2002 – 9/2005  Hyperglycemia was associated with increased hospital mortality independent of ICU type, length of stay and diabetes  Mortality from hyperglycemia varied based on admission diagnosis 12 Falciglia. Crit Care Med. 2009;37(12)3001-09.

13. Adjusted Odds of Hospital Mortality Based on Hyperglycemia 13 Mean Glucose (mg/dL) Odds ratio w/ 95% CI Falciglia. Crit Care Med. 2009;37(12)3001-09.

14.  Acute Myocardial Infarction  Arrhythmia  Unstable Angina  Pulmonary Embolism  Congestive Heart Failure  Stroke  Ischemic and Hemorrhagic  GI bleed  Acute Renal Failure  Pneumonia  Sepsis 14 Falciglia. Crit Care Med. 2009;37(12)3001-09. Admission Diagnosis Associated with Hyperglycemia and Hospital Mortality

15. Admission Diagnosis NOT Associated with Hyperglycemia and Hospital Mortality  Chronic Obstructive Pulmonary Disease  Hepatic Failure  Gastrointestinal Neoplasm  Post Surgical  Coronary Artery Bypass Graft  Peripheral Vascular Disease  Hip Fracture 15 Falciglia. Crit Care Med. 2009;37(12)3001-09.

16. Audience Participation Hyperglycemia has been shown to increase the rates of: A. Acute Kidney Injury B. Infection C. Critical Illness Polyneuropathy D. Hospital Mortality E. All of the above 16

17. We can’t ignore hyperglycemia, so what do we do about it? 17

18. Audience Participation What BG goal is generally targeted in the ICU’s at your institution? A. 80 - 120 mg/dL B. 120 - 140 mg/dL C. 140 – 180 mg/dL D. < 200 mg/dL 18

19. What do the Guidelines Say? OrganizationTarget (mg/dL) NotesStrength of Recommendation AACE/ADA 2009 140 – 180> 110 mg/dL *110 – 140 mg/dL A ACP 2011 140 - 200Not 80 – 110 mg/dL Weak Moderate quality evidence SCCM 2012 100 – 150< 180 mg/dL‘Very low quality of evidence’ 19 *In some Critically Ill patients – level of evidence C 1.Moghissi. Diab Care. 2009;32(6):1119–31. 2.Diab Care. 2012;35(1):S11-63. 3.Qaseem. Ann Intern Med. 2011;154:260-67. 4.Jacobi. Crit Care Med. 2012;40(12):3251-76.

20. Cardiology  Van Den Berghe et al – 2001  63% post cardiac surgery  Decreased mortality with BG 80 – 110 mg/dL compared to 180 – 200 mg/dL (RRR 42%)  High risk for hypoglycemia  5.1 % vs. 0.8 % had BG < 40 mg/dL  Leibowitz et al - 2010  Post Cardiac Surgery  Intervention group targeted 110 – 150 mg/dL  Decreased infection rates from 11 % to 5 % (p = 0.018)  2.5% to 3% hypoglycemia (p = 1.0) 20 1.Van Den Berghe. N Engl J Med. 2001;345:1359 – 67. 2.Leibowitz. Ann Thorac Surg. 2010;90:1825-32.

21. Neurology Treatment arms Hypoglycemia episodes (BG <80 mg/dL) ICU LOS (days) 6 month survival (% ) Bilotta 2008 N = 97 Post severe TBI 80 – 120 mg/dL 151010.4 < 220 mg/dL 77.312.2 P <0.0001P < 0.05NS Bilotta 2009 N = 493 Brain surgery 80 – 110 mg/dL* 8674 < 215 mg/dL* 3872 P < 0.0001P=.0001NS 21 *Converted from mmol. 1. Bilotta. Neurocrit care. 2008;9(2):159-66. 2. Bilotta. Anesthesiology. 2009;110:611-9.

