MEDCAC subcommittee (Devices for Self-management of Type 1 and Insulin-Dependent Type 2 Diabetes)

Committee Chair:
Joseph Ross, MD, MHS
ProfessorMedicine and Public Health
Section of General Internal Medicine
Department of Medicine
Yale University School of Medicine

Committee Vice Chair:
Sanket Dhruva, MD, MHS, FACC
Assistant Professor of Medicine
UCSF School of Medicine
Affiliate Faculty, Philip R.Lee Institute for Health Policy Studies
Staff Physician (Cardiology)
San Francisco VA Medical Center

MEDCAC Members:

Fred Kobylarz MD, MPH
Professor, Department of Family Medicine and Community Health
Parker Division of Geriatrics
Robert Wood Johnson Medical School
Rutgers Biomedical and Health Sciences

Joy H. Lewis, DO, PhD
Professor, Medicine and Public Health
Chair, Department of Public Health
Director, DO/MPH Program
Director, National Center for Community Health Research
A.T. Still University-School of Osteopathic Medicine in Arizona

Patient Advocate

Naftali Z. Frankel, MS, MBEE
Consumer & Patient Advocate
Director of Precision Medicine Innovation, CMR
Adjunct Professor, Johns Hopkins University

Mark D. Carlson, MD, MA
Chief Medical Officer
Sr. Vice President for Clinical Affairs
Covanos

Overview of Clinical Endpoints Review

Clinical Endpoints Review

Detailed Summary of Clinical Endpoints Review

REVIEW QUESTIONS

After reading the full CER and using the foregoing summary in this document, please provide feedback on the following issues. You may type your answers into this document and return it or answer the questions in an email.

1. Do you believe that the search strategy used in the CER is likely to have captured the appropriate literature? (See the “Identifying the Literature” section, starting on page 5 of the full CER). The search strategy used was comprehensive and was able to capture high quality studies available in the literature. Given the dramatic advances in technology in the past few years, I think limiting to the last 5 years is very reasonable.
2. Are the appropriate outcome domains for diabetes management research reflected in the CER findings? (The CER found outcomes in the Clinical, Qualitative, Resource Use and Safety domains). Are important domains missing, and are omissions likely due to deficiencies in the literature search strategy or omissions in the available body of clinical research literature?

   Overall, I agree with the CER focus on Clinical, Qualitative, Resource Use and Safety domains.

   One additional area (which may fit under one of these existing ones but only briefly touched upon) is accuracy of the technology under hypoglycemic conditions. For all current available CGM systems, the MARDs in hypoglycemic ranges are significantly higher than within euglycemic or hyperglycemic ranges. To my knowledge, there are virtually no clinical trials which have addressed this in any of the secondary or post-hoc analyses. Clinical trials using CGM should provide some rationale for choice of device, taking into account accuracy in hypo ranges. Furthermore, in many of the reported trials, CGM output of TIR or % time in hypoglycemia come from the generated AGP (ambulatory glucose profile). There is very little in terms of methodology addressing the analysis of CGM data (accounting for gaps in data, false hypo readings such as from compression lows etc.).

3. Consider the physiological/clinical outcomes in Table A1 of this document.

   1. Should these be considered key outcomes for studies investigating the use of devices for self-management of diabetes in older adults?
      I agree with Table A1 that A1C, level 1, 2 and 3 hypoglycemia as well as level 2 hyperglycemia should be considered key outcomes in the Clinical domain for older adults. I also agree with the consensus guidelines reviewed that the focus should be on A1C, TIR in younger adults but shift to more emphasis on hypoglycemia avoidance in older.
   2. Should additional key clinical outcomes be considered – either outcomes listed in Table A2 or outcomes unrepresented in the CER?
      Time of day of hypoglycemia is not on any of the tables and is an infrequently reported outcome measure. There are few studies looking at this but nocturnal hypoglycemia (and reduction of nocturnal hypos) is a clinically meaningful end point, particularly for older individuals. Ability to detect and treat true lows at night is clinically important, a source of stress and distress for many patients leading to undertreatment of diabetes and associated with greater risk to older adults with underlying CVD.

      Number of episodes of hypoglycemia as opposed to %time in hypo is more difficult to ascertain from CGM data (this is not defined in current AGPs) but is perhaps more meaningful outcome. I would be more concerned about a patient who has discrete episodes of hypo down to 55 rather than a person who hovers around 68-70 for cumulatively for several hours a week (even if they had the same % time in level 1 hypo)

      Hypoglycemia unawareness is also not represented in the CER. Hypo unawareness is a condition where individuals lose the normal, protective physiologic response to hypoglycemia. This usually occurs in individuals with frequent hypoglycemia and is particularly worrisome in older individuals. One important area of investigation is whether CGM/pump/hCL systems can restore awareness to hypoglycemia. There is a NIDDK led consortium that will look at this but the study is in early stages.

