1. A Patient Safety Initiative For Insulin Pumps Standards and recommended practices to improve insulin pump use and medical outcomes These proposals are not final. They will undergo a single editorial review before being distributed for wider signatures. Your editorial comments are needed at this time. All suggestions you have for improvements, additions, or deletions are welcomed and will be attributed. Send comments to: John Walsh, PA, CDE at jwalsh@diabetesnet.com or call (619) 497-0900jwalsh@diabetesnet.com

2. Introduction Over 500,000 insulin pumps are in use around the world, yet no formal guidelines regarding pump settings, safety features, and use have been adopted by the diabetes clinical community and pump manufacturing industry. The slides present a preliminary set of guidelines for standards and recommended pump practices. Most of these cannot be integrated into current insulin pumps, but we recommend that they be seriously considered for all new pump models.

3. These Standards And Recommended Practices Are Supported By: John Walsh, PA, CDE Ruth Roberts, MA Gary Scheiner, MS, CDE Timothy Bailey, MD, FACE Steve Edelman, MD Your name Etc * Any reservations you have about a standard will be modified or noted

4. Definitions TDD – total daily dose of insulin (all basals and boluses) Basal –background insulin pumped slowly through the day to keep BG flat Bolus – a quick surge of insulin as  Carb boluses to cover carbs  Correction boluses to lower high readings that arise from too little basal insulin delivery or insufficient carb boluses Bolus On Board (BOB) – the units of bolus insulin with glucose-lowering activity still working from recent boluses Duration of Insulin Action (DIA) – time that a bolus will lower the BG. This is used to calculate BOB.

5. Why Insulin Pump Guidelines Are Needed These standards and recommended practices are designed to improve: 1.Consistency of pump settings between pump manufacturers 2.Accuracy and safety of carb and correction factors 3.Safety of DIA defaults and increments 4.Consistency in the handling of BOB 5.Improved monitoring for hypoglycemia & hyperglycemia 6.Faster notification that use of correction boluses is excessive 7.Faster identification of control problems related to infusion sets 8.Consistency in the presentation of data in history screens

6. Benefits Pump users and clinicians will be able to quickly make sound clinical decisions when changing between pumps. Clinicians will be able to make better decisions when managing a variety of pumps Adjunctive medical and accessory personnel (ER, surgical, school nurses, etc.) can more easily interact and be trained in pump use

7. Review What % of The TDD Changes The BG? To understand some slides that follow, it helps to know how much of a change in the TDD creates a significant change in the glucose level. Using a conservative 2000 Rule to set the correction factor, 1.25% of anyone’s TDD is sufficient to change the glucose by 25 mg/dl when given as a single dose. About a 5% change in the TDD is equivalent to a 25 mg/dl reduction in the glucose through the day. A change in the carb factor of 5 to 6% (about 2.5% to 3% of the TDD) is sufficient to change the glucose by about 20 mg/dl per meal.

8. Overview #Topic 1Carb Factor Increments 2Correction Factor Increments 3Personal Carb Factors 4Personal Corr Factors 5DIA Settings 6DIA Time Increments 7Severe/Frequent Hypoglycemia #Topic 8.Severe/Frequent Hyperglycemia 9.Correction Bolus Alert 10.Insulin Stacking Alert 11.Automatic BG Entry 12.BOB Handling 13.Linear And Curvilinear DIA 14.Infusion Sets These slides are numbered by topic for easy reference.

9. Issue: Safety is compromised for many pump users because current carb factors lack the precision required to avoid excess hyperglycemia and hypoglycemia. Current carb factor increments are too large to provide accurate carb boluses, especially for those who use smaller carb factors. 1 Carb Factor (CarbF) Increments

10. Most pumps offer 1 gram per unit as their smallest CarbF increment. This increment becomes relatively large for CarbFs below 15 or 20 g/u. For instance, when the carb factor is reduced from 10 to 9 g/u, all subsequent carb boluses are increased by 11.1%. A shift in the carb factor from 1u/5g to 1u/4g causes each subsequent carb bolus to increase by 25%. For most pump users, a change in the a carb factor larger than 5 or 6% would be expected to create more than a 20 mg/dl shift in the glucose following each meal. 1 Example Carb Factor Increments

