Supplementary Appendix This appendix has been provided by the authors to give readers additional information about their work. Supplement to: Bergenstal RM, Klonoff DC, Garg SK, et al. -based insulin-pump interruption for reduction of hypoglycemia. N Engl J Med 2013;369:224-32. DOI: 10.1056/NEJMoa1303576
Supplementary Appendix This appendix has been provided by the authors to give readers additional information about their work. Supplement to: Bergenstal RM, et al. -Based Insulin-Pump Interruption for Reduction of Hypoglycemia Contents Members of the ASPIRE In-Home Study Group... 2 Description of Multiple Imputations Method... 3 Figure S1. Study Visits.... 4 Figure S2. Example of a Sensor-Augmented Pump System... 5 Figure S3. Sensor Glucose Distributions After Nocturnal 2-h Events... 6 Table S1. Primary and Secondary Outcomes... 7 Table S2. Sensor Glucose Distributions... 8 Table S3. Outcomes by Age... 9 Table S4. Outcomes by Glycated Hemoglobin... 10 Table S5. Outcomes by Duration of Diabetes... 11 Table S6. Device Settings and Behaviors; Glucose Values... 12 1
Members of the ASPIRE In-Home Study Group In addition to the authors, the following investigators participated in the ASPIRE In- Home Study: AMCR Institute, Escondido, CA T. Bailey; Rainier Clinical Research Center, Renton, WA R. Brazg; Texas Diabetes and Endocrinology, Austin, TX L. Casaubon; Palm Beach Diabetes and Endocrine Specialists, West Palm Beach, FL B. Horowitz; Rocky Mountain Diabetes and Osteoporosis Center, Idaho Falls, ID D. Liljenquist; Physicians Research Associates, LLC, Lawrenceville, GA O. Odugbesan; Endocrine Research Solutions, Inc., Roswell, GA J. Reed; Diabetes Institute @ Ohio University, Athens, OH F. Schwartz; Arkansas Diabetes Clinic and Research Center, Little Rock, AR J. Thrasher; Iowa Diabetes and Endocrinology Center (IDEC), Des Moines, IA A. Bhargava; The Naomi Berrie Diabetes Center, Columbia University, New York, NY: R. Goland; Joslin Diabetes Center (Syracuse University), Syracuse, NY R. Weinstock. 2
Description of Multiple Imputations Method The multiple imputations method was adopted to handle missing data in glycated hemoglobin measurements. In each round of imputation, the missing glycated hemoglobin at the end of 3- month study was imputed by a random number which was generated by the MI procedure in SAS 9.2 with imputation regression method. The independent variables in the regression model were age, gender, baseline glycated hemoglobin, body-mass index, and treatment group. The missing glycated hemoglobin data were imputed by yˆ z ˆ, where ŷ is the predicted value from regression, z is a standard normal random variable, and ˆ is the estimated standard deviation of the random residual from the regression model. The imputation was performed five times using the MI procedure and the analysis results were combined to form one inference using the MIANALYZE procedure in SAS 9.2. We also performed analysis on the complete case population where the early withdrawals were excluded. The 95% confidence interval of treatment difference remains the same ([-0.05%, 0.15%]). 3
