PREVENTION OF NOCTURNAL HYPOGLYCEMIA USING PREDICTIVE LOW GLUCOSE SUSPEND (PLGS)

Similar documents
What is a CGM? (Continuous Glucose Monitor) The Bionic Pancreas Is Coming

Diabetes Management with Continuous Glucose Monitoring & Multiple Daily Injections. Aaron Michels MD

Diabetes Management: Current High Tech Innovations

Reduction in Hypoglycemia and No Increase in A1C with Threshold-Based Sensor-Augmented Pump (SAP) Insulin Suspension: ASPIRE In-Home

Insulin Pumps and Glucose Sensors in Diabetes Management

The In-Clinic Close Loop Experience in the US

Update on Continuous Glucose Monitoring (CGM) Technology in Diabetes. Elena Toschi, MD November 12, 2016

Artificial Pancreas Device Systems. Populations Interventions Comparators Outcomes. pump. pump

Presented by Dr. Bruce Perkins, MD MPH Dr. Michael Riddell, PhD

Artificial Pancreas Device Systems. Populations Interventions Comparators Outcomes Individuals: With type 1 diabetes

Advances in Technology in the Treatment of Diabetes Mellitus 2017 How far have we come-how far are we going? Is there a final frontier?

Diabetes Management in Toddlers/Preschoolers (<6 yo)

Paolo Di Bartolo U.O di Diabetologia Dip. Malattie Digestive & Metaboliche AULS Prov. di Ravenna. Ipoglicemie e Monitoraggio Glicemico

CAROLINAS CHAPTER/AMERICAN ASSOCIATION OF CLINICAL ENDOCRINOLOGISTS Annual Meeting HILTON HEAD ISLAND FRIDAY PRESENTATION

Supplementary Appendix

Artificial Pancreas Device Systems

Control of Glycemic Variability for Reducing Hypoglycemia Jae Hyeon Kim

Medical Policy. MP Artificial Pancreas Device Systems

Sensor-augmented pump systems provide insulin delivery

Comparing the use of SMBG vs. CGM data to Optimize Glucose Control in T2DM

How I use the latest technology to investigate and treat a person with Type 1 Diabetes. Dr Pratik Choudhary

Preventing Hypoglycemia Using Predictive Alarm Algorithms and Insulin Pump Suspension

Clinical Value and Evidence of Continuous Glucose Monitoring

Inpatient Studies of a Kalman-Filter-Based Predictive Pump Shutoff Algorithm

Insulin Pump Therapy in children. Prof. Abdulmoein Al-Agha, FRCPCH(UK)

What is the role of insulin pumps in the modern day care of patients with Type 1 diabetes?

Artificial Pancreas Device System (APDS)

Hypoglycemia a barrier to normoglycemia Are long acting analogues and pumps the answer to the barrier??

Advances Towards the Bionic Pancreas.

The Diamond Study: Continuous Glucose Monitoring In Patients on Mulitple Daily Insulin Injections

Medical Policy An independent licensee of the Blue Cross Blue Shield Association

More Than 1 Year of Hybrid Closed Loop in Pediatrics. Gregory P. Forlenza, MD Assistant Professor Barbara Davis Center

Diabetes II Insulin pumps; Continuous glucose monitoring system (CGMS) Ernest Asamoah, MD FACE FACP FRCP (Lond)

Medical Policy An independent licensee of the Blue Cross Blue Shield Association

Effective Health Care Program

Continuous or Intermittent Glucose Monitoring in Interstitial Fluid

Florida Network Symposium

When Will CGM Replace SMBG? Roy W. Beck, MD, PhD. JAEB Center for Health Research Tampa, Florida

CGM and Closing The Loop

hypoglycaemia unawareness keystone 18 July 2014

Exercise Prescription in Type 1 Diabetes

Advances in Managing Diabetes in Youth. March 1, 2012

Update on Diabetes Technology

Objectives. Glucose Monitoring Today. Tracking Glucose ISO Standards. Continuous Glucose Monitoring & Diabetes Management 3/7/2015

Insulin Delivery and Glucose Monitoring Methods for Diabetes Mellitus: Comparative Effectiveness

Today s Goals 10/6/2017. New Frontiers in Diabetes Technology. Disclosures

Advances in Diabetes Care Technologies

1. What s the point of a network the case for research? 2. How to use CSII effectively

National Medical Policy

HOW CAN WE PREVENT RECURRENT-DKA?