11. Example Impact On BG From CarbF Adjustments This table shows the average additional fall in glucose after each meal of the day when a carb factor is reduced from 10 grams per unit to 9 grams per unit (for appropriate weight & TDD), and from 5 gr per unit to 4 gr per unit. How A 1-Step Reduction In Carb Factor Impacts Avg. Meal BG Change in CarbF Weight (~TDD) Avg Carb/day Avg carb gms/meal Change per meal Change in BG per meal* 1/10 to 1/9 160 lb (~40 u) 220 gr73 gr + 1.47 u (CorrF = 50) - 73.3 mg/dl 1/5 to 1/4 240 lb (~100 u) 330 gr110 gr + 5.5 u (CorrF = 20) - 110 mg/dl 1 * Calculated as avg. carbs/day – avg. carbs/day X 1 X 2000 new carb factor old carb factor 3 TDD

12. Review Median Carb Factor In unpublished data from the Cozmo Data Analysis Study: The median (middle) carb factor was 11.2 g/u Almost all pumpers used carb factors below 20 g/u 40% or more use carb factors of 10 g/u. 1

13. What Current Changes In CarbFs Do Table shows how subsequent carb boluses are affected by a one-step reduction in the CarbF using different CarbF increments. Yellow area shows values for most current pumps. Green areas show safer increments that impact subsequent boluses less than 5%. 1

14. We recommend that carb factor increments be smaller than 5% of the next larger whole number so that each single step adjustment causes subsequent carb boluses to change by no more than 5% from previous doses. We recommend minimum carb factor increments of: 5 1.0 g/u above 20 g/u 0.5 g/u for 10 to 20 g/u 0.2 g/u for 5 to 9.8 g/u 0.1 g/u for 3 to 4.9 g/u 0.05 g/u for 0.1 to 2.95 g/u Standard For: Carb Factor Increments 1 5 Improved carb factor increments recommended by Gary Scheiner, MS, CDE

15. For similar reasons, we also recommend that correction factor increments be smaller than 5% of the next larger whole number in order that a single step adjustment causes subsequent correction boluses to change by no more than 5% from previous doses. We recommend minimum correction factor increments of: 5.0 mg/dl per uabove 80 mg/dl per u 2.0 mg/dl per ufor 40 to 78 mg/dl per u 1.0 mg/dl per u for 20 to 39 mg/dl per u 0.5 mg/dl per u for 10 to 19.5 mg/dl per u 0.2 mg/dl per u for 5 to 9.8 mg/dl per u 0.1 mg/dl per u for 3 to 4.9 mg/dl per u 0.05 mg/dl per u for 0.1 to 2.95 mg/dl per u Standard For: Correction Factor Increments 2

16. Personal Carb Factors Issue: Many carb factors used in insulin pumps today are poorly tuned to the user’s need. When a carb factor does not match an individual’s need, this will significantly magnify other sources of error in carb bolus calculations. 3

17. Review Carb Factors In Use 1 Avg. carb factors* for 468 consecutive Cozmo insulin pump downloads (>126,000 boluses) are shown in blue Note that they are NOT bell-shaped or physiologic People prefer “magic” numbers – 7, 10, 15, and 20 g/unit – for their carb factors * Determined directly from grams of carb divided by carb bolus units for each carb bolus 7 10 115 20 1 3

18. Review Carb Factors In Use 1 MANY magic carb factors, shown in blue, are inaccurate. A more normal or physiologic distribution is shown in green Use of magic numbers creates major, consistent bolus errors that magnify other sources for error 7 10 115 20 1 3

19. 1.To encourage use of accurate carb factors, we request that research studies be undertaken to determine what range of carb factor rule numbers (CarbF x TDD) provide optimal glucose results (lower eAG and a minimum of hypoglycemia). These rule numbers may vary for various TDD ranges and for the percentage of carbs in the diet. Research results will be published as a reference for clinicians. Recommended Practice For: Personal Carb Factors 3

20. 3.To ensure that carb factor selection is being individual- ized, we ask that insulin pump companies measure and publish each year carb factors from a sufficient number of random downloads of pumps which use carb factors. 4.We recommend that tools be adopted to improve the accuracy of carb factors, specifically automated carb factor recommendations based on the user’s TDD and percentage of carbs in their diet. 5.We recommend that efforts be undertaken to automate carb factor testing. Recommended Practice For: Personal Carb Factors 3