22. Medical patients  Surviving Sepsis guidelines - 2012  Target BG <180 mg/dL  Van Den Berghe - 2006  Prospective, randomized, controlled study in a medical ICU  N = 1200, intention to treat 22 Overall hospital mortality (%) Hypoglycemia (%) Conventional*403.7 IIT (80 – 110 mg/dL)37.318.7 1. Crit Care Med. 2013;41(2):580-637. 2. Van Den Berghe. N Engl J Med. 2006;354:449-61. * Started insulin infusion when BG > 215 mg/dL and titrated down when BG < 180 mg/dL

23. Van Den Berghe – 2006  IIT  Reduced ICU length of stay  Hazard ratio (HR) 1.15 [1.01, 1.32], p = 0.04  Reduced hospital length of stay  HR 1.16 [1.00 1.35] p = 0.05  Reduced duration of mechanical ventilation  HR 1.21 [1.02, 1.44] p = 0.03  Less acute kidney failure  8.9 % to 5.9 %, p = 0.04  Decreased hospital mortality when treated > 3 days  52.5 % to 43.0 %, p = 0.009  But…  Increased rates of hypoglycemia  No mortality benefit  Different nutritional approach  Hard to identify patients > 3 days 23 Van Den Berghe. N Engl J Med. 2006;354:449-61.

24. COIITSS trial - 2010  Multicenter, randomized, 2x2 factorial, open-label trial  IIT (80 – 110 mg/dL) vs. conventional BG control (2004 surviving sepsis guidelines)  All patients with septic shock receiving corticosteroids  n = 509  No significant difference in In-hospital mortality or 90- day mortality  Increased risk of hypoglycemia (BG <40 mg/dL)  72 vs. 44, p <0.001 24 The COIITSS Study Investigators. JAMA. 2010;303(4):341 – 348.

25. Mixed Medical/Surgical Patients  Volume Substitution and Insulin Therapy in Severe Sepsis (VISEP) - 2008  Multicenter, 2x2 factorial trial  Compared IIT (80 – 110 mg/dL) to conventional (180 – 200 mg/dL)  The IIT arm was stopped early due to increased hypoglycemia 25 IITConventionalP – value 28 day mortality (%)24.726.00.74 90 day mortality (%)39.735.40.31 Hypoglycemia (%)174.1<0.001

26. Mixed Medical/Surgical Patients  NICE-SUGAR – 2009  IIT (81 – 108 mg/dL) vs. conventional (< 180 mg/dL)  N = 6104 patients in ICU  Increased mortality and hypoglycemia 26 BG goal (mg/dL) 90-day mortality (%) Hypoglycemia rate BG<40 mg/dL (%) Surgical Subgroup 90 day mortality (%) 81 – 10827.56.824.4 < 18024.90.519.8 P - valueP = 0.02P < 0.001P = 0.10 NICE-SUGAR study investigators. N Engl J Med.2009;360:1283-97.

27. Glucotrol Study  2009, multicenter trial  Medical and surgical patients  IIT (80 – 110 mg/dL) vs. Conventional (140 – 180 mg/dL)  Trial stopped early due to protocol violations  IIT  Increased hypoglycemia (8.7 % vs. 2.7 %, p < 0.0001)  No difference in ICU mortality  Non-significant trend towards increased 28 day and hospital mortality  18.7 % IIT vs. 15.3 % conventional 27 Preiser. Intensive Care Med. 2009;35:1738-48.

28. Blood Glucose Targets StudyControl (mg/dL)IIT (mg/dL) Van Den Berghe 1, 2180 +80 - 110 NICE-SUGAR < 18080 - 110 VISEP180 - 20080 - 110 Glucotrol180 - 20080 - 110 28 OrganizationTarget (mg/dL) NotesStrength of Recommendation AACE/ADA 2009 140 – 180> 110 mg/dL *110 – 140 mg/dL A ACP 2011 140 - 200Not 80 – 110 mg/dL Weak Moderate quality evidence SCCM 2012 100 – 150< 180 mg/dL‘Very low quality of evidence’

29. Recurring Safety Concern is Hypoglycemia… 29

30. Hypoglycemia  Moderate  BG < 70 mg/dL  Severe  BG < 40 mg/dL  Associated with increased morbidity and mortality  Seizures  Brain damage  Depression  Cardiac arrhythmias 30 Donahey. Pharm Pract News. November 2013.

31. NICE-SUGAR  Post-Hoc analysis of 6026 patients  Patients with hypoglycemia had a higher risk of death  Casual? 31 BGHazard Ratio95% CIP - value 41 – 70 mg/dL1.411.21 – 1.62< 0.001 < 40 mg/dL2.101.59 – 2.77< 0.001 The NICE-SUGAR Study Investigators. N Engl J Med. 2012;367:1108-18.