Figure S1. Study Visits. The study included a 2-week Run-In Phase to establish eligibility for the 3-month Study Phase. Blood samples for glycated hemoglobin determination were drawn at Visits 1, 4, and 7. Reported changes in glycated hemoglobin were based on samples drawn at Visits 4 and 7 Pre-Study Phase (6 weeks) Run-in Phase (2 weeks) 2-week run-in phase may be repeated once if Nocturnal Hypoglycemia requirements not met. Visit 2 + 2 weeks Pump Training Pump Start Visit 3 + 4 weeks Sensor Training Sensor Start Visit 4 + 6 weeks Randomization Glycated Hemoglobin Feature Training for Group Visit 5 + 1 week Post randomization Phone call Group Group Visit 5 + 1 week Post randomization Phone call Study Phase (12 weeks) Visit 6 + 4 weeks Post randomization Phone call Visit 6 + 4 weeks Post randomization Phone call Visit 7 + 12 weeks Post randomization End of Study Glycated Hemoglobin Visit 7 + 12 weeks Post randomization End of Study Glycated Hemoglobin Visit 1 Day 0 Baseline Glycated Hemoglobin Screening Informed Consent 4 4
Figure S2. Example of a Sensor-Augmented Pump System The hardware used in the Group and in the Group was identical. The threshold suspend feature was an addition to the software used to control the pump. Glucose sensor Infusion site Transmitter Insulin pump with glucose display Group Standard pump software plus threshold suspend algorithm Group Standard pump software 5
Figure S3. Sensor Glucose Distributions After Nocturnal 2-hour Events There were at total of 1873 2-hour threshold suspend events that had an event marker in pump uploads, of which 1444 were at night. Not all had SG sensor glucose values at 0, 2, and 4 hours after pump suspension: 1438 had an event marker plus sensor glucose values at 0 hours 1385 had an event marker plus sensor glucose values at 2 hours 1136 had an event marker plus sensor glucose values at 4 hours The percentages of sensor glucose values in various ranges following 2-hour threshold suspend events are shown below for the 1385 events with data at 2 hours (left) and for the 1136 events with data at 4 hours (right). See Figure 3 for mean ± SD sensor glucose values before, during, and after the 1438 2-hour nocturnal threshold suspend events that had an event marker plus sensor glucose values. Sensor Glucose Distribution After 2-Hour Events 100% 2.2% 90% 26.0% 80% Percentage of Values 70% 60% 50% 40% 30% 64.8% >200 mg/dl 70 to 200 mg/dl <70 mg/dl 70.2% 20% 33.1% 10% 0% 2 hours (Time of insulin resumption) 3.9% 4 hours (2 hours after insulin resumption) 6
Table S1. Primary and Secondary Outcomes Randomization, Run-In, and Study Phase glycated hemoglobin and hypoglycemia parameters, mean ± SD. Comparisons for glycated hemoglobin and mean AUC of nocturnal hypoglycemic events are shown in Figure 2A and Figure 2B of the article. (n=121) (n=126) Randomization* 7.26 ± 0.71 7.21 ± 0.77 Glycated hemoglobin, % 3-month 7.24 ± 0.67 7.14 ± 0.77 Change 0.00 ± 0.44-0.04 ± 0.42 AUC of nocturnal hypoglycemic events, mg/dl min Nocturnal hypoglycemia events per patient-week AUC of day and night hypoglycemia events, mg/dl min Day and night hypoglycemia events per patient-week Run-In Phase 1547 ± 2035 1406 ± 1950 Study Phase 980 ± 1200 1568 ± 1995 Run-In Phase 2.4 ± 1.2 2.5 ± 1.5 Study Phase 1.5 ± 1.0 2.2 ± 1.3 Run-In Phase 1153 ± 1647 1094 ± 1557 Study Phase 798 ± 965 1164 ± 1590 Run-In Phase 5.0 ± 2.8 5.1 ± 3.0 Study Phase 3.3 ± 2.0 4.7 ± 2.7 * A randomization glycated hemoglobin value was not obtained from one patient in the Group. Based on values from 116 and 124 patients who had available glycated hemoglobin values in the and Groups, respectively. versus, p<0.001. 7