21. To assist users in setting accurate CarbFs, insulin pumps should allow the user to compare their current CarbF to a range of best practices settings derived from CarbF Rule Numbers* gained from pumps with similar TDDs of users whose avg. BGs are < 180 mg/dl without frequent or severe hypoglycemia: 3 Example Sample Carb Factor Setting Guide Sample Carb Factor Rule # Ranges For Carb Factor Guidance ** Carb Factor Rule Number Range Avg. TDDMore aggressiveLess aggressive 40 u or less400550 40 to 80 u425600 Over 80 u500650 * Carb Factor Rule Numbers = carb factor x TDD ** Optimal ranges would be determined from research studies of best practices

22. Personal Correction Factors Issue: Many correction factors used in insulin pumps today are poorly tuned to the user’s need. When a correction factor does not match an individual’s need, this will significantly magnify other sources of error in correction bolus calculations. 4

23. Review Correction Factors In Use 1 Avg. correction factors in use for 452 consecutive Cozmo insulin pump downloads Like carb factors, correction factors in use are NOT bell-shaped or physiologic Users or clinicians appear to frequently select “magic” numbers for correction factors. 7 10 115 20 1 4

24. 1.To encourage use of accurate correction factors, we request that research studies be undertaken to determine what range of correction factor rule numbers (CorrF x TDD) provide optimal glucose results (lower eAG and a minimum of hypoglycemia). Rule numbers may vary for various TDD ranges. Research results will be published as a reference for clinicians. 2.To ensure that correction factor selection is being individualized, we request that insulin pump companies measure and publish each year correction factors from a sufficient number of random downloads of pumps which use correction factors for corrections. Recommended Practice For: Personal Correction Factors 4

25. 3.We recommend that tools be adopted to improve the accuracy of correction factors, specifically automated correction factor recommendations based on the user’s TDD. 4.We recommend that efforts be undertaken to automate correction factor testing. Recommended Practice For: Personal Correction Factors 4

26. To assist users in setting accurate CorrFs, insulin pumps should allow the user to compare their current CorrF to a range of best practices settings derived from CorrF Rule Numbers* derived from pumps with similar TDDs of users whose avg. BGs are < 180 mg/dl without frequent or severe hypoglycemia: 4 Example Sample Correction Factor Setting Guide Sample Corr. Factor Rule # Ranges For Corr. Factor Guidance** Correction Factor Rule Number Range Avg. TDDMore aggressiveLess aggressive 40 u or less17002000 40 to 80 u18002200 Over 80 u18002400 * Correction Factor Rule Numbers = corr factor x TDD ** Optimal ranges would be determined from research studies of best practices

27. DIA Settings Issue: Duration of insulin action measures the glucose-lowering activity of a carb or correction bolus over time. In research studies, inter- individual variation in the pharmaco-dynamic time for rapid insulins is generally about 25%. However, the DIA times recommended by clinicians vary widely from 2 to 6 hours or more. 5

28. Review How Long Do Boluses Lower The BG?  One insulin manufacturer states their insulin’s pharmacodynamics is 3 to 5 hours 10, but numerous studies show rapid insulin lowers the glucose for 5 hours or more.  With Novolog (aspart) at 0.2 u/kg (0.091 u/lb), 23% of glucose lowering activity remained after 4 hours. 12  Another study found Novolog (0.2 u/kg) lowered the glucose for 5 hours and 43 min. +/- 1 hour. 13  After 0.3 u/kg or 0.136 u/lb of Humalog (lispro), peak glucose- lowering activity was seen at 2.4 hours and 30% of activity remained after 4 hours. 11 These times would be longer if the unmeasured basal suppression in pharmacodynamic studies were accounted for. 10 Novolog product labeling information, October 21, 2005. 11 From Table 1 in Humalog Mix50/50 product information, PA 6872AMP, Eli Lilly and Company, issued January 15, 2007. 12 Mudaliar S, et al: Insulin aspart (B28 Asp-insulin): a fast-acting analog of human insulin. Diabetes Care 1999; 22:1501-1506. 13 L Heinemann, et al: Time-action profile of the insulin analogue B28Asp. Diabetic Med 1996;13:683-684. 5

29. A short DIA time hides true BOB level and its glucose- lowering activity. This: Leads to “unexplained” lows Leads to incorrect adjustments in basal rates, carb factors, and correction factors Causes user to start ignoring their “smart” pump’s advice Inappropriately long DIAs overestimate bolus insulin activity. DIA should be based on an insulin’s real action time. Do NOT modify the DIA time to fix a control problem Review Short DIAs Hide Bolus Insulin Activity 5