32. Risk Factors for and Outcomes of Hypoglycemia  Retrospective, case-control analysis (1:3)  Define risk factors that increase the risk for severe hypoglycemia (SH) (< 40 mg/dL)  Assess whether a single occurrence increases risk of death  Results  N = 102 patients had SH out of 5,365 medical, surgical, and cardiac admissions  Risk factors (next slide)  Mortality rates for SH group were 55.9 % compared to 39.5 % in control group (p = 0.057) 32 Krinsley. Crit Care Med. 2007;35(10):2262-67.

33. Risk Factors for Developing Hypoglycemia Risk FactorOdds Ratio (95 % CI)P value Diabetes3.07 (2.03 – 4.63)< 0.0001 Septic Shock2.03 (1.19 – 3.48)0.0096 Mechanical Ventilation 2.11 (1.28 – 3.48)0.0032 Higher APACHE II score 1.07 (1.05 – 1.10)< 0.0001 33 Krinsley. Crit Care Med. 2007;35(10):2262-67.

34. Hypoglycemia and ICU Mortality  Hermanides et al. - 2010  Retrospective database cohort study in a medical/surgical ICU; N = 5961  Increased risk for ICU death up to cutoff BG of 85 mg/dL 34 Hermanides. Crit Care Med. 2010;38(6):1430-34

35. Treating Hypoglycemia  If we can’t avoid it, and it causes harm…we need to know how to treat it  Prevention  Decrease un – planned nutrition interruptions  Be careful with renal, and hepatic dysfunction  Early treatment and recognition 35

36. Treatment  Avoid hyperglycemia…  Avoid hypoglycemia…  How?? 36

37. Decrease Glycemic Variability?  Egi et al. - 2006  Retrospective chart review of 7,049 critically ill patients  Average of 4.2 hourly glucose measurements  Mean + SD of BG  30 + 22 mg/dL in survivors and 40 + 27 mg/dL in non- survivors  Mean and SD were significantly associated with both ICU and hospital mortality (P < 0.001 for both) 37 Egi. Anesthesiology. 2006;105:244-52

38. Date of download: 9/16/2015 Copyright © 2015 American Society of Anesthesiologists. All rights reserved. From: Variability of BG Concentration and Short-term Mortality in Critically Ill Patients Anesthesiology. 2006;105(2):244-252. From: Variability of BG Concentration and Short-term Mortality in Critically Ill Patients Anesthesiology. 2006;105(2):244-252. 38

39. Glycemic Variability  2008 – Retrospective review of 3,252 medical/surgical patients 39 Krinsley JS. Crit Care Med. 2008 Nov;36(11):3008-13

40. A Hypoglycemia Protocol that Minimizes Glycemic Variability?  2013 – Retrospective analysis  N = 772  Nursing driven hypoglycemia protocol  BG < 70 mg/dL  give varying amounts of dextrose 50 %  Less Glycemic variability (GV) than giving full 50 grams  BG rechecked every 15 minutes 40 BG (mg/dL)< 1515 - 2526 - 3536 - 4546 - 6060 - 70 Grams of D50W252017.512.5107.5 Arnold. J Intensive Care Med. 2015;30(3):156-60.

41. Arnold et al. continued Pre-ProtocolPost-ProtocolP-value Coefficient of GV (%)49.340.9.048 Amount of D50W (grams) 21.2*11.5<.001 Degree of BG overcorrection (%) 86.354.5.009 Time to repeat BG (minutes) 6136.003 ICU mortality (%)2522.6NS 41 *Pre-protocol patients generally received 12.5 or 25 grams of D50. Arnold. J Intensive Care Med. 2015;30(3):156-60.

42. Easier Said Than Done…  Avoid hyperglycemia, hypoglycemia, and GV?  How?  Insulin  Intermittent subcutaneous  Intravenous continuous infusion  When?  Persistently elevated BG  > 2 readings >180 mg/dL 42 Jacobi. Crit Care Med. 2012;40(12):3251-76.

43. Insulin Subcutaneous  Pros  Less time  “Set it and forget it”  More types/dosing options  Cons  Less adjustable  Who?  More stable patients  No nutrition interruptions Continuous Infusion  Pros  Most Physiologic  Short half life  Easy titration  Cons  Increased workload  Who?  Hemodynamically unstable  Edematous  Unpredictable nutrition 43

44. Intravenous Insulin Infusion Protocols  Reach and maintain target BG quickly  Often a bolus is used  Monitor BG hourly initially  In range for 2 – 3 hours; monitor every 2 hours  Adjustment based on  Current BG  Rate of change  Result in minimal hypoglycemia 44 Jacobi. Crit Care Med. 2012;40(12):3251-76.