The Impact of Accelerometer Use in Exercise-Associated Hypoglycemia Prevention in Type 1 Diabetes

Report Reference Guide. THERAPY MANAGEMENT SOFTWARE FOR DIABETES CareLink Report Reference Guide 1

King s Research Portal

Technology in Diabetes Management Irl B. Hirsch, MD University of Washington

NEWS BRIEFING Advances in Technology. moderated by: Irl Hirsch, MD University of Washington Medical Center

Clinical practice guidelines in management of type 1 diabetic patients

Limitations of Conventional Methods of Self-Monitoring of Blood Glucose

Emerging Automated Insulin Delivery Systems

The next five years in diabetes technology: closed-loop systems

UNO SGUARDO AL FUTURO. Basta la tecnologia a prevenire l ipoglicemia?

Pumps & Sensors made easy. OPADA ALZOHAILI MD FACE Endocrinology Assistant Professor Wayne State University

Hybrid Closed Loop Status & Practical Challenges in Implementation

Incorporating CGM Into Clinical Decision Making. Etie Moghissi, MD, FACE Clinical Associate Professor, David Geffen School of Medicine UCLA

Supplementary Appendix

Report Reference Guide

Advances in Diabetes Care Technologies

Continuous or Intermittent Monitoring of Glucose in Interstitial Fluid. Original Policy Date

Diabetes Technology Update. Sarah Konigsberg, MD Diabetes & Endocrine Assoc. April 7, 2018

Diabetes Technology Continuous Subcutaneous Insulin Infusion Therapy And Continuous Glucose Monitoring In Adults: An Endocrine Society Clinical

Diabetes Care 34: , 2011

CLINICAL UTILITY OF CONTINUOUS SUBCUTANEOUS INSULIN INFUSION IN PATIENTS WITH TYPE 1 DIABETES: A MACEDONIAN REPORT

Description. Section: Durable Medical Equipment Effective Date: July 15, 2014 Subsection: Original Policy Date: December 7, 2011 Subject:

Non-insulin treatment in Type 1 DM Sang Yong Kim

Sponsor / Company: Sanofi Drug substance(s): insulin glargine (HOE901) According to template: QSD VERSION N 4.0 (07-JUN-2012) Page 1

Artificial Pancreas, shortly close to home? BC Pediatric Diabetes Day, Feb 26, 2016

Diabetes through my eyes. Rick Mauseth, M.D. W.A.D.E. April 2013

The artificial pancreas: the next step in connectivity and digital treatment of type 1 diabetes

WOULD YOU LIKE MORE CONFIDENCE IN YOUR HYPOGLYCAEMIC CONTROL? SMARTGUARD. PROVIDES PREDICTIVE PROTECTION

In general: Hypoglycemia is common in insulin treated diabetes, but may also occur in people on oral medications, especially sulfonylureas/glinides.

Advances in Diabetes Care Technologies

Diabetes Care 34: , with an increased risk of SH. In a

Threshold-Based Insulin-Pump Interruption for Reduction of Hypoglycemia

VIRTUAL PATIENTS DERIVED FROM THE

Continuous Glucose Monitoring

CareLink. software REPORT REFERENCE GUIDE. Management Software for Diabetes

Artificial Pancreas Technologies: New Tools to Improve Diabetes Care Today and Tomorrow

Clinical Policy Title: Artificial pancreas device system

15 th Annual DAFNE collaborative meeting Tuesday 28 th June 2016

Norbert Hermanns, PhD 1,2, Beatrix Schumann, MD 2, Bernhard Kulzer, PhD 1,2, and Thomas Haak, MD 1,2. Original Article

Diabetes Care Publish Ahead of Print, published online September 11, 2007

CGM Use in Pregnancy & Unique Populations ELIZABETH O. BUSCHUR, MD THE OHIO STATE UNIVERSITY WEXNER MEDICAL CENTER

Endocrine Update Mary T. Korytkowski MD Division of Endocrinology University of Pittsburgh

Future Direction of Artificial Pancreas. Roy W. Beck, MD, PhD. JAEB Center for Health Research Tampa, Florida

Efficacy/pharmacodynamics: 85 Safety: 89

Basics of Continuous Subcutaneous Insulin Infusion Therapy. Lubna Mirza, MD Norman Endocrinology Associates 2018