30. Review Duration Of Insulin Action (DIA) 4 hrs 6 hrs 2 hrs 0 Glucose-lowering Activity Accurate bolus estimates require an accurate DIA. DIA times shorter than 4 to 7 hrs will hide BOB and its glucose lowering activity 5

31. Review DIA Large doses (0.3 u/kg = 30 u for 220 lb. person) of “rapid” insulin in 18 non- diabetic, obese people Med. doses (0.2 u/kg = 20 u for 220 lb. person) Apidra product handout, Rev. April 2004a Regular 5

32. Review Does Dose Size Affect DIA? This graphic suggests that smaller boluses do not lower the BG as long as larger boluses. However, this may not be true – see next 2 slides. Size of the injected Humalog dose for a 154 lb or 70 kg person: 0.05 u/kg= 3.5 u 0.1 u/kg = 7 u 0.2 u/kg = 14 u 0.3 u/kg = 21 u Woodworth et al. Diabetes. 1993;42(Suppl. 1):54A 5

33. Review Pharmacodynamics Is Not DIA The DIA time entered into an insulin pump is based on studies of insulin pharmacodynamics. However, the traditional method used to determine the pharmacodynamics of insulin tends to underestimate insulin’s true duration of action. See the next slide. 5

34. Review Pharmacodynamics Underestimates DIA And Overestimates Impact Of Bolus Size To measure pharmaco- dynamics, glucose clamp studies are done in healthy individuals (0.05 to 0.3 u/kg) Because there is no basal suppression, this injected insulin ALSO SUPPRESSES normal basal release from the pancreas (grey area in figure) Basal suppression also makes smaller boluses appear to have a shorter DIA 5

35. Review Pharmacodynamic Time Does Not Equal DIA After accounting for the lack of basal suppression, True DIA times become longer than the pharmacodynamic times derived from typical research At least some of the apparent variation in DIA related to bolus size may also disappear Some of the apparent inter- individual variation in pharmacodynamics may also disappear 5

36. Standards For: Duration Of Insulin Action 1.We recommend that a panel of experts in insulin action review existing pharmacodynamic studies, consider differences between pharmacodynamics and the DIA time used in pumps, and recommend guidelines for acceptable ranges for DIA times in pumps for children and adults. 2.Default DIA times in current pumps vary widely between 3 and 6 hours. We recommend that guidelines be set for consistent and accurate default times in pumps that use different methods to measure DIA. 5

37. Issue: Current DIA time increments vary from 15 minutes to 1 hour in different pumps When a DIA time is changed in a pump, a larger time increment, such as 1 hr, can introduce an excessive change in subsequent estimates of BOB. For example, when the DIA is reduced from 5 hours to 4 hours, subsequent BOB estimates are decreased and recommendations for carb boluses are increased by 25%. 6 DIA Time Increments

38. Most GIR studies suggest that pharmacodynamic action of insulin varies only about 25% between individuals. For a DIA time of 5 hr and 15 min, a 25% range is equivalent to 1 hr and 20 min, such as from 4 hrs and 30 min to 5hr and 50 min. A pump that has only 1 hr DIA increments would enable the user to select only 1 or 2 setting within this physiologic range, while a 30 min increment would allow only 2 or 3 choices that are close to a physiologic range. 6 Review Glucose Infusion Rate (GIR) Studies

39. For safe and accurate estimates of residual BOB, we recommend that DIA time increments be no greater than 15 minutes. Standard For: DIA Time Increments 6

40. Frequent/Severe Hypoglycemia Alert Issue: Although most current insulin pumps contain sufficient data to do so, pumps (and glucose monitors) do not warn users when the user is experiencing patterns of excessively frequent or severe hypoglycemia. 7

41. Recommended Practice For: Hypoglycemia Alert 1.We recommend that insulin pumps which store glucose and insulin dosing data have this data easily accessible on the pump and that the user be alerted when they experience frequent* or severe* hypoglycemia. 2.We recommend that specific guidance be provided regarding likely causes based on the pattern of hypoglycemia and the user’s current settings. 7 * Adjustable setting in pump/controller