45. A Protocol Example  Braithwaite et al. described a tabular, dose-defining protocol for intravenous insulin  Critically ill trauma service patients in surgical intensive care unit  N = 27 runs  Mean pre-infusion BG was 230 + 67.9 mg/dL  BG < 140 mg/dL: 100% of the time; median time of 5 hours  BG < 110 mg/dL: 25/27 runs; median time of 11 hours  Hypoglycemia  < 70 mg/dL: 2.4 % of BG measurements  < 50 mg/dL: none 45 Braithwaite. Diab Technol Ther. 2006;8(4):476-88.

46. Transition to Subcutaneous Insulin  When?  Consistent nutrition  Hemodynamically stable  Stable dose or no corticosteroids  Minimal peripheral edema  How?  Basal – bolus schedule  With corrective scale 46 Jacobi. Crit Care Med. 2012;40(12):3251-76.

47. Basal - Bolus  Based on continuous infusion requirements and carbohydrate intake  80 % of past 24 hour infusion requirement  50 % Basal – long or intermediate acting insulin  50 % Bolus – short acting divided into three doses ‘pre-meal’ insulin  Continuous feeds  Basal (intermediate acting q6h)  Corrective scale  Overlap intravenous insulin and subcutaneous insulin for 2 hours 47 Jacobi. Crit Care Med. 2012;40(12):3251-76.

48. How to Convert?  40 vs. 60 vs. 80% of 24 hour requirement?  Schmeltz et al.; N = 75  % of patients with capillary BG monitoring within 80 – 150 mg/dL during 24 hours after conversion  40 % - 58.7%  60% - 44.4%  80% - 67.6%  Hypoglycemia  < 50 mg/dL: 1 patient in 40 % group  < 70 mg/dL: 8 incidences (2% of total) 48 Schmeltz. Endoc Pract. 2006;12(6):641-649.

49. Patient Case In the last 24 hours MM has required 30 units of insulin via intravenous infusion. The team wants to convert MM to subcutaneous insulin therapy as her clinical status is improving. What dose of basal/bolus would you recommend in addition to a corrective scale? A. 15 basal and 5 bolus TID pre-meal B. 12 basal and 12 bolus TID pre-meal C. 12 basal and 5 bolus TID pre-meal D. 12 basal and 4 bolus TID pre-meal 49

50. BG Monitoring  POC vs. Arterial sampling  Variability has been shown  Convenience  Time  Acceptable error varies  FDA – 20 %  ADA – 5 %  Clinical and Laboratory Standards Institute and International Organization for Standardization  +15 mg/dL for BG < 75 mg/dL  20 % for BG > 75 mg/dL 50 Jacobi. Crit Care Med. 2012;40(12):3251-76.

51. BG Monitoring  POC meter variability  Low hematocrit  Glucose Oxidase based assay  Elevated PO 2  Drugs  Uric acid  Billirubin  Glucose Dehydrogenase based assay  Maltose containing medications  Q1 - 2 hour testing…unrecognized hypoglycemia? 51

52. Continuous Glucose Monitoring  Can we decrease this workload?  Boom et al – 2014 in the Netherlands  N = 78 per group were analyzed 52 *BG levels expressed in mmol in study and converted for purpose of presentation CGMPOCP - value Severe Hypoglycemia detected by CGM (<40 mg/dL*) 7(3/4)0- Time BG in range (90 – 160 mg/dL*) (%) 69660.47 Nursing time (minutes)1736<.001 Cost (Euros)40.7452.89.02 Boom. Crit Care. 2014;18(4):453.

53. Audience Participation Which of the following are used in the ICU(s) at your hospital(s) to control patients’ BG? A. Continuous Glucose Monitoring B. Insulin Infusion Protocol C. Hypoglycemia Protocol D. B and C E. All of the above 53

54. In Summary  Yes, hyperglycemia is still a concern  Exact glycemic goal is still unknown  May depend on population  What we do know  Minimize hyperglycemia (BG > 180 mg/dL)  Minimize hypoglycemia (BG < 40 mg/dL)  Minimize GV  Monitor as closely as possible  Future studies  Define more specific BG goals  Compare GV to mean BG level 54

55. Hyperglycemia: Is This Still a Concern? Lauren E. Healy BA, PharmD, BCPS NYSCHP Downstate: Critical Care Program October 2, 2015