Continuous Glucose Monitoring in Women With Type 1 Diabetes in Pregnancy Trial

Corporate Medical Policy

Ensuring Safe Physical Activity in Children and Adolescents With Type 1 Diabetes

Agreement between Glucose Trends Derived from Three Simultaneously Worn Continuous Glucose Sensors

Transcription:

PREVENTION OF NOCTURNAL HYPOGLYCEMIA USING PREDICTIVE LOW GLUCOSE SUSPEND (PLGS) Pathways for Future Treatment and Management of Diabetes H. Peter Chase, MD Carousel of Hope Symposium Beverly Hilton, Beverly Hills, CA October 11, 2014 No Pertinent Disclosures 1

Severe Hypoglycemia 1. Nocturnal Severe Hypoglycemia i) DCCT = 55% (Diabetes Care, 18:1415, 1995) ii)davis, et al 75% (Diabetes Care, 20:22, 1997) 2. Incidence of Severe Hypoglycemia i) General peds diabetes clinic: 19/100 patient years (Rewers, Chase, et al: JAMA 287:2511, 2002) ii) JDRF-RCT (CSII or MDI): 17.9/100 patient years (N Engl J Med 359:1464, 2008) iii) Star 3 (CSII + CGM) 495 patients = 13.3/100 patient years (N Engl J Med, 363:311, 2010) 2

Incidence of Severe Hypoglycemia per 100 Patient Years- T1D Exchange (Helmsley) (26,000 patients with T1D from 70 diabetes clinics in US) Garg S, et al. Presented at 5 th International Conference on Advanced Technologies & Treatment for Diabetes, Barcelona, 2012. 3

Mean HbA 1c by Age Group Type 1 Diabetes Exchange Registry (Helmsley) Bergenstal R, et al. Presented at 5 th International Conference on Advanced Technologies & Treatment for Diabetes, Barcelona, 2012.

PREVENTION OF HYPOGLYCEMIA IS A KEY TO IMPROVED GLYCEMIC CONTROL Development of the Artificial Pancreas (AP) will be necessary for the prevention of hypoglycemia. I. Low Glucose Suspend (LGS) II. Predictive Low Glucose Suspend (PLGS) III. Overnight AP Glucose Control IV. Complete AP

I. Hypoglycemia Prevention with Use of Low Glucose Suspend (LGS) in Sensor-Augmented Pumps (SAP) Source Patients/ Methods Results 21 youth with T1D from 3 centers in Germany 2 wks=sap 4 wks=sap with LGS Danne, T. et al., Diab. Tech. Ther. 13, 1129, 2011 Choudhary, P. et al., Diabetes Care 34, 2023, 2011 Agrawal, P. et al., J. Diab. Sci Tech. 5, 1137, 2011 31 adults with T1D in. 2 wks=sap 3 wks= SAP with LGS 935 patients 28,401 pt days/278 pts. > 3mos LGS feature in Real World SAP with LGS= 1) No SH or DKA 2) Decrease time spent (and decrease number of episodes) <70mg/dL 3) Positive device satisfaction SAP with LGS= 1) No deterioration of glucose control 2) time of nocturnal hypoglycemia in those with highest quartile of hypoglycemia at baseline 3) All subjects found LGS useful and 93%= (more secure at night) 1) Sensor glucose @ 2 hours=150±69 mg/dl 2) Significant reduction in BG values < 50 mg/dl when LGS in use (p < 0.001). Summary: in initial studies, time spent in hypoglycemia is reduced as a result of the use of the LGS feature. 6

ASPIRE: In Clinic Study I. Randomized Trial of LGS after Exercise in Clinic (Garg S, Brazg RL, Bailey TS, et al. Diabetes Technol Ther. 2012;14:205-209)

ASPIRE In-Clinic Study: Results Mean YSI Glucose Values by Time for LGS On and LGS Off Sessions (Garg S, Brazg RL, Bailey TS, et al. Diabetes Technol Ther. 2012;14:205-209)