42. Example Pump Screen Hypoglycemia Display 7 * Adjustable setting in pump/controller Overview of Low BGs # of weeks1248 # BGs/week23252832 # BGs/week < 50 mg/dl* 54.53.12.7 # BGs/week < 70 mg/dl* 86.75.25.1

43. Frequent/Severe Hyperglycemia Issue: Although most current insulin pumps contain sufficient data to do so, pumps (and glucose monitors) do not warn users when the user is experiencing patterns of frequent or severe hyperglycemia. 8

44. Recommended Practice For: Hyperglycemia Alert 1.We recommend that insulin pumps which store glucose and insulin dosing data have this data easily accessible on the pump and that the user be alerted when they experience frequent* or severe* hyperglycemia. 2.We also recommend that specific guidance be provided regarding likely causes based on the pattern of hyperglycemia and the user’s current settings. 8 * Adjustable setting in pump/controller

45. Example Pump Screen Hyperglycemia Display 8 Overview of High BGs # of weeks1248 # BGs/week23252832 # BGs/week < 50 mg/dl* 67.58.08.4 # BGs/week < 70 mg/dl* 44.56.17.3 * Adjustable setting in pump/controller

46. Correction Bolus Alert Issue: Hyperglycemia is more common than hypoglycemia for most people on insulin pumps. When glucose levels consistently run high, many pump users address the problem by giving frequent correction boluses rather than correcting the core problem through an increase in their basal rates or carb boluses. If the correction bolus % becomes excessive relative to the TDD, this information is often not shown and no alert is given regarding the excess. 9

47. Recommended Practice For: Correction Bolus Alert 1.We recommend that insulin pump companies show in an easily available history what percentage of the TDD is used for correction boluses. 2.We recommend that the pump wearer be able to request that the pump alert them when they are using more than 8% (adjustable) of their TDD for correction bolus doses over a 4 day period (adjustable). 3.We recommend that once an excess in correction bolus is identified, that the user be provided with guidelines in how to safely distribute any excess in correction boluses into carb boluses or basal rates. 9

48. Insulin Stacking Alert Issue: Pump users often bolus for carbs without checking their glucose first. Without a glucose reading, the pump cannot account for BOB, and the bolus is not appropriately adjusted for the BOB or the current BG. However, data in the pump is available at the time the bolus is given whether the BOB is sufficiently large to substantially change a bolus dose. The pump can alert the user to this otherwise unseen insulin stacking. 10

49. Recommended Practice For: Insulin Stacking Alert We recommend that insulin pumps alert* the wearer when there is sufficient insulin stacking to introduce a significant error in a current bolus (such as BOB greater than 1.25%* of TDD). The alert would be on by default once a DIA time is selected to measure BOB, but may be turned off if the user desires. * Adjustable setting in pump/controller for an expected mg/dl drop in glucose with visible, audio, or vibratory output. 10

50. When a carb bolus is planned without a recent BG check, but BOB is more than 1.25% of the average TDD (enough to cause about a 25 mg/dl drop in the glucose), the pump will recommend that the wearer do a BG check due to the substantial presence of BOB. For instance, for someone with: Avg TDD1.25%* of TDD 40 units 0.5 units 50 mg/dl per u (corr factor) X 0.5 u = 25 mg/dl This individual would be alerted when they do not check their glucose and want to give a bolus but have 0.5 u or more of BOB present. Example Insulin Stacking or BOB Alert * Adjustable in pump/controller for a reasonable degree of safety 10

51. Automatic Entry Of BG Values Issue: Pump users often do not enter BG values into their pump if they must do it manually. When the BG is checked but data is not entered, the data becomes unavailable in the pump’s core dataset that is used for analysis of glucose patterns and frequency of hypoglycemia, and the relatationship of these to insulin doses. 11

52. Review Automatic Entry Of BG Values In a study of over 500 insulin pumps where BG values could be entered either manually or automatically, users entered 2.6 BG values per day when the glucose had to be entered manually, compared to 4.1 values per day for pumps where the glucose value was automatically entered. * BOB could typically be taken into account for 1.5 additional boluses per day with automatic entry of BG values. When a meter does not automatically enter glucose values into their pump, BG values are more often not entered for determination of BOB and bolus recommendations. 11 * J. Walsh, D. Wroblewski, T.S. Bailey: unpublished data

53. Review Automatic Entry Of BG Values Pumps without automatic entry of glucose values are at a significant disadvantage to users because boluses are less likely to be adjusted for high or low BGs, and for residual BOB in the bolus calculation. 11