Table S2. Sensor Glucose Distributions Sensor glucose distributions, nocturnal and combined day and night. Comparisons for the <50, 50 to <60, and 60 to <70 mg/dl ranges are shown in Figure 2C of the article. Sensor Glucose Range, mg/dl Nocturnal Combined Day and Night <50 1.2%* 2.8% 0.9%* 1.9% 50 to <60 1.8%* 3.1% 1.6%* 2.5% 60 to <70 3.0%* 4.1% 2.8%* 3.7% 70 to <180 64.7% 62.2% 63.1% 61.4% 180 to <250 21.3% 19.9% 22.3% 21.0% 250 8.0% 7.9% 9.3% 9.4% *, versus, p<0.001 8
Table S3. Outcomes by Age Glycated hemoglobin and nocturnal hypoglycemia event AUC outcomes for patients of different age categories, mean ± SD. Glycated Hemoglobin, % AUC of nocturnal hypoglycemia events, mg/dl min (n=12) Randomization 7.56 ± 0.97 3-month 7.41 ± 0.76 Study Phase 1439 ± 1711 Age 16-24 Age 25-50 Age 51-70 (n=14) 7.64 ± 0.67 7.41 ± 0.89 1921 ± 2625 (n=79) 7.20 ± 0.71 7.23 ± 0.68 925 ± 1117 (n=60) 7.07 ± 0.73 7.02 ± 0.70 1528 ± 1912 (n=30) 7.28 ± 0.60 7.19 ± 0.60 891 ± 1031 (n=52) 7.25 ± 0.82 7.20 ± 0.81 1531 ± 1911 9
Table S4. Outcomes by Glycated Hemoglobin Glycated hemoglobin and nocturnal hypoglycemia event AUC outcomes for patients of different glycated hemoglobin levels at randomization, mean ± SD Glycated Hemoglobin, % AUC of nocturnal hypoglycemia events, mg/dl min Randomization Glycated Hemoglobin 7% (n=50) (n=58) Randomization Glycated Hemoglobin >7% (n=70) (n=68) Randomization 6.61 ± 0.31 6.57 ± 0.34 7.71 ± 0.55 7.75 ± 0.61 3-month 6.75 ± 0.37 6.57 ± 0.38 7.61 ± 0.60 7.64 ± 0.68 Study Phase 1055 ± 1314 1659 ± 2049 898 ± 1063 1438 ± 1908 10
Table S5. Outcomes by Duration of Diabetes Glycated hemoglobin and nocturnal hypoglycemia event AUC outcomes for patients of different diabetes durations, mean ± SD Glycated Hemoglobin, % AUC of nocturnal hypoglycemia events, mg/dl min Duration of Diabetes 25 years (n=57) (n=56) Duration of Diabetes >25 years (n=63) (n=67) Randomization 7.33 ± 0.79 7.40 ± 0.86 7.20 ± 0.64 7.05 ± 0.68 3-month 7.34 ± 0.71 7.21 ± 0.83 7.16 ± 0.61 7.09 ± 0.74 Study Phase 1076 ± 1314 1577 ± 2160 895 ± 1083 1568 ± 1888 11
Table S6. Device Settings and Behaviors; Glucose Values Insulin Delivery, pump and CGM settings and usage, meter blood glucose and calibration frequencies, and glucose values, Study Phase. Insulin Delivery* Total Insulin Dose, U 47.8 ± 19.40 46.5 ± 21.66 Basal Insulin, U 22.7 ± 9.92 23.2 ± 11.07 Bolus Insulin, U 25.1 ± 13.05 23.4 ± 13.91 Bolus Delivery 6.7 ± 1.87 6.2 ± 2.03 Pump and CGM Settings and Usage* Sensor Wear (%) 88.1 ± 8.66 85.5 ± 13.80 Sensor Life (days) 4.8 ± 1.83 4.8 ± 1.84 Bolus Calculator Use 5.6 ± 2.22 4.9 ± 2.29 Meter Blood Glucose Frequency* Study Phase (first 2 weeks) 6.2 ± 2.35 5.9 ± 2.39 End of Study Phase (last 2 weeks) 5.6 ± 2.24 5.3 ± 2.03 Calibration Frequency * Study Phase (first 2 weeks) 4.3 ± 1.24 4.1 ± 1.21 End of Study Phase (last 2 weeks) 4.1 ± 1.10 4.0 ± 1.10 Meter Blood Glucose, mg/dl Run-In Phase 151.4 ± 24.3 151.8 ± 23.6 Study Phase (first 2 weeks) 160.6 ± 24.9 156.7 ± 26.9 End of Study Phase (last 2 weeks) 167.5 ± 24.7 163.9 ± 32.1 Sensor Glucose, mg/dl Run-In Phase 145.5 ± 20.4 146.4 ± 21.0 Study Phase (first 2 weeks) 153.2 ± 20.1 148.1 ± 20.7 End of Study Phase (last 2 weeks) 158.0 ± 20.3 153.8 ± 24.9 Weight, kg Beginning of Study Phase 79.6 ± 15.91 79.1 ± 15.13 End of Study Phase 80.7 ± 17.22 79.7 ± 15.84 Results are given as mean per patient-day (*) or mean of mean per patient per 2 week interval ( ) 12