ASPIRE In-Clinic Study: Results Duration and Severity of Hypoglycemia Duration of hypoglycemia (minutes) Nadir (mg/dl) End-Observation YSI (mg/dl) LGS On 138.5 ± 76.68 59.6 ± 5.72 91.4 ± 41.84 LGS Off 170.7 ± 75.92 57.6 ± 5.69 66.2 ± 13.48 P 0.006 0.015 <0.001 95% CI -57.7 to -7.8 0.3-3.8 12.3-38.1 Mean + SD Garg S, Brazg RL, Bailey TS, et al. Diabetes Technol Ther. 2012;14:205-209

Threshold Suspend Control ASPIRE- Home Study 247 patients were randomized to low glucose (threshold) suspend or control arms for a 3 month study. Age 41.6 ± 12.8 44.8 ± 13.8 % Male 38 39.7 Diabetes Duration 27.1 ± 12.5 26.7 ± 12.7 BMI, kg/m 2 27.6 ± 4.7 27.1 ± 4.3 Bergenstal RM, Klonoff DC, Garg SK et al. N Engl J Med 2013; 369: 224 232 10

ASPIRE In-Home Study: Reduction in Nocturnal Hypoglycemia (Bergenstal RM, Klonoff DC, Garg SK et al. N Engl J Med 2013; 369: 224 232) Mean AUC of Nocturnal Hypoglycemia Events AUC (mg/dl min) 1800 1600 1400 1200 1000 800 600 400 1547 (2035) 980 (1200) 38% reduction TS compared to Control p<0.001 1406 (1950) 1568 (1995) Run-In Phase Study Phase 200 0 Threshold Suspend Control Mean AUC of Nocturnal Hypoglycemia events was 38% lower in the Threshold Suspend Group. Similar A1c values at the End of the Study Similar Benefits in Children and Young Adults 11

ASPIRE In-Home Study: Percentage of time that SG values were in the <50, 50 to <60, and 60 to <70 mg/dl ranges. Bergenstal, Klonoff, Garg et al: NEJM 2013 12

II. Predictive Low Glucose Suspend (PLGS) Studies: Initial CRC Studies Diabetes Care 2010, 33:1013-1017 Two day admissions to CRC Use of Navigator CGM Increase basal rate up to 180% RESULTS: SG <60 mg/dl (<2.2 mmol/l): PLGS System ON: 28% PLGS System OFF: 85% 13

Example Plot: Predictive Low Glucose Suspend (PLGS) 14

Algorithm Description 1. Suspend for no more than 120 out of every 150 minutes. 2. Suspend for no more than 180 min/night. 3. Do not suspend when the glucose is rising. 4. Suspend if CGM <70 mg/dl. 5. Do not suspend if the CGM is above 230 mg/dl. 6. Do not suspend if the CGM is dropping by more than -8 mg/dl/min. 15

Initial outpatient (in-home) safety study 19 pts, ages 18 to 56 yrs with T1D Use of MiniMed and Paradigm Real-time Revel System 21 night randomized trial with PLGS on or off (2:1) PLGS occurred on 53% of 77 intervention nights Hypoglycemia (<70mg/dL; [<3.9mmol/L]) occurred in 16 16% of PLGS nights versus 30% of control nights

System Description Overall System 17

PLGS: Outpatient Study Randomization Flowchart of patients and randomized nights Did not meet hypoglycemia criteria N=4 Enrolled N= 50 Randomized 45 participants 2,007 nights Did not meet complete system use criteria N=1 <4 hrs of CGM data 32 nights Control 1,002 nights Intervention 1,005 nights <4 hrs of CGM data 63 nights Analyzed 970 nights Analyzed 942 nights Study Results 18

Demographics (N=45; 15-45 yo) PLGS: Outpatient Study Characteristic Value Age (yr) median (IQR) 30 (22, 39) Male / Female 21 / 24 Race White non-hispanic 42 (93%) Hispanic 2 (4%) Black 1 (2%) Weight (kg) median (IQR) 70 (62, 85) Height (cm) median (IQR) 172 (163, 178) Body-mass index (BMI) median (IQR) 24 (22, 27) Glycated hemoglobin (%) median (IQR) 6.8 (6.4, 7.6) Diabetes duration (yr) median (IQR) 15 (12, 26) Daily insulin dose (U/day) median (IQR) 45 (32, 60) 19

Example Plot PLGS: Outpatient Study Figure 1. Sample overnight data profile for intervention night System Description 20