54. Standard For: Automatic Entry Of BG Values Due to the benefits (more complete BG history and improved handling of BOB) and the significant decrease in glucose entries seen when BGs are entered manually, we recommend that all pumps be enabled to have wireless or direct entry of BG test results from two or more major brands of meters (to better ensure insurance coverage of strips). 11

55. Issue: Current pumps differ significantly in what is counted as BOB and in whether or not BOB is subtracted from subsequent carb boluses. Most insulin pumps assume that excess BOB does not need to be taken into account when determining the next carb bolus. Though commonly determined in this way, the resulting bolus dose recommendations can cause unexplained and unnecessary insulin stacking and hypoglycemia. BOB Handling 12

56. Example Insulin Stacking With a bedtime BG of 173 mg/dl, is there an insulin deficit or a carb deficit? 6 pm8 pm10 pm12 am Dinner Dessert Correction Bedtime BG = 173 mg/dl 12

57. Review Insulin Stacking Is Common CDA1 Study Results Of 201,538 boluses, 64.8% were given within 4.5 hrs of a previous bolus An accurate DIA shows that some BOB is present for MOST boluses Note that 4.5 hours may underestimate true DIA 4.5 hrs 12

58. Review Bolus On Board (BOB) An accurate measurement of the glucose-lowering activity that remains from recent boluses helps: Prevent insulin stacking Improve bolus accuracy Allows the current carb or insulin deficit to be determined aka: insulin on board, active insulin, unused insulin* * Introduced as Unused Insulin in 1st ed of Pumping Insulin (1989) 12

59. Review How Current Pumps Handle BOB What’s In the BOB & What Is It Applied Against? BOB Includes This Type Of Bolus BOB Is Subtracted From This Type Of Bolus CarbCorrectionCarbCorrection Animas 2020Yes No*Yes Deltec CozmoYes Insulet OmnipodNoYesNoYes Medtronic ParadigmYes No Yes * Except when BG is below target BG * “Yes” is generally safer 12

60. Example Differences In Bolus Recommendations Situation: BOB = 3.0 u and 30 gr. of carb will be eaten at these glucose levels Carb factor = 1u / 10 gr Corr. Factor = 1 u / 40 mg/dl over 100 Target BG = 100 TDD = ~50 u units mg/dl Omnipod bolus cannot be determined - it counts only correction bolus insulin as BOB The graphic shows how widely bolus recommendations vary from one pump to another for the same situation. 12

61. Review Track BOB Or Carb Digestion? For safety after meal and correction boluses, BOB is more important to track than the digestion of meal carbs: When a BG is taken after a meal, the BOB times the correction factor ideally represents the maximum fall in glucose expected. Accounting for the impact of the BOB on the current glucose provides the safest approach to work from Even when a BG is high, warning that carbs may be needed later alerts the user that they may need to take preventive action. Low glycemic index meals, gastroparesis, Symlin, and other issues may counteract a predicted fall in glucose based on BOB, but the user can more easily judge this situation than not being made aware that they may have an excess of BOB. 12

62. Example Unsafe BOB 1 Handling If a pump user gets frustrated with high BGs and they overdose to speed the drop in their BG, or they exercise longer or more intensely than they anticipated, they can acquire a significant excess in BOB. In this situation, most pumps bolus recommend that a bolus be given for all carb intake regardless of how much BOB is present. A subsequent bolus will deliver an excess of insulin if the glucose is not high enough (correction bolus need) to offset the excess BOB. Such a recommendation may introduce an unnecessary risk for hypoglycemia. 1 Pumping Insulin, 1st ed, 1989, Chap 12, pgs 70-73: The Unused Insulin Rule 12

63. Standard For: BOB Handling For safe and accurate BOB measurement, we recommend that: BOB include all carb and correction boluses Residual BOB be subtracted from both carb and correction bolus recommendations delivered within the DIA* * Assumes that the DIA time chosen by the clinician or user is accurate. 12

64. Exceptions To Usual BOB Handling When a second bolus is taken for an unplanned carb intake or a desert that is consumed within 60 minutes or so* of a meal bolus, BOB should not be taken into account for the second bolus because the impact of the first bolus cannot be accurately determined. Given that, it is wise to account for BOB as soon after a meal as possible, such as within 60 to 90 minutes (adjustable), to provide early warning if the bolus given was excessive or inadequate. It is generally safer, though not always more accurate, to account for and apply all BOB in subsequent boluses. 12 * Adjustable setting in pump/controller