   3. Are the thresholds specified for these outcomes appropriate, e.g., <70 mg/dL as the definition of hypoglycemia?
      In my opinion, yes, 70 mg/dl is a pragmatic definition of hypoglycemia to use. I would reference a 2009 editorial written by Dr. Philip Cryer (Diabetologia 2009 52:35) which beautifully summarizes the rationale for using 70 mg/dl as a cut-off. Although technology has advanced, this rationale is evidence based and still the best justification. He writes “this conservative lower limit for individuals with diabetes approximates the lower limit of the postabsorptive plasma glucose concentration range (approximately 3.9–6.1 mmol/l [70–110 mg/dl] [6]) and the glycaemic threshold for activation of glucose counter-regulatory systems (approximately 3.6–3.9 mmol/l [65–70 mg/dl] [6–9]), and is low enough to cause reduced glucose counter-regulatory responses to subsequent hypoglycaemia [10] in non-diabetic individuals. It is higher than the glucose levels required to produce symptoms in non-diabetic individuals (approximately 2.8–3.1 mmol/l [50–55 mg/dl] [6–9]) and substantially higher than those that do so in people with well-controlled diabetes [11], although individuals with poorly controlled diabetes sometimes have symptoms at higher glucose levels [11, 12]. Thus, the recommended glucose alert level of =3.9 mmol/l (70 mg/dl) is data-driven, generally gives the patient time to take action to prevent a clinical hypoglycaemic episode, and provides some margin for the limited accuracy of glucose monitoring devices at low plasma glucose concentrations [5]. The ADA Workgroup-recommended alert value does not, of course, mean that people with diabetes should always self-treat at an estimated plasma glucose concentration of =3.9 mmol/l (70 mg/dl). Rather, it suggests that they should consider actions ranging from repeating the measurement in the short term, through behavioural changes such as avoiding exercise or driving, to carbohydrate ingestion and adjustments of the treatment regimen.”

4. Consider the nonclinical outcomes listed in Table A1.
   1. Which of these should be considered key outcomes for studies investigating the use of devices for self-management of diabetes in older adults?
   2. Should additional nonclinical outcomes be considered?
      

      QOL is an important outcome and I would add that measures of sleep disturbance have not been included in the CER (few studies). Clinically, CGM alarms at night disrupt sleep and are a cause of distress for many patients who use devices. This is less of an issue with the current hCL systems as they tend to be particularly good at stabilizing overnight glucoses.

      Safety/adverse events should always be key outcomes.

5. Consider the quality-of-life instruments listed in Appendix Table B3 of the CER. Are these appropriate for studies investigating the use of devices for self-management of diabetes in older adults, and are any important instruments missing?
   A measure of hypoglycemia awareness should be included. There are several which have been used in studies. Of those, my opinion is the Gold score (Gold et al. 1994 Diabetes Care 17:697) or Hypo A-Q (Speight et al 2016 Diabet Med 33:376) best meet the criteria for patient reported outcome measures.
6. Can you offer definitions of a minimal clinically important difference (MCID) for the physiological/clinical outcomes that you designated to be key outcomes in response to question #3? An MCID is the smallest benefit of value to patients, for example, hypothetically, a 10% reduction in time spent outside an acceptable A1c range. Please keep in mind the typical age of the Medicare population.
   A reduction in level 2 hyperglycemia (>250) by 10% would be clinically important. This is based on some rough estimates that for every 10% change in TIR, it correlates roughly with a 0.8% change in A1C and a 0.5% change in A1C has been defined previously as being a MCID.

   For hypoglycemia, this is more challenging to define in terms of reductions given the relatively low % time spent in hypo to begin with. Thus, thinking in terms of % reduction may not be that meaningful. Also, the number of episodes of hypoglycemia may be more clinically meaningful than the % time spent as described above. For example, in my clinical practice, particularly in my older patients, I aim for 0 episodes of hypo at night. If they have fewer than 1 episode of hypoglycemia (level 1) a month then I am less concerned (and they are less concerned). Rates of severe hypo are quite low in the literature but probably grossly underreported. Severe hypo is associated with greatest morbidity and mortality (and predict additional future episodes of hypo), so ability to achieve 0% level 2 and severe hypoglycemia is a clinically important difference as a binary measure.