Overall Results: 1,912 nights PLGS Study: Outpatient Study Control 970 Nights Intervention 942 Nights P- value Overnight Mean Glucose median 125 (98, 163) 132 (110, 163) <0.001 (IQR) Percentage of glucose between 70-180 75% (46%, 82% (54%, <0.001 mg/dl median (IQR) 93%) 99%) Morning BG mg/dlmedian (IQR) 129 (96, 173) 144 (114, 186) <0.001 % mornings with BG 60 mg/dl 4% <1% <0.001 70 mg/dl 9% 2% <0.001 71-180 mg/dl 70% 70% 0.82 >180 mg/dl 21% 27% 0.001 >250 mg/dl 6% 6% 0.83 21

Duration of Hypo and Hyperglycemia PLGS: Outpatient Study Control 970 Nights Intervention 942 Nights P-value Participants 45 45 Hypoglycemia % nights with glucose 60 mg/dl for > 30 min 24% 12% <0.001 > 60 min 18% 8% <0.001 > 120 min 11% 3% <0.001 > 180 min 8% 1% <0.001 Hyperglycemia % nights with glucose >250 mg/dl for > 30 min 12% 13% 0.80 > 60 min 9% 8% 0.74 > 120 min 5% 6% 0.35 22

Duration of Pump Suspension PLGS: Outpatient Study # Nights 942 Nights with Pump Suspension All durations 719 (76%) 1 30 min 185 (20%) 31 60 min 114 (12%) 61 120 min 262 (28%) >120 min 158 (17%) Shutoff Duration Per Night (minutes) * median (IQR) 71 (29, 115) *Restricted to 719 nights with at least one pump suspension. 23

Overall Results PLGS: Outpatient Study Control 970 Nights Intervention 942 Nights P-value Participants 45 45 Bedtime BG mg/dl median (IQR) 152 (114, 197) 144 (115, 195) # measurements per night 96 (84, 110) 96 (85, 107) % nights with >1 value 50 mg/dl 19% 10% <0.001 % nights with >1 value 60 mg/dl 33% 21% <0.001 % nights with >1 value 70 mg/dl 45% 32% <0.001 % nights with >1 value >180 mg/dl 57% 59% 0.17 % nights with >1 value >250 mg/dl 20% 20% 0.93 % nights with >1 value >300 mg/dl 5% 6% 0.37 24

Morning Blood Ketones vs. Insulin Pump Suspension/ Control Nights PLGS: Outpatient Study Morning Blood Ketones (Precision Meter) System Active All Nights (N=977) System Active 1 Suspension (N=744) 0 0.5 964 (99%) 733 (99%) 966 (99%) 0.6 6 (<1%) 4 (<1%) 5 (<1%) 0.7 2 (<1%) 2 (<1%) 2 (<1%) 0.8 1 (<1%) 1 (<1%) 1 (<1%) 0.9 1 (<1%) 1 (<1%) 1 (<1%) 1.0 2 (<1%) 2 (<1%) 1 (<1%) 1.1 1 (<1%) 1 (<1%) 1 (<1%) System Not Active All Nights (N=977)

Morning Urine Ketones vs. Insulin Pump Suspension/Control Nights PLGS: Outpatient Study Morning Urine Ketones System Active All Nights (N=977) System Active 1 Suspension (N=744) Negative 875 (90%) 656 (88%) 901 (92%) System Not Active All Nights (N=977) Trace 75 (8%) 64 (9%) 57 (6%) Small 25 (3%) 22 (3%) 12 (1%) Moderate 2 (<1%) 2 (<1%) 6 (<1%) Large 0 (0%) 0 (0%) 1 (<1%)

Summary Artificial Pancreas (AP) PLGS: Outpatient Study 1. Hypoglycemia (<60 mg/dl) lasting >2 hrs was decreased by 74% 2. The AUC <60 mg/dl was 81% lower with PLGS nights compared to control nights 3. Overnight mean glucose was higher with PLGS nights (median=132 mg/dl [7.3 mmol/l]) versus control nights (125 mg/dl [6.9 mmol/l]) 4. Median AM BG was 144 mg/dl (7.9 mmol/l) with PLGS and 129 mg/dl (7.2 mmol/l) for control nights 5. The likelihood of developing ketones with 120 to 180 minutes of pump should not be required in future similar suspension studies.