65. Linear And Curvilinear DIA Issue: Pump manufacturers use at least 3 different methods (100% curvilinear, 95% of curvilinear, and straight linear) to measure DIA and BOB. When a realistic DIA time is selected, such as 5 hrs, a linear determination of residual BOB is not as accurate as a curvilinear method. 13

66. Linear And Curvilinear DIA Compared Note how values for the 5 hr linear line in red and the 5 hr curvilinear line in black diverge in value at several points along the graph. 5 hr Linear 5 hr 95% Curvilinear From Pumping Insulin, 4th ed., adapted fom Mudaliar et al: Diabetes Care, 22: 1501, 1999 13

67. The Modified Triple-Linear DIA A linear measurement of DIA becomes more accurate through use of a triple-linear measurement for BOB. 13

68. Example Triple-Linear DIA Times A triple-linear line more closely imitates a curvilinear DIA. For a 5 hr DIA*: 1st 10%** – no change Mid 65%** – fall 75%** Last 25%** – fall 25%** (** adjustable as needed in device) 5 hr Triple Linear 13 * % modification suggested by Gary Scheiner, MS, CDE

69. Recommended Practice For: Linear And Curvilinear DIA We recommend that insulin pumps use either a 100% curvilinear or a triple-linear method to improve accuracy and consistency of BOB estimates. 13

70. Infusion Sets Issue: Infusion set design and inadequate site preparation or training may introduce erratic loss of control for a significant number of pump wearers. 14

71. Review Infusion Set Failure Loss of glucose control caused by infusion set problem or total failure can occur. The most common problem arises when a Teflon infusion set comes loose and some of the infused insulin leaks back to the skin surface. The partial loss of insulin may cause unexplained high readings. A complete loss of control typically is seen when an infusion set is pulled out entirely. Selecting the right infusion set plus good site technique, especially taping the infusion line to the skin, can significantly minimize the number of unexplained highs for many pump wearers. 14

72. Tape The Tubing Most problems with infusion sets come from loosening of the Teflon under the skin, not from a complete pullout. A 1” tape placed on the infusion line: Stops tugging on the Teflon catheter under the skin Prevents loosening of the Teflon catheter under the skin Avoids many “unexplained highs” caused when insulin leaks back to the skin surface Reduces skin irritation And prevents many pull outs 14

73. Tape The Tubing Helps prevent Tugging Irritation Bleeding 14

74. No tape! 14 Examples Lack Of Anchoring Of Sets A review of dozens of pictures of infusion sets online and pump manuals finds that anchoring of the infusion line with tape is rarely recommended or practiced.

75. Review Infusion Set Monitor Many pump wearers experience random erratic readings until they change to a different infusion set or start to anchor their infusion lines with tape to stop line tugging. However, there is currently no tool for clinicians or pump users to detect who may be having problems with their infusion sets. 14

76. Tool Infusion Set Monitor Insulin pumps with direct BG entry can identify those who may be having intermittent loss of glucose control secondary to infusion set failure. The pump: Shows the average time and variation in time of use between reservoir loads or use of the priming function. Shows average BGs for each 12 or 24* hour time interval following set changes (indicated by the prime function) over the last 7 set changes or as soon as statistical significance is reached. 14 * Adjustable setting in pump/controller

77. Standards For: Infusion Sets 1.We recommend that all infusion sets be able to show consistent performance of at least 72 hours without skin problems or a loss of glucose control. 2.We recommend that insulin pumps monitor and record in easily accessible history the duration of infusion set usage, along with average glucose values in 12 or 24 hour* segments, between set changes up to the next set change (reported once statistically significant). 14 * Adjustable setting in pump/controller

78. Recommended Practices For: Infusion Sets 3. We recommend that insulin pump manuals and training cover methods to identify and prevent infusion set failure. 4. We recommend that infusion set designs incorporate easy to use methods to anchor infusion lines and minimize tugging of the infusion line near the infusion site, again to minimize unnecessary loss of control due to set failure. 5. We encourage, as much as possible, use of clear tapes and dressings, and clear set hubs on infusion sets so that skin and set problems can be spotted sooner.* 14 * Recommended by Gary Scheiner, MS, CDE