Future Studies 1) PLGS in 11-14 yo age group 2) PLGS in 7-13 yo age group 3) PLGS in 3-6 yo age group 4) PLGS Pump Suspension with Hyperglycemia Minimization (PHM-1) Studies Thank You 28

Incidence of Biochemical Nocturnal Hypoglycemia JDRF Randomized Clinical Trial: 180 subjects with T1D Use of Navigator or Paradigm RT Systems: 36,000 nights, 2.4 million CGM readings. Frequency of Nocturnal Hypoglycemia 70 mg/dl 60 mg/dl 50 mg/dl ( 3.9 mmol/l) ( 3.3 mmol/l) ( 2.8 mmol/l) % of nights Single value hypoglycemic 25% 15% 7.6% 2 consecutive values hypoglycemic 23% 13% 6.7% 20 consecutive minutes hypoglycemic 18% 9.6% 4.5% 60 consecutive minutes hypoglycemic 9.7% 4.6% 2.0% Two predictors: Lower HbA1c and hypoglycemia on blinded baseline CGM JDRF-RCT: Diabetes Care, 33:1004, 2010 29

PLGS: Outpatient Study Outpatient Glucose Levels Figure 4. % of nights with a CGM 60 mg/dl at a given time (dashed lines) 30

Nocturnal Severe Hypoglycemia (SH) Incidence: 55% of SH in DCCT (Diabetes Care 18,1415,1995) 75% of SH in children (Davis, Diabetes Care 20:22,1997) Causes: 1) Strenuous daytime exercise 2) Attenuated glucagon and epinephrine responses a. Particularly at night b. Particularly with antecedent hypoglycemia or prior exercise c. Reduced cognition: 71% of youth failing to respond to nocturnal alarms (Buckingham, Chase et al. DT&T, 7,440,2005) 3) Lower HbA1C 4) Wrong insulin dose 31

Mean HbA 1c by Age Group Type 1 Diabetes Exchange Registry (Helmsley) Bergenstal R, et al. Presented at 5 th International Conference on Advanced Technologies & Treatment for Diabetes, Barcelona, 2012. 32

Effect of Exercise Mean Glucose Level on Two Study Days (Same 50 Subjects) Impact of exercise on overnight glycemic control in children with type 1 diabetes mellitus. DirecNet: J Pediatr, 2005 Oct; 147(4):528-34 33

Decreased Epinephrine Responses in Children with T1D Subjects: 14 children 4-7 yo; duration 3.3 years 14 adolescents 12-17 yo; duration 6.6 years No epinephrine response (>3SD above baseline) to insulininduced hypoglycemia (<60 mg/dl [<3.3 mmol/l]) 9 of 14 young children (average 47 82 pg/ml) 8 of 14 adolescents (average 27 71 pg/ml) 14 nondiabetic adolescents increased from 77 to 582 pg/ml in response to hypoglycemia Glucagon levels did not increase with hypoglycemia in young children or adolescent subjects with T1D (DirecNet, Diabetes Care, 32:1954, 2009) 34

ASPIRE: I. Randomized Trial of LGS after Exercise in Clinic

PLGS to prevent BG <60mg/dL (<2.2mmol/L) 22 subjects with 2 daytime admissions to CRC Navigator CGM Algorithm development: 80mg/dL threshold 30 minute projection: 60% prevention 45 minute projection: 80% prevention 36

37

Kalman Filter Look at all times after rule is active Activated 46042 times affecting 218 days ID# Morning BG Mean CGM % time < 70 % time > 180 mg/dl Risk With 122 114 10.0 8.6 4.7 without 106 103 21.1 6.4 7.5 Effect of Rule 15 10.9-11.1 2.1-2.8 Simulated Effect of Algorithm Rules 38

Morning Blood and Urine Ketones Control N=45 Participants Intervention N=45 Participants # Mornings 970 942 Morning Measures % mornings with a blood ketone a 0 mmol/l 14% 14% 0.1 mmol/l 64% 63% 0.2 0.3 mmol/l 19% 20% 0.4 0.6 mmol/l 2% 2% 0.7 mmol/l 0.6% 0.6% % mornings with a urine ketone b 40 mg/dl 0.7% 0.2% 80 mg/dl 0.1% 0 a-9 blood ketone measurements in the control group and 10 blood ketone measurements in the intervention group were missing b-12 urine ketone measurements in the control arm and 12 urine ketone measurements in the intervention arm were missing 39